Read Johnson_Drakeley_Plante_Dalton_Trost-ReviewCopy.pdf text version

Naval Engineers Journal

Managing Change on Complex Programs - VIRGINIA Class Cost Reduction

Manuscript ID: Manuscript Type:

Date Submitted by the Author: Complete List of Authors:

r Fo

Journal: Keywords:

Naval Engineers Journal NEJ-2008-11-STP-0111.R2 Symposium: Technical Paper 19-Mar-2009

Johnson, David; US Navy, Deputy Commander for Undersea Technology SEA 073 Drakeley, George; US Navy, NAVSEA PMS 450 Virginia Class Plante, Thomas; GD Electric Boat Dalton, William; PA Consulting Group, Federal and Defense Services Trost, Christopher; PA Consulting Group, Federal and Defense Services change, dynamics, program management

Pe

er

Re

vi ew

Page 1 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Rdml. David C. Johnson, USN, George M. Drakeley, Thomas N. Plante, William J. Dalton, Christopher S. Trost, P.E.

Managing Change on Complex Programs ­ VIRGINIA Class Cost Reduction

ABSTRACT

Today's ship acquisition environment demands the delivery of more ship for less money. The VIRGINIA Class Submarine Program faced this challenge explicitly, with its mandate to achieve cost reduction of approximately $400 million per ship by FY 2012 as a necessary condition for increased production. For VIRGINIA Class, accomplishing this objective meant implementing a number of significant cost reducing changes to a mature design and hundreds of construction processes ­ ensuring that these changes enhanced rather than disrupted the "normal" learning occurring in the serial submarine construction. Managing these changes and their impact on ongoing efforts to achieve good learning from ship to ship (also known as the learning curve) was further complicated by interactions between learning and other factors like changes to production rate, funding profiles, construction schedules and the workforce. Performance of the VIRGINIA Class Program was dynamically simulated to determine in advance the likely full consequences (including indirect impacts) of cost reduction actions in the context of evolving program conditions. The simulation of VIRGINIA Class Program performance captured individual factors whose improvement together drives the classic learning curve ­ for example, design maturity, material availability, staff experience and so on ­ and was validated against historical program performance. This simulation capability was used to evaluate the full impact of potential changes, individually and in combination, along with sensitivity to a variety of different future scenarios. The analysis provided insights regarding the degree of disruption from different cost reduction changes, synergies between these changes, and interactions with other program conditions. This provided an independent "reality check" on the path taken by the VIRGINIA Class cost reduction effort, guidance regarding key sensitivities and risks to be managed, and data that supported contracting for the next block of ships.

r Fo

INTRODUCTION

US Navy shipbuilding and other DoD acquisition programs are under intense cost pressure. Limited discretionary acquisition budgets are competing with war spending as well as other domestic initiatives. Efficient use of today's acquisition budgets is critical to recapitalizing the fleet, and maintaining or rebuilding tomorrow's force levels. The Navy is trying to stop the decline in the number of ships in the fleet, and restore today's fleet of 283 ships back to 313 ships without significantly increasing the shipbuilding budget. To achieve this objective, ships must meet their cost targets and become more affordable - or they will likely be procured in smaller numbers. The 2006 Quadrennial Defense Review (QDR) specifically defined the cost threshold the VIRGINIA Class submarine had to achieve to increase production to two per year, and maintain the 48 submarines needed to fulfil Navy operational requirements. The 2006 QDR directed the Navy to "return to a steady-state production rate of two attack submarines per year not later than 2012, while achieving an average per-hull procurement cost of $2.0 billion (fiscal year 2005 dollars)." This overall objective became known as "2 for 4 in 2012" (Hilarides 2006). Meeting this objective would require changes across the entire program including design, procurement and construction - that could only be achieved through close government­industry collaboration.

Pe er

Re

vi

ew

Naval Engineers Journal

Page 2 of 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Starting point

By Fiscal Year 2005 (FY05) ­ more than 10 years into the VIRGINIA Class Program ­ Congress had authorized seven submarines, and construction had proceeded on those. Two had been christened, and one commissioned ­ USS VIRGINIA (SSN 774). On a comparative recurring cost basis in FY05 dollars, cost performance was improving. But as Figure 1 indicates, the labor hour reductions forecast by a traditional learning curve for continued production of the submarines would not meet the cost threshold by FY12 (for discussion of shipyard learning effects see O'Rourke 1996). While changing the design offered significant potential to reduce costs (Trost 2000), more would have to be done to reach the target and any redesign effort would also have to account for the disruption impacts on ongoing construction. Note too that approximately 30% of the cost of each ship is government furnished equipment (GFE), and comprehensive reduction

of cost needed to be achieved across all stakeholder organizations. Cost reduction program The VIRGINIA Class Program Office and the industry shipbuilders - prime contractor and lead design yard General Dynamics Electric Boat (GDEB), and subcontractor and second delivery yard Northrop Grumman Newport News Shipbuilding (NGNN) ­ had a common goal and strong incentive to achieve "2 for 4 in 2012" in order to allow the increased production rate. GDEB working with NAVSEA established an engineered systematic approach to drive cost from the VIRGINIA Class submarines. This approach started with analysis of the cost drivers from returned construction costs to date, and developed cost targets by area. Based on these targets, a cost framework was used to guide decisions about cost reduction actions, and a comprehensive program plan was established to integrate and implement the effort.

