Edelman Uscg Final

Operations Research Enhanced Supply Chain Management at the US Coast Guard Aircraft Repair and Supply Center

Team Members

• USCG Team – – – –

CDR Carl Riedlin LCDR Mike Shirk LCDR Kent Everingham LCDR Gary Polaski

• Purdue Team – Prof. Vinayak Deshpande – Prof. Ananth Iyer 2

Cultural Transformation Through …

USCG ARSC + Purdue = OR ingrained

+

=

3

Route of Flight • • • •

US Coast Guard Roles & Missions Logistics Network Ops Research & Purdue/Coast Guard Partnership Four Projects: – MIDAS, REAP, CRISP and OPT

• Impact & Organizational Transformation

4

5

Aviation Facility Location & Allocations Port Angeles 3 HH-65C Cape Cod 4 HU-25C 4 HH-60J

Traverse City 5 HH-65C

Astoria 3 HH-60J North Bend 5 HH-65C

Detroit 5 HH-65C

Humboldt Bay 5 HH-65C Sacramento 4 HC-130H San Francisco 4 HH-65C

Elizabeth City 4 HC-130H 5 MH60-J 1 HC-144A

ATC Mobile 4 HU-25A 7 HH-65C 3 MH-60J

Los Angeles 3 HH-65C San Diego 3 MH-60J

Savannah 5 HH-65C

Houston- 4 HH65C New Orleans- 5 HH-65C Kodiak 5 HC-130H 4 HH-65C 4 MH-60J

Corpus Christi 3 HU-25C 3 HH-65C Barbers Point 4 HC-130H Sitka 3 HH-60J 4 HH-65C

Atlantic City 10 HH- 65C Washington 1 C-37 1 C-143

Clearwater 6 HC-130H 9 HH-60J

HITRON 8 MH-68A Miami 6 HU-25D 9 HH-65C

Borinquen 4 HH-65C

6

HC-130: 22 Operational - 5 Support / HU-25:17 Operational - 8 Support / HH-60: 34 Operational - 7 Support / HH-65: 84 Operational - 11 Support

Aircraft Repair and Supply Center (ARSC) • One stop shop for all aviation logistics support – – – – –

Depot level maintenance Supply engineering Spare parts inventory management Component repair Information services

• 60,000 individual parts, – Inventory value over $937 million

• Annual maintenance budget over $154 million for over 6000 parts 7

Un-Scheduled Failures

Air Station Repair

ACMS Data

AMMIS Data

Warehouse

Scheduled Maintenance

Maintenance Shop

Item Managers

- Air Station

Procurement Specialists

- ARSC

Component Repair (Internal)

Vendors (OGA & Commercial)

failed parts good parts

Clear & Present Danger… • Highly complex supply-chain • Various groups focused on a specific task – – – –

Reliability Center Analysis Inventory Replenishment & Budgeting In-House Repairs & Capacity Management Procurement & Best-Value Contracting

• Need for information collaboration between groups • Impending “Brain Drain” in Federal Gov’t – Item Manager / Procurement Specialists Retirement

9

A Partnership is Formed • Purdue a Best Value in Advanced Education – Coast Guard officers in MBA & Structures programs since 1970s

• Initial contact in 2001 with 2 goals – Validate OR capability with ARSC – Lead cultural change in face of budget & people crisis

• Prof. Deshpande & Iyer form a team • Exhaustive review of supply-maint business processes • Concise project definition & contracting deliverables – MIDAS = Turning Data into Gold

10

Project 1: MIDAS (June 2002-April 2003) “Improving Aircraft Service Parts Demand Forecasts and Inventory Management using Scheduled Maintenance Data” 3 Main Tasks 1. Integrate ACMS maintenance and AMMIS demand data 2. Build Demand Forecast Models 3. Policies for effective inventory management using maintenance data 11

MIDAS Methodology and Results

• Gathered extensive maintenance and demand data on 41 critical components consisting of 50% of budget • Created a Linear Programming model to link the maintenance data and demand data for these 41 components

