larry leach 813 240-5534 lleach@advanced-projects 2/8/99

Rev. 2: 2/8/99 1

Larry Leach813 240-5534

lleach@Advanced-Projects.com2/8/99

Rev. 2: 2/8/99 2

• Explain why the critical chain, not the critical path, is often the constraint of a project

• Exploit the constraint (critical chain) of a project• Subordinate everything else to the project goal• Define buffers used in Critical Chain Project

Management• Exploit the multi-project constraint

Objectives

Rev. 2: 2/8/99 3

• 100 % project success rate (scope, schedule, cost)

• Reduced project duration (1/2 or more)• Increased project Throughput with no

resource increase

Critical Chain Users Report:

Rev. 2: 2/8/99 4

•Reduced manager and worker stress

•Minimal investment

•Rapid results

Critical Chain Users Report:

Rev. 2: 2/8/99 5

Successful Users & Clients Include

• Harris Semiconductor

• Lucent Technologies

• Honeywell DAS

• Balfour Beatty

• Israeli Aircraft Company

• Better On-line Solutions

• Saturn Development Corp

Rev. 2: 2/8/99 6

TQM

PMBOK TOCCCPM

Rev. 2: 2/8/99 7

Critical Chain PMBOK Links

ProjectManagement

1Integration

2Scope

3Time

4Cost

5Quality

6Human Resources

7Communications

8Risk

9Procurement

*Critical chain impacts shaded blocks.

Rev. 2: 2/8/99 8

Deming’s System of Profound Knowledge

System

Variation

PsychologyKnowledge

Rev. 2: 2/8/99 9

Deming’s Production System

Products(or Projects)

toCustomers

Raw Materialfrom

Suppliers ProductionFunctions

Customer Feedback

Improve

System

Rev. 2: 2/8/99 10

All Processes Have Variation and Uncertainty

• Common cause variation: variation within the capability of the process

• Special cause variation: variation due to influences outside the process or assignable causes within the process

Rev. 2: 2/8/99 11

Two Mistakes Are Inevitable!

1. Treat common-cause variation as if it were special cause variation.

2. Treat special cause variation as if it were common cause.

It is impossible to eliminate both mistakes.Both mistakes increase variation!

Rev. 2: 2/8/99 12

The Deming/Nelson Funnel Experiment

Rev. 2: 2/8/99 13

Results of Test 1: Do not move the funnel!

Dont Move Funnel

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

Rev. 2: 2/8/99 14

Test 2: Move the funnel by the amount that the ball missed the target.

Funnel location before ball drop

Y

X

Move the funnel to here

Example:•Adjusting gun sights (the funnel) by measuring the deviation of each shot on the target.•Machine auto-adjust

X

Y

Locationof ball after drop

Rev. 2: 2/8/99 15

Test 2 Results: Always worse…treating common cause variation as if it were special cause!

2: Move Relative to Last Position

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

Rev. 2: 2/8/99 16

Test 3: Adjust the funnel by the amount that the ball missed the target; but first move it back to the origin.

Funnel location before ball drop

Y

X

Move the funnel to here

Example: •Periodic calibration•Nuclear proliferation

X

Y

Locationof ball after drop

Rev. 2: 2/8/99 17

Test 3 Results: Move the funnel relative to the target. Worse yet.

3: Move Relative to Zero/Zero

-5

-4

-3

-2

-1

0

1

2

3

4

5

-5 -3 -1 1 3 5

(Note scale increase!)

Rev. 2: 2/8/99 18

Test 4: Place the funnel over the last place the ball dropped.

Funnel location before ball drop

Y

X

Move the funnel to here

Example: •Worker training worker•Estimate activity duration based on last time•This years budget based on last years•Legal system: precedent

X

Y

Locationof ball after drop

Rev. 2: 2/8/99 19

Test 4 Results: Place the funnel over the last place the ball landed. Better to take two aspirins and call in the morning.

4: Place over last position

-5

-4

-3

-2

-1

0

1

2

3

4

5

-5 -3 -1 1 3 5

(Note scale increase!)

Rev. 2: 2/8/99 20

Mistake 1: Treating common cause variation as if it were special cause

• Changing the critical path

• Taking action (e.g. expediting, OT) based on small schedule and cost variances (5%-10%)

• Processing project change actions based on small schedule or cost variances

Rev. 2: 2/8/99 21

Mistake 2: Treating special cause variation as if it were common cause.

