Transcript Using Critical Chain as an Extension to CPM

Critical Chain
as an Extension to C P M
Orlando A. Moreno, PMP
February, 2003
408.656.2498
Orlando Moreno
[email protected]
Overview
•
•
•
•
•
Introduction to Critical Chain
The Use of Buffers for Contingency
Sizing Tasks and Buffers
Simulating Buffers in Microsoft Project
Considerations When Using Critical Chain
Orlando Moreno
[email protected]
Scheduling Issues Addressed by Critical
Chain
• Inherent uncertainty in task duration estimates
• Parkinson’s Law:
Work expands to fill the available time.
• Student Syndrome:
Wait until the last minute to start a task.
Orlando Moreno
[email protected]
Critical Chain Approach to Scheduling
• Account for both resource and precedence
dependencies
• Set task duration for 50% probability of
completing on time
• Add contingency with strategically placed buffers
Orlando Moreno
[email protected]
A Simple Example Will Illustrate the
Use of Buffers
•
•
•
•
Subsystem A requires six tasks (1 - 6)
Subsystem B requires four tasks (7 - 10)
Integration and test task (11)
Subsystem A is on the critical path
Orlando Moreno
[email protected]
Network Diagram
A
1
2
3
4
5
6
11
B
Orlando Moreno
7
8
9
[email protected]
10
End
Critical Chain Adds Buffers
• Feeding Buffer
• Project Buffer
• Resource Buffer
Orlando Moreno
Provides contingency to keep
tasks not on the critical path off
the critical path
Provides contingency for the
entire project
Provides a wakeup call to alert
resources to be ready to work on
critical tasks
[email protected]
Network With Buffers Added
Resource
Buffer
A
1
2
3
4
5
6
11
B
Orlando Moreno
7
8
9
10
Feeding
Buffer
[email protected]
Project
Buffer
End
Putting Feeding Buffers in All Paths
Resource
Buffer
A
1
2
3
4
5
6
Feeding
Buffer
11
B
Orlando Moreno
7
8
9
10
Feeding
Buffer
[email protected]
Project
Buffer
End
Task Sizing
• Create three estimates
– Most Likely
– Optimistic
– Pessimistic
• Calculate mean
• Use mean as an approximation for median
Orlando Moreno
[email protected]
Three Point Estimates with Mean
Mean
Optimistic
Orlando Moreno
Most Likely
[email protected]
Pessimistic
Mean as a Reasonable
Approximation to Median
Optimistic
Most
Likely
Pessimistic
5
5
5
5
10
10
10
10
15
20
25
35
Orlando Moreno
Beta
Distribution
Mean
Median
10
10
11
11
12
11
13
13
[email protected]
Triangle
Distribution
Mean
Median
10
10
12
11
13
13
17
16
Buffer Sizing
• Contingency based on standard deviation of total
path
• Setting buffer to one standard deviation increases
probability of completing on-time to 84%
• std .dev. path 
Orlando Moreno
2


std
.
dev
.
task

[email protected]
Example Path Calculations
Activity Name
Optimistic Most Likely Pessimistic
Initial draft
Gather information
Write sections
Review informally
Total:
Orlando Moreno
8
7
2
9
10
3
22
[email protected]
16
20
6
Mean
Std. Variance
Dev.
11.0
12.3
3.7
1.8
2.8
.8
3.2
7.7
.7
27
3.4
11.6
Simulating Buffers in Microsoft Project
• Add a task for the buffer at the end of the path
• Add a milestone after the buffer
...
Task
Task
Buffer
Milestone
• Set buffer duration based on calculations
Orlando Moreno
[email protected]
Simulating Buffers in Microsoft Project
...
Task
Task
Buffer
Milestone
• Constrain the task type of the milestone to:
Must Finish On
• Change the buffer to a milestone
• Constrain the buffer task type to:
As Soon As Possible.
• The slack for the buffer will now be the buffer size
Orlando Moreno
[email protected]
Simulating Buffers in Microsoft Project
Orlando Moreno
[email protected]
Considerations When Using
Critical Chain
Orlando Moreno
[email protected]
Statistical analysis assumes many
tasks with small random variations
Open ended tasks - the pessimistic estimate is
significantly greater than the most likely - require special
focus
– Identify and include in risk management plan
– Consider changes to cost and scope to reduce
uncertainty
– Consider an iterative development strategy
– Estimate range of possible completion dates
Orlando Moreno
[email protected]
Statistical analysis assumes paths are not
coupled
• Interim deliverables may be needed by other
developers,
• Multiple strongly coupled paths could impact
the statistical calculations
• Identify coupling and add to risk management
plan
• Consider Monte Carlo simulation when there
is a high degree of coupling
Orlando Moreno
[email protected]
Getting the most benefit from 3-point
estimates
• Create 3-point estimates in early top level
schedule
• Identify all assumptions about most likely,
optimistic and pessimistic estimates
• Assess cost and scope impact of these
assumptions
• Include most pessimistic completion estimate,
cost and scope in all negotiations about
schedule
Orlando Moreno
[email protected]