Schedule Risk Analysis ICEAA 2016 PDF

Schedule Risk Analysis
David T. Hulett, Ph.D.

Hulett & Associates, LLC
ICEAA / 2016
© 2016 Hulett & Associates, LLC 1

Introduction
USAF Approach to Schedule Risk
“A Most Probable Schedule (MPS) will be prepared by
assessing the durations presented in the offeror’s MIPS
(this means estimating the longest, the shortest, and
the most likely duration for each task, activity, event,
and milestone) and preparing a network-based Monte
Carlo simulation in order to determine a schedule that
has a 90% probable completion date.”
Integrated Risk Management Guide,
Aeronautical Systems Center (ASC), draft, 9 April 1994

Purpose of a Risk Analysis
 Promote the language of probability and use of its

mathematics in risk analysis
– Why do schedules overrun? Things do not go “according to plan”
 Examine elements of a project in detail, determining
relationships and formulating a model
 Most people are less able to comprehend the whole of the
problem than risk of the elements individually

Purpose of a Risk Analysis (continued)
 Risk analysis strategy

– Describe the risks at the level of the activity
– Use the schedule and Monte Carlo simulation to find the overall
project schedule risk
 The essence is a statement of the probability of program
outcomes
Source: Risk Assessment Techniques,
Defense Systems Management College 1983

Overrun Risk is Not a New Issue
“Initial cost and schedule estimates for major projects have

invariably been over-optimistic. The risk that cost and
schedule constraints will not be met cannot be determined if
cost and schedule estimates are given in terms of single
points rather than distributions”

Overrun Risk is Not a New Issue
(continued)
“A formal risk analysis is putting on the table those problems

and fears which heretofore were recognized but intentionally
hidden.”
Source: “Final Report,”

US Air Force Academy
Risk Analysis Study Team 1973

Some Reasons for Schedule Risk
 Fundamental uncertainty in the work
 Unrealistic baseline schedule
 Natural, geological causes
 Project complexity
 Scheduling abuses
 Relying on participants outside the organization
 Subcontractor late

Some Reasons for Schedule Risk
(continued)
 Design changes
 Staffing Manufacturing problems
 Contracting problems
 Customer (government) not supportive
 Cannot get subcontractor under contract
William Cashman, “Why Schedules Slip…”
Air Force Institute of Technology (AFIT) Master’s Thesis, 1995

Pitfalls in Relying on CPM
 CPM network scheduling is static, not dynamic

 Single-point activity durations known with certainty
 OK only if everything goes according to plan
There are no “facts” about the future
Lincoln Moses, Statistician and Administrator of Energy Information in the US DOE
1977 Annual Report to Congress
 CPM durations are really probabilistic assessments

Risk Analysis Answers
Many Questions that CPM cannot
 Since the inputs are uncertain, the results are

uncertain and we need to make statistical
statements
 Can address questions CPM cannot
 The 3 promises
1. What is the likelihood of meeting schedule?
2. How much schedule contingency do we need to provide?
3. Where is there risk to the project schedule?

First check the Schedule: Apply GAO Best
Practice Scheduling to Agency Schedules
 Can use tools such as Acumen Fuse, Steelray or Oracle
Primavera Risk Analysis (Pertmaster) Schedule Check
Report to perform analysis
– Check the results of any third-party software against the schedule
in its native software – sometimes get incorrect results
 Always use native schedule and filter / sorting capabilities to
check third-party software results
– Can do this inside the schedule or copy to Excel for sorts and
report table creation

http://www.gao.gov/products/GAO-16-89G
Why Conduct Monte Carlo Simulation
 It would help estimate our project if we had data on 5,000
projects exactly like our, just with different risks occurring
– Actual projects, normalized to show % overrun of the schedule
– We may not have such databases
 Monte Carlo simulation creates this database
– Uses our schedule
– Uses our risks
– Risks occur in different combinations, 5,000 times, recording the
results

Risk of an Individual Activity
 Simple activity duration estimates are risky

 Is this activity going to take 30 days? Are you sure?
30d
Design Unit 1

Uniform and Triangular Distributions
Entire Plan : Duration Entire Plan : Duration
100% 240 100% 240
400
200 95% 239 95% 236
90% 238 90% 234
350
180 85% 237 85% 233
80% 235 80% 232
160 75% 234 75% 231

300
70% 233 70% 230
140 65% 231 65% 229
Cumulative Frequency
250
60% 230 60% 228
120
55% 229 55% 227
Hits
Hits
50% 227 200 50% 226
100
45% 226 45% 226
40% 225 40% 225

