Chemical Engineering December 2014 - 39

Cover Story
Quoted ranges in literature, such
as ±10% at sanction, typically reflect
an estimation of process quality
uncertainty, not project risks.
These tight ranges reflect the estimator's
and team's confidence in
how well they quantified and priced
the given scope. They assume nothing
will change, no risk events will
occur and that project control will
be excellent. Of course, these assumptions
are rarely realistic for
owners, although they may be somewhat
more realistic for contractors,
for whom contingency must
only cover uncertainty within the
contract scope.
Figure 1 is not the only story;
there also exists a stark dichotomy
in industry estimation practices
between small and large projects.
Figure 2 depicts an example distribution
of accuracy outcomes for
small projects in a U.S. plant-based
project system, where the project is
managed by an operating facility
rather than a major projects organization.
At the plant level, engineers
may also act as estimator and
project manager. For small projects,
the range is still wide; small projects
do not benefit from the balancing
of underruns and overruns
among a multitude of cost items.
However, they are biased much
further to the low side. The resulting
midpoint is a slight underrun.
This usually reflects conservative
base estimates with a generous use
of " allowances. "
In lean plant-based organizations,
where each project leader has many
projects of short duration, the emphasis
is on getting operations up
and running quickly and safely. One
cannot spend weeks or months arguing
for money to cover an overrun,
so projects are frequently overestimated,
often subconsciously. If excess
money is returned to the business
in a disciplined way, like in the
example shown in Figure 2, this bias
need not be wasteful, although it
does lock up capital for some time.
Those who understand this behavior
will not be surprised that most
predictable owner capital-project
systems tend to be those that are
most accurate. These tight-range
systems tend to have an overfunding
bias, and they spend the excess,
which, on average can be 5-20%
more than disciplined, but less
predictable, systems.
To summarize the above discussion,
there are no standard ranges.
Large projects managed by major
owner-project groups typically experience
greater risk than expected
and face greater cost scrutiny, so
base estimates are tight and contingency
is underestimated. Small
projects managed by plants, while
having their share of risk, are typiActual
CPI estimate accuracy versus reality for typical large projects
cally overestimated. So, when it
comes to improving accuracy and
contingency estimation, one must
pay careful attention to the level of
scope definition, specific risks and
biases. These are discussed in the
following sections.
Know your scope
Table 2 illustrates how estimate accuracy
improves with the level of
scope definition. This is no longer
an arguable topic. Since the 1990s,
almost every major CPI owner company
has implemented a phasegate
project system, and the phases
usually line up with the AACE
Estimate Classes in Table 2. The
topic of scope definition, in terms
of knowing which deliverables are
important, and how to rate their
definition, is well covered in literature.
Example rating schemes for
the CPI include AACE's Classes
(RP 18R-97), the Construction Industry
Institute's (CII; Austin, Tex.;
www.construction-institute.org) Industrial
Project Definition Rating
Index (PDRI) and IPA, Inc.'s (Ashburn,
Va.; www.ipaglobal.com) FrontEnd
Loading Index (FEL). The level of
scope definition is the greatest driver
of cost uncertainty; understanding
that fact is the starting point for
analyzing risks.
Other key risk drivers within the
As estimated
Reality
project scope include the introduction
of new technology in the process
and the level of complexity in
the physical system, as well as the
execution strategy itself. Decades of
empirical industry research on the
impact of scope definition, technology
and complexity on cost accuracy
have proven these points (see sidebar
on RAND and Hackney, p. 42).
AACE refers to these dominant risk
drivers as " systemic " risks because
they are intrinsic to one's project
and physical systems.
Percentage cost over/under estimate including contingency
FIGURE 1. In reality, large CPI projects often miss the mark with contingency estimation
methods, and large discrepancies between actual and estimated values occur
38 ChemiCal engineering www.Chemengonline.Com DeCemBer 2014
Know your base estimate
Per Table 1 definitions, the " base estimate "
excludes risk - it's just the
facts. Base estimating practices are
well covered in literature [3]. Base
estimating is essentially the same,
whether done by an owner or contractor.
Allowances for uncertainty
Relative frequency of occurrence
-20
-10
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
http://www.construction-institute.org http://www.ipaglobal.com http://www.Chemengonline.Com
Chemical Engineering December 2014

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