Submarine Cost Cost Target

r Fo

776 777

Pe er

778 779

3.5

3

Re

2.5

vi

2 $B

ew

1.5 774 775

1

0.5

0

FY99 FY99 FY01 FY02 FY03 FY04 FY05 FY06 FY07 FY08 FY09 FY10 FY11 FY12

FIGURE 1. 2005 outlook indicated assumed learning on VIRGINIA Class would not achieve FY12 cost target of $2B without significant changes

787

Page 3 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Three primary sources of cost reduction were targeted (Johnson, Drakeley, and Smith 2008): · Increasing procurement rate to two ships per year in a multi-year procurement (MYP) contract structure that allowed material to be procured in bulk for the entire block of MYP ships (called "economic order quantity" or "EOQ" material procurement) Improving construction performance (including reducing the construction time span) Executing a significant redesign program to reduce production labor and material costs

same at each location for each submarine, so this phase gains a full submarine's worth of experience on each hull delivered. Since final assembly and test alternates between Groton and Newport News, experience accumulated by each yard on this phase of construction is driven by every other submarine delivered. Redesigning sections of the submarine to reduce cost introduces change into the design, which is widely recognized as a source of cost growth on complex programs. Sean Stackley, Assistant Secretary of the Navy for Research, Development and Acquisition, views design and program stability as key to maintaining affordability, driving program managers to say "no" to design changes (Cavas 2009). Introducing change - whether for cost reduction purposes or capability insertion - disrupts the learning curve for the program, and moves production performance back up the curve (i.e. sets it back) to a less efficient point. This side effect can increase program costs well above the cost of making the change in isolation (Cooper 1994). Changing the design while simultaneously making other changes to the program (e.g. reducing the construction span) can further exacerbate the unintended consequences. Shortening construction time span involves changes to all three major phases of construction: · Early module manufacturing (roughly the first two years of construction) · Major module assembly and outfitting · Final assembly, outfit and test (last two years of construction) A major goal was established on the VIRGINIA Class Program to reduce the new construction time span from 84 months (the lead ship was delivered in 87.5 months) to 60 months by the time the program was to achieve the $2 billion unit cost target. Increasing the amount of work accomplished before final assembly and integration of the hull demands earlier delivery of ship components and subsystems to meet the accelerated assembly schedule. Procurement of ship components and subsystems, however, is linked to funding of the ship at the start of each

·

·

Assessing effects of potential cost reduction and other changes The Navy/design yard/shipbuilder team collaboratively identified numerous potential cost reduction opportunities. These were characterized by the first submarine affected by the change, potential savings for the first and subsequent submarines on which the change was implemented, and investment required (nonrecurring engineering for redesign and production process changes). Cost savings were broken down into material and labor hours. Given the nature of the VIRGINIA Class cost reduction target ­ i.e. achieving a specific cost by a certain ship ­ it would be critical to know the effect on ship-by-ship learning of the collection of cost reduction initiatives applied to different ships preceding the FY12 hull. Determining this likely ship-by-ship learning was complicated by factors like the shipbuilder teaming arrangement, disruption from design changes, construction time span reduction and material availability ­ all of which interact with each other. The VIRGINIA Class submarine has a unique construction teaming arrangement. Submarine modules are built in three different locations ­ GDEB Quonset Point, GDEB Groton and NG Newport News. Final assembly and test alternates between Groton and Newport News as the delivery yard for each submarine alternates. Module construction work is essentially the

r Fo

Pe er

Re

vi

ew

Naval Engineers Journal

Page 4 of 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

fiscal year. Advance procurement funding for long lead time material is typically provided the year before the ship is funded. EOQ items from the MYP provide some additional early funding to the ship, but with proportionally more benefit going to the ships at the end of the MYP. Assembling a ship with lower than expected material availability forces the shipbuilder to work around missing items, and install them out of sequence when they arrive later than planned. Installation of late components are then performed under less efficient conditions, potentially creating rework when nearby components have to be removed for access or are inadvertently damaged during the process. Of course, a shorter construction span reduces fixed costs ­ but at some point, savings from shorter construction spans may be overcome by indirect impacts from poor material availability and resulting rework.

independent "reality check" of the path down which VIRGINIA Class was headed, provided insight into key decisions before they were made, and highlighted areas needing special attention to help keep the program on track.

PROGRAM PERFORMANCE DYNAMICS AND HOW THESE DRIVE THE CLASSIC LEARNING CURVE

One important premise of program performance is that the amount of effort required to execute a given scope of work ­ for example, to build a submarine according to a specified design ­ depends on the conditions under which that work is performed. By definition, "learning" has occurred when the same work scope (represented for example by similar subsequent submarines in serial construction) is accomplished by the expenditure of fewer labor hours. But learning doesn't just happen automatically, and there are no guarantees of its occurrence or magnitude. So what really drives learning, and to what extent can it be managed?

r Fo

Pe er

Key questions

The complexity of the VIRGINIA Class program, along with the significant potential for unintended side effects which could prevent achievement of its cost target for the FY12 ships, raised several important questions: · Given the dynamic interactions described above, how would the identified cost reduction initiatives in combination with other changed program conditions likely affect construction learning, and thus the net labor cost achieved by ship? What are some programmatic considerations in choosing and implementing different changes, and combinations of changes? What were the key risks to manage in order to achieve "2 for 4 in 2012"?

