Upload
dangquynh
View
218
Download
4
Embed Size (px)
Citation preview
Recoupling Forecasting with Inventory Control and
Supply Planning: “Readiness-Driven Supply Networks”
Greg H. Parlier
Colonel, US Army, Ret
2017 FORECAST PRACTITIONER CONFERENCE
NCSU IAA 15 November 2017
-400
-200
0
200
400
600
800
1000
1200
1400
1600
1800
2000
2200
2400
2600
2800
3000
97 98 99 00 01 02 03
Requirement
Funded
UFR
181,047
112,946
38,202
54,397
11,212 12,0422,491 2,491 415
0
50,000
100,000
150,000
200,000
<$10 <$50 <$100 <$500 <$1K <$5K <$10K <$50K >$50K
Cost of Part Requisitioned
71% of NMCS
Requisitions were
for parts under $50
The Immediate Problem: Circa 2002
Situation: Selected (Anonymous) Comments
“All signs are bad”
“Huge disconnect between Log & Ops”
“Wholesale and retail are not integrated”
“There is growing fear that we do not have enough to ensure readiness; that fear is
accompanied with perceptions of tremendous inefficiencies in our system”
“We could spend $100M on spares and see no readiness improvement, or we could
spend $10M on spares (differently) and see it improve!”
“Why am I still throwing billions down this black hole called Spares?”
“We don’t believe the aviation spares requirements numbers”
“The financial system is undermining our ability to do things smart”
“Our incentives are all in the wrong places…”
Following the Cold-War drawdown, as of late 2002 . . .
But then . . . .
“The attacks of September, 11th, 2001, opened a gusher of spending
that nearly doubled the base budget over the last decade, not counting the
supplemental appropriations for the wars in Iraq and
Afghanistan. . .”
And now . . . .
And now (2017), a decade and a half later . . .
“. . . we face a very different set of American fiscal realities . . .
The culture of endless money that has taken hold must be replaced by a culture
of savings and restraint”
Situation: Selected (Anonymous) Comments
Yet . . .
“DoD’s supply chain system has remained stuck in a 20th Century model
because of . . . resistance to change.”
“The Army is facing a [readiness] crises . . . Problems will only get worse.”
“An era of blank-check defense spending is over . . .”
“We need to restructure our demand process and change the algorithm to
meet future demand.”
“DoD has not been able to identify relationships between O&S spending
and the readiness of military units.”
Resources-to-Readiness Challenge
• Investment is increasing, yet back orders are growing and UFRs are
increasing
• “Workarounds” are increasing, readiness is slowly declining
• Readiness reporting appears suspicious, lacks credibility
• Systems are non-operational for relatively inexpensive parts
“Efficient
Frontier”
Ao
$
?
Aligning Supply to Readiness Driven Demand
Wholesale
ReverseLogistics
Retail Unit Mission Demand
SUPPLY DEMAND
Forecast Actually Used
Acquisition
MATERIAL FLOWS
Inventory
Vendors
Inventory
Wholesale
Inventory
Retail
Inventory
Unit
INFORMATION FLOWS
Orders to
Vendors
Orders to
Wholesale
Orders to
Retail
Demand
FUNDING FLOWS
OMA
Funding
Available
OMA
Funding AWCF
Payments to
Vendors and
Depots
ReCap &
ReSet
Demand
ReCap &
ReSet
Funding
Obligation
Authority
Supply Chain FrameworkSupply Chain Framework: Organization, Process,
And Financial “Views” of the Materiel Enterprise
Wholesale
StageDemand
Stage
Retail
Stage
Unit
StageAcquisition
Stage
• OEM’s
• Suppliers
• Supply Depots
• Repair Depots
• OEM’s
• SSAs
• ASLs
“Readiness
Production”
• Retrograde
Operations
• Training
• Operations
Reverse
Logistics Stage
Supply Sources of VariabilityDemand
Uncertainty
σ2 = LσD+ D2σL2 2
Supply Variability and Demand Uncertainty:
Army Supply Chain Model
“Decentralized
”“Centralized
”
K = # of stages031 542
25
0
20
10
15
5
Suppliers RetailWholesale Unit
qk
Lk
q2
L2
q1
L1
q0 = D
Customer
…the “bullwhip effect”
s2 (qk)
s2 (D)
Capacity, Inventory & Knowledge
Capacity:
What we
can do
What we
knowWhat we
have
Substitutable Ingredients of System
Performance
Knowledge: Inventory:
10
A B C
Production
System
Engineering &
Development
Technology
Development
Concept
RefinementO&S
What Happened? What Could Happen? Make it Happen!
