Testing of a Predictive Control Strategy for Balancing ... of a... · Testing of a Predictive Control Strategy for Balancing Renewable Sources in a Microgrid Mattia Marinelli; [email protected]

Testing of a Predictive Control Strategy for Balancing Renewable Sources in a Microgrid

Mattia Marinelli; [email protected]

Center for Electric Power and Energy

DTU Risø Campus

University of California – Santa Cruz

6th Aug 2014

Summer School US-DK

mailto:[email protected]

2 DTU Electrical Engineering, Technical University of Denmark

Outline – I part (storage control strategy and experimental validation)

I. Introduction

II. Microgrid and energy management details

Problem definition

System layout and forecast input

Energy management strategy and microgrid components

Storage sizing and controller sensitivity

III. Experimental testing results and procedures

Experimental results on 3rd July

Experimental results on 7th July

IV. Conclusions and future developments


Introduction

• In several member countries of the European Union, small scale feed-in systems have

increased in popularity due to favorable regulations.

• Due to the fact that the conventional power plants are displaced and thus eventually shut

down, DER plants (including PV and wind) will be required to provide a predictable

production plan and to be able to grant it, even if the meteorological conditions differ

from the forecasted ones.

• It means that every producer will have to provide a day-ahead production plan

with hourly resolution, which is supposed to be granted within a given confidence

interval.

• In such setup energy storage (or demand response) can help in meeting the hourly

production plan.


Microgrid and energy management details Problem definition - hypothesis

• The day-ahead, an hourly energy production plan for the microgrid system

is defined. This plan is calculated by knowing the PV module layout information,

the WT power curve and the weather forecast.

• During the operation day the hourly production plan must be respected within

±1%.

• The storage system can be used to correct the deviations from the plan, but

it is very crucial not to overuse it, because any charge/discharge cycle would

lead to energy losses.


Microgrid and energy management details System layout in SYSLAB facility

10 kW PV Plant:

TF & Poly

11 kW WT Plant:

passive stall turbine

15 kW – 190 kWh:

Vanadium Redox Flow

PCC

Power Transit

Energy Manager

Storage

Power Ref

Estimated Production

DTU Meteo Forecast:

Solar Irr; Temp; Wind Hourly Data Microgrid Component

Simulated Models

400 V 3p

Main Network

Control System

Energy Plan

• SYSLAB facility and used subset of power

components. SYSLAB is a small-scale power

system consisting of real power components

interconnected by a three phase 400V power

grid, and some communication and control

nodes interconnected by a dedicated network,

all distributed over the Risø Campus.


Microgrid and energy management details Forecast input

• Forecasts are given by the Wind Energy Dep. two times per day for the following 48 hours:

– @ 9 am the 48 hourly series starting at 12 GMT

– @ 9 pm the one starting at 24 GMT.

-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 70

0.05

0.1

0.15

0.2

0.25Wind speed forecast error histogram

Rela

tiv

e f

req

uen

cy

(p

u)

Wind speed error (m/s)

PDF

Mean

±STD

-0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.70

0.05

0.1

0.15

0.2

0.25

Solar irradiance forecast error histogram

Rela

tiv

e f

req

uen

cy

(p

u)

Solar irradiance error (kW/m^2)

PDF

Mean

±STD

• 6 months of comparison between the

forecasted values of wind speed and

solar irradiance and the respective

historical data are reported.

• The mean values are respectively

equal to 1.0 m/s and 34 W/m2

• The standard deviations are 2.9 m/s

and 258 W/m2


Microgrid and energy management details Microgrid components models

• In order to evaluate the PV and the WT production, proper models have been developed in Simulink.

• The hourly weather data is used as input for the models

• The PV and WT estimated hourly productions are used for defining the energy production

plan (the one that the microgrid owner is supposed to follow with a precision of ±1% during the day

of operation).


Microgrid and energy management details Energy Management Strategy

• The hourly energy plan is used by the energy manager to build the Energy Ref curve.

• The red and the green dashed curves are the control band. Whenever the energy profile

exceeds the upper or the lower bound (red and green line) the battery is activated:

– the more the distance from the objective value, the deeper the charge/discharge required.

