143
PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail [email protected] Web www.promotion-offshore.net This result is part of a project that has received funding form the European Union’s Horizon 2020 research and innovation programme under grant agreement No 691714. Publicity reflects the author’s view and the EU is not liable of any use made of the information in this report. CONTACT WP 4.7 Deliverable: Preparation of cost-benefit analysis from a protection point of view

WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail [email protected] Web This

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail [email protected] Web www.promotion-offshore.net This result is part of a project that has received funding form the European Union’s Horizon 2020 research and innovation programme under grant agreement No 691714. Publicity reflects the author’s view and the EU is not liable of any use made of the information in this report. CONTACT

WP 4.7 Deliverable: Preparation of cost-benefit analysis from a protection point of view

Page 2: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

i

I. DOCUMENT INFO SHEET Document Name: D4.7 - Preparation of cost- benefit analysis from a protection point of view Responsible partner: SuperGrid Institute Work Package: WP 4 Work Package leader: Dirk Van Hertem (KU Leuven) Task: 4.5 Task lead: Serge Poullain (SuperGrid Institute)

A.1 DISTRIBUTION LIST

Public

A.2 APPROVALS

Name Company Validated by: Dirk Van Hertem KU Leuven Task leader: Serge Poullain SuperGrid Institute WP Leader: Dirk Van Hertem KU Leuven

WP Number

WP Title Person months Start month End month

4 DC Grid Protection System Development 363 4 48

Deliverable Number

Deliverable Title Type Dissemination level

Due Date

D4.7 Preparation of cost- benefit analysis from a protection point of view Report Public 47

Page 3: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

ii

II. LIST OF CONTRIBUTORS

The Work Package and deliverable involve a large number of partners and contributors. The names of the partners, who actively contributed to the realization of the present deliverable, are presented in the following table.

PARTNER NAME

SuperGrid Institute Boussaad Ismaïl

SuperGrid Institute Serge Poullain

SuperGrid Institute Alberto Bertinato

SuperGrid Institute Bruno Luscan

SuperGrid Institute Bertrand Raison

SuperGrid Institute Philippe Chaumès

KU Leuven Jay Dave

KU Leuven Dirk Van Hertem

KU Leuven Hakan Ergun

Page 4: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

iii

III. LIST OF ABBREVIATIONS

ABBREVIATION EXPLANATION AC Alternative Current

ACCB AC Circuit Breaker

DCCB DC Circuit Breaker

Bci Breaking module for converter i

Bri Breaking module for line i

CAPEX CAPital EXpenditures (investment costs)

CBA Cost Benefit Analysis

COE Cost Of Energy

DC Direct Current

DCR DC Reactor

EC Energy Cost

ECER European Community Energy Roadmap

EENT Expected Energy Not Transmitted

EMT Electro-Magnetic Transient

ENTSO-E European Network of Transmission System Operators for Electricity

FB-MMC Full Bridge MMC

FCR Frequency Containment Reserves

FCU Fault Clearing Unit

FFR Fast Frequency Reserve

FS-FDCCB Fully Selective using fast DCCB

FS-SDCCB Fully Selective using slow DCCB

HB-MMC Half Bridge MMC

HDCCB Hybrid DC Circuit Breaker

HSS High Speed Switch

HVDC High Voltage Direct Current

IGBT Insulated-Gate Bipolar Transistor

KPIs Key Performance Indicators

LCC Life Cycle Cost

LOLE Loss of Load expected

MC Monte-Carlo

MCPN Monte Carlo Petri Net

MDCCB Mechanical DC Circuit Breaker

MMC Modular Multilevel Converter

MTDC Multi-Terminal DC

MTTF Mean Time To Failure

MTTR Mean Time To Repair

Page 5: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

iv

NPV Net Present Value

NS-CB Non-Selective using DCCB at Converter output

NS-FB Non-Selective using Converter with fault blocking capability

O Open

OCO Open Close Open

OPEX Operational Expenditures (operational costs)

OPF Optimal Power Flow

PCC Point Common Connections

PCI Project of Common Interest

PF Power Flow

PIR Pre-Insertion Resistor

PMR Preventive Maintenance Rate

PRR Pole Rebalancing Reactor

PtG Pole To Ground

PtP Pole To Pole

RCB Residual Current Breaker

ROCOF Rate of Change of Frequency

RES Renewable Energy Systems

SA Surge Arrester

SCC Short Circuit Current

SCR Short Circuit Ratio

SEW Social Economic Welfare

TIV Transient Interruption Voltage

TYNDP Ten Year Network Development Plan

UFD Ultra-Fast Disconnector

UFLS Under Frequency Load Shedding

VoLL Value of Lost Load

VSC Voltage Source Converter

WP Work Package

WP 4 Work Package 4: DC Grid protection system development - aims to study and develop multivendor DC grid protection systems

WP 4.2 Work Package 4.2: Screening analysis of various protection methods for different topologies

WP 4.3 Work Package 4.3: In depth study of selected protection methods towards practical implementation

WP 4.5 Work Package 4.5: Preparation of cost/benefit analysis of studied protection solutions

WP 7 Work Package 7: Regulation and Financing

WP 12 Work Package 12: Deployment plan for future European offshore grid

Page 6: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

v

IV. LIST OF FIGURES

FIGURE 1: GENERAL FRAMEWORK FOR CBA ASSESSMENT AND COMPARISON ...................................................................... 19 FIGURE 2: TYPICAL OUTPUTS FOR RISK ANALYSIS ASSESSMENT ......................................................................................... 20 FIGURE 3: MODEL REQUIREMENTS FOR INDICATORS CALCULATION ..................................................................................... 21 FIGURE 4: PROPOSED GENERAL ASSESSMENT AND COMPARISON METHODOLOGY ................................................................. 22 FIGURE 5: COST INDICATORS QUANTIFICATION PROPOSED APPROACH ................................................................................ 23 FIGURE 6: CAPEX CALCULATION METHODOLOGY FOR DC GRID PROTECTION EQUIPMENT DEVELOPED WITHIN PROMOTION WP4 24 FIGURE 7: LOSSES CALCULATION METHODOLOGY FOR DC GRID PROTECTION EQUIPMENT DEVELOPED WITHIN PROMOTION WP4 25 FIGURE 8: EENT CALCULATION METHODOLOGY ............................................................................................................. 26 FIGURE 9: EENT CALCULATION MAIN STEPS .................................................................................................................. 26 FIGURE 10 GENERAL FRAMEWORK OF ENERGY NOT TRANSMITTED MONTE-CARLO BASED MODEL ............................................ 27 FIGURE 11 EXAMPLE OF UNAVAILABILITY OF LINE AND CONVERTER DUE TO PROTECTION STRATEGY COMPONENT: A) BR1 FAILURE

LEADS TO UNAVAILABILITY OF LINE L13, B) BC1 FAILURE LEADS TO UNAVAILABILITY OF CONVERTER 1 ............................... 28 FIGURE 12: MAINTENANCE COST CALCULATION METHODOLOGY ........................................................................................ 29 FIGURE 13: PROPOSED INDICATORS FOR OPERATIONAL RISK ANALYSIS............................................................................... 29 FIGURE 14: PROPOSED INDICATORS FOR "SIMPLIFIED" OPERATIONAL RISK ANALYSIS ............................................................. 30 FIGURE 15: MAXIMUM LOSS OF POWER INFEED AND DURATION OF AN AC GRID CONNECTED TO A HVDC GRID [10] ..................... 30 FIGURE 16: PROPOSED INDICATORS FOR "SIMPLIFIED" OPERATIONAL RISK ANALYSIS ............................................................. 31 FIGURE 17: EXAMPLES OF RELIABILITY ASSESSMENT OUTPUTS- PROTECTION SYSTEM FAILURE AND SUCCESS RATE REPARTITION .. 33 FIGURE 18: EXAMPLES OF RELIABILITY ASSESSMENT OUTPUTS- DISTRIBUTION OF “FAULT INTERRUPTION TIME” .......................... 33 FIGURE 19: EQUIVALENT MODEL OF INERTIAL AND PRIMARY FREQUENCY CONTROL RESPONSE [13] [14] .................................... 36 FIGURE 20: (A) ACTIVE POWER DEVIATION CURVE AT PCC (B) FREQUENCY RESPONSE REPRESENTATION ................................. 37 FIGURE 21: FREQUENCY RESPONSE WITH H = 3.2 S, FCR = 1450 MW IN SCENARIO 2025 (RED LINE – MINIMUM PERMISSIBLE

FREQUENCY) .................................................................................................................................................... 39 FIGURE 22: (A) POWER DEVIATION (B) FREQUENCY RESPONSE WITH H = 3.2 S, FCR = 1450 MW IN SCENARIO 2025 ................. 39 FIGURE 23: FREQUENCY RESPONSE MODEL OF NORDIC GRID WITH FFR [19] ....................................................................... 40 FIGURE 24: METHODOLOGY TO DETERMINE FFR REQUIREMENT ........................................................................................ 41 FIGURE 25: METHODOLOGY FOR COST BENEFIT ANALYSIS OF DC PROTECTION IMPACT .......................................................... 43 FIGURE 26: DCCBS INPUT PARAMETERS AND MAIN COST’S OUTPUTS ................................................................................. 47 FIGURE 27: LINE BREAKER CURRENT CAPABILITY ESTIMATION: EXPLANATION OF THE WORST CASE FOR THE LINE BREAKER CLOSE TO

CONVERTER 1 .................................................................................................................................................. 48 FIGURE 28: SUMMARY OF CONVERTER BREAKER STRATEGY DCCBS DESIGN ...................................................................... 49 FIGURE 29: MAIN DEVELOPED TOOLS WITH THEIR RESPECTIVE INPUTS AND OUTPUTS ............................................................ 53 FIGURE 30: ECONOMIC KEY PERFORMANCE INDICATORS FRAMEWORK MODULES .................................................................. 55 FIGURE 31: STOCHASTIC EFFICIENCY AND RELIABILITY KEY PERFORMANCE INDICATORS FRAMEWORK MODULES ......................... 56 FIGURE 32: MATLAB MODULES FOR (A) FREQUENCY QUALITY DEFINING KPIS (B) RESERVE SIZE CALCULATIONS ........................ 57 FIGURE 33: REDISPATCH COST CALCULATION ALGORITHM ................................................................................................ 58 FIGURE 34: BENCHMARK NETWORK FOR SMALL IMPACT SYSTEM (ADOPTED FROM WP2 AND WP3, SEE [1] [3])........................... 61 FIGURE 35: CAPITAL COSTS BREAKDOWN OF DIFFERENT USED DCCBS WITHIN WP4.2 BENCHMARK. ....................................... 63 FIGURE 36: TOTAL COSTS FOR FULLY SELECTIVE (NS-FDCCBS, FS-SDCCB) AND NON-SELECTIVE (NS-CB, NS-FB) PROTECTION

STRATEGIES ..................................................................................................................................................... 63 FIGURE 37: EXPECTED ENERGY NOT TRANSMITTED (% OF 2.4 GW). ................................................................................. 64 FIGURE 38: PART OF GRID COMPONENTS (CONVERTERS TRANSFORMERS AND CABLES), PROTECTION COMPONENTS (DCCBS AND

HSS) ON THE EXPECTED ENERGY NOT TRANSMITTED (EENT). ................................................................................. 64 FIGURE 39: ENERGY LOSSES IN THE WP4.3 PROMOTION BENCHMARK SYSTEM FOR DIFFERENT PROTECTION STRATEGIES (% OF 2.4

GW). .............................................................................................................................................................. 65 FIGURE 40: TEST CASE: DC GRID CONNECTED TO THE NORDIC GRID .................................................................................. 66 FIGURE 41: TEST CASE: DC GRID CONNECTED TO THE NORDIC GRID .................................................................................. 68

Page 7: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

vi

FIGURE 42: LOAD SHEDDING DURATION VS FFR CAPACITY FOR WP 4.3 TOPOLOGY (A) SELECTIVE STRATEGIES (B) NONSELECTIVE STRATEGIES ..................................................................................................................................................... 69

FIGURE 43 GENERIC GRID EXAMPLE TO ANALYZE (A) THE INFLUENCE OF THE INJECTION CAPCITY (B) THE INFLUENCE OF THE NO. OF CONNECTION POINTS. ........................................................................................................................................ 70

FIGURE 44 FFR REQUIREMENT FOR DIFFERENT INJECTION CAPACITIES FOR (A) FS, ∆THVDC = 100 MS & 400 MS (B) NS, ∆THVDC = 100 MS (C) NS, ∆THVDC = 400 MS .................................................................................................................. 72

FIGURE 45 FFR REQUIREMENT FOR DIFFERENT NUMBER OF CONNECTIONS FOR (A) FS, ∆THVDC = 100 MS & 400 MS (B) NS, ∆THVDC = 100 MS (C) NS, ∆THVDC = 400 MS.................................................................................................... 73

FIGURE 46: TEST CASE: DC GRID CONNECTED TO IEEE 118 BUS SYSTEM .......................................................................... 74 FIGURE 47: BRANCH FLOW VIOLATIONS FOR FS AND NS SCHEMES IN PRIMARY SEQUENCE (A) PERCENTAGE OVERLOADS IN

BRANCHES WITH POWER FLOW CONSTRAINT VIOLATIONS (OUT OF 500 SAMPLES) (B) RANGE OF OVERLOADS OCCURRING IN THE BRANCHES ....................................................................................................................................................... 76

FIGURE 48: LIFE CYCLE COST DEPENDING ON UNAVAILABILITY (FROM 0.0003 AND 0.0009)..................................................... 78 FIGURE 49: “BACKBONE” GRID STRUCTURE. .................................................................................................................. 80 FIGURE 50: “DOGGER BANK” RING MESHED VARIATION OF “BACKBONE” GRID STRUCTURE. ...................................................... 81 FIGURE 51: “WIDER” RING MESHED VARIATION OF “BACKBONE” GRID STRUCTURE. ................................................................ 82 FIGURE 52: WITHOUT INTERCONNECTORS VERSION OF “BACKBONE” GRID STRUCTURE .......................................................... 82 FIGURE 53: NON SELECTIVE (CONVERTER BREAKER AND FULL BRIDGE) AND FULLY SELECTIVE (WITH HYBRID AND MECHANICAL

DCCBS) PROTECTION STRATEGY LAYOUTS. ........................................................................................................... 86 FIGURE 54: DCCBS UNIT COSTS USED FOR “BACKBONE”, “DOGGER BANK” RING, “WINDER” RING AND “BACKBONE” WITHOUT

INTERCONNECTOR GRID STRUCTURES. .................................................................................................................. 87 FIGURE 55: FULLY SELECTIVE, NON SELECTIVE AND FULL BRIDGE OPTIONS TOTAL INVESTMENT COSTS. .................................... 89 FIGURE 56: “BACKBONE” GRID STRUCTURE TOTAL COSTS (INCLUDING PROTECTION STRATEGIES) OF ALL PROTECTION STRATEGY

OPTIONS. ......................................................................................................................................................... 90 FIGURE 57: COMPONENT CONTRIBUTION IN TOTAL “BACKBONE” GRID STRUCTURE. ............................................................... 90 FIGURE 58: “BACKBONE” GRID STRUCTURE CAPEX INCLUDING ONLY: ................................................................................ 91 FIGURE 59: “BACKBONE” GRID STRUCTURE MAINTENANCE (INCLUDING PROTECTION STRATEGIES) OF ALL PROTECTION STRATEGY

OPTIONS. ......................................................................................................................................................... 91 FIGURE 60: TOTAL LOSSES FOR “BACKBONE” GRID STRUCTURE (% OF TOTAL LOSSES, FOR NO CONTINGENCIES AND 100% WIND

GENERATION, I.E. 25664 MW). ............................................................................................................................ 92 FIGURE 61: COMPONENT CONTRIBUTIONS TO LOSSES (% OF TOTAL LOSSES, FOR NO CONTINGENCIES AND 100% WIND GENERATION,

I.E. 25664 MW). ............................................................................................................................................... 92 FIGURE 62: EXPECTED ENERGY NOT TRANSMITTED (PER UNIT, PERCENTAGE OF TOTAL ENERGY). ........................................... 93 FIGURE 63: COMPONENT CONTRIBUTIONS TO EENT (% OF TOTAL EENT). ......................................................................... 93 FIGURE 64: DIFFERENT FAULT LOCATIONS CONSIDERED IN EEMT SIMULATION FOR EFFICIENCY KPIS CALCULATION. ................... 95 FIGURE 65: EMT SIMULATION CURVES FOR FAULT LOCATION F3 AND NON-SELECTIVE CONVERTER BREAKER PROTECTION STRATEGY

NS-CB ........................................................................................................................................................... 96 FIGURE 66: NODAL VOLTAGE, ACTIVE AND REACTIVE POWER TIME RESTORATIONS FOR FAULT LOCATION F3 AND NON-SELECTIVE

CONVERTER BREAKER PROTECTION STRATEGY NS-CB. ........................................................................................... 97 FIGURE 67: NODAL VOLTAGE, ACTIVE AND REACTIVE POWER TIME RESTORATION DISTRIBUTIONS FOR FAULT LOCATION F3 AND NON-

SELECTIVE CONVERTER BREAKER PROTECTION STRATEGY NS-CB. ............................................................................ 97 FIGURE 68 EMT SIMULATION CURVES FOR FAULT LOCATION F3 AND FULLY SELECTIVE USING FAST DCCB PROTECTION STRATEGY

FS-FDCCB ..................................................................................................................................................... 99 FIGURE 69: NODAL VOLTAGE, ACTIVE AND REACTIVE POWER TIME RESTORATIONS FOR FAULT LOCATION F3 FULLY SELECTIVE USING

FAST DCCB PROTECTION STRATEGY FS-FDCCB. ................................................................................................ 100 FIGURE 70: NODAL VOLTAGE, ACTIVE AND REACTIVE POWER TIME RESTORATION DISTRIBUTIONS FOR FAULT LOCATION F3 FULLY

SELECTIVE USING FAST DCCB PROTECTION STRATEGY FS-FDCCB. ........................................................................ 100 FIGURE 71: EFFICIENCY KPIS COMPARISON BETWEEN NS-CB, FS-FDCCB AND FS-SDCCB PROTECTION STRATEGIES. .......... 102 FIGURE 72: EMT SIMULATION CURVES FOR GRID SPLITTING STRATEGY USING NS-CB PROTECTION STRATEGY. DC ACTIVE POWER

BEFORE, DURING AND AFTER A FAULT FOR FAULT LOCATIONS F11. ............................................................................ 104 FIGURE 73: EMT SIMULATION CURVES FOR GRID SPLITTING STRATEGY USING NS-CB PROTECTION STRATEGY. DC ACTIVE POWER

BEFORE, DURING AND AFTER A FAULT FOR FAULT LOCATIONS F12. ............................................................................ 105

Page 8: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

vii

FIGURE 74: NODAL ENERGY IMBALANCE (MJ) FOR GRID SPLITTING WITH NS-CB STRATEGY AND FAULT LOCATION F1, F3, F11 AND F12. ............................................................................................................................................................. 106

FIGURE 75: EFFICIENCY KPIS COMPARISON BETWEEN NS-CB AND GRID SPLITTING WITH NS-CB PROTECTION STRATEGIES: PROBABILITY DISTRIBUTION FOR F12 (GRID SPLITTING) AND AGGREGATED FAULT LOCATIONS F1, F3 AND F5 (NS-CB) AND F1, F3 AND F11 (GRID SPLITTING) ........................................................................................................................... 107

FIGURE 76: CONNECTIONS TO NORDIC GRID IN BACKBONE GRID ...................................................................................... 108 FIGURE 77: NORDIC GRID: (A) EQUIVALENT INERTIA OF NORDIC GRID (B) PERCENTAGE RENEWABLES OF THE TOTAL GENERATION

................................................................................................................................................................... 109 FIGURE 78: RATE OF CHANGE OF FREQUENCY (ROCOF) FOR BACKBONE GRID .................................................................. 109 FIGURE 79: FFR REQUIREMENT FOR PLANNING YEAR 2050 (A) ACTIVATION TIME = 0.2 SEC (B) ACTIVATION TIME = 0.1 SEC ........ 110 FIGURE 80: FREQUENCY CONTROL ACTION WITHIN CE [38] ............................................................................................ 120 FIGURE 81: (A) FULLY SELECTIVE PROTECTION STRATEGY WITH BACKUP OPTIONS (B) FAULT CLEARING UNIT (FCU) CONSISTING OF

INDUCTOR (L), HIGH-SPEED SWITCH OR RESIDUAL CURRENT BREAKER (HSS/RCB) AND DC CIRCUIT BREAKER (BR) [1] ...... 121 FIGURE 82: TABLE 46 NON-SELECTIVE PROTECTION STRATEGY WITH FULL BRIDGE MMCS [1]............................................... 122 FIGURE 83: (A) NON-SELECTIVE PROTECTION STRATEGY USING CONVERTER BREAKER (B) FAULT CLEARING UNIT (FCU) CONSISTING

OF HIGH-SPEED SWITCH OR RESIDUAL CURRENT BREAKER (HSS/RCB) AND DC CIRCUIT BREAKER (BR) [1] ...................... 123 1) FIGURE 84: PROCEDURE TO COMPARE ANALYTICAL AND SIMULATIVE APPROACH FOR DCCB SIZING ................................ 124 FIGURE 85 DIFFERENCE IN DCCB COST DUE TO DIFFERENCE IN 𝐸𝐸𝐸𝐸𝐸𝐸 .............................................................................. 125 FIGURE 86: EQUIVALENT CIRCUIT OF A 2L-VSC AT A LINE-TO-LINE SHORT CIRCUIT ON THE DC SIDE ....................................... 125 FIGURE 87: RELIABILITY INDICATORS (MAIN/ BACKUP SEQUENCE PERFORMANCE AND FAILURE). FOR ζDCCBS =0,0003 ................ 128 FIGURE 88: NODAL EFFICIENCY KPIS FOR FAULT LOCATION F1, F3, F5 AND NON-SELECTIVE CONVERTER BREAKER PROTECTION

STRATEGY NS-CB. ......................................................................................................................................... 131 FIGURE 89: NODAL EFFICIENCY KPIS FOR FOR FAULT LOCATION F1, F3, F5 AND FULLY SELECTIVE PROTECTION STRATEGY FS-

FDCCB. ....................................................................................................................................................... 133 FIGURE 90: NODAL ENERGY IMBALANCE DISTRIBUTIONS, NS-FDCCB STRATEGY: .............................................................. 134 FIGURE 91: NODAL EFFICIENCY KPIS FOR FAULT LOCATION F1, F3, F5 AND FULLY SELECTIVE PROTECTION STRATEGY FS-SDCCB.

................................................................................................................................................................... 136 FIGURE 92: NODAL ENERGY IMBALANCE DISTRIBUTIONS, NS-SDCCB STRATEGY: .............................................................. 137 FIGURE 93: EFFICIENCY KPIS COMPARISON BETWEEN NS-CB, FS-FDCCB AND FS-SDCCB PROTECTION STRATEGIES. .......... 138 FIGURE 94: EFFICIENCY KPIS COMPARISON BETWEEN NS-CB, FS-FDCCB AND FS-SDCCB PROTECTION STRATEGIES. .......... 139

Page 9: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

viii

V. LIST OF TABLES

TABLE 1: PROPOSED CLASSIFICATION FOR “SPEED” OF THE PROTECTION STRATEGY INDICATOR .............................................. 34 TABLE 2: PROPOSED CLASSIFICATION FOR MAIN SEQUENCE ROBUSTNESS INDICATOR ............................................................ 35 TABLE 3: PROPOSED CLASSIFICATION FOR BACK-UP SEQUENCES ROBUSTNESS INDICATOR ..................................................... 35 TABLE 4: PROPOSED CLASSIFICATION FOR AC/DC STABILITY ISSUE RISK INDICATOR ............................................................. 35 TABLE 5: FREQUENCY QUALITY DEFINING PARAMETERS ................................................................................................... 37 TABLE 6: PROPOSED CLASSIFICATION FOR FREQUENCY RESPONSE INDICATOR ..................................................................... 38 TABLE 7: PROPOSED MONETARIZATION OF INDICATORS ................................................................................................... 46 TABLE 8: SUMMARY OF CONVERTER BREAKER STRATEGY DCCBS COST DESIGN ................................................................. 49 TABLE 9: SUMMARY OF CONVERTER WITH FAULT BLOCKING CAPABILITY STRATEGY’S DCCBS DESIGN ....................................... 50 TABLE 10: SUMMARY OF HYBRID DCCB SPECIFICATIONS ................................................................................................. 51 TABLE 11: SUMMARY OF MECHANICAL DCCB SPECIFICATIONS .......................................................................................... 52 TABLE 12: AC CB FAILURE RATE AND UNAVAILABILITY PARAMETERS FROM LITERATURE REVIEW .............................................. 60 TABLE 13: SELECTED DCCB UNAVAILABILITY PARAMETERS ............................................................................................. 61 TABLE 14: DCCB FAILURE RATE AND UNAVAILABILITY PARAMETERS FROM LITERATURE REVIEW .............................................. 61 TABLE 15: DCCBS SPECIFICATION FOR COST CALCULATION (USED IN SECTION 5). ................................................................ 62 TABLE 16: EFFICIENCY KEY PERFORMANCE INDICATORS FOR FULLY SELECTIVE FS-FDCCD/FS-SDCCB AND NON-SELECTIVE NS-

FB/NS-CB SUPPORTED BY EMT SIMULATIONS ....................................................................................................... 66 TABLE 17: INPUTS USED IN MONTE-CARLO SIMULATION ................................................................................................... 66 TABLE 18: STRATEGIES INDICATORS FOR ζDCCBS =0,0003 ................................................................................................ 67 TABLE 19: BENCHMARKING INDICATORS FOR ζDCCBS =0,0003 ........................................................................................... 67 TABLE 20: POWER RESTORATION TIME OF DIFFERENT PROTECTION STRATEGIES .................................................................. 69 TABLE 21: FREQUENCY STABILITY INDICATORS AND RESERVE REQUIREMENT ....................................................................... 70 TABLE 22: FREQUENCY RESPONSE PARAMETERS ........................................................................................................... 74 TABLE 23: POWER FLOW VIOLATIONS ........................................................................................................................... 76 TABLE 24: FREQUENCY VIOLATIONS ............................................................................................................................. 76 TABLE 25: ANNUAL REDISPATCH COST ......................................................................................................................... 77 TABLE 26: STUDIES CARRIED OUT FOR EACH GRID STRUCTURE. ........................................................................................ 79 TABLE 27: COMPONENTS’ RELIABILITY DATA USED FOR WP12 CASE STUDIES ...................................................................... 84 TABLE 28: DCCBS SPECIFICATION, USED FOR “BACKBONE”, “DOGGER BANK” RING, “WIDER” RING AND “BACKBONE” WITHOUT

INTERCONNECTOR GRID STRUCTURES: TYPE 1 TO TYPE 4 ......................................................................................... 84 TABLE 29: DCCBS SPECIFICATION, USED FOR “BACKBONE”, “DOGGER BANK” RING, “WIDER” RING AND “BACKBONE” WITHOUT

INTERCONNECTOR GRID STRUCTURES: TYPE 5 TO TYPE 8 ......................................................................................... 85 TABLE 30: DCCBS SPECIFICATION, USED FOR “BACKBONE”, “DOGGER BANK” RING, “WIDER” RING AND “BACKBONE” WITHOUT

INTERCONNECTOR GRID STRUCTURES: TYPE 9 TO TYPE 13 ....................................................................................... 85 TABLE 31: TYPE AND NUMBER OF DCCBS USED FOR FULLY SELECTIVE STRATEGY WITH MECHANICAL DCCBS. .......................... 88 TABLE 32: TYPE AND NUMBER OF DCCBS USED FOR FULLY SELECTIVE STRATEGY WITH HYBRID DCCBS. ................................. 88 TABLE 33: TYPE AND NUMBER OF DCCBS USED FOR CONVERTER BREAKER AND FULL BRIDGE STRATEGIES. .............................. 88 TABLE 34: SUMMARY OF EFFICIENCY KPIS FOR NON-SELECTIVE-CONVERTER BREAKER PROTECTION STRATEGIES (F3) ............... 98 TABLE 35: SUMMARY OF EFFICIENCY KPIS FOR FULL-SELECTIVE FS-FDCCB PROTECTION STRATEGIES (F3) .......................... 101 TABLE 36: SUMMARY OF EFFICIENCY KPIS FOR NON-SELECTIVE-CONVERTER BREAKER PROTECTION STRATEGIES (F1, F3 AND F5)

................................................................................................................................................................... 103 TABLE 37: SUMMARY OF EFFICIENCY KPIS FOR NON-SELECTIVE-CONVERTER BREAKER NS-CB (AGGREGATED FAULT LOCATION F1,

F3, F5) AND GRID SPLITTING WITH NS-CB (AGGREGATED FAULT LOCATIONS F1, F3, F11) PROTECTION STRATEGIES ......... 106 TABLE 38 : COMPARISON OF SEVERAL PROTECTION STRATEGIES FROM CAPEX AND TIME TO RESTORE ACTIVE POWER KPI ....... 108 TABLE 39: KPIS OF FREQUENCY STABILITY AND RESERVE REQUIREMENT .......................................................................... 110 TABLE 40 : IMPACT OF PROTECTION STRATEGY ON AC SYSTEM STABILITY, INFLUENCE OF SYSTEM INERTIA ............................. 115 TABLE 41: HVDC SUBMARINE CABLE RESISTANCE DATA WITH WP4.3 DC GRID BENCHMARK ................................................ 119 TABLE 42: COMPONENTS’ RELIABILITY DATA USED FOR WP4.3 AND WP12 DC GRID CASE STUDIES ....................................... 119

Page 10: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

ix

TABLE 43: AC CB FAILURE RATE AND UNAVAILABILITY PARAMETERS FROM LITERATURE REVIEW ............................................ 119 TABLE 44: INERTIA CALCULATION BASED ON GENERATION TYPE ....................................................................................... 120 TABLE 45: TYPICAL INERTIA VALUES FOR DIFFERENT TYPES OF POWER PLANTS .................................................................. 121 TABLE 46: SIMPLIFIED PROTECTION MATRIX FOR PRIMARY AND BACKUP PROTECTION [1] ...................................................... 122 FIGURE 82: TABLE 46 NON-SELECTIVE PROTECTION STRATEGY WITH FULL BRIDGE MMCS [1]............................................... 122 TABLE 48 SIMPLIFIED PROTECTION MATRIX FOR PRIMARY AND BACKUP PROTECTION [1] ....................................................... 122 TABLE 49: SIMPLIFIED PROTECTION MATRIX FOR PRIMARY AND BACKUP SEQUENCES ........................................................... 123 TABLE 50: COMPARISON BETWEEN ANALYTICAL AND SIMULATIVE APPROACH (THE AVAILABLE MARKET RATINGS ARE MARKED IN BLUE

[25] [26]) ....................................................................................................................................................... 124 TABLE 51: STRATEGIES STOCHASTIC KEY PERFORMANCE INDICATORS FOR ζDCCBS =0,00015 ................................................ 127 TABLE 52: STRATEGIES STOCHASTIC KEY PERFORMANCE INDICATORS FOR ζDCCBS =0,0009 ................................................. 127 TABLE 53: BENCHMARKING INDICATORS FOR ζDCCBS =0,00015 ....................................................................................... 127 TABLE 54: BENCHMARKING INDICATORS FOR ζDCCBS =0,0009 ......................................................................................... 127 TABLE 55: CONNECTION (CABLES) DETAILS FOR “BACKBONE” GRID STRUCTURE (BIPOLAR). .................................................. 129 TABLE 56: ONSHORE AND OFFSHORE NODES CAPACITIES (BIPOLAR). ............................................................................... 130

Page 11: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

10

Content

I. Document info sheet ...................................................................................................................................................... i A.1 Distribution list .......................................................................................................................................................... i A.2 Approvals .................................................................................................................................................................. i

II. List of Contributors ....................................................................................................................................................... ii

III. List of abbreviations .................................................................................................................................................... iii

IV. List of figures ................................................................................................................................................................. v

V. List of Tables .............................................................................................................................................................. viii

1 Executive summary ..................................................................................................................................................... 13

2 Introduction .................................................................................................................................................................. 16

3 Methodology and models for CBA assessment ....................................................................................................... 19 3.1 General methodology and key performance indicators (KPIs) assessment .......................................................... 19 3.2 Economic key performance indicators ................................................................................................................... 21 3.2.1 Investment cost indicators ......................................................................................................................... 23 3.2.2 Losses indicators ....................................................................................................................................... 24 3.2.3 Expected Energy Not Transmitted indicators (unavailability concern) ....................................................... 25 3.2.4 Maintenance indicators .............................................................................................................................. 28 3.3 Operational risk indicators ..................................................................................................................................... 29 3.3.1 Efficiency indicators (supported by EMT simulations) ....................................................................................... 30 3.3.2 Stochastic efficiency and protection scheme reliability indicators ..................................................................... 31 3.4 AC system impact indicators ................................................................................................................................. 35 3.4.1 Frequency stability ..................................................................................................................................... 36 3.4.2 Reserve requirement ................................................................................................................................. 38 3.4.3 Frequency reserves ................................................................................................................................... 40 3.4.4 Redispatch cost ......................................................................................................................................... 41 3.5 Extensibility ............................................................................................................................................................ 44 3.6 Computation of the components reliability ............................................................................................................. 45 3.7 Monetarization: System level aggregated indicators ............................................................................................. 45 3.8 DCCB specification ................................................................................................................................................ 47 3.8.1 Non selective protection DCCBs cost model specification to support WP12 MOGs development ................... 47 3.8.1.1 Protection design methodology for converter breaker strategy ..................................................................... 47 3.8.1.2 Protection design methodology for converter with fault blocking capability strategy ..................................... 50 3.8.2 Fully selective protection DCCB cost................................................................................................................. 50 3.8.2.1 Protection design methodology for fully selective strategy with hybrid DCCB ............................................... 51

Page 12: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

11

3.8.2.2 Protection design methodology for fully selective strategy with mechanical DCCB ...................................... 51

4 Developed tools ........................................................................................................................................................... 53

5 Application to WP 4.2 benchmark system ................................................................................................................ 59 5.1 Objective ................................................................................................................................................................ 59 5.2 Data/assumptions .................................................................................................................................................. 59 5.3 WP 4.3 benchmark ................................................................................................................................................ 61 5.4 Economic key performance indicators ................................................................................................................... 61 5.5 Operational risk indicators ..................................................................................................................................... 65 5.5.1 Efficiency key performance indicators (supported by EMT simulation) ............................................................. 65 5.5.2 Stochastic efficiency and protection scheme reliability indicators ..................................................................... 66 5.6 AC system impact indicators ................................................................................................................................. 67 5.6.1 Frequency stability ............................................................................................................................................. 67 5.6.2 Redispatch cost ................................................................................................................................................. 73 5.6.2.1 Constraint violation ........................................................................................................................................ 75 5.6.3 Key performance indicators monetarization ...................................................................................................... 77

6 Application to WP12 grids development plan .......................................................................................................... 79 6.1 WP 12 benchmark ................................................................................................................................................. 79 6.2 Economic key performance indicators ................................................................................................................... 84 6.2.1 DCCB specification for converter breaker, fully selective (with hybrid and mechanical DCCB), and full bridge protection strategies .......................................................................................................................................................... 84 6.2.2 CAPEX indicator for WP12 grid structure .......................................................................................................... 85 6.2.3 OPEX - maintenance indicator for wp12 grid structures .................................................................................... 91 6.2.4 OPEX - losses indicator for wp 12 grid structures ............................................................................................. 91 6.2.5 OPEX - EENT indicator for WP12 grid structures.............................................................................................. 92 6.3 Efficiency key performance indicators through EMT simulation ............................................................................ 94 6.3.1 Efficiency key performance indicators for non selective converter breaker protection strategy ........................ 95 6.3.2 Efficiency key performance indicators for a fully selective protection strategy using fast DCCBs (FS-FDCCB) 98 6.3.3 Comparison between non selective NS-CB, fully selective FS-FDCCB and fully selective FS-SDCCB protection strategies ........................................................................................................................................................ 101 6.3.4 Efficiency key performance indicators for converter breaker and grid splitting................................................ 103 6.4 AC system impact key performance indicators ................................................................................................... 108

7 Conclusion ................................................................................................................................................................. 112

8 Recommendations for WP 12 ................................................................................................................................... 117

Appendices ........................................................................................................................................................................ 119 A.1 Data and parameters for case studies ................................................................................................................. 119

Page 13: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

12

A.2 Frequency reserve basics ................................................................................................................................... 119 A.3 Inertia values for different generation type .......................................................................................................... 120 A.4 Operation of protection strategies ....................................................................................................................... 121 A.5 Comparison between analytical and simulation approach for breaker sizing ...................................................... 124 A.6 DCCBs short circuit current estimation ................................................................................................................ 125 A.7 Stochastic efficiency KPIs and protection scheme reliability ............................................................................... 126 A.8 WP 12 benchmark (line and node details) ........................................................................................................... 129 A.9 WP 12 efficiency KPIs for full non-selective strategy NS-CB .............................................................................. 130 A.10 WP 12 efficiency KPIs for fully selective strategy FS-FDCCB ......................................................................... 132 A.11 WP 12 efficiency KPIs for fully selective strategy FS-SDCCB......................................................................... 135 A.12 WP12 Comparison (efficiency KPIs) between fully selective FS-FDCCB, FS-SDCCB and non-selective NS-CB strategies 138

VI. References ................................................................................................................................................................. 140

Page 14: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

13

1 EXECUTIVE SUMMARY

This deliverable presents the contributions of Work Package 4.5 of PROMOTioN project’s Work Package 4. It proposes a Cost-Benefit Analysis (CBA) approach from a DC protection point of view, i.e. to investigate the impact of DC grid protection and different DC protection strategies on the overall CBA. The main objective is to investigate to what level of detail DC grid protection needs to be integrated in CBA studies, in particular for the studies performed within the PROMOTioN project, on the topologies developed in WP12. The reasons to specifically consider DC protection in the CBA are as follows:

1. The anticipated component cost of DC protection equipment, particularly DC Circuit Breaker (DCCB) and converters with fault current blocking capability, is significantly higher than that of AC protection equipment,

2. Different protection strategies are considered, which result in fundamentally different protection system layouts and equipment requirements, resulting in significant differences in DC side protection costs,

3. The impact of DC faults on the overall AC/DC system stability might be more important, depending on the protection strategy used and it needs to be assessed.