80 150
35% 224 35% 224
30% 222 30% 223

60
25% 221 100 25% 223
20% 220 20% 222

40
15% 218 15% 221

50
20 10% 217 10% 219
5% 216 5% 218
0 0% 215 0 0% 215
215 220 225 230 235 240 215 220 225 230 235 240
Distribution (start of interval) Distribution (start of interval)

BetaPERT and Normal Distributions
Entire Plan : Duration Entire Plan : Duration
400 100% 240 100% 243
95% 234 95% 235
90% 232 400 90% 233

350
85% 231 85% 232
80% 230 350 80% 231
300 75% 229 75% 230
70% 228 70% 230

300
65% 228 65% 229
250
60% 227 60% 228
250
55% 226 55% 228
Hits
Hits
200 50% 226 50% 227
45% 225 200 45% 226
40% 224 40% 226

150
35% 223 150 35% 225
30% 223 30% 224
100 25% 222 25% 224

100
20% 221 20% 223
15% 220 15% 222

50
50
10% 219 10% 221
5% 218 5% 219
0 0% 215 0 0% 212
215 220 225 230 235 240 215 220 225 230 235 240
Distribution (start of interval) Distribution (start of interval)
© 2016 Hulett & Associates, LLC 16 16

Risk Along a Contiguous Schedule Path
 Path risk is the combination of the risks of its activities
Design Build Test

Test
Start Finish
Unit Unit Unit
Unit

Really Simple Schedule
 This schedule finishes on September 3

– 7-day weeks, like a model changeover, refinery turnaround
 If we can get into trouble with this simple schedule, we can get into
trouble with real project schedules

Add Duration Risk to the Schedule using
Triangular Distributions
This section features Primavera Risk Analysis, formerly Pertmaster,

owned by Oracle

Monte Carlo Simulation Results for Really
Simple Schedule
CPM date is not even the most likely – That’s about 9/13
Entire Plan : Finish Date
100% 19 Oct 11
240
95% 29 Sep 11
220 90% 25 Sep 11
85% 23 Sep 11
200
80% 21 Sep 11
75% 19 Sep 11
180
70% 18 Sep 11
160 65% 16 Sep 11
60% 15 Sep 11
140
55% 14 Sep 11
Hits
80% Target is
120 50% 13 Sep 11
45% 11 Sep 11
100
9/21
40% 10 Sep 11
35% 09 Sep 11
80
30% 07 Sep 11
60 25% 06 Sep 11
20% 05 Sep 11
40
15% 03 Sep 11
10% 01 Sep 11
20
5% 29 Aug 11
0 0% 17 Aug 11
29 Aug 11 18 Sep 11 08 Oct 11
Distribution (start of interval)
CPM date is about 16% Likely to be met

Risk at Merge Points:
The “Merge Bias”
 Many parallel paths merge in a real schedule
 Finish driven by the latest converging path
 Merge Bias has been understood for 40 years
Design Unit 1 Build Unit 1 Test Unit 1
Start
Finish
Design Unit 2 Build Unit 2 Test Unit 2

This Schedule has Three Parallel Paths
Each path has exactly the same structure

CPM says this, too, finishes September 3

Evidence of the Merge Bias
100% 20 Oct 11
Most Likely is now Sept 21 95% 04 Oct 11
90% 02 Oct 11
85% 29 Sep 11
250
80% 28 Sep 11
75% 26 Sep 11
80th
70% 25 Sep 11
200
65% 24 Sep 11
percentile is
60% 23 Sep 11
55% 22 Sep 11
now Sept 28
Hits
150 50% 21 Sep 11
45% 20 Sep 11
40% 19 Sep 11
35% 18 Sep 11
100
30% 17 Sep 11
25% 16 Sep 11
20% 15 Sep 11
50
15% 13 Sep 11
10% 12 Sep 11
5% 09 Sep 11
0 0% 30 Aug 11
08 Sep 11 18 Sep 11 28 Sep 11 08 Oct 11 18 Oct 11
Likelihood of Sept 3 is < 1%


Evidence of Merge Bias (continued)
Distribution Analyzer
Evidence of the
Merge Bias @ P-80
Variation:7
Three Path Schedule
Cumulative Probability
One Path Schedule

