117R-

21

INTEGRATEDCOSTANDSCHEDULE
RISKANAL YSI
SANDCONTI
NGENCY
DETERMI NATI
ONUSI
NG
COMBI NEDPARAMETRI
C
MODELLINGANDMONTE
CARLOSI MULATI
ONOF
ACPMMODEL

Rev
.Augus
t15,2022
 AACE® International Recommended Practice No. 117R-21

INTEGRATED COST AND SCHEDULE RISK ANALYSIS AND

CONTINGENCY DETERMINATION USING COMBINED
PARAMETRIC MODELLING AND MONTE CARLO SIMULATION
OF A CPM MODEL
TCM Framework: 7.6 – Risk Management

Rev. August 15, 2022

Note: As AACE International Recommended Practices evolve over time, please refer to web.aacei.org for the latest
revisions.

Any terms found in AACE Recommended Practice 10S-90, Cost Engineering Terminology, supersede terms defined in
other AACE work products, including but not limited to, other recommended practices, the Total Cost Management
Framework, and Skills & Knowledge of Cost Engineering.

Contributors:
Disclaimer: The content provided by the contributors to this recommended practice is their own and does not necessarily
reflect that of their employers, unless otherwise stated.

Colin H. Cropley (Primary Contributor) John K. Hollmann, PE CCP CEP DRMP FAACE Hon. Life
James E. Arrow, DRMP Dr. David T. Hulett, FAACE
Francisco Cruz Moreno, PE
Copyright © AACE® International AACE® International Recommended Practices
Single user license only. Copying and networking prohibited.

This document is copyrighted by AACE International and may not be reproduced without permission. Organizations may obtain permission
to reproduce a limited number of copies by entering into a license agreement. For information please contact [email protected]
AACE® International Recommended Practice No. 117R-21
INTEGRATED COST AND SCHEDULE RISK ANALYSIS
AND CONTINGENCY DETERMINATION USING
COMBINED PARAMETRIC MODELING AND MONTE
CARLO SIMULATION OF A CPM MODEL
TCM Framework: 7.6 – Risk Management

August 15, 2022

TABLE OF CONTENTS

Table of Contents ..........................................................................................................................................................1

1. Introduction ...............................................................................................................................................................3
1.1. Scope ..................................................................................................................................................................3
1.2. Purpose ...............................................................................................................................................................3
1.3. Background – Source ICSRA Methodologies.......................................................................................................3
1.3.1. Parametric + Expected Value (P+EV) ...........................................................................................................3
1.3.2. CPM-Based ICSRA Assessment of Cost and Schedule Contingencies ..........................................................4
1.3.3. Similarities and Differences Between P+EV ICSRA and CPM-Based ICSRA ..................................................4
1.4. Background – Alternative Classifications of Risk ................................................................................................4
2. Recommended Practice .............................................................................................................................................6
2.1. Combine Parametric and CPM-Based ICSRA Methodologies .............................................................................6
2.1.1. Point at Which Methodologies are Combined ............................................................................................8
2.2. Assess Net Systemic Cost & Schedule Risk Factors using Monte Carlo Simulation (MCS) .................................8
2.3. Map Critical Project-Specific Risk Events to Appropriate Schedule and Cost Elements .....................................8
2.4. Special Considerations for Hybrid Applications ..................................................................................................9
2.4.1. Extreme System Attributes ..........................................................................................................................9
2.4.2. Start-up and Commissioning........................................................................................................................9
2.4.3. Uncertainties that are Project Specific ........................................................................................................9
2.4.4. Impacts to Intermediate Milestones ...........................................................................................................9
2.4.5. Non-linearity ................................................................................................................................................9
2.5. Build and Incorporate Weather Model Where Applicable .................................................................................9
2.6. Run MCS of P+IRA hybrid model.......................................................................................................................10
3. Comparison of the P+IRA Hybrid Method to RP 40R-08 Principles .........................................................................10
4. Summary ..................................................................................................................................................................11
References ...................................................................................................................................................................12
Contributors ................................................................................................................................................................12
Appendix ......................................................................................................................................................................13
A.1. Assess Net Systemic Cost & Schedule Risk Factors Using Monte Carlo Simulation (MCS) ...............................13
A.1.1. Brief Description of Project .......................................................................................................................13
A.1.2. Parametric Analysis of Systemic Cost Growth ...........................................................................................14
A.1.3. CPM-Based ICSRA (IRA) Using Risk Factors Method Analysis of Inherent Risk .........................................15

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 2 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

A.1.4. Assessment of Net Systemic Cost & Schedule Risk Factors by MCS Using SR ...........................................18
A.1.5. Input Distributions .....................................................................................................................................22
A.1.6. Subtraction of Distributions ......................................................................................................................22
A.1.7. Correlation.................................................................................................................................................22
A.1.8. MCS Assessment of Net Systemic Cost Risk Uncertainty Distribution and Risk Factor .............................24
A.1.9. MCS Assessment of Net Systemic Duration Risk Hammocks ....................................................................25
A.1.9.1. Export Percentiles of Both Histograms ...............................................................................................26
A.1.9.2. Combine Histogram Percentiles .........................................................................................................27
A.1.9.3. Upload Combined Percentiles to SR Risk-006 ....................................................................................28
A.1.9.4. Perform MCS to generate inputs for Net Systemic Duration Risk Factors .........................................29
A.1.9.5. Production of Net Systemic Duration Risk Factors (NSDRFs) .............................................................29
A.2. Addition of NSRFs to CPM-Based ICSRA Model ................................................................................................31
A.2.1. Add Net Systemic Risk Factors to CPM-based ICSRA Model .....................................................................31
A.2.2. Map Net Systemic Risk Factors to Tasks and Costs ...................................................................................32

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 3 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

1. INTRODUCTION

1.1. Scope

This recommended practice (RP) of AACE International defines the general practices and considerations for
combining parametric and CPM-based methods for risk analysis and contingency assessment in construction-based
projects. These integrated cost and schedule risk analysis (ICSRA) methods can be combined provided steps are
taken to avoid double-counting sources of risk.

The RP defines the combination of the two methodologies to avoid double-counting of risk, not the source
methodologies themselves. For those, refer to AACE RP 42R-08 [1] and to AACE RP 57R-09 [2].

1.2. Purpose

This document is not intended to be a standard. This document is intended to provide a guideline for combining
parametric and CPM-based ICSRA methods that most practitioners would consider to be good practices that can be
relied on and that they would recommend be considered for use where applicable.

1.3. Background – Source of ICSRA Methodologies

The starting point is a comparison of the methodologies that have been brought together in this RP. Reference is
made to RP 40R-08, Contingency Estimating – General Principles [3]

1.3.1. Parametric + Expected Value (P+EV)

RP 113R-20, Integrated Cost and Schedule Risk Analysis and Contingency Determination using Combined Parametric
and Expected Value [4] describes this hybrid ICSRA methodology. It combines the quantification of systemic risk,
usually the major source of risk in a project, assessed by parametric modelling, with the other source of risk, the
critical project-specific risks normally found in the project risk register. Both are combined by use of Monte Carlo
simulation (MCS) using the technique known as expected value. Once the parametric model is developed [1, 4],
implementing this hybrid methodology is relatively straightforward. It meets one of the key general principles for
contingency estimating in that it explicitly references past project performance [3].

