64R-11 - AACE International

This Recommended Practice (RP) is brought to you as Total Cost
public service by AACE International, the Authority for Management®

Total Cost Management. Framework:
The AACE International Recommended Practices are Total Cost
the main technical foundation of our educational and Management is a
certification products and services. The RPs are a series systematic approach
of documents that contain valuable reference to managing cost
information that has been subject to a rigorous review throughout the life
process and recommended for use by the AACE cycle of any enterprise, program, facility, project, product or service.
International Technical Board. AACE’s flagship publication, the TCM Framework: An Integrated
Approach to Portfolio, Program and Project Management, is a
AACE International is a 501(c)(3) non-profit structured, annotated process map that for the first time explains each
professional association serving the total cost practice area of the cost engineering field in the context of its
management community since 1956. AACE relationship to the other practice areas including allied professions.
International provides its members and stakeholders
with the resources they need to enhance their
performance and ensure continued growth and
success. With over 8,500 members world-wide, AACE
International serves total cost management Visual TCM
professionals in a variety of disciplines and across all Framework:
industries. AACE International has members in 87
countries. If you share our mission to “enable Visual TCM graphically
organizations around the world to achieve their demonstrates the
investment expectations by managing and controlling integration of the
projects, programs, and portfolios and create value by strategic asset
advancing technical knowledge and professional management and
development”, then we invite you to become one of project controls
our members. process maps of the TCM Framework. The Visual TCM application has
been designed to provide a dynamic view of the TCM processes, from
In addition to this and other Recommended Practices, the overall strategy process maps to the mid-level processes and
here are just a sample of the products and services that detailed activities. The processes are hyperlinked, giving the user the
AACE has to offer you: ability to move to and from related process maps and reference
Recommended Practice Recommended Practice

material. This allows for the optimal effectiveness of understanding and management. Comprehensive, well organized, and timely, each PPG is a
using the process and sub-process in the context of and relationship to collection of selected articles covering a particular technical topic area
associated sub-processes that share common strategies and objectives. or industry segment. The PPGs provide an excellent source of reference
Visual TCM allows the user to view and apply TCM section-by-section, material and is a welcome addition to any reference library.
at a sub-process or functional level. Visual TCM is available to members
at no extra fee.
Certification:
Virtual Library: Since 1976, AACE has

been certifying
Members receive free individuals as Certified
access to the Virtual Cost Consultants
Library, an online (CCC)/Certified Cost
collection of over 5000 Engineers (CCE);
complete technical Certified Cost
articles on virtually Technicians (CCT);
every aspect of cost Certified Estimating Professionals (CEP); Certified Forensic Claims
engineering. Search Consultants (CFCC); Earned Value Professionals (EVP); and Planning &
this extensive database and immediately retrieve the best techniques Scheduling Professionals (PSP). In the midst of staggering business and
and potential solutions to the problems confronting you and your economic turmoil, you need all the tools at your disposal to help shore
organization. up your career prospects. AACE certification can help you and the
organizations that rely on you for help!
Professional
Practice Guides Online Learning
(PPGs): Center:
Professional Practice The Online Learning

Guides contain the Center features
most worthwhile modules based upon
contributions to the actual technical
field of total cost presentations captured

at our Annual Meetings. Each recorded unit includes a live audio Discussion
recording of the speaker synchronized to the slides accompanying the Forums:
presentation. Each unit includes the technical paper associated with the
presentation, and a downloadable audio-only version that you may play The discussion forums
on your mobile device or iPod. Completion of each unit earns 0.1 AACE encourage the
recertification credits (i.e. 0.1 CEUs). An electronic certificate of exchange of thoughts
completion will be attached to your profile. and ideas, through
posting questions and
discussing topics. They
provide a great means for networking and interaction with your peers.
Conferences: Participate anytime at your convenience and receive automatic e-mail
notifications on topics that are of interest to you. With several thousand
AACE International’s users, if you have questions or concerns about a technical subject,
Annual Meeting brings program, or project - the forums are a great resource for you.
together the industry’s
leading cost
professionals in a
forum focused on
learning, sharing, and Mentoring
networking. Over 100
hours of technical presentations and an industry tradeshow that will Program:
challenge you to better manage, plan, schedule, and implement
technology for more effective and efficient business practices. Looking to gain more
knowledge from an
The International TCM Conference is a similar event that is held outside experienced
of North America – complete with technical presentations, seminars professional or an
and exhibits. opportunity to help
another professional?
Included with your membership, AACE offers a comprehensive
mentoring program for individuals interested in sharing knowledge with
others or advancing their own careers to the next level.

Career Center: Periodicals
AACE’s career center Members receive a

provides tools and complimentary
resources for you to subscription to the
progress through your Cost Engineering
career. journal, AACE’s
bi-monthly
Looking for the next professionally
rung on the career ladder or to hire the talent necessary to take your peer-reviewed publication. It contains best-in-class technical articles on
firm to the next level? Job seekers, use our services to find your next total cost management related subjects.
job – post your resume, get e-mail notifications of new job-postings, It is published as both a print version and an online version.
and more. Employers, post your current job-openings and search our
extensive resume database to find your next star employee. Our bi-monthly digital publication, Source, focuses on AACE activities
and items of interest to the total cost management community, with
special features for our members.
Salary and
Demographic
Survey:
Conducted annually,
salary survey is a great
resource for
employers that want
to gain a better
understanding of the competitive marketplace for talent and for
employees interested in knowing how their compensation compares
with their peers in the profession.