– the opening of the lines, that means the distance between the upper and the lower band with

reference to the “Energy Ref” curve, determines the stiffness of the control

0 5 10 15 20 25 30 35 40 45 50 55 60-2

024

68

1012

Energy Plan - Microgrid

En

erg

y (

kW

h)

Time (min)

Upper Band

Lower Band

Energy Ref


Microgrid and energy management details a possible way to address the storage control – power smoothing strategy

• A storage (15 kW-190 kWh) plus wind turbine (11 kW) system aiming at smoothing the power

output at the PCC with 1 second communication delay!

0 5 10 15 20 25 30 35 40 45 50 55 60-12

-8

-4

0

4

8

12Battery (black) and Wind Turbine (blue) Powers

Po

we

r (k

W)

Time (min)

0 5 10 15 20 25 30 35 40 45 50 55 6002468

10121416

PCC (blue) and Reference (red) Powers

Po

we

r (k

W)

Time (min)


Microgrid and energy management details a possible way to address the storage control – power smoothing strategy

• …and now with 8 seconds communication delay!!! Very far from being smoothed!

0 5 10 15 20 25 30 35 40 45 50 55 60-12

-8

-4

0

4

8

12Battery (black) and Wind Turbine (blue) Powers

Po

we

r (k

W)

Time (min)

0 5 10 15 20 25 30 35 40 45 50 55 6002468

10121416

PCC (blue) and Reference (red) Powers

Po

we

r (k

W)

Time (min)


Microgrid and energy management details Concerning the power and the energy sizing of the storage

• The design of the control bands has an impact on both storage power and energy size. The two main

variables that affect the sizing of the storage are the forecast precision and accuracy (i.e. if they have

an offset and how much they span).

• Power sizing:

– In the considered setup the control bands are set to 10%, meaning that the lower band (the green line)

crosses at the 6th minute the time axis.

– This means that a storage inverter of 5 kW should be able, in case of null production of the microgrid, to

release full power for the remaining 54 minutes, leading to a maximum energy of 4.5 kWh.

– Considering a microgrid maximum hourly production of 21 kWh, which means full power output from

both PV and WT, with this control band setup, the storage is able to compensate production errors up to

21% of the microgrid hourly production.

• Energy sizing:

– The sizing of the storage capacity depends on the coherence of the forecast errors. It can be noted that

in case of alternate hourly errors, the storage alternatively charges and discharges so that the depletion

of the state of charge depends mainly on the storage internal inefficiencies.

– Therefore the storage energy rating could be relative small. If the error has the same sign for the whole

day, then the storage energy sizing criteria needs to be revisited.


Microgrid and energy management details Controller sensitivity • The relative energy error is the difference between the system energy state and the control bands. Two energy errors can

be identified depending on the crossing of the upper or the lower band.

– eup is the difference between the system energy state and the upper control band, divided by the upper control

band value. This error triggers a storage charge.

– edown is the difference between the system energy state and the lower control band, divided by the lower control

band value. This error triggers a storage discharge.

• The sensitivity of the controller is chosen in order to have the storage to store/release the maximum power, Pmax, if the

relative energy error, e, is equal or greater to 1%.

• If the error amplitude is within 0% and 1%, the storage power set-point is changed by five discrete steps 20% amplitude.

Upper relative energy

error %

Storage reference

power

Lower relative energy

error %

Storage reference

power

eup ≤ 0% 0 edown ≤ 0% 0

0 < eup ≤ 0.2 0 0 < edown ≤ 0.2 0

0.2 < eup ≤ 0.4 -20% Pmax 0.2 < edown ≤ 0.4 20% Pmax



0.8 < eup ≤ 1 -80% Pmax 0.8 < edown ≤ 1 80% Pmax

eup ≥ 1% -Pmax edown ≥ 1% Pmax


Experimental testing Procedure

• The testing days reported are the 3rd July 2013, which was a windy day with frequent clouds

passages and the 7th July, which was a sunny day with few wind.

• The experimental process is here explained:

1. The meteorological forecast data (solar irradiance, wind speed, air temperature) of the

studied day are used to evaluate the PV and WT output. The day-ahead forecasts that mean

the forecasts given at the 9 am of the 2nd July 2013 (for the experiment run on the 3rd July)

are taken in consideration.

2. The PV and WT model outputs are used to build the forecasted energy plan for the studied

day (3rd July).

3. The microgrid setup is prepared and the pc with the Simulink controller, aimed at managing

the storage system, is connected to the SYSLAB facility and the experiment runs.


