Process Development

Developing
Screening Cost Estimates
Richard P. Palluzi, P.E. Making decisions based on estimates
ExxonMobil Research and Engineering
derived from weak data is counterproductive.
Use this established methodology to
prepare scoping and screening cost estimates
for pilot plants quickly and accurately.

N
ew pilot plants often require scoping or screening and poor operation with excessive downtime and mainte-
estimates for budget development, program resourc- nance. Units are often overestimated because they were
ing, contingency planning, or others reasons. Devel- subjected to a less-rigorous cost-benefit analysis or scrutiny
oping these estimates, however, can sometimes be difficult. for cost-effectiveness.
The most basic challenges include: lack of adequate Scoping estimates receive little or no validation after they
unit definition, which makes it very difficult (or impos- are made — each estimate stands on its own and is as uncer-
sible) to accurately identify cost elements in any detail; tain as the last. At best, a cursory analysis of past estimating
lack of adequate time to develop the design, in which case performance (i.e., how accurate previous scoping estimates
a guess for the cost is made based on similar units; and were relative to the final cost of the pilot plant) may be
lack of an established methodology to deal with this type conducted, but this often only shows the range of uncertainty.
of cost-estimating need. Although this information may be statistically correct, it is
There are also several more subtle, and potentially usually of little or no use in the decision-making process.
somewhat more adverse, challenges. Most pilot plants are Major decisions often must be made based on weak or
so novel that trying to estimate costs based on similarity is bad data. The basic conundrum is that trying to develop an
rarely effective, and often this approach leads to gross errors. estimate too quickly often produces gross errors, but waiting
Differences in construction time and local labor rates can for the necessary information and expending the significant
easily be identified and adjusted; differences in materials, time and effort needed to produce a realistic estimate is
instrumentation, and control levels can often be addressed, often too time-consuming or unnecessary early in a project.
albeit with more difficulty and uncertainty. However, differ- The results of an accurate scoping estimate can lead to the
ences in productivity, design philosophies, local site require- decision to abandon some projects, and although expending
ments and preferences, safety rules, equipment robustness, significant time and effort for a negative result is valuable, it
wiring practices, and many other issues are generally almost is rather counterproductive to overall effectiveness.
impossible to correct for with any degree of accuracy. The
scoping estimate is often poor, with wide limits of error. This Assess the proposed unit
leads to insufficient confidence in the result and complicates An established methodology can produce accurate
the decision-making process. results with a reasonably high confidence level. It should
There is also a tendency for the result to match the per- not require a large amount of detail (which is unlikely to be
ception, resulting in a badly underestimated unit that must be available) and should be relatively fast and easy to use.
completed for an often unrealistically low cost. This can lead The procedure outlined here is to:
to poor design and construction decisions, extended startups, 1. Define the unit to the extent possible at the stage that
This article is based on a paper presented at the AIChE Annual Meeting, the estimate is needed.
Salt Lake City, UT, Nov. 2010. 2. Assess the new unit in terms of three basic criteria:

Copyright © 2011 American Institute of Chemical Engineers (AIChE) CEP  July 2011  www.aiche.org/cep  39
Process Development
Table 1. Consider these process features when assessing whether
the proposed pilot plant is of low, average, or high complexity.

size, complexity, and constructability. Process

Feature Low Average High
3. Use this assessment and a data-
base of previous unit costs to develop a Instrumentation Less than normal About normal (6–8 More than normal
transducers, 10–12
preliminary scoping estimate. Standard Nonstandard equipment
equipment thermocouples)
4. Add appropriate specialty costs for General-purpose equip-
Standard equipment ment in a classified area
features unique to the proposed new unit.
5. Add an appropriate contingency. Control System Stand-alone Standard computer Larger system
Size is the easiest criterion to under- controllers Nonstandard system
stand. A larger unit will, in general, cost Standard small
more than a smaller unit. To account for computer
size, determine how many standard-size Control Loops Fewer than normal About normal (8–10 More than normal
modules or sections the proposed new Standard loops loops) Nonstandard loops
unit will require. These modules can be Standard loops Advanced controls
somewhat flexible. For example, skids Area Electrical General purpose Class I Division 2 Class I Division 1
or pallets 6 ft by 3 ft by 6 ft high are Classification Some Class I Class II or Class III
common. The modules selected can be Division 2 Zone 0 or Zone 1
smaller or larger depending on what the
Reactors Single Single Multiple
organization typically constructs. Larger
Simple More complexity High complexity
units can be estimated as multiples of a
number of smaller modules. Packaged auto- Larger autoclave Adiabatic
clave or equivalent
Complexity compares the envisioned Special or novel
unit to a series of standards, based on Reactor Standard electric Multiple standard Nonstandard heater
such factors as shown in Table 1 that Temperature heater electric heaters High complexity
Control
account for the effort required in the Fluidized-bed heater Adiabatic
design and construction. A more-complex Hot or cold oil Special or novel
unit will clearly require more effort. system
Constructability refers to how hard Process Vessels Less than normal Normal (4–6) More than normal
or easy the envisioned unit will be Specialty design
to build relative to the factors shown Novel design
in Table 2. This includes important, Higher complexity
but often overlooked, factors, such as Rotating Less than normal Normal (1–2) More than normal
construction density (space available Equipment Standard Standard equipment Nonstandard equipment
vs. space required), construction height equipment Solids feeding
(ladders or scaffolds required vs. at grade
level), construction methods, and so on. Heat Tracing None to very few Limited (1–4) Many (>4)
(or Cooling)
In assessing both complexity and
constructability, one could, in theory, Specialty None Limited (1–2) Many
Designs
assign as many levels as desired.
However, the methodology cannot Online Analytical None or at most a Normal (1–2) More than normal
Equipment single standard
realistically accommodate many levels, Standard equipment Nonstandard equipment
because the granularity achievable is Gas Novel equipment
chromatograph
simply not high enough. Hence, it is General-purpose equip-
most effective to have a maximum of ment in a classified area
three levels — the average level, which Safety Interlock Fewer than normal Normal (10–20) More than normal
implies that the proposed module will be Systems Limited sequenced Significant sequenced
fairly similar to most other units of this actions actions
type of construction, plus a lower level High-hazard shutdowns
and a higher level. Pressure* <500 psig Up to 3,000 psig >3,000 psig
Using Tables 1 and 2 is very straight- Temperature† <250°F <450°F >450°F
forward: Assess the proposed pilot plant
* May be organization-specific, but there is a clear break point at 3,000 psig.
one module at a time and determine
† May be organization-specific, but there is a clear break point at 450°F.
which of the three categories most likely

