System Dynamics - Model - The Rework Cycle

ESD.36 System Project Management

Instructor(s)

+

- The Rework Cycle

Dr. James Lyneis

Lecture 7

Copyright 2012 James M Lyneis.

+

-

2

Todays Agenda

Overview: Causes of Project Dynamics

The Rework Cycle

Integrating Tools in Project Planning

Simple Model of Project Dynamics, Pt. 1

+

- Typical Behavior Modes on a Project

3

ProjectStaffing

Time

TypicalPlan

Productivity(Normalised)

Time

1

2TypicalPlan

2001 John Wiley & Sons, Ltd. All rights reserved. This content is excluded from ourCreative Commons license. For more information, see http://ocw.mit.edu/fairuse.

+

-

4

Drivers of Project Dynamics

The rework cycle

Feedback effects

Negative, controlling

Positive, re-enforcing, often vicious circles

Knock-on or domino effects within or between work phases

Knock-on or domino effects between projects

+

-

5

Todays Agenda


The Rework Cycle



+

-

Network Diagram for NMM Case (from 2010 Homework #1)

a, Start b, Req

c,DieselSpecs

d, ElecDriveSpecs

e,Layout

f,Electronics

Design

g, Software Specs

h, Software Dev

i,DieselDev

j, ElecDriveDev

k,ChassisDesign

l, BodyDesign

m,InternalFittings

s, Body/Chassis

Prototype

p,Electronics

Delivery

q,ElectronicsSoftware

Integ

n,Diesel

Delivery

o,Integra

tionr,

ChargingAssembly

t,Power-Vehicle

Integ

U, FinalAssembly

v,DynamoLab Test

w,RoadTest

x, CarShowSetup

y, Finish

6

+

-

7

The Network View of a Program Task Accomplishment

Task Accomplishment Task X

Work to Do Work DoneWork Being

Accomplished

Productivity Staff

7

+

-

Work to Do Work DoneWork Being

Accomplished

Productivity Staff

Network Diagram for NMM Case

a, Start b, Req

c,DieselSpecs

d, ElecDriveSpecs

e,Layout

f,Electronics

Design

g, Software Specs

h, Software Dev

i,DieselDev

j, ElecDriveDev

k,ChassisDesign

l, BodyDesign

m,InternalFittings

s, Body/Chassis

Prototype

p,Electronics

Delivery


Integ

n,Diesel

Delivery

o,Integra

tionr,

ChargingAssembly

t,Power-Vehicle

Integ

U, FinalAssembly

v,DynamoLab Test

w,RoadTest

x, CarShowSetup

y, Finish

Two stocks and one flow subsumed within each task of network diagram

8

+

-

Work Backlogs and Flow

1,000 Tasks100 Tasks/Month

500 Tasks50 Tasks/Month

0 Tasks0 Tasks/Month

0 5 10 15 20 25 30 35 40 45 50Time (Month)

Work to Do : Traditional View Tasks

Work Done : Traditional View Tasks

Work Being Accomplished : Traditional View Tasks/Month

Aggregate Model of Network View

9

Initial Work to Do = 1000 tasks Productivity = 1 task/mo/person Staff = 40 people

Work Done Work to Do

Work Being Accomplished

Work to Do Work Done

StaffProductivity

Initial Workto Do

Work BeingAccomplished

+ +

+

-

10

OR...

Data Often Tell a More Complex Story

Work to Do Work Done

StaffProductivity


+ +

10

PRODUCTIVITY?

+

-

Work to Do WorkDone

Rework

x yWork Being

Accomplished

StaffProductivity Fraction Correct

and Complete

z

ReworkCreation

ReworkExecution

Effort is being spent elsewhere

Sep 11, 2007 11

(0-1: fraction not to be reworked)

+

- Definitions

Productivity --

Work accomplished per hour of effort,

regardless of completeness or correctness

Fraction Correct & Complete --

Fraction of work just accomplished that will

not need rework. Work Quality

Later we will revisit the iron triangle to

consider delivered quality.

12

+

-

OriginalWork to Do Work Done

UndiscoveredRework

Progressx

Fraction Correctand Complete

Time to Discover Rework

Rework toDo

ReworkDiscovery

- +

Productivity

EffortApplied

Rework

ProgressRework

Generation

++ -

The Rework Cycle

Delay!

13

Delay in Discovering Rework

+

-

Work Done

1,000

750

500

250

0

0 5 10 15 20 25 30 35 40 45 50Time (Month)

Ta

sks

Work Done : Traditional With ReworkWork Done : Traditional View

14

Adding Rework Delays Completion

No Rework

With Rework

Undiscovered Rework

60

45

30

15

0

0 5 10 15 20 25 30 35 40 45 50Time (Month)

Tasks

Undiscovered Rework : Traditional With Rework

+

- So, What Happens on Projects?

Staff

Effort on Initial

Work

TIME

Effort on Rework

15

+

-

Rework creates and extended staffing tail

16

17

Simulated Actual

Year

800.

