planning and executing fast-track projects › scv-tems › files › 2016 › 12 › tems... ·...

Planning and Executing Fast-Track Projects

Dr. Raymond LevittProfessor of Civil & Environmental EngineeringAcademic Director, Advanced Project Management ProgramStanford University

Presentation to IEEE SCV Engineering Management Society October 29, 2003

2

Planning and Executing Fast Track Projects

Converting Strategy into Action

Outline

Fast Track Project Case Study: The Lockheed Martin Launch VehicleTechniques for Planning Fast Track Projects

PERT SimulationThe Critical ChainSystems Dynamics SimulationThe VDT/SimVision Project Design Approach

AppendixTrajectory of Ongoing Project Design Research

3



Fast-Track Project Case Study: Lockheed Martin Launch Vehicle

Goal: Shrink time-to-market for LMLV by 80% vs. Trident missile!Highly Concurrent: many interdependent activities must be scheduled concurrentlyKey components will be outsourced to minimize cost

4



Organization of Avionics PDT

P ow er D ist.P anel

S ub Team (3)

F iring U n itS ubT eam (3)

O peratio nalIn terlocks

S ub Team (4)

F ligh t B o xes S u b Team

P rogramM anager

S E & IP ro ject

M anager

P ropu ls ionP ro ject

M anag er

N G & CP ro ject

M anager

Facilities& S E

E ngineeringM anager

B usin essD evelopm ent

M anager

O perationsM anager

TestE ng ineerin g

M anager

S tructu resP ro ject

M anager

T otal: (15)

O ff-shelf D epartm ent

C ab les C o ntractor

P ackag ing D epartm ent

E lectron ics D epartm en t

F ligh t B oxesD epartm ent

P rocesso r fo rP ackage

S ub Team (5 )

C ab lesS ubT eam

(3)

O ff-sh elfS ub Team

(5)

P ackag ingS u bTeam

(1.5)

E lectron icsP arts

S ubTeam (6)

A vion icsP ro ject

M anager

Fu nctio nal G u idance

P ro ject O vers igh t- M on ito ring-E xception H and ling

Fu nctio nal G u idance

P ro ject O vers igh t- M on ito ring-E xception H and ling

S T S ubT eamP D T P roduct D evelopm ent T eamS E & I S yste m s E ng ineering and In teg ra tionN G & C N av iga tion , G u idance and C ontro lS E S upport E qu ipm ent

(Actor N am e) A cto r N o t R epresen tedin A vio n ics M od el

(Actor N am e)(# FTE s) A ctor R ep resen ted

in A vion ics M o del

5



Activity Workflow for Avionics PDT

P re d ec es so r -S u cc es so rre la tio n sh ip

A c tiv ity

(T o ta l H o u rs )

A c tiv ity

(T o ta l H o u rs )

(s ta rt)m ile s to n e

m ile s to ne(fin is h )V e h ic le

A v io n ic sC o n c e p t

S ys te mIn te g ra tio n a n d

T e s t

Id e n tify P a rtsR e q u ire d

S e a rc h fo rV e n d o rs

P ro c u re m e n tS u p p o rt

P re p a reD o c u m e n ta tio n

P rin te d W irin gB o a rd D e s ig n

P rin te d W irin gA s s e m b ly

E n c lo s u reD e s ig n T o p A s s e m b ly

P ro c u re m e n t

D e ve lo p in gS u b -c o n tra c ts

T e a m in gA g re e m e n ts

R a n g eR e q u ire m e n ts

V e h ic leIn te rc o n n e c t

L a yo u t

D e fin eIn te rfa c e s

D e ta ile dC a b le D rw .