On the other hand, P+EV only quantifies overall cost and schedule contingencies. It does not quantify subproject
cost contingencies nor schedule contingencies of intermediate milestones. Furthermore, the methodology is unable
to optimize risk, it simply assesses it. In addition, the success of the methodology depends on two equally important
requirements:
• The parametric model represents the various major drivers of uncertainty realistically so that assessments
of, for example, scope definition or strengths and weaknesses of the project team produce valid changes
to cost and schedule uncertainty.
• The assessments of the various major drivers of uncertainty are as objective as possible, with minimal
introduction of biases by the participants in the assessments.

The vulnerability of the P+EV methodology to failure to fully achieve each of these requirements can be considered
disadvantages.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 4 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

1.3.2. CPM-Based ICSRA Assessment of Cost and Schedule Contingencies

This class of contingency assessment methodologies includes the CPM-based ICSRA risk drivers method described in
RP 57R-09, Integrated Cost and Schedule Risk Analysis Using Risk Drivers and Monte Carlo Simulation of a CPM Model
[2] and its near-variant the risk factors method (in which the identified risk drivers are not necessarily based on root-
cause analysis and may require correlation between related factors).

These methods include analysis of risk events which may be identified during risk workshops or confidential one-on-
one risk interviews and are compiled in the project risk register.

The methodology involves comprehensive MCS analysis of the full project scope enabling reporting on all cost and
schedule aspects. It also enables optimizing schedule risk (and thus time-dependent cost) through identifying and
ranking the highest schedule drivers and subsequently documenting changes to the schedule to progressively reduce
schedule risk.

Disadvantages include the time and effort required for this full analysis and the fact that the methodology does not
explicitly include past project performance. More definitively, the absence of reference to past project performance
means that the methodology relies almost completely on subjective assessments of probabilities and impact
distributions. This allows for reference to productivity data for some schedule and cost impact distributions and to
data on the frequency of occurrence of some contingent risk events. In these project model cases, most input
distributions and probabilities are likely to be subjective. This typically results in output probability impact
distributions that underestimate the pessimistic “tail” of the distribution, usually substantially. The output
probability impact distributions are also likely to underestimate the optimistic end, compared with real project
outcomes data.

1.3.3. Similarities and Differences Between P+EV ICSRA and CPM-Based ICSRA

Both methodologies start with 100% probability impact uncertainty and add in <100% probability discrete risks or
risk events. Both methodologies refer to the project risk register.

Parametric modelling assesses a particular kind of uncertainty, driven by the system in which the project is being
delivered. The discrete risks incorporated in the P+EV methodology are also of a particular kind, identified as project-
specific and relating to project execution, which usually comprise a subset of the risks in the risk register.

To take this comparison to a more useful point, it is helpful to step back and compare the different ways in which
risk is classified using the two methodologies.

1.4. Background – Alternative Classifications of Risk

CPM-based ICSRA divides project cost and schedule impact risk according to:
• Whether the risk is always present (100% probability), also described as inherent risk, or
• Whether the risk may or may not be present (<100% probability), also described as contingent risk (because
the effect of the risk is contingent on the risk’s occurring).

P+EV ICSRA divides risk according to the way in which it is assessed:

• Risk originating from the project delivery system – so-called systemic risk (as described in section 1.3 of [4])
and
• Project-specific risk events assessed as large enough to be designated critical, where this term is defined in

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 5 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

RP 65R-11 [5] for AACE Estimate Classes 3, 4 & 5 as causing cost variances > 0.5% of project capital
expenditure (capex) or >5% of project profit (a more difficult metric to apply in contingency assessments).
For detailed Estimate Classes 1 & 2, the capex threshold is > 0.2%.

The P+EV methodology uses these critical risk cutoffs on the basis that any project to be included in a parametric
modelling database would be normalized for inclusion by removing such project-specific risks which would otherwise
distort the data. Thus, such risks must be added back in when assessing overall project risk.

These alternative classifications of risk overlap with each other. Understanding how this overlap occurs is key to
combining the two methodologies. A Venn diagram for this purpose is shown in Figure 1.

Figure 1 – Venn Diagram Comparing Risk Classifications

The two classes of risk used in CPM-based ICSRA are shown as the darker discs:
• Inherent Risk - duration and cost uncertainty
• Contingent Risk – risk events

The types of risk expressed in P+EV ICSRA, based on the way they are quantified are shown in Figure 1 as follows:
• Systemic Risk – as expressed in project delivery risk (measured by parametric modelling) usually
encompasses all inherent risk covered by cost estimate and schedule duration uncertainties, plus some of
the risk events in the risk register (namely general or systemic risks and less-than-critical project-specific
risks).
• Critical-scale project-specific risk events are represented by the risk events not overlapped by the project
delivery risk disc in the Venn diagram. Under the P+EV methodology, these critical risk events represent
the risk responses developed by the risk team attending the expected value workshop as part of the P+EV
process.

This RP describes how the net systemic risk not overlapping the inherent risk is quantified and how the risk events
not covered by systemic risk are identified and added into the parametric + CPM-based ICSRA model.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 6 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

2. RECOMMENDED PRACTICE

2.1. Combine Parametric and CPM-Based ICSRA Methodologies

The combination of parametric and CPM-based ICSRA methodologies described by this RP, for convenient reference
is abbreviated to P+IRA. A flowchart is provided in Figure 2 to visualize the complete sequence of events. Note that
escalation and currency risks are not covered in this RP. These could be substantial risks in the large and complex
projects for which this methodology is recommended (see RP 58R-10, Escalation Estimating Principle and Methods
Using Indices [10]; or RP 68R-11, Escalation Estimating using Indices and Monte Carlo Simulation [11]).

Step 8
Step 1 Step 9
Map net systemic risk factors to
Receive / critique project all activities and costs and map Build and incorporate
schedules, estimates & risk critical project-specific residual weather model
register risk events to appropriate in P+IRA model
(including strategy, alignment schedule and cost elements in (as applicable)
and technical quality) P+IRA model

Step 2 Step 7 Step 10

Run base IRA analysis, capture
Conduct schedule and cost distributions for cost & schedule, Run integrated cost & schedule
inherent risk workshops deduct from parametric risk analysis of P+IRA model
(through assignment of risk distributions using MCS; Build including net systemic risk
factors), systemic risk and cost & schedule P10/P50/P90s for factors.
project-specific risk assessment net systemic risk factors for P+IRA Run validation checks and
workshops analysis prepare initial reporting.

Step 6 Step 11
Step 3 Map labor costs to schedule in
structured cost hammocks and If required, incorporate project
Incorporate required schedule team feedback, re-analyze, run
changes, ensure schedule & materials costs to costs sheet.
Include cost risk factors in the quantitative exclusion analysis
estimate are aligned. Sensitivities, prepare final
Review risk register. IRA model to express the base
estimate model report.

Step 4
Step 5
Configure parametric model for
Incorporate schedule risk
systemic risk from workshop
factors to drive duration
inputs;
uncertainty with automatic
Determine probability
correlation in the IRA model to
distributions for cost growth &
form the base schedule model.
schedule slip

Figure 2 – Methodology Flowchart for Parametric + CPM-Based ICSRA (P+IRA) using Risk Factors

Much of the content of Figure 2 is covered in other RPs [1] [2]. Most of the focus of this RP is on Step 7. This is
described in more detail in the logic diagram shown in Figure 3. The methodology was first published in a paper
presented to the 2019 AACE Conference and Expo [6].