40  www.aiche.org/cep  July 2011  CEP Copyright © 2011 American Institute of Chemical Engineers (AIChE)
 Table 2. Consider these aspects of the process when assessing the pilot plant’s constructability.
Consideration Low Average High
Piping Size <0.25 in. 0.25 in. to 1 in. >1 in.
Piping Type Tubing Tubing Pipe
Some pipe
Fittings Compression Compression Welded
Some threaded All threaded
Cone and collar
Specialty fitting
Construction Access* Access from all sides; or Access from 3–4 sides; or Access from only one side; and
Able to be modularized Able to be modularized Not able to be modularized
Space Available More than required Satisfactory Less than required
Construction Density† Less than normal Normal Higher than normal
Significant open space Only moderate open space Little to no open space
Most components within Majority of components within reach, but Significant components require
reach may require ladders or stools means to access
No significant in-depth Moderate in-depth placement Significant in-depth placement
placement Some potential problems
Construction Height‡ 6–8 ft 8–20 ft >20 ft
Utility Feeds Close Within 25 ft >25 ft
Access requires only ladders Access requires only ladders Requires scaffolding or lift to ac-
Some existing supports Some existing supports cess
No existing supports
Power Available nearby Available locally Available within building
Mostly 120 VAC, 1 phase Mostly 120 and 240 VAC, 1 or 3 phase > 480 VAC
240 VAC or less 480 VAC or less Requires multiple transformers or
No transformer, panel feed Requires single transformer 45 kVA or less transformer > 45 kVA

* Modularized means able to be built in a shop and then moved into place and connected.
† In-depth placement means the equipment is not accessible without reaching through piping and/or components.
‡ Height above grade or existing platform.

describes the module. Note that this requires an assessment The cost database
(or guess) of what the final design will involve. A database can be developed from actual costs of
For example, assume there will be a fixed-bed reactor previously built units by dividing the existing units into
made from tubing with a recycle compressor, flow and pres- standard-size modules, assessing the complexity and con-
sure control loops, and several sample points as one module. structability of each module, and developing the average
This might be assessed as shown by the highlighting in cost for each category.
Tables 3 and 4 and the results summarized as in Table 5. This approach, while very useful, raises several con-
This suggests that the envisioned module is of about cerns. It requires a large amount of accurate data, because
average complexity and lower-than-average constructabil- the database has to cover the entire range of classes, with
ity. However, the assessor should use his or her judgment enough examples of each category, to give adequate con-
and not rely solely on a simple mathematical summary. For fidence in the accuracy. This requires obtaining multiple
instance, if the control system for the recycle loop and the costs for each class. The database also requires an accurate
temperatures involved will be significantly more complex assessment of each module against some predefined criteria
than average, then assessing the module as above average is to ensure adequate uniformity in the results. In many cases,
warranted. This demonstrates the need for an experienced cost data — or enough cost data — are not readily available.
designer/estimator to make the assessment. In either case, it may take some time to gather enough data.
It is often useful to document and retain this information so The existing cost data will almost certainly require some
that as the actual design progresses, the screening estimate can analysis to make them useful. While this is particularly
be updated if the assumed basis starts to deviate significantly common for older data gathered before the decision to create
from the evolving design. This helps to prevent surprises later. the database was made, even newly gathered data may have