12

Database Match Example ACMS maintenance Database

AMMIS demand Database

RCi

ii

ri

cj

sj

pj

TRBL

11/25/97

11/25/97

10/16/97

10/17/97

12

TIME

07/09/98

07/09/98

11/20/97

11/24/97

12

TIME

07/30/98

07/30/98

07/07/98

07/07/98

2

TRBL

10/28/98

10/28/98

10/29/98

12/02/98

5

TIME

01/05/99

01/05/99

03/11/99

04/13/99

5

TIME

04/15/99

04/15/99

03/18/99

03/22/99

5

TIME

04/22/99

04/22/99

11/08/99

11/15/99

5

TIME

12/14/99

12/14/99

08/24/00

08/25/00

2

TRBL

08/28/00

08/28/00

05/01/01

05/23/01

2

TIME

06/06/01

06/06/01

06/15/01

06/25/01

2

TRBL

06/29/01

06/22/01

06/25/01

07/10/01

2

TRBL

07/20/01

06/29/01

03/05/02

04/19/02

5

TIME

05/08/02

05/08/02

TRBL

08/08/02

08/08/02

13

Young Parts

ACMS Data

Old Parts

Good parts at Warehouse

Failed parts at Warehouse L1

L2

IM’s

AMMIS Data

Component Re-Supply

failed parts good parts re-supply order

Part-Age Distribution of Installed Parts Cumulative Distribution

100% 80% 60%

Oct. '98 Oct. 2000

40% 20% 0% 0

500

1000

1500

2000

2500

3000

TSO

Time Since Overhaul

15

SIGNAL Dependent BASE STOCK • SIGNAL each period = # parts beyond Threshold age • Correlate the SIGNAL and Demand over lead time • Use the conditional distribution and costs to set BASE STOCK = Function(SIGNAL) • Evaluate the optimal THRESHOLD • Empirical Results showing cost impact on Inventory levels • Proactive Inventory Management 16

Data Becomes Gold (1) LP tools used to match maintenance records and inventory (2) Part Age (accumulated flight hours) information improves demand forecast (3) Part Age based triggers for advance orders (4) Empirical results show cost reductions ranging from 20% to 70% for over 90 % of the parts examined (5) The advance orders enable separation of supply processes for replenishment and advance orders and can be used for budgeting 17

MIDAS Project Organizational Impact • Establishment of OR Cell – 4 new positions & summer interns – Expanded budget planning-execution authority

• Demand forecasting & budgeting using MIDAS – Partnering with Item Managers

• Supply Chain Management Business Solution – – – –

Scalable, repeatable, supportable solution Up-to 100,000 stocking units Data warehouse AMMIS – ACMS bridge 18

19

Project 2: REAP (June 2003-May 2004) “Improving Scheduling of Repairs of Parts using Scheduled Maintenance Data” Main tasks: (1) Understand current repair release approaches using empirical data (2) Develop Component Repair Capacity Planning Models to choose the optimal repair mix including internal vs vendor repair choices. (3) Estimate the impact of adjusting releases on performance of the repair shops 20

Components vs TSO

The data shows that as part age increases, the number of components to be replaced and labor content increases

21

Role of LP Models • The optimization models try to minimize costs while maintaining safety stock of parts and completing contracted number of part repairs in the shop • Models capture the impact of shop costs using vendor capacity, using overtime, coordinating repair releases with IMs etc.

22

REAP Optimization Model Minimize Sum of in  repair, vendor Min ∑ ∑ CGi X it1 + ∑ ci X it2 + ∑ CO j OR j  house repair and overtime t  i i j  capacity costs s.t ∑ aij X it1 ≤ R tj + OR tj Resource availability constraint i

I ti −1 + X it1 + X it2 − ( DAit + DRit ) = I ti I ti ≥ Z ⋅ σ ( DRit )

t c X ∑∑ i i 2 ≤ B t

i

Demands composed of advance and regular orders

Safety stock only needed for regular orders Budget constraint

Where i is for NIIN and j for shop or resource.