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 7 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Logic Diagram for Preparation of Net Systemic Risk Factors in P+IRA (Risk Factors) Methodology
Receive Project Documents & Review
Parametric Modelling IRA Modelling
Review of schedule quality (request
Identify Execution Base Cost & Duration changes if required)
(excluding contingencies); Review input
documents Review estimate & schedule docs, check
for alignment; review risk register; import
into RM database

Prepare Workshop input documents

Workshop, Interviews for Schedule Risk Factors (SRFs) workshop

Parametric model inputs Cost Risk Factors (CRFs) workshop
Risk Register (RR) review workshop

Review & complete parametric Build IRA Base Model:

model inputs Base Schedule
Overlay Base Estimate

Apply SRFs and CRFs and any

Parametric model generates: applicable correlations between RFs
Mean cost & log-normal distribution,
Mean duration & log-normal distbn Duration Run Base IRA Model MCS

Cost Produce Base Estimate distribution;

Base Schedule distribution
NET SYSTEMIC COST DISTRIBUTION NET SYSTEMIC DURN DISTRIBUTION

Is Mean Is Mean Parametric

Pa ra metric Cos t > No No Durn > mea n IRA
mea n IRA Cost? Durn?

Yes Yes
Create MCS model to subtract Create MCS model to subtract
IRA Cost Distribution from IRA Duration Distribution from
Parametric Cost Distribution Parametric Duration Distribution

Run Cost Distributions Run Duration Distributions

Subtraction model MCS Subtraction model MCS

Produce Net Systemic Produce Net Systemic

Cost Distribution Duration Distribution

Take P10/P50/P90 values as 3 point Take P10/P50/P90 values as 3 point

distribution values of Net Systemic distribution values of Net Systemic
Cost Risk Factor (NSCRF) Duration Risk Factor (NSDRF)

Apply NSCRF to all project execution Apply NSDRF to all project execution
time Independ costs in P+IRA model normal tasks in P+IRA model

Map treated risks into P+IRA model

Apply probabilistic weather calendars

Run MCS Analysis

Produce P+IRA Outputs and Report

Figure 3 – Logic Flow Diagram for Preparation and Use of Net Systemic Risk Factors

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 8 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

The upper part of the logic diagram shows the parallel development of the parametric and CPM-based cost and
schedule models. The lower part describes the calculations for the net systemic distributions and risk factors for
cost and schedule. It allows for the possibility that one or both of the cost and schedule parametric model mean
values may not exceed the corresponding CPM-based model mean values. Because the CPM-based models are
mainly or exclusively built on subjectively assessed distributions, it is possible for them to produce larger mean values
than the parametric model if the workshop assessment teams or interview subject matter experts are more
pessimistic than the parametric model. Practical experience has shown that this can happen. It can be argued that
even if the mean parametric cost or schedule duration does not exceed the corresponding CPM-based mean value,
the net systemic risk factors should still be determined and applied because the distribution spread effects will still
be important. As noted in section 1.3.2, it would be unusual for the CPM-based distribution to be broader than the
parametric distribution, especially at the pessimistic end.

2.1.1. Point at Which Methodologies are Combined

Workshop sessions have been held for systemic risk and for reviewing the risk register to identify the critical project-
specific risk events (including assessment of responses to the occurrence of the risks) which are to be included in the
analysis. All required information to complete the parametric model has been obtained and the systemic cost
growth and schedule slip distributions have been generated by the parametric model.

Under the CPM-based IRA methodology, workshop sessions have also been held to identify and scale the risk factors
driving the project schedule activity durations and the project cost line items. During these risk factor workshop
sessions, further potentially critical project-specific risk events may be identified to add to the risk register.

The base estimate has been overlaid on the base schedule and the 100% probability duration and cost risk factors
identified in the workshop sessions applied to enable the base schedule distribution and base estimate distribution
to be produced.

2.2. Assess Net Systemic Cost & Schedule Risk Factors using Monte Carlo Simulation (MCS)

To illustrate the methodology and provide essential details for understanding and using the methodology, an
example project using the risk factors method is described.

Please refer to the Appendix for a detailed set of instructions on quantifying the net systemic cost risk factor and the
net systemic duration risk factor for a project and mapping those risk factors into the project P+IRA model.

2.3. Map Critical Project-Specific Risk Events to Appropriate Schedule and Cost Elements

As per Step 8 of the flowchart in Figure 2, the project-specific critical risk events are then mapped into the P+IRA
model to the activities to which they apply (for duration impacts) or to the time-independent cost elements (for cost
impacts) of treated risks or risk responses agreed to be applied in the event of the risk occurring.

Note that due to the inclusion of net systemic risk impacts, in accordance with the Venn diagram in Figure 1, any
systemic (general) risk events in the risk register are already covered by the net systemic risk factors. The net
systemic risk factors also cover sub-critical scale project-specific risk events.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 9 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

2.4. Special Considerations for Hybrid Applications

Attention is drawn to the section with this heading in RP113R-20, Section 3 [4] covering the P+EV hybrid method.
Similar considerations apply to this hybrid methodology, except where the use of a CPM schedule provides
functionality unavailable in the P+EV methodology. Some comments follow on topics covered in Section 3 of [4].

2.4.1. Extreme System Attributes

Refer to this topic in [4]. Extreme system attributes, if critical to project economics can and should be included as
project-specific risks and mapped into the P+IRA model.

2.4.2. Start-up and Commissioning

This post-mechanical completion phase is excluded from the research underlying parametric modelling because the
uncertainty associated with the phase often exceeds the construction phase. Sometimes it is difficult to exclude this
phase or distinguish it from mechanical completion in the project schedule. It is recommended that net systemic
cost and duration risk factors are applied to this phase also and that explicit project-specific risk events applicable
to this phase are mapped into the P+IRA model where they reach the critical risk threshold.

2.4.3. Uncertainties that are Project Specific

Refer to this topic in [4]. Labor productivity can be readily allowed for in duration risk factors and, if necessary, in
cost risk factors as well. Weather variability is normally allowed for in probabilistic weather modelling, covered in
Section 2.5 below. Extreme weather events that go beyond the weather modelling and exceed critical risk thresholds
are covered by project specific risk events.

2.4.4. Impacts to Intermediate Milestones

Such situations are where P+IRA has a clear advantage over P+EV. Project specific schedule risk impacts to
intermediate milestones are readily mapped to the CPM logic preceding the intermediate milestones. Schedule float
and logic determine the effective impact of such schedule risks.

2.4.5. Non-linearity

This topic in [4] covers the known issue that weak project systems and high complexity exacerbate the impact of
major risk events. If the net systemic risk factors are high, this could require major risks impacts to be increased at
the pessimistic ends of their impact distributions. Another possibility is placing high positive correlations on
combinations of risk event impacts. Explicit non-linear effects are beyond the scope of this RP.