AACE International Recommended Practice No. 64R-11
CPM SCHEDULE RISK MODELING AND ANALYSIS:

SPECIAL CONSIDERATIONS
TCM Framework: 7.6 – Risk Management
Rev. May 2, 2012

Note: As AACE International Recommended Practices evolve over time, please refer to www.aacei.org for the latest revisions.
Contributors:
Christopher P. Caddell, PE CCE (Author) Leonard Enger, CCE
James E. Arrow Ambrish Gupta, CCE
Chris Beale Dennis R. Hanks, PE CCE
Peter R. Bredehoeft, Jr., CEP John K. Hollmann, PE CCE CEP
Christopher W. Carson, PSP Shannon A. Katt
Dr. Ovidiu Cretu, PE H. Lance Stephenson, CCC
Michael W. Curran Anthony M. Woodrich, CCE
Larry R. Dysert, CCC CEP
Copyright © AACE® International AACE® International Recommended Practices

AACE® International Recommended Practice No. 64R-11
CPM SCHEDULE RISK MODELING AND ANALYSIS:
SPECIAL CONSIDERATIONS
TCM Framework: 7.6 – Risk Management
May 2, 2012
INTRODUCTION
Scope
This recommended practice (RP) of AACE International defines general practices and considerations for the various
aspects of conducting a project schedule risk analysis using a critical path method (CPM) network of activities and
Monte Carlo methods to estimate contingency and/or to understand the project’s behavior in consideration of
risk. This RP does not present a standalone methodology, but is an extension of other RPs that present CPM-based
approaches to schedule risk analysis and contingency estimating. This RP discusses key procedural, analytical and
interpretive considerations in preparation and application of a CPM model; considerations that were not covered
in the broader methodological RPs.
A quantitative schedule risk analysis is an important aspect of risk management on a project. It can help project
teams understand how project risks and uncertainty may impact the project schedule and when key milestones
will be achieved. The analysis should be conducted by a skilled risk analyst. This analysis is typically performed
during project development prior to key approval points, but can also be used during project execution to assess
the current status of the project schedule risks.
Most schedule risk analyses utilize a CPM network as the base tool for conducting a Monte Carlo type simulation of
project schedule variability. The CPM model for risk analysis must be properly constructed and realistically reflect
how the identified risks may impact the project activities and overall duration. The identified schedule risks may be
linked to the activities in the model in a variety of ways, depending on the software used and user preference.
Regardless of how the risks are linked to activities in the software, the analysis needs to be based on a
comprehensive list of schedule risks and an understanding of how they may impact the project. Understanding the
compromises, assumptions and basis of the analytical methods and what the resulting schedule risk analysis
means are key to developing appropriate risk treatment plans, contingency estimates, and making well supported
value adding project decisions.
This RP is applicable to any industry or project where the CPM approach is used. It addresses considerations for
risk analysis as they relate to the CPM model and not to any integration with cost risk analysis.
Outline
The following is an outline of this RP’s content:

• Minimum Conditions of Satisfaction
• Schedule Model Building
• Summary vs. Detailed CPM Models
• What to Include in a Summary CPM Model
• How to Summarize Schedule Activities
• Logic Tie Considerations
• Constraint Considerations
• Schedule Work Calendars
• Assignment of Resources and Costs
• Probabilistic and Conditional Branching

64R-11: CPM Schedule Risk Modeling and Analysis: Special Considerations 2 of 14
May 2, 2012
• Alternative Scenarios
• Associating Risks to Activities
• Linking Risk Registers to Activities
• Duration Ranges on Activities
• Combination of Linked Risks and Duration Ranges
• Establishing Correlations
• Interpreting Results
• Typical Simulation Outputs
• Merge Bias
• Duration Cruciality vs. Schedule Sensitivity
• Conclusion
Background
The application of schedule risk analysis for projects is a growing practice. The practice is used at various stages of
project development and execution. The practice serves multiple purposes, including:
• Assessing the potential variability in project duration (in individual activities, various sub-networks, or the
overall completion milestone) resulting from identified project risks;
• Assessing how risks, including threats and opportunities, in the project schedule may influence project
economics;
• Understanding which paths in the schedule have the highest probability of influencing the schedule
completion or key milestones; and
• Understanding which risks have the most influence on overall schedule variability.
While multiple methods exist for conducting a schedule risk analysis, the most common method utilizes a CPM
network model. There are multiple categories of schedule risks, from duration uncertainty to specific impact
events to CPM network risks. The project risks with the potential distribution of impacts are either explicitly linked
to the schedule activities or more subjectively reflected in duration variability that is assigned directly to each
activity. Risks that are inherent in the network logic may show up as merge bias or as weaknesses in the CPM
network. A Monte Carlo type simulation of the schedule model is then conducted, the results analyzed and
feedback provided to the project team and/or management for further action/revision/acceptance. While there
are other ways to handle the various categories of risks, the Monte Carlo simulation can be used for all types.
Without use of a Monte Carlo simulation, there are ways to provide schedule risk analysis. Specific impact event
risks can be modeled directly in the CPM network by use of what-if scenarios, performed individually or
progressively (by adding one event after another). Network risks may be evaluated through detailed and thorough
review of the schedule from a technical network perspective. Duration uncertainties cannot easily be evaluated
except through analysis such as three point duration estimates. Effective use of project controls practices will
provide proactive engagement so that projects will not allow sequential activities to progress at their pessimistic
durations.
What-if scenarios developed directly from the CPM model without use of a Monte Carlo simulation, are performed
using the same process as described in AACE International Recommended Practice No. 52R-06 “Time Impact
Analysis – As Applied in Construction”.
Various software products are available which support Monte Carlo simulation and risk analysis using a CPM
network model. Each product is different based on which scheduling programs they support, how risks are linked
to activities, and how the results are presented. This recommended practice is not based on any specific software
product but intended to support general practices with any available standard schedule risk analysis software.