2002 2003 2005
Phase 1• Segment the Logistics
Structure & Processes
for Analysis
• Adapt Enterprise
Supply Chain
Framework for
Integration
[~$200K]
Phase 2• Identify "Readiness
Production Function"
• Develop "Mission
Based Forecasting"
• Validate "Readiness
Based Sparing"
• Incorporate
Multi-Echelon
Optimization &
"Synchronized
Retrograde
Operations"
• DDSN & LEWS
[$1.0M]
Phase 3• Provide COTS RBS
Solutions for PSI
• Develop Large-Scale
MOD & SIM Capacity
for SC Enterprise
• Implement CILS
Organizational Design
• Strategic outreach &
Research Partnerships
for Continuous
Improvement
[$2.2M]
Expanded
Market
Opportunities
CILS Provides:• Product Support Integration
• Supply Chain Optimization
• Logistics System Readiness
• SNL (RECAP)
• IDA ( Reliability
Design to Readiness)
• UAH (OEM Supplier Analysis)
Acquisition
Reverse
Logistics
UnitWholesale Retail Demand
Task Organization for Research and Analysis
• LMI (Peak Policy & ICAAPS)
• PNNL (VLD)
• USMC (TOC)
• IDA
• RAND
• LOGSA
• SAIC
• RAND (EDA)
• LMI
• AMSAA
• PMs
• AMSAA (RBS)
• MCA
Transforming US Army Supply Chains (TASC): Project Phases
11
Working Towards Solutions
Innovation Catalysts:• Defining the Readiness Equation
• Mission Based Forecasting
• Connect CBM to the Supply Chain
• Readiness Based Sparing
• Readiness Responsive Retrograde
• Leveraging Lessons Learned & Best Practices
RETROGRADE
ACQUISITION DEMANDWHOLESALE UNITRETAIL
Analyzing Root Causes and Prescribing Innovation
Catalysts Across the Supply Chain
(1) lack of an aviation readiness production function which induces both uncertainty and
variability at the point of consumption in the supply chain resulting in inappropriate
planning, improper budgeting, and inadequate management to achieve readiness objectives;
Supply Availability Demand Requirements
MTBF
MLDT
MTTR
- NMCS
- NMCM
Operational
Availability (Ao)
[ER] - MC
- FMC
- PMC
[AS] [TS]
• Deployment Missions (DEPTEMPO)
- Patterns of Operation
Duration
Profile
- Environmental Conditions and
Locations
• Training Requirements (OPTEMPO)
Readiness – related Measures / Metrics
[ER] – Equipment Readiness (Ao)
• FMC • NMCS
• MC (PMC) • NMCM
[AS] – Assigned Strength
[TS] – Trained Strength
Personnel
Manning and
Skill Levels
Weapon
System
Reliability
Supply
Support
Capability
Training
Resources
(OPTEMPO $)
“Production Function”: Components of Readiness
La
bo
rL
ab
or
CapitalCapital
90% MC90% MC
AO =
=
Uptime
Total Time
WhereMTBF = Mean Time Between
Failures (Reliability)
MTTR = Mean Time To Repair
(Maintainability)
MLDT = Mean Logistics Delay Time
(Supportability)
MTBF
MTBF + MTTR + MLDT
80% MC80% MC
Extract from research results:
- The longer the delay, the more likely a
workaround . . .15% of deadline requisitions
for wholesale backorders were satisfied by
workarounds.
- “Labor” (MMH) increasingly substituting for
“Capital” . . . If workarounds were eliminated,
readiness would decline by 33%.
- “Consumption” data is not systematically
collected by current MIS
Research Goals:
- Define and empirically measure the
“readiness equation” for Ao
- Determine readiness “driver” marginal
values, and evaluate contributions and
costs for potential solutions.
Innovation Catalyst: Analyzing the Readiness EquationInnovation Catalyst: Analyzing the Readiness Equation
and Measuring True “Customer Demand”
15
Working Towards Solutions
Innovation Catalysts:• Defining the Readiness Equation
• Mission Based Forecasting
• Connect CBM to the Supply Chain
• Readiness Based Sparing
• Readiness Responsive Retrograde
• Leveraging Lessons Learned & Best Practices
RETROGRADE
ACQUISITION DEMANDWHOLESALE UNITRETAIL
Analyzing Root Causes and Prescribing Innovation
Catalysts Across the Supply Chain
(2) limited understanding of mission-based, operational demands and associated spares
consumption patterns which contribute to poor operational and tactical support planning and
cost-ineffective retail stock policy;
3 observed force-on-force forms
Effects-based operational forms:
Continuous fronts
Disintegrations
Disruptions
+ Stability operations
P A G E 16
Secondary attack
Fix
Fix
Fix
Main attack
Main attack
Secondary attack
Continuous FrontContinuous front
Stability OpnMain attack
Fix
Fix
Main attack
Main attack
Fix
Fix
Fix
Secondary attack
Disruption (high-order)
Disruption(high-order)
Main attack
Secondary attack
Fix
Fix
Main attack
Secondary attack
Fix
Fix
Fix
Fix
Disintegration
Disintegration
Analyzing Operational Forms
and Empirical Patterns
STRATIFIED SAMPLING
POPULATION OF SIZE N DIVIDED INTO K STRATA
n
xPRSM
ˆRANDOM SAMPLING:
STRATIFIED SAMPLING:
nx
Pk
kk1
THEN:
NPN
P
k
ikk
STRAT
1
1
ˆ
USUALLY:
ˆˆRSMPOPSTRAT
VarVarVar
“. . .compare forecasting methods [to establish] regions of
superior performance, then categorize demand patterns
[in order to] select the most appropriate estimation procedure.”
From “On the Categorization of Demand Patterns”, by
Syntetos, Boylan, and Croston JORS, 2005
1
8
Mission Demand
Operation Type/Duration
Environmental Conditions
Force Size/Composition
Center for Systems ReliabilityReadiness & Sustainment DepartmentSandia National Laboratories (SNL)
Albuquerque, NM 87185
Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company,
for the United States Department of Energy under contract DE-AC04-94AL85000.
