1. Project Management
Critical Chain Method

2. PM: A Common Problem Meeting all specifications & other requirements
Within budget
On time

3. PM: Overspending (PM Discipline established in the 1950s)
Cost overrun of 111 transport projects (constant prices)
Adapted from Flyvbjerg. See Chapter 8

4. PM: A Common Problem
Why is it so difficult to manage projects to deliver on time, within budget and meet the specifications?
Why do operations mangers perform better than project managers?
Uncertainty

5. PM: A Proposition / Hypothesis (Tentative Statement)
Projects are (by definition) unique endeavours 
Projects always involve unknowns
Projects are high-risk endeavours
The risks cause projects to be late and to overspent

6. PM:Is The Problem Really Uncertainty?
If uncertainty is the source of the problem, then all high-risk projects (projects with much uncertainty) should be late and over budget
But, there are high-risk projects that were completed well within budget and schedule:
Notes:

7. PM: Is The Problem Really Uncertainty?
U2 Reconnaissance Airplane
Notes:

8. PM: Is The Problem Really Uncertainty?
U2 Reconnaissance Airplane (1955):
Met all specifications
Flew higher than any airplane before (more than 12 miles above the earth) – uncertainties!
Flew over the USSR within 8 months of project approval
Spent less than budgeted
Notes:

9. PM: Mozal Aluminum Smelter

10. PM: The Mozal Smelter - a successful project that had a high level of uncertainty
The state-of-the-art smelter (that uses as much electricity as a whole city) has been erected in Mozambique, one of the poorest countries in the world …
Risks:
Stability of government
Legal & commercial
framework
Language
Security
Medical facilities & health
Weather
Lack of infrastructure
Logistics e.g. customs clearance
Labour
Work ethic
Skills
Unions
Health
Expatriate conditions
Notes:

11. PM: Alternative hypothesis …
Cases such as the U2 and Mozal Smelter indicate that uncertainty is not the cause of our three problems.
This hypothesis is wrong!
What could explain the three problems and the success of projects such as the Mozal Smelter?
Perhaps it has to do with the way we manage under conditions of uncertainty!
Notes:

12. PM: A Common Problem It’s an age-old problem in several industries …
Yet, the essential principles of managing contingency reserves are often sadly neglected

13. PM: A Possible Solution
The Critical Chain Method claims to offer a solution to managing contingency reserves
Various aspects of the solution have been known for a long time
These aspects were put together during and after a PhD project
Popularized in a business novel in 1997

14. PM: A Possible Solution

15. Theory of Constraints (TOC)
The critical chain method is a Theory of Constraints (TOC) application to projects
Project duration is considered as the constraint:
The objective of a project is to deliver something that would generate income (or provide some other benefit)
The project itself costs money
The sooner the income (or other benefit) can materialize, the better

16. TOC: Contingency Reserves – Do We Need them?
An activity such as driving to some destination

17. TOC: Contingency Reserves – Do We Need them?
If you have to make a commitment about duration, how much time will you allow?
And if there is a heavy penalty if it takes longer?

18. TOC: Contingency Reserves – Do We Need them?
If you have to make a commitment about duration, you need to build in a contingency reserve
The project manager always has to make a commitment about the duration of the project
A common idea is that everyone responsible for an activity should also make a commitment

19. Contingency Reserves – Do We Need them?
In general, substantial reserves are built into project schedules (and budgets)
Finishing late implies criticism and even penalties
If we expect that the estimate will be cut by seniors, we build in even more reserve
Contingency reserves are built in at more than one level of the WBS

20. TOC: Contingency reserves at more than one WBS level
How much reserve in total?
(It works like compound interest)
= 2.01
The reserve is doubled

21. TOC: Contingency Reserves – Need to be Managed
Contingency reserves are essential
Contingency reserves are of significant size
Contingency reserves need to be managed

22. TOC: Contingency Reserves – Need to be Managed
Conclusion: In general significant amounts of contingency reserves are built into project schedules and budgets