What are the Highest Risk Activities?
 Monte Carlo simulation

– Activities on critical path in most iterations
 Path delaying the project may not be the critical path
identified by CPM
 This is “Schedule Critical” not necessarily Technically
Critical
– Combination of risk and low float (slack)

What are the Highest Risk Activities?
Units 1 & 3 are Shorter and Not Risk Mitigated
“Critical” Unit 2 is Identified for Risk Mitigation

Criticality in Pertmaster
Three Path Project
Criticality Index: All tasks
000002 - Start 100%
000015 - Finish 100%
000004 - Design Unit 1 45%
000005 - Build Unit 1 45%
% of iterations on the critical path

000006 - Test Unit 1 45%
designates the path most likely to
000012 - Design Unit 3 40% delay the project (new definition of
000013 - Build Unit 3 40%
“critical”)
000014 - Test Unit 3 40%
000008 - Design Unit 2 19%
000009 - Build Unit 2 19%

Duration Sensitivity – Correlation Concept
Three Path Project

Duration Sensitivity: Entire Plan - All tasks
000006 - Test Unit 1 47%
000014 - Test Unit 3 42%
000004 - Design Unit 1 34%
000012 - Design Unit 3 29%
000005 - Build Unit 1 21%
000013 - Build Unit 3 18% Correlation coefficients not much

help to project management in
000008 - Design Unit 2 10%
making decisions
000010 - Test Unit 2 8%
000009 - Build Unit 2 6%

Cruciality = Criticality x Sensitivity
Three Path Project

Duration Cruciality: Entire Plan - All tasks
000006 - Test Unit 1 22%
000014 - Test Unit 3 17%
000004 - Design Unit 1 15%
000012 - Design Unit 3 12%
000005 - Build Unit 1 9%
000013 - Build Unit 3 7%
Helps eliminate the spurious

000008 - Design Unit 2 2%
correlation for activities that are not
000010 - Test Unit 2 2% frequently on the risk critical path
000009 - Build Unit 2 1%

What if We May Fail a Test or Inspection?
Probabilistic Branching
 Many projects have points where there is a possibility of

failure, a discontinuous event
 Have to model the likelihood of the failure and its
consequence for the schedule
 Called “Probabilistic Branching”

Model the Probabilistic Branch
 Typically do not include failure in schedule
– Include FIXIT and Retest with 0 duration to reserve the 9/3 date
 Enter ranges for the new activities if they occur

Typical Bi-Modal Result Distribution
100% 02 Jan 12
1100
95% 07 Dec 11
1000 90% 01 Dec 11
900
85% 26 Nov 11
“Shoulder” is at
70% for a 30%
80% 21 Nov 11
75% 16 Nov 11
800
likely risk
70% 20 Oct 11
65% 04 Oct 11
700
60% 30 Sep 11
55% 27 Sep 11
600
The 80% date falls

Hits
50% 25 Sep 11
500 45% 23 Sep 11
400
40% 21 Sep 11
35% 19 Sep 11
in the Failure Mode
30% 17 Sep 11
300
25% 16 Sep 11
20% 14 Sep 11
200
15% 12 Sep 11
10% 10 Sep 11
100
5% 07 Sep 11
0 0% 28 Aug 11
28 Sep 11 17 Nov 11 06 Jan 12

Effect of Mitigating The Risk of Test Failure on an
80% Company Distribution Analyzer
Variation:46
If taking 10 Days Longer in Build reduces probability to
10%, can save 46 days for an 80th percentile company

Causes of Correlation
 Correlation between activity durations
– When two activities’ durations “move together”
– They are driven by a common risk driver
– Using Pearson correlation, not Spearman Rank Order
Software
Designing
Technology
State-of-
the- Art
Software
© 2016 Hulett & Associates, LLC Coding 34
Effect of Correlation
 Without correlation there is a lot of cancelling-out between

long and short durations in each iteration
– The result is for moderate uncertainty risk overall
 With correlation, long durations will occur together on an
iteration and reinforce each other as long durations are
added down the path
– Same for short durations
– There is reinforcing long-long and short-short so get higher-highs
and lower-lows

This Matrix shows High Correlation
 Correlation is usually shown in a Correlation Matrix like this
one
Design Build Test
Design 1.0 .9 .8
Build .9 1.0 .7
Test .8 .7 1.0
 Shows high correlation between all 3 activities

Install Correlation Coefficients between
Activity Durations

Effect of Correlation is to Increase the
Standard Deviation along the Path Distribution Analyzer
100%
Not much 90%
change near Variation:5