Innovation Catalyst: Mission Based Forecasting (MBF)
CONUS vs IraqStabCONUS vs IraqMCO
59% 41%
73% 27%
CONUS IraqStab
59% 41%59% 41%
73% 27%73% 27%
CONUS IraqStab
Ratios of Demands: Common PartsRatios of Demands: Common Parts1
2
3
4
0
4.4 x CONUS5
6
1
2
3
4
0
4.4 x CONUS5
6
1
2
3
4
0
3.9 x CONUS
5
6
1
2
3
4
0
3.9 x CONUS
5
6
12%88%
CONUS IraqMCO
71% 29%
12%88%
CONUS IraqMCO
12%88% 12%88%
CONUS IraqMCO
71% 29%71% 29%
111 84 87111
AH-64DAH-64D
1.01.0
Note: original Views
(2005 data)
Note: original Views
(2005 data)
2833 2833
Research Goal:
Our major hypothesis states: “If empirically-derived Class IX usage patterns, profiles and/or trends
can be associated with various operational mission types and environmental conditions, then
operational planning, demand forecasting, and budget requirements can be significantly improved
to support a capabilities based force”.
Innovation Catalyst: Mission-Based Forecasting (MBF)
Note: New (2006) data
Note: New (2006) data
Note: New (2006) data
Actual NSNs Used
Comparing Forecast Methods:Accuracy versus over- and under-forecast
Cost
Found
Accurate
Error(over-forecast)
Error(under-forecast)
Forecasted NSNs
NSN lines [“breadth”]
Errors Found Errors(over-forecast) (under-forecast)
Forecast Actual
Part
Analysis
Actual
QuantityQuantity
[“depth”]
Quantity
Analysis
Forecasted
Quantity
Forecasted
Quantity
Actual
Quantity
The Analysis addresses two perspectives
Phase 2 Membership (Lines) - Case 3
without JLAT (3 hr)
0
500
1,000
1,500
2,000
2,500
ODDP G4 OSRAP CIF JLAT (1 hr)
Method
Pa
rt C
ou
nt
Over-forecast
Found (correctly forecast)
Under-forecast
AH-64D Parts Count Forecast (Breadth of NSNs):
MBF Compared to Current MethodsCase 3, Stability Ops (mid-level threat), 12 months, 104 tails
MBF reduces part over-forecast, and under-forecast
& improves forecast part-accuracy% “Found” out of total forecast parts (breadth)
MBF A B C D
These current methods (A, B, C,
D)
use supply requisitions data
Measuring Forecast Accuracy:Reducing Error Sources
Improving Forecast Accuracy: Reduces Forecast Errors, Increases Readiness, Reduces Excess, and Minimizes Burden
$0
$50,000,000
$100,000,000
$150,000,000
$200,000,000
$250,000,000
ODDP Combo
PLL+ASL
Combo PLL PLL1 PLL2 PLL3 PLL4
Phase 2 Cost (Parts) - Case 3
Excess $ for incorrectly predicted parts
Excess $ for correctly predicted parts
Shortage of $ for missed parts
Shortage of $ for correctly predicted parts
Correctly predicted actual $
AH-64D Parts Quantity Forecast (Depth of NSNs):
MBF Compared to Actual On-Hand Stocks
$200M
$138M
Bn
(24 tails)
Bn
(24 tails)
Bn
(24 tails)
Bn
(24 tails)PLL
Rollup
for
4Bn’s(96 tails)
Actual
On-hand
Rollup
for
4Bn’s(96 tails)
ODDP-
Forecasted
Parts rollup
for
104 tails
Subset: 4 Bn PLLs
(Bn-level stocks)
Savings:
for 1 year
(breadth & depth)
= $62M
Intermittent Demand
Professional Judgment
Moving Average/Exponential Smoothing
Poisson Methods (Croston)
Markov Bootstrap (Smart-Willemain)
Ongoing Research
Varieties of Intermittent Demand
7/24/2015 29
0 5 10 15 20 25 30 35
02
46
8
company 17 item 221
Index
y[i,
]
0 5 10 15 20 25 30 350
10
20
30
company 17 item 222
Index
y[i,
]
0 5 10 15 20 25 30 35
02
46
8
company 17 item 223
Index
y[i,
]
0 5 10 15 20 25 30 35
050
150
company 17 item 224
Index
y[i,
]
0 5 10 15 20 25 30 35
020
60
company 17 item 225
Index
y[i,
]
0 5 10 15 20 25 30 35
010
20
company 17 item 226
Index
y[i,
]
0 5 10 15 20 25 30 35
040
80
company 17 item 227
Index
y[i,
]
0 5 10 15 20 25 30 35
0.0
0.4
0.8
company 17 item 228
Index
y[i,
]
0 5 10 15 20 25 30 35
0.0
0.4
0.8
company 17 item 229
Index
y[i,
]
0 5 10 15 20 25 30 35
02
4
company 17 item 230
Index
y[i,
]
0 5 10 15 20 25 30 35
01
23
4
company 17 item 231
Index
y[i,
]
0 5 10 15 20 25 30 35
040
80
company 17 item 232
Index
y[i,
]
0 5 10 15 20 25 30 35
05
15
company 17 item 233
Index
y[i,
]
0 5 10 15 20 25 30 35
02
46
company 17 item 234
Index
y[i,
]
0 5 10 15 20 25 30 35
0.0
1.0
2.0
company 17 item 235
Index
y[i,
]
0 5 10 15 20 25 30 35
0400
1000
company 17 item 236
Index
y[i,
]
0 5 10 15 20 25 30 35
02
46
8
company 17 item 237
Index
y[i,
]
0 5 10 15 20 25 30 35
0.0
0.4
0.8
company 17 item 238
Index
y[i,
]
0 5 10 15 20 25 30 35
0.0
1.0
2.0
3.0
company 17 item 239
Indexy[i,
]
0 5 10 15 20 25 30 35
0.0
1.0
2.0
company 17 item 240
Index
y[i,
]
Unclassified. Smart Software, Inc.