Re

Underlying constructs related to learning

Two key constructs relate to the accomplishment of work and the effort that's needed to accomplish it. The first of these is shown in Figure 2, which portrays how work is typically accomplished (Cooper 1993).

vi

ew

·

·

As part of the overall Navy/industry team effort, design and construction performance of the VIRGINIA Class program was dynamically simulated to rigorously and quantitatively capture each of these issues, and to perform advance "what if" analysis of specific potential program actions. This analysis provided an

FIGURE 2. The "Rework Cycle" representation of program work accomplishment captures behavior observed on complex programs

Page 5 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

In this representation, at the start of a program or program phase all work resides in the upper left backlog of work to be done. As the program or work phase begins and progresses, changing numbers of people working at some productivity together determine the pace of progress. Productivity here is defined as the output per unit effort (for example, tons of steel welded per labor hour). Work is executed at varying (and usually less than perfect) levels of "quality" which is defined here as the correctness and completeness (for any reason whatsoever) of the work being performed at this point in time. This fractional value of quality determines the amount of the work being done that will enter the backlog of work really done, which will not need revisiting. The rest will subsequently need some rework, but for a (sometimes substantial) period of time it remains in a backlog of undiscovered rework - work that contains as yet unknown errors, and is therefore perceived as being done. Errors or omissions are often discovered by downstream efforts or testing, so this rework discovery may occur months or even years later ­ during which time subsequent work has incorporated these errors or technical derivations thereof. Once discovered, known rework becomes part of the total scope remaining and requires application of people beyond those needed to complete the original work. Executed rework enters the flow of work being done and is subject to similar productivity and quality variations as original work. Some reworked items may flow through this "rework cycle" one or more subsequent times.

following (note that even under the best possible but realistic program conditions most of these factors will tend to have some impact). · Design and planning maturity/quality ­ design and associated detailed planning (work package) information which is available when needed at a high degree of maturity or quality generally enables higher construction productivity and work quality than if the information is late relative to need and/or of low quality, and work proceeds anyway Material availability ­ material which is on hand when needed to support the build plan in the proper construction sequence generally enables higher construction productivity and quality than if some material is not available, which therefore requires temporary accommodations of missing material along with completion of those areas of work in a separate subsequent pass under less optimal conditions Workforce experience ­ a workforce with a high average level of experience or skill (defined relative to the type of work involved, with more complex work demanding higher levels of experience) generally enables higher construction productivity and quality than a workforce which is inexperienced or lacks familiarity with the specific work at hand

Drivers of productivity and quality and resulting dynamics

Both productivity and quality may vary substantially over the course of a program, depending on the conditions under which the work is done. Learning occurs when productivity increases (the same tasks are accomplished with less effort) and/or work quality increases (rework needed is reduced so that achieving the desired end result requires less iteration and therefore effort). Common drivers of variations over time in construction productivity and quality include the ·

r Fo

·

Pe er

· ·

Re

vi

Prior construction work availability and quality ­ prior or "upstream" work which is available when needed to support planned downstream work and is of high quality generally enables higher construction productivity and quality on that downstream work than if the upstream work is late and/or deficient Out of sequence work ­ work which is performed in the originally planned

ew

Naval Engineers Journal

Page 6 of 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

sequence generally enables higher construction productivity and quality than work which is performed out of sequence due to necessary workarounds and ad hoc modification to planned processes · Sustained overtime ­ a workforce operating on a "normal" schedule generally enables higher construction productivity and quality than a fatigued workforce which has worked substantial amounts of overtime for a sustained period of time Schedule pressure ­ working under "normal" levels of concern for the schedule generally enables higher work quality to be achieved, though more time may be taken to accomplish it and therefore productivity may be lower. Conversely working under high degrees of schedule pressure may increase productivity but decrease work quality since "haste makes waste"

Design changes

·

Workforce morale ­ a workforce with high morale (for example caused by good working conditions, leadership and/or clear feasible and consistent objectives) generally enables higher productivity and work quality than one whose morale is impacted by adverse conditions and expectations

·

Figure 3 illustrates at a high level how these influences on productivity and work quality contribute to key program dynamics, which in turn determine performance over time in a given phase of program work. For example (at the right hand side of the figure), current perceived progress and the hours expended to accomplish this progress together determine the expected hours at completion. Expected hours at completion (and the expected hours remaining to work, not shown explicitly here) along with the scheduled completion time drive the number of staff requested at any point in time. To the extent that staff are available, they are applied to program work, and this in part determines the rate of further progress. If not enough staff are

r Fo

Workforce morale Schedule pressure

Pe

Design and planning maturity/quality

er

Work To Be Done

Re

Workforce experience

Hiring Attrition

vi

Sustained overtime

Staff

ew

Scheduled completion time

Quality Productivity

Staff on project

Staffing r equested

Out-of- sequence work

Progress

Hours expended to date

Work Really Done

Expected completion Material time availability

Expected hours at completion

Prior construction work availability and quality

Known Rework Rework Discovery

Undisc. Rework

Material funding timing

Perceived progress

FIGURE 3. Construction productivity and quality vary with program conditions and resulting work accomplishment drives subsequent program condition