F a b ric a te a n dT e s t C a b le s

D e fin eR e q u ire m e n ts

N ewE n g in e e rin g

(C a b le s )

R e e n g in e e rin gE x p e rie n c e s

P ro d u c tio nE n h a n c e m e n ts

B u ild a n d T e s tF lig h t U n its

B re a d B o a rdA n d P h ys ic a l

M o c k u p

A p p ly in gE x is tin g

A p p lic a tio n s

O ffs h e lf S u b T ea m (S T ) (5 )

C a b les S T (3 )

F lig h t B o x es S T (15 )

E le c tro n ic P arts S T (6 )

P a cka g in g S T (1 .5 )

P ro jec t M a n a g er (1 )

C a b le s S u b Tea m (3 F T E )

F lig h t B o x e s S T (1 5 F T E )

E le c tro n ic P a rts S u b T e a m (6 F TE )

P a ck a g in g S u b T ea m (1 .5 F TE )

P h ys ic a lM o c k u p

A v io n ic sD raw in g s

O ffs h e lf S u b Te a m (5 F T E )

6



Case Study Results :Lockheed Martin Launch Vehicle

LMLV1 launched in mid-April 1996 – almost 4 months later than plannedLaunch vehicle “departed controlled flight” and had to bedetonated by AF safety officerAnalysis of telemetry data indicated most likely cause of failure to be a misrouted cablethat shorted out!

7



Analysis Tools Can Enable “Project Design”Conceptual

DesignDetailed Design and Execution

Closeout &Learning

Level of InfluenceLevel of Influence

Expenditure of Funds Expenditure of Funds

OutcomeKnowledgeAnalysis ToolsOutcome

Predictions

100%

0%

8



Fast Track Projects Are Information-Intensive

High performance,complex product has

high level of inter-dependency between

its subsystems

Fast-track schedule triggers unplanned coordination and rework for project

organization

Process OrganizationProduct

Project team must process large amount of information under

extremely tight time constraints

9



The Challenge of Fast-Track Projects:“Concurrent Engineering” Incurs Large Overheads

CPM view of traditional project work – Sequential ActivitiesCPM View of Fast-Track Project work—Overlapped Activities

Reality of fast-track project work!

Increased CoordinationIncreased Rework

10



Outline




11



PERT SimulationAssume each activity has a variable duration that is described by a probability distribution (Gamma) as follows:

Perform a large number of PERT simulations (~1,000)Independently sample each activity’s durationPerform a standard CPM analysis, using the sampled duration for each activityUse multiple (~1,000) CPM analyses to compute probability distributions of project duration and activity criticality

moml

mp Project Duration

12



Nov95 Dec95 Jan96 Feb96 Mar96 Apr96

100%

0%

Probability ofCompletion 50%

What Would PERT Simulation Have Told Lockheed Managers?

“There is uncertainty in project completion time”

“Some near-critical activities may become critical”Fast-track projects usually have multiple near-critical pathsA “criticality index” is computed for each activity, equal to the % of simulation trials in which it was critical

13



& of PERT Simulation for Fast-Track ProjectsShows how uncertainty in task durations affects uncertainty of project completion date

A straightforward extension of CPM approach and tools

Assumes that activity durations vary independently Does not model fundamental causes of variation in activity durations (e.g., poor designs, key skill deficits, bad weather, …)Does not reflect that fact that positive or negative risk factors (“knights and villains”) will impact multiple activitiesGives managers no guidance about where/how to intervene

Assumes no effects of executing activities in parallel vs. in sequence

Provides no insights about the hidden cost of more aggressive fast-tracking (concurrent task scheduling)

14



Outline




15



Critical Chain* Concepts—1Remove hidden safety from task estimates

Eliminate multitasking

* Sources: Eliyahu Goldratt, “The Critical Chain,” and Scitor Corporation Web Site)

16



Critical Chain* Concepts—2Plan backward from required completion date Calculate the Critical Chain (the resource constrained critical path)

* Source: Eliyahu Goldratt, Scitor Corporation Web Site)

17



Critical Chain* Concepts—3Insert Project Buffer at end of critical chain; and insert Feeding Buffer at end of all non-critical chains

Track consumption of buffers during project

* Source: Eliyahu Goldratt, Scitor Corporation Web Site)

18



What Could Critical Chain Analysis Have Told Lockheed Managers?

During the Planning Stage“Start some project activities earlier!” Earlier start time may not have been feasible.