Output of Markov Bootstrap
307/24/2015 Unclassified. Smart Software, Inc.
Markov Bootstrap Algorithm
1. Code historical demands as 0 or X>0.2. Fit a 1st- order binary Markov model to data.3. Use Markov model to project demand
sequences over the replenishment lead time.4. Replace X’s in scenarios with random samples
from the set of observed nonzero demands (“bootstrap”).
5. Sum the projected lead time demands.6. Repeat steps 3-5 many times to build empirical
estimate the distribution of lead time demand.
7/24/2015 Unclassified. Smart Software, Inc.
Step 1: Parse History into Two Pieces
• Piece #1: Sequence of zero/nonzero demands.
7/24/2015
• Piece #2: List of all nonzero demand values.
Unclassified. Smart Software, Inc.
Step 2: Estimate Transition Probabilities
7/24/2015 Unclassified. Smart Software, Inc.
Step 3: Generate Demand Scenarios
7/24/2015 Unclassified. Smart Software, Inc.
Output of Markov Bootstrap
357/24/2015
Unclassified. Smart
Software, Inc.
Assessing ID Forecast Accuracy•ID is not “unforecastable”.•Goal is to support inventory management by forecasting the sum of ID over an item’s replenishment lead time
Need a “distribution forecast” not “point forecast”.•Traditional forecasting metrics not applicable.•Accuracy should be measured by the “calibration” of the forecast distribution.
Ex: If weather forecast says “80% chance of rain”, then it should rain on 80% of days with that forecast.Ex: After ID forecasts identify 95%iles of lead time demand, 95% of items should have demand <= their predicted 95th %tiles.
•Markov Bootstrap algorithm has excellent calibration.
7/24/2015 Unclassified. Smart Software, Inc.
Working Towards Solutions
Innovation Catalysts:• Defining the Readiness Equation
• Mission Based Forecasting
• Connect CBM to the Supply Chain
• Readiness Based Sparing
• Readiness Responsive Retrograde
• Leveraging Lessons Learned & Best Practices
RETROGRADE
ACQUISITION DEMANDWHOLESALE UNITRETAIL
Analyzing Root Causes and Prescribing Innovation
Catalysts Across the Supply Chain
Prognostic Demand
Condition Based Maintenance (CBM)
“Connecting” CBM to the Supply Chain
Remaining Useful Life (RUL)
39
RETROGRADE
ACQUISITION DEMANDWHOLESALE UNITRETAIL
CBM Data Warehouse
Regime Recog
Failure Analysis
Logistics Req’s
Prognostics
Environment
In Theater
PLM+ LMP
GCSS-A
Fleet Managers
IMMC
Platform PMs
LOGSA
Historical
Supply
Chain Data
OEMs
Condition & Health
Usage & Operations
Maintenance
Integration Opportunity:
Connecting CBM to the Supply Chain“Connecting” CBM to the Supply Chain: A Conceptual View
40
t3 t2 t1
A W R U
t3 t2 t1
A W R U
Downtime
Xf
MTBF
MLDT MTTR
Xf ?OST1 OST2 OST3
Reactive Repair Proactive Replacement
MTBR
OST MTTR
Xr
Down
timeMTBR
XrOST1 OST2
Reactive Repair Proactive Replacementvs.
“Connecting” CBM to the Supply Chain: A Mathematical View
Remaining
Useful
Life
Available
ost
Mean,Std. dev.
0,1
Normal Distribution
x
de
nsi
ty
-10 -8 -6 -4 -2 0 2 4 6
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time
TaTr
λ
Tf
CBM “Alert”(Ta)
Prognostic
Component
Replacement (Tr)
Tr – Ta > OST
Failure
Rate
Curve
From
Prognostic
Algorithm
Connecting CBM to the Supply Chain
(Tr)
(Ta)
Early demand
signal can drive
down retail
stockage
= Time to Replacement
Benefits of “Connecting” CBM to Forward Supply Chain
Wholesale
ReverseLogistics
RetailRDA UnitMissionDemand
CBM+
•Anticipatory requisitioning for proactive maintenance
•Supply Forecasting - Readiness Based Sparing (RBS)
•Reduced Enterprise Requirement Objective (RO) for Cost-Wise Readiness
Contributes to Achieving Cost-Wise Readiness
CBM+ = Early Warning
SSA
Inventory
LMP
RIMFIRE
Connecting CBM to the Supply ChainProject Process
Project/Process Flow:
Analyze Existing,
Actual Data
Baseline
Metrics
Develop
Prognostic
Simulation
Validate
Algorithm
Develop
Predictive
Algorithm
ASAP Data
DA Form
2410
DSC
Failure Rate Curve: Component HealthPrognostic Simulation Tool
Actual InventoryInventory Inventory CostReduction 0 2 4 6 8 12 16 Level Savings
0% 110.7 77.4 48.1 34.9 22.8 7.4 1.6 37 $0
2% 130.1 99.6 66.3 45.8 27.1 12.0 2.9 36 $51,801
5% 150.7 120.5 81.9 52.0 36.6 13.1 5.6 35 $103,602
10% 196.4 155.6 113.4 89.2 61.1 27.3 9.2 33 $207,204
15% 230.7 196.4 166.4 134.8 105.4 51.9 22.3 31 $310,806
20% 256.2 232.8 201.8 180.2 147.7 78.7 44.9 29 $414,408
25% 287.4 265.4 244.1 223.7 180.6 128.0 59.2 27 $518,010
Days Ordered Early (compared to historical requisition times)