Page 7 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

immediately available, some of the shortage may be made up temporarily via use of overtime. This increases the number of effective staff on the program, but (as described above) may also impact productivity and work quality. Alternately (and especially in the longer term), any shortage of staff may drive additional hiring, either externally or internally from other programs. Hiring also increases the effective number of staff working on the program, but may also cause changes to average experience levels if the new hires are of different average experience levels than the existing workforce. Through dynamics like these, progress resulting from the applied resources and the productivity and quality at which they work feed back to determine other program conditions which in turn influence the productivity and quality of downstream work.

require less effort to build than the previous ones.

Potential threats to learning

But in truth learning is not guaranteed, and can even be undone to some degree by change and related disruption on the program. The achieved rate of improvement in productivity and quality i.e. the magnitude of learning - depends heavily on the stability of the design and other factors described above, as well as the culture of continuous improvement in the organization, contractor incentives, supply chain, the construction processes and business systems used to manage the work. Realistically, some change is inevitable and in moderation is even desirable ­ for example design changes to add new capability to a stable design or to accelerate cost reduction, changes to new suppliers with higher performance or more robust commercial status, changes to remove specific bottlenecks in current construction processes etc. ­ so managing effectively is not a matter of simply avoiding change at all cost. But keeping a mature program on the right track does mean balancing the nature and degree of changes (and their potentially disruptive effects) against their likely benefits. Achieving the right balance with respect to change was critical for VIRGINIA Class, where past construction performance was considered excellent - but at the same time it seemed clear that further "organic" learning (without specific cost reduction actions) would be insufficient to achieve the explicit target of "2 for 4 in 2012." The following section describes the approach taken on VIRGINIA Class to provide a rigorous and independent view of the likely outcome of proposed design and construction process changes, and how these would interact with other changed program conditions including funding profiles, production rates and construction schedules.

How traditional learning curves arise

Many of these factors influencing construction productivity and quality tend naturally to improve over the course of a serial construction effort. For example, problems with the initial Class design for a new ship tend to be identified and resolved in the course of building the lead or first few ships ­ so design maturity is higher for construction of subsequent ships. Similarly, potential technical, commercial and logistical issues with subcontractors supplying material tend to get resolved relatively early on in the program. The workforce gains useful experience (both program specific and generally related to a trade) by building early units, so that productivity is improved on subsequent units. As the construction sequence and process are fine tuned or optimized, productivity and work quality further increase. Higher productivity and quality mean faster progress (more consistent with target schedules), which may mean less sustained overtime is needed to resolve problems or make up delays ­ so productivity and quality improve further. The natural tendency for these influences on productivity and quality to improve over the course of serial construction (and of course lots of empirical evidence showing that it generally happens) drives expectations for the classic learning curve - whereby each subsequent unit is expected to

r Fo

Pe er

Re

vi

ew

Naval Engineers Journal

Page 8 of 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ANALYZING VIRGINIA CLASS PROGRAM PERFORMANCE

VIRGINIA Class Program performance was simulated using a sophisticated, quantitative System Dynamics model made up of several thousand equations, which explicitly represented cause and effect relationships and program dynamics like those described above, and was rigorously validated against data for historical program performance through the first five delivered ships. The same dynamic simulation approach and toolset (including customizable model structures, numerical parameters etc.) has been successfully applied to over 100 complex design-build programs since the late 1970's. Many of these have been in US naval shipbuilding (e.g. LHA, LHD, CG 47, DDG 51, SSN 688, CVN 65, DD(X)) or in aerospace (e.g. F/A-18 E/F mod, B-2 mod, A-12, C-17, AC130U, F-35 Joint Strike Fighter), as well as many other types of government and commercial development programs. Applications have occurred across the entire life

cycle of programs, from early program costing and bid support or evaluation, through retrospective diagnosis of program performance to support dispute resolution or lessons learned.

VIRGINIA Class Program simulation model structure

The basic structure of the VIRGINIA Class Program simulation model is shown in Figure 4. Design activity (shown at the left) is represented separately for the Class design and groups of subsequent changes which are first applicable to the submarines indicated by arrows. Procurement of material depends on design status, and is simulated separately for each ship (lower left). Direct construction activity (which depends on both the design and material procurement activities) is simulated separately for five phases for each of the eighteen ships in Blocks 1 through 3 ­ including module construction at EB Quonset Point, module construction at EB Groton, module construction at Newport News, final assembly and test at

SSN 791 SSN 790 SSN 789 SSN 788 SSN 787 SSN 786 SSN 785 SSN 784 SSN 783 SSN 782 SSN 781 SSN 780 SSN 779 SSN 778 SSN 777 SSN 776 SSN 775 SSN 774

