Transcript Critical Chain An application of the Theory Of Constraints

File:CCPGS05
CC present Page: 1
ETXPGS 2000-09-14
How do we manage uncertainty in projects?
12
36
6
12
4
22
1. To know when the project can be finished we estimate the individual
task times and add them up in a project schedule.
-> We follow the additive rule
2. We follow the progress of the activities and take action when
problems occur.
-> How do we estimate the duration of an activity?
File:CCPGS05
CC present Page: 2
ETXPGS 2000-09-14
How do we estimate the duration of an activity ?
“I also do
other things”
Probability
100%
concentration
median time with 100%
concentration
File:CCPGS05
CC present Page: 3
ETXPGS 2000-09-14
A realistic
estimate (ca 90%)
We build in safety that protect each activity
12
36
6
12
4
22
 But why are so many activities still late?
 That should indicate that our estimates do not have safety, it
indicates they are overoptimistic - Right?
 Let´s look at the effect of high probability estimates.
File:CCPGS05
CC present Page: 4
ETXPGS 2000-09-14
Effects of high probability time estimates...
1. Unreported Early Finishes
Activity A
10 days
Activity B
10 days
Finishing early but not reporting until the scheduled completion - for several reasons, e.g:
...The next resource isn’t ready anyway!
...It’s not quality if it’s finished
the time is up.
One effectbefore
is:
...Work expands to fill Finishing
its time.early but not reporting
Probabili
ty
- What’s the rush?
- The “3 Minute Egg Rule”
- Parkinson’s Law
until the scheduled completion.
Parkinson’s Law and The 3 Minute Egg Rule - Two effects from the same root:
...the lack of good criteria defining when a task is complete.
5%
(High risk)
File:CCPGS05
CC present Page: 5
ETXPGS 2000-09-14
50%
(Medium risk)
90%
(Low risk)
2. The student syndrome
1. We have a margin until we reach the estimated completion
time for the activity.
Activity intensity
2. We have other urgent issues to deal with.
Murphy
Planned
start
File:CCPGS05
CC present Page: 6
ETXPGS 2000-09-14
Planned
completion
With more Complexity - Priorities become Unclear...
It’s not always possible to
complete one task until it is
required to shift to another
task (especially true for
limited key resources).
Many things to do at
the same time
Unclear Priorities
Current mode of operation: (Multi Tasking)
Single tasking:
File:CCPGS05
CC present Page: 7
ETXPGS 2000-09-14
More
Complexity
Summary:
 The additive rule does not work in projects
 We add safety in our time estimates to protect for
uncertainty and the effects of variability and
dependencies
 We consume safety time
File:CCPGS05
CC present Page: 8
ETXPGS 2000-09-14

Not taking advantage of positive variation (error reporting)