# of Times an Aircraft was Down for More Than One Day
Expected Results Based on Improved Predictive Ordering
• 2 Variables, 7 levels each, 49 options, 90 simulation runs per option = 4410 total runs
Calibrated using actual 2410 data for AH-64D Nose Gear Box
Connecting CBM to the Supply ChainCBM Prognostics Simulation Model - Initial Results
Actual InventoryInventory Inventory CostReduction 0 2 4 6 8 12 16 Level Savings
0% 110.7 77.4 48.1 34.9 22.8 7.4 1.6 37 $0
2% 130.1 99.6 66.3 45.8 27.1 12.0 2.9 36 $51,801
5% 150.7 120.5 81.9 52.0 36.6 13.1 5.6 35 $103,602
10% 196.4 155.6 113.4 89.2 61.1 27.3 9.2 33 $207,204
15% 230.7 196.4 166.4 134.8 105.4 51.9 22.3 31 $310,806
20% 256.2 232.8 201.8 180.2 147.7 78.7 44.9 29 $414,408
25% 287.4 265.4 244.1 223.7 180.6 128.0 59.2 27 $518,010
Days Ordered Early (compared to historical requisition times)
# of Times an Aircraft was Down for More Than One Day
Expected Results Based on Improved Predictive Ordering
Calibrated using actual 2410 data for AH-64D Nose Gear Box
CBM Prognostics Simulation Model –Initial Results
Working Towards Solutions
Innovation Catalysts:• Defining the Readiness Equation
• Connect CBM to the Supply Chain
• Mission Based Forecasting
• Readiness Based Sparing
• Readiness Responsive Retrograde
• Leveraging Lessons Learned & Best Practices
RETROGRADE
ACQUISITION DEMANDWHOLESALE UNITRETAIL
(4) failure to proactively synchronize and manage reverse logistics which contributes significantly
to increased DLR RO, excess inventory, increased delay times (order fulfillment), and reduced
readiness while simultaneously precluding the enormous potential benefits of a synchronized,
closed-loop supply chain for DLRs;
Analyzing Root Causes and Prescribing Innovation
Catalysts Across the Supply Chain
Benefits of “Connecting” CBM to Reverse Pipeline
RDA Wholesale
ReverseLogistics
Retail UnitMissionDemand
CBM+
• Improve DLR induction forecast
•Forecast consumable Class IX
requirements maintenance workload
•Enable synchronized closed loop supply
chain for Maintenance Repair & Overhaul
(MRO) depots
Contributes to Synchronized Retrograde Process
CBM+ = Early Warning
Benefits of “Connecting”CBM to Demand Signal
RDA Wholesale
ReverseLogistics
Retail UnitMissionDemandCBM+
•Capture consumption/replacement data at unit
•Adopt point-of-effect demand segmentation
•Forecast Demand = f (Mission Based
Forecasting + Intermittent Demand + CBM+)
Contributes to Readiness Driven Supply Network (RDSN)
The Benefits of ConnectingCBM+ to the Supply Chain
CBM+
ReverseLogistics
Wholesale RetailRDA Unit Demand
Forward Supply Chain
Demand Signal
Reverse Pipeline
Contributes to Achieving Cost-Wise
Readiness
Contributes to Readiness Driven Supply Network
Contributes to Synchronized Retrograde
Process
Benefits
Quantifying the Benefits
Metrics Forward Supply Chain Reverse Pipeline Demand Signal
Readiness Return on Net Assets
Operational Availability
Materiel Availability
Backorders
Forecast Error
Metrics Forward Supply Chain Reverse Pipeline Demand Signal
Inventory (RO)
Inventory Value/Aircraft
Inventory Turns
Excess
Forecast Error
READINESS
INVENTORY
BURDEN
Metrics Forward Supply Chain Reverse Pipeline Demand Signal
Workarounds
Forecast Error
Draft Pre-Decisional Working Papers
Ultimate Goal
51
Improved alignment between maintenance and supply will
reduce excess and improve performance
Reduce Reduce
Right Part
Right Time
Backo
rders
Excess P
arts
Readiness
Risk
Underestimated
Demand
Overestimated
Demand
Fiscal
Risk
Army Consumption Data=Commercial POS Data
Data flow
Consumption
(parts in
aircraft
maintenance)
Recommended
Forecast Data Source
52
Adopting Mission Based Forecasting (MBF):
Key enabler for a “Readiness-Driven” Supply Network (RDSN)
Supplier Distribution
CenterStore
Room
Point
of Sale
(POS)
Supplier “Wholesale”
supply, aggregated
orders (requisitions)
“Retail”
supply(inventory across
multi-echelons)
Advanced Commercial Supply Chain
Army Supply ChainCurrent Forecast
Data Source
Current Forecast
Data Source
Focus:
Top-down
approach geared
toward meeting
inventory level
targets
Focus:
Bottom-up, POS
based approach
geared toward
meeting customer
demands
Commodity flow
Data flow
Parts flow
Data flow
Guiding Principles for Readiness-Driven Supply Networks
1. The purpose of the materiel enterprise is to sustain current readiness and generate
future capability.