SSN 778 Block 3 Mods SSN 776 Virginia Class Class

r Fo

SSN 786 SSN 785 SSN 784 SSN 782 SSN 780

Pe er

Re vi ew

Design Design

Final Assembly and Test (Lead Yard)

EB Modules/Construction (Groton) Support Material

NN Modules/Construction (Newport News)

Key:

Block 1 Block 2 Block 3

Support Support

FIGURE 4. VIRGINIA Class Program simulation model structure represents design, material fabrication and delivery, direct construction effort, and support activities (for Blocks 1-3)

Post Delivery/PSA

EB Modules/Construction (Quonset Point)

Page 9 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

whichever shipyard delivers that ship, and postdelivery/PSA activity at the appropriate location. Support effort (which is roughly the same magnitude as the direct hours) is also simulated separately for each ship. Each of the design and direct construction phases simulated incorporates the structures described in the previous section to represent the accomplishment of work and rework, drivers of and variations in productivity and quality, and related program conditions and management decision-making. Effects analogous to those described above for construction determine the productivity and quality of the design work over time. Somewhat simpler structures simulate material procurement activities over time, as well as the need for and application of support labor in construction (driven by direct labor and fixed/facility considerations). The application of labor and accomplishment of work is by four separate resource groups ­ EB design engineering, EB construction (Quonset Point), EB construction (Groton) and Newport News construction.

determined via structured group interviews ­ was also reflected in the calibration of the model. As a result of these efforts, the VIRGINIA Class Program simulation model closely recreates what is known to have happened on the program to date, and for the right reasons. Figure 5 provides representative examples (with vertical scales removed) of the correspondence of simulated variables to actual historical data, for engineering and construction at various levels of aggregation. In each case the solid lines are the model simulation, and the dashed red line is the actual data used for calibration and validation (but not to drive the simulation). At the time this work was performed the actual data extended through early calendar year 2007.

Simulation model development and validation

Key assumptions used as inputs to the VIRGINIA Class Program simulation model ­ for example, target schedules, work scopes and shares between shipbuilders, budgets, technical interdependencies between phases, etc. ­ were carefully determined from available data or estimated, and then reviewed with knowledgeable Navy and/or shipbuilder staff. Behavior of the program simulation model was also extensively numerically calibrated and validated against a wide range of actual timephased historical data, for metrics like drawings and revisions accomplished, labor hours expended (engineering and construction direct and support) and construction progress achieved. Finally, more qualitative information about the history of the program ­ including "recollections" by construction managers regarding variations in the influences on construction productivity and quality,

r Fo

Reference simulation of future VIRGINIA Class performance

Besides simulating history, the VIRGINIA Class Program model simulates performance on the remainder of the program through the end of Block 3. The reference (or baseline) program simulation defined likely construction labor hours expended on each ship before any cost reduction or other program changes (see Figure 6). Traditional learning curves applied on a total hour, non parity adjusted basis (representing 91% learning in solid green, and 89% learning in dashed red) are included here for illustration only. Figure 6 illustrates departures from smooth, theoretical learning curves in the simulated construction labor hours by individual ship. These variations were driven by the dynamic effects on productivity and work quality, reflecting unique conditions simulated for specific ships or groups of ships arising from: · Restarting submarine construction at NG Newport News after a long hiatus (involving the rebuilding of experience levels, which affected productivity and work quality)

Pe er

Re

vi

ew

Naval Engineers Journal

Page 10 of 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Design Drawing Revisions

Simulated

20,000 15,000 10,000 1,000 5,000 0 1994 500 1998 2002 2006 2010 2014 2018 2022 0 1994

Data

Design Staffing

Simulated

2,500 2,000 1,500

Data

1998

2002

2006

2010

2014

2018

2022

Total Construction Staffing

Simulated

8,000 6,000 4,000 2,000 0 1994

Data

Block 1-3 Total Eq Heads

Block 1 Simulated Block 2 Simulated Block 3 Simulated

8,000 6,000 4,000 2,000 0 1994

Block 1 Data Block 2 Data

r Fo Pe

2018 2022 2006 2010 2014 2006 2010 2014 2018 2022

2006 2010 2014 2018 2022

1998

2002

1998

2002

2006

2010

2014

2018

2022

er

Total EB Staffing

Simulated

4,000 3,000 2,000 1,000 0 1994

Data

Total EB Direct Staffing

Simulated

4,000

Data

Re

3,000 2,000 1,000 0 1994

vi ew

1998

2002

1998

2002

2006

2010

2014

2018

2022

Total Staffing for SSN774

Simulated

2,000 1,500 1,000 500 0 1994

Data

Total NNS Build Staffing for SSN774

Simulated

2,000 1,500 1,000 500 0 1994

Data

1998

2002

1998

2002

2006

2010

2014

2018

2022

FIGURE 5. Simulation plots illustrate the correspondence between simulated variables (solid lines) and the actual historical data (red dashed lines)

Page 11 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

18 16 14 Labor Hours (millions) 12 10 8 6 4 2 0 774 775 776

FIGURE 6. VIRGINIA Class Program model simulates drivers and resulting performance for each ship including construction labor hours expended

·

Applying most design changes to the lead ship in a block (so impacts from design maturity on productivity and work quality were worst at the beginning of the block)

·

Effectively receiving material and construction funding earlier for the second ship in a fiscal year than the first, since all of the funding for a given year becomes available at the same point in the year (and relatively earlier material therefore benefits productivity and work quality on the second ship)

This reference simulation of likely program performance provided a starting point for analysis of the likely full consequences of cost reduction initiatives and sensitivity to other program conditions and changes.