• Including special cause events in the contingency analysis (or PERT or Monte Carlo type simulations)

• Increasing schedule estimates to respond to delays in another project (where the cause of the delay could be removed through root cause elimination)

Rev. 2: 2/8/99 22

Rev. 2: 2/8/99 23

Resource Leveled Critical (?) Path

E CB

A

6. Feeding Buffer

4. Project Buffer

A B C

C D

2. Critical Chain

1. Resource conflictsremoved.

3. Reduced Task Times

RB RB RB RB

5. Resource Buffers

7. ‘Late’ start

Rev. 2: 2/8/99 24

Goldratt’s Theory of Constraints

Products(or Projects)

toCustomers

Raw Materialfrom

Suppliers ProductionFunctions

System Throughput Limitedby a Constraint

Rev. 2: 2/8/99 25

Theory of Constraints (TOC)

• IDENTIFY the constraint

• EXPLOIT the constraint

• SUBORDINATE everything else to the constraint

• Only then, ELEVATE the constraint

• Do not let mental INERTIA stop you here…do it again!

Rev. 2: 2/8/99 26

Reference (Deterministic) Critical Path Project Plan

A B C

E CB

A C D

Resource Leveled Critical (?) Path

A B C

E CB

A C D

Critical Path Resource

Time

Rev. 2: 2/8/99 27

IDENTIFY Constraint:• Longest project path• Includes resource constraint• Never changes (reduce mistake 1!)

A B C

C D

System

Rev. 2: 2/8/99 28

• Project duration too long*

• Not enough of the right resources*

• Schedule over-run*

• Budget over-run

• Under scope

• Too many changes

Rev. 2: 2/8/99 29

Prepare a Task Duration Estimate

• Write down the time, in minutes.

• Leave this room, purchase specified item, and return it to this room.

• Specification: 4 inch three ring binder, white, with clear cover, binding, and back (slip-in), and slip-in pockets inside.

Rev. 2: 2/8/99 30

How do you feel about your estimate?

• Do you feel your estimate is pretty accurate?

• Do you feel that this task is simpler than most of your preventive and corrective action tasks?

• Do you feel your ability to estimate this task is as good, or better, than your ability to estimate your project task times?

Rev. 2: 2/8/99 31

Example Group Result

Histogram

0

1

2

3

4

5

6

5 10 20 30 40 50 60 70 80 MoreBin

Fre

qu

en

cy

.00%

20.00%

40.00%

60.00%

80.00%

100.00%

120.00%FrequencyCumulative %

Rev. 2: 2/8/99 32

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

Activity Time or Cost

P

All Project Activity Times Are Uncertain Estimates

Minimum Time/Cost

Most Likely Time/Cost

Long ‘tail’ means no definite upperlimit

Rev. 2: 2/8/99 33

How do you define contingency?

Rev. 2: 2/8/99 34

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

Activity Time or Cost

P

Contingency Definition Must Specify Base

Contingency?

Rev. 2: 2/8/99 35

Rev. 2: 2/8/99 36

Focus On Milestones And Variances Drives Low-risk (90%+) Activity Duration Estimates

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

ScheduledDuration

CumulativeCompletionProbability

Rev. 2: 2/8/99 37

CDeliver projectsin shortest time

ASuccessful

career

BKeep my

commitments

DInclude contingency in

my estimates

D’Do not include

contingency in my estimates

Current situation:Resources focus on D while

management pushes for D’.

Rev. 2: 2/8/99 38

Solution Direction:Align individual success (policy,

measures) with the needs forproject success by effectivemanagement of uncertainty.

Feasibility: Many projects completein one half or less the previous time, all of the time, with people reportinggreater satisfaction.

CDeliver projectsin shortest time

ASuccessful

career

BKeep my

commitments

DInclude contingency in

my estimates

D’Do not include

contingency in my estimates

Because: The ‘only way’ to complete the project on timeis to plan for and complete each task on time.

Rev. 2: 2/8/99 39

Critical Path Schedule Hides Contingency In Each Activity

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

Activity 1 Activity 2 Activity 3 Activity 4 Activity 5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

Rev. 2: 2/8/99 40

Date-driven Human Behavior Uses Scheduled Activity Time

0

1

2

3

4

5

6

0 0.5 1 1.5 2 2.5Activity Time or Cost

P

1 = Plan Time or Cost

Rev. 2: 2/8/99 41

IDENTIFY the Project Constraint

• Uncertainty:– Variation in estimate– Variation in task performance

• Dependent events: resources

Rev. 2: 2/8/99 42

EXPLOIT variation by taking contingency out of each task, and moving it to the end of the chain.