2. Since readiness is “produced” by tactical (and training) units, these tactical
“consumers” represent the ultimate “customer”.
3. Actual consumer demand needed to produce “readiness” for training and
operational missions should drive the materiel enterprise - these are customer
“requirements” .
4. These requirements must be systematically measured and accurately forecasted at
the “point of sale” where readiness is produced by the consumer.
5. Demand planning across the enterprise must focus on meeting these requirements
(for effective performance) while reducing forecast error (efficient performance).
Align the Class IX supply chain to “real” customer demand,
then pursue Continuous Performance Improvement efforts and initiatives
focusing on “Cost-Wise Readiness” for Army Materiel Transformation
Working Towards Solutions
Innovation Catalysts:• Defining the Readiness Equation
• Mission Based Forecasting
• Connect CBM to the Supply Chain
• Readiness Based Sparing
• Readiness Responsive Retrograde
• Leveraging Lessons Learned & Best Practices
RETROGRADE
ACQUISITION DEMANDWHOLESALE UNITRETAIL
Analyzing Root Causes and Prescribing Innovation
Catalysts Across the Supply Chain
Readiness Based Sparing (RBS)
Ao
$ Spares
Marginal Analysis Includes:
• Cost of Parts
• Frequency of Use/Need
• Part Impact on Readiness
-
-
-
6th A
11th B
2nd C
12th B
1st D
7th A
-
-
-
-
-
-
1,600
2,300
10, 400
2,300
13,800
1,600
-
-
-
-
-
-
0.388
0.352
0.312
0.283
0.154
0.144
-
-
-
-
-
-
101.600
103.900
114.300
116.600
130.400
132.000
-
-
-
-
-
-
66.67
66.69
66.74
66.76
66.78
66.79
-
-
-
Item UnitCost($)
AddedAircraft/$10K
TotalCost($)
AvailabilityRate(%)
Shopping List
65
70
75
80
85
90
95
100
0 1 2 3 4 5
Readiness
Target
ASL Investment in $M
$.5M
$.6M
$.8M
$1.2M
$2.2M
$4.8M
Source: AMSAA
RBS Curve:
“The Efficient Frontier”
Results
x
?
Analytical Demonstration:
Readiness Based Sparing: 101st ABN DIV UH-60
57
Innovation Catalyst: Readiness Based Sparing
65
70
75
80
85
90
95
100
0 1 2 3 4 5
Readiness
Target
ASL Investment in $M
$.5M
$.6M
$.8M
$1.2M $2.2M
$4.8M
Source: AMSAASource: AMSAA
RBS Curve:
“The Efficient
Frontier”
Analytical Demo 101st ABN DIV UH-60 Results
xx
Conditions:
- Low $ parts were causing Army weapon systems NMC
- “Readiness Based Sparing” (RBS), developed at RAND
and LMI, had not been tested for Army Aviation
Research Results:
- Analytical Demo revealed significant potential to reduce
costs and relate investment levels to Ao. . . RBS later
adopted at Fort Rucker.
- Multi-Echelon RBS exhibits tremendous potential for
cost savings and relating resources to Ao fleetwide.
0
1
2
3
4
5
0 50 100 150 200 250 300 350 400
Percent
Increase
In
Readiness
Percent Increase in
Investment at Wholesale
Fill Rate Safety Level Readiness
70 189M 84.7%
75 210M 85.9%
80 256M 86.7%
85 340M 87.3%
90 505M 87.7%
95 857M 88%
Baseline:
Impact of Increased Investment at Wholesale on Blackhawk Equipment Readiness at 101st Airborne
Source: AMSAASource: AMSAA
RBS Impact?
Innovation Catalyst: Multi-Echelon Readiness Based Sparing
Part III. Enterprise Integration: Prescriptive Analytics for
Efficiency, Resilience, and Effectiveness
Achieving “Efficiency” in the Cost -
Availability Trade Space
“Efficient Frontier”
Ao
$
Gain in
“Efficiency”
Increasing “Effectiveness” in the
Cost -Availability Tradespace
“Efficient Frontier”
Ao
$
4
32 1
Cost Benefits Alternatives:
1. Improved effectiveness with increased costs
2. Improved effectiveness at
same costs
3. Improved effectiveness at
reduced costs
4. Same effectiveness at
significantly reduced costs
… however, magnitude of each
depends upon where you are on
the current efficient frontier!
… and the expansion trace of the
improved frontier
12. Achieving Efficiency: An Integrated Multi-Echelon Inventory Solution
13. Designing for Resilience: Adaptive Logistics Network Concepts
14. Improving Effectiveness: Pushing the Logistics Performance Envelope
59
NICP
DDOC
SSA
ASL
ASL
ASL
A
Design for Resilience: Demand Driven Supply
Network (DDSN)
DDOC
DDOC
SSA SSA
SSA
SSA
SSA
SSA
SSA
SSA
• RBS reduces cost
• Inventory pooling reduces both
cost and risk
• Lateral supply decreases
requisition delay time & increases
Ao
RBS
Stock
List
$
Hi Cost-
Low Demand
DLRs
• Low Cost
Consumables
• Hi Demand
Parts
RBS
Cost/Item
List
Design for Structural Resilience: Readiness Driven Supply Network
60
Pursuing Cost-Effective Readiness:
Pushing the Performance Envelope
Increasing “Effectiveness” in the
Cost -Availability Tradespace
“Efficient Frontier”
Ao
$
4
32 1
Cost Benefits Alternatives:
1. Improved effectiveness with increased costs
2. Improved effectiveness at
same costs
3. Improved effectiveness at
reduced costs
4. Same effectiveness at
significantly reduced costs
… however, magnitude of each
depends upon where you are on
the current efficient frontier!