Determining the full impacts of cost reduction initiatives

One of the primary uses of the VIRGINIA Class Program simulation model was to determine the likely full-up impact of potential cost reduction initiatives, net of any indirect effects (including

r Fo

777 778 779

780 781 782 783 784 785 786 Reference Learning @ 91% Learning @ 89%

787

788

789

790

791

Pe

either beneficial synergy or adverse disruption) these might have on the program. Each of the potential cost reduction initiatives were characterized by estimates for: · · · · · Engineering investment (and degree of change to the design) involved First ship on which the action was to be implemented Nominal labor hour savings on first and subsequent ships Phase of work and organization (e.g. EB, NN, suppliers) primarily affected Non-labor (e.g. contractor-furnished material or government-furnished equipment) dollar savings

er

Re

These estimates were translated into specific model inputs for each initiative, which defined changes to the design, construction work scopes, budgets and design/construction timing as appropriate. Starting from the reference simulation, the full and cumulative impact of these initiatives was simulated by adding them one by one to the simulation, with order determined chronologically and by any logical precedence of different categories of initiatives.

vi

ew

Naval Engineers Journal

Page 12 of 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The resulting impact on forecast labor hours by ship for all cost reduction initiatives combined is shown in Figure 7, where the solid blue line ("Reference") indicates labor hours by ship before the cost reduction initiatives, and the dashed pink line ("Should Cost") the cumulative effect of implementing all the likely cost reduction initiatives. The marginal impact on construction hours expended over time on each ship for each initiative could then be calculated by comparing the simulation into which that initiative was first added with the preceding simulation in the sequence, which excluded it. Other program conditions assumed - e.g. funding profiles, targeted construction schedules etc. - were those considered most likely (keeping in mind that there was not yet a firm contractual plan for Block 3). The analysis showed that some specific initiatives caused little or no indirect impact or disruption, and so it was likely that essentially all of the estimated labor hour savings for that

16 15 14 Labor Hours (Millions) 13 12 11 10 9 8

initiative would be realized. This was generally true for relatively straightforward changes to construction processes where the end product was largely unchanged, as well as for design changes which were modest in extent. Other initiatives ­ especially those which involved more extensive design changes, or which were executed in the midst of other changes ­ caused much more indirect impact. For these initiatives it was likely there would be significant disruption that would attenuate the savings realized, especially for the first ships on which they were implemented. Figure 8 shows the proportion of estimated savings likely to be achieved for different categories of cost reduction initiatives, where the bottom blue portion of each bar indicates simulated savings realized and the top purple portion any nominal savings not achieved due to indirect impacts or disruption. Note that this figure sums savings across all ships affected by those initiatives ­ in many cases disruption on the first ships affected was much higher than on later ships.

774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791

FIGURE 7. Total VIRGINIA Class labor hours by ship showing marginal net savings arising from implementation of all the cost reduction initiatives

r Fo

Pe er

Reference

Re vi

Should Cost

ew

Page 13 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Labor Hours (Millions)

3.5 3.0 2.5 2.0 1.5

Savings Achieved

Target Savings Not Achieved

r Fo Pe

Other Block II Block II ManTech Other In Block II Approved and III and III Approved Process and III PMS450 DFA PMS450 Learning Approved Learning Approved In Process Block II Other Legacy Current Capability and III Near LEAN LEAN Reduction DFA In Term DFA Initiatives Initiatives (Backup) Process In Process

1.0 0.5 0.0

FIGURE 8. Indirect impacts (disruption) associated with some categories of initiatives attenuate their realized savings, while other categories realize full nominal savings

Throughout this sequence of program simulations there were also dynamic interactions between the individual initiatives being added. The sequence described above associates any indirect impacts from these interactions with the marginal impact of the new initiative being added, since its full impact is simulated assuming the presence of all preceding initiatives.

Sensitivity of cost savings and learning to other program conditions

Analysis was also performed to determine how sensitive the likely realization of labor hour savings from cost reduction initiatives was to a number of other program conditions, including: · · Timing of long lead material funding Timing of engineering investment to technically define changes

er

Re

· · ·

Overall construction schedules (construction spans) Construction workforce attrition levels Potential errors in investment and savings estimates