2.5. Build and Incorporate Weather Model Where Applicable

Step 9 of the flowchart in Figure 2 allows for the inclusion of probabilistic weather calendars in the P+IRA model
where historical weather records show significant seasonal variation in conditions that could reduce availability of
work periods for construction activities for some project disciplines. This allows for non-linear changes in project
productivities as delays move project activities into more adverse weather conditions.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 10 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

2.6. Run MCS of P+IRA hybrid model

In accordance with Step 10 of the flowchart in Figure 2, once the hybrid model is set up, CPM-based ICSRA can be
performed as per RP 57R-09 [2].

3. COMPARISON OF THE P+IRA HYBRID METHOD TO RP 40R-08 PRINCIPLES

RP 40R-08, Contingency Estimating – General Principles, provides objective principles against which one can assess
the suitability of a contingency estimating method [3]. The following are the RP’s general principles that any
methodology developed or selected for quantifying risk impact should address:

• Meet client objectives, expectations and requirements

• Part of and facilitates an effective decision or risk management process (e.g., TCM [7])
• Fit-for-use
• Starts with identifying the risk drivers with input from all appropriate parties
• Methods clearly link risk drivers and cost/schedule outcomes
• Avoids iatrogenic (self-inflicted) risks
• Employs empiricism
• Employs experience/competency
• Provides probabilistic estimating results in a way the supports effective decision making and risk
management

Table 1 summarizes how the P+IRA hybrid method performs in respect to the 40R-08 principles.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 11 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

General Principles Hybrid P+IRA Methodology Characteristics

Where an acceptable quality schedule is available, along with the project
estimate and risk register, P+IRA can be applied using a parametric model
Meets client objectives and
configurable to the client’s organization and project. Client objectives for
requirements
realistic risk and contingency assessments have been demonstrated to be
met.
Parametric modelling aligns with industry phase-gate scope development and
decision-making processes. Identifies and models deficiencies in the project
Part of a risk and decision system process. The mapping of project-specific risk events to appropriate
management process schedule and cost elements is a fully developed process for applying risks at
the project level from the risk register to the activity and cost item level that
automatically adjusts for project logic and float.
Can be used in projects from concept phase onwards, where estimates,
Fit-for-use schedules and risk registers have been defined, using appropriate software
(i.e., spreadsheet and capable CPM-based ICSRA software).
Systemic risk drivers are explicitly identified and rated (including bias), along
Starts with identifying risk
with driving risk factors. Critical project-specific risks are individually assessed
drivers
and mapped into the schedule model.
Driver-to-outcome linkages are explicit for all risk types. The review of risk
events ensures that risk treatments to reduce probabilities of occurrence and
Links risk drivers and
impacts have been accounted for in the project estimate and schedule, while
cost/schedule outcomes
risk responses focus on planning for mitigations incl. cost/schedule trading.
Supports joint cost and schedule confidence levels (JCL) developed by NASA.
The parametric model incorporates only those systemic risks found to be
Avoids iatrogenic (self- statistically significant by empirical research. The risk review and mapping
inflicted) risks process only quantifies critical project-specific risks and reduces the quantity
of risk events handled.
Parametric, multi-variable linear regression (MLR) based modeling is explicitly
Employs empiricism based on historical experience and may be calibrated to the user’s
experience.
P+IRA requires experienced risk analysis facilitation. Risk factor-driven CPM-
Employs experience
based ICSRA method requires experienced planner/scheduler and estimator
/competency
inputs.
Provides probabilistic All integrated methods covered by AACE RPs produce probabilistic outcomes.
estimating results Supports JCL evaluations.
Table 1 – P+IRA Hybrid Method versus RP 40R-08 General Principles

4. SUMMARY

The P+IRA hybrid methodology combines the realism of data from past project performances with the full analysis
and schedule risk optimization capabilities of CPM-based ICSRA.

Advantages: The use of risk factors improves the speed and responsiveness of the P+IRA modelling process and the
ability to identify and optimize driving risks. Its usefulness for large and complex projects is particularly emphasized.
Sophisticated parametric models properly used introduce needed quantification of systemic risk, the largest
contributor of cost and schedule risk in most projects. While complex, the methodology is no more difficult than
CPM-based ICSRA but benefits from the explicit inclusion of empiricism.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 12 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Disadvantages: The methodology is the most complex to use for contingency assessment and requires the most time
and effort. It works best where the inputs are of good quality, such as for project sanction gates. It is less justifiable
for small projects and for those in the early stages of project development.

REFERENCES

[1] AACE International, Recommended Practice No. 42R-08, Risk Analysis and Contingency Determination Using
Parametric Estimating, Morgantown, WV: AACE International, Latest revision.
[2] AACE International, Recommended Practice No. 57R-09, Integrated Cost and Schedule Risk Analysis Using
Risk Drivers and Monte Carlo Simulation of a CPM Model, Morgantown, WV: AACE International, Latest
revision.
[3] AACE International, Recommended Practice No. 40R-08, Contingency Estimating – General Principles,
Morgantown, WV: AACE International, Latest revision.
[4] AACE International, Recommended Practice No. 113R-20, Integrated Cost and Schedule Risk Analysis and
Contingency Determination using Combined Parametric and Expected Value, Morgantown, WV: AACE
International, Latest revision.
[5] AACE International, Recommended Practice No. 65R-11, Integrated Cost and Schedule Risk Analysis and
Contingency Determination Using Expected Value, Morgantown, WV: AACE International, Latest revision.
[6] C. H. Cropley, "Combining Parametric and CPM-Based Integrated Cost-Schedule Risk Analysis," in AACE
International Transactions, Morgantown, WV, 2019.
[7] H. L. Stephenson, Ed., Total Cost Management Framework: An Integrated Approach to Portfolio, Program
and Project Management, 2nd ed., Morgantown, WV: AACE International, Latest revision.
[8] J. K. Hollmann, Project Risk Quantification, Sugarland, TX: Probabilistic Publishing, 2016.
[9] AACE International, Recommended Practice No. 10S-90, Cost Engineering Terminology, Morgantown, WV:
AACE International, Latest revision.
[10] AACE International, Recommended Practice No. 58R-10, Escalation Estimating Principles and Methods Using
Indices
[11] AACE International, Recommended Practice No. 68R-11, Escalation Estimating Using Indices and Monte Carlo
Simulation

CONTRIBUTORS

Disclaimer: The content provided by the contributors to this recommended practice is their own and does not
necessarily reflect that of their employers, unless otherwise stated.

Colin H. Cropley (Primary Contributor)

James E. Arrow, DRMP
Francisco Cruz Moreno, PE
John K. Hollmann, PE CCP CEP DRMP FAACE Hon. Life
Dr. David T. Hulett, FAACE

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 13 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

APPENDIX

A.1. Assess Net Systemic Cost & Schedule Risk Factors Using Monte Carlo Simulation (MCS)

To illustrate the methodology, an example project using the risk factors method and Safran Risk1 is described. This
is a rail transport project. The parametric model used was developed based on the RAND model developed in the
1980s using a process plant construction projects database. However, as demonstrated in Chapter 15 of Project Risk
Quantification [8, pp.299-300], parametric model inputs describing the characteristics of such a rail transport project
can be applied to produce a reasonable fit using the process model.