600.

400.

200.

0 0 3 6 9 12 15 18 21

ApproximateOriginal Plan

Disguised results from

actual aerospace

project

Staffing experiences an extended tail

0 3 6 9 12 15 18 21

800.

600.

400.

200.

0

Work Assignments of Staff to...

Original Work

Rework

Total

2001 John Wiley & Sons, Ltd. All rights reserved. Thiscontent is excluded from our Creative Commons license.For more information, see http://ocw.mit.edu/fairuse.

+

- or a second staffing peak

17

20

Program Staff, Simulated vs. Data (Equivalent Staff)

400.

300.

200.

100.

TIME

0 Year 1 Year 2 Year 3 Year 4 Year 5 Year 6

Simulated Original Plan Actual


actual vehicle project

A second staffing peak

Work Assignments of Staff to...

400.

300.

200.

100.

TIME

0 Year 1 Year 2 Year 3 Year 4 Year 5 Year 6

Original Work Rework


actual vehicle project

Rework

Original Work

2001 John Wiley & Sons, Ltd. All rights reserved. Thiscontent is excluded from our Creative Commons license.For more information, see http://ocw.mit.edu/fairuse.

+

-


UndiscoveredRework

Progressx


Rework toDo

ReworkDiscovery

Productivity

EffortApplied

Rework

Progress+

+Rework

Generation

-

Consequences of Undiscovered Rework

1.Overestimates of Progress

18

+

-

19

Additional Metrics (Used in Model)

Fraction Reported to be Complete =

(Work Done + Undiscovered Rework) / Initial Work to Do

Fraction of Work Really Complete =

(Work Done) / Initial Work to Do

+

-

Progress

1

0.75

0.5

0.25

0

0 5 10 15 20 25 30 35 40 45 50Time (Month)

Fra

ctio

n

Fraction Reported Complete : Traditional With Rework3

Fraction Really Complete : Traditional With Rework3

20

Undiscovered rework generates progress-measuring errors

Reported

Really

Undiscovered Rework

+

-

Rework Cycle Creates The Lost-Year...

Design Progress (Percent Complete)

Simulated 100.

75.

50.

25.

0.

Disguised results from actual aerospace project

Time

Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9

The Lost Year

Real Progress

21

Source: Ken Cooper

+

-

22

and the 90% Syndrome --- Design Progress (Percent Complete)

Simulated100.

75.

50.

25.

0.

TIME1/85 1/89 1/91 1/93 1/95 1/97 1/99 1/01

Disguised results

from actual

aerospace project

Year 1 Year2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8

+

- Survey Question 1

Does your organization measure rework?

1. Yes, we keep data on how much rework is being discovered

2. Yes, we keep data on how much work being done is rework (vs original work)

3. Yes, both 1 and 2

4. No, we do not keep data on rework being discovered

23

+

- Survey Question 2

Does your organization recognize rework in executing the project?

1. Yes, by adjusting progress estimates to reflect expected remaining undiscovered rework

2. Yes, by building rework tasks into the project graph

3. Both 1 and 2

4. No, we only react as rework is discovered

24

+

-


UndiscoveredRework

Progressx


Rework toDo

ReworkDiscovery

Productivity

EffortApplied

Rework

Progress+

+Rework

Generation

-

Consequences of Undiscovered Rework

25

1. Overestimates of Progress

2. Build Errors Into Downstream Work

+

- Survey Question 3

How significant is the Errors on Errors feedback on typical projects in your organization?

1. Very

2. Modest

3. Weak

4. None

26

+

- Units of Work SD Model

Tasks -- ideally, of uniform size and fungible

In actual applications, might use things like:

Drawings, work packages, lines of code,

Tons of concrete, feet of steel, feet of wiring, ...

Parts designed

Sometimes %

Can represent precedence constraints

27

+

- Quality

Need to distinguish between:

Instantaneous work quality (fraction correct and complete)

Delivered quality remaining bugs (undiscovered rework) when product shipped

Product capabilities, fit and finish (a part of scope in model)

28

+

- Fraction Correct and Complete

Represents unplanned iterations; planned iterations are separate tasks

Sources of errors many fold:

Mistakes from inexperience, fatigue,

Technical complexity/uncertainty

Work done correctly but ultimately needing rework because it builds on Incorrect prior work

Assumptions about technology or customer requirements which prove incorrect

29

+

- Two Types of Iteration

Planned Iteration Caused by needs to get it

right the first time.

We know where these iterations occur, but not necessarily how much.

Planned iterations should be facilitated by good design methods, tools, and coordination.

Separate tasks in the rework

cycle

Unplanned Iteration Caused by errors and/or

unforeseen problems.

We generally cannot predict which unplanned iterations will occur.

Unplanned iterations should be minimized using risk management methods.