23

REAP Methodology and Key Results • Data Analysis to link – AMMIS – ACMS – Extended WO data

• Link TSO to labor, material data • Optimal component repair capacity planning LP models • Use models to project a 10% savings in repair costs

24

REAP Project Organizational Impact • Developed 2 bill of material (BOM) lists based on TSO (young vs old failures) • Linked BOM to extended work order • Used BOM to assist with budget builds • Shadow price of $ 6500 per hour for specific resources • Adjusted resources and skills – Release shop repairs to optimally use scarce shared repair resources – Coordinate IM management – Use material usage data to create realistic BOM 25

Upgrade of HH65 “Dolphin” Helicopter

26

27

Restore Unrestricted Ops • Restore safety, increase payload and operational flexibility • Retrofit HH65 with “Turbomeca” engines • Improve gearbox durability – upgrade 135 units • Urgent requisition - 2 year completion mandate • How do we accomplish this?

28

Project 3: CRISP (June 2004-May 2005) •

CRISP required a model of the impact of: – – – – – –

Availability of overhaul kits and parts Planned overhaul vs modification Associated spare parts required Overhaul interval Future overhaul Different levels of aircraft operation (C with G4, C with G2, switch back and forth, etc.) – Resources available

29

States and Transitions t

t+L1+1

t+L2

t+L2+1

t+L2+MTTF2

t+2

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

RFI G2

5

5

5

5

5

5

5

5

5

5

RFI G4

6

6

6

6

6

6

6

6

6

6

G2 on B Aircraft

7

7

7

7

7

7

7

7

7

7

G2 on C Aircraft

8

8

8

8

8

8

8

8

8

8

G4 on C Aircraft

9

9

9

9

9

9

9

9

9

9

NRFI G2

t+L1

t+L1+MTTF1

t+1

t+L2+MTTF3

NRFI G4 In Repair G2 In Repair G4

Grounded B

1 0

1 0

1 0

1 0

1 0

1 0

1 0

1 0

1 0

1 0

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

1 1

Grounded C In mod-line In PDM line

1 2 1 3

1 2 1 3

1 2 1 3

1 2 1 3

1 2 1 3

1 2 1 3

1 2 1 3

1 2 1 3

1 2 1 3

1 2 1 3

Features of the MIP Model • Model possible states of individual aircraft and individual components • Evolution of configuration over time • Impact of overhaul or modification schedule • Impact of constraints on level of flexibility (B,C-G2,CG4, switchover) • Weighted by upgrade level over horizon to maximize aircraft uptime • Mixed Integer Program with network substructures

31

Number of Flying Aircraft by Type Over Time

32

Results of the Model • Model results highlighted the impact of part availability constraints on upgrade process • Quantified impact of: – level of aircraft operation flexibility on performance – manufacturer suggested mean time to overhaul for new components – spare parts availability

33

34

CRISP Results and Organizational Impact • Bottlenecks reduced by productivity changes – Dual conversion paths – Lean manufacturing

• Building block for additional analysis – Spend plans – Fleet sparing

• Catalyst for a successful on-time conversion

35

Deferred Maintenance Crisis

36

Project 4: OPT (June 2005-Present) • Coast Guard’s “Big Iron” – 300 NM radius • essential to cover EEZ • Alaska and Caribbean – Critical role in massive disaster relief • As showcased during Hurricane Katrina

• ARSC HH60 Product Line

– PDM – corner stone process – every 4 yrs – Logistics and Engineering Support

• Impending failure – Mission scope creep following 9/11 – Aging airframe and extensive corrosion

• Diminishing overhaul throughput – Dropped from 9 in 1999 to low of 5 in 2005 37

Impending Disaster

• The HH-60 deferred maintenance burden on USCG reached an all time high of $23.6 Million • The Impending Train Wreck Operational Groundings – Starting Mar 07 – 24% of the Coast Guard’s operational fleet

• In order to begin a road to recovery, the HH-60 Product Line needed to rapidly increase its throughput

38

Original PDM Line Base Case Disassembly (7 days)

Start Aircraft Flies in from Field Unit

Strip (7 days)

Hull (58 days)