During the Execution Stage“ Cable team project and feeding buffers are being consumed by activity overruns!” Analysis could have alerted managers earlier in the project to bring in extra cable resources

19



& of Critical Chain for Fast-track ProjectsHighlights latest startsShows impact of eliminating multi-taskingTracks impact of activity delays on buffersDoes not relate size of buffer in Feeder Chain or Critical Chain to degree of complexity or interdependence of activities in that chain Does not predict relative schedule risks of particular activities or chains in advance—vs. “task criticality” in PERT Simulation

20



Outline




21



System Dynamics ApproachesModel projects as “stocks and flows” of work, resources, information, motivation, etc.Express relationships between variables as arbitrarily simple or complex finite difference equations

* Source: “The Rework Cycle: Why Projects Are Mismanaged” by Kenneth Cooper, PMNet, 1993

22



What Would a System Dynamics Model Have Told Lockheed Managers

Showed impacts of positive and negative feedback loops on performanceShow impacts of delayed feedback loops—(oscillation)

Could provide insights about overall schedule risks due to fast-tracking this projectUnlikely to have identified specific problems in this case

* Source: “The Rework Cycle: Why Projects Are Mismanaged” by Kenneth Cooper, PMNet

23



& of System Dynamics for Fast-Track Projects

System Dynamics is a broadly applicable simulation language

SD has been applied to problems as diverse as business supply chains (e.g., “The Beer Game”) and natural ecosystems (e.g., sustainability of fisheries, forests, …)

System Dynamics is a broadly applicable simulation language

No built-in objects or behaviors to model projects in detail

Insights it can provide about projects tend to be generic and high level (e.g., rework example)

“Stocks & flows” architecture is ideally suited for modeling flows of goods & info in ecosystems or supply chains

24



Outline




25



Project Design ApproachModel planned fast-track work process and proposed organization realisticallySimulate organization executing work process to predict schedule/quality risksCompare predicted performance vs.plan, and “intervene” to mitigate risksIterate to find “optimal” project design

VDT Project Design Case Study: Lockheed Martin Launch Vehicle

26



LMLV Project Avionics Team: VDT/SimVision Model

EngineeringManager

0.515

15

15

15

1515

5

5

5

5

5

3

3

32

2

1

11

1

1

1

11

1

0.4

5

Start

Finish

Vehicle AvionicsConcept Re-engineering

ExperiencesApply ExistingApplications

Bread Board andPhysical Mockup

Avionics Drawings

ProductionEnhancements

Build and TestFlight Units

DefineRequirements

DevelopingSubcontracts

TeamingAgreements

RangeRequirements

Physical Mock up

Procurement

New Engineering

Vehicle InterconnectLayout

Define Interfaces

Detailed CableDrawings

Fabricate andTest Cables

Identify PartsRequired

Search forVendors

PrepareDocumentation

ProcurementSupport

Printed WiringBoard Design

Printed WiringAssembly

Enclosure design

Top Assembly

System Integration& Test

ProjectManagement

Avionics PM

Packaging ST Flight Box ST ElectronicParts ST Cables ST Off-Shelf ST

27



LMLV Project Avionics Team: Gantt Chart

1995 1996

28



LMLV Project: Actor Backlogs

29



LMLV Project: Process Quality Risks

Project Design Guides Managerial Interventions

“What-if Analysis” of LMLV Avionics Team

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%

Increase Cable Subteam staffing

from 3-5 engineers

Replace 3 Cable Subteammembers with 3 more

experienced engineers

Cost

Duration

Except's

Bet

ter

Wor

se

31



Lockheed Martin Launch Vehicle: Project Design Results

Simulated organization executing work process to predict schedule and quality risks

VDT/SimVision predicted launch date delay to within a few days, one year ahead!VDT/SimVision identified cable team quality risk that ultimately caused LMLV to fail!