80%
the mean High correlation adds 70%
5 days at the P-80
High
60%
correlation 50%
subtracts 40%
5 days at 30%
the P-20
Variation:5
20%
10%
0%
14 Aug 12 19 Aug 12 24 Aug 12 29 Aug 12 03 Sep 12 08 Sep 12 13 Sep 12 18 Sep 12 23 Sep 12 28 Sep 12 03 Oct 12 08 Oct 12 13 Oct 12 18 Oct 12 23 Oct 12
Not likely to be much help for real schedules that have parallel paths

This Matrix is Not Feasible
 Correlation Matrix may be infeasible

 Eigenvalue (determinant) test used to test feasibility
Design Code Test
Design 1.0 .9 .8
Code .9 1.0 .2
Test .8 .2 1.0

Software Performs
Eigenvalue Test And Alerts the User
This plans to adjust the

coefficients permanently so
we can see what they
should be to just pass the
Eigenvalue test.

Correlation Coefficients Change
before the Simulation

Gather Good Quality Risk Data
 3-point estimates of Impact (now used for uncertainty)

– Another method, the Risk Driver method, uses the probability that
the discrete risk will occur in addition to the impact range
– We combine uncertainty with discrete risks
 Credibility comes from quality risk data
 The benefits of collecting data about project risk
– Gain better understanding of the project
– Build stronger project teams
– Communicate better about project problems

Main Ideas about Risk Data
 Input data are gathered from people’s expert judgment

– There are often no company or industry data bases on risk
 Concepts of risk are usually new to the project participants
– Collecting “data” about the future should not be new to most
people who have put together a schedule
 Some are reluctant to participate -- uncomfortable
 Need corporate culture to be supportive
 Need time and budget for the risk analysis – make risk
understood as part of the team’s job, not an extra

Independent Data Collection Organization
 Consider making the risk analysis activity independent of

the project
Management Team
Risk Project Project

Analysis One Two

The Confidential Risk Interview
 Arrange for enough time to interview

– Usually take 2 hours per person
 Promise confidentiality – No information will be connected to the
individual interviewee
 Interviews give the team members focused discussion on risk
– Unlike the workshop where only one person can talk at a time
while the others are waiting to be heard
 They appreciate that the interviewer is interested in their ideas

The Risk Interview (continued)
 During the risk interview challenge interviewees’ risk ranges

– Risk is usually underestimated
 Ask open-ended questions to identify risks not necessarily
included in the Risk Register
– Some are the Unknown Unknowns that were not unknowable
– Some are the Unknown Knowns, that were known but not talked
about since they are adverse to the project or the company

Biases can Make Risk Data Collection
Difficult
 Motivational bias
– Some people do not want to cooperate
– Others want to bias the results, usually so the project looks good.
They have their own agenda
 Cognitive bias
– Even people who want to respond accurately may find it difficult at
first
– Thinking about things going wrong is not easy

Sources of Motivational Bias
 Want the project to look good

 Not willing to jeopardize the project
 Unwilling to admit uncertainty or inability to do the job
 Afraid of telling people the estimates are not “solid”
 Identified with a specific number, result
 Afraid of “shoot the messenger” response

A Politically-Set Corporate Ceiling
Ceiling Amount
Likelihood
Risk Denied
Element Cost

Cognitive Bias in
Quantifying Project Risk
 Cognitive bias is common
– Even though you want to estimate the risk you find it hard
– People new to risk analysis are more likely to have this bias
– Most difficult to estimate pessimistic scenarios, failure
Underestimation of risk is quite common

Anchoring and Adjusting Bias
 “Anchor” on the baseline estimate

– This estimate has been carefully compiled, documented
– The anchor takes importance beyond its credibility
 “Adjust” the extreme ranges only slightly from the anchor
 People find it difficult to imagine values very far away from
the anchor estimate
See: A. Tversky and D. Kahneman, “Judgment under

Uncertainty: Heuristics and Biases,” Science, Sept. 26, 1974

Anchoring and Adjusting Bias
Range
Unbiased Range
Anchored on
Most Likely
Likelihood
Activity Duration

Use “Trigen” Function to Correct for
Underreporting of Extreme Ranges
Compare Triangle and Trigen (205,216,245)

10th 90th
5 Percentile Percentile
3
Triangle
Trigen
2
0
190 220 250

Unstated Assumptions Bias
Unspoken True High Range

Likelihood
Recognized
High Range
Activity Duration

Availability Bias can Increase the
Perception of Project Risk
Original Distribution
Dramatic
Event
Likelihood
Element Cost