23. TOC: Contingency Reserves – Why Still Late?
If it is true that there is so much reserve built into project schedules and budgets, why do so may projects still go over budget and over the due date?
Let us explore human behaviour during project execution

24. TOC: Reasons for not Finishing Early
No incentive for early finish
Keep on improving the work:
Enjoy the work
Reduce risk of poor deliverable
Often leads to adding unnecessary “bells & whistles”
Argued for long duration - reporting early finish could jeopardise credibility

25. TOC: Reasons for not Finishing Early
Money on budgets are sometimes spent merely because we have an approved budget
Likewise, work sometimes expands to fill approved schedules

26. TOC: Not Finishing Early -- Variability and Deadlines
Most likely duration
Possible duration

27. TOC: The “Student’s Syndrome”
Deadline
Effort
Most likely duration

28. The first law of hydraulics …..
No pressure - no flow!

29. TOC: Reserves Get Wasted
“Student syndrome” is the first way that we waste reserves
A second way is by jumping from one task to another
When one critical task is late, it delays the whole project, but when a task finishes early, it has no effect
Let us see how 2 and 3 above work …

30. TOC: Reserves Get Wasted --Multitasking
Notes:
Setup time refers to the time you spend to get ready to do a task. Setup time is a common term in manufacturing operations where machine tools such as lathes or milling machines need to be set up before the manufacturing operation can begin. Most tasks involve “setup” or an initial period where progress is slow.
Multitasking cause activities that succeed X and Y to start later

31. Without multi-tasking
TOC: Reserves Get Wasted--Multitasking
Without multi-tasking
With multi-tasking
Progress
Time
Progress
Time

32. TOC: Reserves Get Wasted -- Delays Add Up but Finishing Early Does not Help
End D E

33. TOC: Reserves Get Wasted -- Delays Add Up but Finishing Early Does not Help
Even where tasks are being done in series (not in parallel) gains are not always passed on:
Other resources may not be ready to start earlier than planned (peoples’ diaries, venues and equipment have been booked, …)
Resources are busy with other work
Notes:

34. TOC: Reserves Get Wasted – Early Completion Not Reported
Parkinson’s Law: Work expands to fill the time available
Because the person negotiated to get the time, he/she could be embarrassed if it is now done in much less time. This could lead to loss of credibility and invite pressure
Notes:

35. TOC: Reserves Get Wasted -- Reserves Get Wasted: Summary
We build in a lot of contingency reserve (“fat”) and then it gets wasted in one or more of the following ways:
Student syndrome
Multi-tasking
Delays accumulate, but gains (working faster) don’t
Notes:

36. TOC: Reserves Get Wasted -- Potential for Improvement
There are substantial reserves in project schedules
Reserves should be provided at project level only
Even if reserves are only built in at the lowest level, the principle of aggregation offers potential for significant reduction of the reserves

37. TOC: Reserves Get Wasted -- Potential for Improvement
The Principle of Aggregation
The basis of, for example the insurance industry
Sometimes correctly applied in project cost management
Since 1997 increasingly being applied to project scheduling (as a result of the “Critical Chain” methodology)

38. TOC: Reserves Get Wasted -- Aggregation
The variance of the sum equals the sum of the variances
V = V1 + V2 + … + Vn
If V1 = V2 = … = Vn = V, then V = n V
 indicates risk
2 = V Therefore:  = (n)1/2∙
In the absence of aggregation:  = n∙
 aggregated risk is significantly smaller than the sum of the individual risks

39. TOC: Reserves Get Wasted -- Aggregation
To aggregate project reserves:
 Make provision only at project level and nowhere else

40. TOC: Reserves Get Wasted -- Aggregation
Previous example: 15% reserve required for a specific activity
Less than 15% required for that activity if provided at project level