Task 1 Task 2 Task 3 Task 4

Task 1 Task 2 Task 3 Task 4

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.5 1 1.5 2 2.5

•Combining variances•Central limit theorem

Rev. 2: 2/8/99 43

(Smaller) aggregated Buffer at end of chain (Exploit)

Critical chain differs from critical path by:• Resolving resource contentions first (Identify)• Using 50 % probable activity times (Exploit)

A B C

C D

Rev. 2: 2/8/99 44

Buffer Provides AnticipatoryMeasure to Exploit Constraint

1 2 3 4 5 BUFFER

WatchWatch PlanPlan ActAct

And, eliminates many type 1 mistakes!

Rev. 2: 2/8/99 45

Buffer Tracking Predicts Action Need

Time0

1/3

2/3

1

xx

x

xx

xx

xxx

Plan

Act

Project Buffer

Rev. 2: 2/8/99 46

Subordinate Merging Paths With ‘Feeding Buffers’

FB

FB

Project Buffer

Isolates the critical chain from common cause variation in feeding paths!

Rev. 2: 2/8/99 47

Exploit With Roadrunner Task Performance

•Start as soon as input is available•Work 100% on the project task•Turn in work as soon as it is complete

Rev. 2: 2/8/99 48

• Critical chain task gets priority over non critical chain task

• Priority to task with buffer in greatest jeopardy for tasks on like chains

• Non-project work lowest priorityPsychology

Exploit With Buffer Management

Rev. 2: 2/8/99 49

Subordinate to the Critical Chain

• Eliminate start and stop time for each activity (Only start dates for chains, and end of Project Buffer!)

• Late start feeding chains

• No intermediate milestones (But...there is a way to meet client or regulator demands.)

Psychology

Rev. 2: 2/8/99 50

• Information tool• Does not add time to project schedule• Notifies resources and resource managers when they

will be needed on the project• May use incentives for subcontractors• Reduces need to change critical path (chain) due to

common cause variation

Rev. 2: 2/8/99 51

Resource Leveled Critical (?) Path

E CB

A

6. Feeding Buffer

4. Project Buffer

A B C

C D

2. Critical Chain

1. Resource conflictsremoved.

3. Reduced Task Times

RB RB RB RB

5. Resource Buffers

7. ‘Late’ start

Rev. 2: 2/8/99 52

Rev. 2: 2/8/99 53

• Resources must multi-task

• Projects are late

• Quality deteriorates

Rev. 2: 2/8/99 54

The Project Team’s Dilemma:Management Imposes New ‘Priority’ Project

Because?

Successfulemployee

Meet my priorcommitment

Respond to thenew priority

Complete firstproject task

Start newproject task

Rev. 2: 2/8/99 55

Rev. 2: 2/8/99 56

Project ExerciseFour tears vertical

Five tearshorizontal

Rev. 2: 2/8/99 57

Task A Task B Task C

Task A

Task B

Task C

Rev. 2: 2/8/99 58

Consider Unloading Ships...

Five ships arrive. Each requires 5 person-days to unload.Each owner wants his ship unloaded ASAP. You have fivepeople to unload the ships. Simple…assign one to each ship.

Starting each one right away (the sooner you start…)each ship is unloaded on the end of the fifth day.

Rev. 2: 2/8/99 59

Put all five resources on ship 1 the first day, ship 2 thesecond day, etc. Result:Ship New (days) Old (days) Saved 1 1 5 4 2 2 5 3 3 3 5 2 4 4 5 1 5 5 5 0

Nobody loses. Four of five clients done sooner. Cost = $ 0

Rev. 2: 2/8/99 60

• IDENTIFY: Multi-project resource constraint

• EXPLOIT: Prioritize projects (Drum)• SUBORDINATE: Stagger project start• ELEVATE• INERTIA

Rev. 2: 2/8/99 61

Delaying Projects Accelerates Completion!

