EM 540 Operations Research

1

• EM 540 Operations Research/
• DecS 581 Operations Management

Why are Projects Invariably Late?The
Problems (Supplement to Chapter 13)
2
Intro to Critical Chain Project Management

• Projects Are
• Unique
• Dependent on Precedence
• Activities Not Well Known
• Highly Variable
• Share Resources
• Concurrent with Other Projects
• Valued by Scope, Schedule and Cost

3
Undesirable Effects of Projects

• Projects Are
• Usually Late
• Have Too Many Changes
• Often Over Budget
• Lots of Rework
• Many Priority Battles
• Resources Not Available When Needed
• Jeopardize Scope for Cost or Schedule

4
Projects are Balancing Acts
5
Then things Combine
Quality andScope
Timing andSchedule
BudgetedCosts
Precedence Structure
Statistical Variation
HumanBehavior
6
And Reality Sets In
Quality andScope
Timing andSchedule
BudgetedCosts
Bumpy Road of Reality
7
Due Date Problems
Thanks to Rees Furbeck who prepared some of
these graphics
8
Change Problems
9
Rework Problems
10
The Core Problem (Constraint)
Compensate for
Do whatever it takes
early mis-
to meet an
estimations /
endangered original
mis-calculations
commitment
Meet original
commitments
Not compensate
Not jeopardize any
for early mis-
other original
estimations /
commitment
mis-calculations
11
How To Understand Project Problems

• It is hard to examine one project and find a
  pattern to the problems.
• There are too many excuses.
• There is so much variability in tasks that the
  solution (problem) changes with every look.
• Its better if we could examine the same project
  executed over and over again to look for trends.
• Simulation can do this
• Thru-Put Critical Chain Simulator!

12
Understanding Variability
Before we go to the Simulator
What is 8 times 8?
64? Are you sure?
8? Or 81?
What is 8?
So, what is 81 times 81?
Somewhere between 49 and 81! With 64 being the
most likely.
13
Good Statistics
Variability Adds as the Square Root of the Sum of
the Squares (Central Limit Theorem Variability
reduces when combined)



Watch Simulation 01
The Thru-Put Simulator material is provided with
permission of Eli Goldratt. Thru-Put is now
Realization _at_ www.realization.com
14
Bad Statistics
Assembly If one is late, they are all late.



Watch Simulation 02
15
Resource Conflict
Watch Simulation 04
16
Look at a real projectSim 5
Commitment 70 Days
Watch Simulation 05
17
One Simulation Run
Commitment 70 Days
Actual72 Days
18
Results of 1000 Runs
Commitment 70 Days
19
Sim 05 Treats Tasks as Normal Distributions
Task Duration Distribution for Activity
A1 Theoretical and Actual Simulation Data
20
But, Are Project Tasks Normal Activities?
How long did it take you to drive to work this
morning? What is the least time it has ever
taken? What is the most time it has ever
taken? What is the average time to drive to
work? Is the average closer to the shortest? Is
the average closer to the longest?
21
Skewed Distribution for Sim 6a
Task Duration Distribution for Activity
A1 Theoretical and Actual Simulation Data
22
Results of 1000 Simulations 6a (skewed)
Scheduled _at_50 Probability
Commitment 70 Days
23
Try Adding Safety time to each Task Scheduling 11
Days per Task
Scheduled _at_60 Probability
Commitment 77 Days
24
Maybe 13 Days per Task would do it?
Scheduled at75 Probability
Commitment 91 Days
25
Try giving 16 days per Task as a big safety
cushion
Scheduled _at_ 85 Probability
Commitment 112 Days
26
Getting Better at Hitting our Projections, But...
Rees Furbecksclever graphic
27
Engineering Optimism
Question If you have 16 days to to a 10 day
project, when do you start?
Immediately! Or, After 6 days. Or, After 10
days (since you know you are faster than average
and can probably do it in 6 days).
28
I hate Student Syndrome!
It wears People out!And, they think its their
own fault!
Self Imposed Overtime
Level of Effort
Normal level of effort
Assigned
Due Date
29
There is another Problem
Question You fought tooth and nail to get the
16 days you wanted to do the 10 day project. If
you finish in 14 days, will you go around
advertising early completion? What if you finish
in 10 days? How about 8 days? Wonder of wonders 6
days?
This is called Erroneous Reporting of Completion
time! A second similar rule Parkinsons
Law--Work expands to fill the time available
30
With 75 Erroneous Reporting
Sim 06, Tasks Scheduled _at_ 16 days, only 25
report early completion
Appearance Boy are we Good! We made our Due
Dates!
31
Now Add Student Syndrome
Sim 06, Tasks Scheduled _at_ 18 days (85
confidence), only 25 report early completion,
and late starting (Student Syndrome and
Parkinsons Law)
Expected Completion718 day126 Days
32
Student Syndrome and 75 Erroneous Reporting
Distribution used to simulate individual task
duration. Only 25 of the time were true times
reported.
33
Commitment 126 Days
(Scheduled 18 daysfor each Task)
85 Probability
34
Increasing Safety to 19 days
Commitment 133 Days
(Scheduled 19 daysfor each Task)
35
Huge Safety -- 22 Days
Commitment 154 Days
(Scheduled 22 daysfor each Task)
36
A 90 Estimate of this 70 Day Project is 154
Days
37
Consider Three Individual 70 Day Projects in
Parallel!
Simulation 9aMulti-ProjectStudent Syndrome75
Erroneous Reporting,90 EstimateResources Shift
and stay until complete (one task delays
another).
38
One Project Delays Another
90 Probability
Commitment 154 Days
Project 1 as first Priority Scheduled 154Task
Duration 22 Days
39
Rather than delaying, try Alternating
My Assignments
Task 1
Task 2
Task 3
10 days
10 days
10 days
40
I Hate Multi-Tasking Even More!
41
Painful Results!