Predicted performance impact of two potential managerial interventions(although these results were not used)

32



“Exception”

(Jay Galbraith, 1974)

But Project Participants also Coordinate. And they Generate & Handle Exceptions

+

Project ParticipantsPerform Assigned Tasks

33



Info. Volume is Derived from Project Tasks:Direct Work, Communications, Rework

Start

Ready toExcavate

ChooseConstruction

Methods

Long LeadPurchasing

Apply Exc Permit

Seek ZoningVariance

Provide GMP

SelectSubconsultants

Arch Program

Select Key Subs

Estimate Time

Define Scope

ProjectCoordination

Estimate Cost

Choose facadematerials

Choose Struct.System

GMPAccepted

DesignCoordination

Start

Ready toExcavate

ChooseConstruction

Methods

Long LeadPurchasing

Apply Exc Permit

Seek ZoningVariance

Provide GMP


Arch Program

Select Key Subs

Estimate Time

Define Scope

ProjectCoordination

Estimate Cost



GMPAccepted

DesignCoordination

Start

Ready toExcavate

ChooseConstruction

Methods

Long LeadPurchasing

Apply Exc Permit

Seek ZoningVariance

Provide GMP


Arch Program

Select Key Subs

Estimate Time

Define Scope

ProjectCoordination

Estimate Cost



GMPAccepted

DesignCoordination

34



Team Information Processing Capacity is Derived From:# Actors; Skill Set; Experience; Structure; Policies

Client PM

Design PM Construction PM

ArchitecturalDesign Subteam

StructuralDesign Subteam

ProjectEngineers

ProcurementTeam

Client PM




ProjectEngineers

ProcurementTeam

35



VDT/SimVision Information Processing Model:Team IP Capacity >= Task IP Demand?

Client PM




ProjectEngineers

ProcurementTeam

Start

Ready toExcavate

ChooseConstruction

Methods

Long LeadPurchasing

Apply Exc Permit

Seek ZoningVariance

Provide GMP


Arch Program

Select Key Subs

Estimate Time

Define Scope

ProjectCoordination

Estimate Cost


1

1

1


0.5

1

1.5

1

1.5

1

0.2

0.5

0.5

2

1GMP

Accepted

Design-Build Biotech Project

DesignCoordination

1

ProjectCoordination

Meeting

1

1 1

1

36


Converting Strategy into ActionPredictions from VDT/SimVision Project Design Approach