Student Syndrome

Multi-tasking
Developing the Critical Chain Single Project Solution...
D
D
12
Critical Chain
D
D
36
6
D
D
4
D
D
D
D
1. IDENTIFY the System’s Constraint(s).
D
D
12
22
DD = Due Date
What is the constraint of a project?
The longest chain of all dependent events...
- the Critical Chain!
File:CCPGS05
CC present Page: 9
ETXPGS 2000-09-14
2. Decide How to EXPLOIT the System’s
Constraint(s).
3. SUBORDINATE Everything Else to the
Above Decision.
4. ELEVATE the System’s Constraint(s).
CC Four Step Systems
Improvement Process
Project Buffer - Protecting the Project Completion
D
D
612
D
D
18
Critical
CriticalChain
Chain
4
22
3 36
6
D
D
4
D
D
6
1. IDENTIFY the System’s Constraint(s).
D
D
D
PB (33
days) 12
D
22
2. Decide How to EXPLOIT the System’s
Constraint(s).
3. SUBORDINATE Everything Else to the
Above Decision.
4. ELEVATE the System’s Constraint(s).
CC Four Step Systems
Improvement Process
Put all the safety in the Critical Chain
tasks in an aggregated Project Buffer!
File:CCPGS05
CC present Page: 10
ETXPGS 2000-09-14
Feeding Buffers - Protecting the Critical Chain
1. IDENTIFY the System’s Constraint(s).
6
18
3
Critical Chain
D
D
6
PB (33 days)
24 11 22
FB (13)
2. Decide How to EXPLOIT the System’s
Constraint(s).
3. SUBORDINATE Everything Else to the
Above Decision.
4. ELEVATE the System’s Constraint(s).
CC Four Step Systems
Improvement Process
Put all the safety in the non-critical feeding
paths tasks in an aggregated Feeding
Buffer!
File:CCPGS05
CC present Page: 11
ETXPGS 2000-09-14
Relay Runner
1. IDENTIFY the System’s Constraint(s).
6
18
3
Critical Chain
2 11
FB (13)
DD
6
PB (33 days)
2. Decide How to EXPLOIT the System’s
Constraint(s).
3. SUBORDINATE Everything Else to the
Above Decision.
4. ELEVATE the System’s Constraint(s).
We also need to make it possible for people to act in an appropriate way
that supports task times with (much) less safety in them - Relay Runner!
File:CCPGS05
CC present Page: 12
ETXPGS 2000-09-14
Reduced Buffer Size...
1. IDENTIFY the System’s Constraint(s).
6
18
3
Critical Chain
DD
days)(33 days)
6 PB (17PB
2 11 FB FB
2 11
(13)
DD
DD
2. Decide How to EXPLOIT the System’s
Constraint(s).
3. SUBORDINATE Everything Else to the
Above Decision.
4. ELEVATE the System’s Constraint(s).
From statistics we know:
“The overall project estimate has much less variability than each of
the individual task estimates that have been added together to create
it”.
File:CCPGS05
CC present Page: 13
ETXPGS 2000-09-14
Subordinate everything else...
Minimise the wasting of safety times:
 Accept that task times have a natural variation
 Use short but not impossible duration times (50/50)
 Protect the critical chain with feeding buffers, do not start
activities earlier than that in order to avoid multi-tasking.
 Do not use fixed intermediate milestones for controlling the
project. The only fixed point is the Project due date.
 Resources finish the activity they are working on unless our
control mechanism tells us otherwise, i.e no bad multi-tasking.
File:CCPGS05
CC present Page: 14
ETXPGS 2000-09-14
The Control Mechanism - Buffer Management
We no longer track progress against
Milestones - But we still need to know
how we are doing!
1. Report estimates of remaining duration
periodically (daily, bi-weekly, weekly)
+ when any task is started/completed.
2. Update penetration into buffers.
=
+
+
+
3. Base task priority decisions and corrective actions on buffer penetration data.
=
+
Project Buffer
4. Do not re-plan, unless there is a major
change in scope, dead-line etc
Feeding Buffer
Region III Region II
OK
File:CCPGS05
CC present Page: 15
ETXPGS 2000-09-14
Watch
& Plan
Region I
Act
“Life Before Critical Chain”
”We are ready when we
are ready!”
 Tight time plan
 Insufficient resources

From start

Delays in other projects
 Late input
 Input with unknown and varying
quality
 Unclear when we are ready
 Much re-work
File:CCPGS05
CC present Page: 16
ETXPGS 2000-09-14
“Life After Critical Chain”
”We are ready when
exit criteria met”
File:CCPGS05
CC present Page: 17
ETXPGS 2000-09-14
 Resources available when
needed
 Avoid Multitasking
 Use buffer penetration as
indication for action
 All input in place
 Less stress
 Less re-planning
 Less rework
 Define Exit Criteria
 More clear priority between
projects
 More easy to follow
progress
And remember:
 All risks are not eliminated!

You should be more immune to “common cause variation”

You still have to manage “special cause variation”
 Stick to the plan, keep the momentum up

It’s easy to fall back in traditional behaviour
 Question all intermediate milestones, and avoid
committing to dates for them
File:CCPGS05
CC present Page: 18
ETXPGS 2000-09-14