Chemical Engineering December 2014 - 43

Cover Story
risk ratings (OR). A, B and C are
the respective coefficients, which
along with the regression constant,
are developed from regression analysis
of historical data or published
models in literature.
All the team has to do is rate
their project's systemic risks and
bias, and insert the ratings into
the equation. The challenging part
is developing a risk-rating scheme
and an equation. Most companies
do not have historical cost and
risk data with which to develop an
equation using regression analysis.
Fortunately, rating schemes and
reliable equations are available,
based on 50 years of CPI research.
Two time-honored rating schemes
and equations are built into working
Microsoft Excel tools, which are
publicly available on AACE's website,
and documented in RP 43R-08
(see sidebar on RAND and Hackney).
For those looking for a quicker
answer, Table 5 provides systemic
contingency values similar to the
RAND model, but instead applies a
model derived by the author.
The values in Table 5 are average
allowances (about p55). They
assume a large gas or liquids CPI
project with 20% equipment cost
and typical project-management
system maturity and biases. For
Class 3 scope definition, it assumes
that 5% of the total cost has already
been spent on engineering, and that
the class is over-rated. Few projects
achieve ideal Class 3; most are
the RaND aND hackNey paRametRic
cOst Risk mODels
O
n its website, AACE International provides two working Microsoft Excel models for
quantifying systemic risks (RP 43R-08). One model is by the late John W. Hackney,
a founder of AACE, as first published in his 1965 landmark text " Control and
Management of Capital Projects, " the 2nd edition of which is now published by AACE
International. Hackney studied actual projects from his experience, and developed an
elaborate contingency model based on the rating of many systemic risk factors. In 1981,
the RAND Corp. (Santa Monica; Calif.; www.rand.org) published a groundbreaking
report entitled " Understanding Cost Growth and Performance Shortfalls in Pioneer Process
Plants " (Rand R-2569-DOE). That report includes an empirically based parametric
model of cost growth, including statistical parameters. The study's lead author, Edward
Merrow, went on to found IPA, Inc., and Hackney was a consultant to the 1981 study.
These two publications should be on every CPI cost estimator's bookshelf.
As estimated
Reality
❑
Case study of actual CPI estimate accuracy for small projects
funded closer to Class 4. As seen in
Table 5, a wide range of complexity
is taken into account, with low-complexity
projects having less than
three block-flow process steps with
a simple execution strategy, and
high-complexity projects having
more than six continuously linked
process steps or a complicated execution
strategy. The technology
basis varies from less than 10% of
capital expenditure (capex) for a
process step with commercially unproven
technology to greater than
50% of capex for new, R&D or pilot
scaleup steps. Engineers can make
adjustments to the indicated table
value as shown. The fixed-price adjustment
allows for less uncertainty
for that element, but there is no
such thing as fixed price for owners;
change and risk events will inevitably
happen. Per the precedents in
Table 5, an example of best/worst
case (p10/p90) application is as
If management chooses a 50% confidence level,
then contigency is $100,000 ($1.1-1.0 million)
Frequency
of
occurence
c
o
n
t
i
n
g
e
n
c
y
Base estimate
$1.0 million
Value at p50
$1.1 million
FIGURE 4. A probabilistic cost distribution can be used to determine contingency
42 ChemiCal engineering www.Chemengonline.Com DeCemBer 2014
Consider that a workshop is held
in which key team members rate
the systemic risks for application of
Table 5 and then discuss the critical
risks and quantify their probability
and most likely impacts if they occur.
For example, assume the team estimates
there is a 20% chance that
The goal is to come up
with a distribution of
possible outcomes
considering all
identified risks
follows: if systemic contingency is
+10% (Class 4, low complexity, no
new technology), the corresponding
best and worst cases would be -5%
(-0.5 × 10) and +30% (3 × 10), respectively.
Note that the best case
under-runs the base estimate in
this scenario.
After quantifying the systemic
risks, the Expected Value method
is used for each critical project-specific
(PS) risk to determine the average
contingency cost to add to the
base estimate. The exception is for
Class 5 estimates, for which specifics
are unknown and systemic contingency
can be used alone. If risks
have been largely mitigated during
early design, and risk screening is
done correctly, there will typically
be less than 10 critical project-specific
risks. Never quantify all of the
risks in a register, of which there
may be hundreds - the systemic
allowance covers typical risk " noise "
for non-critical risks. The simple
Expected Value expression for determining
the mean PS contingency
value (MPSC) percentage is shown
in Equation (2), as the summation
of the products of probability of risk
occurring (Prisk) and the cost impact
(in $) if the risk occurs (CI).
MPSC = ∑(Prisk × CI)
(2)
Relative frequency of occurrence
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
10
20
30
40
50
http://www.rand.org http://www.Chemengonline.Com
Chemical Engineering December 2014

Table of Contents for the Digital Edition of Chemical Engineering December 2014