Quality problems in the rework cycle

30

+

-

31

Qualitative Insights

Undiscovered Rework is one of the most important single factors driving schedule and budget overruns

Most management reporting systems overestimate real progress and discourage reporting of rework

+

-


UndiscoveredRework

Progressx

Quality

Time to Discover Rework

Rework toDo

ReworkDiscovery

- +

Productivity

EffortApplied

Rework

ProgressRework

Generation

++ -

Lessons: Recognize the rework cycle and minimize its consequences

Fraction Correct and Complete

Increase

Reduce Be Realistic

Overestimates of Progress

Build Errors Into Downstream Work

32

Delay in Discovering Rework

+

-

33

Dynamics of Project Performance

The rework cycle Fraction correct and complete

Undiscovered rework

Feedback effects on productivity and fraction correct (Next class)



Knock-on effects between work phases (Next)

Availability and quality of work products

Progress to discover upstream rework

+

-

34

Todays Agenda

Causes of Project Dynamics

The Rework Cycle



+

-

35

Fundamental Approaches

Network-based (graph theory) methods CPM, PERT, . Task is a node or an arc

Matrix-based methods DSM - Tasks are columns and rows Interrelationships are off-diagonal entries

System Dynamics Feedback loops, causal relationships Stocks and flows simulation Tasks that are done or waiting to be done

are stocks amount of work Doing project work causes a flow

a b

c d e

x x x x x

x x

x x

x

W OR K

BEING DO NE

P EO P LE PRODUCTIVITY

W ORKTO BEDONE

W ORKDONE

What have we learned thus far?

+

-

40

1

2

3

7 5 6

4

8

9

10

11

12

13

14

15 16

17

18

19

23

21

22

20

24

28

25

29

27

26 30

31

32 33

35 38 34

37 36

39

Network View of CityCar Project

+

-

Network Diagram for NMM Case (from 2010 Homework #1)

a, Start b, Req

c,DieselSpecs

d, ElecDriveSpecs

e,Layout

f,Electronics

Design

g, Software Specs

h, Software Dev

i,DieselDev

j, ElecDriveDev

k,ChassisDesign

l, BodyDesign

m,InternalFittings

s, Body/Chassis

Prototype

p,Electronics

Delivery


Integ

n,Diesel

Delivery

o,Integra

tionr,

ChargingAssembly

t,Power-Vehicle

Integ

U, FinalAssembly

v,DynamoLab Test

w,RoadTest

x, CarShowSetup

y, Finish

Note: Diesel, Elec Drive, and Software do not show separate design and build tasks.

37

+

- Organizing the Tasks by Metatask

a, Start b, Req

c,DieselSpecs

d, ElecDriveSpecs

e,Layout

f,Electronics

Designg, Software Specs

h,Software

Dev

i,DieselDev

j, ElecDriveDev

k,ChassisDesign

l, BodyDesign

m,InternalFittings

s, Body/Chassis

Prototype

p,Electronics

Delivery


Integ

n,Diesel

Delivery

o,Integra

tionr,

ChargingAssembly

t,Power-Vehicle

Integ

U, FinalAssembly

v,DynamoLab Test

w,RoadTest

x, CarShowSetup

y, Finish

Requirements Definition

Electronics & Software Design

Body & Chassis Design

Diesel & Electric Drive Design

Vehicle & Power- train Integration

Final Assembly

Testing & Completion

38

+

-

Gantt Chart with Metatasks Shown Note: dates approximate; JML estimates of Diesel & Electric, Software Design/Build Split

2010 2011

Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec

Requirements Definition Requirementsa Start 1/1/10 Design

b Requirements 60 1/1/10 3/25/10 a Build/Test

c Diesel Specs 30 3/25/10 5/6/10 b Coupled

d Elec Drive Specs 30 3/26/10 5/6/10 b

g Software Specs 72 3/26/10 7/5/10 b

Body & Chassise Layout 40 3/26/10 6/1/10 b

k Chassis Design 102 6/2/10 10/21/10 e

l Body Design 90 6/2/10 10/5/10 e

s Body Chassis Prototype 48 10/22/10 12/28/10 k,l

Electronics & Softwaref Electronics Design 90 5/7/10 9/9/10 b,c,d

h Software Dev 150 9/10/10 4/7/11 g,f

p Electronics Delivery 72 9/10/10 12/20/10 g,f

q Electronics Software Int 30 4/8/11 5/19/11 h,p

Diesel & Electric Drivei Diesel Dev 180 5/7/10 1/13/11 c

j Elec Drive Dev 90 5/7/10 9/9/10 d

n Diesel Delivery & Check 12 1/14/11 1/31/11 i

Vehicle & Power Train Into Integration 60 2/1/11 4/25/11 j,n

r Charging Assembly 12 4/26/11 5/11/11 j,o

t Power Vehicle Int 60 5/20/11 8/11/11 q,r,s

Final Assemblym Internal Fittings 48 10/22/10 12/28/10 k,l

u Final Assembly 30 8/12/11 9/22/11 m,t

Testing & Completionv Dynamo Lab Test 30 9/23/11 11/3/11 u

w Road Test 72 11/4/11 2/13/12 v

x Car Show Setup 45 11/4/11 1/5/12 v

y Finish 2/13/12

39

+

-

What Messages Come From DSM and SD regarding the network/Gantt plan?