41. TOC: Reserves Get Wasted -- Aggregation
Project managers sometimes take direct control of contingency reserves in budgets
Until the advent of critical chain, few project managers utilized the same principle when scheduling projects
Realization of typical human behavior during planning and execution as well as the effect of aggregation form cornerstones of Critical Chain project scheduling

42. TOC: Reserves Get Wasted -- Aggregate Reserves
Consider the critical path (project constraint):
People responsible for activities normally build in reserves
Activity A
Activity B
Activity C
Remove the reserves from activity level and give it to the project manager: A B
Activity C
Reserve
PM’s
commitment
Team’s
objective

43. TOC: Reserves Get Wasted -- Aggregate Reserves
Contingency reserves at project level is referred to as a “project buffer”
Two reasons why a project buffer can be smaller than the sum of individual reserves:
The principle of aggregation
When a schedule indicates less time, less time is wasted (students’ syndrome is minimized)

44. TOC: Reserves Get Wasted -- Aggregate Reserves
Reduce the size of the Project Buffer
Activity A
Activity B
Activity C A B
Activity C
Project Buffer A B
Activity C
P B
Project duration is reduced
Adapted from Goldratt, Critical Chain

45. TOC: Reserves Get Wasted -- Aggregate Reserves in Practice
Reduce the size of the Project Buffer:
Existing project: cut activity durations
New project: can work on estimates with very little or no contingency reserve

46. TOC: Reserves Get Wasted –Aggregate Reserves
The fact that there is a deadline for the project does not imply a deadline for each activity.

47. TOC: Reserves Get Wasted -- Aggregate Reserves
Implication:
Only the project manager makes commitments on due dates (and project cost) – everybody else only makes realistic estimates without reserves
This normally requires a change in project culture – the only difficult aspect of critical chain management

48. TOC: Reserves Get Wasted -- Project Buffer - Implication
Should people responsible for activities be aware of the project buffer?
Yes, if not, they will tend to build in contingency reserves at activity level as well
They must trust the project manager to “bail them out” in the case of an unforeseen event

49. TOC: Critical Chain versus Critical Path
The critical chain is similar to the critical path, but it takes into account that a resource can do only one activity at a time

50. TOC: Critical Chain versus Critical Path
8 days
Ann B
10 days
John C
5 days
Peter D
5 days
John
Critical path: ABC (23 days)
Critical chain:
ABDC or ADBC (28 days)

51. TOC: The Function of Feeding Buffers
If a non-critical path is late, we do not want it to delay the critical chain
Therefore, we build in feeding buffers
(half sum of non-critical activities durations)

52. TOC: Feeding Buffers
Feeding buffers prevent the project from being delayed if a non-critical activity takes longer than planned.
Adapted from Goldratt, Critical Chain

53. TOC: Critical Chain versus Critical Path
A resource is often needed in two places at the same time. Consider a small project:
A10 B5 C8
Notes:
E10
Project Buffer 17 D8 B8
FB 8
Adapted from training material of Goldratt Institute

54. TOC: Critical Chain versus Critical Path
The critical chain takes into account that a resource can do only one thing at a time.
A10 B5 C8
E10
Project Buffer 17 D8 B8
FB 8
A10 C8 B5 D8 B8
E10
Project Buffer 20
FB5
Critical chain
Notes:
A definition of “critical path”: The series of activities which determine the early completion of a project.
“The critical path will generally change from time to time as activities are completed ahead or fall behind schedule” (P<BOK)
The fact that the critical path changes, makes it difficult to control a project because the project manager does not know what activities are critical. The feeding buffers prevent the critical chain form changing – thus the project plan is much more rigorous and it is easier to control. The project manager is also in a position to predict the project completion date better.
Adapted from training material of Goldratt Institute

55. TOC: Critical Chain versus Critical Path
In traditional project scheduling, the critical path changes because activities often take longer or shorter than the estimated durations.
Feeding buffers ensure that the critical chain remains stable.
With critical chain, the project manager always knows on which activities to focus his/her attention.