Figure 9 shows the sensitivity of labor hours achieved by ship with cost reduction initiatives to an increase in the FY09 advanced procurement funding for Block 3. Two potential Congressional "plus up" funding amounts were assessed ­ an additional $79 million, and twice that ($158 million). It was assumed that the funding was provided to the program up front, and was not "paid back" via a smaller subsequent FY appropriation until the final funding increment for the last ship in the block. Further sensitivity testing was performed to determine how best to apply the funds ­ for example, applying all additional funding to the unfunded amount of the next ship or splitting it up between multiple ships. The analysis

vi

ew

Naval Engineers Journal

Page 14 of 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

16 Should Cost 15 14 Labor Hours (Millions) 13 12 11 10 9 8 $79M FY09 Plus UP $158M FY09 Plus UP

774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791

FIGURE 9. Earlier funding for material improves availability to construction and accentuates cost savings/learning

showed that maximum savings were achieved by applying all extra funding to the earliest unfunded ship. Splitting this additional funding between the first two ships produced 30% less savings than applying all of the incremental funding to the first ship. The analysis also indicated that "paying back" the incremental funding before the last ship in the block substantially reduced the savings realized. Not surprisingly, the primary dynamic by which this earlier funding enhanced construction savings was improved material availability leading to fewer work-arounds in construction, resulting in higher productivity and work quality. Interestingly, the benefits of the earlier funding are not shared evenly between ships. As the earliest unfunded ship, SSN 785 receives the most benefit from the FY09 "plus up" funding, especially given that early ships in a block tend to have more material problems than later ones. The odd numbered ships from then on receive little additional benefit, because they are the second, later ships in their fiscal year and

r Fo Pe er Re

already relatively well off with respect to funding timing (since the funding for both ships in a fiscal year becomes available at the same absolute point in time).

Contribution to the VIRGINIA Class Program

Dynamic simulation of the VIRGINIA Class Program provided important feedback to the program office and the shipbuilders in two primary areas. First, it provided a separate, independent and rigorous analytical evaluation which confirmed that with the planned cost reduction initiatives ­ even fully considering likely indirect and disruptive consequences ­ the program was indeed likely to realize the labor savings necessary to reach its goal of "2 for 4 in 2012." Second, it provided a broad and flexible capability to quantitatively analyze effects of specific cost reduction initiatives and other program changes, individually and in combination. This "what if" capability to

vi

ew

Page 15 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

predict VIRGINIA Class Program performance under a wide range of assumed conditions helped identify key risks to be managed, and supported specific decisions. For example, analysis showed that although compressing construction time spans could save cost by shortening the duration over which fixed costs accrued, too much compression ­ pushing up against the limits of design maturity in the wake of changes, and material availability given limits on advance procurement funding ­ actually drives cost back up. As a result, the Cost Reduction Program was careful to properly plan the schedule reduction. Other simulations showed the impact that timing of engineering investments, timing of construction starts, and construction workforce attrition levels would have on labor cost by acting on productivity and work quality, and helped the program focus attention on high leverage areas.

program - recognizing that once underway, adverse indirect consequences of change (delay and disruption) can cascade through the program with impacts that far exceed the apparent magnitude of the change itself, or the ability of reactive steps to mitigate. Change which is not managed effectively can disrupt learning via its effect on the factors that drive productivity and work quality ­ with the result that hard won gains are lost and need to be "clawed back" with additional time and effort.

Drivers of learning ­ effects on productivity and quality

The program team should keep in mind that: · Design maturity, material availability and other factors affecting productivity and work quality which tend to improve over the course of the program are key drivers of learning Net benefits of individual changes vary substantially with the degree of disruption they cause The benefits of design and process changes also depend on other program conditions and changes The amount of cumulative program change can exceed the capacity to effectively mitigate disruption

One key element of the success of the VIRGINIA Class Program dynamic simulation effort was the integrated approach and information sharing between the Navy, the shipbuilders GDEB and NGNN, and PA Consulting Group. The program simulation model was built and validated by PA with significant input from both Navy and shipbuilder stakeholders at critical steps along the way. All analysis results were briefed to both the Navy and the shipbuilders, which helped stimulate and advance discussions about top priorities for the cost reduction and other program changes, and the best path by which to reach the common goal of cost reduction.

CONCLUSIONS - MANAGING CHANGE AND LEARNING

Change in the course of complex programs is inevitable - whether through the realization of risks, the need to replace obsolete components or suppliers, or in response to evolving requirements such as new cost or performance targets. How change is managed is a critical determinant of success or failure for most complex programs. Change needs to be proactively managed in the context of the entire

r Fo

Pe er

·

·

Re

·

Executing a comprehensive re-design effort intended to reduce construction labor and material costs carries with it the risk of higher cost associated with reduced design maturity and other disruptive effects. Implementing multiple changes results in interactions between individual initiatives, and the total effect is not just the sum of the parts. This is especially true when other aspects of the program are changing at the same time ­ and the timing of changes is always a key consideration. In general, it will be helpful to think through in advance the likely impact of change on the drivers of productivity and work quality, and how these might best be managed or mitigated.

vi

ew

Naval Engineers Journal

Page 16 of 17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Understanding likely consequences in advance

A thorough understanding of the likely consequences of change in advance of making decisions provides the opportunity to identify effective actions to manage or mitigate undesirable impacts on program performance and learning. On the VIRGINIA Class Submarine Program this understanding was provided by dynamic simulation of the program to determine likely outcomes with cost reduction initiatives and other program changes implemented. This rigorous analysis helped the program understand better likely change impacts, how they interacted, and how to mitigate the risks associated with these while moving forward toward "2 for 4 in 2012."