Experimental testing test on July 3rd

0 30 60 90 120 150 180

02468

101214

Energy State and Plan

En

erg

y (

kW

h)

Time (min)

0 30 60 90 120 150 180-6

-4

-2

0

2

4

6Storage Power

Po

wer

(kW

)

Time (min)

Energy State

Upper Band

Lower Band

Energy Ref

0 20 40 60 80 100 120 140 160 1800

1

2

3

4

5Relative Energy Error (5% limited)

En

erg

y E

rro

r (%

)

Time (min)

• The experiment starts at 13:00 GMT (15:00

local time) and lasts for 3 hours.

• The overall energy plan, that means the expected

production for PV and WT, for the 3 hours

experiment is shown in the first plot. The

microgrid energy production can be also observed

in this plot (black curve).

• The storage, whose output is depicted in the

second plot, is able to compensate the deviations

from the predicted production.

• The relative deviation from the expected

production, which means the difference between

the production and the forecasts, referred to the

energy plan is reported in the third plot.


Experimental testing test on July 3rd – detail

120 125 130 135 140 145 150 155 160 165 170 175 180-202468

1012

Energy State and PlanE

nerg

y (

kW

h)

Time (min)

Energy State

Upper Band

Lower Band

Energy Ref

120 125 130 135 140 145 150 155 160 165 170 175 180-6

-4

-2

0

2

4

6Storage Power

Po

wer

(kW

)

Time (min)


Experimental testing test on July 3rd

• The power profiles of the different

components can be observed in the first

plot.

• While the overall microgrid production,

that means the power transit at the PCC, is

reported in the second plot.

0 20 40 60 80 100 120 140 160 180-5

-4

-3

-2

-1

0

1

2

3

4

5

6

7

8

9

10

11

12PV, WT and Storage outputs

Po

wer

(kW

)

Time (min)

PV

WT

Storage

0 20 40 60 80 100 120 140 160 18002468

1012141618

Microgrid PCC power transit

Po

wer

(kW

)

Time (min)


Experimental testing test results on July 3rd

3rd July – 13:00-16:00 GMT

GMT hours 13-14 14-15 15-16 13-16

PV+WT forecasted (kWh) 7.85 12.61 9.61 30.07

PV+WT produced (kWh) 8.49 10.75 11.35 30.58

PV+WT error (kWh) 0.64 -1.86 1.74 *4.24

Storage usage (kWh) -0.51 2.48 -4.10 *7.09

PCC energy transit (kWh) 7.88 12.60 9.61 30.08

PCC energy absolute error (kWh) 0.027 -0.014 -0.003 *0.044

PCC energy relative error (%) 0.35 -0.11 -0.03 **0.15

• PV+WT forecasted is the forecasted hourly production, it is the value of the energy plan at the end of the hour.

• PV+WT produced is the hourly energy produced by the renewable sources: photovoltaic and wind turbine.

• PV+WT error is the difference between energy produced and forecasted. The error is positive if there is an excess of

production.

• Storage usage is the total work of the storage that means the sum of charge and discharge regardless of the sign.

• PCC energy transit is the hourly production of the whole microgrid, thus PV, WT and storage and including the cable losses

(almost negligible due to their size).

• PCC energy absolute error is the difference between the production of the microgrid and the forecasted one.

• PCC energy relative error is the PCC energy error relative to PV+WT Forecasted.

• *calculated as the sum of the hourly errors regardless of sign;

• ** calculated as the weighted average of the hourly errors regardless of sign


Experimental testing test on July 7th

• The experiment starts at 9:00 GMT (11:00

local time) and lasts for 6 h.

• The overall energy plan, that means the

expected production for PV and WT, for the 3

hours experiment is shown in the first plot.

The microgrid energy production can be also

observed in this plot (black curve).

• The storage, whose output is depicted in the

second plot, is able to compensate the

deviations from the predicted production.

• The relative deviation from the expected

production, which means the difference

between the production and the forecasts,

referred to the energy plan is reported in the

third plot.