Hull (58 days)

Intermediate Paint (4 days)

Assembly (54 days)

Assembly (54 days)

Aircraft Returns to Field Unit End

Final Paint (9 days)

Ground Turns (12 days)

Fuel Ops (11 days)

39

OPT Project Goals

• Capture the links between decisions regarding – – – –

resources inventory repair rules lead times

• Identify bottlenecks and the benefits of implementing improvements • Design plans for MH-60T aircraft conversion 40

ARENA Simulation Model • Captured three sets of flows in the system – Aircraft cycles of field missions and PDM overhaul – Component flows on flying aircraft, inventory and repair – Modules (within components) flow on components, field failures, inventory and vendor repair

• Each flow has different criteria • Model includes component repair, vendor repair lead-times, contracts, priority rules, repair triggers, etc. 41

OPT : Methodology • Closed-loop Queuing and Inventory simulation model in ARENA – Identified bottleneck processes • Hull rework • Final Assembly – Capture impact of changes on the production line – Impact of different rules for triggering module repair – Impact of improving processing times through lean events – Impact of inventory positioning (ARSC vs field) – Impact of WIP inventory changes – Impact of resource level changes 42

Original PDM Line Base Case 9.14

9.64

Assembly stations 3

Hull stations 3

9.14 6.55 4 aircrafts

8.06 5 aircrafts

2 Hull Repair VV 2 Assembly 0 hull inventory 6aircrafts

9.31

CC

1 hull

9.29

7 aircrafts

9.29

Module life 500

9.33 2 hulls

9.2 Module life 0

9.2

8 aircrafts 9.31

43

OPT Results and Organizational Impact • Reduced process cycle time – 200 (+) working days down to an impressive 145 – Eliminated $5M annual outsourcing initiative

• Analysis used to drop plans to add 2 hulls at a cost of $ 10 million (low throughput impact) • This resulted in an increase in throughput by 80% and a drop in deferred maintenance burden from $23.6M to a mere $6.5M

44

Four Projects

Missions

Air-station CRISP Aircraft Type

Upgrade

OPT

MIDAS

Repair

ARSC/E-city

Inventory

ARSC Repair

Vendor REAP

Research Impact • Each project led to an innovative idea with solid OR analysis as summarized in the table

Project

Concept

MIDAS

Part Age based Inventory Levels (published in Operations Research)

REAP

Repair portfolio of ages

CRISP

Linking upgrade rates to future part demand, including effect of flexibility

OPT

Inventory, Capacity, contracts and closed loop supply chains 46

At the End of The Day Quantifiable, measurable, tangible benefits… • MIDAS – inventory reduction for critical parts • REAP – 10% cost avoidance with $200MM budget • CRISP - Unrestricted H65 Operations • CRISP – Capacity constraints highlighted – Catalyst for Lean Manufacturing with $1.2MM savings

• CRISP - Redirect $9.9MM for component sparing – Gear box, engine control system with long lead-time

• OPT – Ended H60 deferred maintenance crisis – 80% increase in overhaul throughput 47

Organizational Impact of OR Projects • Establishment of an Operations Research cell – Several new employees and interns hired – Provides critical decision support tools for planning repair and maintenance activities – Overwhelming increase in requests to analyze logistics issues – All new projects expected to be grounded with OR analysis – Supply Chain Management System (SCMS) being implemented to leverage information sharing and OR applications across the enterprise

• Future Initiatives: Aircraft availability simulation

48

Social Impact of OR Projects • Total cost savings in excess of $ 70 million • CRISP returned the H-65 aircraft to a safe, reliable platform • Prevented grounding of 9 HH-60 aircraft which would have resulted in loss of 6300 mission flight hours – Replacing these aircraft would have cost USCG $270 million • Long procurement lead-times and budgetary realities would make that infeasible

• True impact of grounded aircraft would have been a drop in mission readiness from 100% to 96% and mission execution drop to 4%. • The social cost would have been our inability to respond to natural disasters such as Hurricane Katrina – 33,000 rescues, 5,000 by the HH-60

49

50

Questions?

51