A.1.1. Brief Description of the Project

A definition phase rail project involving a 15km extension to a suburban passenger line using an existing freight rail
line alignment, including 2 bridges and 2 stations, has been used for this purpose for which a schedule, estimate and
risk register were available. The base estimate is summarized in Table 2.

Element Totals Labor Plant Materials Subcontractors

Base Estimate $530,528 $158,170 $66,908 $126,424 $177,526
Table 2 – Summary of Rail Project Base Estimate ($000s)

Labor, plant, and subcontractor costs were all treated as time dependent (TD) costs, with materials treated as time
independent (TI) costs. A portion of the summarized base schedule is shown in Figure 4.

Figure 4 – Portion of Summarised Rail Project Schedule

The planned or deterministic duration of the base schedule was from the funding approval milestone of 24May18
to the planned date for practical completion of 17Mar23, a duration of 1,758 calendar days.

The characteristics of the summary rail project schedule used are shown in Figure 5. The purpose of including the
schedule statistics is to provide otherwise unavailable information about the schedule and to demonstrate that the
schedule is fully linked. If the ratio of links to total activities falls significantly below 2, open ends (of activities) are
likely to reduce the usefulness of a CPM-based ICSRA or an SRA.

1
Note: The sample project was created using Safran Risk, however this does not imply an endorsement by AACE International.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 14 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Figure 5 – Characteristics of Rail Project CPM Schedule

A.1.2. Parametric Analysis of Systemic Cost Growth

Systemic risk was evaluated in accordance with RP 42R-08, Risk Analysis and Contingency Determination Using
Parametric Estimating. The parametric model used was one made available to readers of JK Hollmann’s Project Risk
Quantification [8]. Log normal distributions for cost growth and schedule slip were generated around the mean
values produced by the parametric model, using the total base estimate of $530.5m and the base schedule duration
of 1,758 days to frame the distributions.

The systemic cost growth probability distribution around the mean project cost of $605m (114% of the base estimate
of $530.5m) was interpolated from Table 3, which was generated by the parametric model.

Systemic Cost Contingency Probability Distribution

Base Estimate Cost ($m): $ 530.5
Proj Cost (m) P-value
$ 496 10%
$ 526 20%
$ 550 30%
$ 572 40%
$ 594 50%
$ 617 60%
$ 643 70%
$ 675 80%
$ 725 90%
Table 3 – Systemic Cost Growth Probability Distribution

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 15 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

The systemic schedule slip probability distribution around the mean project duration of 2004 days (114% of base
duration of 1,758 days) was interpolated from Table 4, which was generated by the parametric model.

Systemic Sched Contingency Probability Distribution

Base Sched Dur (cal days): 1,758 57.8
Proj Dur (d) P-value (Months)
1,776 10%
1,840 20%
1,890 30%
1,936 40%
1,981 50%
2,031 60%
2,087 70%
2,153 80%
2,248 90%
Table 4 – Systemic Schedule Slip Probability Distribution

A.1.3. CPM-Based ICSRA (IRA) Using Risk Factors Method Analysis of Inherent Risk

Cost hammocks were applied to the schedule shown in Figure 4. To these were applied the time-dependent costs
summarized in Table 2. The time-independent costs summarized in Table 2 were applied via the Costs tab in Safran
Risk (SR). All these costs were made subject to uncertainty through a series of 100% probability risk factors, such as
those shown in Figure 6. Each risk factor has a three-point distribution applied to it for schedule and for cost (as
applicable) that has been assessed by the project team in a risk factors workshop. An example is provided in Figure
7 for RF009 Track construction uncertainty.

Figure 6 –Sample of Risk Factors Applied to Rail Project

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 16 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Figure 7 – Example Schedule and Cost Impact Risk Factor Distributions

Cost and schedule probability distributions based on (inherent) risk factors only are shown in Figure 8 and Figure 9
respectively. As per Figure 2 methodology Steps 5 & 6, these base distributions of inherent cost and schedule risk
are defined and, as per Step 7, are subtracted from the parametric distributions using MCS.

These same base distributions of inherent cost and schedule uncertainty then form the foundations of the P+IRA
model when the net systemic risk factors produced from the MCS subtractions in Step 7 are added in Step 8.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 17 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Figure 8 – Rail Project Base Estimate Inherent Cost Risk Probability Distribution

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 18 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Figure 9 – Rail Project Base Schedule Inherent Duration Risk Probability Distribution

A.1.4. Assessment of Net Systemic Cost & Schedule Risk Factors by MCS Using SR

To produce net systemic risk factors, a simple separate project is set up in SR, as shown in Figure 10.

The project subtracts:

• The base IRA schedule impact duration distribution from the parametric duration impact distribution and
• The base IRA cost impact distribution from the parametric cost impact distribution using MCS in multiple
stages.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 19 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

For convenience, all tasks have the same remaining duration as the planned Rail Extension Project execution
duration of 1,758 days. This duration is necessary for the duration-related tasks, but not for the cost-related tasks,
which are assumed to be 100% non-time-dependent.

Figure 10 is shown during an iteration of a simulation in which the duration of hammock activity 00030 is positive
with a duration of 387d because the systemic risk duration in that iteration is greater than the IRA base plan risk
duration. Hammock activity 00040, with the reverse linkages of hammock 00030 would have a duration of -387d
except that under CPM rules, negative durations are not permitted, so the duration of hammock 00040 shows as 0.

Figure 10 –Rail Project Net Systemic Risk Factors (NSRF) Assessment Schedule in Safran Risk

The purposes of each of the activities in the NSRF schedule are described and defined in Table 5 below.

Outline of Net Systemic RFs Assessment Process

Note that there are several stages to the buildup of the net systemic risk factors:
• 3 MC simulations for the net systemic duration RF.
• 2 MC simulations for the net systemic cost RF.
• The stages are noted under each activity ID.
• The net duration RF process is independent from the net cost RF process; they are included in the same net
systemic risk Factors project schedule for convenience.
• The inputs to and outputs from the duration and cost processes are noted in the Comments column.
• The input RF distributions are loaded through the SR Project Risks tab, as described in section A.1.5.
• The output distributions produced by MCS are produced in the SR Distribution Graph tab, as shown in
Figure 15 and Figure 20. Their output (statistical) Information is used to produce candidates for NSCRF and
NSDRF. Two alternative types, Trigen (10,90) and LogNormal are input to the final cost and final duration
stages. Comparisons are made in each of the cost and duration processes as to which distribution type best
matches the NETCOST and NETSYSDUR output distributions, using Excel (see Table 6 and Table 7).
• Note that the selection between two candidate distribution types is not necessary. Just the Trigen (10,90)
NSCRFT and NSDRFT RFs could have been calculated and used in the Rail Project IRA model (or just the
LogNormal RFs).