DSM ?

SD ?

How/when is rework discovered?

40

+

-

DSM Identifies Coupled Tasks & Areas of Unplanned Iterations (from Homework #2)

a, Start b, Req

c,DieselSpecs

d, ElecDriveSpecs

e,Layout

f,Electronics

Designg, Software Specs

h,Software

Dev

i,DieselDev

j, ElecDriveDev

k,ChassisDesign

l, BodyDesign

m,InternalFittings

s, Body/Chassis

Prototype

p,Electronics

Delivery


Integ

n,Diesel

Delivery

o,Integra

tionr,

ChargingAssembly

t,Power-Vehicle

Integ

U, FinalAssembly

v,DynamoLab Test

w,RoadTest

x, CarShowSetup

y, Finish

Requirements Definition

Electronics & Software Design

Body & Chassis Design

Diesel & Electric Drive Design

Vehicle & Power- train Integration

Final Assembly

Testing & Completion

Coupled Tasks Possible planned Iterations (in design) Unplanned Iterations (why not back to design phase?)

41

+

- Gantt Chart With Coupled Tasks Shown

2010 2011

Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec

Requirements Definition Requirementsa Start 1/1/10 Design

b Requirements 60 1/1/10 3/25/10 a Build/Test

c Diesel Specs 30 3/25/10 5/6/10 b Coupled

d Elec Drive Specs 30 3/26/10 5/6/10 b

g Software Specs 72 3/26/10 7/5/10 b

Body & Chassise Layout 40 3/26/10 6/1/10 b

k Chassis Design 102 6/2/10 10/21/10 e

l Body Design 90 6/2/10 10/5/10 e

s Body Chassis Prototype 48 10/22/10 12/28/10 k,l

Electronics & Software

f Electronics Design 90 5/7/10 9/9/10 b,c,d

h Software Dev 150 9/10/10 4/7/11 g,f



Diesel & Electric Drive

i Diesel Dev 180 5/7/10 1/13/11 c

j Elec Drive Dev 90 5/7/10 9/9/10 d

n Diesel Delivery & Check 12 1/14/11 1/31/11 i

Vehicle & Power Train Int

o Integration 60 2/1/11 4/25/11 j,n

r Charging Assembly 12 4/26/11 5/11/11 j,o

t Power Vehicle Int 60 5/20/11 8/11/11 q,r,s

Final Assemblym Internal Fittings 48 10/22/10 12/28/10 k,l

u Final Assembly 30 8/12/11 9/22/11 m,t

Testing & Completionv Dynamo Lab Test 30 9/23/11 11/3/11 u

w Road Test 72 11/4/11 2/13/12 v

x Car Show Setup 45 11/4/11 1/5/12 v

y Finish 2/13/12

Observation/ Question: no coupling between major design phases?

42

+

-

a, Start b, Req

c,DieselSpecs

d, ElecDriveSpecs

e,Layout

f,Electronics

Design

g, Software Specs

h,Software

Dev

i,DieselDev

j, ElecDriveDev

k,ChassisDesign

l, BodyDesign

m,InternalFittings

s, Body/Chassis

Prototype

p,Electronics

Deliveryq,

ElectronicsSoftware

Integ

n,Diesel

Delivery

o,Integra

tionr,

ChargingAssembly

t,Power-Vehicle

Integ

U, FinalAssembly

v,DynamoLab Test

w,RoadTest

x, CarShowSetup

y, Finish

i*,DieselBuild

j*, ElecDriveBuild

f*,Electronics

"Build"

h*,Software"Build"