May 2, 2012
The key to deriving the best value from schedule risk analysis is to use a schedule risk model that correctly
represents the project plan while also properly reflecting how the potential project risks may impact the schedule.
Practicality requires tradeoffs that must be considered; these are covered in this RP.
This recommended practice should be considered complementary to RP 57R-09, “Integrated Cost and Schedule
Risk Analysis Using Monte Carlo Simulation of a CPM Model”, which discusses how to conduct a combined cost and
schedule risk assessment in a single CPM model. This recommended practice addresses considerations for
conducting only a schedule risk analysis using the CPM model, not combining it with cost risk analysis.
RECOMMENDED PRACTICE
Minimum Conditions of Satisfaction
To properly conduct a schedule risk analysis and provide quality results, the project team needs to have developed
the following items consistent with the current stage of project development:
• A development of scope of work, execution plan, cost estimate, and schedule that are consistent with
each other;
• A quality schedule that includes activities for all project scope; has realistic durations for activities; has
logic links that appropriately reflect the sequence of work and the dependences between activities;
properly developed with full allowance for available, especially limited, resources and space for those
resources to work; and realistic critical and near-critical paths;
• A team-approved schedule (where schedule readiness is acknowledged only after project team buy-in
around such factors that include, but are not limited to, the validity of precedence logic, extent of broken
logic, missing logic or open-ends, logic density or logic to activity ratio, and the number or type of
constraints);
• A risk register with a comprehensive list of all identified project-specific risks, an understanding of the
likely impact of each of those risks, and risk response actions for each risk; and
• An understanding of the systemic risks, if not captured in the risk register, which may impact the project’s
performance, including schedule completion.
If these items do not exist at the start of a schedule risk analysis, the team will need to address those issues and
either resolve them as a part of the analysis, determine how the gaps should be reflected in the analysis, or defer
the analysis until the items are satisfactorily modeled in the schedule.
Schedule Model Building
Summary versus Detailed CPM Models
Two basic approaches are used in formulating a CPM schedule model for risk analysis. One uses the detailed
project schedule in its entirety, which may involve thousands of activities. The other uses a summary CPM model
to represent the detailed project schedule, typically with only a few hundred activities or less. There are
proponents to each approach. Risk analysis benefits from team input. Also, conducting a schedule risk analysis
with a summary model is likely to be more understandable to the team while making better use of their time (and
to take less time overall) than with a detailed project schedule. Summarized schedules are also useful for
addressing more strategic risks and considerations. Use of a detailed project schedule may be more appropriate if
it is expected to be incorporated into ongoing schedule management (i.e., a definitive diagnostic, tactical or
control approach) for the project.

May 2, 2012
While impact event risks may be evaluated easily and effectively with risk analysis performed on a summary
schedule, the full impacts to activities in the detailed schedule will not be fully assessed from a duration
uncertainty and logic network risk standpoint. The risk assessment of these schedule components will only be fully
evaluated with use of a detailed CPM schedule.
In either case, the schedule model needs to be properly formulated to reflect how the duration risks for each of
the activities, or groups of activities, flow through to project completion or key milestones. Modifications are likely
to be required in the detailed project schedule (risk analysis used this way may result in more robust, risk-tolerant
schedules). These considerations are discussed in this document.
What to Include in a Summary CPM Model
If a summary CPM model is being utilized, the quality of the result is dependent upon how well the summary
model reflects the important logic elements and activity inter-relationships in the detailed project schedule. Not all
activities from the detailed schedule need to be reflected in the summary schedule; however it should include all
activities that have the potential of significantly impacting the project completion date or any key milestone being
analyzed. At a minimum, this approach would incorporate the following activities or schedule paths:
• Critical Path Activities: All activities on the critical path that progresses from the starting point of the
analysis to schedule completion need to be included in the model.
• Near Critical Path Activities: All activities that have total float values close to the critical path need to be
included in the model. The definition of “near critical path activities” will vary depending on the overall
duration of the project. One guideline is to include all activities with total float values that are less than
five percent of the overall schedule duration. The impact to float values by calendars and constraints must
be understood when determining which sequences to model. The expected size of the risk impact being
analyzed should also be taken into account when limiting the criticality of activities.
• High-Risk Activities: All activities that have a high absolute level of duration variability due to the
identified project risks need to be included in the model, even if they are associated with higher total float
values.
How to Summarize Schedule Activities
If a summary CPM model is being utilized, two or more activities in the detailed schedule are often combined to
reduce the number of activities required in the summary model. This summarization is similar to the practice of
having a “Level 2” summary schedule versus a “Level 3” detailed schedule.
Activities should be combined when practical. Activities should not be combined if the risks that face each of the
activities are significantly varied and combining the activities would result in a misrepresentation of the risk
impacts. Basic activity combinations include sequences of activities or parallel activities.
Combination of a Sequence of Activities: A string of activities that have no (or insignificant) logic links with other
activities at intermediate steps may be combined. Examples include such activities as: 1) procurement activities to
solicit, prepare, evaluate, and award supply contracts; 2) forming, placement of rebar, and placement of concrete
for a foundation; and 3) detailed commissioning steps for a system.
Combination of Parallel Activities: Detailed schedules may contain multiple activities that occur in the same
timeframe and are integrated - these may be combined. Examples include such as activities as: 1) procurement
activities for different pumps that will be issued under one purchase order; 2) a contractor’s rough-in of different
piping systems in the same building area; and 3) installing multiple foundations in the same area.