The contributions of this report are:

1. A generic methodology for DC protection CBA is defined and implemented in dedicated tools. This methodology relies on the computation of a dedicated set of Key Performance Indicators (KPIs) to support protection strategies assessment and comparison. The proposed generic methodology can be applied to a large range of DC grid case studies. .

2. Different protection strategies are compared, showing deviations in terms of Capital Expenditure (CAPEX), Operational Expenditure (OPEX) and performance.

The proposed KPIs consist in economic KPIs (CAPEX and OPEX), efficiency KPIs (fault interruption time, voltage restoration time, active power restoration time, reactive power restoration time and transient energy imbalance), failure KPIs (primary sequence failure probability and protection strategy failure probability) and AC impact KPIs (frequency stability, fast frequency reserve requirement, redispatch costs). These KPIs have been calculated using several dedicated tools developed in Matlab, Julia or Python environments.

The methodology is applied to two case studies, i.e. a “small” size Multi-Terminal DC (MTDC) grid and a “large” size MTDC grid. Several protection strategies have been considered: (i) a fully selective (FS) fault clearing strategy using fast DCCB, named FS-FDCCB strategy; (ii) a fully selective fault clearing strategy using slow DCCBs, named FS-SDCCB strategy; (iii) a non-selective (NS) fault clearing strategy using DCCB at each converter terminal, named NS-CB strategy; and (iv) a non-selective fault clearing strategy using converter with fault current blocking capability, in particular using FB-MMC converter, named NS-FB strategy (see [1] and [2] for more information about selective and non-selective fault clearing strategies).

From the knowledge acquired within the PROMOTioN project that can be found within this report and the deliverables [1] and [3], some takeaways can be highlighted. Note that these takeaways stand only for offshore DC grids with cables.

Analysis, supported by EMT simulation, has shown that all considered protection strategies are applicable for the studied DC grids, with the stability of the DC grid being ensured before and after fault clearing and during the restoration process

CAPEX of protection systems which use DC circuit breakers is dominated by DC circuit breaker CAPEX, consisting of the cost of the breaking devices, DC reactors and surge arresters for energy dissipation. Additionally, the costs related with the DCCB weight and volume, especially if the DCCBs are installed on an offshore platform, are not negligible (an increase of more than 100% of the DCCB CAPEX can be observed if large DC reactor and high energy to be dissipated are required). The CAPEX for NS-FB strategy needs to consider the additional costs incurred by such converters (i.e. difference in cost between half-bridge and full bridge converters). From CAPEX

Page 15: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

14

point of view, there is an advantage for NS protection system using DCCB at converter terminals: the NS-CBS strategy.

OPEX indicators are related to losses, maintenance, Expected Energy Not transmitted (EENT) and need for fast frequency reserves (FFR) to stabilize AC grid frequency during the protection process. The protection system contribution to total losses is quite negligible (less than 4% of the total grid system losses) except for the NS-FB strategy where FB converters bring significant additional losses. Protection strategies having long active power restoration time could potentially require more FFR. The NS-CB protection system exhibits a higher EENT contribution (around 10% of the total grid system EENT). Therefore from OPEX consideration, it seems that both NS protection strategies incur higher OPEX compared to FS strategies.

For both case studies, the total cost (CAPEX and OPEX) of protection systems never exceeds 9% of the total grid cost in the two considered case studies. The range of contribution of protection system cost to the total grid cost varies from 5% to 9%. While not negligible, the protection system cost is not seen as prohibitive in respect to the total grid cost. As a consequence, it seems that no protection system is considered inapplicable or impossible due to its excessive cost. However, a cost saving of some percent of the total grid system cost can be significant and would have to be fully considered through the selection of the “right” protection system.

Computation of efficiency, failure and AC impact KPIs has demonstrated that impacts on both DC and AC grids are dependent on protection strategies. As such, computation of dedicated KPIs is fundamental when assessing the performance of protection systems.

There is no single preferred protection strategy which outperforms all other protection strategies and the protection system design needs a full assessment (no short cut can be applied). However, some elements for supporting the selection of a protection strategy can be highlighted.

The impact on both frequency and transient stabilities of surrounding AC systems is a key element to consider when designing the protection system of a DC grid. On one hand, this impact depends on the AC system inertia and operating conditions. On the other hand, it depends on the temporary interruption of the active power flowing through the DC grid (in terms of quantity and duration) and exchanged with the AC systems. For an AC system with high inertia (typically H>2-4 s), it is expected that both NS and FS strategies would not incur major impacts on the AC system stability. The situation could be more critical with an AC system with low inertia (typically H<2-4 s). The temporary interruption of the active power through the DC grid (in terms of quantity and duration) is highly related to the design of the protection system and more particularly in relation with the time to restore the active power. Converter control performance plays an important role and coordination between converter controls is necessary in order to ensure a fast restoration time. Indeed, in fully selective strategies the presence of line inductors induces power oscillations that need to be damped by dedicated converter controls. In non-selective strategies, the major difficulty is related to the coordination of the power ramp-up among all converters after a full DC grid temporary stop.

Non-selective protection strategies:

Non-selective strategies lead to a temporary full interruption of the active power flowing through the DC grid during the protection process (the entire DC grid is de-energized). As such, they could impact stability (both frequency and transient) of low inertia AC systems, more particularly when high power is exchanged through the DC grid prior to fault inception.

Non-selective protection strategies (NS-CB or NS-FB) using high speed switches (HSS) to isolate the faulty line show quite high time to restore active power (> 400 ms) and can be considered as applicable to only a limited number of AC/DC systems due to potential AC system stability issues and as non-applicable for large size DC grids.

Replacing HSS with DCCBs at each line end leads to a real reduction of the time to restore active power and also allows efficient protection back-up sequences. As such, these strategies seem to be applicable to small size to medium size DC grids, mainly depending on AC system stability requirements. Their application to a large size DC grid is even possible but a special attention must be paid to potential AC system stability issues.

NS-CB strategy exhibits a gain in protection system CAPEX. As such, it can provide a recommended cost-effective solution when the impact on AC system stability is acceptable.

Page 16: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

15

Due to quite high CAPEX compared to the NS-CB strategy, it seems that the NS-FB strategy would not be the preferable one in many cases.

Fully-selective protection strategies:

Fully-selective strategies lead to no or temporary partial interruption of the active power flowing through the DC grid during the protection process. As such, both studied fully-selective protection strategies have a lower AC grid impact and can be applied to any DC grid configuration (small size to large size DC grids). However, during backup protection sequence, higher times to restore active power are observed and FS strategies could also entail transient and frequency instabilities in a low inertia AC system.

The fully-selective protection strategy seems to be the preferable solution for large size DC grid application.

Considering the studied large DC grid case, there is no direct evidence allowing the selection between using fast DCCBs or slow DCCBs.

Using fast DCCBs leads to the best performance but at the highest cost.

Using slow DCCBs leads to a cost saving (about 25%) for the protection system together with a slightly degraded performance. Thus, this solution might be considered as a good trade-off between cost and performance. However, using slow DCCBs results in implementing quite large DC reactors (around 300 mH at each DCCB location compared to 140-200 mH with fast DCCBs), the impact of which on control performance shall be fully analysed for a large range of fault location scenario.

Grid splitting based protection strategies:

For large DC grids, the grid splitting protection strategy could be a solution to reduce grid impact when implementing NS protection strategies. At a quite low additional cost (i.e. cost of some additional fast DC circuit breakers used as a firewall between different zones), grid impacts can be significantly reduced.

Grid splitting could also enable to implement different protection strategies in different zones of the grid and therefore facilitate the HVDC grid extensibility.

While the method is interesting to consider and showed considerable merit, the application is very case dependent. The applicability and efficiency of the grid splitting option still need to be fully demonstrated and was out of the scope of the present work.

Page 17: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

16

2 INTRODUCTION

When performing investments, investments options are evaluated using a cost-benefit analysis (CBA). According to Investopedia1:

• A cost-benefit analysis (CBA) is the process used to measure the benefits of a decision or taking action minus the costs associated with taking that action.

• A CBA involves measurable financial metrics such as revenue earned or costs saved as a result of the decision to pursue a project.

• A CBA can also include intangible benefits and costs or effects from a decision such as employee morale and customer satisfaction.

In a power systems, the costs are related to on the one hand the capital expenditures (CAPEX), or the costs associated with the development, investment and installation of equipment, and on the other hand with the operational expenditures (OPEX), or namely the costs associated with running the system (e.g. losses and maintenance). Other costs could also be considered such as development expenditure (DVEX), replacement expenditure (REPEX) and decommissioning expenditure (DECEX). The latter is more particularly considered within the WP 7. The benefits of grids are enhanced market efficiency (increased social welfare), increased reliability, increased RES penetration, etc. In practical CBA approaches [4] [5], a number of assumptions are made to be able to perform the assessment, typically:

• Scenarios are used to evaluate different futures • Selected years are evaluated (typically the economic lifetime of a project is considered) • Reduced power system detail (focus on the area close to the foreseen investment )

In the scope of the PROMOTioN project, CBAs are carried out at different levels, within different work packages. WP 7 has proposed a cost-benefit analysis methodology for offshore grids in Deliverable D7.10 [6]. WP12 is in charge of defining the deployment plan for future European offshore grids. To support such offshore grid design, a CBA based on the methodology developed in D7.10 is applied to assess both costs and benefits. Cost are related to the CAPEX and OPEX associated to grid components and layout. The benefits are captured through a classical approach aiming to optimize the generation costs at European scale, resulting from RES integration and trans-border interconnections.

In traditional power system CBA assessments, AC protection is disregarded as it is considered to be of little impact. This is due to the fact that all types of investments will have comparable protection costs due to the standardized protection approach and these costs in all cases have a relatively small (negligible) contribution on the overall investment.

This deliverable investigates the effect of DC protection strategies on the overall CBA. The reasons for considering DC protection in the CBA are the following:

• The anticipated component cost of DC protection equipment, such as Circuit Breakers (DCCBs) and converters with fault blocking capability, is significantly higher than that of AC protection equipment,

• Different protection strategies are considered, which result in fundamentally different protection system layouts, resulting in significant differences in DC side protection costs,

• The impact of DC faults on the overall AC/DC stability system might be more important.

As such, in this deliverable, the first main objective is to investigate how to include DC protection systems in the cost-benefit analysis. Note that it is not the objective to develop a full CBA method for DC grids, but rather to focus on the difference caused by the protection system, through the use of different protection technologies or different protection strategies. As such, the proposed CBA approach is seen as complementary to the CBA implemented

1 Investopedia is an American website based in New York City that focuses on investing and finance education along with reviews, ratings and comparisons of various financial products.

Page 18: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

17

in WP122. As a second objective, the method allows to compare different DC grid protection strategies, as defined in other tasks of this work package. It aims to propose an assessment/comparison framework to capture costs related to protection systems as well as to be able to assess the performance related to those protection systems. Doing so, it is possible to “give a cost to performance”, allowing to discriminate between protection schemes from a cost and performance consideration, within a given AC/DC grid setup, and finally to help selecting the most appropriate protection strategy for a given AC/DC system, that is to say, the right protection at the right cost. As such, the proposed approach can be seen as a DC protection design support methodology.

The proposed methodology is structured around economic and technical key performance indicators. Economic KPIs are related to the CAPEX and OPEX associated to the protection systems. The methodology is also able to compute CAPEX and OPEX of the overall grid, using the same economic KPIs. As such, it is possible to estimate the relative contribution of the protection system in the total CAPEX and OPEX. OPEX is captured through the computation of losses, maintenance costs and impact of unavailability of the grid, due to component failure, on the transmitted energy (Expected Energy Not Transmitted, EENT). CAPEX and OPEX can be aggregated through a global economic indicator such as Life Cycle Cost (LCC) or Net Present Value (NPV).

Technical KPIs are defined in order to capture impacts incurred on both DC and AC grids during the whole protection process, i.e. DC fault clearing and DC grid restoration. Following technical KPIs are considered:

• Efficiency KPIs as defined within the scope of WP 4.3: Efficiency indicators are the KPIs related to how the protection strategy manages the fault clearing process and the grid restoration. The main goal of the efficiency indicators is to measure the impact of the protection strategy performance on the system. These indicators are also useful to estimate how much the AC grid will be disturbed by the fault clearing in the DC grid. Five efficiency indicators are proposed in this deliverable:

o Fault interruption time o Voltage restoration time o Active power restoration time o Reactive power restoration time o Transient energy imbalance.

Efficiency indicators can be computed by running a set of EMT simulations introducing several fault locations and several power flow scenarios, as proposed in WP4.3. Besides this scenario-based EMT simulation approach, a methodology using an assessment framework relying on a stochastic approach is proposed by considering a distribution of efficiency KPIs. A set of probabilistic indicators is then built and can be used to assess and compare protection strategies. This approach has been widely used within the scope of WP4.2 to provide support for the broad comparison of protection strategies presented in Deliverable D4.2 [1].

• Failure indicators as defined within the scope of WP 4.3: The failure indicators are KPIs that measure the features related to the malfunctioning of a protection strategy. These KPIs are primary sequence failure probability and protection strategy failure probability. A stochastic approach is used for computing these indicators. The methodology relies on both Petri Nets and Monte Carlo models.

AC grid impact indicators: The impact is analysed in dynamic as well as in steady state conditions. In the dynamic condition, the DC grid protection can influence the stability of the connected AC grid. This stability issue is addressed through assessing the frequency stability. The methodology also computes the costs incurred by the mobilization of fast frequency reserves (FFR) as response to frequency constraint violation in direct relation to the protection strategy timings and amount of instantaneous power deviation induced by a strategy

2 In WP12, we use a classical methodology developed from the ENSTO-E approach in WP7. It is worth noting that the Value of the CBA is measured as maximising Social Economic Benefit for all stakeholders. In the real world, a party may wish to protect his part of the grid and will make a CBA for protection of his portion of the grid.

Page 19: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

18

The proposed methodology has been illustrated on two specific case studies consisting of HVDC grids of different sizes:

• The “small” 4 terminal meshed DC grid case study proposed in the framework of WP4.1-3. Four protection strategies are considered in the scope of this application:

o A fully selective fault clearing strategy using fast DCCB (FS-FDCCB) o A fully selective fault clearing strategy using slow DCCB (FS-SDCCB) o A non-selective fault clearing strategy using DCCB at converter terminals (NS-CB) o A non-selective fault clearing strategy using converter with fault blocking capability (NS-FB)

All proposed KPIs are addressed through this case study application. Both primary and back-up protection sequences are considered when computing KPIs.

• A “large” 35 terminal radial backbone grid proposed by WP12: The analysis is performed on the basis of a restricted set of KPIs. The CAPEX KPI is determined for the four protection strategies defined above. Their performance is captured through a set of technical KPIs, some of which were computed by means of EMT simulations. Only primary protection sequences are considered here when computing technical KPIs. Only the following three protection strategies are considered for the technical KPIs assessment3:

o A fully selective fault clearing strategy using fast DCCB o A fully selective fault clearing strategy using slow DCCB o A non-selective fault clearing strategy using DCCB at converter terminals

This report is organized as follows:

Section 3 presents the proposed methodology and models that are used to compute both economic and technical KPIs. Note that in the scope of the WP 4.5, several DC circuit breaker parametric cost models have been developed together with work package 6. Both fast DCCB (associated to so-called hybrid-type technology) and slow DCCB (associated to so-called mechanical type technology) cost models are proposed. More details about those models can be found in [7] and are not be presented in this report.

Section 4 briefly introduces the different tools that have been developed to implement the proposed methodologies.

Section 5 presents an application to the 4 terminals meshed DC grid case study proposed in the framework of WP4.3.

Section 6 is dedicated to a specific application to a 35 terminals radial backbone grid proposed by WP12. Such case study is the opportunity to show that the proposed CBA approach is applicable to large size DC grids. The analysis is performed on the basis of a restricted set of KPIs. Performance is captured through a set of technical KPIs, some of them being computed through EMT simulations.

Section 7 provides conclusions regarding the proposed methodologies and highlights the key takeaways learnt from the case study applications.

Finally, section 8 makes some recommendation to support the CBA analysis of the deployment plan for future European offshore grid performed by WP12.

3 EMT simulations for efficiency KPIs calculation have not been performed for the NS-FB strategy within the WP4.5 framework due to the lack of dynamics model of FB converter with its control in EMTP-RV environment which has been used in WP4.5 to perform EMT simulation (this FB converter EMT dynamics model has only been developed within the scope of WP4.3 by University of Aachen (RWTH) in the PSCAD environment). Developing a dynamics model of FB converter with its control in EMTP-RV environment has been considered out of the scope of the WP4.5 work.

Page 20: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

19

3 Methodology and models for CBA assessment

This section is dedicated to introduce CBA objectives and methodology that are proposed within Task 4.5 for protection systems assessment and comparison. First, the general assessment methodology is presented. Associated indicators are then briefly introduced through three points that are CBA multi-criteria assessment, proposed monetarization of indicators and proposed risk analysis approach.

3.1 GENERAL METHODOLOGY AND KEY PERFORMANCE INDICATORS (KPIS) ASSESSMENT

The proposed assessment/comparison framework is based on the computation of “costs” and “performance” relative to each protection schemes. Doing so, it will be possible to “give a cost to performance”, allowing to discriminate between protection schemes from a cost and performance consideration, within a given AC/DC grid setup.

It is important to emphasize that, as far as the proposed performance assessment is mainly related to system operations, the proposed CBA framework is more an operational based CBA assessment. It differs from a standard CBA framework which has as main objective to evaluate an investment project from an economic point of view, as proposed by ENTSO-E for assessment of Project of Common Interest (PCI) within the scope of the Ten Year Network Development Plan (TYNDP) proposed by European Community Energy Roadmap ECER. In fact, this assessment should be seen as a complement to or sensitivity analysis of those methods.

Figure 1: General framework for CBA assessment and comparison

Figure 1 presents the general framework for CBA assessment and comparison of protection schemes used within the framework of task 4.5 and this deliverable. The CBA core is indicated with a grey background colour. This framework interfaces with other tasks and work packages, as indicated in Figure 1. In particular:

Page 21: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

20

• System operation and components specification from WP 4.2 and WP 4.3: Information about system protection architecture, protection algorithm as well as operating points will be necessary for CBA assessments

• Reference grids and associated benchmarks should give information about the grid architecture and would allow to replace the protection schemes within its environment. It is then possible to assess the protection system relative to the total system from a CBA assessment point of view. Two levels of reference grids are considered here: The first level is in relation to the reference grids used within WP 4.2 and WP 4.3, where quite “simple” grids are sufficient for the technical evaluation of different protection systems. However, for a full CBA assessment and comparison, it seems necessary to extend reference grids to more realistic and complex network topologies which would be defined within the framework of the roadmap for meshed offshore HVDC grids developed in WP 1 and WP 12.

From the Figure 1 CBA framework, it is proposed to asses protection strategies from two points of view:

• Protection strategy performance. At this stage of the study, the protection strategies are assessed by evaluating their performance to protect DC and their potential impact on the operation of the surrounding AC grids. Two main key performance indicator categories are considered:

Operational risks indicators. These KPIs aim to evaluate the protection strategy’s performance from their ability/speed to clear the fault and to restore power flow through the DC grid: probability of success of the main/backup sequences, efficiency indicators such fault interruption time, DC voltage and power restoration times, energy imbalance at AC/DC interfaces during the protection process. Typical outputs for risk analysis are illustrated in Figure 2, where the fault interruption time distribution could be compared with the maximal time that an AC system can withstand power interruption.

AC impact indicators. These indicators aim to evaluate the impact (cost and performance) of the protection strategy on the surrounding AC grids. Four indicators are proposed: frequency stability, frequency reserve, reserve requirement and redispatch costs indicators.

Other performance indices coming from WP 4.2 and WP 4.3 technical assessment could then complete the analysis.

• Economic indicators: the objective of such indicators is to assess the protection strategies by computing the costs associated to CApital EXpenditures (CAPEX) and Operational EXpenditures (OPEX), considering maintenance, losses and Expected Energy Not Transmitted EENT, relative to the protection system and its components. These costs are then compared to the overall “direct” costs relative to the full DC grid and its components (cables, converters, platforms, etc.). Operational costs (OPEX) indicators will be first defined in physical units (e.g. MWh/year) for supporting a multi-criteria analysis. Then, a monetarization of these indicators could be introduced in order to perform fully aggregated cost indicators, such as Life-Cycle-Cost (LCC) or Net Present Value (NPV), see section 3.7.

(a) (b)

Figure 2: Typical outputs for risk analysis assessment

Page 22: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

21

(a): reliability analysis (main/backup sequence success and protection failure), (b): distribution of fault interruption times (for main and backup sequences)

To compute these key performance indicators, some models are required. Figure 3 introduces the relation between the different indicators to be quantified and the model requirements. Within this work, it is essential to start with the most detailed models in order to be able to draw the relevant and unbiased conclusions. It will allow, in a later stage, to propose the appropriate level of detail for CBA system including DC systems and their protection system. More details related to model requirements are presented in [8].

Figure 3: Model requirements for indicators calculation

The last performance indicator which will be addressed within WP 4.5 is the Extensibility Feature of a given protection scheme. The extensibility property is understood here as the ability for a protection scheme to be adapted to an expansion of the DC grid, without major modification of the existing protection scheme and device specifications. Up to now, the best way to qualify and quantify this indicator needs to be further analysed. It is not in the scope of this report to go on such deeper analysis. In a first step, aligned within the WP 4.2 timeline, it could rely on the shallow assessment methodology proposed by WP 4.2. Section 3.5 gives a short summary of the work done in WP 4.2. More details can be found in [1] [3].

3.2 ECONOMIC KEY PERFORMANCE INDICATORS

This section aims to further introduce the different economic indicators (CAPEX and OPEX), and their computation. The main focus lies on the link between system operational state, components specification and requirements in terms of models (cost, losses, reliability). Investment cost (CAPEX) and operational costs (losses, expected energy not transmitted (EENT), maintenance) indicators principle and models are briefly presented.

The objective here is not to give a full detailed presentation of processes and associated equations which are involved within the indicator calculation, but more to give a general perception of the way how these indicators are proposed to be quantified. For a more detailed indicators computation process, the reader can refer to the report [8].

Figure 4 presents the proposed general assessment and comparison methodology. For a given protection scheme embedded within a given reference grid, the methodology is organized as follows:

Page 23: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

22

• A level of performance is assessed through the computation of relevant key performance indicators. This level could be required to be improved in order to reach acceptable performance. If it is not possible to reach a mandatory level of performance, the protection scheme could be declared as not suitable for application to the reference grid (e.g. a temporary loss of X GW during the fault clearing process could be a discarding criteria). This can be seen as an approach where a base case (e.g. without protection system or with a minimal protection level) is considered and where the increase of performance will be characterized by its marginal additional cost.

• A cost will then be associated to each level of performance (i.e. what is the cost of performance?). In all cases, investment cost (i.e. CAPEX, which is considered as the main cost driver) will be quantified. Assessment and comparison will be done from direct or relative considerations. The direct case would consider only the costs due to the protection strategy, whereas a relative one would compare the relative contribution of the protection strategy to the overall system costs, i.e. HVDC grid and protection system. The latter will then give an importance level of comparison.

• Additional indicators (mainly associated with OPEX: Losses, maintenance, EENT) will be computed and compared to the total system’s indicators. Associated with CAPEX KPI, these indicators will be also used for supporting multi-criteria or full monetized assessment and comparison.

Figure 4: Proposed general assessment and comparison methodology

Several indicators are proposed related to both investment and operation costs:

• Investment costs: Total CAPEX + Additional CAPEX due to protection system

• Losses: Total system losses + Additional losses from protection components

Page 24: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

23

• Expected Energy Not Transmitted (EENT) due to unavailability of components: EENT from total system components + EENT due to protection system components

• Maintenance operation costs: Maintenance for total system + Maintenance for protection system components

Additionally, some operation costs incurred on the surrounding AC grids are considered. These costs are associated with Fast Frequency Reserve requirement and generation redispatch. There are fully detailed in section 3.4.

From a quantitative point of view, some of these indicators can be given in “original” units (e.g. MW, MWh). On another side, they can be monetized. Monetarization would offer the opportunity to characterize the protection systems through aggregated cost indicators such as Life Cycle Cost (LCC), see section 3.7. Figure 5 illustrates these two alternative quantification approaches, which are both developed within the WP 4.5 development framework.

It is important to note that the focus of the analysis performed for this task is on criteria (indicators) mainly dependent on MTDC parameters (e.g. topology, power transfer capacity, etc.). Considered indicators mainly dependent on external parameters (AC side, market) (e.g. Social Economic Welfare (SEW), Value Of Lost Load (VOLL)) would require building representative reference case studies on which such CBA indicators could be computed. While that extension is conceptually small, it is considered outside the scope of the here presented work.

Figure 5: Cost indicators quantification proposed approach

3.2.1 INVESTMENT COST INDICATORS

PRINCIPLE

Investment costs4 are related to the calculation of the Capital Expenditures (CAPEX). These costs are mainly related to both manufacturing and installation/commissioning costs of the different components of the system. Figure 7 presents the general methodology associated to CAPEX calculation, highlighting in green the associated model involvement. The main idea is to be able to quantify the relative contribution to CAPEX of each component of the DC grid system. It is then proposed to compute two main CAPEX indicators, i.e. CAPEX of the total system (including all components such as cables, converters, transformers as well as protection system devices) and

4 Investment costs here do not consider the planning aspect (i.e. “moment in time” approach). This is out of the scope of deliverable 4.7 where the spread of the protection costs over time is not considered. However, the proposed approach remains applicable if the grid development planning is given. Indeed, this aspect have an impact on the monetarization step (discounted costs) which is described in section 3.7.

Page 25: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

24

CAPEX of the protection system including only costs related to protection devices. Doing so, it will be possible to have a clear idea of the relative contribution of each studied protection scheme CAPEX within the total CAPEX. Same approach will be applied for all the cost indicators (i.e. losses, EENT, maintenance).

As shown in Figure 6, investments costs are related to cost models for each component. These costs will mainly depend on components specifications defined from WP 4.2 or WP 4.3 outputs. Because such specifications could be different for each protection scheme, it is essential to be able to calculate components’ costs through parametric models. This will apply for each type of component, not only for the protection system devices. The costs of grid component costs (including converters, transformers, DC cables, AC switchgear) are gathered from the “Cost Collection Task [9]”; a database set up by different Work Packages participants within the PROMOTioN project in the framework of WP12. For protection system based on non-selective fault clearing strategy using converter with fault blocking capability, it is proposed to consider the difference between the cost of a converter with fault blocking capability (e.g. full bridge MMC) and the cost of a converter without fault blocking capability (e.g. half bridge MMC) as a part of the CAPEX of the protection system.

It should also be noted that there may be a relationship between the cost of the grid components and the type of protection system chosen (e.g. cables will have to be rated to deal with the DCCB TIV, some fault current withstand ratings can potentially be reduced for fast clearing schemes, etc.). Even if this can lead to cost savings in real projects, it is hard to be quantified as it would be linked with a full optimization of the DC grid with its protection system and is out of the scope of the present work.

Figure 6: CAPEX calculation methodology for DC grid protection equipment developed within PROMOTioN WP4

MODEL

All the component cost models/data used in the current deliverable are coming from dedicated “Cost data collection task” within PROMOTioN project, except for DC circuit breakers (DCCBs). Regarding these last, their costs model are coming from a specific dedicated task and are not presented here in the current deliverable. More details about DCCBs cost models can be found in [7].

3.2.2 LOSSES INDICATORS

PRINCIPLE

Figure 7 presents the general methodology for DC grid system active energy losses calculation. Losses will be computed from the system operation state, i.e. power flows 𝑃𝑃𝑡𝑡, through dedicated parametric loss models, 𝑔𝑔𝑖𝑖(P𝒕𝒕 , Oi , Sti ) based on operational and/or structural parameters of grid components (Oi , Sti ) (e.g. the impedance of cables). Formally, the losses related to component i could be expressed by equation 3-1, where the time T is the time window considered for the analysis (e.g. some years).

Page 26: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

25

𝐿𝐿𝐿𝐿𝐸𝐸𝐸𝐸𝑖𝑖 = � 𝑔𝑔𝑖𝑖(P𝑡𝑡 , Oi , Sti )𝑇𝑇

0𝑑𝑑𝑑𝑑 3-1

From Figure 7, the calculated losses indicators either could be used for multi-criteria analysis, where losses will be expressed in the original unit, i.e. MWh, or could be monetarized in order to be integrated within a fully aggregated cost integrator (e.g. LCC).

Losses calculation is highly dependent on the power flow scenario. Those scenario will have to be defined for each considered DC grid case study. It is the case in this report for the two studied DC grid benchmarks analysed in sections 5 and 6.

Figure 7: Losses calculation methodology for DC grid protection equipment developed within PROMOTioN WP4

MODEL

The component loss models used in WP4.5 (transformers, converters) are based on data generated by a SuperGrid Institute internal tool. However, as no physical cable parameters are available within WP4.5, the cable losses are approximated by the 𝑅𝑅𝑅𝑅2 formula, where R is the resistance given for a standard 70° C operation temperature.

3.2.3 EXPECTED ENERGY NOT TRANSMITTED INDICATORS (UNAVAILABILITY CONCERN)

PRINCIPLE

The general methodology for the Expected Energy Not Transmitted (EENT) indicators calculation is introduced in Figure 8. EENT indicators are related to the unavailability of components within the DC grid. As part of this methodology, EENT will be calculated using a Monte-Carlo (MC) probabilistic approach whose main steps are described in the algorithm given in Figure 9 and Figure 10. From Figure 8, it is proposed that the calculated EENT indicators are either used for multi-criteria analysis, where EENT will be expressed in original unit, i.e. MWh, or are monetized in order to be integrated within a fully aggregated cost integrator (e.g. life cycle cost LCC).

The proposed methodology relies on reliability models for each component. For each of these components, reliability parameters associated with unavailability assessment requirements are as follows:

• DFR: “Detected Failure” Rate of components (e.g. Mean Time To Failure, MTTF)

• MTTR: Mean Time To Repair, after a failure occurred and was detected

These values will have to be defined for each considered DC grid case study and are not detailed in this section dedicated to the proposed methodology. For the two studied DC grid benchmarks of sections 5 and 6, they are given in appendix A.1 for grid components (cables, converters) and in section 5.2 for protection components.

Page 27: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

26

Figure 8: EENT calculation methodology

Figure 9: EENT calculation main steps

MODEL

In order to compute the undelivered energy EENT, a reliability model is required. To achieve that, a Monte-Carlo based approach is proposed. By denoting:

• 𝑃𝑃𝑃𝑃𝑖𝑖𝑜𝑜𝑜𝑜𝑜𝑜: the set of protection devices out of service (including devices in maintenance operation) of the 𝑖𝑖𝑡𝑡ℎ

protection scheme,

• 𝑃𝑃𝑃𝑃𝑖𝑖𝑜𝑜𝑜𝑜: the set of healthy protection devices of the 𝑖𝑖𝑡𝑡ℎ protection scheme,

• 𝐶𝐶𝑔𝑔𝑔𝑔𝑖𝑖𝑔𝑔𝑜𝑜𝑜𝑜𝑜𝑜 : the set of grid components (except protection devices) out of service (this could include transformers,

converters and cables),

• 𝐶𝐶𝑔𝑔𝑔𝑔𝑖𝑖𝑔𝑔𝑜𝑜𝑜𝑜 : the set of healthy grid components (except protection devices).