Network Diagram With Build Steps Added and Major Work Phases Highlighted

Requirements Design Build/Test

43

+

-

In an SD Model, Each Phase of Work Could Be Represented by a Rework Cycle

OriginalWork to Do

Work Done

UndiscoveredRework

Original Work Done Correctly

Rework Generation onOriginal Work


Staff

Productivity

Original Work BeingAccomplishmed

Rework BeingAccomplishmed

Rework to Do

ReworkDiscovery

ReworkGeneration on

Rework

Staff on Rework

Staff onOriginal Work

44

+

- Expanded to Three Phases

45

RqmntsOriginal

Work to Do

RqmentsWork Done

RqmentsUndiscovered

Rework

Rqmnts Original Work Done Correctly

Rqmnts ReworkGeneration on Original

Work

RequirementsFraction Correct and

Complete

Requriements Staff

Requirements Productivity

Rqmnts Original WorkBeing Accomplishmed

Rqmnts ReworkBeing

Accomplishmed

RqmntsRework to Do

Rqmnts ReworkDiscovery

Rqmnts ReworkGeneration on

Rework

Rqmnts Staff onRework

Rqmnts Staff onOriginal Work

Design

OriginalWork to Do

Design WorkDone

DesignUndiscovered

Rework

Design Original Work Done Correctly

Design ReworkGeneration on Original

Work

Design FractionCorrect andComplete

Design Staff

Design Productivity

Design Original WorkBeing Accomplishmed

Design ReworkBeing

Accomplishmed

DesignRework to Do

Design ReworkDiscovery

Design ReworkGeneration on

Rework

Design Staff onRework

Design Staff onOriginal Work

Build/TestOriginal

Work to Do

Build/TestWork Done

Build/TestUndiscovered

Rework

Build/Test Original Work Done Correctly

Build/Test ReworkGeneration on Original

Work

Build/Test FractionCorrect and Complete

Build/Test Staff

Build/Test Productivity

Build/test OriginalWork Being

AccomplishmedBuild/Test Rework

Being Accomplishmed

Build/TestRework to Do

Build/Test ReworkDiscovery

Build/Test ReworkGeneration on Rework

Build/Test Staffon Rework

Build/Test Staff onOriginal Work

Requirements Design Build/Test

Assumptions: Scope = 100 Tasks Scope = 1000 tasks Scope = 1000 tasks Staff = 4, Staff = 20 Staff = 40 Productivity = 2 tasks/month/person Productivity = 4 tasks/month/person Productivity = 1 tasks/month/person Duration = 12.5 months Duration = 12.5 months Duration = 25 months

+

-

Project Staff

60

45

30

15

0

0 6 12 18 24 30 36 42 48 54 60Time (Month)

Pe

op

leRequirements Staff : Three P Four Stock No Rework Discrete Start

Design Staff : Three P Four Stock No Rework Discrete Start

"Build/Test Staff" : Three P Four Stock No Rework Discrete Start

Project Staff : Three P Four Stock No Rework Discrete Start

Work Done

2,000

1,500

1,000

500

0

0 6 12 18 24 30 36 42 48 54 60Time (Month)

Ta

sk

Requirements Work Done : Three P Four Stock No Rework Discrete Start

Design Work Done : Three P Four Stock No Rework Discrete Start

"Build/Test Work Done" : Three P Four Stock No Rework Discrete Start

Simulation of project assuming no rework

Requirements

Design

Build/Test Requirements

Design

Build/Test Total

46

+

- Sources of Rework -- Categories

1. Classical Quality or design mis-execution.

2. Technical complexity/uncertainty; customer uncertainty.

3. Work done correctly but ultimately needing rework because it builds on

47

+

- Classical Quality

There are two (or more) important sources of these errors People factors such as fatigue, inexperience, skill mismatches,

etc.

Coupled tasks that require shared information but which are done independently

Organizational size & complexity?

These types of errors can in theory be discovered by further design work, such as doing downstream design work, QA, design reviews, planned iteration, etc.

48

+

-

Technical complexity/uncertainty; customer uncertainty.

Technical: novelty of the project

Customer: Same logic might apply to novel/new uncertain customer requirements? I.e., where say a software product is being developed and the customer may not know what they want until they see it in practice.]

These types of errors can only be discovered by doing build and test work. Fixing these errors may just require redoing tasks, but may also require new tasks to make the technology work (e.g., and additional controller).

49

+

- Knock-on Rework

Work done correctly but ultimately needing rework because it builds on Incorrect prior work

Assumptions about technology or customer requirements which prove incorrect.

Exogenous changes in requirements, scope, etc.

These errors could be discovered in design or build/test, depending on the source.

50

+

- How does Design Affect Build?

51

DesignOriginal

Work to Do

Design WorkDone

DesignUndiscovered

Rework



Work


Design Staff

Design Productivity


Design ReworkBeing

Accomplishmed

DesignRework to Do



Rework



Build/TestOriginal

Work to Do

Build/TestWork Done


Rework



Work


Build/Test Staff




Being Accomplishmed






Design FractionReported Complete

Design Fraction of WorkDone Correct and

Complete

Build/Test FractionReported Complete

+

- How is Design Rework Discovered?

52

DesignOriginal

Work to Do

Design WorkDone

DesignUndiscovered

Rework



Work


Design Staff

Design Productivity


Design ReworkBeing

Accomplishmed

DesignRework to Do



Rework



Build/TestOriginal

Work to Do

Build/TestWork Done


Rework



Work


Build/Test Staff




Being Accomplishmed








Complete


+

-

Project Staff

80

60

40

20

0

0 6 12 18 24 30 36 42 48 54 60Time (Month)

People

Requirements Staff : Three Phase Four Stock V1

Design Staff : Three Phase Four Stock V1

"Build/Test Staff" : Three Phase Four Stock V1

Project Staff : Three Phase Four Stock V1

Simulation Incorporating Rework and Rework Discovery

Work Done

1,000

750

500

250

0

0 6 12 18 24 30 36 42 48 54 60Time (Month)

Ta

sk

Requirements Work Done : Three Phase Four Stock V1

Design Work Done : Three Phase Four Stock V1

"Build/Test Work Done" : Three Phase Four Stock V1

Requirements

Design

Build/Test

Requirements Design

Build/Test Total

Notes: (1) Details of simulation do not correspond to NMM Case; (2) so far, we do not represent project control actions and effects, so staff/hours worked are constant.