May 2, 2012
Logic Tie Considerations
How duration variability in any sequence passes from one activity to another depends on the logic ties that are
used to connect these two activities. The types of logic ties used in the schedule risk model will have a significant
impact on the results of the model in a Monte Carlo simulation. The following describes how duration variability is
treated between activities based on the type of logic tie being used. The description is based on the absence of any
other logic ties that may impact either the preceding or succeeding activity. The triangles included in the figures
below represent the duration variability or distribution of the activity. The placement of the triangle indicates
whether the finish date of the triangle will vary or the start date of the activity will vary.
Finish-to-Start Logic Ties: In general, a finish-to-start logic tie is preferred in a schedule risk model. It allows the
duration variability from the preceding activity (A) to impact the start of the succeeding activity (B), which then
adds its own duration variability. See Figure 1.
B
Figure 1 – Finish-to-Start Logic Tie
Finish-to-Finish Logic Ties: The finish-to-finish logic tie will negate the impact of the duration variability in the
succeeding activity. The duration variability of the preceding activity (A) impacts when the succeeding activity (B)
will complete. If the succeeding activity has no other preceding activities, then the duration variability of that
activity (B) will only alter when the activity will start and will not impact when the activity will finish. See Figure 2.
B
Figure 2 – Finish-to-Finish Logic Tie
Start-to-Start Logic Ties: The start-to-start logic tie will negate the impact of the duration variability of the
preceding activity (A) on the succeeding activity (B). The duration variability on the preceding activity will impact
when that activity will finish, but not impact when that activity will start. Since the succeeding activity is linked to
the start of the preceding activity, the duration variability of the preceding activity will have no impact on when
the succeeding activity will start. See Figure 3.

May 2, 2012
B
Figure 3 – Start-to-Start Logic Tie
Start-to-Finish Logic Ties: The start-to-finish logic tie, which is rarely used in scheduling practice, negates both the
variability of the preceding activity (A) and the succeeding activity (B). The logic tie to the start of the preceding
activity negates the variability of the preceding activity on the start or finish of the succeeding activity. The logic
tie to the finish of the succeeding activity impacts when that activity will complete. The duration variability of the
succeeding activity, in the absence of other logic ties, will then only impact when that activity will start and have
no impact on its completion date. See Figure 4.
B
Figure 4 – Start-to-Finish Logic Tie
Logic Tie Lags: At the time this recommended practice was issued, no software products allowed assigning
variability to time lags placed in logic ties. If a start-to-start logic tie has an X-day lag assigned to it, the succeeding
activity will start X days after the preceding activity started. In actual practice the lag to the start of the succeeding
activity has some risk to it due to multiple factors. Use of lags in logic ties should be minimized or eliminated. One
alternative to use of a lag in a logic tie is to segment activities so that finish-to-start logic ties with no lags can be
used. See Figure 5 for an alternative to a start-to-start logic tie with a lag.
A1 A2
B
Figure 5 – Alternative to Start-to-Start Logic Tie with Lag, Using Finish-to-Start Logic Ties
Another alternative is to create a separate task to represent the lag.
The general rule of thumb is that every activity in the schedule needs at least one predecessor that ties to the start
of that activity and one successor that ties to the finish of that activity. Redundant logic, links between activities
that are already linked through other activities, should be avoided to help easily trace paths through the schedule.

May 2, 2012
Constraint Considerations
Recommended industry practice in the development of project schedules is to have minimal use of constraints on
activities in the schedule. Nevertheless, some project schedules have a large number of constraints on activities.
In many cases the constraints reflect project preferences rather than actual constraints. If the detailed project
schedule is utilized as the schedule model, the constraints included in the model may impact the schedule risk
results by negating or limiting the variability in associated activities. The user needs to review each constraint
included in the network to determine if it should be deleted, modified, or retained. Any constraint that restricts
the finish date of an activity from going beyond the constraint date should be removed. One of primary goals of
the schedule risk analysis is to determine if the schedule is likely to go beyond the desired completion date or key
milestone dates. Leaving finish constraints in the model will negatively impact the quality of the simulation.
Use of constraints in a detailed project schedule will also alter total float values for affected activities. If a summary
schedule model is being created, the user needs to review each constraint in the detailed project schedule to
understand how it is being used and how the total float values are impacted to determine what activities need to
be reflected in the summary model. Constraints should not be included in the summary schedule if at all possible,
except to potentially constrain the start of an activity due to an external requirement. For example, “activity X
cannot start until after March 1st when the spring thaw starts.” In such instances, constraints may also be
addressed through other means, such as by the use of a special calendar or dummy activities that provide the
required delay/lag in the schedule.
Different software packages have various types of constraints that may impact the schedule risk analysis
differently. If constraints are used in the schedule model, the user needs to understand how those constraints may
impact the results of the simulation.
Schedule Work Calendars
Detailed project schedules may contain multiple work calendars to reflect different work schedules for different
trades or phases of work and to reflect restrictions on when certain work can be performed. Total float values for
an activity are calculated based on the work calendar assigned to the activity. This is an important consideration
when determining what activities need to be included in a summary model. Certain activities may have higher or
lower total float values than others but have similar “criticality” based on longest path determination. For
example, an activity that is on a five-day work week schedule will have five days of total float for each week, while
another activity on a four-day work week schedule will have four days of total float for each week.
If practical, a schedule model should use a single work calendar for all activities. This practice will facilitate
evaluating different activity paths against each other for how close they are to being critical. Reflecting detailed
project activities that use different work calendars in a summary model that uses a single work calendar may
require translation of the activity duration to match the original detailed work schedule. For example, a four day
activity on a four-day work week calendar will have a duration of five days in a summary model that uses a five-day
work week as the common work calendar for all activities.
Assignment of Resources and Costs
Certain projects have to deal with specific limitations that need to be addressed by the schedule risk analysis, such
as limited availability of resources or cashflow limitations per period. Certain software programs will allow the user
to set those limitations as a part of the risk simulation settings.
In those cases, the schedule must be loaded with all the resources or costs that will be subject to the limitation. If