0 30 60 90 120 150 180 210 240 270 300 330 360

02468

101214

Energy State and Plan

En

erg

y (

kW

h)

Time (min)

Energy State

Upper Band

Lower Band

Energy Ref

0 30 60 90 120 150 180 210 240 270 300 330 360-6

-4

-2

0

2

4

6Storage Power

Po

wer

(kW

)

Time (min)


Experimental testing test on July 7th

0 30 60 90 120 150 180 210 240 270 300 330 360-5

-4

-3

-2

-1

01

2

3

4

5

67

8

9

10

1112

PV, WT and Storage outputs

Po

wer

(kW

)

Time (min)

PV

WT

Storage 0 30 60 90 120 150 180 210 240 270 300 330 36002468

1012141618

Microgrid PCC power transit

Po

wer

(kW

)

Time (min)

0 30 60 90 120 150 180 210 240 270 300 330 36002468

1012141618

PV and WT power production

Po

wer

(kW

)

Time (min)

• It is straightforward to note that the forecasts were

rather optimistic; however the storage succeeds in

managing the microgrid energy production.

• the microgrid power output smoothing is not

pursued.


Experimental testing test results on July 7th • PV+WT forecasted is the forecasted hourly production, it is the value of the energy plan at the end of the hour.

• PV+WT produced is the hourly energy produced by the renewable sources: photovoltaic and wind turbine.

• PV+WT error is the difference between energy produced and forecasted. The error is positive if there is an excess of

production.

• Storage usage is the total work of the storage that means the sum of charge and discharge regardless of the sign.

• PCC energy transit is the hourly production of the whole microgrid, thus PV, WT and storage and including the cable losses

(almost negligible due to their size).

• PCC energy absolute error is the difference between the production of the microgrid and the forecasted one.

• PCC energy relative error is the PCC energy error relative to PV+WT Forecasted.

• *calculated as the sum of the hourly errors regardless of sign;

• ** calculated as the weighted average of the hourly errors regardless of sign

7th July – 9:00-15:00 GMT

GMT hours 9-10 10-11 11-12 12-13 13-14 14-15 9-15

PV+WT forecasted (kWh) 11.56 12.78 12.82 12.40 11.46 9.67 70.69

PV+WT produced (kWh) 8.90 9.78 9.41 9.42 9.58 6.93 54.02

PV+WT error (kWh) -2.66 -3.00 -3.41 -2.98 -1.88 -2.73 *-16.66

Storage usage (kWh) 2.56 5.36 8.63 11.56 13.36 16.00 57.47

PCC energy transit (kWh) 11.48 12.60 12.70 12.38 11.38 9.57 70.11

PCC energy absolute error (kWh) -0.08 -0.18 -0.12 -0.03 -0.07 -0.10 *-0.58

PCC energy relative error (%) -0.71 -1.41 -0.90 -0.22 -0.60 -0.99 **0.80


Conclusions… • The management of the energy production of energy systems with renewable generation sets will

get more and more important in the near future and the correct prediction of the expected

production will be crucial in order to be able to manage the overall system resources.

• Improving the forecast techniques is of great importance; however errors cannot be totally

avoided. The compensation of these errors will be critical and a proper management of the

controllable resources, such as storage systems or controllable loads, is essential.

• This presentation showed a management strategy capable of controlling a storage system (limited

to 5 kW) in order to grant the forecasted energy production plan of a microgrid, formed by a 10

kW PV and an 11 kW WT.

• The overall idea consists in, by knowing the meteorological forecast for the next 24 h, controlling

the storage system in order to compensate the hourly deviations from the day-ahead energy

plan, minimizing the storage usage.


…and future developments • A first evaluation of the correlation between the meteorological forecasts errors and the sizing in

term of power and energy of the storage is analyzed even if deeper analyses are needed.

• The range of experiments will be extended across the days, taking also advantage of the fact that

the whole system can be studied in the simulated environment.

• Finally, further analysis will aim at extending the described strategy in a larger microgrid setup,

including building appliances, such as heating systems and water heaters, and electric vehicles.


Outline – II part (simulations over 1 day and storage sizing)

I. Introduction

II. Model Description

– Problem definition

– System layout and management strategy analysis

III. Simulation Results

– Simulation procedure

– Scenarios analysed (different storage power size)

IV. Conclusions and Future Developments


Model Description System layout – components are simulated

10 kW PV Plant: TF & Poly

VRB storage system

PCC

Power Transit

Energy Manager

Storage

Power Ref

Estimated Production

Meteo Forecast:

Solar Irr; Temp; Wind Hourly Data PV Plant Model

400 V 3p

Main Network

Control System

Energy Plan


Model Description Photovoltaic model

• The DC power produced by the module mainly

depends on the incident solar radiation and on

the temperature, which for instance is function

of air temperature, wind speed and radiation

itself.