The skeleton of the Monte-Carlo model are summarized as shown in Figure 10. The sequences i) to viii) of the flowchart in Figure 10 are roughly summarized by computing a power flow according to the availability (or not) at time t of the grid components and protection devices involved in the 𝑖𝑖𝑡𝑡ℎ protection scheme.

Page 28: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

27

Figure 10 General framework of energy not transmitted Monte-Carlo based model

However, it can be noted that the power flow obtained in the operation vi) of the flowchart would be influenced by a given protection scheme i in two cases:

1. If at least one protection device d is unavailable (i.e: d ∊ 𝑃𝑃𝑃𝑃𝑖𝑖𝑜𝑜𝑜𝑜𝑜𝑜) and its availability affects other grid

components (a cable for example). This could happen, for example, if the loss of a given protection device leads to the disconnection of a given cable for operational purpose.

2. If a given protection device d is unavailable (i.e: d ∊ 𝑃𝑃𝑃𝑃𝑖𝑖𝑜𝑜𝑜𝑜𝑜𝑜) when it receives an order to trip (i.e. so-called

hidden failure).

Within deliverable D4.7, only the first case (detectable failure) will be considered. This can be supported by assuming that a continuous diagnostic of the protection strategies’ components is possible. In this case, if one (or more) protection components located at line ends (respectively downstream of a converter) is not available, this line becomes unavailable (respectively, the converter becomes unavailable). Furthermore, only the components involved in the primary sequence lead to unavailability of a line (respectively a converter). Figure 11 shows an example of such assumptions. This figure shows that in case of the failure of line breaker (Br1 failure, figure 11(a)), the entire grid section up to wind farm 1 is disconnected from the rest of the grid. Similarly, if a converter breaker fails (Bc1 failure, figure 11(b)), onshore AC grid 1 is disconnected from the associated bus.

Page 29: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

28

Figure 11 Example of unavailability of line and converter due to protection strategy component: a) Br1 failure

leads to unavailability of line L13, b) Bc1 failure leads to unavailability of Converter 1

Finally, the total undelivered energy EENT𝐠𝐠𝐠𝐠𝐠𝐠𝐠𝐠,𝐏𝐏𝐏𝐏𝐠𝐠 involving the 𝑖𝑖𝑡𝑡ℎ protection scheme 𝐏𝐏𝐒𝐒𝐠𝐠 includes two types of undelivered energy (see equation 3-2):

• Undelivered energy due to the 𝑖𝑖𝑡𝑡ℎ protection scheme 𝐏𝐏𝐒𝐒𝐠𝐠 : EENT𝐏𝐏𝐒𝐒𝐠𝐠 , • Undelivered energy due to the electrical grid components: EENT𝐠𝐠𝐠𝐠𝐠𝐠𝐠𝐠 .

EENTgrid,Psi = EENTPSi + EENTgrid 3-2

3.2.4 MAINTENANCE INDICATORS

PRINCIPLE

Proposed methodology for maintenance cost calculation is presented in Figure 12. Main inputs are the system components state (faulty or not) and the maintenance policy. Maintenance costs are mainly divided in two categories:

• Corrective maintenance costs: made after a failure occurs and was detected, mainly linked with system components state (failure probability)

• Preventive maintenance costs: made when an intervention is planned in advance, linked with the maintenance policy

To compute corrective maintenance costs, a grid reliability analysis is required. That is achieved by simulating the stochastic aspect of the grid component’s failures, as discussed in EENT section.

Regarding the preventive maintenance, the frequency of interventions is obtained through a pre-defined maintenance plan (or budget).

Then, the maintenance costs are obtained by multiplying the intervention frequency by the unit cost of the intervention, based on cost parametric models of components (if a component spare is required) and cost of maintenance operation model which could depend on system environment (e.g. onshore, offshore)

Reliability parameters associated with maintenance assessment requirements are as follows:

• DFR: “Detected Failure“ Rate of components (e.g. Mean Time To Fail, MTTF)

Page 30: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

29

• MTTR: Mean Time To Repair, after a failure occurred and was detected

• PMR: Preventive Maintenance Rate (preventive maintenance operation duration / time period between two preventive maintenance operations)

Figure 12: Maintenance cost calculation methodology

MODEL

As for grid components costs and as agreed with other Work Packages participants (WP1, WP2, WP4 and WP12), the used maintenance model/data within WP4.5 studies are coming from the “Data collection task”.

3.3 OPERATIONAL RISK INDICATORS

In this section, some operational risk indicators, are proposed. They are called technical key performance indicators, and are related to the impact on AC/DC system. They are divided into two categories: efficiency/reliability indicators and AC system impact indicators (see Figure 13). These key performance indicators intend to strike a balance between cost and performance of a given protection strategy.

Considering efficiency key performance indicators, two complementary methodologies can be used. The first one consists of performing electromagnetic transient simulations (EMT) for different fault locations in the DC grid. This method is adopted in [3]. The second one consists of using a simplified model (from minimal EMT studies if possible) to derive distributions of the different efficiency KPIs by performing stochastic studies on the DC grid.

Figure 13: Proposed indicators for operational risk analysis

Page 31: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

30

3.3.1 EFFICIENCY INDICATORS (SUPPORTED BY EMT SIMULATIONS)

Efficiency indicators as defined in [3] includes five metrics (see Figure 13):

• Fault interruption time: this metric expresses the time needed to isolate the faulty component. In other words, this corresponds to the needed time span to clear the fault current within the faulty component.

• Voltage restoration time: This metric corresponds to the time span needed to restore the DC grid voltage (generally, within ±15% of the steady state values).

• Active power restoration time: This is defined as the needed span time to restore the active power (generally, within ±10% of the steady state values in the post-fault grid configuration).

• Reactive power restoration time: This is defined as the needed span time to restore the reactive power (generally, within ±10% of the steady state values in the post-fault grid configuration).

• Energy imbalance: During the protection process, deficit or surplus energy (energy imbalance) could occur at the Point Common Coupling (PCC). Hence, this leads to an AC transient behaviour (acceleration/deceleration of the generators). It is proposed to use this energy imbalance during the protection process as a metric for the protection strategy impact on the connected AC system(s).

Figure 14: Proposed indicators for "simplified" operational risk analysis

The aim of different time restorations (fault interruption, voltage, and active/reactive power) is to evaluate the protection strategy performance during different phases of the fault clearing processes. For more details about these indicators, readers can refer to [3]. They give information to assess the impact on both DC and AC grids.

These time indicators are to be put in relation with functional requirements from AC systems on the DC grids protection system as defined in Deliverable 4.1 [10], mainly with times t1 and t2 defined in Figure 15.

Figure 15: Maximum loss of power infeed and duration of an AC grid connected to a HVDC grid [10]

Efficiency Indicators

(supported by EMT simulation)

Fault InterruptionTime

Voltage Restoration Time

Active Power Restoration Time

Reactive Power Restoration Time Energy Imbalance

Page 32: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

31

PRINCIPLE

The indicators introduced in this section are supported by extensive EMT simulations. They are calculated for each protection sequence (main and backup sequences) and for different fault locations. These fault locations are selected in such a way as to cover the extreme possible cases (in terms of fault interruption times and energy imbalance values). The more fault locations are considered, the higher accuracy the efficiency indicators can be computed. Moreover, if the component failures are considered during the EMT simulation, these indicators will be quite similar to those presented in section 3.3.2.

The final indicators are calculated either per nodes (converter) or per zone. The maximum value between all fault locations and protection sequences (main/backups) is then considered.

MODEL

As said above, the efficiency indicators are mainly determined by EMT simulations. So, in addition to the protection sequences model (implemented on different relays), physical models (cables, transformers, converters, etc.) and control models are also required. More details about protection system efficiency computations by EMT simulations can be found in [3].

3.3.2 STOCHASTIC EFFICIENCY AND PROTECTION SCHEME RELIABILITY INDICATORS

From a risk analysis point of view, it seems essential to quantify the operational risk associated to a given protection scheme, i.e. how far a given protection scheme could affect operation of the system in terms of security and reliability. Such an operational risk assessment would then reflect a level of performance to put in perspective of a given cost.

It is proposed to assess the protection scheme reliability through two main indicators:

• Failure indicators: Robustness and reliability of the protection strategy. The failure rate in terms of fault clearing will be computed considering both material and algorithmic reliability rates. Through a sensitivity analysis, requirements (component failure rate, redundancy level) to meet acceptable protection system failure rates could be defined.

• Efficiency indicators (supported by stochastic simulations): It is proposed to use restoration time distributions for a stochastics analysis of efficiency indicators (see Figure 16).

Figure 16 introduces the main Key Performance Indicators considered within WP 4. Efficiency indicators include fault interruption time, voltage restoration time, active/reactive power restoration times and energy imbalance and are already introduced in [1] [3].

Figure 16: Proposed indicators for "simplified" operational risk analysis

Stochastic efficiency and

protection reliability indicators

Failure rate computation

(main, backup sequences)

Efficiency Indicators

(supported by MC

simulation)

Fault Interruption

Time

Voltage Restoration

Time

Active Power Restoration

Time

Reactive Power

Restoration Time

Page 33: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

32

PRINCIPLE

The fault clearing performance indicators considered in the risk analysis approach aim to evaluate two main indicators: protection system failure rates (main and backup sequence performance) and restoration time distribution (time to restore active power). These two indicators are highly related to:

• The succession of operations (e.g. fault detection, fault clearing, system power restoration …) involved in the protection sequence, both for primary protection and back-up sequences. This is described using the protection system flowchart (algorithmic sequences, see [11] and [3] for more details).

• The time duration associated to each of these operations, mainly described by the protection system time chart (see [11] and [3] for more details).

The proposed approach to quantify these two indicators is based on a Monte Carlo Petri Net methodology (MCPN). Petri Net representation simplifies modelling of the protection system dynamics (succession of events) on the basis of both flowchart and time chart descriptions. Moreover, it also allows to describe, on the same base support, the protection system architecture including physical links between the different components (relays, DCCB, High Speed Switch HSS, telecommunication …). So doing, a complete representation of protection system architecture and operation could be defined as a base model for the Monte Carlo analysis of the protection system behaviour, from a probabilistic point of view. Main outputs are then:

• The distribution of success or failure of protection system, from which information such as protection system failure rate, main sequence failure rate, etc. can be deduced, as illustrated in Figure 17.

• The distribution of fault interruption time, as illustrated in Figure 18, from which the risk to be close to “maximum loss of power duration of an AC grid connected to a HVDC grid” requirement can be evaluated.

To carry out such analysis, it is required to have quite representative reliability models, and associated parameters/data, for each protection system component. Critical components from an operative point of view are those which are involved in active operations during both main and back-up protection sequences, i.e. DCCB, relays, HSS, converters etc. (in this deliverable, the converter with its control system is considered as fully reliable during the protection process). For each of these components, it would be required to know the following parameters:

• Failure rate (probability) during operation • Undetectable idle failure rate

Both of these failure rates will play a role in the protection system reliability analysis. Detectable idle failure rate is not considered at reliability level analysis as far as it will be expected that a line (or a grid partition), on which protection components are known to be non-operational, will not be used by MTDC grid operator. In other words, it means, for example, that if we know that a line breaker is out of service, the line is considered as disconnected. So doing, the failure of the protection components, associated to this line, has no more impact on the success or the failure of the protection system when a fault occurs.

Page 34: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

33

Figure 17: Examples of reliability assessment outputs- Protection system failure and success rate repartition

Figure 18: Examples of reliability assessment outputs- Distribution of “fault interruption time”

MODEL

From the indicator outputs of the MCPN model and simulation (sequence and time distributions), some specific first level indicators can be computed. These specific indicators should be relevant enough to allow a discrimination (assessment, comparison and benchmarking) between risks associated to each protection scheme. Five specific indicators are proposed. The two first ones are linked with the sequence distribution pattern:

• IMAIN: Probability to be in back-up sequences (i.e. main sequence failure rate): This indicator will give information about the performance of the main sequence and, as a consequence, about the importance to have relevant back-up sequences

• IBACKUP: Probability to be in All AC CB back up sequence (i.e. main and back-up sequences failure rate): This indicator will give information about the performance of all back-up sequences and about the importance of the “ultimate” back-up sequence, which is suspected to lead to a sustained non-selective

Page 35: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

34

outage of the DC grid. This is an important risk indicator as sustained outages could impact the AC system stability.

The three other ones are related to time distribution patterns and try to extract information regarding time sequence duration. ITR, time to return to operation, is here defined as the time to clear the fault and to be ready to operate (ready to restore voltage, active and reactive powers).

• ITR_20: Probability to have a time to return to operation tR > 20 ms

• ITR_50: Probability to have a time to return to operation tR > 50 ms

• ITR_200: Probability to have a time to return to operation tR > 200 ms

ASSESSMENT, COMPARISON AND BENCHMARKING INDICATORS

It is proposed to assess, compare and benchmark protection strategies on the basis of four main items:

1. “Speed” of the protection strategy, i.e. related to fault interruption time (time to be ready for operation)

2. Performance of the main sequence

3. Performance of the back-up sequences (except the “All AC CB back up” ultimate sequence)

4. Risk associated to the AC/DC stability issue

A. “SPEED” OF THE PROTECTION STRATEGY INDICATOR

It is intended here to try to classify the protection strategies regarding the duration (“speed”) to clear the fault and be ready to operate. The analysis is based on the indicators ITR_20, ITR_50 and ITR_200 as proposed in Table 1. This information could be related to the relative compatibility of the protection strategy with the constraints of small, medium and large grids, as introduced in WP4.2 [1].

Table 1: Proposed classification for “speed” of the protection strategy indicator

INDICATOR VALUE CLASSIFICATION

ITR_20: prob (tR > 20 ms) Low (< 10%) Very fast

ITR_50: prob (tR > 50 ms) Low (< 10%) Fast

ITR_200: prob (tR > 200 ms) Low (< 10%) Medium fast

ITR_200: prob (tR > 200 ms) High (> 10%) Slow

B. PERFORMANCE OF THE MAIN SEQUENCE INDICATOR

It is proposed here to examine the protection strategy from the probability that back-up protection sequences are necessary, in case of the failure of the main protection sequence to clear the fault. From the indicator IMAIN, the protection strategies could be ranked as proposed in Table 2 where IMAIN_mean is given by equation 3-3.

𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀_𝑚𝑚𝑚𝑚𝑚𝑚𝑜𝑜 = 1𝑁𝑁𝑁𝑁𝑆𝑆

�𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀_𝑆𝑆𝑖𝑖𝑆𝑆𝑖𝑖

3-3

with 𝑁𝑁𝑁𝑁𝑆𝑆 is the number of protection strategies to study, 𝐸𝐸𝑖𝑖 is the protection strategy i index and 𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀_𝑆𝑆𝑖𝑖 is the indicator IMAIN for protection strategy 𝐸𝐸𝑖𝑖.

Page 36: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

35

Table 2: Proposed classification for main sequence robustness indicator

INDICATOR STATE CLASSIFICATION

IMAIN 𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀_𝑆𝑆𝑖𝑖 <𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀_𝑚𝑚𝑚𝑚𝑚𝑚𝑜𝑜 +

IMAIN 𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀_𝑆𝑆𝑖𝑖 > 𝑅𝑅𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀_𝑚𝑚𝑚𝑚𝑚𝑚𝑜𝑜 -

C. PERFORMANCE OF THE BACK-UP SEQUENCES INDICATOR

Similarly to what is proposed for the performance of the main sequence, it is proposed to rank the protection strategies according to the performance of back-up sequences, i.e. the probability that the “All AC CB back up” ultimate protection sequence is required, where AC/DC stability issue could be more critical (Table 3). From the indicator 𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵, the protection strategy could be ranked as proposed in Table 3 where 𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵_𝑚𝑚𝑚𝑚𝑚𝑚𝑜𝑜 is given by equation 3-4.

𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵_𝑚𝑚𝑚𝑚𝑚𝑚𝑜𝑜 = 1𝑁𝑁𝑁𝑁𝑆𝑆

�𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝑆𝑆𝑖𝑖 𝑆𝑆𝑖𝑖

3-4

with 𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝑆𝑆𝑖𝑖 is the indicator 𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵, for protection strategy 𝐸𝐸𝑖𝑖.

Table 3: Proposed classification for back-up sequences robustness indicator

INDICATOR STATE CLASSIFICATION

IBACKUP 𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝑆𝑆𝑖𝑖< 𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵_𝑚𝑚𝑚𝑚𝑚𝑚𝑜𝑜 +

IBACKUP 𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝑆𝑆𝑖𝑖 > 𝑅𝑅𝐵𝐵𝑀𝑀𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵_𝑚𝑚𝑚𝑚𝑚𝑚𝑜𝑜 -

D. AC/DC STABILITY ISSUE RISK INDICATOR

As said before, the AC/DC stability issue is mainly related to the duration of the interruption of real power transmission within HVDC grid (and mainly with the time threshold t2 defined in Figure 15). It is proposed here to assess the protection strategy from the risk (probability) that the time to return to operation is higher than a given threshold, chosen here as equal to 200 ms. From the indicator ITR_200, the protection strategies could be ranked as proposed in Table 4.

Table 4: Proposed classification for AC/DC stability issue risk indicator

INDICATOR STATE CLASSIFICATION

ITR_200: prob (tR > 200 ms) Low (< 1%) ++

ITR_200: prob (tR > 200 ms) Medium (> 1% and < 10%) +

ITR_200: prob (tR > 200 ms) High (> 10%) -

3.4 AC SYSTEM IMPACT INDICATORS

This section focuses on the AC grid impact to determine the cost-benefit of employing a particular protection scheme. The impact is analysed in dynamic as well as steady state conditions. In the dynamic condition, the choice of DC grid protection can influence the stability of the AC grid. The stability can be classified in three categories namely frequency, voltage and rotor angle stability. We focus on the frequency stability in this

Page 37: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

36

deliverable since it is a primary concern for the future low-inertia grids. Rotor angle and voltage stability are not considered in this report as they would require to have a detail dynamics model of the AC grids and cannot be analysed with a “simplified” equivalent model as it can be done for frequency stability. The present-day AC grids are equipped with a certain amount of frequency reserves to ensure the stability against the worst-case fault (sizing incident) scenario. However, with the reducing grid inertia due higher share of renewables, the current reserves may not be sufficient. At first, section 3.4.1 offers some typical KPIs to indicate the frequency stability and then, 3.4.2 proposes a method to calculate the required frequency reserves. The method determines a trade-off between the required reserve capacity and the required load shedding due to frequency instability.

In section 3.4.3, the impact in the steady state condition is analysed. After the DC fault clearing, the AC grid generators need to be redispatched to compensate for the loss of infeed from the DC grid and/or to establish new power flow set points. The redispatch may cause additional outages in the AC grid, if line flows and bus voltages exceed the protection settings, resulting in a modified AC grid topology. The optimal dispatch cost of the modified AC/DC grid topology is likely to be higher than the cost before the fault. The difference between these two costs gives the redispatch cost due to a DC contingency. Moreover, the cost of load shedding due to the frequency instability is added to the redispatch cost. The amount of required load shedding is determined after equipping the grid with the calculated reserves from former section 3.4.2. The N-1 reliability criteria is used for both analyses. The DC line outages are considered as a single credible contingency.

3.4.1 FREQUENCY STABILITY

The frequency response model from [12] is adopted in this work. The model includes the inertial and primary frequency responses. Immediately following a disturbance (load/ generation loss) in a grid, the change in system frequency is mainly determined by the system inertia. Following the initial period, the speed controllers of the generators participate in the regulation of the frequency. This is called primary frequency control (PFC). Each generator participates in PFC according to its own droop characteristic.

Δωr1MS + D-+

11 + sTg

1R

ΔPL

Figure 19: Equivalent model of inertial and primary frequency control response [13] [14]

Figure 19 depicts the equivalent model used to obtain the frequency response of an AC grid, where M = 2*H is the equivalent inertial constant of the grid, R is the equivalent droop constant of the grid, Tg is the aggregated turbine-governor time constant and D is the system’s representative load-frequency characteristic which represents the percentage change in load per % change in frequency. The value of D is typically 1-2 %, The equivalent inertia constant of the system can be given as [13]

𝐻𝐻 =1𝐸𝐸𝑠𝑠𝑠𝑠𝑠𝑠

�𝐸𝐸𝑖𝑖 𝐻𝐻𝑖𝑖

𝑀𝑀𝑔𝑔

𝑖𝑖=1

3-5

Where 𝐸𝐸𝑖𝑖 is the nominal MVA capacity of 𝑖𝑖𝑡𝑡ℎ generator, 𝐸𝐸𝑠𝑠𝑠𝑠𝑠𝑠 is the MVA capacity of the system, 𝐻𝐻𝑖𝑖 is the inertia of 𝑖𝑖𝑡𝑡ℎ generator and 𝑁𝑁𝑔𝑔 is the total number of generators in the system.

Page 38: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

37

Similarly, the equivalent droop constant of the AC system is [13]

1𝑅𝑅

= 1𝐸𝐸𝑠𝑠𝑠𝑠𝑠𝑠

�𝐸𝐸𝑖𝑖𝑅𝑅𝑖𝑖

𝑀𝑀𝑔𝑔

𝑖𝑖=1

3-6

Where 𝑅𝑅𝑖𝑖 is the droop constant of individual generator and 𝑓𝑓0 is the nominal frequency of the system.

The input to Figure 19 is a power deviation considering pre-fault, during fault and post-fault time. A typical power deviation curve at the PCC can be depicted as shown in Figure 20 (a). ∆PHVDC,dyn is the instantaneous/dynamic power dip immediately after the fault occurrence and ∆PHVDC is the steady state power deviation. ∆THVDC is the power restoration time depending on the protection method under study. An example frequency response to the depicted power deviation is shown in Figure 20 (b).

(a) (b) Figure 20: (a) Active power deviation curve at PCC (b) Frequency response representation

Table 5 shows the frequency quality defining parameters (ROCOF, ∆𝑓𝑓𝑖𝑖𝑜𝑜𝑠𝑠𝑡𝑡, ∆𝑓𝑓𝑠𝑠𝑠𝑠 ) for four different synchronous areas in Europe. Note that the maximum allowed ROCOF is not yet defined for all synchronous areas in Europe. The Irish system (IE/NI) defines the limit to 0.5 Hz/s (over a 500 ms time window) which is expected to be increased to 1 Hz/s in future. The Great Britain (GB) system has recently upgraded the limit from 0.125 Hz/s to 0.5 Hz/s (over a 500 ms window) for the existing synchronous generators and to 1 Hz/s (over a 500 ms window) for all new synchronous generators plus new or existing non-synchronous generators [15]. Although there is no defined limit for Continental Europe (CE), the current system can successfully sustain a ROCOF in the range of 0.5 Hz/s to 1Hz/s as stated by ENTSO-e [16].

Table 5: Frequency quality defining parameters

Continental Europe Great Britain Ireland and

Northern Ireland Nordic

ROCOF range not defined 0.5 Hz/s – 1Hz/s 0.5 Hz/s, changing to 1 Hz/s Not defined

Minimum frequency (∆𝑓𝑓𝑚𝑚𝑖𝑖𝑜𝑜) [17] 49 Hz 49 Hz 49 Hz 49 Hz

Maximum steady state frequency deviation (∆𝑓𝑓𝑠𝑠𝑠𝑠 ) [17] 200 mHz 500 mHz 500 mHz 500 mHz

Out of the mentioned parameters to determine the frequency stability, ROCOF and 𝑓𝑓𝑚𝑚𝑖𝑖𝑜𝑜 are used as KPIs in this work since they are determining parameters for under frequency load shedding. Additionally, the average

Page 39: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

38

equivalent inertia of the system is proposed as an indicator. Table 6 shows the classification of all three proposed indicators, keeping the Nordic grid as a reference.

Table 6: Proposed classification for frequency response indicator

INDICATOR STATE CLASSIFICATION

ROCOF < 0.5 Hz/s > 0.5 Hz/s and < 1 Hz/s > 1 Hz/s ++ + -

Fmin < 49 Hz - > 49 Hz - ++

Heq > 3 sec < 3 sec and > 2 sec < 2 sec ++ + -

3.4.2 RESERVE REQUIREMENT

Any imbalance between supply and demand in the power system can cause the grid frequency to deviate from the nominal value. To counteract such imbalances and to keep the frequency at its nominal value, each synchronous area is equipped with sufficient operating reserves. The fundamental theory on frequency reserves is covered in appendix A.2. We focus particularly on the frequency containment reserves (FCR) since they are responsible to limit the frequency deviation during the time frames of protection operation. During the planning horizon of this project (2020-2050), the FCR requirement can be affected by the following mechanisms:

1) Effects of reducing inertia and different protection schemes:

With the integration of more renewable resources, the system inertia is expected to drop. The reducing inertia is identified to be the primary concern of future European power system [18]. The reduction of the system inertia leads to higher ROCOF values that may cause the frequency nadir to decrease below 49 Hz before the primary control activates. This work deals with different offshore HVDC grid topologies during a planning horizon of 2025 – 2050, and some of these topologies are expected to infeed substantial power to the synchronous areas. With a low inertia in the AC system, a fault in such DC grid can cause a sudden loss of infeed from the DC grid which might lead to a frequency nadir violating the permissible limit.

In addition, if the offshore DC grid is connected at more than one point in a synchronous area, the amount of sudden power loss due to a DC grid fault will be different for selective and non-selective protection strategies, possibly causing different minimum frequency violations in each case.

2) Effects of different protection timings:

The proposed protection schemes within the project are expected to restore the power within a few hundred milliseconds whereas the system frequency reaches its nadir in a few seconds (e.g. 4-5 sec in the Nordic grid). Without the power restoration process, the frequency will drop from the instance of the disturbance until the activation of the primary control. However, due to a quick restoration time offered by the protection schemes, this drop will be constrained. Since the restoration time is different for each strategy, the maximum instantaneous frequency deviation will be different for each protection strategy. Since this deviation is responsible for under frequency load shedding (UFLS), each protection scheme can have different load shedding.

The following subsections discuss both effects in the Nordic system without changing FCR reserves in the current system.

INFLUENCE OF REDUCED INERTIA

The ENTSO-e study [19] provides the inertia evaluation for the years 2020 – 2025. Note that the available kinetic energy in the grid at a time does not depend on the installed capacity or type of generation, but on the connected rotating mass [20]. Therefore, the inertia can change throughout the year depending on the generation mix used at a time. The minimum kinetic energy in the Nordic for year 2025 are given as 80 GWs, for low load and reduced

Page 40: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

39

nuclear generation, and 102 GWs, corresponding to the base case scenario. The corresponding power production is between 24.5 GW and 26.5 GW. Based on these data, the minimum and maximum inertia constants (H = M/2) are calculated as 3.2 and 3.8 s using equation 3-5.

A brief case study is conducted here to check if a reduction in system inertia has any impact on the frequency stability. The simplified frequency response model introduced in section 3.4.1 is extended in order to include hydro power plants as the main FCR resource in the Nordic grid (see [19] for more details). The FCR-D limit is set at ±1450 MW. Figure 21 shows the frequency response for different power deviations for a minimum system inertia (inertia constant = 3.2 s). For any power imbalance bigger than 1400 MW, the frequency nadir goes below the current limit of 49 Hz, causing under frequency load shedding (UFLS). The frequency stability can be an issue even at lower power imbalances when inertia further drops further in a post 2025 scenario. Note that we have considered permanent loss of infeed here. Next section will analyse the effect of temporary loss.

Figure 21: Frequency response with H = 3.2 s, FCR = 1450 MW in scenario 2025 (red line – minimum permissible frequency)

INFLUENCE OF DIFFERENT PROTECTION STRATEGIES

Two different DC grid protection strategies with different power restoration times are considered here to analyse the influence of different protection strategies on the frequency nadir. Although the timings do not correspond to a particular protection strategy, they represent the lower and upper range of power restoration times among the studied DC grid protection strategies. A hypothetical infeed to the Nordic grid is taken as 2000 MW which is assumed to be restored back to the full level after the power restoration as shown in Figure 22(a).

(a) (b)

Figure 22: (a) Power deviation (b) Frequency response with H = 3.2 s, FCR = 1450 MW in scenario 2025

Page 41: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

40

The frequency nadirs for both protection restoration times are different which indicates that different protection schemes might lead to different minimum frequency violations and consequently, different reserve requirements to avoid these violations. By comparing Figure 22(b) to Figure 21, it is worth to point out that the permanent loss and temporary loss of infeed have a different influence on the frequency response. During the temporary loss, the minimum frequency is well within the permissible limits whereas a permanent loss of even lower infeed leads to frequency instability. For example, the minimum frequency is ~49.7 Hz for the temporary loss of 2000 MW whereas it is lower than 49 Hz for the permanent loss of 1600 MW.

3.4.3 FREQUENCY RESERVES

Multiple measures can be employed to avoid load shedding caused by minimum frequency violations as listed in [21]. One category of measures is to implement the fast frequency reserves (FFR) that can be activated swiftly compared to the traditional FCR reserves and contain the frequency within the allowed limit. Note that although the energy storages are out of scope of the project, we see them as a key solution to reliability and stability challenges, caused by high amount of wind power penetration. The FFR can be supplied by multiple technologies and its model would depend on the specific technology used. In order to comment on the FFR requirement in general, the FFR can be modelled in a simple manner, a function with an activation time delay (𝑇𝑇𝑔𝑔 ) and a ramp rate (MW/s). The response from FFR is assumed to be proportional to the frequency deviation as shown in Figure 23. Both ramp rate and activation time are important factors to determine the FFR capacity [22]. Energy storage (battery or flywheel), demand curtailment and HVDC interconnectors can provide higher ramp rates in comparison to synchronous condensers. A sufficiently high ramp rate (1000 MW/s set as limiter) is assumed to focus only on the reserve sizing. The activation time (𝑇𝑇𝑔𝑔 ) of FFR is expected to be below 0.5 sec [22].

Δωr1MS + D-

ΔPL

Hydro PowerPlant

FCR limiter

FFR limiter

-

FFR

+

Grid

Figure 23: Frequency response model of Nordic grid with FFR [19]

METHODOLOGY

The FFR requirement can be defined deterministically based on the minimum inertia and a fixed power deviation stemming from an N-1 offshore DC grid contingency. The calculated capacity ensures that there is no under frequency load shedding for all scenarios. The calculated reserve capacity can be higher than the cost of load shedding which might be expected for a few hours in a year. A probabilistic method, that finds a trade-off between the cost of capacity reserves and the benefits received in terms of lowered load shedding, can provide a more cost-effective solution. Figure 24 shows the proposed probabilistic methodology.

Page 42: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

41

Figure 24: Methodology to determine FFR requirement

The methodology starts with hourly generation data (including offshore wind power) of different power plants types (i.e. nuclear, hydro, renewable etc.). These data, along with typical inertia values for each generation type (see appendix A.3), are used to calculate the total system inertia at each hour based on equation 3-5. Using the calculated system inertia and the available wind generation at each hour, the frequency response for each DC contingency is calculated. The value of the frequency nadir is extracted from frequency response and if the value is less than 49 Hz, a load shedding instance is noted. The load shedding instances are summed up over one year and weighted according to the probability of the contingency (e.g. unavailability of the DC cable). The obtained load shedding per annum is compared against the acceptable limit defined by TSOs. If the accumulated load shedding over one year is higher than the acceptable limit, the FFR capacity is increased and the whole method is repeated till the load shedding duration is within the limit of the defined increased loss of load expectation (ΔLOLE). The basic LOLE is defined by TSOs based on statistical analyses of all contributing factors to it. The proposed method calculates the additional contribution (ΔLOLE) from a DC line contingency to the basic LOLE.

3.4.4 REDISPATCH COST

This section proposes a methodology to determine the benefits of employing a particular DC protection scheme and network topology in terms of AC grid redispatch (see Figure 25). The method begins by running a power flow on a hybrid AC/DC network and determining the power exchange between the AC and DC grids at their connection points. The active power set points of the generators are set to their optimal dispatch values during the power flow run (corresponding to the wind sample). This in turn would provide an optimal cost of dispatch at the beginning of the method when the system is healthy. Then, contingencies are created on different lines of the DC grid (e.g. N-1 criterion). The DC faults are assumed to be pole-to-ground type in a bipolar grid. After disconnecting the faulty

Page 43: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

42

line, the power flow is calculated again and the post-fault power exchange between the AC and DC grid is obtained. At this point, the power deviation is calculated by taking the difference between the pre-fault and post-fault power exchanges and distributed equally among the active generators within the AC grid. In a real grid, the participation of each generator can be different and depends on the droop constant of the individual generator. However, the participation is assumed to be equal here for simplification.

Power flow simulation outcomes are then used to check the constraint violations. Three types of constraint violations in the AC grid are checked.

a) Voltage limit violations b) Branch flow limit violations c) Frequency limit violations (using Simulink model)

The voltage and branch flow limit violations are determined based on the post-fault power flow solution. On the other hand, the frequency limit violations are determined based on the frequency response model shown in section 3.4.1.

Each kind of violation is dealt with by different philosophies:

a) Branch flow limit violations: Once a DC line fault is cleared, the power exchange between AC and DC grid will settle to a new value (assuming DC grid is not n-1 contingent). The post fault flows will be different in both AC and DC grids compared to those before the fault. The new power flows might cause overloads in the connected AC grid. The AC lines can be operated with a certain overload for a short duration, depending on the thermal characteristic of the line. Meanwhile, the generators can be redispatched to remove the overloads. However, if the line overload is beyond the allowed temporary limit, the line protection will trigger and isolate the line from the rest of the network. In this work, we assume that the lines can be overloaded up to 50% for a short duration and that such overloads are removed using redispatch. When a branch flow is higher than this limit, the line is isolated from the grid. The overloading is analysed only in the AC network whereas the DC network overloading is not considered.

If one or more branches connected to a node are opened, the remaining branches may not have sufficient capacity to deliver the load connected to that node, resulting in load shedding. The cost of load shedding is considered as 40 €/MWh.

b) Voltage limit violations: Similar to line flow limits, it is assumed that the power system buses can temporarily sustain overvoltage condition. The temporary upper and lower voltage limits are set to typical steady state limits as 1.1 pu and 0.9 pu respectively. If a bus voltage is beyond these limits, the corresponding node is isolated from rest of the network. The isolated bus could be connected to a load. In such a case, the cost of load shedding is accounted.

c) Frequency limit violations: If the frequency nadir drops below the 49 Hz, under frequency load shedding is activated. The recommended amount of ULFS differs for each country and is defined in steps over frequency ranges. In this study, 5 % of load is shed once the frequency reaches 49 Hz [23].

Page 44: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

43

Figure 25: Methodology for cost benefit analysis of DC protection impact

Both branch flow and voltage limit violations would modify the AC network with opening the branches or isolating the nodes that consequently will result in a new optimal dispatch cost of generators. This cost will be less economical than the optimal dispatch cost before the fault. If any frequency violation occurs, the load shedding cost will be added to the new dispatch cost. Therefore, the final dispatch cost in event of any kind of violation must be higher than the initial dispatch cost.