… and the expansion trace of the
improved frontier
61
“Optimizing” the System: Applying a Dynamic (Multi-Stage) Programming Model
0
1
2
3
4
5
0 50 100 150 200 250 300 350 400
65
70
75
80
85
90
95
100
0 1 2 3 4 5
Acquisition
Design Stage
Wholesale
Stage
Retail
Stage
Unit Production
Stage
$$ $$ $$ $$ AAoo
Perc
en
t In
cre
ase
In
Read
iness
Percent Increase in Investment at Wholesale
Re
ad
iness
Targ
et
Investment in $M
$.5M
$.6M
$.8M
$1.2M
$2.2M
$4.8M
Lab
or
Lab
or
CapitalCapital
90% MC90% MC
80% MC80% MC
112233NN
Automated Monitoring
Management
Assessment
Policy Response
Feedback
Alert Warning
- Readiness trends and forecasts
- Supply chain metrics
- Logistics system readiness parameters
- Corroborate and validate alerts
- Assess near and long-term implications
- Integrate empirical evidence with human judgment
- HQDA reviews
- Analyze and implement cost-effective options
- Minimize recognition and response bags
- PPBES implications (resources-to-readiness)
Logistics Readiness and Early Warning SystemLogistics Readiness Early Warning System
The regression relating Mission Capable rates (MC) to age lagged 5 months, shown in
the equation below, indicates that a one-month increase in backorder average age
leads to a reduction of 2.8 percentage points in MC rate 5-months hence. The
coefficient is highly significant (at the one percent level), and the R2 is 63 percent.
MC = 0.97 – 0.028 (Age lagged 5 months)
CONUS Europe
Pacific
PMO
Apache
SWA
Ao
Reset
Available
Ready
$
Integration Opportunity: ARFORGEN Synchronization -
MBF and RBS
CONUS vs IraqStabCONUS vs IraqMCO
59% 41%
73% 27%
CONUS IraqStab
59% 41%59% 41%
73% 27%73% 27%
CONUS IraqStab
12%88%
CONUS IraqMCO
71% 29%
12%88%
CONUS IraqMCO
12%88% 12%88%
CONUS IraqMCO
71% 29%71% 29%
111 84 87111
2833 2833
La
bo
r
Capital
80% MC60% MC
90% MC
RBS Curve:
“The Efficient
Frontier”
Integration Opportunity: RBS and MBF for the
Army’s new Regionally Aligned Force Concept
Integration Opportunity: Logistics Support for
Capabilities Based Planning
REVISED CLASS IX STOCKAGE POLICY
*EG: IDEEAS, JANUS,
JCATS CASTFOREM,
JTLS
MISSION
SCENARIOS
ReadinessEquation
CONUS vs Iraq Stab
73% 27%
CONUSIraqStab
59% 41%
Integration Opportunity: “Advanced Analytics” for a Capabilities Based Force
Defense Planning Guidance
Scenarios
Integration Opportunity: Product Support Integration (PSI) for PBL
Aligning PBL Incentives to Readiness Outcomes
Value of
Cost or
Output
(Return)
Decision or Stopping Points
(Iterations)
Maximum difference
between Total Revenue
and Total Cost or
Maximum Profit
Total Cost Function
Production Function
(A0 * VPC)
Steep here means
shallow here
Tangents
of equal
slope
PBL Contract Scoring Regime Results
0
0 5025
10
2015105 403530 45
9
3
2
1
6
5
4
8
7
($ 000,000s)
Inventory Value* ($ 000,000s)
Legend
Award Fee
Cost
Profit
Max Profit
The Fallacy of ‘Fill Rate’ as an Incentive
for SC Performance
0
0 5025
1.0
2015105 403530 45
.9
.3
.2
.1
.6
.5
.4
.8
.7
Ao/FR/Score
Inventory Value* ($ 000,000s)
0
100
90
30
20
10
60
50
40
80
70
Legend
Ao
Fill Rate Avg. Delay
Max Profit
Score
What the
customer got
What the
customer
wanted
Avg. Delay
Integration Opportunity: Product Support Integration for
Performance Based Logistics (PBL)
Draft Pre-Decisional Working Papers
MVC Principles
Point of Effect
Data
Point of Effect
Maintenance
Planning and
Forecasting
Inventory
Demand
Planning
Close
Collaboration
Between Supply
& Maintenance
Functions
Supply Chain
(Asset) Visibility
Performance
Measures
Utilized
Level
1
Collect data at point
of effect using ad hoc
data collection
systems
Forecasting
scheduled
maintenance using
OPTEMPO as
primary driver
Forecast based
predominantly on
historical demand
Supply and
Maintenance
functions act nearly
independently;
limited comms
between functions
No visibility of
inventory levels
across wholesale
and retail levels
Metrics that only
focus on wholesale
supply goals (i.e. Fill
Rate)
Level
2
Authoritative data
sources, made
available to
authorized users;
quality improvements
to data collected
Level 1 +
Forecasting
unscheduled
maintenance using
mission based
forecasting (MBF)
Forecasting models
plus analytics for
different types of
demand at different
echelons
Scheduled
maintenance
forecasts are shared
with supply planners
Inventory visibility
across unit and retail
levels, within units
supported by SSA
Metrics that focus on
supply goals at all
echelons
(operational, retail
and wholesale)
Level
3
Systemic processes
to enhance
collection, cleansing,
and auditing; Data is
used to inform
decision making at
multiple levels
Level 2 +
Forecasting
unscheduled
maintenance using
segmented MBF (i.e.