Representative Bias can Increase Project
Risk – may need Correction for Priors
Revised Distribution
Original
Distribution New
Information
Likelihood
Bayesian statistics tell

how to combine priors
with new information
Activity Duration for a blended forecast

Limitations of the Traditional
3-point Estimate of Activity Duration
 Traditional schedule risk analysis starts with the activity that

is impacted by risks
 Estimates the 3-point estimate for optimistic, most likely and
pessimistic duration
– Creates a probability distribution for activity duration
– Performs Monte Carlo simulation
 This is the image of the risk on the objective, activity
duration
– Which risks influence the duration are not carefully sorted
– Just “give me a +20%, -10%”
Some Problems with Traditional Approach
 Can tell which activities are crucial, but not directly which
risks are driving
 Makes poor use of the Risk Register that is usually available
 Cannot decompose the overall schedule risk into its
components BY RISK
– Ability to assign the risk to its specific risk drivers helps with
communication of risk causes and risk mitigation

Uncertainty, Estimating Error
and Estimating Bias
 Uncertainty, the inherent variability in project activities that

arise because people and organizations cannot do things
reliably on plan
 Estimating error – attaches to all types of estimates
 Estimating bias – estimates may be slanted, usually

toward shorter durations, to make desired project results

Inherent Variability is Similar to Common
Cause Variability
 Inherent variability is similar to “common cause variation”
described by Walter A. Shewhart and championed by W.
Edwards Deming in process control
 Common cause variability is a source of variation caused by
unknown factors that result in a steady but random distribution of
output around the average of the data
 Common cause variation is a measure of the process’s potential,
or how well the process can perform when special cause
variation is removed
 Common cause variation is also called random variation, noise,
non-controllable variation, within-group variation, or inherent
variation. Example: Many X’s with a small impact.
http://www.isixsigma.com/dictionary/common-cause-variation/ cited February 6, 2015

Estimating Error (1)
 Estimating error is often attributed to a lack of information

concerning specific issues needed to make up a duration or
cost estimate for a WBS element
– We may not have specific vendor information until the vendors
bid. Vendor information is required for completed engineering
– Ultimately we do not necessarily have contractor bids
 Each of these sources of information can be helpful to
narrow the estimating error. Still, the estimates and even
contractor bids are uncertain

Estimating Error (2)
 The estimating range is often related to the “class” of
estimate, determined by the level of knowledge and the
method of estimating
 With less knowledge the “plus and minus” range would be
large, but as more information is known it may become
smaller
 Research shows that the actual range of uncertainty
around estimates is larger than recommended by
professional associations (including AACEI)
John Hollmann, 2012 AACE INTERNATIONAL TRANSACTIONS,
RISK.1027: Estimate Accuracy: Dealing with Reality)
Estimating Bias is Common for Schedule
and for Cost
“I want it NOW!”
 “Schedule pressure dooms more megaprojects than any other single
factor”
 Ambitious managers see early completion as ways for promotions. But,
every megaproject has an appropriate pace that becomes known early.
Pronouncements do not change this pace
“We need to shave 20 percent off that cost number!”

 Cost reduction task force is a counterproductive exercise
 May just reduce estimates, this is foolish
 Alternatively, may actually identify scope to come out, but the scope
needs to be added back in later, so only temporary reduction in cost
Edward W. Merrow, Industrial Megaprojects (Wiley, 2011)

Ask Yourself these Questions
about the Duration Estimates
 Was there pressure put on the estimator or scheduler
– By prior expectations
– Statements by management or
– By the customer, or
– Was pressure for early finish implicit in the competitive process?
 How long would this scope of work take if no pressure for an
earlier date were brought to bear?
– Contractors often claim that the schedule would take longer without
pressure, “But, we can do it!”