56. TOC: Critical Chain versus Critical Path
“The critical path will generally change form time to time as activities are completed ahead or fall behind schedule”
PMI 1996
And it does so without warning! 
You do not know where to focus

57. TOC: Critical Chain Example

58. TOC: Critical Chain Example
The constraint is the path P-Q-R

59. TOC: Critical Chain Example
Remove reserve from activities and give to the project manager:

60. TOC: Buffers Reduced
Reduce buffer size (as motivated earlier)

61. TOC: A Resource Does One Activity at a Time
A resource (the technician) is overloaded

62. TOC: A Resource Does One Activity at a Time
Decide which activity the Technician should do first

63. TOC: The Critical Chain
The path S-T-Q-R-Z is the critical chain
The Critical Chain is similar to the Critical Path but takes into account that a resource can do only one activity at a time

64. TOC: The Critical Chain
Alternative critical chains
As you might wish to verify, the project can be done slightly faster if the Technician does Activity Q before Activity T

65. TOC: Finishing an Activity Early
When a project activity finishes early, successor activities are often not expedited
In the case of relay races, however, a runner starts early if an earlier runner runs faster than planned
For critical activities only: Give a count-down to the start in order to expedite if a predecessor works faster than planned

66. TOC: Early Finishing of an Activity
Resource buffers are early warnings on critical activities (flagging the need for early starts)

67. TOC: Buffers - Summary

68. TOC: Buffers
Buffers provide a practical way to convert a complex stochastic problem into a deterministic one
Feeding buffers stabilize the critical chain – unlike the critical path, the critical chain does not change (unless a feeding buffer is depleted)
The stability ensures better forecasting of end dates
The project manager always knows where to focus her attention
Monitoring of feeding buffers (discussed in Chapter 11) ensures that an early warning is received in the event that the critical chain would change

69. TOC: Buffers
Feeding buffers provide the answer to the question of when to do non-critical activities: as early as possible or as late as possible…
As late as possible, but with an appropriately sized buffer

70. TOC: As Early as Possible or Not
During scheduling and during project execution, non-critical activities start as late as possible but with buffers
During execution, critical activities start as soon as possible (and are expedited by means of resource buffers when predecessors finish early)

71. TOC: The Change Process
To convince people to give estimates without significant reserve built in, everybody in the organisation must understand that the durations on the schedule are merely estimates
This means that only the project manager makes a commitment on due date
There are no due dates (deadlines) on activities

72. TOC: The Change Process
Executives, clients must understand the basics
If they don’t understand it, they may for example want to eliminate the project buffer and insist on delivery on the date at the beginning of the buffer
Executive support is necessary for successful implementation
Working with subcontractors adds another level of complications

73. TOC: Software and Critical Chain
Can you simply insert buffers as “activities” in all project scheduling software and schedule non-critical activities as late as possible?
No. While popular software allows you to schedule activities as late as possible, resource leveling then does not work because resource leveling works on moving activities later (not earlier)

74. TOC: Monitoring buffers
0 to 33% of buffer consumed. OK
Green zone
33 to 67% of buffer consumed.
WATCH
& PLAN
Yellow zone
67% to 100%
of buffer consumed
ACT
Red zone

75. TOC: Monitoring buffers

76. TOC: Software That Supports Critical Chain
MSProject plus Prochain
Sciforma
Concerto

77. Allocating Resources
Different tasks within a project typically rely on the shared resources (equipment and staff)
Different projects within an organization (especially a matrix organization) also share resources
Resources must not have unrealistic workloads
Functional managers prefer more or less uniform workloads on their resources

78. Allocating Resources - Complexity
Say you have to do 10 tasks and you can start with any one. How many possible schedules exist?
10 x 9 x 8 x 7 x 6 x 5 x 4 x 3 x 2
 3.6 million
And only one resource is involved in this example