• The dependence of the panel output with

different sunlight intensity and the dependence

in function of the temperature have been

evaluated in order to evaluate the reduction

from the nominal efficiency, having taken in

account that the nominal data are provided for

standard meteorological conditions (1000 W/m2

and 25 °C)

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 200

12345

6789

10PV Production 10 kW - 1 minute sampled

Ou

tpu

t (k

W)

Time (h)

Model Output

Historical Data

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 200

12345

6789


Ou

tpu

t (k

W)

Time (h)

Model Output

Historical Data


Model Description storage model

• The modelled dynamics regard the State-of-Charge (SOC) behaviour, the electrochemical conversion

and the thermal characterization.

• The main state variables are therefore the state of charge, the temperature and the

voltage: all the characteristic elements of the storage system (as open circuit voltage, internal

resistances, limitations and protections thresholds) present some kind of dependence from these

state variables.


Model Description Energy Management Strategy

• The input of the controller is the power measured

at the PCC, which is not directly used in the control

loop; but it is used to compute the correspondent

energy reference within the hour.

0 5 10 15 20 25 30 35 40 45 50 55 60-1

0

1

2

3

4

5

6

7

8Energy Plan - Microgrid

En

erg

y (

kW

h)

Time (min)

Upper Band

Lower Band

Energy Ref

6 6.25 6.5 6.75 7-0.5

0

0.5

1

1.5

2

2.5

3Energy Plan

En

erg

y (

kW

h)

Time (h)

Energy State

Upper Band

Lower Band

Energy Ref

6 6.25 6.5 6.75 7-1.5

-1

-0.5

0

0.5

1

1.5Battery Power

Po

wer

(k

W)

Time (h)

Ref Power

Power Output

• The red and the green lines form the control

band: whenever the energy profile exceeds the

upper or the lower bound (red and green line)

the battery is activated:

– the more the distance from the objective

value, the deeper the charge/discharge

required.


Simulation Results Simulation Procedure

• The simulation day chosen is the 4th May 2013, which was a sunny day with

frequent clouds passages.

• The simulation process follows:

1. The forecast meteorological data (solar irradiation, wind speed, air

temperature) are used to evaluate the PV output. The day-ahead forecasts

(that mean the forecasts given at the 8 am of the 3rd May 2013) are taken in

consideration.

2. The PV model output is used to build the energy plan for the studied day

3. Since the sensitivity analysis aim at finding the suitable size of the storage

system, several simulations are run using the historical production data of

the PV installed in the SYSLAB and using the Simulink validated model of the

storage.


5 6 7 8 9 10 11 12 13 14 15 16 17 18 19-1

-0.5

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

6

6.5

7

7.5

8

8.5

9Energy Plan

En

erg

y (

kW

h)

Time (h)

Upper Band

Lower Band

Energy Ref

Simulation Results Simulation Procedure

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 200

1

2

3

4

5

6

7

8

9


Ou

tpu

t (k

W)

Time (h)

Forecast Output

Historical Data

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 200

1

2

3

4

5

6

7

8

9


Ou

tpu

t (k

W)

Time (h)

Forecast Output

Historical Data

PV Forecasted profile (red)

PV historical data (blue)

The total energy produced by

the PV for the studied day was

64.6 kWh, while, according to

the forecasts, it would have

been 71.8 kWh (+11%).


Simulation Results Base case – no storage

6 7 8 9 10 11 12 13 14 15 16 17 180

1

2

3

4

5

6

7

8

9System Hourly Energy

En

erg

y (

kW

h)

Time (h)

Hourly System Energy

Hourly Energy Error

6 7 8 9 10 11 12 13 14 15 16 17 180

1

2

3

4

5

6

7

8Hourly Energy Relative Error

Err

or

(%)

Time (h)

6 7 8 9 10 11 12 13 14 15 16 17 18-1

0

1

2

3

4

5

6

7

8

9Energy Plan

En

erg

y (

kW

h)

Time (h)

Energy State

Upper Band

Lower Band

Energy Ref

The relative average error

is equal to 17.3% and the

maximum one is greater

than 50%.