Backlog

Quality

Schedule

Cost

Start Fab Test& Deliver

DesignCoordination

DevelopSpecification

Insert Scan

Partition Chip

Gen Test Suite

PlaceRoute

FullChipSynth

Verify RTL

FloorPlanning

Write B1RTL

Sim GatesAssemble RTL

PhysVerifn

Verify B1RTL

Generate TestVectors

Synth_B1RTL

Activities


Project Lead

Marketing Team Chip Architect TestEngineering St Foundry Lead

Logic DesignTeam 1 Foundry Test

EngineerFoundry Layout

Engineer

VerificationTeam

DesignCoordination


Insert Scan

Partition Chip

Gen Test Suite

PlaceRoute

FullChipSynth

Verify RTL

FloorPlanning

Write B1RTL


PhysVerifn

Verify B1RTL


Synth_B1RTL

Organization


Project Lead




Engineer

VerificationTeam

DesignCoordination


Insert Scan

Partition Chip

Gen Test Suite

PlaceRoute

FullChipSynth

Verify RTL

FloorPlanning

Write B1RTL


PhysVerifn

Verify B1RTL


Synth_B1RTL

Communication


Project Lead




Engineer

VerificationTeam

DesignCoordination


Insert Scan

Partition Chip

Gen Test Suite

PlaceRoute

FullChipSynth

Verify RTL

FloorPlanning

Write B1RTL


PhysVerifn

Verify B1RTL


Synth_B1RTL

Rework


Project Lead




Engineer

VerificationTeam

DesignCoordination


Insert Scan

Partition Chip

Gen Test Suite

PlaceRoute

FullChipSynth

Verify RTL

FloorPlanning

Write B1RTL


PhysVerifn

Verify B1RTL


Synth_B1RTL

ManagementMeeting

Architecture TeamMeeting

Foundry TeamMeeting

1

1 1 1 1

1

1

1

1

1

1

Meetings


Project Lead




Engineer

VerificationTeam

DesignCoordination


Insert Scan

Partition Chip

Gen Test Suite

PlaceRoute

FullChipSynth

Verify RTL

FloorPlanning

Write B1RTL


PhysVerifn

Verify B1RTL


Synth_B1RTL

0.8

4

4

4

4

1

1

1

1

1 1

1

1

1

1

1

Assignments 1


Project Lead




Engineer

VerificationTeam

DesignCoordination


Insert Scan

Partition Chip

Gen Test Suite

PlaceRoute

FullChipSynth

Verify RTL

FloorPlanning

Write B1RTL


PhysVerifn

Verify B1RTL


Synth_B1RTL

0.8

4

4

4

4

1

1

1

1

1 1

1

1

1

1

ManagementMeeting

Architecture TeamMeeting

Foundry TeamMeeting

1

1 1 1 1

1

1

1

1

1

1

1

Model Simulation Results

37



Steps in VDT/SimVision Project DesignIdentify client’s key business issues and risksDevelop Flexibility Matrix and trade-offsModel “Baseline Case”

Lay out 5-10 key business milestones per projectIdentify and sequence 5-10 activities per milestone

Lay out organization:– structure, positions, capacity, skills, decision making policies

Assign each task to one responsible position“Flight-simulate” Baseline Case

Diagnose backlogs, schedule and quality risksExplore potential interventions to mitigate risks

Choose a project design that is likely to succeed

38



Level of Effort for Project/Program (Re)Design

Elements of Fast-Track ProgramDesign and Redesign

ClientEffort

(FTE-days)

AnalystEffort

(FTE-days)

Gather data from client about business objectives, 0.5 0.5milestones, tasks, costs, staffing, known risks, etc.

Build “straw-man” as-planned, baseline model 0 1-2

Discuss and refine model 0.5 1-2

Diagnose risks with baseline case 0.25 0.5

Evaluate multiple potential interventions 0.25 0.5

Produce recommendations and report 0 1.0

Ongoing Redesign (Tracking) per cycle 0.5 1.0

39



VDT Fast-Track Project Design: ExamplesReduce time to market for complex manufacturing facilities

Facilitate roll-out of new wireless telecom infrastructure across multiple regions

Develop best practices template toaccelerate factory start-ups

Identify and correct subcontractor management problem that would have delayed project 4 mo.

Help to meet ship milestone date required to close sale with large customer

Align goals and accelerate rollout of innovative consumer product by 3 months

Identify and mitigate critical quality risks to accelerate rollout of new server product

Help to define scope, schedule and organization for strategic IT projects

40



& of VDT/SimVision for Fast-Track ProjectsVDT/SimVision uniquely highlights impact of fast-track work process on cost, schedule and process quality

VDT/SimVision shows impacts of differences in participants skills & experience on project outcomes

Small models and graphical inputs/outputs engage executives in project design process

Models only organizational risks—not technical or market risks (these risks require separate “scenarios”)

VDT organizational model assumes hierarchical exception handling

41



Analysis Tools Can Support “Fast-Track Project Design”Conceptual

DesignDetailed Design and Execution

Closeout &Learning

0%

100%

Level of InfluenceLevel of Influence

Expenditure of Funds Expenditure of Funds

OutcomeKnowledge

OutcomePredictions

Analysis Tools- PERT Simulation- Critical Chain- System Dynamics- Virtual Design Team

42



Outline




43



Trajectory of Past VDT Project Design Research

99-02: LambertBuettner

97-00: Miller

96-99: Fridsma/Cheng

More Impact of IT/ Communication

technologies

More Impact of IT/ More Impact of IT/ Communication Communication

technologiestechnologies More Impact of Social ProcessesMore Impact of More Impact of

Social ProcessesSocial Processes

More Innovative Work ProcessesMore Innovative More Innovative Work ProcessesWork Processes

95-99: ThomsenSalazar-Kish

90-94: Cohen/Christiansen

Single Project

Program

Enterprise

Communities ofPractice, LearningOrganizations

Single Project

Program

Enterprise

Communities ofPractice, LearningOrganizationsMore Flexible

OrganizationsMore FlexibleMore FlexibleOrganizationsOrganizations

44



Ongoing Stanford Project Design Research

Direct Costs

Complex, Fast-Track “Engineering Projects”

CoordinationCosts

* Effects of Coordination complexity

21 stCentury “Global Change” Projects

* Effects of Institutional Complexity

InstitutionalCosts

$, ¥, €, …

45



planning and executing fast-track projects › scv-tems › files › 2016 › 12 › tems... ·...

Documents