At the last step of the method, the cost of redispatch due to a DC grid contingency can be determined. During each iteration of the method, the redispatch cost for each wind sample (𝑘𝑘) and corresponding contingency (𝑖𝑖) is calculated as:

𝐶𝐶𝐷𝐷𝐵𝐵_𝑐𝑐𝑜𝑜𝑜𝑜𝑡𝑡(𝑘𝑘, 𝑖𝑖) = 𝐶𝐶𝑝𝑝𝑜𝑜_𝑜𝑜𝑚𝑚𝑡𝑡𝑛𝑛𝑜𝑜𝑔𝑔𝑛𝑛(𝑘𝑘, 𝑖𝑖) − 𝐶𝐶𝑖𝑖𝑜𝑜𝑖𝑖_𝑜𝑜𝑚𝑚𝑡𝑡𝑛𝑛𝑜𝑜𝑔𝑔𝑛𝑛(𝑘𝑘, 𝑖𝑖)

3-7

𝐶𝐶𝑝𝑝𝑜𝑜_𝑜𝑜𝑚𝑚𝑡𝑡𝑛𝑛𝑜𝑜𝑔𝑔𝑛𝑛 is the optimal dispatch cost (including load shedding) in the post fault network after the violation removal and 𝐶𝐶𝑖𝑖𝑜𝑜𝑖𝑖_𝑜𝑜𝑚𝑚𝑡𝑡𝑛𝑛𝑜𝑜𝑔𝑔𝑛𝑛 is optimal dispatch cost in the initial network before the fault. The method runs multiple samples, taking the hourly wind conditions throughout a year into account. The yearly operational cost of a particular protection strategy is given as:

Page 45: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

44

𝐿𝐿𝑃𝑃𝐸𝐸𝑋𝑋𝑔𝑔𝑚𝑚𝑔𝑔𝑖𝑖𝑠𝑠𝑝𝑝/𝑦𝑦𝑦𝑦 =8760𝑁𝑁

��𝐶𝐶𝐷𝐷𝐵𝐵_𝑐𝑐𝑜𝑜𝑜𝑜𝑡𝑡(𝑘𝑘, 𝑖𝑖) ∗ 𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑖𝑖𝑢𝑢𝑢𝑢𝑁𝑁𝑖𝑖𝑢𝑢𝑖𝑖𝑑𝑑𝑦𝑦(𝑖𝑖) 𝑀𝑀𝑙𝑙

𝑖𝑖=1

𝑀𝑀

𝑛𝑛=1

3-8

Where 𝑁𝑁𝑙𝑙 is the total number of DC contingencies and 𝑁𝑁 is the total number of samples. The annual cost due to DC contingencies are weighted according to the probability of DC cable failures based on the unavailability data for cables shown in appendix A.1. The proposed method is a more detailed version of the expected energy not transmitted (EENT) analysis. The redispatch cost of generators provides a realistic cost of N-1 contingency. The method also respects the network limits (for line, bus and generator) and stability limits (frequency stability) during the redispatch.

3.5 EXTENSIBILITY

A preliminary analysis on extensibility of each protection strategy is already covered in deliverable 4.2 [1]. We summarize it here for the completeness of the document. A detailed analysis is out of the scope of this deliverable.

The extensibility is considered in three cases: • Addition of a DC line • Addition of a converter to the same or different AC grid • Change of the AC grid requirement: the requirement from AC grid is considered one of three types, i.e.

temporary stop, permanent stop or continuous operation of the converters

The extensibility of the four DC grid protection strategies are commented here. The readers are referred to [1] for other protection strategies.

For addition of line or converter, both fully selective strategies, using fast DC breaker (FS-FDCCB) and slow DC breaker (FS-SDCCB) remain unchanged since the converters can operate in continuous mode. The DCCB capacity might need to be upgraded. If the AC grid requirements are changed, the primary protection components do not need changes. However, the backup protection components can not comply if the new requirement is continuous operation. Note that none of the protection strategies (FS or NS) can offer continuous operation of converters in the backup sequence.

For the non-selective strategies, if a new converter is added to different AC grid, there is no change required in the protection. However, if a new converter is connected to the same AC grid, the temporary loss of infeed changes to a higher value. If the new value is acceptable based on AC grid requirements, there is not fundamental change needed in the strategies. However, if the new value of temporary loss of not acceptable, the non-selective strategies may not be useful. A grid splitting option is attractive in this case. The NS strategies cannot meet continuous operation requirements from AC grid and are not suitable for high impact DC grids.

The extensibility requirement is highly dependent on the impact of the extended DC grid and the AC grid requirement at the time of extension. A detail study for individual AC/DC grids will be needed to quantify such requirement. The frequency stability/reserve requirement study within this deliverable provides a good example of it where we have considered three different test cases with three different DC grid (WP 4, WP 12 and a generic grid) and the changing requirements in AC grid (inertia variation till 2050), reaching different results for each case (e.g. no specific trend among either slower to faster or non-selective to selective protection. In some cases, the slower and faster selective protection have the same load shedding duration). In general, there is a benefit of using fully selective over the non-selective strategies, but the difference between two remains dependent on the test case. Therefore, the question of how big the difference must be answered by studying each case individually.

Page 46: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

45

3.6 COMPUTATION OF THE COMPONENTS RELIABILITY

The availability 𝐸𝐸 of a given system (or given component) is defined as the probability that this system (or component) is available (works normally) for a given time period. Similarly, unavailability ζ of a given system is defined as the probability that this system is not available during a given time period; as a consequence:

Unavailability (ζ) = 1 – Availability (𝐸𝐸)

Basically, the most common reliability parameter is the Mean Time To Failure (MTTF), assuming exponential distribution of the failure occurrence. For repairable systems (which is often the case), Mean Time To Repair (MTTR) could be necessary to perform reliability studies. These two parameters could be obtained from failure rate λ (number of failures per unit of time) and repair rate μ (repair time per unit time). When the failure rate λ is stable (which is the case of power electronic components) and failures occur randomly, MTTF and MTTR might be approximated by exponential probability density function. In this case, MTTF and MTTR are given by equations 3-9 and 3-10 respectively.

𝑀𝑀𝑇𝑇𝑇𝑇𝑀𝑀 = 1/𝜆𝜆 3-9

𝑀𝑀𝑇𝑇𝑇𝑇𝑅𝑅 = 1/𝜇𝜇 3-10

Finally, unavailability ζ may be expressed by equation 3-11.

𝜁𝜁 = 1 − 𝐸𝐸 =𝑀𝑀𝑇𝑇𝑇𝑇𝑅𝑅

𝑀𝑀𝑇𝑇𝑇𝑇𝑀𝑀 + 𝑀𝑀𝑇𝑇𝑇𝑇𝑅𝑅 3-11

This equation is used in the following sections to compute unavailability of protection strategy components from their reliability data parameters (MTTF and MTTR).

3.7 MONETARIZATION: SYSTEM LEVEL AGGREGATED INDICATORS

It is proposed to compute monetarized indicators from the different indicator calculation methodologies, from which it is possible to compute system level aggregated cost indicators such as Life Cycle Cost (LCC). LCC is a system level indicator generally used in project economic assessments. One main advantage of LCC is that it has a linear formulation (see equation 3-12) and could then be applied at component level, which is to say that an individual LCC could be associated with a restricted set of components, e.g. protection system components. This property enables the separate computation of a LCC indicator related to the full DC grid system (LCCTS, see equation 3-13) and a LCC indicator related to the protection system (LCCPS, see equation 3-14), allowing relative assessments and comparisons to be carried out.

𝐿𝐿𝐶𝐶𝐶𝐶 = �(𝐶𝐶𝑖𝑖 + 𝐿𝐿𝑖𝑖 + 𝑀𝑀𝑖𝑖)

(1 + 𝑦𝑦)𝑖𝑖

𝑀𝑀

𝑖𝑖=1

3-12

where

• 𝑪𝑪𝑖𝑖 is the investment costs of year i • 𝑶𝑶𝑖𝑖 is the operation costs which are here costs related to losses and EENT of year i • 𝑴𝑴𝑖𝑖 is the maintenance costs of year i • r is the discount rate (from European project studies, a typical discount rate is 8% per year) • N is the total number of years within the system life duration

𝐿𝐿𝐶𝐶𝐶𝐶𝑇𝑇𝑆𝑆 = �(𝐶𝐶𝑇𝑇𝑆𝑆𝑖𝑖 + 𝐿𝐿𝑇𝑇𝑆𝑆𝑖𝑖 + 𝑀𝑀𝑇𝑇𝑆𝑆𝑖𝑖)

(1 + 𝑦𝑦)𝑖𝑖

𝑀𝑀

𝑖𝑖=1

3-13

𝐿𝐿𝐶𝐶𝐶𝐶𝐵𝐵𝑆𝑆 = �(𝐶𝐶𝐵𝐵𝑆𝑆𝑖𝑖 + 𝐿𝐿𝐵𝐵𝑆𝑆𝑖𝑖 + 𝑀𝑀𝐵𝐵𝑆𝑆𝑖𝑖)

(1 + 𝑦𝑦)𝑖𝑖

𝑀𝑀

𝑖𝑖=1

3-14

Page 47: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

46

The main drawback of LCC is that it requires monetarisation of all indicators, i.e. it requires to define a Cost Of Energy (COE) to be applied within both losses and EENT indicators. Table 7 presents proposed monetarization approaches for these indicators along with the monetarization guidelines proposed by ENTSO-E in [5].

Table 7: Proposed monetarization of indicators

Indicator (original units) Proposed Monetarization ENTSO-E recommended Monetarization

CAPEX (€)

Direct monetarization Direct monetarization

Losses (MWh/year)

Cost of Energy in €/MWh Cost of Energy could be issued from forecasted marginal costs estimated from Market assumptions to be defined within reference cases

Cost of Energy in €/MWh “Monetisation of losses is based on forecasted marginal costs in the studied horizon. These marginal costs are derived from market studies”.

EENT (MWh/year)

Option 1: Cost of Energy in €/MWh (could be issued from forecasted marginal costs estimated from Market assumptions to be defined within reference cases) Option 2: Based on the computation of the Value of Lost Load (VOLL). Contribution could be how far VOLL is modified by EENT due to protection scheme Option 3: Based on the computation of Social Economic Welfare (SEW). Contribution could be how far SEW is modified by EENT due to protection scheme. SEW approach would naturally include RES spillage effect. .

No monetarization of EENT is recommended. “The monetisation of system unreliability and security of supply using VOLL cannot be performed uniformly on a Union-wide basis. There is a large variation in the value that different customers place on their supply and this variation can differ greatly across the Union, as it depends largely on regional and sectorial composition and the role of the electricity in the economy. Additional factors such as time, duration and number of interruptions over a period also influence VOLL. Given the high variability and complexity of the VOLL, calculating project benefit using market-based assessment will only provide indicative results which cannot be monetised on a Union-wide basis. VOLL will therefore not be used as a basis for comparative EENT calculations“.

Maintenance (€/year)

Direct monetarization from maintenance cost models and parameters (e.g. percentage of CAPEX per year? cost of spare components + cost of maintenance operation?)

Direct monetarization

Monetarization of EENT is the main questionable issue. Indeed, EENT related monetized costs are directly linked to the consequence that energy not transmitted has on the interconnected AC/DC system. Such consequences are generally not easy to be quantified as they would generally require to make some assumptions on interconnected AC/DC grids, defined within reference cases studies. As proposed in Table 7, several options are possible. Both options 2 and 3 (i.e. SEW based and VOLL based indicators) are typically those which are very

Page 48: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

47

sensitive to case studies assumptions. So, in a first step of these CBA work, it is proposed to rely on option 1, i.e. consider a fixed cost of energy, to give a monetarization value to EENT indicators.

3.8 DCCB SPECIFICATION

The objective of this section focuses on the proposed methodology related to the protection strategy design (for further cost estimation) in order to support the deployment plan for future European offshore grid defined within WP 12. The main idea is to set up simplified DCCB design methodologies which will not require extensive EMT simulations. As such, those methodologies will allow a fast DCCB design and facilitate application to a large range of DC grids. This methodology consists of designing the protection strategy and to provide a rough component specification (including DCCBs, Surge Arrestors, energy dissipation and DC reactors). In other words, the objective is to give a representative DCCBs specification based on analytical models.

The presented methodology is firstly validated on WP 4.3 benchmark (The application to WP12 benchmarks is presented in section 6.1) and on four protection strategies: two Full-Selective Strategies (with mechanical and hybrid DC Circuit Breakers) and two Non-Selective Strategies (Full-Bridge and Converter Breaker based strategies). Consequently, a given generic methodology does not necessary work for all the protection strategies. In this section, two methodologies for non-selective (section 3.8.1) and fully Selective (section 3.8.2) will be presented.

As said in section 3.2.1, the DCCB cost model takes several parameters as inputs which are summarized in Figure 26. Consequently, the proposed methodology should estimate these parameters without doing any detailed EMT simulation, which are mainly dependent on the Grid topology/physical parameters and on the operating points.

Figure 26: DCCBs input parameters and main cost’s outputs

3.8.1 NON SELECTIVE PROTECTION DCCBS COST MODEL SPECIFICATION TO SUPPORT WP12 MOGS DEVELOPMENT

This section gives a general methodology to specify DCCBs in non-selective protection strategies including converter breaker NS-CB based and converter with fault blocking capability NS-FB based fault clearing strategies.

3.8.1.1 PROTECTION DESIGN METHODOLOGY FOR CONVERTER BREAKER STRATEGY

For the converter breaker protection strategy, two currents are mainly required to design the DCCBs:

1. Rated breaking current capability (𝐸𝐸𝐶𝐶𝐶𝐶𝑡𝑡): which depends mainly on the AC grid contribution to fault current. This current will define both to the converter breaker’s breaking current capability and its short time withstand current (see bellow).

Page 49: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

48

2. Rated short time withstand current: short time withstand current is the very short term transient current that the breaker will have to support, but not to break. It mainly depends on the cable discharging currents.

Consequently, knowing the maximum AC contribution (i.e. on the closest busbar connecting this converter to the AC grid) of each converter breaker allows to design its current capability and short time withstand current to this contribution. Following ways might be used to estimate this current:

1. By performing Electro Magnetic Transient (EMT) simulations. 2. By using approximate analytical methods. 3. Approximating this current using human expertise and simple rules of thumb.

In this current deliverable, the third option is adapted. The converter breaker‘s current capability is assumed to be equal to 20 kA (for 525 kV converters). For converters connected to a weak AC grid (short circuit ratio SCR< 3 p.u) 16 kA could also be used.

This assumption has been validated by performing EMT simulations throughout works carried out in WP 4.3 (the results of such study are presented in details in [1] and [3]). Furthermore, an analytical methodology to estimate converter breaker short circuit current (𝐸𝐸𝐶𝐶𝐶𝐶𝑡𝑡) is also proposed and described in appendix A.6 but has not been implemented.

The line breaker rated current breaking capability is set to the maximum of the converter rated breakers current capabilities. Indeed, the line breaker must be designed for the worst case (worst line fault and configuration). This is explained in Figure 27. For example, if the converter breaker 1 fails, the maximum short circuit current of the line DCCBs close to converter 1 is obtained when a fault occurs close to the converter 1 at line L14 (see upper of the Figure 27). Once the short circuit current is computed for each converter failure, the line breaking current capability (connecting converter 1 to converter 4 and close to converter 1) is set to the maximum value. In other words, if the converter breaker‘s rated breaking current capabilities are fixed to 20 kA and 16 kA for a strong nodes (onshore) and weak nodes (windfarms) respectively, the line breaker rated breaking current capability is set to 20 kA (maximum between 20 kA and 16 kA). This reasoning is then repeated for each line breaker.

Figure 27: Line breaker current capability estimation: Explanation of the worst case for the line breaker close to converter 1

Page 50: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

49

In non-selective converter breaker protection strategy, the breaker located at each AC/DC terminal will first interrupt the AC grid contribution to fault current. Then, when all AC contributions are interrupted, the faulty line can be isolated using a line breaker located at each line end. In this strategy, no DC reactor are implemented on DC lines. The only current rise limiting reactors are the converter reactors. At a consequence, the rated energy absorption for both converter and line breakers are defined by the converter reactor.

SUMMARY

Table 8 and Figure 28 summarizes the estimated input parameters of the DCCBs (DCCBs specification) for converter breaker protection strategy.

Table 8: Summary of Converter Breaker strategy DCCBs cost design

Parameter Converter Breakers (CB) Line Breakers (LB)

Rated breaking current capability (kA) Short Circuit Current (SCC) see 3.8.1.1

Maximum SCC see 3.8.1.1

Rated current limiting reactor (mH) 0 0

Rated energy absorption (MJ) 32 × 𝐿𝐿𝐵𝐵𝑜𝑜𝑜𝑜𝐶𝐶𝑚𝑚𝑔𝑔𝑡𝑡𝑚𝑚𝑔𝑔 × 𝐸𝐸𝐶𝐶𝐶𝐶2

32 × 𝐿𝐿𝐵𝐵𝑜𝑜𝑜𝑜𝐶𝐶𝑚𝑚𝑔𝑔𝑡𝑡𝑚𝑚𝑔𝑔 × 𝐸𝐸𝐶𝐶𝐶𝐶2

Open-close operation O-C-O O-C-O Rated transitent Interruption Voltage (TIV) (kV) 1.5 × 𝑈𝑈𝑜𝑜 1.5 × 𝑈𝑈𝑜𝑜

Rated short time withstand current (kA) Short Circuit Current (SCC) 𝑢𝑢 ×𝑈𝑈𝑜𝑜𝑍𝑍𝑐𝑐

Rated DC voltage (kV) 𝑈𝑈𝑜𝑜 𝑈𝑈𝑜𝑜 Rated DC current (kA) 𝑅𝑅𝑜𝑜𝑐𝑐 𝑅𝑅𝑜𝑜𝑖𝑖

Where,

𝐿𝐿𝐵𝐵𝑜𝑜𝑜𝑜𝐶𝐶𝑚𝑚𝑔𝑔𝑡𝑡𝑚𝑚𝑔𝑔: Converter inductor 𝑢𝑢: Number of lines connected to the DCCB 𝑈𝑈𝑜𝑜: Nominal voltage 𝑍𝑍𝑍𝑍: Characteristic impedance O-C-O: Open-Close-Open O: Open

Figure 28: Summary of Converter Breaker strategy DCCBs design

Page 51: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

50

3.8.1.2 PROTECTION DESIGN METHODOLOGY FOR CONVERTER WITH FAULT BLOCKING CAPABILITY STRATEGY

The methodology for estimating breaking current capability for non-selective protection strategy using converter with fault blocking capability is quite simple since converter with fault blocking capability can control the DC voltages and hence, reduce the AC fault current contributions. However, the line DCCBs should break some residual currents. In this deliverable, we assume that the DCCB can break the nominal current (around 2 kA). Such value has been defined within the scope of work of WP 4.3. As far as quite low breaking current capability is assumed, the energy to be dissipated is also quite low. Considering results from WP 4.3, the energy is assumed lower than 1 MJ.

In WP4.3 [3], several options for isolating the faulty line have been analysed. Using High Speed Switches (HSS) would lead to quite long times to restore active power through the DC grid. It is then proposed to use a low voltage DC breaker, with a rated breaking current capability about 2 KA (about the rated DC current) for the line isolation (typically with a TIV about 80 kV as the DC voltage can be controlled by the converter). This leads to a drastic reduction of the time to restore active power (divided by more than two). The HSS is then only used to break the low value residual current.

SUMMARY

Table 9 summarizes the estimated input parameters of the DCCBs (DCCBs specification) for converter with fault blocking capability protection strategy.

Table 9: Summary of converter with fault blocking capability strategy’s DCCBs design

Parameter Line Breakers (LB) Rated breaking current capability (kA) 2

Rated current limiting reactor (mH) 0

Rated energy absorption (MJ) < 1

Open-close operation O-C-O

Rated transitent Interruption Voltage (TIV) (kV) 1.5 × 𝑈𝑈𝑜𝑜

Rated short time withstand current (kA) 𝑢𝑢 ×𝑈𝑈𝑜𝑜𝑍𝑍𝑐𝑐

Rated DC voltage (kV) 𝑈𝑈𝑜𝑜

Rated DC current (kA) 𝑅𝑅𝑜𝑜𝑖𝑖

where,

𝐿𝐿𝐵𝐵𝑜𝑜𝑜𝑜𝐶𝐶𝑚𝑚𝑔𝑔𝑡𝑡𝑚𝑚𝑔𝑔: Converter inductor 𝑢𝑢: Number of lines connected to the DCCB 𝑈𝑈𝑜𝑜: Nominal voltage 𝑅𝑅𝑜𝑜𝑖𝑖: Nominal current 𝑍𝑍𝑍𝑍: Characteristic impedance O-C-O: Open-Close-Open O: Open

3.8.2 FULLY SELECTIVE PROTECTION DCCB COST

Both fully selective strategies, using mechanical DCCB and hybrid DCCB are explained in D 4.2 [1]. To isolate a DC line fault in primary sequence, one fault clearing unit (FCU) per pole is required at each line end. The FCU is composed of a fault limiting inductor, a high-speed switch (HSS) and a DC breaker. In addition, one FCU per pole at converter end is required to isolate the fault in backup sequence.

Page 52: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

51

The cost of HSS can be calculated in a straightforward manner whereas the ratings of the DC reactor and DC breaker are interdependent. The DC reactor is used to limit the rate of rise of fault current. The required minimum inductance value can be calculated as [24]:

𝐿𝐿𝐷𝐷𝐵𝐵 = 𝑉𝑉𝐷𝐷𝐵𝐵

(𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 − 𝑖𝑖0)/𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛 3-15

where 𝑉𝑉𝐷𝐷𝐵𝐵 is the DC voltage, 𝑖𝑖0 is the pre-fault current, 𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛 is the fault interruption time and 𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 is the peak fault current. The peak fault current is taken equal to the breaker rated breaking current capability (𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛). It will ensure that the fault current remains below 𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛 at time 𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛.

The energy stored in the DC reactor and from nearby converters and feeders must be absorbed by the surge arrester placed in the absorption branch of DC breaker. The required amount of energy absorption can be approximated as [24]:

𝐸𝐸𝑆𝑆𝑀𝑀 = 1.5 𝐿𝐿𝐷𝐷𝐵𝐵𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛2 3-16

3.8.2.1 PROTECTION DESIGN METHODOLOGY FOR FULLY SELECTIVE STRATEGY WITH HYBRID DCCB

The hybrid DCCB is assumed to be rated at 9 kA and its breaker operation time is taken as 2 ms [25]. Table 10 shows the proposed specifications for the DCCB.

Table 10: Summary of hybrid DCCB specifications

Parameter Converter Breakers (CB) Line Breakers (LB) Rated breaking current capability (kA) 9 9

Rated current limiting reactor (mH) 𝑉𝑉𝐷𝐷𝐵𝐵 × 10−6

(𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 − 𝑖𝑖0)/ 2𝑚𝑚𝑚𝑚 𝑉𝑉𝐷𝐷𝐵𝐵 × 10−6

(𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 − 𝑖𝑖0)/2𝑚𝑚𝑚𝑚

Rated energy absorption (MJ) 32 × 𝐿𝐿𝐷𝐷𝐵𝐵 × 𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛

2 32 × 𝐿𝐿𝐷𝐷𝐵𝐵 × 𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛

2

Open-close operation O-C-O O-C-O Rated transitent Interruption Voltage (TIV)

(kV) 1.5 × 𝑈𝑈𝑜𝑜 1.5 × 𝑈𝑈𝑜𝑜

Rated short time withstand current (kA) Short Circuit Current (SCC) 𝑢𝑢 ×𝑈𝑈𝑜𝑜𝑍𝑍𝑐𝑐

Rated DC voltage (kV) 𝑈𝑈𝑜𝑜 𝑈𝑈𝑜𝑜 Rated DC current (kA) 𝑅𝑅𝑜𝑜𝑐𝑐 𝑅𝑅𝑜𝑜𝑖𝑖

3.8.2.2 PROTECTION DESIGN METHODOLOGY FOR FULLY SELECTIVE STRATEGY WITH MECHANICAL DCCB

The mechanical DCCB is assumed to be rated at 16 kA and its breaker operation time is taken as 8 ms [26]. Table 11 shows the proposed specifications for the DCCB.

Page 53: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

52

Table 11: Summary of mechanical DCCB specifications

Parameter Converter Breakers (CB) Line Breakers (LB) Rated breaking current capability (kA) 16 16

Rated current limiting reactor (mH) 𝑉𝑉𝐷𝐷𝐵𝐵 × 10−6

(𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 − 𝑖𝑖0)/ 8𝑚𝑚𝑚𝑚 𝑉𝑉𝐷𝐷𝐵𝐵 × 10−6

(𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 − 𝑖𝑖0)/8𝑚𝑚𝑚𝑚

Rated energy absorption (MJ) 32 × 𝐿𝐿𝐷𝐷𝐵𝐵 × 𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛

2 32 × 𝐿𝐿𝐷𝐷𝐵𝐵 × 𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛

2

Open-close operation O-C-O O-C-O Rated transitent Interruption Voltage (TIV)

(kV) 1.5 × 𝑈𝑈𝑜𝑜 1.5 × 𝑈𝑈𝑜𝑜

Rated short time withstand current (kA) Short Circuit Current (SCC) 𝑢𝑢 ×𝑈𝑈𝑜𝑜𝑍𝑍𝑐𝑐

Rated DC voltage (kV) 𝑈𝑈𝑜𝑜 𝑈𝑈𝑜𝑜 Rated DC current (kA) 𝑅𝑅𝑜𝑜𝑐𝑐 𝑅𝑅𝑜𝑜𝑖𝑖

Page 54: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

53

4 DEVELOPED TOOLS

In this section, a brief description of the main tools which are developed within WP 4.5 are presented. The interaction between the tools required inputs and their outputs are presented in Figure 29. Four tools are developed:

1. Economic key performance (tool A): used for economic KPIs computation including CAPEX, maintenance, expected energy not transmitted EENT and losses presented in section 3.2. The tool is developed in the python environment.

2. Stochastic efficiency and protection reliability (tool B): used for time restoration and protection strategy reliability KPIs (performance of main/backups, see section 3.3.2). The tool is developed in the python environment.

3. Frequency stability (tool C): used to compute inertia and ROCOF KPIs (see sections 3.4.1 and 3.4.2). This tool is developed in the Matlab environment.

4. Redispatch (tool C) tools: used to compute redispatch costs KPI (see section 3.4.4). The tool uses two different modules which are developed in the Matlab and Julia environments.

Only the redispatch costs KPIs needs inputs from the stochastic efficiency and protection reliability tools (i.e. time restoration distributions and main/backup probabilities). The general framework of each tool is presented in the following sections.

Figure 29: Main developed tools with their respective inputs and outputs

A) ECONOMIC KEY PERFORMANCE INDICAROTS

The general tool framework used to compute the economic key performance indicators (see 3.2) is presented in Figure 30. The tool needs three main inputs and includes five main modules:

• Grid topology (inputs 1): at this phase, the DC grid node (i.e. converter) connectivity (or grid topology) and physical component parameters (line resistances/lengths, line and converter rating powers, nominal voltage) are populated from a graphical user interface (GUI). This information is managed by associated attributes (parameters).

• Load/generation profile (input 2): the load/generation profile may be of two different natures (options): the user could specify either a parametric model or a time series of set points of load/generation profiles (hourly generation/load profiles). It should be noted that in the scope of the current deliverable, only wind Weibull distribution parameters are possible for the first option. Regarding the second option, the time series (set points of loads/generation) are provided in an Excel or CSV file. Moreover, the nature of a node (i.e. converters or inverters) is stored on a given attribute and is initially provided in “input 1”

Page 55: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

54

phase from the GUI. The converter operating point (inverter or converter) is dynamically managed during the simulation depending on the generation/load profile.

• Reliability data (inputs 3): at this phase, the mean time to repair MTTR and mean time to failure MTTF are provided by the user. This is done at the same time as the grid topology input phase by storing these parameters in a dedicated attributes.

• Protection strategy layout (module 0): this module has a duty of defining the protection components (DCCBs, switches, etc.) and their location in the grid. For example, setting the protection strategy scheme to fully selective with FS-FDCCB and the type of used DCCBs (fast or slow), this module automatically adds the protection strategy components to the grid.

• Protection strategy specification (module 1): this module has as a duty of specifying the DCCBs parameters (Current breaking capability, energy absorption, short time withstand current, etc., see section 3.8).

• Grid configuration (module 2): this module has as a duty to generate, based on reliability data, a set of credible contingencies, using a Monte Carlo (MC) sampling approach. These last could be either the result of grid component failures (line, converters) or protection scheme related assumptions (see section 3.2.3).

• Load/generation profile (module 3): this module has as a duty to generate representative load/generation profiles, using a Monte Carlo (MC) sampling approach. In case where the generation/load profile is provided in an Excel (or CSV) file, this module simply screens the file (if the number of profiles is relatively small) or generate set points from this file (if the computing time does not permit to screen the Excel file).

• DC optimal power flow DC OPF (module 4): based on the grid configuration and load/generation profile, this module performs an optimal power flow (regarding given criteria such as evacuating maximum of wind power). Line power flows, currents and node voltages are the main outputs of this module.

The general functioning of the economic KPIs tool consists first, to provide the different inputs, secondly to generate credible contingencies, and finally to generate representative load/generation profiles and performing the DC OPF until a level of representative load profile is reached (see Figure 30). From this process, two main KPIs are then derived: expected energy not transmitted EENT and grid losses (which are displayed and stored in an Excel files). For CAPEX and maintenance calculations, module 2, 3 and 4 in Figure 30 are bypassed.

Page 56: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

55

Figure 30: Economic key performance indicators framework modules

B) STOCHASTIC EFFICIENCY AND RELIABILITY KEY PERFORMANCE INDICATORS

The general tool framework used to compute the stochastic efficiency and reliability key performance indicators (see section 3.3.2) is presented in Figure 31. In addition to “inputs 1”, “inputs 3” and the protection strategy layout module presented above, this tool receives as an input the protection strategy timeline (“inputs 4”, main sequence, backup sequences temporally ordered) and uses two main modules:

• Petri net builder (module 5): this module uses “inputs 1”, “inputs 3”, “inputs 4” and the protection strategy layout (module 0) to automatically build a stochastic petri net (i.e. builds the petri net transitions and places with associated time transitions). It should be noted that this module is interfaced with TINA open software [27]. In other words, the “module 5” can use a TINA model as a petri net input without using the grid inputs and reliability data inputs (provided in Figure 30) or timeline inputs (inputs 4 provided in Figure 31). In this case, the petri net used in the Monte-Carlo simulations is directly build from TINA software. Moreover, the “module5” also automatically saves a TINA filename if the petri net is built from this module. In this case, a simulation can also be made on TINA software. Finally, the petri net file can also be written in a text file without using TINA or the petri net builder “module 5”.

• Monte Carlo Petri Net MCPN (module 6): this module manipulates the generated (or given in a text file) petri net in order to make stochastic simulations on different initial conditions.

To conclude, the general functioning of this tool consists of building a petri net (from different inputs) in order to take into account the time domain aspect of the main and backup protection strategy sequences and then take into account the stochastic aspect by modifying the initial conditions of the petri nets (place initial marking). The main outputs are the efficiency KPIs (time distributions) and the protection strategy reliability performance (probability to be in main, in backup or in ACCB backup). These are directly displayed by the tools and stored in Excel files.

Page 57: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

56

Figure 31: Stochastic efficiency and reliability key performance indicators framework modules

C) FREQUENCY STABILITY AND RESERVE SIZING CALCULATIONS

Two different MATLAB tools are developed to analyse the impact of DC grid protection on the AC system, namely the reserve sizing tool and the redispatch cost calculations tool.

C1) FREQUENCY STABILITY AND RESERVE SIZING CALCULATIONS

Figure 32 represents the details of the reserve sizing tool. The inputs are the hourly generation data and converter set points for each planning year. The generation data is segregated in different types (thermal, nuclear, renewable etc.) and the equivalent inertia of the grid is calculated depending on the typical inertial constant of each generation type. The converter set points provide details on the wind power injection in the onshore grid at different nodes. In order to analyse the influence of a particular DC grid topology on a specific synchronous area, the power injection at each interconnection node is required. The frequency stability KPIs that include inertia and ROCOF are calculated using the obtained data (Figure 32 (a)).

Hourly Generation data (csv file)

Hourly converter setpoints data (csv

file)

Planning year (i)

Total generation in a synchronous area

(Ptotal)

Extract wind power injection

(P_OWF_nodes)

Output: Heq/hour, ROCOF/hour, Ptotal, P_OWF_nodes

1

i = i+1

Set FFR capacity (start with 0)

Monte Carlo samples for hourly scenario

Simulink Model for Nordic grid

Load shedding duration

Next samplePower deviation curve

Output : FFR capacity

FFR = FFR + 100 MW

Accetable limit?

1

ΔTHVDC

Yes

Page 58: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

57

(a) (b) Figure 32: MATLAB modules for (a) frequency quality defining KPIs (b) reserve size calculations

These outputs are saved (indicated by 1) and used for the reserve sizing mechanism shown in Figure 32(b). The methodology sets the FFR capacity to zero at the first iteration to check if there is even the need of any FFR reserves. The first iteration also calculated the amount of load shedding duration (a KPI) without any FFR.

In each iteration, for a fixed FFR capacity, the algorithm iterates 500 generation scenarios (can be changed). Depending on the current sample, the values of inertia (Heq), offshore power injection (P_OFF_nodes) and total generation (Ptotal) are used that were obtained from module 1 (on the left). At this point, the power deviation curve is prepared depending on the protection scheme under study. For a selective scheme, the deviation is assumed to be equal to the disconnection of one of the nodes from DC grid. For a non-selective scheme, the deviation is set equal to the power of all the injection points from a common DC grid. The input of power restoration time (∆THVDC) is taken as per the protection scheme under study.

The power deviation curve is input to the Simulink model for frequency response. The minimum frequency is obtained and if less than 49 Hz, it is noted as a load shedding instant. The loop is repeated with incremental FFR capacity until an acceptable load shedding duration is obtained. The corresponding FFR capacity is taken to the output. Both modules are simulated for the whole planning horizon, determining FFR capacity for the planning year.

C2) REDISPATCH COST

The redispatch cost algorithm uses two different modules, one developed in MATLAB and another developed in Julia. MATACDC [28], a MATLAB based tool, is used to solve the power flow problem of AC/DC network whereas PowerModels ACDC [29], a Julia based tool, is used to perform an optimal power flow run of an AC/DC network. Hence, the MATLAB module computes the violations and Julia module computes the dispatch cost.