Handling Estimating Bias
 When talking with project participants (management, team

leaders, SMEs) we often find that they do not believe the
values in the schedule
– Motivational bias and cognitive bias are present
 With a range represented by optimistic, most likely and
pessimistic values, these people present that the “most
likely” duration or cost is not the value in the schedule for
activities or estimate for cost elements
– Often the “most likely” multiplier is 1.05 or 1.1 or more, indicating
that the estimates are viewed as being 5%, 10% or more above
those in the project documents
Summary of Inherent Variability
and Estimating Error / Bias
 These sources of uncertainty have already occurred and are

“baked in the cake” of the schedule and cost estimate being
risked
 They are 100% likely so they can be represented by a 3-
point estimate (min, most likely, max) of multiplicative
factors applied directly to activities’ durations
 Under-reporting may be corrected and 3-point estimates
may be correlated

Introducing the Gas Production Platform
Schedule
Three year+ schedule costing $1.57 billion. Using Polaris®

© Booz
from 2016 Allen
HulettHamilton
& Associates, LLC 67 67
Applying Different Uncertainty Reference
Ranges to Categories of Tasks
Each category of activity may
have different levels of
uncertainty, called “reference
ranges.”
Five of the ranges have “most

likely” values that differ from the
durations in the schedule
Three (Engineering, Drilling and

HUC) use the Trigen function to
correct for suspected under-
reporting impact ranges

Risk on the Offshore Gas Production Platform -
Reference Range Uncertainties
With Uncertainty by category of task
representing:
• Inherent variability
• Estimating error
• Estimating bias
The CPM date is 20 March 2017
The P-80 date is 30 July 2017 for a

contingency just with Uncertainty of 4
+ months
This is very likely irreducible. It

represents the base that cannot be
mitigated

Discrete Risks is Similar to
Special Cause Variation
 Unlike common cause variability, special cause
variation is caused by known factors that result in a
non-random distribution of output. Also referred to as
“exceptional” or “assignable” variation. Example: Few
X’s with big impact.
 Special cause variation is a shift in output caused by a
specific factor such as environmental conditions or
process input parameters. It can be accounted for
directly and potentially removed and is a measure of
process control.
http://www.isixsigma.com/dictionary/variation-special-cause/ cited February 6, 2015)

Introducing the Risk Driver Method for Causing
Additional Variation in the Simulation
Four risk drivers are specified. The first is a general risk about engineering productivity,
which may be under- or over-estimated, with 100% probability. It is applied to the two
Design activities

100% Likely Risk Driver’s
Effect on Design Duration
With a 100% likely risk the
probability distribution of the
activity’s duration looks like a
triangle. Not any different
from placing a triangle
directly on the activity

Risk Driver with
Risk at < 100% likelihood
With this risk, the Construction Contractor may or may not be familiar with the technology,
the probability is 40% and the risk impact if it happens is .9, 1.1 and 1.4. It is applied to the
two Build activities

With a 40% Likelihood, the “Spike” in the
Distribution Contains 60% of the Probability
Here is where the Risk

Driver method gets
interesting. It can create
distributions that reflect:
• Probability of occurring
• Impact if it does occur
Cannot represent these two
factors with simple
triangular distributions
applied to the durations
directly

Risk Drivers Models
how Correlation Occurs
 Correlation can be caused by identifiable risks that are

assigned to two different activities
– If the risk occurs it occurs for each activity
– If the risk impact multiplier is X% it is X% for each activity
 We are not very good at estimating correlation coefficients,
so generating them within the simulation is a better
approach
 There still may be correlations among uncertainty (3-point
estimates)

Risk Drivers Generate Correlation between
Activities (1)
Risk 1: Probability 100% Impact .9, 1.05, 1.3
Activity 1 Activity 2
Correlation (Activity 1, Activity 2) = 100%

Activities (2)
But there is no such thing as 100% correlation! OK!
Risk 1: Probability 100% Impact .9, 1.05, 1.3
Uncertainty Not Correlated: .85, 1, 1.2
Adding uncorrelated uncertainty reduces

correlation (Activity 1, Activity 2) to 86%

Activities (3)
Risk 2: Probability 40% Risk 1: Probability 100% Risk 3: Probability 65%

Impact .9, 1.1, 1.4 Impact .9, 1.05, 1.3 Impact .9, 1.15, 1.5
Correlation (Activity 1, Activity 2) = 64% (without uncertainty)

Activities Can be Influenced
by More than One Risk Driver
An Organizational Risk has been added to the mix, assigned to all

activities in the Offshore Gas Production Platform schedule
Adding Risk Drivers
to Every Activity
With all risk Drivers

including the
Organizational Risk the
P-80 result is 25 January
2018, an additional 7
months
With Uncertainty the P-80

was 30 July2017
The scheduled date is 20

March 2017

Parallel and Series Risks
Multiplicative with Risk Drivers
If these two risks cannot be recovered from simultaneously, they
are entered in series
Risk 1 1.2 factor Risk 2 1.05 factor Use (1.2 x 1.05 = 1.26) Factor,
multiply the two
If recovery from two risks can be accomplished simultaneously,

they are entered in parallel
Risk 1 1.2 factor Use 1.2 Factor, the largest

Risk 2 1.05 factor factor only

Assign the Risks To Design 1 in Series
and to Design 2 in Parallel
Specify each risk in parallel for Design 2, Series for Design 1