May 2, 2012
manpower resources are limited, either due to available space on site or available workers in the local market, all
of the required resources need to be loaded on the affected activities in the schedule model. If cashflow
limitations have been placed on the project, all the project costs need to be loaded in the schedule model. To
properly load the resources or costs on the schedule model activities, the model may need to be expanded to
represent not only the risk sensitive sequences, but all work requiring limited resources or all project work to
assign basic cost functions. Once the model has been expanded and loaded with the resource or cost, the resource
or cost profile of the model should be compared against the original project schedule to confirm that the model
provides a reasonably sound representation of the original schedule.
When running the Monte Carlo simulation, if the resource or cost limitations are exceeded, the program will level
the schedule according to priorities that the user establishes. The results will reflect the impact of schedule
leveling to avoid the resource or cost limitations being exceeded. This may mean that the CPM model no longer
calculates accurate total float values, so the user should be careful in depending on this resource-leveled schedule
for further analysis.
Probabilistic and Conditional Branching
Most software products allow establishing probabilistic branching in the schedule model. Some software products
allow conditional branching. These features may be needed in a schedule model to properly reflect how the
project will progress, depending on certain things that may occur and are particularly useful in assessing specific
impact event risks.
Probabilistic branching is used when certain activities may or may not occur depending on a particular event. A
common example is a test. If the system fails a test, some rework and retest will be required. If the system passes
the test, these activities will not be required. The probability that the test will fail is entered as the percentage of
simulation runs that will go through the rework and retest activities. The balance of the simulation runs will go
straight from the initial testing activity to the post testing activities.
Conditional branching is used to address a response to a specific occurrence. It can be considered an “if, then”
approach. Project teams often develop work around plans in the event some activity takes longer than planned or
if some other event occurs. Certain software products will allow this type of situation to be built into the schedule
model. A common example is based on the delivery of some key piece of equipment: “if the delivery of this tower
is later than X, we have developed an alternate erection sequence for this area.” Once the delivery time of the
tower goes beyond a certain date or takes more than X number of days, this alternate sequence of activities is
triggered. The inclusion of conditional branches may be the result of iterative application of the risk analysis model
(i.e., supports development of “contingency plans”). If a risk analysis run results in an unacceptable outcome, a
conditional branch may be added that results in a more risk-tolerant plan.
When a major risk event occurs, the risk response often results in changed schedule logic after that event. This is a
challenge to model (i.e., allowing for conditional branches for each risk at each possible time of occurrence is not
practical). This is a situation where some individuals may advocate use of non-CPM expected value risk analysis
methods where subjective consideration of alternate paths is assumed.
Alternative Scenarios
In the early development stages of projects, alternative execution strategies may be developed and the team may
evaluate the different scenarios using schedule risk analysis. Alternate scenarios can also be used in lieu of
conditional branching if the software does not support that feature. Using alternative scenarios involves creating
different versions of the schedule model, with only the necessary activity changes to reflect the alterations from

May 2, 2012
the base scenario to the alternative scenario. This is often facilitated by using summary schedules. Certain
software products will support a comparison of the results of different versions to highlight the variations. The
variability in outcomes between the different scenarios can be evaluated to determine if one scenario offers a
higher probability of meeting a specific target date or has less variability in overall completion or a key milestone.
Associating Risks to Activities
A first principle of risk analysis in support of risk management is that impacts must be as explicitly related to the
risks as possible. Various software products treat the linking of risks to activities differently - some differences are
slight and some are major. Two primary methods are as follows:
• Enter a list of risks in a register in the software and then link each risk to each of the activities that would
be impacted by that risk; or
• Establish a range of potential durations that reflects the range of potential impacts from the combined
risks that might affect that activity.
Regardless of how the risk impacts are assigned to activities in the model, the user needs to have a good
understanding of the how the durations in the original schedule were developed. Activity durations that are based
on very limited information will have more variability than activities that are based on more refined information,
such as direct manhours, crew size, and a detailed execution plan. That higher level of uncertainty may or may not
be reflected in the risk register. Another consideration is the confidence level or bias that was used to develop the
schedule. Some projects have very “aggressive” schedules that reflect optimistic durations on the premise that
everything will go well. Other projects have very “conservative” schedules that include time allowances for various
undefined impacts resulting in more pessimistic durations. Understanding this characteristic of the underlying
project schedule is essential to determining the range of potential impacts due to the risks. Quantitative validation
or comparison to historical metrics should be performed before the risk analysis to ascertain the degree of bias, if
any.
In either case, the user will need to select an appropriate probability distribution function (PDF) for the range of
impacts. The selection of the PDF should be based on a realistic assessment of how the risks will impact the
activities.
The other consideration in determining the potential impacts of risk is how the project team may respond if certain
activity durations start to extend. If the project is schedule driven, the project team may be willing to expend
additional funds to add labor to compress work activity durations or pay expediting costs to limit delivery delays. If
the project is cost driven, the project team may not be willing to expend additional funds to compensate for any
activity extension. It is important to make note of the risk responses that have been assumed in the analysis and
ensure that these responses are accepted by management and are consistent with project objectives.
Linking Risk to Activities
The first step in linking the identified project risks to the activities in the schedule model is to review the itemized
risks. The risk register needs to be reviewed for the following attributes:
• The register needs to be comprehensive and include risks for all remaining phases and aspects of work on
the project.
• The register needs to include not only project specific risks, but also systemic risks that may affect all
projects in the organization.
• Each risk should be unique, i.e. not have overlapping causes or impacts. Two risks in a register that are
both due to the same root cause may need to be combined as one risk.