Page 59: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

58

Wind scenario (i)

Power flow of AC/DC grid using optimal

setpoints

Remove DC branch (k)

Check voltage, current and frequency

violation

Store the elements to be isolated

2

k = k+1

Prefault optimal dispatch cost

Disconnect the elements2

Post fault optimal distpatch cost

Redispatch cost

MATLAB module Julia module Figure 33: Redispatch cost calculation algorithm

The algorithm begins with a power flow run using the optimal active power and voltage set-points at the generator buses. The initial power flow computes the pre-fault power exchange between onshore AC and offshore DC grids. Then, a contingency is created on each DC branch in sequential simulation runs. A second power flow run checks for constraint violations of voltage and thermal limits in the AC grid after isolating the DC fault. The Simulink based frequency response model is used to determine the frequency violations. The elements (branch number, bus number or the frequency violation) are stored. The loop is run for a number of wind scenarios, e.g. 500 samples per year. For each sample, a fault is simulated on every DC line (k).

The Julia based module (on the right) uses the same wind samples as the MATLAB module. This module determines the redispatch cost based on the difference between pre- and post-fault network. The pre-fault network is the same as the MATLAB module (without DC contingency). The post fault network is prepared after isolating the elements that had violations in the MATLAB module. The calculated redispatch cost is weighed according to the fault probability.

Page 60: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

59

5 APPLICATION TO WP 4.2 BENCHMARK SYSTEM

5.1 OBJECTIVE

This section aims to show how the proposed methodology in section 3 could be applied and validated. It is important to highlight that it is not intended to classify the different protection strategies regarding their costs or performance. So, this is only a use case application based on input assumptions. Indeed, depending on these assumptions (DDCB reliability data, assumptions on EENT calculation, energy monetization, etc.) the conclusions could be different. The main assumptions and data used in this section are summarized in section 5.2. The key performance indicators computation including economic indicators, operational risk indicators and AC system impact are presented in sections 5.4, 5.5 and 5.6 respectively.

5.2 DATA/ASSUMPTIONS

Today it seems difficult to have relevant values of failure rates for some critical components such as DCCBs. In order to deal with this lack of input data, a comparison approach is developed to deal with this issue. The proposal is based on the two following main ideas:

• To carry out a relative assessment and comparison (not considering absolute value of indicators)

• To have as far as possible a “Technology Neutral“ approach: To consider the same failure rates for all DCCB technologies. This point could be discussed as very different breaker technologies are developed (e.g. hybrid, mechanical). Having no real information about breaker reliability, any extrapolation such as considering close technology devices (e.g. VSC converter valves for hybrid breaker) could be possible but should be considered with caution as the operating conditions can be quite different. An alternative approach, which is here chosen, is to deploy a sensitivity analysis considering different values for breaker reliability parameters.

The main idea is to carry out analyses with several values of DCCB failure rates. Applying such approach enables the assessment of the robustness of the protection strategies regarding failure rate variation, with two possible situations:

• If the comparison ranking is not significantly modified by failure rate values: Conclusions can be drawn about the impact of protection scheme architecture and associated sequences (main, back-ups)

• If the comparison ranking is significantly affected by failure rate values: It will be more difficult to conclude. It may however provide clues about the suitability of breakers with certain failure rates to a particular protection scheme

Some preliminary assumptions are made which will be applied to all case studies (for all protection strategies analysis):

• All measurement (voltage, current sensors) and control (relays) devices are considered as fully reliable (zero failure rate)

• All related control and protection algorithms are considered as fully reliable (sufficiently redundant control systems and algorithms are supposed)

• The blocking state operation is considered as fully reliable for both half bridge (HB) and full bridge (FB) MMC. Furthermore, as the current limiting state operation is based on control action, it is considered as fully reliable for FB MMC. As a consequence, a zero failure rate is considered for FB MMC in operational risk and AC system impact KPIs studies (section 5.5 and 5.6).

• All passive components (e.g. DC reactor, capacitance, etc.) are considered as fully reliable (zero failure rate)

Page 61: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

60

• Considered non zero failure rate components are:

• AC circuit breakers (AC CB)

• High speed switches (HSS)

• DC circuit breakers (DCCB)

• Cables, converters for the study of grid system availability with computation of Expected Energy Not Transmitted (EENT) (see appendix A.1 for both converter and cable reliability data).

AC CIRCUIT BREAKERS

Unavailability parameters for AC CBs are based on parameters found in the literature, and are shown in Table 12. Regarding the unavailability (p.u.) from Table 12, the following value is selected for all studies and analysis:

Unavailability: ζACCB = 0.0003.

Table 12: AC CB failure rate and unavailability parameters from literature review

FAILURE RATE (/YEAR)

REPEAR TIME (HOURS)

UNAVAILABILITY (HOURS) UNAVAILABILITY REFERENCE

0,01 200 2 0,000228 [30]

0,0121 200 2,42 0,000276 [31]

0,0281 312 8,7672 0,001 [31]

0,003 200 0,6 6,845E-05 [32]

0,01654 200 3,3083 0,000377 [33]

HIGH SPEED SWITCHES

The HSS unavailability parameter is assumed to be similar to the AC breaker one. This assumption seems natural as HSS are likely to be based on similar technology as AC breakers.

Unavailability: ζHSS = 0.0003.

DC BREAKERS

Regarding DC breakers, the technical literature gives only very little information and vendors generally reply that reliability is a design criterion (which comes at a cost). As a consequence, it is proposed to set the DCCB unavailability parameters for DCCBs from the value selected for AC CBs seen as a reference value. Three cases will then be analysed as proposed in Table 13, to assess the protection scheme within a quite large range of variation of DCCB unavailability parameters. Note that the values from Table 13 are in the same order of magnitude as some of the values that can be found in the technical literature, given here for information in Table 14, without any more information on the technology of DCCBs.

Page 62: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

61

Table 13: Selected DCCB unavailability parameters

CASE LOWER THAN AC CB SIMILAR TO AC CB HIGHER THAN AC CB

Unavailability parameter ζDCCB = ζACCB/2 = 0.00015 ζDCCB = ζACCB = 0.0003 ζDCCB = ζACCB*3 = 0.0009

Table 14: DCCB failure rate and unavailability parameters from literature review

FAILURE RATE (/YEAR)

REPEAR TIME (HOURS)

UNAVAILABILITY (HOURS)

UNAVAILABILITY ζ REFERENCE

0,015 192 2,88 0,00033 [34]

0,0091 192 1,7472 0,0002 [34]

0,045 192 8,64 0,00099 [34]

5.3 WP 4.3 BENCHMARK

The topologies considered within WP 4.3 for the benchmark networks are classified as “small, medium and large impact”-benchmark networks [35]. In this section, only small impact networks will be considered. The basic topology for this grid is shown in Figure 34. Furthermore, only subsea cables will be considered, pole to ground PtG faults in bipolar configuration are simulated for non-selective protection strategies, whereas pole to pole PtP faults in monopolar configuration are simulated for full-selective protection strategies. Converter and cable rating powers are set to 1200 MW except for the cable connecting “converter 1” to “converter 4” which is set to 600 MW.

Four protection strategies are considered in this deliverable (described in A.4 and [1] [3]):

• Non selective protection strategies: with breaker at converter output (NS-CB) and using converter with fault blocking capability (NS-FB).

• Fully selective protection strategies: using fast DCCBs (FS-FDCCB) and using slow DCCBs (FS-SDCCB).

Figure 34: Benchmark network for small impact system (adopted from WP2 and WP3, see [1] [3])

5.4 ECONOMIC KEY PERFORMANCE INDICATORS

Table 15 presents the DCCBs specifications used for Full-Selective strategies (with mechanical FS-SDCCB and hybrid DCCBs FS-FDCCB) and Non-Selective protection strategies (Full-Bridge and Converter Breaker). This table is derived by applying the methodology described in section 3.8.

Page 63: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

62

The associated unit DCCB capital costs (CAPEX) are presented in Figure 35 (a). It should be noticed that the blue part (for extra costs part) corresponds to the additional platform costs, so only offshore DCCBs include these extra costs. This figure shows that:

• Offshore extra costs (blue colour) associated to the DCCBs are quite high. These extra cost are mainly related to the volume and weight of the DC CBs, including DC reactor and surge arresters (energy dissipation) (see [7]). The volume and the weight of the DC reactor and the surge arresters are quite significant and thus higher is the DC reactor and associated energy to be dissipated, higher is the extra cost. It is why, FS-SDCCB protection strategy, which uses large DC reactors, exhibits higher extra cost.

• The CAPEX of DCCBs used for FS-FDCCB strategy is twice the CAPEX of DCCB used for NS-CB protection strategy. This is due to the important number of IGBTs used for hybrid DCCBs (i.e. material costs).

• The CAPEX of DCCBs of the NS-FB protection strategy is significantly smaller compared to the other. In fact, in the full bridge protection strategy an 80 kV TIV DCCB with current breaking capability of 2 kA is sufficient to isolate the faulty line (see section 3.8.1.2 and [3]).

• It can be observed from Figure 35 (b) that the range of the total CAPEX of DCCBs for the different protection systems varies approximately by a factor of 2.3. For the NS-FB protection strategy, as stated in section 3.2.1, the additional CAPEX of the FB converter (regarding a half-bridge converter) is considered as part of the protection system. That leads to a quite high CAPEX for the NS-FB (between the two fully selective protection systems). The large variation of the CAPEX makes important to be able to relate those costs to performance in order to get a full decision criteria for selecting the most appropriate protection system.

Table 15: DCCBs specification for cost calculation (used in section 5).

Input parameter Unit

FS - FDCCB FS-SDCCB NS- FB NS-CB

Technology Hybrid Mechanical Mechanical Mechanical

Rated DC current kA 2 2 2 2

Rated Breaking current capability kA 9 16 2 20

Rated DC voltage kV 320 320 50 320

Rated transitent Interruption Voltage (TIV) p.u 1.5 1.5 1.5 1.5

Rated energy absorption MJ 10 50 1 1

Breaker opening time at maximum DC breaking current ms 2 8 8 8

Current limiting DC reactor mH 125 200 0 0

Open-close operation - OCO OCO OCO OCO

Directionality - Bi-directional Bi-directional Bi-directional Bi-directional

Rated short time withstand current kA 9 16

Bc1, Bc2, Bc3: 20 Br1, Br3, Br4: 45 Br5, Br6, Br7, Br8: 30 Br2: 20

Page 64: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

63

Figure 35: Capital costs breakdown of different used DCCBs within WP4.2 benchmark.

a) Unit DCCB costs, b) Protection strategies total costs and full bridge additional CAPEX.

Figure 36 shows the Capital Expenditures indicators (CAPEX) and maintenance indicators of the Figure 34 grid benchmark including the protection strategy and grid components (converter transformers, cables and platforms). It shows that the relative proportion of the protection CAPEX is quite low compared to the total grid CAPEX (2.8%, 4.2% 5.4% and 5.8% for NS-CB, FS-SDCCB, NS-FB and FS-FDCCB respectively). As expected, the converter breaker NS-CB protection strategy is the most cost effective one and the FS-FDCCB is the more expensive one.

Figure 36: Total costs for fully selective (NS-FDCCBS, FS-SDCCB) and non-selective (NS-CB, NS-FB) protection strategies

a) Capital Expenditures (CAPEX) Indicator, b) Maintenance Indicator.

Figure 37, Figure 38 and Figure 39 respectively show the expected energy not transmitted (EENT), the influence of protection strategy on EENT and grid losses. The following conclusions can be drawn from these figures:

• The expected energy not transmitted seems to be quite similar for all protection strategy options. The values are between 2.1% and 2.3% of the total generation capacity (2.4 GW). These EENT values are

Page 65: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

64

mostly due to both cable and converter unavailability’s, as shown in Figure 38, where the contribution of protection system components to total EENT is depicted.

• The contributions of protection system components to total EENT never exceeds 10%.

• Non selective protection strategy NS-CB incurs slightly more EENT compared to FS-SDCCB, FS-FDCCB and NS-FB. This is due to the fact that more DCCBs are required during the main sequence in case of NS-CB (assumption taken in section 3.2.3). However, the difference is still very low (around 0.1%).

All protection strategies exhibit negligible losses associated to DCCBs. However, for protection NS-FB, full bridge converters additional losses have to be considered (around 30% of converter additional losses).

Figure 37: Expected Energy Not Transmitted (% of 2.4 GW).

Figure 38: Part of Grid components (converters transformers and cables), protection components (DCCBs and HSS) on the Expected Energy Not Transmitted (EENT).

a) FS-FDCCB, b) FS-SDCCB, c) NS-CB, d) NS-FB

Page 66: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

65

Figure 39: Energy losses in the WP4.3 Promotion benchmark system for different protection strategies (% of 2.4 GW).

5.5 OPERATIONAL RISK INDICATORS

In this section, the key performance indicators considered for risk analysis are applied for the case study presented in section 5.3. The focus is on stochastic efficiency and protection scheme reliability indicators. Regarding efficiency indicators supported by EMT simulation, readers can refer to [3].

5.5.1 EFFICIENCY KEY PERFORMANCE INDICATORS (SUPPORTED BY EMT SIMULATION)

As indicated before, EMT simulations on benchmarks presented in section 5.3 are already performed within WP 4.3. So, the objective here is not to give a detail about KPIs computed from EMT simulations but only to remind these results. For more details about different efficiency key performance indicators related to the benchmark, readers can refer to [3]. In this document, in-depth studies of several protection strategies taking into account different aspects (configuration technology, different fault locations, different physical grid parameters, etc.) have been carried out. One of the objectives was to identify and assess efficiency KPIs (including fault interruption time, voltage restoration time, active and reactive powesr restoration times as well as energy imbalance) for different sets of protection strategies.

Table 16 summarizes the results in [3] for fully selective FS-FDCCB, FS-SDCCB and non-selective NS-CB, NS-FB protection strategies. It should be noted that within WP 4.3, different converter controls are used for these strategies. Consequently, the active power restoration times and transient energy imbalance KPIs in Table 16 should be interpreted and appreciated with caution.

The main conclusions which could be drawn from Table 16 are the following:

• Fault interruption times seem to be quite low for all protection strategies, with however higher values for non-selective protection strategies (NS-CB and NS-FB).

• Active power restoration time for FS-SDCCB and FS-FDCCB, and following energy imbalance, are quite high. This is mainly due to converter control performance which has been limited in order to deal with stability issues due to the presence of DC reactors. With a more appropriate control, those times would be reduced, as it is the case for the large size grid application presented in section 6.

• Non selective protection strategies exhibit quite similar active power and voltage restoration times.

• As far as full bridge converters remain operational during all the protection process, reactive power support to AC system can be maintained with NS-FB protection strategy.

Page 67: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

66

Table 16: Efficiency key performance indicators for fully selective FS-FDCCD/FS-SDCCB and non-selective NS-FB/NS-CB supported by EMT simulations

KPI FS-FDCCB FS-SDCCB NS-FB NS-CB

Fault interruption time (ms) 10.3 18.5 26 29.5 Voltage restoration time (ms) 42 84.7 102 73.7 Active power restoration time (ms) 218.6 507.3 134 150.0 Reactive power restoration time (ms) 65.1 188.6 0 47.1

Transient energy imbalance (MJ) 233 543.64 -143 -94.4

5.5.2 STOCHASTIC EFFICIENCY AND PROTECTION SCHEME RELIABILITY INDICATORS

The different protection strategies are simulated with the MCPN model. Simulated faults are faults located on lines of the base reference grid (faults located on busbars are not considered here). 50000 simulations are carried out for each strategy and parameters set, in order to be able to extract both sequence and time duration distributions.

The main parameters used in simulation results are summarized in Table 17. “Time response” corresponds to the time reaction of DCCBs, ACCBs, Full Bridge MMC and HSS devices (e.g. DCCBs opening time) where “Failure time” corresponds to the necessary time to detect if these devices do not react and hence, pass to the next step of the protection strategy (backup sequences). Furthermore, the time values given in Figure 40 are deduced from the timeline of each protection strategy’s benchmark, corresponding simulation results and from discussions with work package partners. DCCBs technologies for FS-FDCCB, FS-SDCCB, NS-CB and NS-FB protection strategies are conventionally (within WP4.2) classified regarding their time response (opening/closing time) and failure. Figure 40 shows such timeline.

In the following Monte-Carlo results, the time response and component failure time are randomly drawn between corresponding minimal and maximal values.

Figure 40: Test case: DC grid connected to the Nordic grid

Table 17: Inputs used in Monte-Carlo simulation

COMPONENT TIME RESPONSE (MS) FAILURE TIME (MS) UNAVAILABILITY ζ

DCCBs FS-FDCCB: [2,3],

FS-SDCCB: [7,10], NS-CB and NS-FB: [13,17],

FS-FDCCB: [5,7], FS-SDCCB: [12,14],

NS-CB and NS-FB: [18,20],

For all DCCBs: 0.00015, 0.0003, 0.0009,

ACCBs [40,60] [60,70] 0.0003

Full bridge [0,0] [0,0] 0

Simulation results are here given for the case of a DCCB unavailability of ζDCCB = 0.0003. The cases with lower and higher values of ζDCCB are presented in appendix A.7. The values of each indicator for all strategies are given in Table 18.

Page 68: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

67

Table 18: Strategies indicators for ζDCCBs =0,0003

STRATEGY IMAIN (%) IBACKUP (%) ITR_20 (%) ITR_50 (%) ITR_200 (%)

Selective strategies

FS-FDCCB 0.060 0.059 0.059 0.059 0.059

FS-SDCCB 0.060 0.059 0.117 0.059 0.059

Non-Selective strategies

NS-FB 0.060 0.059 100 0.059 0.059

NS-CB 0.060 0.089 87.50 0.089 0.089

From indicators IMAIN, IBACKUP, ITR_20, ITR_50 and ITR_200 (see section 3.3.1), a benchmarking table have been completed as proposed in section 3.3.1.

Table 19: Benchmarking indicators for ζDCCBs =0,0003

STRATEGY “SPEED” OF THE PROTECTION STRATEGY CLASSIFICATION

PERFORMANCE OF THE MAIN SEQUENCE CLASSIFICATION

PERFORMANCE OF THE BACK-UP SEQUENCES CLASSIFICATION

AC/DC STABILITY ISSUE RISK INDICATOR

CLASSIFICATION

Selective strategies

FS-FDCCB Very fast + + ++

FS-SDCCB Very fast + + ++

Non-Selective strategies

NS-FB fast + + ++

NS-CB fast + - ++

From benchmarking Table 19, selective protection strategies including Full-Selective with mechanical DCCB (FS-SDCCB) and Hybrid DCCBs (FS-FDCCB) are ranked as “very fast”. Non-selective protection strategies including Full-Bridge (NS-FB) and Converter Breaker (NS-CB) are ranked as “medium fast”. From main and backup performance point of view, all protection strategies have a similar performance (with “+” ranking, i.e. performant). The reliability protection scheme indicators including IMAIN , IBACKUP and ACCB backup (protection failure) are presented in appendix A.7.

Selective (FS-FDCCB, and FS-SDCCB) and Converter Breaker (NS-CB) protection strategies have quite a similar level of risk related to the AC/DC stability issue (with “++” ranking, i.e. high performance). On the other hand, the full Bridge (NS-FB) protection strategy requires a larger time duration to recover and be ready for operation (fault current suppression) which is ranked “+”.

5.6 AC SYSTEM IMPACT INDICATORS

The AC system impact indicators are calculated on two different example AC grids. The KPIs of frequency stability and reserve requirement are calculated using the Nordic grid model as onshore grid since a frequency response model is available for this grid. However, the redispatch cost indicator is calculated on the IEEE 118 bus system as onshore system due to availability of the line flow and node voltage limits.

5.6.1 FREQUENCY STABILITY

The methodology is demonstrated using the DC grid example from WP4. The DC grid is assumed be connected to two onshore points within the Nordic grid as shown in Figure 41. The original windfarm ratings are 1200 MW which is increased to 4000 MW (2x2000 MW per pole in bipolar grid) here to demonstrate the methodology. The original rating does not result in any frequency violations.

Page 69: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

68

Figure 41: Test case: DC grid connected to the Nordic grid

The first step of the proposed methodology requires hourly generation samples of the Nordic grid to calculate the inertia of the system, as well as hourly samples of offshore wind power generation to calculate the power deviation during a DC grid contingency. An example generation scenario from year 2045 is used here (from WP 12). In total, 500 wind samples are used to represent yearly generation scenario. The average inertia is calculated as 1.81 sec as shown in Table 21.

The ROCOF value is calculated based on the average inertia and average total generation in the grid. For the selective case, the ROCOF value is calculated when one line is disconnected whereas for nonselective cases, it is calculated when the whole DC grid is disconnected. Note that we do not calculate hourly ROCOF values because the hourly ROCOF is observed to reach to very high values in case inertia is near zero (since ROCOF = ΔP/ 2H), and taking average of hourly values do not provide clear comparison between selective and nonselective schemes.

The minimum frequency violation is calculated based on the current FCR reserves in the grid, i.e. 1450 MW. The power restoration time for different protection strategies from WP 4 are used (see Table 20). The N-1 DC contingency analysis (L-G faults) is performed for the fault on each DC cable. Except for the fault on line L1, the power (injection in AC grid) is assumed to be restored to its pre-fault injection, with sufficient capacity of L3. For the fault on line L1, the power injection is assumed to be restored to the wind farm-2 generation. It was observed that for a very high renewable penetration (>90%), the inertia value drops near to zero, causing ROCOF to reach a very high value (in thousands). In such scenario, no amount of reserves would be sufficient to arrest the frequency drop. Therefore, a minimum inertia (M) of 0.5 sec is assumed to be presented in the AC grid. Such inertia can be maintained using multiple measures [21]. As seen in Table 21, the frequency violation duration varies from 16 hr/year to 36 hr/yr, depending on the power restoration time. Note that the power restoration time for NS-FB is the least, hence it has the least duration of frequency violation duration.

Next, the FFR size is calculated based on the proposed methodology. The acceptable amount of load shedding is set to 3 hr/year. The FFR is assumed to have a high ramping rate (i.e. 1000 MW/s) and instant activation (i.e. 0 sec). This will ensure that the FFR releases its full capacity for any assumed size before the frequency nadir is reached. Figure 42 (a) and (b) show the load-shedding duration for selective and nonselective strategies, as the reserve sizing methodology iterates through incremental size of the reserve. The selective strategy using fast DCCB requires less amount of reserves (i.e. ~1125 MW) than the selective strategy using slow DCCB (i.e. ~1260 MW). Similarly, the nonselective strategy with higher restoration time (NS-CB) requires a higher amount of reserves than the NS-FB. In general, non-selective strategies require a higher amount of reserves than the selective strategies, even in case they are faster than the selective strategy.

Page 70: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

69

(a)

(b) Figure 42: Load shedding duration vs FFR capacity for WP 4.3 topology (a) selective strategies (b) nonselective strategies

Table 20: Power restoration time of different protection strategies

Protection strategy Primary

Fully selective fault clearing with FDCCB 124 ms

Fully selective fault clearing with SDCCB 382 ms

Non-selective fault clearing with full-bridge MMC 106 ms

Non-selective fault clearing using converter breaker 150 ms

Page 71: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

70

Table 21: Frequency stability indicators and reserve requirement

INDICATOR FS - FDCCB FS-SDCCB NS- FB NS-CB

Heq,avg (sec) 1.81 (not affected by protection schemes)

ROCOF (Hz/s) 0.6 0.6 1.21 1.21

Fmin, duration (hr/year) 24 34 16 36

FFR requirement (MW) 1125 1260 2100 2400

Next, a generalized test system is introduced to analyse the influence of two main factors: higher offshore wind power injection and the number of connection points between AC and DC grids. As the injected offshore power increases, the temporary loss of infeed will have a bigger impact on the connected synchronous area. On the other hand, as the number of connection points increases, there should be a bigger difference between FS and NS strategies. Figure 43 shows the generic networks used for the investigation. To investigate the influence of the injection capacity, the offshore dc network is assumed to have 2 DC lines (of 4000 MW bipolar link, temporary loss of 2000 MW for L-G fault) connected to the Nordic grid (see Figure 43 (a)). This is a minimal model to differentiate between FS and NS protection strategies (that result temporary loss of 1 and 2 lines respectively) while analysing the influence of offshore wind power injection. The influence of number of connections is analysed using model shown in Figure 43 (b) where the offshore dc grid connections to the Nordic grid are varied between 1 to 4 (of 4000 MW bipolar links). A FS protection strategy will result in temporary loss of one line whereas a NS protection strategy will result in the temporary loss of the whole dc grid. The activation time of reserves is taken as 0.1 sec compared to 0 sec in previous case. The nonzero value helps in analysing the impact of both speed and capacity of reserves. The time is estimated based on a preliminary analysis of the time to reach nadir (49 Hz) of all generation samples. In order to provide a generic overview and to avoid EMT analysis, rather than taking a fixed power restoration time, it is taken as upper and lower limit of the restoration times for both types of protection strategies from previous case study. Four combinations are created as:

• FS with ∆THVDC = 100 ms • FS with ∆THVDC = 400 ms • NS with ∆THVDC = 100 ms • NS with ∆THVDC = 400 ms

Each line is considered to be 100 km long and unavailability is calculated based on the parameters provided in previous section. The unavailability of the cables within the dc offshore grid is not considered as the topology of the network is unknown. The impact of NS strategies is expected to be higher as the total length of dc network increases since any fault in the dc network will result in the disconnection of whole offshore grid.

(a) (b)

Figure 43 Generic grid example to analyze (a) the influence of the injection capcity (b) the influence of the no. of connection points.

Page 72: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

71

INFLUENCE OF THE INJECTION CAPACITY

Figure 44 shows the influence of increasing injection capacity on FFR requirement. The FS strategy with both power restoration times, 100 ms and 400 ms, results in the same FFR requirement (Figure 44 (a)) whereas the requirement is different for NS strategies with high and low power restoration times (Figure 44 (b) and Figure 44 (c)). For the 2 GW injection, the FFR requirement is nearly the same for all protection strategies. At higher power injections, a higher load shedding is observed for NS strategies in general. The NS strategy with 400 ms power restoration time has the most load shedding. An acceptable load shedding is assumed to be a few hours (less than 3 hours/year here) rather than zero, a criterion similar to LOLE (loss of load expectation) defined for generation adequacy in different synchronous areas. The FS strategy can meet the load shedding criterion till 4 GW injection whereas NS strategies can meet the criterion only in case of 2 GW injection (see red markers). This is due to the fact that the frequency drops faster for NS strategies, needing faster reserves. The activation time of 0.1 s is not sufficient to constraint the load shedding to 3 hours for this case.

(a)

(b)

Page 73: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

72

(c)

Figure 44 FFR requirement for different injection capacities for (a) FS, ∆THVDC = 100 ms & 400 ms (b) NS, ∆THVDC = 100 ms (c) NS, ∆THVDC = 400 ms

INFLUENCE OF THE NUMBER OF CONNECTIONS

If there is only one connection between an ac and a dc grid, there is no difference between FS and NS strategies with similar power restoration time (Figure 45). Also, there is no difference between two non-selective strategies with high and low power restoration times (Figure 45 (b) and Figure 45 (c)); hence the influence of the restoration time is negligible at lower number of connections. However, as the number of connections increases, higher restoration time leads to increased load shedding. The fully selective strategy will have same load shedding, irrespective of the number of connections. Note that in this case, the power deviation is realized only at the converter connected to the faulted line. We do not observe any difference between the restoration time of 100 ms and 400 ms within FS strategy. Up to 3 connections, the NS protection strategies can still be used with a certain amount of reserves if permissible load shedding is up to 3 hours.

(a)

Page 74: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

73

(b)

(c)

Figure 45 FFR requirement for different number of connections for (a) FS, ∆THVDC = 100 ms & 400 ms (b) NS, ∆THVDC = 100 ms (c) NS, ∆THVDC = 400 ms

In summary, the impact of two variables, namely offshore power injection and the number of AC/DC grid connections is analysed. Both FS protection strategies, with restoration times 100 ms and 400 ms, perform better than NS strategies. The NS strategy with higher restoration time has higher load shedding and the reserve requirements. For higher number of connections, if a permissible load shedding is limited 3 hours, NS can still be used for 2 or 3 connections, whereas zero load shedding cannot be reached even with very high FFR. This can also be witnessed for the FS strategy with 4 GW injection, where zero load shedding cannot be achieved but can be limited to 3 hours. It is not possible to reach zero load shedding for the aforementioned cases without reducing FFR activation time further or employing the additional measures e.g. keep inertia to a minimum value. For a larger DC grid (with higher no. of connections to AC grid) or with high impact DC grid (more injection to AC grid), a FS strategy can be a better choice (grid splitting can be an option here). However, for smaller or low impact DC grids, the NS protection strategies can still achieve the required load shedding duration.

5.6.2 REDISPATCH COST

The performance of the proposed methodology (in section 3.4.4) is demonstrated using a test case. The offshore DC grid model for the test case is shown in Figure 46. It connects two offshore wind farms to onshore AC grids using a meshed bipolar DC network. The Weibull distribution is assumed for the wind speed. The test case

Page 75: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

74

assumes both onshore AC grids as IEEE test systems. The total load of the IEEE 118 bus system is 4242 MW. The converter 1 is connected to node 47 of IEEE bus system 1 and node 68 of IEEE bus system 2. Two identical IEEE bus systems are connected at node 102 of each.

Figure 46: Test case: DC grid connected to IEEE 118 bus system

The bus voltage and branch flow limits are modified to prepare the test case and to obtain a feasible power flow for every wind scenario. The optimal power flow problem for 500 wind samples is solved for this purpose and the maximum power flowing through each branch is calculated. The OPF problem also provides the initial set points (generator P and V) for the proposed method. Each branch flow limit is taken as 1.5 times the maximum possible power flow in the branch. The upper and lower voltage limits are taken as 1.1 pu and 0.9 pu respectively.

The frequency response model of the Nordic grid is used for the IEEE118 grid with the hydro generation as the FCR. For the selective protection strategy with a fast restoration process, there is no FFR required. For the selective protection strategy with a slow restoration time and for non-selective strategies, 250 MW of FFR is considered.

The dynamic parameters of the IEEE test systems are available in [36]. Using these data, the equivalent inertia and droop of the IEEE 118 bus system are calculated using equations 3-5 and 3-6 respectively. The parameters used for the frequency response model are shown in Table 22. The largest generation of IEEE 118 grid is 500 MW which is taken as the FCR capacity.

Table 22: Frequency response parameters

Parameter name Parameter value Equivalent system inertia (H) 3.11 s Equivalent system droop (R) 0.05 p.u. Load-frequency characteristic (D) 1 p.u. Turbine-governor time constant (Tg) 2 s FCR 500 MW

This section will cover the general overview of protection strategies and emphasize the relevant parameters for the proposed CBA methodology. Out of various protection strategies that have been assessed in Deliverable 4.2 [1], the present work selects two selective and two non-selective strategies for further analysis. The rationale for selection is explained in D4.3 [3]. The selected strategies are:

1. Fully selective fault clearing using fast DC circuit breaker (FS- FDCCB) 2. Fully selective fault clearing using slow DC circuit breaker (FS-SDCCB) 3. Non-selective fault clearing using full-bridge MMC (NS-FB) 4. Non-selective fault clearing using converter breaker (NS-CB)

In this work, the fast and slow DCCBs correspond to hybrid and mechanical breakers, respectively. The mechanism of each protection strategy is explained in appendix A.4. Only the primary sequence is taken into

Page 76: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

75

account since the probability of backup sequence is quite small, contributing only a fraction of the total operational cost.

PROTECTION PARAMETERS

A selective and a nonselective strategy have the same power deviation (∆PHVDC) in the power deviation curve depicted in Figure 22 (a). Therefore, the redispatch cost based on voltage and power flow constraint violations would be the same for both protection strategies. The difference between the two can be realized through taking the speed of protection into account, which would be translated into the resulting frequency deviation. The speed of protection is considered through the power restoration time which represents the time required for power flow to return to a post fault steady-state value (+/-10 %) from the time of fault occurrence [3]. The power restoration time for different protection strategies is given in Table 20.

5.6.2.1 CONSTRAINT VIOLATION

This section provides an analysis of voltage, power flow and frequency violations at the end of the primary protection sequence for different protection schemes. These violations would be eventually converted to a redispatch cost as explained in the methodology. The intermediate results can provide some insights about how the cost is affected by different violations. Note that since at the end of power restoration in the primary sequence, both selective and nonselective strategies would end up isolating only the faulted line in the primary sequence, the steady state violation (power flow and voltage) plots would remain the same for both.

Figure 47 show the power flow violations for FS-FDCCB in the primary sequence, which also represents the violations for the primary sequences of other FS and NS strategies. The IEEE 118 test system has 186 branches. The total number of branches in both AC grids combined is 373 (2*186 + a branch connecting two grids). The overload is determined out of 500 wind samples and the DC fault on each line is simulated for each wind sample. Figure 47(b) shows the range of overloads and the average value of overload on each branch as box plots. Figure 47 (a) shows the number of times each branch is overloaded in terms of percentage of total wind samples. No violations were observed in node voltages.

(a)

Page 77: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

76

(b)

Figure 47: Branch flow violations for FS and NS schemes in primary sequence (a) percentage overloads in branches with power flow constraint violations (out of 500 samples) (b) range of overloads

occurring in the branches

Table 23: Power flow violations

Protection type and sequence Branch flow violations

No.

All strategies 43

The frequency violations are given in Table 24. The following assumptions are made to obtain the frequency response:

1. Both selective and nonselective protection strategies would follow the simplified power deviation curve shown in section 2.2.1. Please refer to power deviation plots in D 4.2 [1] for examples of actual power deviation with time. The approximation adopted here enables the study of the impact on the power deviation curve due to a frequency deviation using simple models and avoids EMT simulation.

2. The power flowing between AC and DC grids is zero between the fault occurrence time until the time when the power restoration is completed.

3. With fast hybrid breakers, the unaffected converter (not directly connected to the faulted line) is assumed to be working continuously for this sequence. It should be noted that the power deviation may change largely based on the protection strategy, topology, current limiting inductor and fault interruption time. For example, Deliverable D4.2 [1] shows ∆PHVDC,dyn ranging from < 200 MW to > 1000 MW for selective strategies for 1200 MW pre-fault power flow to the onshore grid (refer figure 5.11 of D 4.2).

Table 24: Frequency violations

Protection type and sequence Avg (Hz) Percentage violations

FS - FDCCB_primary 0 0

FS- SDCCB_primary 48.95 Hz 9 %

NS - FB_primary 48.96 Hz 9 %

NS - CB_primary 48.96 Hz 9 %

Page 78: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

77

The frequency violations are absent for the primary sequence of FS – FDCCB scheme since the available FCR has a sufficient capacity to maintain the frequency stability. The violations remain at the same level (i.e. 9 % of total samples) in most cases since the power restoration curve is similar for all of these cases except minor differences in restoration time. The yearly operational cost of DC contingencies is calculated using equations 3-8 and weighted against the probability of sequence occurrence (see Table 25).