Three Risk Drivers Applied to
One Activity In Series, To Another In Parallel
Risks in
Parallel on
Design 2
Risks in Series on Design 1
Risks in series often lead to very long durations, especially if there are many risks on the activity

Risk Prioritization Method
 Risks should be prioritized through the project schedule and

the Monte Carlo simulation method to inform the risk
mitigation exercise
 For management we need to identify those risks by “days
saved” if they were fully mitigated so management can do
benefit/cost
 For management we should identify “days saved” at the
target level of certainty, say P-80

Two Approaches to Risk Prioritization
using Quantitative Methods
 Typical Tornado Diagram with Risks (not activities or paths)

as the arguments help to prioritize the risks
 However, with the structure of the schedule the Tornado
Diagram is instructive but not definitive
– The order of the risks’ importance can change when one is
removed, since that exposes other paths that were “risk slack
paths” before
 Tornado Diagrams are not reliable with risks that are not
100% likely

Tornado Highlighting Risks, Not Activities
or Paths
This special tornado diagram
focuses on the entire impact of the
risks, including their probability,
impact range and the activities to
which they are assigned
Still, it is based on correlation

concepts
It shows Drilling Risk as an

Opportunity, negatively correlated
with finish date.
Correct to focus on risks but still

measure is correlation
Benefits of the New
Risk Prioritization Approach
 Identify the level of uncertainty desired (P-80)
 Simulate the schedule as many times as there are risks
 Then Identify the risk that saves the most days when it is
eliminated
 Focus on the risks at the P-80 level of certainty and
measure impact in “days saved”
– These are 3 good measures for management to use when
determining the mitigations that make sense

Iterative Approach to
Prioritizing the Risk
 Purpose: determine which risks contribute the most days at
the P-80 level
 Compute the Baseline with All Risks In
 Pass # 1: Simulate with each risk disabled in turn, recording
the P-80 date
– The risk with the earliest P-80 date is 1st priority
– Take it out for Iteration # 2
 Pass # 2: Simulate the remaining risks, disabling each in turn,
recording P-80, choose earliest. Take it out for Iteration # 3
 Etc.

Picture of Prioritized Risks
Selected by their Days Saved at P-80
Iterative Approach to Prioritizing Risks (Based on Days Saved at P-80)
Risk # 1 2 3 4 5 6 7 8
Resources
Offshore Coordinati
Priority Level Abusive Suppliers Fab Geology Problems may go to
design on during
(Iteration #) Bids Busy productivity unknown at HUC other
firm Installation
projects
1 X X X X X X X 1
2 X X X 2 X X X
3 X 3 X X X X
4 X X X X 4
5 X 5 X X
6 X X 6
7 7 X
8 8
This is an example of the simulation strategy for a project with 8 risks. It requires 8 + 7 + 6
+….+1 or 36 separate simulations. This would take a long time by hand. At this point only
Polaris has automated the process.
Schedule Risk Tornado
with Risks Prioritized by Days Saved
Unlike typical activity tornado diagrams

showing activities and based on correlation
coefficients, this one is based on risk and is
calibrated in days saved and computed at
the P-80

Table Showing Risks’ Days Saved
Gas Platform-1 - Risk Prioritization (80%)
Days
Risk UID Name
Saved
The organization has other priority projects so personnel and funding may
8 102
be unavailable
4 Fabrication yards may experience lower Productivity than planned 34
2 Engineering may be complicated by using offshore design firm 15
7 Fabrication and installation problems may be revealed during HUC 15
3 Suppliers of installed equipment may be busy 9
6 Installation may be delayed due to coordination problems 4
1 Bids may be Abusive leading to delayed approval 0
5 The subsea geological conditions may be different than expected -1
TOTAL DAYS SAVED WITH FULL MITIGATION OF RISKS 178
Uncertainty (inherent, estimating error / bias) 130
TOTAL CONTINGENCY DAYS WITH UNCERTAINTY & RISKS 308
Target for Mitigations is 178 days, risk-by-risk. Uncertainty alone accounts for 130 days.
Total schedule contingency to ©P-80
2016 Hulett & Associates, LLC
is 308 days. 91 91
Risk Mitigation Workshop(s)
 This is a workshop with the project manager, deputy PM, team

leads, controls personnel, SMEs with experience
 Use the prioritized risk list
– Start at the top
– Working on risks lower on the priority list will not be effective. Those
risks are not important until the top risk is dealt with as much as
possible
– Determined by the structure of the schedule and which paths are risk
critical – changes as risks are mitigated