May 2, 2012
The next step, if not already accomplished, is to assign the probability of occurrence and range of potential impact
associated with each risk. This step requires extensive input from the project team, the owner’s organization,
contractors, and other stakeholders.
Finally, each risk needs to be linked to every activity that would be affected by that risk, according to the method
required by the software. Each activity should have at least one risk assigned to it, or utilize duration ranging, as
discussed in the next section. Systemic risks tend to be overarching and may impact all activities.
Duration Ranges on Activities
In the case that a single duration range is assigned to an activity in the model (as opposed to linking risks to
activities), that duration range needs to account for all uncertainties (threats and/or opportunities) that may drive
the activity duration to be longer or shorter. In many cases, a practical approach is to assign three durations to
each activity: an optimistic duration, a most likely duration, and a pessimistic duration. This approach is similar to
the PERT scheduling approach. The most likely duration may not be the same as the planned duration. Then a PDF
is assigned to each activity to best reflect the profile of the potential risk impacts. A beta or triangular type
distribution is commonly used for most duration ranges.
The optimistic and pessimistic durations may reflect the absolute extremes, P0 and P100, or a less extreme point,
such as P10 and P90. Many people are unable or unwilling to acknowledge the absolute best case scenario and
absolute worst case scenario which may be quite extreme. As a result, using some point within the absolute
extremes is typically the preferred approach. If extremes are considered (and it is recommended that all potential
impacts be considered just short of absurdity), make sure the PDF does not artificially accentuate the impact.
The most important concern in this approach is that the probabilistic duration range assigned to an activity has to
consider all risks that may impact it. The risk register needs to be reviewed to determine which risks need to be
considered for each activity. If a number of independent risks may impact a single activity, the range and PDF
should not reflect the additive impact of all the optimistic durations and the additive impact of all the pessimistic
durations, but something less than that. With multiple independent risks, it is rare that all would occur at their
extreme ranges concurrently. Also, in any given simulation of the project, some of the pessimistic risk impacts may
occur but be offset by some optimistic risk impacts. As a consequence, combining the potential impacts of all the
risks into a single range is difficult to assess. The resulting probabilistic duration range will be something more than
the single greatest risk but less than the impact of all risks combined. An exception to this is when so many risks
occur that their impact has a compounding effect. This situation is more typical with very large, complex projects.
The risk analyst must balance these considerations. Given the subjectivity of risk impact quantification, some
advocate the use of non-CPM expected value schedule risk analysis methods where subjectivity is addressed by
the methodology. (Not covered in this RP. See AACE International Recommended Practice No. 44R-08 “Risk
Analysis and Contingency Determination Using Expected Value”.)
If an individual risk event is identified for an activity that is not considered in the base schedule, it may be modeled
by including an activity in the risk schedule with a duration of zero. This activity can then be assigned a probability
of occurring and a duration range. The base schedule model reflects the schedule without any impact of that
event. If the identified risk activity occurs in a simulation, the duration will be selected from the duration range
and PDF assigned to that activity. A common use of this is to simulate weather impacts on the schedule.
Combination of Linked Risks and Duration Ranges
On certain projects, the project risk register is not reflective of all known risks on a project. The register may only

May 2, 2012
reflect project-specific risk events and not include either systemic risks or general uncertainty risks. In those cases,
it may be appropriate to use a hybrid of the two methods described above. The project-specific event risks can
each be assigned to the appropriate activities in the schedule model. The systemic risks and general uncertainty
may be reflected by activity ranges that are assigned directly to each of the activities. In using this approach, the
important consideration is that the duration ranges assigned to each of the activities do not duplicate the potential
impacts of certain risks in the risk register, resulting in a duplication of impacts.
By definition, systemic risks are overarching and only impact the bottom-line or overall duration. Further,
empirically-based parametric methods are recommended for quantifying their impact. The output of a separate
parametric systemic risk analysis can be incorporated into the CPM model by including the systemic risk as an
activity at the completion of the model, using the PDF output of the parametric model as its duration. Systemic
risks also have a compounding affect that should be considered. For example, a risk event that is innocuous in a
disciplined project management culture may introduce chaotic results where discipline is weak (which partially
explain the poor track record of complex mega-projects).
Establishing Dependency and Correlations
It is important to evaluate and describe each risk’s relationship with other risks when a risk is identified. In other
words, it is important to establish risk conditionality. Risk conditionality has two components: (1) risk dependency,
and (2) risk correlation.
Risk dependency expresses the relationship between risks as events. There are 3 situations: (1) mutually inclusive,
(2) mutually exclusive, and (3) independent. The default of most software is that risks are independent of each
other.
Mutually inclusive events:

If risk X occurs only and always when risk Y occurs than we have a total mutually inclusive relationship. When risk X
may occur only when another risk Y occurs then we would say that they are in partial mutually inclusive
relationship. In this case we must define the probability that risk X occurs when risk Y occurs.
Mutually exclusive events:

If risk X occurs only and always when risk Y never occurs than we have a total mutually exclusive relationship.
When risk X may occur only when another risk Y never occurs then we would say that they are in partial mutually
exclusive relationship. In this case the risk X must have it own probability of occurrence that may be applied when
risk Y does not occur.
Independent events:
If there is no relation between the occurrences of risks X and Y they are considered to be independent.
Correlation establishes how a risk’s impact is sampled during the simulation process. Correlation applies only when
non-independent risks occur on the same iteration. Correlation is represented by a “correlation factor” that may
take any value from -1 to +1 [-100% to +100%].
Negative correlation occurs between two related risks in which an increased level of risk in one event will result in
a decreased level in the other event; e.g. a relationship between two risks in which one increases as the other
decreases, and vice versa. Perfect negative correlation would be represented by a correlation factor of -1.
Positive correlation occurs between two related risks in which an increased level of risk in one event will also result
in an increased level in the other event; e.g. a relationship between two risks in which both increase or decrease
together. Perfect positive correlation would be represented by a correlation factor of +1.

May 2, 2012
The default of most software is that risks have no correlation, i.e. a correlation factor of 0.
The assignment of risk dependency and correlation between two or more activities should be performed according
to the software product instructions. This method varies between different products.
Failure to assign dependencies and correlations to risks and activities where appropriate will result in the schedule
variability of project completion or certain key milestones being understated. This step is critical in developing a
viable schedule model that accurately reflects the variability of the project schedule.
Interpreting Results
Typical Simulation Outputs
Each software product has a different approach to presenting the results of the Monte Carlo type simulation; yet
each presents the same basic information to the user for analysis. Typical results include the following:
• A histogram and cumulative probability curve of dates for the start and/or finish date of each activity in
the schedule model;
• A criticality percentage number that indicates how often an activity was on the critical path in all the
simulation runs of the model; and
• Duration or schedule sensitivity scores that indicate how much an activity may have contributed to the
overall variability of the project or some key milestone.
Most software products will produce graphical tools for use in sensitivity analysis of various project constraints
called “tornado diagrams”. These tornado diagrams highlight those activities that are identified as critical most
often, the activities with the most variability, and/or the risks with the most impact on the overall project duration.
These tornado diagrams are critical for focusing attention on those elements of the schedule that are subject to
the most risk and those risks which contribute the most variability to schedule duration.
One of the important steps in analyzing the risk analysis results is to perform a quality review of the results and the
model to ensure that the results make sense and errors did not occur in the development of the schedule model.
For example, for results to pass the common sense test, the risk analyst should be able to provide unambiguous
answers to reasonable questions such as:
• Why does the distribution have such long/short tails?
• Why are the max/min values so high/low?
Merge Bias
In the performance of a schedule risk analysis, the phenomena of “merge bias” impacts most project results. This
CPM network risk impact is typically poorly understood by project teams and management, the intended
beneficiaries of the analysis. Simply put, merge bias is the impact of having two or more parallel paths of activities,
each with its own variability, merge at one milestone. This schedule logic structure has a significant impact on
probabilistic dates for the milestone and any of its successors. The number of merging parallel paths and the level
of overlap between them produce an increasing merge bias impact to the schedule.
The expected completion date of a milestone that has two or more similar parallel paths merge into it is later than
the expected completion of each path. In a project schedule, the milestone date will always be based on the path
that finishes last. In other words the milestone will always reflect the worst case of the multiple paths that go into
that milestone. The resulting probability profile of a milestone with two or more merging paths will be to the right

May 2, 2012
of the probability profile of each path.
In analyzing the risk results, this concept needs to be understood to be able to track the increasing variability in the
schedule model from the beginning of the model to project completion. However, keep in mind that in building a
CPM-based risk analysis model, there are many subjective considerations such as the PDF profile, the cumulative
impact of multiple risks on an activity and the impact of risk responses on logic.
Duration Cruciality vs. Schedule Sensitivity
Some software products offer more than one method for characterizing how each activity can influence the
variability in the overall schedule. The purpose of this data is to define which activities have the most influence on
the overall schedule or key milestones. The different methods for calculating this influence can result in different
rankings of activities and risks. This difference could result in different priorities being established in risk response
actions for different activities.
Two common methods for ranking activities are duration cruciality index (DCI) and schedule sensitivity index (SSI).
Each method is typically calculated as follows:
Standard Deviation of Activity Duration (or Risk Event) x Activity Criticality Index
SSI =
Standard Deviation of the Overall Project Duration
DCI = Duration Sensitivity of Activity x Activity Criticality Index
The schedule sensitivity index method is a simpler method that can be equated to a qualitative measure of
probability of occurrence times impact. SSI is only useful for identifying overall project risk drivers because the
standard deviation of the entire project and not the standard deviation between activities is being used.
The duration sensitivity of each activity is based on a detailed calculation to measure the correlation between the
variability of the activity and the overall schedule.
The duration cruciality index method is a complex calculation, based directly on the simulation results in
comparing the duration of an activity to the overall schedule duration in each simulation run. DCI attempts to
show if an activity is not only a driver to another activity, but also whether it is on the overall critical path. It may
be a key driver to another activity, however it may not have an impact on overall critical path.
In most cases, the differences in the top rankings from either method are minor. Typically the same activities will
appear in the top quintile of activities, but the ranking may vary slightly. There are proponents for each method of
measurement, however both methods should be utilized to develop a comprehensive ranking of top activity risks.
Conclusion
A CPM-based schedule risk analysis can be a complicated exercise which can be easily flawed if:
• the project team has not first deemed the schedule mature enough or ready for review;
• the model is not well formulated to reflect the planned flow of work for the project; or
• the model does not realistically reflect how the risks, both project-specific and systemic, may impact the
schedule.
Prior to publishing schedule risk analysis results for the use in any decision analysis, the risk analyst must first