Table 25: Annual redispatch cost

Total 𝐿𝐿𝑃𝑃𝐸𝐸𝑋𝑋𝑀𝑀−1/year (M€)

FS – FDCCB 0.37

FS – SDCCB 5.59

NS – ConvBlk 5.59

NS – ConvBrk 5.59

5.6.3 KEY PERFORMANCE INDICATORS MONETARIZATION

In section 3.7, an aggregated method for the monetarization of economic key performance indicators has been proposed. In this section, an example of results which could be obtained from this methodology is shown. However, the purpose is not to draw a final conclusion about the protection strategies (i.e. classify these protection strategies) but only to show what the comparison between protection schemes could be. Indeed, depending on the assumptions (discounted rate, life time of the grid, EENT assumptions calculation, etc.) and uncertainties related to protection strategy costs, the conclusions could be radically different.

In this section, the following additional assumptions are taken into account:

• Wind farm generation is modelled considering a Weibull wind speed distribution • Grid lifetime is set to 25 years

Figure 48 shows the aggregated life cycle costs LCC indicator for the whole DC grid system with different protection strategies and for different DCCB unavailability rates and costs of energy. The monetised EENT indicator is shown under Option 1 in Table 7. This figure shows that the FS-FDCCB, FS-SDCCB and NS-FB life cycle cost ranking is still the same regardless of the DCCB unavailability. This is due to the fact that the DCCB unavailability is considered the same for all protection strategies.

Figure 48 also shows that the higher the DCCB unavailability is, the more the ranking between NS-CB and the other protection strategies changes. This is due to the fact that the life cycle cost of the EENT in NS-CB is more important compared to the other protection strategies (due the assumption related to the main sequence, see section 5.4). This example illustrates that the adverse impact on EENT of a given protection strategy could counterbalance the benefit of lower CAPEX, as it is for the NS-CB strategy, for high DCCB failure rates.

Page 79: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

78

Figure 48: Life cycle cost depending on unavailability (from 0.0003 and 0.0009).

a) price energy = 50 M€, b) price energy = 150 M€

Page 80: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

79

6 APPLICATION TO WP12 GRIDS DEVELOPMENT PLAN

This section presents an application of the methodology to a large DC grid. From obtained results, it is expected to bring information allowing to draw some recommendations to support WP12 deployment plan for a future European offshore grid.

Different case studies resulting from WP12 analysis are considered and introduced ("radial" and meshed versions of the so-called backbone DC grid). Economic indicators including CAPEX, OPEX (losses, maintenance and EENT) are evaluated for the radial backbone case study and presented in section 6.2. In that section, EENT indicator is used to assess the radial case study as well as two meshed variations. Section 6.3 is dedicated to efficiency KPIs (supported by EMT simulations) calculation. In that section, some protection strategies (non- selective and fully selective) are assessed on the radial backbone case study. Then, a protection strategy based on grid splitting is studied. It consists on implementing one firewall (i.e. fast DCCB) in the grid (at one line end), thus splitting the grid into two areas, each of them protected by a non-selective converter breaker protection strategy. Finally, AC system impact indicators are presented in section 6.4.

6.1 WP 12 BENCHMARK

The topologies developed within WP12 are aimed at evacuating all offshore wind production to shore while minimizing the CAPEX. These topologies have been developed considering some design criteria described in the WP1 deliverables and in later agreements by PROMOTioN partners.

In this chapter, application to 4 potential grid structures provided by WP12 is carried out. These 4 grid structures are the following:

• “Backbone“ grid structure (see Figure 49). • Two meshed version of “backbone“ grid structure:

o “Dogger bank“ ring grid structure (see Figure 50), o “Wider“ ring grid structure (see Figure 51),

• “Backbone“ grid structure without country interconnectors (see Figure 52).

It should be noted that the Key Performance Indicators (CAPEX, OPEX, efficiency KPIs) are not calculated for all these grid structures. Indeed, the first reason is related to the computation time needed for all these indicators, in particular for Efficiency KPIs (time to restore V, P and Q, and energy imbalance), which need extensive EMT simulation. The second reason is due to the fact they are quite similar grid structures with same number of nodes (converters), which are constructed around a common grid: “Backbone“ structure. Table 26 summarizes the kind of study carried out for each grid structure.

Table 26: Studies carried out for each grid structure.

CAPEX LOSSES EENT MAINTENANCE EFFICIENCY KPIS

“Backbone“ yes yes yes yes yes

“Dogger bank“ ring yes yes yes NO NO

“Wider“ ring yes yes yes NO NO

“Backbone“ without interconnectors yes yes yes NO NO

Page 81: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

80

The “backbone” grid-structure (see Figure 49) consists of a multi-terminal connecting 5 countries (Norway, Denmark, Germany, Netherlands and United Kingdom). The grid structure includes:

• 29 lines (bipolar pairs of cables), for a total cable length of 5452 km (one pole). The lines include cables of 1 GW, 0.8 GW, 0.7 GW and 0.5 GW capacity (one pole). The detail of lines (length, connection and resistance) are summarized in appendix A.8 (Table 54).

• 35 converter nodes divided into 16 onshore converters and 19 offshore converters. The offshore converters (19 nodes) represent the generators of the “backbone“ grid structure connecting the offshore wind farms. The wind farms maximum power output varies between 549 MW and 2000 MW. The generator capacities are summarized in appendix A.8 (Table 55).

Figure 49: “Backbone” grid structure.

The meshed variations “Dogger Bank” and “Winder” ring topologies are proposed based on engineering judgment and on the results of the topology development. It must therefore be noticed that the economic benefits of these variations has not been assessed by WP12. These topologies represent elements of the larger optimisation done by WP12 to develop the different concept topologies for evacuating Offshore Wind Energy to shore. In this chapter, some additional economic indicators (protection strategy costs, losses and expected energy not transmitted) are included in the analysis for these topologies.

Furthermore, a reference case study “Backbone” without interconnectors is introduced by WP4.5 (see Figure 52). This case study aims to illustrate the interest of connecting the windfarms area (NL, UK, DE, NO and DK) from EENT point of view.

The first proposed variation connects the dots in the “Dogger Bank” area where the sea depth is shallow to form a small ring. It can also be seen that the distances between offshore platforms in the Dogger Bank are relatively small, which lead to reduce additional cost of cable for meshing. The proposed meshing consists of three

Page 82: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

81

additional cables connecting the wind farms UK_OFF31-UK_OFF36 and DE_OFF01. The cable length required to close that mesh is around 270km.

Figure 50: “Dogger bank” ring meshed variation of “backbone” grid structure.

The “Wider” ring variation needs longer cable than the “Dogger Bank” ring but can potentially be built earlier because of the shorter distance to shore and the fact that future wind farms are likely to be installed first close to shore and later on in the Dogger Bank area. This ring requires to connect UK_OFF11 to the NL_OFF20/NL_OFF22 hub and then to the NL_OFF16/NL_OFF17 hub for a total length of 425km.

Page 83: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

82

Figure 51: “Wider” ring meshed variation of “backbone” grid structure.

Figure 52: Without interconnectors version of “Backbone” grid structure

Page 84: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

83

ASSUMPTIONS/DATA

The considered assumptions are as follow:

The reliability data used for Expected Energy Not Transmitted computation are summarized in Table 27 and are also given in appendix A.1. These data comes from Cigre expectations for protection components. The DCCBs reliability data are assumed to be close to ACCBs (which match the limited data found in literature [30] [31] [32] [33] [34] ). In this section, the unavailability of 0.0006 (twice ACCBs unavailability) is considered for offshore DCCBs.

Only a simple bus bar configuration is considered.

The bus bars are considered as fully reliable. For EENT calculations, this assumption does not affect the final results. Indeed, the bus bar reliability is so high that EENT caused by bus bar failures is low even if the power losses is high. This is due to the very high availability of the bus bar. However, from a security point of view, the bus bar configuration could be an important factor to take into account, for instance where the topology is altered in the case where a single bus bar failure (however unlikely) would not lead to a particular maximum EENT.

The hybrid and mechanical DCCBs reliability data are considered similar. This assumption is accepted by all PROMOTioN partners and helps ensure the confidentiality between (manufacturer) partners. To a large extent, component reliability is a design criterion, which comes at a cost. For the current studies, data provided in Table 27 is used. Indeed, we could expect that the DCCBs reliability will be high compared to cables and converters. So, the impact of the protection strategy in whole EENT will be small (around 5 to 10 % off total EENT for all protection strategies as seen in section 5.4). So, based on this assumption, only the converter breaker protection strategy is considered for EENT studies.

The maximum wind farm power is considered as a criterion for EENT studies. The EENT is the difference between the available wind power (depending on wind speed), and the effective extracted power by the onshore converters. Note that for a full assessment of EENT, it should be necessary to also consider impact on cross border exchanges. This has not been done within WP4.5 scope due to lack of AC system physical and market data.

Generation input for EENT calculation in this section are provided as time series with 8760 hourly snapshots for each converter (derived from 52 sites distributed over the North Sea). It should be noted that only 12 wind profiles are used for the 19 “Backbone” offshore converters. This is due to the fact that some wind farms are quite close to each other. In principle, for a grid with large number of wind farms, to consider 8760 hourly power generation values is not sufficient to cover all the scenarios. However, due to observed correlation between the 12 wind profiles, the 1-year hourly (8760 hours) wind generation snapshots is considered to be representative for North Sea.

Four fault clearing strategies are considered in the scope of this application:

• A fully selective using fast DCCB (FS-SDCCB)

• A fully selective using slow DCCB (FS-SDCCB)

• A non-selective using DCCB at converter output (NS-CB)

• A non-selective using converter with fault blocking capability (NS-FB)

CAPEX KPI is applied considering the four here above defined fault clearing strategies. Performances are captured through a set of technical KPIs, some of them being computed through EMT simulations. Only protection primary sequences are here considered when computing technical KPIs. Three fault clearing strategies are considered for technical KPIs assessment:

• A fully selective using fast DCCB

• A fully selective using slow DCCB

• A non-selective using DCCB at converter output

Page 85: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

84

Table 27: Components’ reliability data used for WP12 case studies

ONSHORE OFFSHORE

MTTF (hours) MTTR (hours) MTTF (hours) MTTR (hours)

Cables - - 219150 2316

Converters 6257 48 6257 6

Switches (HSS) 160000 48 160000 6

DCCBs (Hybrid/Mechanical) 80000 48 80000 6

6.2 ECONOMIC KEY PERFORMANCE INDICATORS

In this section, economic Key Performance Indicators including, CAPEX, OPEX (losses, expected energy not transmitted) will be computed for “backbone”, “dogger bank” ring, “wider” ring and “backbone” without interconnector grid structures.

6.2.1 DCCB SPECIFICATION FOR CONVERTER BREAKER, FULLY SELECTIVE (WITH HYBRID AND MECHANICAL DCCB), AND FULL BRIDGE PROTECTION STRATEGIES

To compute the protection strategy costs, protection components (mainly DCCBs) should be specified. Table 28, Table 29 and Table 30 summarize the DCCBs specifications used for Full-Selective strategies (with mechanical and hybrid DCCBs) and Non-Selective protection strategies (Full-bridge and converter breaker). These specifications are derived by applying the methodologies presented in section 3.8. Thirteen different breakers specification are identified (DC CB type 1 to 13) for covering the four protection strategies requirements.

Table 28: DCCBs specification, used for “backbone”, “dogger bank” ring, “wider” ring and “backbone” without interconnector grid structures: Type 1 to Type 4

DCCB Type 1

DCCB Type 2

DCCB Type 3

DCCB Type 4

Parameter Unit

Converter breaker

Mechanical DCCB, 20 kA

Converter breaker

Mechanical DCCB 16 kA

Full bridge Mechanical DCCB 2 kA

(80kV)

Fully selective

Mechanical DCCB 16 kA

Rated DC voltage kV 525 525 80 525

Rated DC current kA 2 2 2 1.9

Rated breaking current capability kA 20 16 2 16 Breaker operation time (including delay

time) ms 8 8 8 8

Directionality - Bidirect Bidirect Bidirect Bidirect

Open-close operation - O-C-O O-C-O O-C-O O-C-O

Current limiting DC reactor H 0 0 0 0.3445 Rated transitent Interruption Voltage

(TIV) p.u 1.5 1.5 1.5 1.5

Rated energy absorption MJ 3 3 1 132.3

Page 86: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

85

Table 29: DCCBs specification, used for “backbone”, “dogger bank” ring, “wider” ring and “backbone” without interconnector

grid structures: Type 5 to Type 8

DCCB Type 5

DCCB Type 6

DCCB Type 7

DCCB Type 8

Parameter Unit

Fully selective

Mechanical DCCB 16 kA

Fully selective

Mechanical DCCB 16 kA

Fully selective

Mechanical DCCB 16 kA

Fully selective

Mechanical DCCB 16 kA

Rated DC voltage kV 525 525 525 525

Rated DC current kA 1,52 0,95 1,33 0,7

Rated breaking current capability kA 16 16 16 16 Breaker operation time (including delay

time) ms 8 8 8 8

Directionality Bidirect Bidirect Bidirect Bidirect

Open-close operation O-C-O O-C-O O-C-O O-C-O

Current limiting DC reactor H 0,3243 0,298 0,315 0,2875 Rated transitent Interruption Voltage

(TIV) p.u 1.5 1.5 1.5 1.5

Rated energy absorption MJ 124,5 114,4 121 110,4

Table 30: DCCBs specification, used for “backbone”, “dogger bank” ring, “wider” ring and “backbone” without interconnector grid structures: Type 9 to Type 13

DCCB Type 9

DCCB Type 10

DCCB Type 11

DCCB Type 12

DCCB Type 13

Parameter Unit

Fully selective Hybrid DCCB 16 kA

Fully selective Hybrid DCCB 16 kA

Fully selective Hybrid DCCB 16 kA

Fully selective Hybrid DCCB 16 kA

Fully selective Hybrid DCCB 16 kA

Rated DC voltage kV 525 525 525 525 525

Rated DC current kA 1,9 1,52 1,33 0,95 0,7 Rated breaking current capability kA 9 9 9 9 9 Breaker operation time (including delay time) ms 2 2 2 2 2

Directionality Bidirect Bidirect Bidirect Bidirect Bidirect

Open-close operation O-C-O O-C-O O-C-O O-C-O O-C-O

Current limiting DC reactor H 0,2023 0,1765 0,1658 0,148 0,138 Rated transitent Interruption Voltage (TIV) p.u 1.5 1.5 1.5 1.5 1.5

Rated energy absorption MJ 24,6 21,4 20,1 18 16,8

6.2.2 CAPEX INDICATOR FOR WP12 GRID STRUCTURE

The CAPEX of the grid structures presented in section 6.1 are already calculated in WP12 scope without considering any protection strategy costs. In this section, the whole CAPEX (including grid components –Converters, Platforms, Cables, and Transformers – and protection strategy components) is calculated. The aim of considering the whole grid CAPEX is to compare the part of the protection strategy CAPEX.

Page 87: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

86

Based on DCCBs specification introduced in section 6.2.1, and based on the DCCB costs model developed in [7] the unit cost of all DCCBs used for WP 12 grid structures (“Backbone” with its 2 meshed version and without interconnector, see section 6.1) is shown in Figure 54. This figure shows that the costs of DCCB vary from 10 M€ to 20 M€ depending on specifications. It should be noticed that the DCCBs used for full bridge protection strategy is quite low (around 1.7 M€, for type 3).

Grid layouts with the necessary components associated with converter breaker and fully selective (with hybrid and mechanical DCCBs) protection strategies are shown in Figure 53. The same layout for converter breaker, fully selective with mechanical DCCBs and fully selective hybrid DCCBs is considered since it is considered a converter DCCB based back-up strategy for selective protection systems (backup with converter breakers).

In full bridge strategy, AC backup is considered. Therefore, the layout will be as in Figure 53 without converter DCCB (black squares).

Figure 53: Non selective (Converter breaker and full bridge) and fully selective (with hybrid and mechanical DCCBs)

protection strategy layouts.

Page 88: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

87

Figure 54: DCCBs unit costs used for “Backbone”, “Dogger bank” ring, “Winder” ring and “Backbone” without

interconnector grid structures.

Page 89: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

88

Table 31, Table 32 and Table 33 present the number and type of DCCBs used for fully selective strategy with mechanical DCCBs, fully selective strategy with hybrid DCCBs and non-selective converter breaker/full bridge protection strategies respectively. In total, 7 protection strategies are assessed from cost point of view. It should be noticed that AC backup is only considered for fully selective strategy (which is studied in deliverable D4.2 and D4.3). In this case, the corresponding layout does not include the converter breaker (the black square in Figure 53).

Table 31: Type and number of DCCBs used for fully selective strategy with mechanical DCCBs.

DCCB type4

DCCB type5

DCCB type6

DCCB type7

DCCB type8

Option1.1 AC Backup Onshore 14 2 14 2 0

Offshore 30 26 14 14 0

Option1.2 Converter breaker backup Onshore 28 4 28 4 0

Offshore 36 40 16 24 6 Table 32: Type and number of DCCBs used for fully selective strategy with hybrid DCCBs.

DCCB type9

DCCB type10

DCCB type11

DCCB type12

DCCB type3

Option.2.1 AC Backup Onshore 14 2 14 2 0

Offshore 30 26 14 14 0

Option.2.2 Converter breaker backup Onshore 28 4 28 4 0

Offshore 36 40 16 24 6 Table 33: Type and number of DCCBs used for converter breaker and full bridge strategies.

DCCB type1

DCCB type2

DCCB type3

Option3.1 Converter breaker strategy Onshore 64 0 -

Offshore 122 0 -

Option3.2 Converter breaker strategy Onshore 64 0 -

Offshore 84 38 -

Option4 Full bridge strategy Onshore - - 32

Offshore - - 84

From Table 31, Table 32 and Table 33, the corresponding total protection strategy investment costs of option1.1, option1.2, option2.1, option2.2, option3.1, option3.2 and option4 are shown in Figure 55. This figure shows that, as expected, the converter breaker protection strategy, which uses cost effective DCCBs, has lower CAPEX compared to the fully selective strategies. Option 1.1 is an exception as it uses fewer DCCBs because of ACCBs backup consideration. For the same reason, option 2.1 using hybrid DCCBs seems to be more cost effective than option 2.2. As expected, the full bridge protection strategy corresponds to the lowest CAPEX in protection

Page 90: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

89

equipment, with a CAPEX not exceeding 1.7 M€. However, in this comparison the additional CAPEX of full bridge converters is not included as it should be done.

Figure 55: Fully selective, Non selective and full bridge options total investment costs.

Figure 56 and Figure 57 show the “Backbone” grid structure overall CAPEX (including protection strategies) and maintenance costs respectively. In Figure 57, only protection strategies with converter breaker backup are considered for fully selective protection strategies (option 1.2 and option 2.2), and only option 3.1 is considered for the non-selective protection strategy.

These figures show that the total CAPEX for all protection strategies varies from 5% to 9% of the total system CAPEX (5% for the NS-CB, 7% for the NS-FB and the FS-SDCCB and 9% for the FS-SDCCB) (see Figure 57 and Figure 58). As assumed in section 3.2.1, the additional CAPEX of Full Bridge converter has to be included in the CAPEX of NS-FB protection strategy. As expected, the non-selective using converter breaker (NS-CB) protection strategy is the most cost effective from the CAPEX point of view.

Page 91: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

90

Figure 56: “Backbone” grid structure total costs (including protection strategies) of all protection strategy options.

Figure 57: Component contribution in total “Backbone” grid structure.

Page 92: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

91

Figure 58: “Backbone” grid structure CAPEX including only:

6.2.3 OPEX - MAINTENANCE INDICATOR FOR WP12 GRID STRUCTURES

The Figure 59 shows the respective maintenance cost contributions of protection strategies to total maintenance costs. As far as maintenance costs are evaluated as a fixed proportion of CAPEX, it can be noted that these respective contributions are similar to the contributions to total CAPEX.

Figure 59: “Backbone” grid structure maintenance (including protection strategies) of all protection strategy options.

6.2.4 OPEX - LOSSES INDICATOR FOR WP 12 GRID STRUCTURES

Figure 60 and Figure 61 presents the total grid losses and respective contributions of protection strategies, computed for 100% wind generation and no contingencies scenario. It can be observed that these contributions are quite low (from ~0.08% for NS-FB strategy to ~4% for FS-SDCCB strategy), as far as the main contributors to losses are converters. Note that the main contributors to losses for protection systems are the DC reactors. As

Page 93: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

92

such, the maximum contribution to losses are obtained with fully selective using slow DCCBs which needs larger DC reactors. Moreover, the total grid losses are around 2.5% for FS-FDCCB, FS-SDCCB and NS-CB strategies, and to the order of 3% for NS-FB strategy in which additional losses (around 30% more than a half-bridge converter) for full bridge converter are observed.

Figure 60: Total losses for “backbone” grid structure (% of total losses, for no contingencies and 100% wind generation, i.e.

25664 MW).

Figure 61: Component contributions to losses (% of total losses, for no contingencies and 100% wind generation, i.e.

25664 MW).

a) FS-DCCB, b) FS-SDCCB, c) NS-FB, d) NS-CB.

6.2.5 OPEX - EENT INDICATOR FOR WP12 GRID STRUCTURES

To compute the expected energy not transmitted, the maximum wind power Optimal Power Flow (OPF) is considered (this corresponds to RES integration criteria). An hourly wind power time series is taken as an input

Page 94: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

93

power generation (8760 points are considered). This means that for each grid configuration (contingency), 8760 OPF are performed. Hence, the cable, transformer and converter losses are derived from the resulting OPF flows.

Expected Energy Not Transmitted study is performed to the grid structures presented in section 6.1. Figure 62 and Figure 63 show the EENT and component contributions of these grid structures respectively.

Figure 62: Expected Energy Not Transmitted (per unit, percentage of total energy).

Figure 63: Component contributions to EENT (% of total EENT).

a) Backbone without interconnectors, b) “Wider” ring, c) “Dogger bank” ring, d) “Backbone”.

Following conclusions can be drawn from these figures:

• From EENT point of view, there is clearly an interest to interconnect the country wind farms (NO, UK, DK, NL and DE). Indeed, a difference about 1400 GWh of EENT per year can be observed between “Backbone” without interconnectors and the other grid structures (see Figure 62). In addition to this gain in term of reduction of RES curtailment, it should be necessary to also consider benefits from cross

Page 95: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

94

border exchanges (this has not been done in the scope of WP4.5). These benefits shall also be related to additional investment costs which are mainly due to additional cables and protection equipment. In the considered case study, additional cables CAPEX is around 1300 M€. Additional protection equipment cost is dependent on the implemented protection strategy (e.g. around 800 M€ for a non-selective converter breaker protection strategy and around 1200 M€ for a fully selective using fast DCCBs).

• The “Wider” ring grid structure seems to be interesting from EENT point of view with around 250 GWh and 180 GWh (per year) less than “Backbone” and “Dogger Bank” ring grid structures. The additional costs of the “Wider” ring and “Dogger Bank” ring grid structures comparing to “Backbone” is around 1050 M€ and 700 M€ respectively.

• The contribution of protection strategy to the total EENT is very low (around 4%) (see Figure 65). The grey color in Figure 63 corresponding to “Both EENT” is due to both grid (converters, cables) and protection components unavailability.

6.3 EFFICIENCY KEY PERFORMANCE INDICATORS THROUGH EMT SIMULATION

In this section, two main studies are carried out:

1. Non selective and fully selective protection strategy comparison: In this application, efficiency KPIs are computed for fully selective strategy (FS-FDCCB and FS-SDCCB) and non-selective strategy converter breaker NS-CB.

2. Grid splitting efficiency KPIs calculation: the objective of this study is to show that either a grid splitting using the same protection strategy (or combining different protection strategies) is possible or not. In this section, the study is carried out by considering converter breaker protection strategy for the whole grid and setting one firewall (i.e. fast DCCB) in the grid (at one line end), so splitting the grid in two areas, each of them implementing a converter breaker protection strategy.

To compute the efficiency key performance indicators (time to restore V, time to restore P, time to restore Q and energy imbalance), five fault locations are selected labeled F1, F3, F5, F11 and F12 (see Figure 64). Protection strategy comparison between fully selective and non-selective strategy is performed on fault locations F1, F3 and F5 while the grid splitting efficiency KPIs calculation is carried out considering F1, F3, F11 and F12. Moreover, the following additional assumptions are taken into account:

Bipolar with metallic return is considered. For EMT simulations, only one pole is considered. It is then assumed that the healthy pole is not affected by the fault (No bundled cables are considered).

Droop control is considered for all converters.

Pole to ground fault is simulated.

Only one power set point (load flow) is considered. This load flow (for one pole) is provided by the WP 12.

The post fault operation point is assumed to be acceptable for both AC and DC systems. In other word, the new set point is assumed to be N-1 secure.

Only the primary sequence is considered for KPIs calculation.

These fault locations are chosen in such a way that it includes the main cases for the protection strategies. Fault location F1 aims to represent both furthest fault locations and fault located at onshore nodes. Fault location F3 is supposed representing the middle grid fault location and the large hubs (connecting several converters and a powerful wind farms). Fault locations F11 and F12 are chosen in such a way that a fault occurs on one side of the firewall or the other side (zone 1 or zone 2, see Figure 64).

Converter breaker NS-CB strategy and fully selective FS-FDCCB strategy are detailed in this section. For fully selective protection FS-SDCCB strategy, simulation details and results can be found in appendix A.11 (see Figure 91 and Figure 92).

Page 96: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

95

Figure 64: Different fault locations considered in EEMT simulation for efficiency KPIs calculation.

6.3.1 EFFICIENCY KEY PERFORMANCE INDICATORS FOR NON SELECTIVE CONVERTER BREAKER PROTECTION STRATEGY

In this section and in section 6.3.2, non-selective NS-CB and fully selective FS-FDCCB protection strategies are first assessed based on fault location F1 and F3. A full comparison between all protection strategies NS-CB, FS-FDCCB and FS-SDCCB is carried out at the end. Furthermore, the comparison is performed considering four key performance indicators: time to restore voltage, time to restore active power, time to restore reactive power, and energy imbalance indicators. The time restoration is calculated per node and as the maximum time to keep the node voltage and active/reactive power within a bandwidth (within ± 10%). Similarly, the energy imbalance is calculated in the same manner considering a band width of ± 10% of the nominal values. Moreover, the new set point (after fault clearing process) is taken as a reference to compute these values. For example, the voltage time restoration for a given converter will correspond to the needed time to stabilize the voltage at this converter between ± 10%.

Figure 65, Figure 66 and Figure 67 show outputs of EMT simulations (voltage, active/reactive powers), nodal values (time to restore voltage, times to restore active/reactive powers) and the time to restore voltage, times to restore active/reactive powers and magnitude of energy imbalance distributions respectively for the fault location F3 and non-selective protection strategy NS-CB. It should be noted that Figure 67 is drawn from Figure 66 and shows the frequency (number of converters) and the cumulative probability distribution CDF of time to restore voltage, times to restore active/reactive powers and magnitude of energy imbalance. The objective of these three figures is to show how the voltage, active and reactive power look like during the clearing process, how these values look like at each converter node and how they are distributed over the DC grid.

Page 97: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

96

Figure 65: EMT simulation curves for fault location F3 and non-selective converter breaker protection strategy NS-CB

a) DC Voltage, b) active power, c) reactive power

Page 98: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

97

Figure 66: Nodal voltage, active and reactive power time restorations for fault location F3 and non-selective converter

breaker protection strategy NS-CB.

Figure 67: Nodal voltage, active and reactive power time restoration distributions for fault location F3 and non-selective

converter breaker protection strategy NS-CB.

Page 99: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

98

From Figure 65, Figure 66 and Figure 67 following conclusions can be drawn:

• All node voltages, active/reactive powers and energy imbalance restoration times are affected by the fault F3 (see Table 34). This is shown by the fact that the CDF curves (see Figure 67) start from 0 for each key performance indicators.

• Voltage restoration times are between 75ms and 85ms for all nodes, which indicates that these time restorations are smoothly distributed over the nodes (see blue curve of Figure 66). It can be also observed that 50% of the total converters (the median) have a restoration time less than 77ms, and 90% have a restoration time which do not exceed 82ms.

• Active power (respectively reactive power) time restoration is in range of 130ms to 190ms (respectively, 10ms to 85ms). Figure 67 shows that half of converters have an active power restoration time (respectively, reactive power restoration time) less than 152ms (respectively, 13ms), and 90% less than 177ms (respectively 70ms). Furthermore, it can be observed that the active power time restoration seems to be influenced more for onshore nodes (with large power capacity, and connected to the onshore grid) comparing to the offshore grid. Moreover, these values are also more distributed in case of onshore converters by contrast to offshore converters which value are smoother (distributed around 130ms see red curve of Figure 66).

• The range of energy imbalance KPI is quite high (between 20 MJ and 150 MJ), which indicates that some converters are less impacted than others (this is highlighted by Figure 69.d). So, some converters seem to have a quite high energy imbalance (150MJ, the maximum value, 95MJ for half of converters).

• The worst fault locations for the NS-CB protection strategy (regarding voltage time restoration) seem to be the fault in the middle grid close to a hubs (F3). For instance, 33% of converters are not affected regarding voltage time restorations in case of fault location F1 while all converters are affected in case of fault location F3 (see figure a and c of Figure 90, appendix A.9). In other words, more the fault occurs far from the middle grid, less the number of converters are impacted by the fault occurrence.

• The protection strategy NS-CB capacity to restore reactive power on the grid is high (90% of converters restored in less than 70ms).

Table 34 summarizes the voltage, active and reactive power restoration times and the energy imbalance key performance indicators for non-selective NS-CB protection strategy in case of fault location F3.

Table 34: Summary of efficiency KPIs for non-selective-converter breaker protection strategies (F3)

Voltage restoration time

Active power restoration time

Reactive power restoration time

Energy Imbalance

Not affected converters (%) 0 0 0 0

Range (lowest/highest values) (ms/MJ)

[75,85] [130,190] [10,85] [20,150]

Median (ms/MJ) 77 152 13 95 90% threshold (ms/MJ) 82 177 70 145

6.3.2 EFFICIENCY KEY PERFORMANCE INDICATORS FOR A FULLY SELECTIVE PROTECTION STRATEGY USING FAST DCCBS (FS-FDCCB)

Figure 68, Figure 69 and Figure 70 show of EMT simulations (voltage, active/reactive powers), nodal values (time to restore voltage, times to restore active/reactive powers) and the time to restore voltage, times to restore active/reactive powers and magnitude of energy imbalance distributions respectively for the fault location F3 and fully selective protection strategy FS-FDCCB.

Page 100: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

99

Figure 68 EMT simulation curves for fault location F3 and fully selective using fast DCCB protection strategy FS-FDCCB

a) DC Voltage, b) active power, c) reactive power.

Page 101: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

100

Figure 69: Nodal voltage, active and reactive power time restorations for fault location F3 fully selective using fast DCCB

protection strategy FS-FDCCB.

Figure 70: Nodal voltage, active and reactive power time restoration distributions for fault location F3 fully selective using fast DCCB protection strategy FS-FDCCB.

Page 102: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

101

Table 35 summarizes the voltage, active and reactive power restoration times and the energy imbalance key performance indicators for fully selective FS-FDCCB protection strategy in case of fault location F3.

From Table 35, Figure 68, Figure 69 and Figure 70 following conclusions can be drawn:

• By contrast to non-selective converter breaker NS-CB protection strategy, only voltage, active and reactive power time restorations KPIs of the converters close to the fault location are affected. This is also shown by the non-zero values in Table 35. All converter are still affected regarding energy imbalance KPI. However, the impact on this KPI seems to be quite low. Indeed, the values corresponding to the energy imbalance in Table 35 are quite small.

• For affected converters, voltage restoration times are relatively small (less than 45 ms) and 80 % of voltage converters re still in bandwidth of ±10% of the nominal value during the fault clearing process. Furthermore, the majority of converter voltages (90%) are restored quite quickly (less than 15 ms).

• About 75% of converters are not affected from active power restoration KPI point of view, and 90% of converter active power are restored in less than 47ms.

• Around 90% of converters are not affected from reactive power KPI point of view and most of them have a reactive power restoration time of less than 6 ms.

Table 35: Summary of efficiency KPIs for full-selective FS-FDCCB protection strategies (F3)

Voltage restoration time

Active power restoration time

Reactive power restoration time

Energy Imbalance

Not affected converters (%) 80 74 89 0

Range (lowest/highest values) (ms/MJ) [0,45] [0,100] [0,15] ]0,30]

Median (ms/MJ) 0 0 0 4

90% threshold (ms/MJ) 15 47 6 12

6.3.3 COMPARISON BETWEEN NON SELECTIVE NS-CB, FULLY SELECTIVE FS-FDCCB AND FULLY SELECTIVE FS-SDCCB PROTECTION STRATEGIES

In this section, non-selective converter breaker, fully selective FS-FDCCB and FS-SDCCB protection strategies are compared. Figure 71 shows time to restore voltage, times to restore active and reactive power and magnitude of energy imbalance KPIs cumulative probability distributions for these protection strategies. A threshold is also displayed in this figure. The intersection of this threshold and the other curves (blue, green and red curves) projection on x-axis corresponds to the value in which 90% of converters have a KPI less than this value. For example, the intersection of the 90% threshold with the red curve in Figure 71 (a) (corresponding to the voltage time restoration for FS-FDCCB protection strategy) indicates that 90% of the converters have a voltage time restoration less than 10 ms. Table 36 summarizes the voltage, active and reactive power restoration times and the energy imbalance key performance indicators for non-selective NS-CB, fully selective FS-FDCCB and fully selective FS-SDCCB protection strategies. Figure 71 and results in Table 36 are drawn from different fault location by aggregating F1, F3 and F5. In other words, 105 values (3 multiply by 35 converters) for each KPIs and each protection strategy are included.

The following conclusions can be drawn from Figure 71 and Table 36:

• Fully selective FS-FDCCB and FS-SDCCB are clearly more efficient than non-selective converter breaker protection strategy NS-CB for all KPIs. This is highlighted in Table 36 by the low values of the range, median and 90% threshold, the high percentage value of converter which are not affected in case

Page 103: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

102

of fully selective FS-FDCCB and FS-SDCCB protection strategies. This is also confirmed by the fact that the blue curves, which corresponds NS-CB protection strategy, is closer to the x-axis than the FS-FDCCB and FS-SDCCB curves (green and blue curves).

• Fully selective FS-SDCCB slightly degrades all the KPIs comparing to fully selective FS-FDCCB. This can be observed by the fact that the percentage of converters not affected, the range and the 90% threshold is slightly for benefit of FS-FDCCB protection strategy.

• All converters are affected from energy imbalance KPI point of view whatever the protection strategy is. However, the impact is quite low for fully selective FS-FDCCB and FS-SDCC protection strategies. Indeed, the median and 90% threshold values are less than 6MJ and 15MJ for fully selective strategies whereas these values are respectively around 85MJ and 135MJ for non-selective protection strategy NS-CB.

• Reactive power restoration times is relatively small for all protection strategies. Particularly, about half of converters are not affected (or have a time restoration close to 0) and the 90% threshold do not exceed 65ms for all protection strategies. So, these indicators seem to be not so much discriminant between protection strategies.