Sample Risk Mitigation Entry
Risk: The organization has other priority projects so personnel and funding may be unavailable
Most P-80 Cost
Probability Low High P-80 Date
Likely ($ billions)
Pre-Mitigated parameters 65% 95% 105% 125% 1/22/2018 $2.13
Mitigation Establish this project as top priority - needs top management action and
Action commitment
10/20/201
Post -Mitigated parameters 15% 95% 100% 115% $1.99
7
Cost
Risk Owner: S. Smith Days saved
Saved
Date of
Within 1 month Results 94 $0.14
Action:
Risk Action Cost of
Risk is not completely
B. Blake mitigated. Cost saved is the reduction of cost contingency reserve$0.02
held
Owner: Mitigation
for schedule risk. For Net Cost Saved subtract the $20 million cost of mitigation

Compare the Baseline and One Risk
Mitigation Scenario – Finish Date
The high priority risk has been mitigated. The finish date moves to 20
October 2017 from 22 January 2018, about a 3-month improvement
Partially Mitigate all Risks – Finish Date
All risks mitigated
Pre-mitigation results
One risk mitigated
Mitigating all risks (Here, just reducing the probability by half) moves the P-80 date
back to 10 August, 2017 for a total mitigation time of about 5 months.

Schedule Risk Analysis
David T. Hulett, Ph.D.

Hulett & Associates, LLC
ICEAA / 2016

Schedule Risk Analysis ICEAA 2016 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Schedule Risk Analysis ICEAA 2016 PDF

Uploaded by

Copyright:

Available Formats

Schedule Risk Analysis

David T. Hulett, Ph.D.

© 2016 Hulett & Associates, LLC 1

© 2016 Hulett & Associates, LLC 2

 Promote the language of probability and use of its

© 2016 Hulett & Associates, LLC 3

 Risk analysis strategy

© 2016 Hulett & Associates, LLC 4

“Initial cost and schedule estimates for major projects have

© 2016 Hulett & Associates, LLC 5

“A formal risk analysis is putting on the table those problems

Source: “Final Report,”

© 2016 Hulett & Associates, LLC 6

© 2016 Hulett & Associates, LLC 7

© 2016 Hulett & Associates, LLC 8

 CPM network scheduling is static, not dynamic

 CPM durations are really probabilistic assessments

© 2016 Hulett & Associates, LLC 9

 Since the inputs are uncertain, the results are

© 2016 Hulett & Associates, LLC 10

© 2016 Hulett & Associates, LLC 11

© 2016 Hulett & Associates, LLC 13

 Simple activity duration estimates are risky

© 2016 Hulett & Associates, LLC 14

200 95% 239 95% 236

90% 238 90% 234

80% 235 80% 232

160 75% 234 75% 231

140 65% 231 65% 229

40% 225 40% 225

30% 222 30% 223

20% 220 20% 222

15% 218 15% 221

© 2016 Hulett & Associates, LLC 15

95% 234 95% 235

90% 232 400 90% 233

80% 230 350 80% 231

300 75% 229 75% 230

70% 228 70% 230

45% 225 200 45% 226

40% 224 40% 226

30% 223 30% 224

100 25% 222 25% 224

15% 220 15% 222

© 2016 Hulett & Associates, LLC 16 16

 Path risk is the combination of the risks of its activities

Design Build Test

© 2016 Hulett & Associates, LLC 17

 This schedule finishes on September 3

© 2016 Hulett & Associates, LLC 18

This section features Primavera Risk Analysis, formerly Pertmaster,

© 2016 Hulett & Associates, LLC 19

220 90% 25 Sep 11

160 65% 16 Sep 11

CPM date is about 16% Likely to be met

Design Unit 1 Build Unit 1 Test Unit 1

© 2016 Hulett & Associates, LLC 21

Each path has exactly the same structure

© 2016 Hulett & Associates, LLC 22

Most Likely is now Sept 21 95% 04 Oct 11

150 50% 21 Sep 11

Likelihood of Sept 3 is < 1%

© 2016 Hulett & Associates, LLC 23