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 20 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Activity ID Description Purpose Comment

NETDUR Net duration distributionGroups duration impact tasks Grouping function: distribution
determination together not required
00010 Rail extn execution Input task to which systemic Project input risk factor Risk-001
(1st dur stage) duration (systemic duration impact risk distribution is applied (correlated +80% to
distribution) is applied Risk-002)
00020 Rail extn execution Input task to which IRA inherent Project input risk factor Risk-002
(1st dur Stage) duration (IRA base plan duration impact risk distribution is applied (correlated +80% to
distribution) is applied Risk-001)
00030 Hammock systemic - IRA Collects positive differences from Output distribution from 1st durn
(2nd dur stage) base plan iterations when systemic duration MCS, to be combined with that
> IRA duration (otherwise 0) from task 00040
00040 Hammock IRA base plan Collects positive differences from Output distribution from 1st durn
(2nd dur Stage) – systemic iterations when IRA duration > MCS, to be combined with that
Systemic duration (otherwise 0) from task 00030
NETSYSDUR Net systemic duration Task on which net systemic Key output from 2nd durn MCS, of
(3rd dur stage) distribution (based on duration distribution produced RF Risk-006, produced from
correlation +0.8 between from combining hammocks 00030 combination of hammocks 00030
Risk-001 & Risk-002) & 00040 (through Risk-006). & 00040 built in Excel.
NETSYSDRFT Net systemic duration Task duration distribution derived Net systemic duration RF Risk-007
(4th dur stage) risk factor (Trigen) from distribution parameters of distribution, built from
NETSYSDUR duration distribution. NETSYSDUR output values using
Trigen (10,90).
NETSYSDRFN Net systemic duration Task duration distribution derived Net systemic duration RF Risk-010
(4th dur stage) risk factor (LogNormal) from distribution parameters of distribution built from
NETSYSDUR duration distribution. NETSYSDUR output values using
LogNormal.
NETCOST Net cost distribution Groups cost impact tasks Correlation negative because base
(2nd cost from MCS (based on together and collects cost impact & systemic costs have opposite
stage) correlation difference distribution between signs. Key output net cost risk
-0.8 between Risk-004 & systemic cost risk and base cost distribution generated on the
Risk-005) estimate risk distributions summary task by the MCS
iterations.
00080 Rail extn systemic risk Task to which systemic cost Input RF Risk-004 is applied
(1st cost stage) cost distribution impact risk distribution is applied (correlated -80% to Risk-005)
00090 Rail extn IRA base plan Task to which IRA inherent cost Input RF Risk-005 is applied
(1st cost stage) execution cost distbn impact risk distribution is applied (correlated -80% to Risk-004)
as negative costs.
NETSYSCRFN Net systemic cost risk Cost distribution derived from
Net systemic cost RF Risk-008
(3rd cost stage) factor (LogNormal) distribution parameters of
distribution built from NETCOST
NETCOST cost risk distribution.
output values using LogNormal.
NETSYSCRFT Net systemic cost risk Cost distribution derived from
Net systemic cost RF Risk-009
(3rd cost stage) factor (Trigen) distribution parameters of
distribution built from NETCOST
NETCOST cost risk distribution.
output values using Trigen
(10,90).
Table 5 – Purposes of Duration and Cost Related Net Systemic Risk Factors (NSRF) Activities

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 21 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

In the SR risk factors ICSRA tool, uncertain input task durations and cost line items are driven by risk factors. This
requires that risk factors be defined for each uncertain input duration task and cost, as shown from the Project Risks
tab of SR in Figure 11.

Figure 11 – Risk Factors in NSRF Project

Some RFs are inputs and some outputs, as noted in Table 5. RFs are assigned to tasks in the Risk Mapping tab in SR
shown in Figure 12. Note that checkmarks are assigned to task/RF intersections to create assignments. Black squares
indicate a summary task for which sub-tasks have been checked.

Figure 12 – Mapping of Risk Factors to Activities

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 22 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

A.1.5. Input Distributions

Each of the four distributions to be input to Risk-001 – Risk-005 are imported to SR as histograms: discrete
distributions, made up of discrete values, interval by interval and their probabilities of occurring within each interval.

The discrete systemic duration distribution to be input to Risk-001 can be input to SR from 1% interval values
interpolated from a curve generated from Table 4. The corresponding discrete systemic cost distribution to be input
to Risk-004 can be input to SR from 1% interval values interpolated from a curve generated from Table 3. The discrete
duration distribution to be input to Risk-002 can be input to SR from the IRA project base probability duration
distribution shown in Figure 9 exported from SR to Excel at 1% intervals. The discrete cost distribution to be input to
Risk-005 can be input to SR from the IRA project base probability cost distribution shown in Figure 8 – exported
from SR to Excel at 1% intervals.

Once these four distributions are input to SR, MCS analysis can be initiated, and outputs obtained.

A.1.6. Subtraction of Distributions

The analysis requires the subtraction of the inherent risk base schedule distribution from the systemic risk schedule
distribution. For duration, that involves subtracting the base ICSRA Inherent duration risk distribution (Figure 9)
from the systemic duration risk distribution (interpolated from Table 4) using two hammock activities to measure
the differences between iteration values.

For cost, the base ICSRA inherent cost distribution (Figure 8) is subtracted from the systemic cost risk distribution
(interpolated from Table 3). This is achieved by making the inherent cost distribution negative and adding the
inherent cost distribution to the systemic cost distribution, automatically storing the combined net distribution in
the summary task NETCOST.

A.1.7. Correlation

To ensure that the comparison is like for like, high correlation must be input between the pairs of distributions, so
that if a high or low systemic value is chosen, a similarly high or low Inherent risk value is to be selected for the same
iteration. For Risk-001 to Risk-002, both distributions are positive so the correlation coefficient must also be positive.
The Risk-001: Risk-002 pair was assigned a correlation coefficient of +0.8. This is shown from the SR Correlations
tab in Figure 13.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 23 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Figure 13 – Correlation Pairings for MCS Subtractions

For Risk-005 (IRA base cost distribution), the cost distribution is negative which results in the largest value (at P100)
(value -672,906) being considered mathematically as smaller (more negative) than the P001 value
(-535,154), as shown in Figure 14.

This means that when a large systemic cost value from Risk-004 is selected for an iteration, a large magnitude but
mathematically small value is to be selected from the base IRA cost distribution to produce a credible net cost
distribution. The Risk-004 to Risk-005 distribution pair was thus assigned a correlation of -0.8. Both pairs are shown
in the correlation matrix in Figure 13.

The choice of 0.8 was made as a compromise between choosing a high enough coefficient to produce the desired
effect but not so high (say 1.00) so as to produce exact linear matching between pairs of values. When there are
limited values available, this can produce a “step-like” relationship between the pairs of values. The user is
encouraged to see what effect higher correlation coefficients than 0.8 may produce.

Figure 14 – IRA Base Cost Distribution (Negative Values)

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 24 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

A.1.8. MCS Assessment of Net Systemic Cost Risk Uncertainty Distribution and Risk Factor

With distributions loaded into Risk-004 (systemic cost) and Risk-005 (base IRA cost), an MCS can be run using the
SR Analyze tab. Using 5,000 iterations, the distribution shown in Figure 15 was obtained.

Figure 15 – NETCOST Net Systemic Cost Risk Distribution

The NETCOST cost risk distribution represents the probabilistic difference between systemic cost risk and the base
estimate cost uncertainty. It has a deterministic value of $0 and a probabilistic distribution that can be represented
by a suitable distribution shape and characterizing values.

To produce a useful net systemic cost risk factor that can be applied to project costs, it is necessary to express the
risk factor as percentages of the deterministic cost (base estimate). This is done by adding the base estimate to the
NETCOST absolute values in the distribution and dividing the summed values by the base estimate.