79. Allocating Resources - Complexity
Even with modern computers, attempts to develop optimal schedules for multiple projects require intolerably large amounts of computing time
The practical way is to use heuristic rules to allocate resources (project scheduling software use such rules)

80. Allocating Resources : Heuristic Rules
Schedule activities as early as possible
Analyze the schedules for resource loading
When a resource is needed at more than one place at the same time, (a resource is overloaded) use a heuristic rule to decide to which activity the resource should be allocated
If one project has a high priority, it makes sense to give preference to that project when allocating resources

81. Allocating Resources -- A Common Heuristic Rule: Least Slack
If an activity is on the critical path, it should get preference when allocating resources
Critical activities have the least slack
Activities on near-critical paths should also have some priority
Least slack rule: Activities with zero slack have priority, then ones with one day slack, and so on

82. Allocating Resources -- A Common Heuristic Rule: Shortest Task Time
Activities with shortest duration get priority
It has motivational value (perception that work is getting done) – but that could be misleading!
Succeeding activities can start early. This reduces the total waiting time:

83. Allocating Resources -- A Common Heuristic Rule: Shortest Task Time
Total waiting time is reduced

84. Allocating Resources -- Several Rules ExistD

85. Allocating Resources -- Exercise
Use the Shortest Task Time rule to schedule the following small project:

86. Allocating Resources -- Exercise (Cont’d)
Then use the Least Slack rule to schedule the same project:
H Steyn (Ed). Project Management – A Multi-disciplinary Approach

87. Allocating Resources : Exercise -Solution
Shortest Task Time Rule

88. Allocating Resources -- Exercise -Solution
Least Slack Rule:
All activities to be performed by John (as well as the work Ann has to do on Section 1) have zero slack
This rule does not indicate with which one John should start
This is called a “tie” between the activities
A secondary rule is needed to break a tie

89. The TOC Method for Multiple Projects
5-step process
Step 1: Identify the constraint / bottleneck
Step 2: Decide how to exploit (utilize) the constraint
Step 3: Subordinate all non-constraints to the decision made in Step 2
Step 4: Elevate the constraint
Step 5: Return to Step 1 to identify new constraint

90. The TOC Method for Multiple Projects
Constraint for individual project: duration
Goal of organization handling multiple projects: maximize flow of projects through the system
Step 1: Identify the constraint
Constraint may be a specific resource that limits the number of projects that can be handled

91. The TOC Method for Multiple Projects
Step 1 (Cont’d):
Constraint for planning and execution sometimes not the same
For planning a set of projects a rule may be used as proxy for the constraint
Example of rule: three projects in execution phase

92. The TOC Method for Multiple Projects
Example of rule for planning:
Three projects in execution phase 1 2 3

93. The TOC Method for Multiple Projects
Step 1 (Cont’d):
Constraint for executing work may be the time that managers have available to spend on monitoring projects
Step 2: Decide how to exploit the constraint
For rule: Three projects in execution phase, insert Capacity Buffers to stagger projects
If management time is constraint during execution, they should not spend time on activities such as attempting to keep all resources busy all the time

94. The TOC Method for Multiple Projects
Step 4: Elevate constraint
This could imply adding additional capacity
For the constraint Three projects in execution phase it could imply additional capacity to increase the number of projects in execution from 3 to 4
As this is costly, it is done only after Step s 2 and 3
Elevate management time: simplify management systems

95. The TOC Method for Multiple Projects
Step 5: Return to Step 1
Adding additional capacity might remove the constraint and a new constraint may emerge
Sometimes taking a new constraint into account could be disruptive and the decision may be made not to take another constraint into account

96. The TOC Method for Multiple Projects
Three rules used by consultancy that implement the TOC method for multiple projects:
During planning, stagger the release of projects
Plan aggressive durations, using project buffers /3 of critical chain length
During execution:
Priorities determined by buffer status (Chapter 11)
Minimize buffer consumption by performing all work as soon as possible
Source: Training material of Realization Technologies Inc.