53

+

-

Design Stocks

1,000

750

500

250

-0.0001

0 6 12 18 24 30 36 42 48 54 60Time (Month)

Ta

sk

Design Original Work to Do : Three Phase Four Stock V1

Design Undiscovered Rework : Three Phase Four Stock V1

Design Rework to Do : Three Phase Four Stock V1

Design Work Done : Three Phase Four Stock V1

Original Work

Undiscovered Rework

Rework

Work Done

54

0

+

-

Design Fraction Complete

2

1.5

1

0.5

0

0 6 12 18 24 30 36 42 48 54 60Time (Month)

Fra

ctio

n

Design Fraction Original Work Complete : Three Phase Four Stock V1

Design Fraction Reported Complete : Three Phase Four Stock V1

Design Fraction Really Complete : Three Phase Four Stock V1

Fraction Design Rework Discovered

0.8

0.6

0.4

0.2

0

0 6 12 18 24 30 36 42 48 54 60Time (Month)

Fra

ctio

n

Fraction Design Rework Discovered : Three Phase Four Stock V1

Original Work

Reported

Really

Discovery by design

Discovery by build

55

Progress and Rework Discovery

+

-

Rework Increases Cost and Delays Finish

Cumulative Effort

2,000

1,500

1,000

500

0

0 6 12 18 24 30 36 42 48 54 60Time (Month)

Pe

rso

n*M

on

ths

Cumulative Reqmnts Effort : Three Phase Four Stock V1

Cumulative Design Effort : Three Phase Four Stock V1

"Cumulative Build/Test Effort" : Three Phase Four Stock V1

Cumulative Effort : Three Phase Four Stock V1

Requirements

Design

Build/Test Total

With Rework

Cumulative Effort

2,000

1,500

1,000

500

0

0 6 12 18 24 30 36 42 48 54 60Time (Month)

Pe

rso

n*M

on

ths

Cumulative Reqmnts Effort : Three P Four Stock No Rework

Cumulative Design Effort : Three P Four Stock No Rework

"Cumulative Build/Test Effort" : Three P Four Stock No Rework

Cumulative Effort : Three P Four Stock No Rework

Without Rework

56

+

-

Revised Network/Gantt to reflect rework discovery (by build, not design)

Insert some Rework Tasks in Design, more in Build/Test

57

2010 2011

Apr May June July Aug Sept Oct Nov Dec Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec

Design R ReworkElectronics & Software -- Original Build/Test

f Electronics Design 90 5/7/10 9/9/10 b,c,d Coupled

h Software Dev 150 9/10/10 4/7/11 g,f



Electronics & Software -- Rework Discovered in Build/Testf Electronics Design 90 5/7/10 9/9/10 b,c,d R R R R Rh Software Dev 150 9/10/10 4/7/11 g,f R R R R R R Rp Electronics Delivery 72 9/10/10 12/20/10 g,f R R R R R Rq Electronics Software Int 30 4/8/11 5/19/11 h,p R R

+

- Management Issues

Planned Iterations (vs dealing with unplanned iterations later) How many iterations?

When to start build phases how much overlap with design phases.

Factors to Consider:

Type of project (determines amount of rework discoverable in design, amount of rework, number of tasks which must be repeated for each iteration)

Relative cost of design vs build/test (determines the relative cost of spending more in design)

58

More on these later in term

+

- Role of Tools

SD:

Evaluate the macro tradeoffs in terms of impact on cost and schedule

Coming up with a good staffing plan

DSM:

Determine where design iterations are needed and how many tasks would need to be repeated per iteration

Network:

Guide operational task planning and resource scheduling

59

+

-

60

Dynamics of Project Performance

The rework cycle Fraction correct and complete

Undiscovered rework

Feedback effects on productivity and fraction correct (Next class)



Knock-on effects between work phases

Availability and quality of work products

Progress to discover upstream rework

+

-

61

Knock-on Between Phases: A system dynamics model usually represents several phases of work

S ystem

E n g in eerin g

S o ftw are C o d e

an d T es t

H ard w are

D es ig n

H ard w are B u ild

an d T es t

In teg ra te

an d T es t

S o ftw are

D es ig n

Productivity and Fraction Correct Effects (availability and quality of upstream work)

Rework Discovery Effects

Notes: Suppliers might be a phase

+

-

62

Why Represent Separate Phases?