Figure 71: Efficiency KPIs comparison between NS-CB, FS-FDCCB and FS-SDCCB protection strategies.

a) Voltage, b) active power, c) reactive power, d) energy imbalance. Aggregated fault location F1, F3 and F5.

Page 104: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

103

Table 36: Summary of efficiency KPIs for non-selective-converter breaker protection strategies (F1, F3 and F5)

Strategy Voltage restoration time

Active power restoration time

Reactive power restoration time

Energy Imbalance

Not affected converters (%)

NS-CB 10 0 37 0 FS-FDCCB 90 75 92 0 FS-SDCCB 75 55 83 0

Range (lowest/highest values, ms/MJ)

NS-CB [0,90] [120,200] [0,90] [10,150] FS-FDCCB [0,70] [0,100] [0,50] ]0,30] FS-SDCCB [0,120] [0,150] [0,60] ]0,40]

Median (ms/MJ)

NS-CB 75 140 15 85 FS-FDCCB 0 0 0 2 FS-SDCCB 0 0 0 6

90% threshold (ms/MJ)

NS-CB 90 180 65 135 FS-FDCCB 10 70 0 10 FS-SDCCB 41 100 31 15

6.3.4 EFFICIENCY KEY PERFORMANCE INDICATORS FOR CONVERTER BREAKER AND GRID SPLITTING

In this section, non-selective converter NS-CB and grid splitting (using NS-CB protection strategy in all protected zones) protection strategies are compared. The objective is first, to show that a grid splitting for a large grids is possible, and secondly to show that some key performance indicators presented in 3.3.1 and 3.3.2 could be improved by implementing a grid splitting protection strategy. So, the optimal grid splitting is not addressed in this deliverable, although the methodology developed can be easily extended to perform such optimization. It is not intended in this deliverable and within WP4.5 to perform a detailed study and a full demonstration of a grid splitting based protection system. Indeed, this would require to develop a deep analysis of protection system (protection algorithms, component specification and performances, primary and back-up sequences ...) and is out of the scope of the WP4.5 work. Additionally, the EMT simulation of some grid splitting protection applied to the WP 4.3 DC grid benchmark can be found in appendix of [3].

Figure 72 and Figure 73 show the active power outputs obtained by EMT simulations for protected zone 1 and zone 2 (see Figure 64) and fault location F11 and F12 respectively. These figures show clearly that the zone which does not include the faulty component is practically not affected by the fault occurrence. Indeed, unlike the faulty zone (which includes the faulty component, upper Figure 72 and lower Figure 73) converter powers in the healthy zone (lower Figure 72 and upper Figure 73) remain within nominal values during the fault clearing process (except at the fault beginning, during ~5 ms). This means that the grid splitting protection philosophy is clearly respected hence, and that the grid splitting approach on large DC grids could be a possibility in order to build protection schemes with a lower cost and responding to a specific operational constraint (e.g. protect a sensitive zone).

This conclusion is also supported by Figure 74 which shows the nodal energy imbalance indicator for all fault locations F1, F3, F11 and F12. Indeed, the nodes’ energy imbalance in case of F1, F3 and F11 (included in zone 1, blue, green and red colours in Figure 74) is quite high for converter in zone 1 while it is close to zero for converters included in protected zone 2 (vice versa in case of fault location F12, included in protected zone 2).

Page 105: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

104

Figure 72: EMT simulation curves for grid splitting strategy using NS-CB protection strategy. DC Active power before, during and after a fault for fault locations F11.

a) converters in zone 1, b) converters in zone 2.

Page 106: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

105

Figure 73: EMT simulation curves for grid splitting strategy using NS-CB protection strategy. DC Active power before,

during and after a fault for fault locations F12. a) a) converters in zone 1, b) converters in zone 2

Page 107: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

106

Figure 74: Nodal energy imbalance (MJ) for grid splitting with NS-CB strategy and fault location F1, F3, F11 and F12.

Figure 75 shows time to restore voltage, times to restore active and reactive power and magnitude of energy imbalance KPIs cumulative probability distributions for non-selective NS-CB without grid splitting and non-selective NS-CB with grid splitting protection strategies. Only faults locations in zone 1 (i.e. F1, F3, F5 for NS-CB, and F1, F3, F11 for NS-CB with grid splitting) are considered to compare the protection strategies (blue and red curves).

The green curve corresponding to the fault location F12 in zone 12 is not used for comparison due to the fact that fault F12 has not been simulated for NS-CB without grid splitting. However, the green curve is displayed as an illustration of what could be the KPIs if a fault occurs in zone 2.

Table 37 summarizes the voltage, active and reactive power restoration times and the energy imbalance key performance indicators. Figure 71 and results in Table 37 are drawn from different fault location by aggregating F1, F3 and F5 for non-selective NS-CB and F1, F3, F11 for NS-CB with grid splitting. So, 105 values (3 multiplied by 35 converters) for each KPIs and each protection strategy are included.

The results clearly show that the key performance indicators of NS-CB protection strategy are quite improved by implementing a grid splitting strategy based on converter breaker NS-CB. This is also supported by Figure 75 which shows that the grid splitting strategy curve (in red colour) is mainly above to the non-selective NS-CB protection strategy curve (in blue colour).

Table 37: Summary of efficiency KPIs for non-selective-converter breaker NS-CB (aggregated fault location F1, F3, F5) and grid splitting with NS-CB (aggregated fault locations F1, F3, F11) protection strategies

Strategy Voltage restoration time

Active power restoration time

Reactive power restoration time

Energy Imbalance

Not affected converters (%)

NS-CB 10 0 37 0 Grid splitting 36 22 52 0 NS-CB [0,90] [120,200] [0,90] [10,150]

Page 108: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

107

Strategy Voltage restoration time

Active power restoration time

Reactive power restoration time

Energy Imbalance

Range (lowest/highest values) (ms/MJ)

Grid splitting [0,110] [0,220] [0,90] [0,150]

Median (ms/MJ)

NS-CB 75 140 15 85 Grid splitting 50 130 0 55

90% threshold (ms/MJ)

NS-CB 90 180 65 135 Grid splitting 100 175 20 130

Figure 75: Efficiency KPIs comparison between NS-CB and grid splitting with NS-CB protection strategies: probability

distribution for F12 (grid splitting) and aggregated fault locations F1, F3 and F5 (NS-CB) and F1, F3 and F11 (grid splitting) a) Voltage, b) active power, c) reactive power, d) energy imbalance.

Some considerations about the CAPEX of protection system related to the grid splitting option:

A methodology to estimate the CAPEX related to grid splitting solution is here proposed. It assumes that, in each area, the protection system and the rating of associated components are specified and that the operational KPIs are acceptable. Protection system CAPEX are then computed for each area and the full system in order to build a relation between full DC grid protection system design, specification and associated CAPEX.

The methodology is applied to large radial DC grid use case considering the here simulated protection systems, i.e. a NS using converter breaker protection system (NS-CB), a FS using fast DCCBs protection system (FS-SDCCB), a FS using slow DCCBs protection system (FS-SDCCB) and a partially selective (PS) splitting the grid

Page 109: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

108

in 2 areas and using NS-CB protection strategy in each area (PS with NS-CB). From those results, an extrapolation of PS with NS-CB with 11 areas is done. Table 38 presents a comparison of those protection strategies from both CAPEX and time to restore active power KPIs. The base case is the NS-CB protection strategy which is the more cost effective one. It can be seen that the CAPEX of protection system varies with a factor 2 and that splitting the grid in several areas would allow to reduce the constraints on the surrounding AC grids (less converters exhibiting active power disturbance) at a moderate additional cost.

Table 38 : Comparison of several protection strategies from CAPEX and time to restore active power KPI (* Observed in simulation ** Expected, not simulated)

Protection system CAPEX (p.u.)

Part of protection

system CAPEX relative to full

DC grid CAPEX (%)

Max Number of converter

terminals with active power disturbance

Max time to restore

active power (ms)

NS-CB (1 area= full DC grid) 1.0 5.0 35* 190* PS with NS-CB (2 areas) 1.02 5.1 28* 200* PS with NS-CB (11 areas) 1.1 5.4 15** ~200**

FS-SDCCB (29 areas = 29 single lines) 1.6 7.0 15* 150*

FS-SDCCB (29 areas = 29 single lines) 2.0 9.0 9* 100*

6.4 AC SYSTEM IMPACT KEY PERFORMANCE INDICATORS

Out of the proposed indicators for AC grid impact, we analyse the frequency stability and reserve requirement indicators for bigger grid example, after discussion with partners from WP 12. The grid model includes offshore backbone grid connected to the Nordic system. The simplified representation of the network is shown in Figure 76, where two offshore nodes are connected to the onshore grid at different points. Both nodes are connected to the onshore grid starting from 2045 within the planning horizon 2025-2050.

Nordic gridOffshore grid

DK_OFF01

NO_OFF02

AskaerDK_OFF25

Stokkeland

Figure 76: Connections to Nordic grid in backbone grid

The equivalent inertia is calculated using the hourly generation data obtained through the market simulations (input from other WP). The assumed inertia values for different types of power plants are shown in appendix A.2. Using equation 3-5, the calculated equivalent inertia of the Nordic grid is depicted in Figure 77 (a). From 2025 to 2050, the average inertia decreases from 3.36 sec to 1.36 sec whereas the average renewable generation increases from 17 % to 56 %. Note that the inertia values are valid for all NAT topologies and are not limited to the backbone grid. Table 39 indicates the least of yearly average inertia within planning horizon.

The reduced inertia leads to higher ROCOF values. Figure 78 shows the possible ROCOF values for selective and non-selective strategies. For selective strategy, the power injection is taken as the maximum of two node injections, whereas for the nonselective strategy, the power injection is taken as the sum of two node injections. We present results for the fault on DK_OFF01 to Askaer line since it is rated at a higher capacity, causing higher ROCOF than the Stokkeland connection. We observe that ROCOF may not be an issue if ROCOF limit is increased to 1 Hz/s or even higher in future. Although, if the current ROCOF limit (0.5 Hz/s) remains unchanged,

Page 110: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

109

ROCOF becomes a concern for both the 2045 and 2050 scenarios, especially for the non-selective scheme. Table 39 shows the ROCOF values for 2050 as KPI.

(a) (b) Figure 77: Nordic grid: (a) Equivalent inertia of Nordic grid (b) Percentage renewables of the total generation

Figure 78: rate of change of frequency (ROCOF) for backbone grid

Next, the minimum frequency violation duration is calculated. The power restoration time for different protection strategies are shown in Table 39. The missing power restoration time for NS-FB is assumed to be equivalent to NS-CB. The N-1 contingency is performed on two DC lines directly connected to the Nordic grid as well as on the remaining lines in the DC grid (see Figure 49). The power deviation curve, the input to frequency response model, is derived based on following approximations: 1). For a fault on the two directly connected lines, the post fault power injection to the AC grid is assumed to be restored to the set-point of the converter connected to the healthy line. 1) For a fault on the remaining lines, the power injection in AC grid is assumed to be restored to the pre-fault value. 2) For the converters located away from the fault, the power restoration time is assumed as zero in case of selective strategies.

The frequency violation duration for all strategies are shown in Table 39. With equal power restoration times for nonselective strategies, the duration of violation is equal among them. The same can be observed for two FS strategies even though the power restoration times are different for both.

Next, the reserve size is calculated. All the assumptions made in section 5.6.1 are maintained except the activation time of reserves that is assumed to be 0.2 sec to offer more realistic representation of the reserves. This time is determined based on N-1 analysis for 500 samples (generation and power injection scenarios) and finding the lowest time to frequency nadir.

Page 111: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

110

Table 39: KPIs of frequency stability and reserve requirement

INDICATOR FS - FDCCB FS-SDCCB NS- FB NS-CB

Heq,avg (sec) 1.36

ROCOF (Hz/s) 0.63 0.63 1.06 1.06

Fmin (duration/yr) 20.2 20.2 54.6 54.6

FFR requirement (MW) ~175 ~175 ~435 ~435

Table 39 shows the FFR requirement for the backbone grid in year 2050 (also in Figure 79). The acceptable load shedding requirement is set to 3 hr/year. The protection times do not contribute much, hence the load shedding duration vs FFR graph is similar among the protection groups. The selective strategies need FFR capacity of 175 MW to meet the requirement. However, the non-selective strategy cannot meet the requirement even with 450 MW reserves of or higher. The nonselective strategy requires even the faster reserves than 0.2 sec to remove the remaining violations. Figure 79 (b) shows the FFR reserve requirement after the reserve activation time is reduced to 0.1 sec. The selective and nonselective strategies require ~175 MW and ~435 MW of reserves to meet the requirement.

(a)

(b) Figure 79: FFR requirement for planning year 2050 (a) activation time = 0.2 sec (b) activation time = 0.1 sec

Page 112: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

111

In summary, the duration of frequency limit violations are different for selective and non-selective strategies. The non-selective protection strategy results in more annual frequency violations due to two reasons: 1) higher number of frequency minimum instances due to high temporary power loss 2) higher probability of event occurrence as for fault at any location in the DC grid will disconnect the whole DC grid temporarily. We assume that for the fault on the remote location (not on the line connected to the AC grid), the total pre-fault power injection is restored causing a small amount of violations from remote faults. However, if the power is not restored fully, the non-selective protection strategy will cause even a higher load shedding.

The load shedding duration is highly dependent on both the power restoration time and the amount of power deviation. Out of three cases analysed so far, the load shedding vs FFR curves were different for WP 4.3 (section 5.6.1) – small grid, the curves were identical among FS group of protection strategies for the generic example (section 5.6.1) and the curves were similar among the FS and NS groups of protection strategies in this section. Therefore, a generic conclusion that the power restoration time is less determining parameter, cannot be drawn. But a generic conclusion that the FS protection strategies require the lower amount of FFR reserves that the NS protection strategies, holds true in all cases. A detail study of each AC/DC grid topology is required to find the difference in FFR requirement among protection strategies.

Page 113: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

112

7 CONCLUSION

This deliverable presents the contributions of Work Package 4.5 of PROMOTioN project’s Work Package 4. It proposes a Cost-Benefit Analysis (CBA) approach from a DC protection point of view, i.e. to investigate the impact of DC grid protection and different DC protection strategies on the overall CBA. The main objective is to investigate to what level of detail DC grid protection needs to be integrated in CBA studies, in particular for the studies performed within the PROMOTioN project, on the topologies developed in WP12. The reasons to specifically consider DC protection in the CBA are as follows:

• The anticipated component cost of DC protection equipment, particularly DC Circuit Breaker and converters with fault current blocking capability, is significantly higher than that of AC protection equipment,

• Different protection strategies are considered, which result in fundamentally different protection system layouts and equipment requirements, resulting in significant differences in DC side protection costs,

• The impact of DC faults on the overall AC/DC system stability might be more important, depending on the protection strategy used and it needs to be assessed.

The proposed CBA approach is a complement to the CBA implemented in WP12 and proposes an assessment/comparison framework which enables to “compute a cost to performance”. As such, the work in this deliverable makes it possible to discriminate protection schemes from a cost and performance trade-off vantage point, within a given AC/DC grid setup.

The contributions of this report are:

• A generic methodology for DC protection CBA is defined and implemented in dedicated tools. This methodology relies on the computation of a dedicated set of Key Performance Indicators (KPIs) to support protection strategies assessment and comparison. The proposed generic methodology can be applied to a large range of DC grid case studies. .

• Different protection strategies are compared, showing deviations in terms of Capital Expenditure (CAPEX), Operational Expenditure (OPEX) and performance.

The proposed methodology is structured around economic and technical key performance indicators. Economic KPIs are related to the CAPEX and OPEX associated to protection systems as well as to the CAPEX and OPEX of the overall grid, using the same economic KPIs. Technical KPIs are related to efficiency KPIs (fault interruption time, voltage restoration time, active power restoration time, reactive power restoration time and transient energy imbalance), failure KPIs (primary sequence failure probability and protection strategy failure probability) and AC impact KPIs (frequency stability, fast frequency reserve requirement, redispatch costs).

Simplified DCCB design methodologies, applicable to any DC grid, have been proposed in order to define a rough specification of DCCBs involved in a given protection strategy. They enable the assessment of the DCCBs main specifications without running EMT simulations. Such specifications are mandatory in order to be able to compute the costs of DCCBs. Several types of protection strategies are addressed: Fully-selective using fast DCCB (named FS-FDCCB) or fully-selective using slow DCCB (named FS-SDCCB), non-selective using (slow) DCCBs at the converter terminals (named NS-CB) and non-selective using converter with fault current blocking capability (named NS-FB).

DC circuit breaker parametric cost models have been developed for both fast (so-called hybrid-type technology) and slow (so-called mechanical type technology) DCCBs. These DCCB models include costs related to the number of breaking cells, DC reactor and surge arresters for energy dissipation as well as an estimated installation and commissioning cost. It is shown that the cost of the DC reactor is not negligible. Additional costs due to installation of DCCB on offshore platform are also estimated. Depending on volume and size of DCCBs, these platform costs have a significant contribution to a total cost. These models are not presented in the present report. More details can be found in [7].

Page 114: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

113

Dedicated tools have been built to implement the proposed methodologies and associated KPIs. Those tools define a general workflow for the evaluation of protection strategies. As such, they facilitate the application to a large range of DC grid case studies.

The proposed methodology has been illustrated on two specific case studies consisting of HVDC grids of different sizes: The small 4 terminal meshed DC grid case study proposed in the framework of WP4.1-3 and a large 35 terminal radial backbone grid proposed by WP12. Preliminary EMT simulations on a large radial DC grid has demonstrated that the considered protection strategies are applicable for such a DC grid, with stability of the DC grid being ensured before and after the fault clearing and grid restoration process.

The main conclusions and learnings from the case study analysis are summarized here under in the key takeaways section. Note that these takeaways stand only for offshore DC grids with cables.

KEY TAKEAWAYS RELATED TO COSTS/PERFORMANCE ANALYSIS OF PROTECTION SYSTEMS

Methodologies and tools:

Methodologies to support a CBA assessment of protection systems for DC grids have been proposed and developed. Those methodologies cannot be classed under the classical CBA approach but more as risk-based approaches. As such, they are considered as an efficient way to complement CBA analysis as it is performed in WP12. A set of economic and technical key performance indicators are defined. Those indicators have been demonstrated as relevant to distinguish between protection systems from both cost and performance perspectives.

These methodologies are implemented in several tools or modules which embed the models and parameters to compute the KPIs. The generality of those tools enables a wide range of application, as is illustrated through two analysed case studies.

Capital expenditure (CAPEX):

The CAPEX of protection systems using DC circuit breakers is dominated by DC circuit breakers CAPEX, including DC reactors and surge arresters for energy dissipation. Additional costs due to the installation of DCCB on offshore platform are not negligible. The CAPEX for NS-FB strategy needs to consider the additional costs incurred by such converters (i.e. difference in cost between half-bridge and full bridge converters). Generally, the required CAPEX is lower for systems with fewer DCCBs. Lower CAPEX is associated with NS-CB strategy, whereas higher CAPEX is related to FS-FDCCB strategy. The CAPEX between those two strategies varies by approximately a factor two. This is observed for both DC grid case studies (small size DC grid and large size backbone DC grid). CAPEX related to other protection strategies are within this range. The CAPEX of NS-FB strategy is in the same range that the CAPEX of the FS-SDCCB strategy. From CAPEX point of view, there is an advantage for non-selective protection system using DCCB at converter terminals.

Operational expenditure (OPEX):

Additional losses due to the choice of protection system seem to be negligible compared to the losses incurred within cables and converters. Indeed, the contribution of the protection system to total system losses is observed to be less than 4% (i.e. the losses related to the protection system is lower than 4% of the total system losses). However, in case of NS-FB strategy, additional converter losses shall be considered (e.g. the total system losses increase from 2.5% to 3% of the total system maximal generation if full bridge converters are used, for the large DC grid case study).

Maintenance costs are estimated as being proportional to CAPEX. As such, contribution of protection system to total maintenance costs is similar as contribution to total CAPEX, i.e. in the two case studies between 4% and 9%.

Expected Energy Not Transmitted (EENT) is here computed as the generated offshore wind power which could not be transmitted due to the unavailability of grid components. As such, it can be seen as an estimation of

Page 115: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

114

curtailed wind power generation. EENT due to protection systems is not insignificant but appears to be quite low compared to the total EENT incurred by failures of other grid components (i.e. cables and converters). The maximal EENT due to protection system failures does not exceed 10 % of the total EENT.

EENT due to the NS-CB strategy is slightly higher than for other protection strategies. This can counterbalance the CAPEX benefit of such strategy in case of DCCBs with high failure rate.

Transient frequency constraint violations incurred during DC fault clearing and DC grid restoration processes can be counterbalanced by using fast frequency reserves (FFR). It seems that protection strategies with long active power restoration time potentially require more FFR and would then incur additional OPEX.

From OPEX consideration, it seems that both non-selective protection strategies incur higher OPEX. The NS-CB strategy exhibit more EENT contribution, the NS-FB strategy has higher losses and both can potentially require more FFR.

Total cost:

For both case studies, the cost (CAPEX and OPEX) of protection systems never exceeds 9% of the total grid cost in the two considered case studies. The range of contribution of the protection system cost to the total grid cost varies from 5% (for NS-CB strategy) to 9% (for FS-FDCCB strategy). While not negligible, the protection system cost is not seen as prohibitive in respect to the total grid cost. As a consequence, it seems that no protection system is considered inapplicable or impossible due to its excessive cost.

Technical Key Performances Indicators (KPIs):

Computation of technical KPIs has demonstrated that the impact on both DC and AC grids depend on the choice of protection strategies. As such, the computation of technical KPIs is fundamental when assessing the performance of protection systems. This has been shown during WP4.3 works and is here confirmed for the large size backbone radial DC grid.

In the case of the large size radial backbone DC grid, it is shown that, if well-adapted converter controls are implemented, fully-selective protection strategies will have less grid impact. On the other hand, non-selective protection strategies result in a de-energization of the full grid, which could be unacceptable regarding the stability constraints of the connected AC systems. However, this last statement has to be analysed on a case by case approach depending on the specific characteristics of these AC systems.

For large DC grids, the grid splitting design could be a solution to reduce grid impact when implementing non-selective protection strategies. At a quite low additional cost (i.e. cost of some additional fast DC circuit breakers used as a firewall between different zones), grid impacts can be significantly reduced. Optimal grid splitting will then be driven by defining the number and location of areas which result in acceptable grid impacts from an AC grid side perspective. It is not intended in this deliverable and within WP4.5 to perform a full demonstration of a grid splitting based protection system. Indeed, this would require to develop a deep analysis of protection system (protection algorithms, component specification and performances, primary and back-up sequences ...) and is out of the scope of the WP4.5 work.

Some elements to support selection of a protection strategy From the knowledge acquired within the PROMOTioN project that can be found within this report and the deliverables [1] and [3], it can be highlighted some elements for supporting protection strategy selection:

1. Impact on DC grid

Fault interruption time can vary between few ms for the FS-FDCCB strategy up to 10-20 ms for the FS-SDCCB and the NS strategies. For FS strategies, a lower fault interruption time brings benefits in terms of fault current stresses on DC grid equipment (converters, cables, DCCB), but it requires more complex and costly DCCB whereas a higher interruption time leads to an upsizing of DC reactor (DCR). For NS strategies without DCR at line ends, the inrush currents coming from the discharge of the adjacent cables flowing into the faulty line could cause very high short time currents (so-called rated short time withstand current) that need to be taken into account during the design of the equipment. In case of large DC grid, these inrush

Page 116: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

115

currents could lead to some technical limitations for the DCCBs technology. Adequate solutions would then to be defined to limit such inrush currents.

Time to restore power can vary from several tens of ms up to 100-200 ms depending on the strategies and their backups. Converter control performance plays an important role and coordination between converter controls is necessary in order to ensure a fast restoration time. Indeed, in FS strategies the presence of line inductors induces power oscillations that need to be damped by dedicated converter controls. In NS strategies the major difficulty is related to the coordination of the power ramp-up among all converters after a temporary stop.

2. Impact on AC grid

DC grid protection strategies could impact the frequency stability and the transient (voltage and rotor angle) stability of the AC transmission systems surrounding the DC network due to the temporary stop of the DC power flow that can reach values of several GW for 100-200 ms. The impact on AC system stability is strongly dependent on the exchanged power prior to fault inception and on the value of the system inertia as summarised in Table 40. It is important to note that due to the aggregated contribution of physical inertia (e.g. synchronous generators), synthetic inertia and machines with fast frequency response, the concept of inertia may change in the next future. Hence, the actual behaviour of future systems with low inertia is not yet really known.

Table 40 : Impact of protection strategy on AC system stability, influence of System Inertia

High value of System Inertia (H>2-4s) with

reasonable DC power exchange

Low value of System Inertia (H<2-4) and high DC power exchange

Rotor angle transient stability

No major impact

Critical Time to Return to Operation5 [37] could be exceeded when employing NS strategies if AC system is operated close to its stability limits. FS strategies could also entail transient instability during backup sequence. Temporary compensation of power loss could be a solution to avoid instability.

Frequency stability No major impact

Frequency limits can be exceeded when employing NS strategies. FS strategies could also entail frequency instability during backup sequence.

A precise assessment of the AC stability issues in Table 40 requires detailed data about AC system connected to the DC grid, specifically including detailed generator models.

3. A non-selective protection strategy using AC DCCB exhibits very large time to restore active power (i.e. >1 s). Considering potential impact on AC system stability, this protection strategy is discarded for most of the AC/DC systems.

4. Non-selective protection strategies (NS-CB or NS-FB) using high speed switches (HSS) to isolate the faulty line show quite high time to restore active power (> 400 ms) and can be considered as applicable to only a limited number of AC/DC systems due to AC system stability issues. For large DC grids, it seems that these strategies cannot be applied. Replacing HSS with DCCBs at each line end has been shown to divide the time to restore active power by more than a factor 2. As such, these strategies seem to be applicable to small size to medium size DC grids, mainly depending on AC system stability requirements. Their application to a large size DC grid is even possible but a special attention must be paid to potential AC system stability issues. In addition, implementing a DCCB at each line end allows efficient protection back-up sequences.

5 Critical Time to Return to Operation is the time beyond which the stability of the AC system cannot be restored after a contingency and the AC system loses synchronism. It is not constant and mainly depends on AC system inertia, pre-fault AC system condition (close or not to its stability limits) and the exchanged power prior to fault inception.

Page 117: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

116

5. NS-CB strategy exhibits a gain in protection system CAPEX. As such, it can provide a recommended cost-effective solution when the impact on AC system stability is acceptable.

6. CAPEX and OPEX (mainly additional losses due to FB converter) related to NS-FB strategy are significantly higher regarding NS-CB strategy, with quite similar performance (observed here only on the small DC grid benchmark). As a consequence, it seems that NS-FB strategy would not be the preferable one in many cases.

7. Both fully-selective protection strategies have a lower grid impact and seem to be preferable for large DC grid application. Considering the large DC grid case study, it is not obvious to select between using fast DCCBs or slow DCCBs. Using fast DCCBs leads to the best performance but at the highest cost. Using slow DCCBs leads to a slightly degraded performance with a real cost saving (about 25%) for the protection system (but not as high as expected mainly due to bulky and heavy DC reactors and associated energy dissipation devices (surge arresters) that are required with slow DCCBs and which incur significant additional offshore platform CAPEX). This solution might be considered as a good trade-off between cost and performance. However, slow DCCBs results in implementing quite large DC reactors (around 300 mH at each DCCB location compared to 140-200 mH with fast DCCBs), the impact of which on control performance shall be fully analysed for a large range of fault location scenario.

8. For large DC grids, the grid splitting design could be a possible solution. It allows to reduce grid impact when implementing non-selective protection strategies at a quite low additional cost (i.e. cost of some additional fast DC circuit breakers used as a firewall between different zones), grid impacts can be significantly reduced. Grid splitting could also enable to implement different protection strategies in different zones of the grid and therefore facilitate the HVDC grid extensibility. It is important to note that such solutions are very case specific.

ADDITIONAL CONTRIBUTIONS TO SUPPORT DEPLOYMENT PLAN FOR FUTURE EUROPEAN OFFSHORE GRID WITHIN WP12

The EENT indicator, which doubles up as an indicator for wind power generation curtailment, has been used for assessing two meshed variations of the large size radial backbone grid, as well as a backbone grid structure without country interconnectors. It is shown that adding country interconnectors results in a significant reduction of curtailed wind generation. Meshed options also bring a gain in terms of EENT. However, the additional EENT benefits of such meshed grid options needs to be balanced against the additional CAPEX incurred by the implementation of new lines. Note that for a full assessment of EENT, it should be necessary to also consider the impact on cross border exchanges. This has not been done within the WP4.5 scope due to the lack of AC system physical and market data.

Page 118: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

117

8 RECOMMENDATIONS FOR WP 12

This section is dedicated to recommendations for protection systems design, costs and performance. It is expected that these recommendations support the definition of the deployment plan for future European offshore grid performed within WP 12.

A large set of protection strategies are technically applicable to small size and large size DC grids

It has been shown, from the implemented case studies, that both fully-selective and non-selective protection strategies can be applied to both small size and large size DC grids. The performance of the protection strategies does not only depend on DC circuit breakers and protection equipment performance but also highly relies on converter control behaviour. This is essential for ensuring a fast grid restoration after fault clearing and line isolation. More particularly, control shall be able to deal with the presence of large DC reactors in case of fully selective protection strategies. If not, a significant increase of time to restore the grid can be observed.

There are no preferred protection strategies suitable for all DC grid systems

There is not one generally best performing strategy to protect DC grids and the selection of the most suitable protection system will depend on many parameters. The first objective of a protection system is to ensure system security as well as to guarantee system stable operation during and after fault clearing process. Cost considerations are then addressed. To this end, it is important for a grid developer to have the opportunity and means to assess different protection system options through a set of consistent indicators, in order to select the most appropriate protection system at the right cost. Although not investigated in detail in this deliverable, the authors do not see a reason why a mix of protection systems cannot exist (potentially requiring some additional engineering).

However, considering the case study presented in this report, it seems that fully-selective or non-selective strategies based on grid splitting could be preferable for large size DC grids. Indeed, grid impacts, captured through technical KPIs, are clearly reduced with such strategies when primary protection sequences act. As such, it seems that a grid splitting design could be a good way to use non-selective protection strategies for large grids. To fully confirm such a trend, it would be necessary to include the impact of back-up protection sequences on technical KPIs, which has not been done in the present work for the large DC grid case study.

Protection system design needs a full assessment and no short cut to design a protection system can be applied.

It has been demonstrated through the entire WP4 that protection system design is not a straightforward task. Many parameters have to be considered. Both technical performance and costs have to be addressed and relevant key performance indicators need to be defined and used to perform a full assessment of protection system. KPIs shall be carefully considered for appropriate selection of protection strategies. As such, updating and adapting relevant KPIs is a good way to include new constraints such as new grid codes requirements.

Methodologies and tools developed throughout WP4.3 and WP4.5 define a relevant framework and can be seen as efficient means of design support. As such, they form a general workflow for the evaluation of protection systems and will facilitate protection system design for new DC grid case studies.

Designing protection systems will then result in protection system components specification. Specification of DC circuit breakers will play a central role regarding the cost of protection systems as it has been shown in this report that costs are mainly DCCB driven.

Protection system costs are mainly driven by the cost of DC circuit breakers which can vary in a wide range

DCCBs may have different costs. From the case studies addressed, it can be observed that the cost of DCCBs varies approximately with a factor of two, depending on the breaker requirements. The CAPEX required for non-

Page 119: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

118

selective using converters with fault blocking capability protection strategy needs to consider the additional costs incurred by such converters (i.e. difference in cost between half-bridge and full bridge converters).

From cost considerations, the general transmission system (DC grid) layout is not so much affected by the selection of protection system

While not negligible, protection system cost is not seen as prohibitive in respect to total grid cost. This has been observed for both small size and large size DC grid case studies. As a consequence, it seems that no protection system is impossible due to its excessive cost. However, as far as costs could vary in a factor of 2, it is important to select the right protection strategy at the right cost. That could imply a significant cost saving (some percent of the total grid cost).

Page 120: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

119

APPENDICES

A.1 DATA AND PARAMETERS FOR CASE STUDIES

Table 41: HVDC submarine cable resistance data with WP4.3 DC grid benchmark

Resistance 0.032 Ω/km

Table 42: Components’ reliability data used for WP4.3 and WP12 DC grid case studies

ONSHORE OFFSHORE

MTTF (hours) MTTR (hours) MTTF (hours) MTTR (hours)

Cables - - 219150 2316

Converters 6257 48 6257 6

Switches (HSS) 160000 48 160000 6

DCCBs (Hybrid/Mechanical) 80000 48 80000 6

Table 43: AC CB failure rate and unavailability parameters from literature review FAILURE RATE

(/YEAR) REPEAR TIME

(HOURS) UNAVAILABILITY

(HOURS) UNAVAILABILITY REFERENCE

0,01 200 2 0,000228 [30] 0,0121 200 2,42 0,000276 [31] 0,0281 312 8,7672 0,001 [31] 0,003 200 0,6 6,845E-05 [32]

0,01654 200 3,3083 0,000377 [33]

A.2 FREQUENCY RESERVE BASICS

The classification of operating reserves differ in the literature but in general, they can be classified in three groups: 1) Frequency Containment Reserves (FCR) corresponding to Primary Frequency Control (PFC) 2) Frequency Restoration Reserves (FRR) corresponding to Secondary Frequency Control 3) Frequency Replacement Reserves (RR) corresponding to Tertiary Frequency Control.

Figure 80 shows the use of these reserves in a frequency control action in CE synchronous area. If an imbalance occurs in a grid, the PFC acts to recover the frequency to a permissible limit using FCR. The response time of these reserves is within 30 sec. The frequency will still be away from its nominal value. Then, the secondary control brings back the system frequency to its nominal value and re-establishes the power exchange with adjacent control areas to their pre-defined values. It also ensures that the full FCR is available for next contingency by freeing them. The FRR are utilized during this action that are activated roughly between 30 s and 15 min following a disturbance. The tertiary control, supported by RR, also works roughly in the same time frame. The aim of this control is to free up all the utilized FRR reserves so that the secondary control is ready to tackle the next disturbance.

Page 121: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

120

The interest of this work is to define the reserve requirement to maintain the frequency stability in case of a contingency. The FCR is preliminarily responsible for this function. The capacity of FCR is defined in different manners in each synchronous area but at fundamentals, it is calculated based on a reference incident, which represents the worst-case scenario for the system. The dimensioning of FCR in the Nordic grid is explained here since it is used for the analysis in later chapters. The FCR is made of two parts in the Nordic system:

• FCR-N (Frequency containment reserves for normal operation) and • FCR-D (frequency containment reserves for disturbance)

Figure 80: Frequency control action within CE [38]

FCR-N is activated during normal fluctuations in load and generation. It is fully activated at frequencies outside the permissible range of 49.9 Hz to 50.1 Hz. The sizing of FCR-N depends on the power/frequency characteristic of the synchronous area. It is equal to 6000 MW/Hz for the Nordic grid. The required FCR-N for permissible ± 0.1 Hz frequency deviation amounts to 600 MW (= 6000 MW/Hz * 0.1 Hz).