BUFFER

Individual Project Schedules (Note resource contention)

Synchronized Multiple Projects Accelerate and ReduceResource Contention

BUFFER

Synchronizing (Constraint) Buffer

Constraint Resource

BUFFER

BUFFER

Rev. 2: 2/8/99 62

Multiple Projects

•Select a Drum resource•Identify project priority•Stagger project start to the drum schedule•Use Capacity Constraint Buffers to separate project starts

Rev. 2: 2/8/99 63

Selecting the Drum Resource.

THE PURPOSE OF IDENTIFYING THE DRUM IS TOSTAGGER THE PROJECTS.

The drum candidates should be highly loaded resources which are expensive or can not be easily elevated.

THERE IS NO DRUM FOR SINGLE PROJECT IMPLEMENTATION!

THERE IS LITTLE DOWNSIDE TO SELECTING THE ‘WRONG’ DRUM. You will stagger the projects nearly correctly, and the ‘real’ drum will assert itself.

Rev. 2: 2/8/99 64

Prioritize the Projects

Prioritize projects based on their contribution to the company goal: for profit making companies, to make money now and in the future. Considerations may include:•Expected Throughput of the project.•Market•Competition•Future client relationship•Company visionTHE PURPOSE OF THIS PRIORITY IS SIMPLY TO STAGGER THE START OF THE PROJECTS.

Rev. 2: 2/8/99 65

Creating A Drum Schedule

1. Identify the drum resource.2. Prioritize the projects.3. Lay out the plan for each project.4. Collect the drum demands.5. Assign the resources according to the priority.6. Schedule the project start as late as possible based on drum availability.7. Schedule project completion downstream from the drum resource start.

Rev. 2: 2/8/99 66

Drum Schedule Example

B B B

C C

A A A

TimeResource available = 2Priority:

1. A2. B3. C

Highest Priority

Lowest Priority

Stack the drum demand for each project, assuming it started today.

ResourceSupply

# of

dru

m r

esou

rce

Rev. 2: 2/8/99 67

Drum Schedule Result

A & B can start immediately.

C start date determined by ‘backing up’ fromthis point: the constraint use date.

B B BC

CA A A

Push the overflow later in time, untilyou can ‘drop it in’ to start with a CCB

CCBDrop into next slot

# of

dru

m r

esou

rce

Rev. 2: 2/8/99 68

Adding A New Project

•Commitments exist for the yellow, green, and blue projects.•The Orange Project comes up.•Management decides it’s priority is higher than blue, and lower than green.•The resource can not be elevated.

Rev. 2: 2/8/99 69

Adding A New Project

B B BC C

A A A

TimeNow

D

CCB

CCB

Earliest available, withoutimpacting A or B.

Scheduled start oforange project

No impact on C

# of

dru

m r

esou

rce

Rev. 2: 2/8/99 70

Management Behavior

• Only create feasible single project plans; with rational buffers and end date determined by the plan

• Select the Drum resource• Prioritize existing projects• Create the drum schedule• Schedule new projects:

– Create the single project plan

– Set the new project in the priority list

– Update the drum schedule

– Schedule the new project

Rev. 2: 2/8/99 71

Management Behavior (Ctd.)

•Enable resources to avoid (bad) multi-tasking•Avoid actions that may encourage date driven behavior•Enable resources to engage in roadrunner task performance•Give positive feedback as long as the resources engage in roadrunner behavior, especially when they over-run 50-50 estimated duration•Use the buffer report to allocate resources•Use the buffer report to make project decisions

Rev. 2: 2/8/99 72

Buffer Summary

• Project Buffer– One per project– Size using duration estimate differences

• Feeding Buffer– One for each non-critical chain– Size using duration estimate differences

• Resource Buffer– Critical chain only– Size to resource need

Rev. 2: 2/8/99 73

Buffer Summary

• Capacity Constraint Buffer– Between projects– Equal to the preceding task

• Drum Buffer– Before first use of CCR– Size as feeding buffer for preceding chain

Rev. 2: 2/8/99 74

• Explain why the critical chain, not the critical path, is often the constraint of a project

• Exploit the constraint (critical chain) of a project• Subordinate everything else to the project goal• Define buffers used in Critical Chain Project

Management• Exploit the multi-project constraint

Objectives

Rev. 2: 2/8/99 75

Implement effective process for:

• Management leadership and behavior

• Client alignment

• Resource behavior

• Buffer Management

• Organization specific obstacles

• Potential unintended consequences

Path ForwardPath Forward