Simulation Results Different storage power size (with reference to the PV installed capacity)

Results with the storage size

of 3 kW (30% scenario)

6 7 8 9 10 11 12 13 14 15 16 17 18-1

0

1

2

3

4

5

6

7

8

9Energy Plan

En

erg

y (

kW

h)

Time (h)

Energy State

Upper Band

Lower Band

Energy Ref

6 7 8 9 10 11 12 13 14 15 16 17 18-3

-2.5-2

-1.5-1

-0.50

0.51

1.52

2.53

Battery Power

Po

wer

(k

W)

Time (h)

Ref Power

Power Output

6 7 8 9 10 11 12 13 14 15 16 17 180

1

2

3

4

5

6

7

8

9System Hourly Energy

En

erg

y (

kW

h)

Time (h)

Hourly System Energy

Hourly Energy Error

6 7 8 9 10 11 12 13 14 15 16 17 180

1

2

3

4

5

6

7

8Hourly Energy Relative Error

Err

or

(%)

Time (h)


Simulation Results Different storage power size (with reference to the PV installed capacity)

• It is interesting to observe the

behavior of the power measure at the

PCC (first plot) which for certain

aspects is worsen because of the

power steps which can be induced by

the storage operation.

• However it has to be kept in mind that

the objective of the control strategy

proposed is driven from an energy

perspective due to the need to respect

the scheduled energy plan.

5 6 7 8 9 10 11 12 13 14 15 16 17 18-10123456789

1011

PCC (Battery plus PV) transit

Po

wer

(k

W)

Time (h)

5 6 7 8 9 10 11 12 13 14 15 16 17 18-10123456789

1011

PV PowerP

ow

er (

kW

)

Time (h)


Simulation Results Storage usage and system performances

Storage Size Storage usage

(kWh)

Energy released (kWh)

Average hourly error

Max hourly error

Base case size 0% PV

0 0 17.3% >50.0%

5% 2.64 1.90 10.3% >50.0%

10% 4.45 3.54 7.2% 41.2%

15% 6.03 5.07 4.1% 19.8%

20% 7.43 6.46 1.5% 6.1%

25% 7.99 6.99 0.7% 2.6%

30% 8.10 7.10 0.6% 2.2%

50% 8.28 7.27 0.5% 1.4%


Conclusions and future developments

• The proposed work describes an energy management strategy for a 10 kW PV plant coupled with

a storage system.

• By knowing the meteorological forecast for the next 24h, the objective is to dispatch the PV

system and to be able to grant the scheduled hourly energy profile by a proper management of

the storage.

• The energy manager controls the storage in a predefined way in order to ensure that the hourly

energy production plan is respected, compensating the forecast errors and minimizing the storage

itself usage.

• The study is intended to provide also a methodology for the evaluation of the size of the storage

system in terms of power and energy.

• Further analysis will aim at extending the study along the whole year and at evaluating the

behaviour of the energy manager in case of different forecast horizons.


For further readings…

• M. Marinelli, F. Sossan, G. T. Costanzo, and H. W. Bindner, “Testing of a Predictive Control Strategy for Balancing

Renewable Sources in a Microgrid,” Sustainable Energy, IEEE Transactions on, vol.PP, no.99, pp.1-8, Jan. 2014.

• M. Marinelli, F. Sossan, F. Isleifsson, G.T. Costanzo, and H. W Bindner, “Day-Ahead Scheduling of a Photovoltaic

Plant by the Energy Management of a Storage System,” Universities Power Engineering Conference (UPEC), 2013

48th International, pp.1-6, Dublin, 2-5 Sep. 2013.

• F. Baccino, M. Marinelli, P. Nørgård, F. Silvestro, “Experimental testing procedures and dynamic model validation for

vanadium redox flow battery storage system,” Journal of Power Sources, Volume 254, 15 May 2014, Pages 277-286.

• F. Baccino, M. Marinelli, F. Silvestro, O. Camacho, F. Isleifsson, and P. Nørgård, “Experimental validation of control

strategies for a microgrid test facility including a storage system and renewable generation sets,” Integration of

Renewables into the Distribution Grid, CIRED 2012 Workshop, pp.1-4, 29-30 May 2012

• F. Baccino, S. Grillo, M. Marinelli, S. Massucco, and F. Silvestro, “Power and Energy Control Strategies for a

Vanadium Redox Flow Battery and Wind Farm Combined System,” Innovative Smart Grid Technologies (ISGT

Europe), 2011 2nd IEEE PES International Conference and Exhibition on, pp.1-8, 5-7 Dec. 2011.

Documents

Testing of a Predictive Control Strategy for Balancing ... of a... · Testing of a Predictive Control Strategy for Balancing Renewable Sources in a Microgrid Mattia Marinelli; [email protected]