Trost, Christopher S. "Moving Towards the Next Milestone of Submarine Design," Naval Engineers Journal, March 2000.

REFERENCES

Cavas, Christopher P. "Carrier Numbers, Budget Issues Await Obama," Navy Times, January 12, 2009. Cooper, Kenneth G. "The Rework Cycle: Why Projects Are Mismanaged," Project Management Institute PM Network, February 1993.

Rear Adm. David C. Johnson is the Commander, Naval Undersea Warfare Center, Deputy Commander for Undersea Technology (SEA 073) and Deputy PEO Submarines, OHIO Replacement Program (PEO SUB-OR). He was the Program Manager for the VIRGINIA Class Attack Submarine Program (PMS 450) from 2005 to 2008. Rear Adm. Johnson's previous shore assignments include Assistant Program Manager for USS JIMMY CARTER (SSN 23) and VIRGINIA Class Program Manager's Representative at SUPSHIP Groton. Rear Adm. Johnson, a native of Pensacola, Fla., is a 1982 graduate of the United States Naval Academy. Rear Adm. Johnson also holds a Naval Engineer degree and a Master of Science in Mechanical Engineering from the Massachusetts Institute of Technology (1989). George M. Drakeley is the Special Assistant for Acquisition in the VIRGINIA Class Program Office (PMS 450). He began his naval career as a Nuclear Qualified Officer aboard USS ARKANSAS (CGN 41) and the RADCON Officer on USS ACADIA (AD 42). Upon leaving active duty service, Mr. Drakeley joined the OHIO Class Program Office at the Naval Sea Systems Command. Following this assignment, Mr. Drakeley went on to serve as the VIRGINIA Class Test and Evaluation Manager and the Deputy Program Manager for the VIRGINIA Class Program. Mr. Drakeley received a Bachelor of Science in Nuclear Fusion Engineering from the Massachusetts Institute of Technology in 1978. He also holds a Master of Physics From George Mason University.

Cooper, Kenneth G. "THE $2000 HOUR : How Managers Influence Project Performance Through the Rework Cycle," Project Management Journal, March 1994. Hilarides, William. "2 For 4 in 2012 - The Virginia Class Road Ahead." Proceedings, June 2006. Johnson, David C. CAPT, G. Drakeley, G. Smith. "Engineering the Solution: VIRGINIA Class Submarine Cost Reduction," ASNE Proceedings: Engineering the Total Ship (ETS) Symposium 2008, September 2008. O'Rourke, Ronald. "Navy Major Shipbuilding Programs and Shipbuilders: Issues and Options for Congress." CRS Report # 96-785F, September 24, 1996.

r Fo

Pe er

Re

Thomas N. Plante is the VIRGINIA Class Program Manager for Technology Insertion at General Dynamics Electric Boat Corporation in Groton, CT. He served in the Navy as a junior officer aboard USS Boston (SSN703) from 19811984, and joined Electric Boat in September 1984 where he has served in a variety of positions with increasing responsibility. Since 2006, he has worked with the Navy to develop

vi

ew

Page 17 of 17

Naval Engineers Journal

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the Virginia Class Cost Reduction Program aimed at reducing Virginia Class unit cost through redesign and process improvement. He also led proposal development for the third block of eight Virginia Class Submarines which began construction in 2009. Mr. Plante graduated from Rensselaer Polytechnic Institute in Troy New York in 1980 with a Bachelor of Science degree in Mechanical Engineering, and also received a Masters of Science degree in Computer Science from Rensselaer Polytechnic Institute Hartford Graduate Center in 1990. William J. Dalton is a Managing Consultant in PA Consulting Group's Federal and Defense Services Practice in Cambridge, MA. He joined PA in 1985, and in addition to simulation of VIRGINIA Class Submarine Program performance has lead numerous assignments on large, complex programs in aerospace, automotive, civil construction, electric power and shipbuilding (naval and commercial). Prior to joining PA, he worked as an engineer and project manager for Offshore Devices Inc. He received a Bachelor of Science in Ocean Engineering (1980) and a Master of Science in Ocean Systems Management (1985) from the Massachusetts Institute of Technology. Christopher S. Trost, P.E. is a Principal Consultant in PA Consulting Group's Federal and Defense Services Practice in Cambridge, MA and has been an active member of ASNE since 1997. He has used simulations to provide strategic assistance to the VIRGINIA Class Submarine Program as well as US Navy submarine maintenance planning. Prior to joining PA, he worked for Raytheon as a Systems Engineer on the DD(X) program and was Program Manager for the Littoral Combat Ship (LCS) program. As a military officer in the U.S. Navy, he was a Submarine Officer and Engineering Duty Officer supporting ship design, procurement, maintenance and repair. He was awarded a Naval Engineer and Master in Ocean Engineering from MIT, an MBA from Old Dominion University and a B.S. in Nuclear Engineering from Penn State.

r Fo

Pe er Re vi ew

Information

18 pages

Find more like this

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

376578


You might also be interested in

BETA
Navy Littoral Combat Ship (LCS): Background and Issues for Congress
Pg1_Pamphlet_V2_CMYK.cdr