CBM+, intermittent
demand, etc)
Driven by Point of
Effect Level 3
forecasting
Systemic feedback
processes to monitor
progress,
performance, and
cross-functional
effectiveness
Transparency of
inventory levels
across all echelons
of supply chain,
including In-transit
Visibility
Metrics focused on
organizational goals
(i.e. readiness);
Metrics that
incentivize cost
savings
DRAFT MVC Principles Maturity Model
67
Bringing it All Together
Other Uses of the MLNPS
• Plan Fuel Networks
• Reverse Logistics
Logistical Planning tool is needed
• Must support fast-paced, frequently
changing expeditionary operations.
• Must rapidly determine capability
requirements as operational
requirements change.
RAND Study
MLNPS
ERP
System
Expeditionary
Operation
Mission-
Based
Forecasting
Current State
of System
Network
MDMP
Constraints
Mission
Specific
Demand
Forecast
DP
• Decision Point for Commander
• Forecasted Information for Subordinate
Units & Critical Nodes
• Reduce Uncertainty (Fog of War)
MLNPS as the app
There’s an app for that!
Sustaining Innovation While
Linking Execution to Strategy
Capacity, Inventory & Knowledge
Capacity:
What we
can do
What we
knowWhat we
have
Substitutable Ingredients of System
Performance
Knowledge: Inventory:
Management Innovation as a Strategic Technology
Customer
Needs
Methodology
Advancement
Technology
Enablers
Management
Innovation:
•MERBS1
•MBF2
•R33
•DSLP4
•LREWS5
Technology
Innovation:
•CBM6
•RFID7
•TAV8
•ERP9
6Condition Based Maintenance
7Radio Frequency Identification
8Total Asset Visibility
9Enterprise Resource Planning
1Multi Echelon Readiness Based Sparing
2Mission Based Forecasting
3Readiness Responsive Retrograde
4Dynamic Strategic Logistics Planning
5Logistics Readiness and Early Warning
System
71
Academia
Corporate
Research
Tactical Units Academic
Institutions
Commercial
SectorPrivate
(1)
Magnet, Filter and
“Repository”
for
“Good Ideas”
(2)
Modeling,
Simulation
& Analysis of
Complex
Systems
Public
(3)
Transforming
Organizations &
Managing
Change
DoD
Organizations Professional
Societies:
INFORMS
FFRDC’s
Non-Profits
• Organizational Design
• Supply/Value Chain
• Workforce Development
• Technology Implications
• Innovation & Productivity
Gain
• R & D
• System Dynamics Modeling
• Large Scale System
Design, Analysis, and
Evaluation
• Systems Simulation,
Modeling and Analysis
• Repository for validated
models & analytical tools
• Cost Benefit Analyses
• Risk Reduction & Mitigation
• Research, Studies, and
Analysis
• Education & Training
• Technical Support
• Change Management
An “Engine for Innovation”:
The Center for Innovation in Logistics Systems (CILS)
http://www.nap.edu/catalog/18832/force-multiplying-technologies-
for-logistics-support-to-military-operations
Reasons for the Book (from Preface):
1. Resurrect traditional Operations
Research (OR) for the US Army.
2. Apply “advanced analytics” to our
materiel enterprise challenges.
3. Link operational, technical,
educational, scientific, and
analytical communities.
4. Demonstrate “Management Innovation
as a Strategic Technology”.
5. Document a case study for:
analytically-driven,
transformational change;
a comprehensive, collaborative
effort by many contributors.
Recoupling Forecasting with Inventory Control and
Supply Planning: “Readiness-Driven Supply Networks”
Greg H. Parlier
Colonel, US Army, Ret
2017 FORECAST PRACTITIONER CONFERENCE
NCSU IAA 15 November 2017
Mission Based ForecastingMBF BENEFITS
•Ability to accurately predict tactical-level demand
•Measure the actual cost of operational requirements
•Align/manage inventory to readiness-driven demand
•“Connect” CBM prognostics to forward supply chain
•Synchronize retrograde and depot repair operations
•Defend resources needed for mission Ao requirements
•Reduce tactical-level “burden” and work-arounds
•Enable “early warning” for sustainment enterprise
•Cost savings estimates: tens of billions $
•MBF investment ROI of several orders of magnitude
•Significantly improve tactical unit-level operational Ao
TECHNOLOGY SOLUTION
To identify potential “catalysts for innovation”, the US Army established the project to Transform Army Supply Chains (TASC). Mission Based Forecasting (MBF), a critical enabling catalyst, is a new concept for demand planning which capitalizes on big data, the Internet of Things, and predictive analytics to support military operations. MBF testing suggests order-of-magnitude reductions in forecast error, inventory savings of billions of dollars, and reduction of manpower-intensive work-arounds at the tactical level.
PROBLEM STATEMENT
The US Department of Defense (DoD) operates the
most complex global supply chains in the world.
However, effectively integrating production
planning, maintenance operations, inventory
systems, and distribution policies has been a
strategic challenge. According to the GAO, DoD
supply chain management is both wasteful and hi-
risk due to poor demand forecasting, ineffective
inventory management, and inadequate strategic
planning.