May 2, 2012
ensure the model is not only free from errors but both practical and accepted by the project team as being
reflective of the project.
If conducted properly, a schedule risk analysis can yield valuable predictive powers to mitigate risk events and
impacts as the project progresses. It can help uncover unknown schedule issues, drive decisions, and focus the
team’s actions.
REFERENCES
1. AACE International, Recommended Practice 10S-90, Cost Engineering Terminology, AACE International,
Morgantown, WV, (latest revision).
2. AACE International, Recommend Practice 57R-09, Integrated Cost and Schedule Risk Analysis Using Monte
Carlo Simulation of a CPM Model, AACE International, Morgantown, WV, (latest revision).
3. AACE International, Recommend Practice 42R-08, Risk Analysis and Contingency Determination Using
Parametric Estimating, AACE International, Morgantown, WV, (latest revision).
4. AACE International Recommended Practice No. 52R-06, Time Impact Analysis – As Applied in Construction,
AACE International, Morgantown, WV, (latest revision).
5. AACE International Recommended Practice No. 44R-08, Risk Analysis and Contingency Determination Using
Expected Value, AACE International, Morgantown, WV, (latest revision).
6. AACE International, Recommend Practice No 65R-11, Integrated Cost and Schedule Risk Analysis and
Contingency Determination Using Expected Value, AACE International, Morgantown, WV, (latest revision).
7. Hollmann, John K., PE CCE, Editor, Total Cost Management Framework: An Integrated Approach to Portfolio,
Program and Project Management, AACE International, Morgantown, WV, (latest revision).
8. Valdahl, Jeffrey and Caddell, Christopher, Understanding Merge Bias in Schedule Risk Analysis, AACE
International Transactions, 2010, AACE International, Morgantown, WV, 2010.
9. Hulett, David, Practical Schedule Risk Analysis, Gower Publishing Limited, Burlington, VT, 2009.
10. Kendrick, Tom, Identifying and Managing Project Risk, 2nd Edition, American Management Association, New
York, New York, 2009.
11. Bartlett, John, Project Risk Analysis and Management Guide, Association for Project Management, 2004.
CONTRIBUTORS
Christopher P. Caddell, PE CCE (Author)

James E. Arrow
Chris Beale
Peter R. Bredehoeft, Jr., CEP
Christopher W. Carson, PSP
Dr. Ovidiu Cretu, PE
Michael W. Curran
Larry R. Dysert, CCC CEP
Leonard Enger, CCE
Ambrish Gupta, CCE
Dennis R. Hanks, PE CCE
John K. Hollmann, PE CCE CEP
Shannon A. Katt
H. Lance Stephenson, CCC
Anthony M. Woodrich, CCE

64R-11 - AACE International

Uploaded by

Document Information

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

64R-11 - AACE International

Uploaded by

Copyright:

Available Formats

This Recommended Practice (RP) is brought to you as Total Cost

public service by AACE International, the Authority for Management®

Recommended Practice Recommended Practice

Virtual Library: Since 1976, AACE has

Professional Practice The Online Learning

Recommended Practice Recommended Practice

Recommended Practice Recommended Practice

AACE’s career center Members receive a

Recommended Practice Recommended Practice

CPM SCHEDULE RISK MODELING AND ANALYSIS:

Rev. May 2, 2012

Copyright © AACE® International AACE® International Recommended Practices

The following is an outline of this RP’s content:

Copyright © AACE® International AACE® International Recommended Practices

Copyright © AACE® International AACE® International Recommended Practices

Minimum Conditions of Satisfaction

Schedule Model Building

Summary versus Detailed CPM Models

Copyright © AACE® International AACE® International Recommended Practices

What to Include in a Summary CPM Model

How to Summarize Schedule Activities

Copyright © AACE® International AACE® International Recommended Practices

Logic Tie Considerations

Copyright © AACE® International AACE® International Recommended Practices

Another alternative is to create a separate task to represent the lag.

Copyright © AACE® International AACE® International Recommended Practices

Schedule Work Calendars

Assignment of Resources and Costs

Copyright © AACE® International AACE® International Recommended Practices

Probabilistic and Conditional Branching

Copyright © AACE® International AACE® International Recommended Practices

Associating Risks to Activities

Linking Risk to Activities

Copyright © AACE® International AACE® International Recommended Practices

Duration Ranges on Activities

Combination of Linked Risks and Duration Ranges

Copyright © AACE® International AACE® International Recommended Practices

Establishing Dependency and Correlations

Mutually inclusive events:

Mutually exclusive events:

Copyright © AACE® International AACE® International Recommended Practices

Typical Simulation Outputs

Copyright © AACE® International AACE® International Recommended Practices

of the probability profile of each path.

Duration Cruciality vs. Schedule Sensitivity

DCI = Duration Sensitivity of Activity x Activity Criticality Index

Copyright © AACE® International AACE® International Recommended Practices

Christopher P. Caddell, PE CCE (Author)

Copyright © AACE® International AACE® International Recommended Practices

You might also like