Correspond to gate reviews and deliverables

Different units of work (and therefore data)

Different types of labor

Disproportionate increase in costs of fixing errors as move downstream

Address issues of overlap/concurrency

Opportunity to adjust plans, delay start, between phases

+

-

63

Todays Agenda


The Rework Cycle



+

-

64

Hard tools force us to be more explicit, and accurately simulate the consequences of our models ...

Soft tools --

behavior-over-time graphs

cause-effect diagramming

mental simulation

Tools for describing dynamics

Hard tools --

computer models

computer simulation

calibration to data

sensitivity and what-if analyses

Tools for quantifying dynamics

+

- We will use two models

Simple rework cycle model with project control feedbacks

HW#3 develop simple model without feedbacks

Feedbacks added in class, given in HW#5

Full rework cycle model with two phases of work

No project control feedback

Model given to you for HW#3 and HW#5

65

+

- Development of Simple Project Model 1

Rework cycle model (HW#3)

Three stocks

Variable rework discovery time

Errors Build Errors Feedback

Project control & Side Effects (HW#5)

Work Intensity/Schedule Pressure & Haste Makes Waste

Staffing & Experience Dilution

Slip Schedule

66

+

- Two Views of the Rework Cycle

Work to Do Work DoneUndiscovered

Rework

Work Accomplishment

Rework Generation

Rework Discovery

Complete Model

Simplified Version

The simplified version assumes that rework tasks require the same effort as original tasks, and that it is not important to distinguish between original work and rework.

Fraction Correct & Complete


UndiscoveredRework

Progressx

Rework toDo

ReworkDiscovery

Rework

Progress

67

+

- Simple Rework Cycle Model

68

Steps 1-4 of Homework #3


Rework

Work DoneCorrectly

Rework Generation

Rework Discovery


-

Staff

Productivity

Potential Work RateBased on Staffing

Levels

+

+

Project Finished

CumulativeWork Done

Rate of Doing Work

Initial Work toDo

CumulativeEffort

ExpendedEffort Expended

Time to DiscoverRework

-

+


Maximum Work RateBased on Tasks

Available

Minimum Time toFinish a Task

+

+

+

Fraction Completeto Finish

+--

+

Normal Staff

+

+

- Rework Discovery Depends on Progress

69

+

- Complete Simple Model 1

70

Errors on Errors Feedback


Rework

Work DoneCorrectly

Rework Generation

Rework Discovery


Staff

Productivity

Potential Work RateBased on Staff

Project Finished

CumulativeWork Done

Rate of Doing Work

Initial Work toDo

Work Believed toBe Done

Fraction ReallyComplete

CumulativeEffort

ExpendedEffort Expended

Fraction of Rework Discovered


Maximum Work RateBased on Tasks

Available

Minimum Time toFinish a Task

Delay inDiscovering Rework

Fraction ReportedComplete

Fraction Completeto Finish

Normal Staff

Effect of UndiscoveredRework on Fraction

Correct

Fraction of WorkBelieved Done Correct

and Complete

Normal FractionCorrect and Complete

Maximum Effect ofUndiscovered Rework on

Fraction Correct

Sensitivity of FractionCorrect to Undiscovered

Rework

This reflects the average delay in

discovering discoverable rework,

such as from QA activities or

downstream work.

This reflects the fraction of

undiscovered rework that is

discoverable at any point in the project

based on the activities tak ing place.

+

-

Effect of Undiscovered Rework on Fraction Correct:

71

Note: The effect of undiscovered rework on fraction correct is assumed to be

proportional -- an error in past work creates an error in current work. Given that in this

simple model fraction correct represents several effects of work errors, this strong

relationship may be reasonable.