Two alternative distribution shapes have been investigated:

• Trigen (10,90), creating Risk-009 inputs to activity NETSYSCRFT (using NETCOST P10, P50 & P90)
• LogNormal, creating Risk-008 inputs to activity NETSYSCRFN (using NETCOST mean & std. deviation)

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 25 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

The closeness of these NSCRFs to the original NETCOST distribution are compared for the example project in Table
6. The two distributions based on the NETCOST distribution characteristics (mean and standard deviation for Risk-
008 LogNormal; P10, P50 & P90 for Risk-009 Trigen (10,90)) are used to generate distributions in activities
NETSYSCRFN and NETSYSCRFT respectively.

Comparison of Log-normal & Trigen Cost Risk Factor Distributions with NETCOST distribution
Determ Cost = ($000s): 530,500 Risk-008 Inputs, NETSYSCRFN Accuracy Risk-009 Outputs, NETSYSCRFT Accuracy
NETCOST (NETCOST + LogN% /
NETCOST LogNormal LogNormal Trigen (10,90) Trigen (10,90) Trigen% /
Distribution: + DetCost Det Cost)/ NETCST% NETCST%
Parameter ($000s) DetCost (%) NETCOST (% of DetCost) NETSYSCRFT (% of DetCost)
Mean 18,499 548,999 103.5% 548,999 103.5% 553329 104.3%
Std Devn 64,907 64,907 12.2% 64,907 12.2% 66027 12.4%
Check Outputs from Risk-008 Inputs to Risk-009 Trigen profile
P10 -64,187 466,313 87.9% 471,252 88.8% 466,313 87.9%
P50 10,719 541,219 102.0% 546,094 102.9% 541,219 102.0%
P90 115,982 646,482 121.9% 634,360 119.6% 646,482 121.9%
Check Risk-008 Outputs Avg: 100.6% Check Risk-009 Outputs Avg: 101.4%
P20 -41,937 488,563 92.1% 495,053 93.3% 101.3% 493,712 93.1% 101.1%
P30 -23,068 507,432 95.7% 513,481 96.8% 101.2% 515,506 97.2% 101.6%
P40 -6,550 523,950 98.8% 530,384 100.0% 101.2% 533,728 100.6% 101.9%
P60 29,574 560,074 105.6% 563,133 106.2% 100.5% 568,261 107.1% 101.5%
P70 51,287 581,787 109.7% 581,769 109.7% 100.0% 589,554 111.1% 101.3%
P80 77,027 607,527 114.5% 603,286 113.7% 99.3% 613,370 115.6% 101.0%
Table 6 – NETCOST Discrete Distribution + Base Estimate Compared with Lognormal & Trigen (10,90) RFs

This analysis shows that both the LogNormal and Trigen (10,90) distributions can produce acceptable net systemic
cost risk factors to apply to all time-independent costs in the project (relative to the deterministic base estimate), as
follows:
• LogNormal NSCRF: Mean 103.5%; Standard Deviation 12.2%
• Trigen (10,90) NSCRF: Low 88%; Likely: 102%; High: 122%

The systemic risk distributions are based on LogNormal functions and it can be seen from the accuracy columns of
Table 6 that the LogNormal distribution NSCRFN produces results slightly closer to the NETCOST distribution (100.6%
average of P20 to P80 % ratios) than the Trigen (10,90) NSCRT (101.4% average of P20 to P80 % ratios), but both are
acceptable because the differences in accuracy are masked by the fact that SR will only accept whole number
percentages. So, when the LogNormal RF has to be defined as Mean: 104%, Standard Deviation 12%, it produces
results less accurate than the Trigen (10,90) RF defined by Low 88%, Likely 102%, High 122%.

Given the uncertainties involved in the forecasting of project cost and schedule outcomes, the accuracies of these
net systemic cost RFs to be applied across all fixed costs are within acceptable limits.

The time dependent costs in the project are subject to the net systemic duration risk factor which is next to be
evaluated.

A.1.9. MCS Assessment of Net Systemic Duration Risk Hammocks

The assessment of the net systemic duration risk factor is more complicated than the cost risk factor because
negative durations are not permissible in CPM schedules. It is necessary to use two hammock activities to track the

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 26 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

difference in durations between the systemic risk schedule slip activity 00010 and the base schedule inherent risk
activity 00020, as driven by Risk-001 and Risk-002 respectively. The two hammocks produce two parts of a single
distribution defining the net systemic duration uncertainty distribution, but the two parts have to be combined in
Excel to deal with the respective negative portions, expressed in the hammock duration distributions as 0 duration.

To reduce the data handling load in Excel, a 1,000 iteration MCS of the rail project NSRF schedule model shown in
Figure 10 is performed.

The two duration distributions for Hammock tasks 00030 and 00040 are shown in Figure 16.

Figure 16: Duration Distributions for Hammock activities 00030 & 00040 (1,000 iterations)

A.1.9.1. Export Percentiles of Both Histograms

SR enables the duration percentiles from 0% to 100% at 1% intervals to be exported to Excel from each of the two
hammock activity distributions. Upon examination in Excel, it becomes apparent that the large number of zero
duration “hits” in each distribution almost entirely represent the negative values which are displayed as positive in
the other hammock distribution.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 27 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

A.1.9.2. Combine Histogram Percentiles

It is necessary to manipulate the non-zero data from Hammock 00040 to express it as negative and reverse the order
of the numbers of that they can be combined with the Hammock 00030 data. This process is shown first in the export
file from Hammock 00040 (Figure 17), from where the data highlighted in yellow is combined in the export file from
Hammock 00030 (Figure 18).

Figure 17 – Positive Tail of Hammock 00040 is Converted to Negative and the Order Reversed

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 28 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Figure 18 – Combining Negative Tail from Hammock 00040 with Main Positive Section of Distribution in
Hammock 00030

A.1.9.3. Upload Combined Percentiles to SR Risk-006

The yellow highlighted columns in Figure 18 (1% probability values adding up to 100%) are pasted into SR (partially
shown in Figure 19), as has been done with costs.

Figure 19 – Combined histogram percentiles uploaded to Risk-006

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 29 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Note that the percentile values are uploaded as numbers, but SR adds “d” suffixes.

This creates the net systemic risk distribution used to generate the Risk-006 distribution around the deterministic
duration of the rail project of 1758 days.

A.1.9.4. Perform MCS to generate inputs for Net Systemic Duration Risk Factors

Another MCS is run on the rail extension net systemic risk project to produce an output distribution in the task
NETSYSDUR from the Risk-006 input distribution, as shown in Figure 20.

A.1.9.5. Production of Net Systemic Duration Risk Factors (NSDRFs)

The NETSYSDUR distribution provides the input metrics to produce net systemic duration risk factors, LogNormal or
Trigen, as was done for cost RFs. However, whereas the net systemic cost distribution is around a deterministic
value of $0, the net systemic duration distribution is around the non-zero deterministic duration of 1,758 days.

SR adds absolute distributions to the deterministic value but multiplies relative distributions by the deterministic
value. As relative distributions in SR are rounded to the nearest whole percent values, the precision of the relative
distribution is reduced compared with the absolute distribution.

The metrics of the two duration RFs are compared with the parent NETSYSDUR distribution in Table 7.