FCR-D is activated in response to a disturbance in a synchronous area (e.g. disconnection of a generator). The sizing of FCR-D is calculated based on the reference incident that can cause the largest possible imbalance of in the grid. Such incident can include 1) a loss of generator 2) a loss of HVDC interconnector 3) a loss of load facility. In the Nordic grid, the required size of FCR-D is the largest loss of active power minus 200 MW that is supported by load-frequency dependency. The largest imbalance in a synchronous area is called “dimensioning incident”. The current dimensioning incident in the Nordic grid is the loss of Oskarshamn 3 nuclear power plant in Sweden with 1450 MW capacity.

A.3 INERTIA VALUES FOR DIFFERENT GENERATION TYPE

Table 44: Inertia calculation based on generation type

Plant type Typical inertia value Hydro 3

Nuclear 6,3 Thermal 4

Wind 0 Solar 0

Page 122: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

121

Table 45: Typical inertia values for different types of power plants

Types of power plants Typical inertia value (sec) Gas

4

HardCoal Oil

Lignite Biofuels

Oil_2045_GCA Other non-RES

Nuclear 6.3 Solar PV

0 Solar thermal wind offshore

0 wind onshore

wind OPF Other-RES

Hydropump

3 Hydrorun Hydroturbine

A.4 OPERATION OF PROTECTION STRATEGIES

FULLY SELECTIVE FAULT CLEARING WITH FDCCB AND SDCCB The fully selective clearing strategy uses DCCB (mechanical or hybrid) as the main protection while the backup protection can be provided by one of three options: 1) using converter AC breakers 2) using fault blocking converters 3) using DCCB at the converter terminal. The third option is used for the CBA analysis in this work.

= ~

= ~

=

~

=

~

Converter 1

Converter 2

Offshore windfarm 1

Offshore windfarm 2

Converter 3

Converter 4

FCUC1BrAC3

BrAC4

Line L1

Line L2Line L3

Line L4

Bus 3Bus 1

Bus 2 Bus 4

FCUC2

FCU1

FCU3

FCU5

FCU6

FCU7

FCU2

FCU4

FCU8

FCUC4

Onshore AC Grid 1

~

~

BrAC1

BrAC2

Onshore AC Grid 2

BrLBus Line

HSS/RCB

FCU

Figure 81: (a) Fully selective protection strategy with backup options (b) Fault clearing unit (FCU) consisting of inductor (L), high-speed switch or residual current breaker (HSS/RCB) and DC circuit breaker (Br) [1]

Page 123: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

122

Table 46 shows the responsible breakers for clearing the fault in primary and backup sequences. In the primary protection sequence, the line breakers on the faulted line would open to clear the fault. If a line breaker fails (marked in red), the adjacent converter terminal DCCB would open to clear the fault in the backup sequence.

Table 46: Simplified protection matrix for primary and backup protection [1]

Fault on Primary Backup L1 Br1, Br 2 Br 1, Br 2, HSS1 L2 Br 3, Br 4 Br 3, Br 4, HSS3 L3 Br 5, Br 6 Br 5, Br 6, HSS5 L4 Br 7, Br 8 Br 7, Br 8, HSS7

The protection matrix here is modified from D4.2 in order to represent only the post fault steady state condition. There would be more elements disconnected temporarily in order to recover the power flow after a DC fault. In the CBA methodology explained in section 5.6.2, the above protection matrix is used at step 2 to determine what part of DC grid to be disconnected. This would decide post fault AC/DC network for the power flow analysis. For example, if Br 1 fails (part of FCU1), the neighbouring breakers will open (Br 3, Br 5, BRc1) to limit the fault current and finally, opening HSS1 (part of FCU1) will isolate the fault.

Out of two types of breakers that can be used for fully selective strategy, the use of fast (hybrid) DCCB would result in a shorter power restoration time compared to use of slower (mechanical) DCCB for the same fault location and same system configuration. On the other hand, mechanical DCCB would be less expensive solution than hybrid DCCB.

NON SELECTIVE FAULT CLEARING WITH FULL-BRIDGE MMC (CONVERTER FAULT CURRENT LIMITING) This protection strategy uses full-bridge MMCs to control the fault current contribution from AC grid during the fault. In the primary protection sequence, all converters are blocked/controlled to reduce the DC fault current close to zero. The high speed switch (HSS) then opens the faulted line by interrupting the residual current. If a high speed switch fails to operate, the fault is cleared in backup sequence by opening the adjacent HSSs.

= ~

= ~

=

~

=

~

Converter 1

Converter 2

Offshore windfarm 1

Offshore windfarm 2

Converter 3

Converter 4

BrAC3

Line L1

Line L2Line L3

Line L4

Bus 3Bus 1

Bus 2 Bus 4

HSS1

HSS3

HSS5

HSS6

HCU7

HSS2

HSS4

HSS8

BrAC2

Onshore AC Grid 1

~

~

BrAC1

BrAC2

Onshore AC Grid 2

Figure 82: Table 47 Non-selective protection strategy with full bridge MMCs [1]

The protection matrix that is used in the CBA methodology is given in the table below. As for selective strategy, the protection matrix from D 4.3 is modified to represent only the steady state condition after the fault clearing.

Table 48 Simplified protection matrix for primary and backup protection [1]

Fault on Primary Backup (Blk = permanently blocked) L1 HSS1, HSS2 HSS1, HSS2, HSS3, HSS5, Blk-C1 L2 HSS3, HSS4 HSS3, HSS4, , HSS1, HSS5, Blk-C1 L3 HSS5, HSS6 HSS5, HSS6, HSS1 , HSS3, Blk-C1 L4 HSS7, HSS8 HSS7, HSS8, HSS6, Blk-C2

Page 124: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

123

NON SELECTIVE FAULT CLEARING USING CONVERTER BREAKER For this protection strategy, the line and converter breakers are mechanical DCCBs. In the primary protection sequence, all converter breakers opens on the fault detection and suppresses the contribution from AC side. Then, the fault is isolated by opening the line breakers and HSSs of the faulted line. The power flow can be restored by reclosing all converter breakers. If a converter breaker fails to open, the line breaker still breaks the fault current (still below the line breaker capacity due to opening of remaining converter breakers). Then, the HSS in series with the failed breaker opens and the closing of healthy converter breakers would restore the power flow.

= ~

= ~

=

~

=

~

Onshore AC Grid 1

Converter 1

Converter 2

Offshore win

Offshore win

Converter 3

Converter 4

~

~

FCUC1BrAC1

BrAC2

BrAC3

BrAC4

Onshore AC Grid 2

Line L1

Line L2Line L3

Line L4

Bus 3Bus 1

Bus 2 Bus 4

FCUC2

FCU1

FCU3

FCU6

FCU7

FCU2

FCU4

FCU8

FCUC4

BrBus Line

HSS/RCB

FCU

Figure 83: (a) Non-selective protection strategy using converter breaker (b) Fault clearing unit (FCU) consisting of high-

speed switch or residual current breaker (HSS/RCB) and DC circuit breaker (Br) [1]

The protection matrix is given below.

Table 49: Simplified protection matrix for primary and backup sequences

Fault on Primary Backup L1 Br1, HSS1, Br2, HSS2 Br1, HSS1, Br2, HSS2, Bc1, HSSc1 L2 Br3, HSS3, Br4, HSS4 Br3, HSS3, Br4, HSS4, Bc1, HSSc1 L3 Br5, HSS5, Br6, HSS6 Br5, HSS5, Br6, HSS6, Bc1, HSSc1 L4 Br7, HSS7, Br8, HSS8, Br7, HSS7, Br8, HSS8, Bc2, HSSc2

Page 125: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

124

A.5 COMPARISON BETWEEN ANALYTICAL AND SIMULATION APPROACH FOR BREAKER SIZING

Figure 84 shows the procedure used to compare the analytical approach with the simulations.

1) Figure 84: Procedure to compare analytical and simulative approach for DCCB sizing

The PSCAD model from D4.2 is repurposed here. At first, the DC reactor value is calculated using equation 3-15 based on different values of 𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛 and 𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛. The obtained 𝐿𝐿𝐷𝐷𝐵𝐵 and assumed 𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛 are used to perform the simulations. The DC line faults at each line end is simulated and following points are analysed:

• If the maximum fault current, 𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 is below the breaker’s interruption capacity, 𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛 • The difference of calculated surge arrester energy 𝐸𝐸𝑆𝑆𝑀𝑀 from the simulation output

The results for the DC fault on L12 near converter 1, which leads to the highest fault current and surge arrester energy, are represented in Table 50.

Table 50: Comparison between analytical and simulative approach (the available market ratings are marked in blue [25] [26])

Calculation Simulation 𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛 (𝑘𝑘𝐸𝐸) 𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛 (𝑚𝑚𝑚𝑚) 𝐿𝐿𝐷𝐷𝐵𝐵 (𝑚𝑚𝐻𝐻) 𝐸𝐸𝑆𝑆𝑀𝑀 (𝑀𝑀𝑀𝑀) 𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 (𝑘𝑘𝐸𝐸) 𝐸𝐸𝑆𝑆𝑀𝑀 (𝑀𝑀𝑀𝑀)

7 2 125 9,2 6,3 7,4 7 3 187 13,8 6,4 10,5 7 4 250 18,4 6,4 13,51 9 2 90 10,9 7,7 7,77 9 3 135 16,4 7,4 10,9 9 4 180 21,8 7,8 13,9 12 7 221 47,8 9,5 24,1 12 8 253 54,6 9,8 27,3 12 9 284 61,4 9,6 30,7 14 7 185 54,3 11,3 24,77 14 8 211 62,1 11.20 28 14 9 238 69,8 11.17 31,4 16 7 159 60,9 12,3 26,1

16 8 181 69,6 12,3 29,5

16 9 204 78,3 12,2 32,8

The highest fault current during simulation is below the interruption capacity of breakers for all cases. It indicates that the calculated DC reactor is able to limit the fault current within the breaker rating. However, the calculated 𝐸𝐸𝑆𝑆𝑀𝑀 is higher than the simulation ones in all cases. The difference between two is significant for higher 𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛 and 𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛. In the next step, the influence of this difference on the total breaker cost is analysed in order to determine if the analytical approach can provide the total cost within an acceptable tolerance.

Figure 85 shows the percentage increase in DCCB cost due to higher calculated 𝐸𝐸𝑆𝑆𝑀𝑀. The difference is small for hybrid DCCB since 𝐸𝐸𝑆𝑆𝑀𝑀 requirement is quite small for lower 𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛 and 𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛 and the SA cost will form only a small percentage of the total breaker cost. The cost of SA is less than 1 % of the total HDCCB cost in most cases. For

calculate required 𝐿𝐿𝐷𝐷𝐵𝐵 based on

assumed 𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛, 𝑑𝑑𝑏𝑏𝑔𝑔𝑛𝑛

use calculated 𝐿𝐿𝐷𝐷𝐵𝐵and assumed Δ𝑑𝑑 for

simulations

Simulate for different fault

locations

Compare 1) 𝑖𝑖𝑜𝑜,𝑝𝑝𝑚𝑚𝑚𝑚𝑛𝑛 (simulation) vs 𝑖𝑖𝑏𝑏𝑔𝑔𝑛𝑛 (assumed)2) 𝐸𝐸𝑆𝑆𝑀𝑀 (calculation vs simulation)

Page 126: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

125

mechanical breaker (7 ms and higher), the calculated cost can be up to 6 % higher. This is due to a higher SA cost for mechanical breakers which is almost 4 – 7 % of total MDCCB cost for the above ratings.

Figure 85 Difference in DCCB cost due to difference in 𝐸𝐸𝑆𝑆𝑀𝑀

A.6 DCCBS SHORT CIRCUIT CURRENT ESTIMATION

The analytical methodology for estimating the 𝐸𝐸𝐶𝐶𝐶𝐶𝑡𝑡 is an adaptation to a MTDC grid of the approach detailed in [39]. This approach focuses on two types of VSCs: 2L-VSC and MMC. The equivalent circuit of a 2L-VSC at a line-to-line short circuit is shown in Fig. 17. The short circuit current of such a converter consists of a contribution of the bulk capacitors 𝐶𝐶𝐷𝐷𝐵𝐵 on the DC side and a contribution fed by the AC grid.

Figure 86: Equivalent circuit of a 2L-VSC at a line-to-line short circuit on the DC side

The steady-state short circuit current of a 2L-VSC is determined according to (11). After the discharge of the DC capacitors only the AC grid contributes to the short circuit current.

Page 127: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

126

0-1

where:

• 𝑍𝑍𝑀𝑀𝐵𝐵 represents the equivalent AC grid impedance, which includes the AC short circuit impedance, the transformers and inductors connecting the AC and DC grids.

• 𝑈𝑈𝑜𝑜 represents the rms value of the line-to-line AC voltage.

The peak short circuit current 𝑅𝑅𝐷𝐷 depends on the short circuit location. If the short circuit occurs next to the 𝐿𝐿𝐷𝐷𝐵𝐵 converter terminal (𝐿𝐿𝐷𝐷𝐵𝐵 < 0.2 H), the short circuit current is almost solely determined by the contribution of the DC capacitors. In this case 𝑅𝑅𝐷𝐷 can be calculated with the following approach based on the RLC-circuit on the DC side.

For DC inductances 𝐿𝐿𝐷𝐷𝐵𝐵 > 0.2 𝐻𝐻 equations 0-2, 0-3 and 0-4 can be used for the calculation of 𝑅𝑅𝐷𝐷. Uncertainties occur because of the recharging and discharging of the capacitors, when 𝑈𝑈𝐷𝐷𝐵𝐵≠0.

0-2

0-3

0-4

In case of line-to-earth short circuits the same approach according to equation 0-5 can be applied for the calculation of the peak short circuit current. The steady-state short circuit current is zero.

0-5

Note that in case of Multi level Modular Converters (MMCs), the AC impedance 𝑍𝑍𝑀𝑀𝐵𝐵 includes two parallel arms inductors (i.e. includes half of arm inductor). This AC impedance short circuit (SCI) can be calculated as in equation 0-6.

𝐸𝐸𝐶𝐶𝑅𝑅 =3 ∗ 𝑈𝑈𝑜𝑜2

𝐸𝐸𝐶𝐶𝐶𝐶 0-6

where:

• SCC is the AC short circuit capacity

For the line converters short time withstand current, this is estimating using the cable characteristic impedance 𝑍𝑍𝑐𝑐 as in equation 0-7.

𝑢𝑢𝑙𝑙𝑑𝑑 𝑑𝑑ℎ𝑦𝑦𝑟𝑟𝑢𝑢𝑔𝑔ℎ 𝑍𝑍𝑢𝑢𝑦𝑦𝑦𝑦𝑙𝑙𝑢𝑢𝑑𝑑 =𝑈𝑈𝑜𝑜𝑍𝑍𝑐𝑐

0-7

where: • 𝑈𝑈𝑜𝑜 represents the DC nominal voltage

A.7 STOCHASTIC EFFICIENCY KPIS AND PROTECTION SCHEME RELIABILITY

Efficiency key performance indicators (presented in section 3.3.2) simulation results in case of DCCB unavailability ζDCCB = 0.00015 and ζDCCB = 0.0009 are given in Table 51 and Table 52 respectively. Associated benchmarking are given in Table 53 and Table 54 respectively. These complementary results are given to show the impact of a higher and lower values of ζDCCB on the study curried out in 5.5.2.

Page 128: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

127

Table 51: Strategies stochastic key performance indicators for ζDCCBs =0,00015

STRATEGY IMAIN (%) IBACKUP (%) ITR_20 (%) ITR_50 (%) ITR_200 (%)

Selective strategies

FS - FDCCB 0.031 0.062 0.062 0.062 0.062

FS-SDCCB 0.031 0.062 0.090 0.062 0.062

Non-Selective strategies

NS- FB 0.031 0.062 100 0.062 0.062

NS-CB 0.031 0.074 87.50 0.074 0.074

Table 52: Strategies stochastic key performance indicators for ζDCCBs =0,0009

STRATEGY IMAIN (%) IBACKUP (%) ITR_20 (%) ITR_50 (%) ITR_200 (%)

Selective strategies

FS - FDCCB 0.185 0.064 0.064 0.064 0.064

FS-SDCCB 0.185 0.064 0.240 0.064 0.064

Non-Selective strategies

NS- FB 0.185 0.064 100 0.064 0.064

NS-CB 0.185 0.148 87.50 0.148 0.148

Table 53: Benchmarking indicators for ζDCCBs =0,00015

STRATEGY “SPEED” OF THE PROTECTION STRATEGY CLASSIFICATION

PERFORMANCE OF THE MAIN SEQUENCE CLASSIFICATION

PERFORMANCE OF THE BACK-UP SEQUENCES CLASSIFICATION

AC/DC STABILITY ISSUE RISK INDICATOR

CLASSIFICATION

Selective strategies

FS - FDCCB Very fast + + ++

FS-SDCCB Very fast + + ++

Non-Selective strategies

NS- FB fast + + ++

NS-CB fast + - ++

Table 54: Benchmarking indicators for ζDCCBs =0,0009

STRATEGY “SPEED” OF THE PROTECTION STRATEGY CLASSIFICATION

PERFORMANCE OF THE MAIN SEQUENCE CLASSIFICATION

PERFORMANCE OF THE BACK-UP SEQUENCES CLASSIFICATION

AC/DC STABILITY ISSUE RISK INDICATOR

CLASSIFICATION

Selective strategies

FS - FDCCB Very fast + + ++

FS-SDCCB Very fast + + ++

Non-Selective strategies

NS- FB fast + + ++

NS-CB fast + - ++

From analysis of the different benchmarking Table 53 and Table 54, it can be observed that the classification of the different protection strategies are only slightly modified by a variation of the DCCB unavailability ζDCCB. Modifying the value of ζDCCB would change absolute value but not really impact the relative comparison conclusions. It can then be said that the proposed study allows to carry out a comparative analysis, mainly

Page 129: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

128

considering strategies from both system architecture (layout, component) and sequences (main and back-up) flow chart perspectives.

Figure 87 shows the reliability indicators (main backup sequence performance) for non-selective and fully selective protection strategies considered within WP4.5. This figure shows that:

• All the protection strategies are quite reliable from main and failure sequence probability success.

• Fully selective FS-FDCCB, FS-SDCCB and non-selective NS-FB protection strategies have quite similar reliability.

• Non-selective protection strategy NS-CB is slightly less reliable than FS-FDCCB, FS-SDCCB and NS-FB strategies from backup sequence probability success.

Figure 87: Reliability indicators (main/ backup sequence performance and failure). for ζDCCBs =0,0003

a) FS-FDCCB, b) FS-SDCCB, c) NS-FB, d) NS-FB.

Page 130: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

129

A.8 WP 12 BENCHMARK (LINE AND NODE DETAILS)

Table 55 and Table 56 present the details of line connections and power ratings of the “backbone” grid structure (presented in 6.1). Inter symbol in “country” column refers to the interconnectors between country’s wind farms.

Table 55: Connection (cables) details for “Backbone” grid structure (bipolar).

rate length Resis-tance

country Node1 Node2

2000 380 3,31 DE DE_OFF01 Diele

2000 521 4,53 DK DK_OFF01 Askaer

1600 288 3,14 NL NL_OFF09/NL_OFF12/NL_OFF13 NL_OFF16/NL_OFF17

2000 124 1,08 NL NL_OFF09/NL_OFF12/NL_OFF13 NL_OFF20/NL_OFF22

1600 288 3,14 NL NL_OFF16/NL_OFF17 NL_OFF09/NL_OFF12/NL_OFF13

2000 126 1,10 NL NL_OFF16/NL_OFF17 Meeden

2000 126 1,10 NL NL_OFF16/NL_OFF17 Meeden

1000 103 2,24 NL NL_OFF20/NL_OFF22 Beverwijk

2000 180 1,57 NL NL_OFF20/NL_OFF22 Lelystad

1600 54 0,59 NL NL_OFF28/NL_OFF29 NL_OFF20/NL_OFF22

2000 124 1,08 NL NL_OFF28/NL_OFF29 Bleiswijk

2000 124 1,08 NL NL_OFF28/NL_OFF29 Bleiswijk

1400 264 3,83 NO NO_OFF02 Stokkeland

1600 23 0,25 UK UK_OFF31 UK_OFF16/UK_OFF20

1600 118 1,28 UK UK_OFF32/UK_OFF33 UK_OFF11

1600 408 4,45 UK UK_OFF36 Blyth

1600 49 0,53 UK UK_OFF16/UK_OFF20 UK_OFF32/UK_OFF33

1400 367 5,31 UK UK_OFF16/UK_OFF20 West Burton

1400 367 5,31 UK UK_OFF16/UK_OFF20 West Burton

1400 284 4,12 UK UK_OFF16/UK_OFF20 Grimsby

1400 284 4,12 UK UK_OFF16/UK_OFF20 Grimsby

1400 149 2,16 UK UK_OFF11 Bramford

1400 149 2,16 UK UK_OFF11 Bramford

1000 35 0,77 Inter. DK_OFF01 DE_OFF01

1000 35 0,77 Inter. DK_OFF01 DE_OFF01

2000 91 0,80 Inter. NL_OFF09/NL_OFF12/NL_OFF13 DE_OFF01

2000 141 1,23 Inter. NL_OFF09/NL_OFF12/NL_OFF13 UK_OFF36

2000 69 0,60 Inter. NL_OFF09/NL_OFF12/NL_OFF13 UK_OFF32/UK_OFF33

1000 182 3,97 Inter. NO_OFF02 NL_OFF09/NL_OFF12/NL_OFF13

Page 131: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

130

Table 56: Onshore and offshore nodes capacities (bipolar).

Onshore Node Offshore Node Rate Node name Rate Node name

2000 Diele 2000 DE_OFF01 2000 Askaer 2000 DK_OFF01 2000 Meeden 1530 NL_OFF09/NL_OFF12/NL_OFF13 2000 Meeden 1132 NL_OFF09/NL_OFF12/NL_OFF13 1000 Beverwijk 1600 NL_OFF09/NL_OFF12/NL_OFF13 2000 Lelystad 730 NL_OFF16/NL_OFF17 2000 Bleiswijk 949 NL_OFF16/NL_OFF17 2000 Bleiswijk 2000 NL_OFF20/NL_OFF22 1400 Stokkeland 750 NL_OFF20/NL_OFF22 1600 Blyth 1720 NL_OFF28/NL_OFF29 1400 West Burton 549 NL_OFF28/NL_OFF29 1400 West Burton 1388 NO_OFF02 1400 Grimsby 1575 UK_OFF36 1400 Grimsby 1150 UK_OFF16/UK_OFF20 1400 Bramford 874 UK_OFF16/UK_OFF20 1400 Bramford 990 UK_OFF11

1528 UK_OFF32/UK_OFF33 1600 UK_OFF32/UK_OFF33

A.9 WP 12 EFFICIENCY KPIS FOR FULL NON-SELECTIVE STRATEGY NS-CB

Figure 88 shows the results of the nodal KPIs including voltage, active power, reactive power time restorations and energy imbalance of the non-selective NS-CB protection strategy for fault locations F1, F2, and F5.

Page 132: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

131

Figure 88: Nodal efficiency KPIs for fault location F1, F3, F5 and non-selective converter breaker protection

strategy NS-CB. a) Voltage, b) Active power, c) Reactive power, c) Energy imbalance.

Page 133: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

132

A.10 WP 12 EFFICIENCY KPIS FOR FULLY SELECTIVE STRATEGY FS-FDCCB

Figure 89 shows the results of the nodal KPIs including voltage, active power, reactive power time restorations and energy imbalance of the fully selective FS-FDCCB protection strategy for fault locations F1, F2, and F5, while Figure 90 shows the nodal energy imbalance distributions and frequencies for the same fault locations.

Page 134: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

133

Figure 89: Nodal efficiency KPIs for for fault location F1, F3, F5 and fully selective protection strategy FS-

FDCCB. a) Voltage, b) Active power, c) Reactive power, c) Energy imbalance.

Page 135: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

134

Figure 90: Nodal energy imbalance distributions, NS-FDCCB strategy:

a) fault location F1, b) fault location F3, c) and fault location F5

Page 136: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

135

A.11 WP 12 EFFICIENCY KPIS FOR FULLY SELECTIVE STRATEGY FS-SDCCB

Figure 91 shows the results of the nodal KPIs including voltage, active power, reactive power time restorations and energy imbalance of the fully selective FS-SDCCB protection strategy for fault locations F1, F2, and F5 while Figure 92 shows the nodal energy imbalance distributions and frequencies for the same fault locations.

Page 137: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

136

Figure 91: Nodal efficiency KPIs for fault location F1, F3, F5 and fully selective protection strategy FS-SDCCB.

a) Voltage, b) Active power, c) Reactive power, c) Energy imbalance.

Page 138: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

137

Figure 92: Nodal energy imbalance distributions, NS-SDCCB strategy:

a) fault location F1, b) fault location F3, c) and fault location F5

Page 139: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

138

A.12 WP12 COMPARISON (EFFICIENCY KPIS) BETWEEN FULLY SELECTIVE FS-FDCCB, FS-SDCCB AND NON-SELECTIVE NS-CB STRATEGIES

Figure 93 and Figure 94 show the comparison between fully selective and non-selective protection strategies for fault locations F1 and F3 respectively. Cumulative probabilities distributions for time restorations (voltage, active/reactive powers) and energy imbalance are displayed in these figures.

Figure 93: Efficiency KPIs comparison between NS-CB, FS-FDCCB and FS-SDCCB protection strategies.

a) Voltage, b) active power, c) reactive power, d) energy imbalance. Fault location F1.

Page 140: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

139

Figure 94: Efficiency KPIs comparison between NS-CB, FS-FDCCB and FS-SDCCB protection strategies.

a) Voltage, b) active power, c) reactive power, d) energy imbalance. Fault location F3.

Page 141: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

140

VI. REFERENCES

[1] PROMOTioN, “Deliverable 4.2 - Broad comparison of fault clearing strategies for DC grids,” 2018. [2] PROMOTioN Workpackage 4, “Deliverable 4.3 Report on Performance, interoperability and failure modes

of selected protection methods,” 2019. [3] PROMOTioN, "Deliverable 4.3 - Report on Performance, interoperability and failure modes of selected

protection methods," 2019. [4] PROMOTioN, “Deliverable D7.3 - Economic framework for offshore grid planning,” 2017. [5] ENTSO-E, “Guideline for Cost Benefit Analysis of Grid Development Projects,” 2014. [6] PROMOTioN, “Deliverable D7.1 Cost-benefit analysis methodology for offshore grids,” 2018. [7] PROMOTioN, “DCCB cost model,” 2019. [8] WP4.5, “Overview of Models Requirements for CBA Assessment,” PROMOTioN Project, 2016. [9] PROMOTioN, “Cost data collection,” 2018. [10] WP4.1, “D4.1: Definition of test cases and functional requirements for DC grid protection methodologies,”

PROMOTioN Project, 2016. [11] PROMOTioN, “WP4.2 - Broad comparison of fault,” 2017. [12] M. Abedrabbo and et al., “Impact of DC grid contingencies on AC system stability,” 2017. [13] G. C. Tarnowski, “Coordinated frequency control of wind turbines in power systems with high wind power

penetration,” Technical University of Denmark (DTU), 2012. [14] P. Kundur, N. J. Balu and a. M. G. Lauby, Power system stability and control, New York: McGraw-hill,

1994. [15] Australian Energy Market Operator, “International review of frequency control adaptation,” New South

Wales, 2016. [16] ENTSO-e, “Frequency stability evaluation criteria for the synchronous zone of Continental Europe,”

Brussels, 2016. [17] “COMMISSION REGULATION (EU) 2017/1485 - establishing a guideline on electricity transmission

system operation,” Official Journal of the European Union, 2017. [18] T. Breithaupt, D. Herwig and L. Hofmann, “MIGRATE WP1 Deliverable D1. 1 Report on Systemic Issues,”

Bayreuth, 2016. [19] E. Ørum and et al., “Future system inertia - tech report,” ENTSOE, Brussels, 2015. [20] P. Tielens and D. V. Hertem, “Grid inertia and frequency control in power systems with high penetration of

renewables,” in Young Researchers Symposium in Electrical Power Engineering, Delft, The Netherlands, 2012.

[21] E. Ørum and et al., “Future system inertia 2 - technical report,” ENTSO-E, Brussels, 2016. [22] Q. Hong and et al., “Fast frequency response for effective frequency control in power systems with low

inertia,” The Journal of Engineering, no. 16, pp. 1696-1702, 2019. [23] S. De Boeck and et al., “Review of defence plans in Europe: current status, strengths and opportunities,”

CIGRE Science & Engineering, vol. 5, pp. 6-16, 2016. [24] A. J. Far and D. Jovcic, “Design, modeling and control of hybrid DC circuit breaker based on fast

thyristors,” IEEE Transactions on Power Delivery, vol. 33, no. 2, pp. 919-927, 2017. [25] M. Callavik and et al., “The hybrid HVDCbreaker: an innovation breakthrough enabling reliable

HVDCgrids,” ABB grid systems, 2012. [26] K. Tahata and et al., “HVDC circuit breakers for HVDC grid applications,” in 11th IET International

Conference on AC and DC Power Transmission, , Birmingham, 2015. [27] LAAS-CNRS, “Petri net analyser - Tina toolbox,” 2015. [28] J. Beerten and R. Belmans, “Development of an Open Source Power Flow Software for HVDC Grids and

Hybrid AC/DC Systems: MatACDC,” IET Generation Transmission and Distribution, vol. 9, no. 10, pp. 966-974, 2015.

Page 142: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

141

[29] H. Ergun and e. al., “Optimal power flow for ac/dc grids: Formulation, convex relaxation, linear approximation and implementation,” EEE Transactions on Power Systems, vol. 34, no. 4, pp. 2980 - 2990, 2019.

[30] C. W. A3, “Reliability of High Voltage Equipment: SF6 Circuit Breakers,” 2012. [31] C. W. A3, “ Reliability of High Voltage Equipment,” 2014. [32] A. a. M. D. a. S. C.-E. Janssen, “International surveys on circuit-breaker reliability data for substation and

system studies,” IEEE Transactions on Power Delivery, vol. 29, no. 2, pp. 808--814, 2014. [33] M. Stanek, “Model-aided diagnosis for high voltage circuit breakers,” Vienna University of Technology,

Austria, 2000. [34] S. K. MERZ, “Calculating Target Availability Figures for HVDC Interconnectors,” 2012. [35] W. Promotion, D4.1 Definition of test cases and functional requirements for DC grid protection

methodologies, 2016. [36] P. Demetriou and et al., “Dynamic IEEE test systems for transient analysis -

http://www.kios.ucy.ac.cy/testsystems/index.php/dynamic-ieee-test-systems,” IEEE Systems Journal, vol. 11.4, p. 2108, 2017.

[37] J. C. Gonzales, V. Costan, G. Damm and et.al., “HVDC protection criteria for transient stability of AC systems with embedded HVDC links,” in IET DPSP, 2018.

[38] “UCTE operation handbook: Appendix 1: Load frequency control and performance”. [39] H. B. A. R. E. B. Ricardo Vidal-Albalate, “Analysis of the Performance of MMC Under Fault Conditions in

HVDC-Based Offshore Wind Farms,” IEEE TRANSACTIONS ON POWER DELIVERY, 2016. [40] A. Schärling, “Décider sur plusieurs critères,” Presses Polytechniques, 1985. [41] M. Zeleny, “Multiple Criteria Decision Making,” McGraw-Hill, Columbia University, 1982. [42] CIGRE WG B4/B5-69, “Protection and local control of DC grids,” 2017. [43] W. Leterme and D. Van Hertem, "Classification of fault clearing strategies for HVDC grids," in CIGRE,

Lund, 2015. [44] “Challenges and opportunities for the Nordic power system - Technical Report,” Statnett, Fingrid, 2016. [45] Q. e. a. Hong, “Fast frequency response for effective frequency control in power systems with low inertia,”

The Journal of Engineering, no. 16, pp. 1696-1702, 2019. [46] Economics, London, “Final report for OFGEM and DECC - the value of lost load (VoLL) for electricity in

Great Britain,” 2013. [47] “The Value of Lost Load,” Toulouse School of Economics, 14 July 2018. [Online]. Available:

https://www.tse-fr.eu/value-lost-load#_ftn2. [Accessed 02 August 2019]. [48] “Continental Europe Operation Handbook: P1—Policy 1: Load-Frequency Control and Performance,”

Union for the Co-ordination of Transmission of Electricity (UCTE), 2009, v3.0, rev15. [49] “Supporting Document for the Network Code on Operational Security,” European Network of Transmission

System Operators for Electricity (ENTSO-E), Brussels, 2013, rev. 2. [50] PROMOTioN, “Deliverable 2.1 - Grid topology and model specification,” 2016. [51] J. Beerten, “Modeling and control of DC grids-Dissertation,” KU Leuven, 2013. [52] PROMOTioN, “D4.1: Definition of test cases and functional requirements for DC grid protection

methodologies,” 2017. [53] V. Knap and et al., "Sizing of an energy storage system for grid inertial response and primary frequency

reserve," IEEE transactions on Power Systems, no. 31.5, pp. 3447-3456, 2016. [54] ENTSO-e, “Commission regulation on frequency quality defining and target parameters - article 127,”

2017. [55] ENTSO-e, “Position paper: dispersed generation impact on Continental Europe region security,” Brussels,

2014. [56] H. Ergun and e. al., “Optimal power flow for ac/dc grids: Formulation, convex relaxation, linear

approximation and implementation,” EEE Transactions on Power Systems, vol. 34, no. 4, pp. 2980 - 2990, 2019.

[57] P. Tielens and D. Van Hertem, “Grid inertia and frequency control in power systems with high penetration of renewables,” in Young Researchers Symposium in Electrical Power Engineering, Delft, The Netherlands, 2012.

Page 143: WP 4.7 Deliverable: Preparation of cost-benefit analysis ...€¦ · PROMOTioN – Progress on Meshed HVDC Offshore Transmission Networks Mail info@promotion-offshore.net Web This

Preparation of cost-benefit analysis from a protection point of view

142

[58] M. Abedrabbo, M. Wang and et.al., “Impact of DC grid contingencies on AC system stability,” in IET ACDC , 2017.

[59] PROMOTioN Workpackage 4, “WP 4.7 Deliverable: Preparation of cost-benefit analysis from a protection point of view,” 2020.

[60] PROMOTioN Workpackage 4, “Deliverable 4.2 Broad comparison of fault clearing strategies for DC grids,” 2017.

[61] J. Gonzales, V. Costan, G. Damm and and al., “HVDC protection criteria for transient stability of AC systems with embedded HVDC links,” IET DPSP, 2018.

[62] “PROMOTioN Work Package 4.5, Deliverable D4.7: Preparation of cost-benefit analysis from a protection point of view,” 2019.