Average Work Quality

Graph Lookup - Table for Effect of Prior Work Quality on Quality

1

0

0 1

Effect of Undiscovered Rework on Fraction Correct

Fraction of Work Believed Done Correct and Complete

+

-

Work Flows & Staffing in Simple Two Phase Model

72

Design Startup

Project Staff


Build/TestStartup

Build/Test Delay inDiscovering Rework


Normal DesignFraction Correctand Complete

Build/Test Effect of DesignUndiscovered Rework On

Fraction Correct


Complete

Design WorkBelieved to Be Done

Sensitivity of Build/TestFraction Correct to Design

Undiscovered Rework

Normal Build/TestFraction Correct and

Complete

Build/Test WorkBelieved to Be Done

Start DesignRampup

End DesignRampup

Fraction DesignComplete to Start Build

Rampup

Duration of BuildRampup

DesignOriginal

Work to Do

Design WorkDone

DesignUndiscovered

Rework

Design OriginalWork Done Correctly


Work


Design Staff

DesignPlanned Staff

Design MinimumTime to Finish a Task

Design Fraction ofEffort to Rework


Design ReworkBeing

Accomplishmed

Design Maximum WorkRate Based on OriginalWork Tasks Available

DesignRework to Do



Rework


DesignProductivity on

Rework

Design Relative EffortRequired for Rework


Design ReworkDone Correctly

DesignIndicated Staff

Design Maximum WorkRate Based on Rework

Tasks Available

Design Priority toOriginal Work

Build/TestOriginal

Work to Do

Build/TestWork Done


Rework

Build/Test OriginalWork Done Correctly


Work


Build/Test Staff

Project FinishedSwitch

Build/TestPlanned Staff

Build/Test MinimumTime to Finish a Task

Build/Test Fractionof Effort to Rework

Build/Test OriginalWork Being

Accomplishmed

Build/Test ReworkBeing Accomplishmed

Build/Test Maximum WorkRate Based on OriginalWork Tasks Available




Staff on Rework

Build/TestProductivity on

Rework

Build/Test RelativeEffort Required for

ReworkBuild/Test Staff on

Original Work

Build/Test ReworkDone Correctly

Build/TestIndicated Staff

Build/Test Maximum WorkRate Based on Rework

Tasks Available

Build/Test Priorityto Original Work

Build/Test Effect of BuildUndiscovered Rework on

Fraction Correct

Build/Test Fraction of WorkBelieved Done Correct and

Complete

Sensitivity of Build/TestFraction Correct to BuildUndiscovered Rework


Fraction Correct

Design InitialWork to Do

Design Effect ofUndiscovered Rework on

Fraction Correct

Sensitivity of DesignFraction Correct to Design

Undiscovered Rework

Maximum Sensitivity ofBuild/Test FCC to Design

Undiscovered Rework

Maximum Sensitivity ofBuild/Test Fraction Correct toBuild Undiscovered Rework

Maximum Sensitivity ofDesign Fraction Correct to

Design Undiscovered Rework

Design StaffTotal

Switch to IncludeRework

Time to StartBuild Ramp Up

Design Build/Test

Four Stock Rework Cycle

+

-

73

Design Phase of Work (Build/Test Similar)

Design Startup

Project Staff




Complete


Start DesignRampup

End DesignRampup


Rampup

DesignOriginal

Work to Do

Design WorkDone

DesignUndiscovered

Rework



Work


Design Staff

DesignPlanned Staff




Design ReworkBeing

Accomplishmed


DesignRework to Do



Rework



Rework






Tasks Available




Fraction Correct


Undiscovered Rework



Design StaffTotal


Design Staff Allocation

of Staff Based on Backlogs, etc.

Rework Discovery

Errors Create Errors Feedback

+

- Phase Interconnections

74

Design Startup

Project Staff


Build/TestStartup

Build/Test Delay inDiscovering Rework



Build/Test Effect of DesignUndiscovered Rework On

Fraction Correct


Complete


Sensitivity of Build/TestFraction Correct to Design

Undiscovered Rework

Normal Build/TestFraction Correct and

Complete

Build/Test WorkBelieved to Be Done

Start DesignRampup

End DesignRampup


Rampup

Duration of BuildRampup

DesignOriginal

Work to Do

Design WorkDone

DesignUndiscovered

Rework



Work


Design Staff

DesignPlanned Staff




Design ReworkBeing

Accomplishmed


DesignRework to Do



Rework



Rework






Tasks Available


Build/TestOriginal

Work to Do

Build/TestWork Done


Rework

Build/Test OriginalWork Done Correctly


Work


Build/Test Staff

Project FinishedSwitch

Build/TestPlanned Staff

Build/Test MinimumTime to Finish a Task

Build/Test Fractionof Effort to Rework

Build/Test OriginalWork Being

Accomplishmed

Build/Test ReworkBeing Accomplishmed

Build/Test Maximum WorkRate Based on OriginalWork Tasks Available




Staff on Rework

Build/TestProductivity on

Rework

Build/Test RelativeEffort Required for

ReworkBuild/Test Staff on

Original Work

Build/Test ReworkDone Correctly

Build/TestIndicated Staff

Build/Test Maximum WorkRate Based on Rework

Tasks Available

Build/Test Priorityto Original Work

Build/Test Effect of BuildUndiscovered Rework on

Fraction Correct

Build/Test Fraction of WorkBelieved Done Correct and

Complete

Sensitivity of Build/TestFraction Correct to BuildUndiscovered Rework


Fraction Correct



Fraction Correct


Undiscovered Rework

Maximum Sensitivity ofBuild/Test FCC to Design

Undiscovered Rework

Maximum Sensitivity ofBuild/Test Fraction Correct toBuild Undiscovered Rework



Design StaffTotal


Time to StartBuild Ramp Up

Design Build/Test

Build/Test Progress On Design Rework Discovery

Design Errors (Rework) on Build/Test Fraction Correct

Design Rework on Build/Test Productivity

MIT OpenCourseWarehttp://ocw.mit.edu

ESD. 6\VWHP3URMHFW0DQDJHPHQWFall 2012

For inforation about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.

Documents

System Dynamics - Model - The Rework Cycle