Comparison of Log-normal & Trigen Duration Risk Factor Distributions with NETSYSDUR distribution
Determ Durn (DD, days): 1,758 Risk-010 Inputs, NETSYSDRFN Accuracy Check NETSYSDRFT outputs Accuracy
LogN% / % Trigen% /%
NETSYSDUR NETSYSDUR LogNormal LogNormal Trigen (10,90) Trigen (10,90)
Distribution: NETSYSDUR NETSYSDUR
Parameter (days) / DD (%) As analysed (% of DD) As analysed (% of DD)
Mean 1,876 106.7% 1,878 106.8% 1,889 107.5%
Std Devn 124 7.1% 124 7.1% 127 7.2%
Check NETSYSDRFN Outputs Outputs from NETSYSDRFT
P10 1,728 98.3% 1,720 97.8% 1,725 98.1%
P50 1,857 105.6% 1,873 106.5% 1,880 106.9%
P90 2,064 117.4% 2,040 116.0% 2,066 117.5%
Avg: 100.5% Check NETSYSDRFT outputs Avg: 101.0%
P20 1,762 100.2% 1,773 100.9% 100.6% 1,774 100.9% 100.7%
P30 1,794 102.0% 1,811 103.0% 100.9% 1,815 103.2% 101.2%
P40 1,824 103.8% 1,844 104.9% 101.1% 1,849 105.2% 101.4%
P60 1,892 107.6% 1,908 108.5% 100.8% 1,914 108.9% 101.2%
P70 1,940 110.4% 1,939 110.3% 99.9% 1,955 111.2% 100.8%
P80 1,992 113.3% 1,979 112.6% 99.3% 2,006 114.1% 100.7%
Table 7 – NETSYSDUR Discrete Distribution Around Base Duration Compared with LogNormal & Trigen RFs

As was the case for modelling of the NETCOST distribution in Table 6, the modelling of the NETSYSDUR distribution
by the LogNormal distribution, NETSYSDRFN, based on a mean of 106.8% and a standard deviation of 7.1% gave
slightly better accuracy (100.5% of P20 to P80 % ratios) than the Trigen (10,90) distribution NETSYSDRFT (101.0% of
P20 to P80 % ratios), based on low, medium & high percentages of 98.1%, 106.9% & 117.5% respectively.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 30 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

However, the constraint in SR of only allowing whole percentage numbers to be input results in the LogNormal RF
being defined as mean 107%, standard deviation 7%, and the Trigen (10,90) RF being defined by low 98%, likely 106%
& high 117%. With these inputs, the Trigen RF is slightly more accurate than the LogNormal RF.

Figure 20 – Net Systemic Duration Risk Distribution Produced by MCS of Risk-006 Acting on Task NETSYSDUR

As discussed above, the net systemic duration risk factors analysis in Table 7 – when constrained by SR to whole
percentage inputs results in the Trigen (10,90) distribution giving a slightly closer NSDRF representation of the
NETSYSDUR distribution than the LogNormal distribution, but that both are acceptable because of the overall
uncertainties in the inputs to the methodology. The two NSDRFs are as follows:
• Lognormal NSDRF: Mean 106.8%; Standard Deviation 7.1%, rounded by SR to 107% / 7%.
• Trigen (10,90) NSDRF: Low 98.3%; Likely 105.6%; High 117.4%., rounded by SR to 98%, 106%, 117%.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 31 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

A.2. Addition of NSRFs to CPM-Based ICSRA Model

Referencing Figure 2, with the definition of the two NSRFs, the process arrives at Step 8.The two NSRFs firstly are
added as project risks to the rail extension CPM-based ICSRA model, their impacts are defined then they are mapped
to the applicable activities and costs, as described below.

A.2.1. Add Net Systemic Risk Factors to CPM-based ICSRA Model

Two new project risks are added to the Project Risks tab:
• NSDRF – Net systemic duration risk factor. From above, using Trigen (10;90;98%;106%;117%)
• NSCRF - Net systemic cost risk factor. From above, also using Trigen (10;90;88%;102%;122%)

These are shown below in Figure 21 – after definition in the Risks tab of SR.

Figure 21 – Net Systemic Risk Factors After Addition to Project Risks tab in SR

Their impact distributions are made visible when they are highlighted (as above) in the Project Risks tab in SR:
The NSCRF impact distribution is shown below in Figure 22.

Figure 22 – NSCRF Relative Cost Impact Distribution Defined in SR Project Risks Tab

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 32 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

A.2.2. Map Net Systemic Risk Factors to Tasks and Costs

The chosen NSDRF risk factor is applied to the durations of all normal activities (non-zero duration tasks taking part
in critical path calculations) in the Rail Extension project schedule. This will drive project duration uncertainty and
time-dependent cost uncertainty.

The NSDRF risk factor is mapped into the ICSRA model using the SR Risk Mapping tab. It only requires one click to
map the risk factor to all the activities in the schedule, as shown in Figure 23, due to the availability of an hierarchical
breakdown structure of the activities.

The chosen NSCRF risk factor is applied to all the time-independent costs in the CPM-based ICSRA model. With the
release of SR Version 21, a hierarchical cost breakdown structure is definable in the Costs tab of SR (as shown in
Figure 24) and can be used for allocating risks, as explained after Figure 23.

Figure 23 – Mapping of NSDRF to All Activities in Schedule (Clicking on Top Project Button)

The project costs are divided into time dependent (ID = Labor) and time independent (ID = Materials) costs in the
Cost tab of the project. Columns are displayed using the Field Chooser icon to the left of the Description Title field.
The time dependent costs are linked to the activities to which they apply by clicking the Schedule Connection
checkbox. This shows the activities to which the time-dependent (variable) costs have been allocated and the
percentage of the cost assigned to each task (100% unless multiple activities have been allocated).
Time independent costs must not be linked to activities, otherwise they are treated as variable costs.

In SR version 21 onwards, risk factors can be added at the highest summary level to which they are to apply (as is
required in this case) or manually to each material cost line-item ID in the Risks field, by clicking on the rectangle
with green information triangle (on right of Risks column in blue highlighted row), shown in Figure 24.

When the green triangle is clicked, a selection window appears “Pick the risks that affect the cost elements under
the summary”. Click on the ID header to put the RFs into alphabetical order and ensue that the NSCRF net systemic
risk factor appears at the top. Click on the Selected check box as shown in Figure 25 and OK. The NSCRF RF will then
be applied to every time independent cost line item under the Materials ID.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.
117R-21: Integrated Cost and Schedule Risk Analysis and Contingency Determination Using 33 of 33
Combined Parametric Modelling and Monte Carlo Simulation of a CPM Model

August 15, 2022

Figure 24 – Selection of Time-Independent Costs (Id Materials) for Assignment of NSCRF Cost Risk Factor

Figure 15: Selection drop-down for picking the risks to apply to the cost elements under the summary

After the NSDRF and NSCRF risk factors have been quantified and mapped into the project schedule and cost items,
the P+IRA model can be completed as shown in Step 8 of the flowchart in Figure 2. Please return to the main body
of the recommended practice, section 2.2.1 Map Critical Project-Specific Risk Events to Appropriate Schedule and
Cost Elements of the RP.

Copyright © AACE® International AACE® International Recommended Practices

Single user license only. Copying and networking prohibited.