Copyright © 2011 American Institute of Chemical Engineers (AIChE) CEP  July 2011  www.aiche.org/cep  41
Process Development
Table 3. The fixed-bed reactor module is of about average complexity.
Process
apparent problems. Cost data Feature Low Average High
often appear inconsistent or Instrumentation Less than normal About normal (6–8 More than normal
suspect, and require work to Standard equipment transducers, 10–12 Nonstandard equipment
thermocouples)
resolve the apparent concerns. General-purpose equipment in
In certain cases, some data Standard equipment a classified area
may prove useless because Control System Stand-alone controllers Standard computer Larger system
the problems cannot be fully Standard small Nonstandard system
explained; in these instances, it computer
is usually better to accept that Control Loops Fewer than normal About normal (8–10 More than normal (Recycle
these data points are outliers Standard loops loops) compressor controls)
and not include them in the Standard loops Nonstandard loops
database. Advanced controls
Expect to spend significant Area Electrical General purpose Class I Division 2 Class I Division 1
effort resolving/normalizing Classification Some Class I Division 2 Class II or Class III
the cost data. Available cost
Zone 0 or Zone 1
data are not very specific,
particularly older information; Reactors Single Single Multiple
perhaps only total costs are Simple More complexity High complexity
available, and they have little Packaged autoclave or Larger autoclave Adiabatic
detail. It may be necessary to equivalent Special or novel
apply normalizing factors or Reactor Standard electric Multiple standard Nonstandard heater
guidelines to make the data Temperature heater electric heaters High complexity
useable (at least in the begin- Control Fluidized-bed heater Adiabatic
ning). As data continue to be Hot or cold oil Special or novel
added based on knowledge of system
what is needed, the database Process Vessels Less than normal Normal (4–6) More than normal
will become more accurate and Specialty design
the effort involved will be less. Novel design
The assessment also Higher complexity
requires a high degree of
Rotating Less than normal Normal (1–2) More than normal
uniformity. During initial Equipment Standard Standard equipment Nonstandard equipment
development, it is more effec-
equipment Solids feeding
tive for one person or group to
assess all the units in the data- Heat Tracing None to very few Limited (1–4) Many (>4)
base. If this is not possible, use (or Cooling)
predefined criteria and actual Specialty None Limited (1–2) Many
examples. Recognize that Designs
some people are better at this Online Analytical None or at most a Normal (1–2) More than normal
type of assessment than others. Equipment single standard Standard equipment Nonstandard equipment
Training and experience helps, Gas chromatograph Novel equipment
but does not always close this General-purpose equipment in
innate gap. a classified area
Costs for items such as spe- Safety Interlock Fewer than normal Normal (10–20) More than normal
cialty analytical instrumenta- Systems Limited sequenced (Compressor recycle system
tion, packaged systems, special actions and interlocks)
structures, facility modifica- Significant sequenced actions
tions, etc., need to be added to High-hazard shutdowns
the basic module costs. A cost Pressure* <500 psig Up to 3,000 psig >3,000 psig
database for common specialty Temperature† <250°F <450°F >450°F
items can be maintained, using
* May be organization-specific, but there is a clear break point at 3,000 psig.
either budget quotes from
† May be organization-specific, but there is a clear break point at 450°F.
contractors or vendors or costs

42  www.aiche.org/cep  July 2011  CEP Copyright © 2011 American Institute of Chemical Engineers (AIChE)
 Table 4. The fixed-bed reactor module is expected to be easier to construct than an average module.
Consideration Low Average High
Piping Size <0.25 in. 0.25 in. to 1 in. >1 in.
Piping Type Tubing Tubing Pipe
Some pipe
Fittings Compression Compression Welded
Some threaded All threaded
Cone and collar
Specialty fitting
Construction Access* Access from all sides; or Access from 3–4 sides; or Access from only one side; and
Able to be modularized Able to be modularized Not able to be modularized
Space available More than required Satisfactory Less than required
Construction Density† Less than normal Normal Higher than normal
Significant open space Only moderate open space Little to no open space
Most components within Majority of components within reach, but Significant components require
reach may require ladders or stools means to access
No significant in-depth Moderate in-depth placement Significant in-depth placement
placement Some potential problems
Construction Height‡ 6–8 ft 8–20 ft >20 ft
Utility Feeds Close Within 25 ft >25 ft
Access requires only ladders Access requires only ladders Requires scaffolding or
Some existing supports Some existing supports lift to access
No existing supports
Power Available nearby Available locally Available within building
Mostly 120VAC, 1 phase Mostly 120 and 240 VAC, 1 or 3 phase > 480 VAC
240 VAC or less 480 VAC or less Requires multiple transformers or
No transformer, panel feed Requires single transformer 45 kVA or less transformer > 45 kVA

* Modularized means able to be built in a shop and then moved into place and connected.
† In-depth placement means the equipment is not accessible without reaching through piping and/or components.
‡ Height above grade or existing platform.

developed from national estimating services such as Table 5. Summary of the complexity and constructability
RS Means. In general, if the special need can be defined, analyses for the fixed-bed reactor module.
an estimate can be made; at worst, recognize that it Assessed as Assessed as
remains an area of very soft cost accuracy. Lower than Assessed Higher than
Contingency is an allowance for historically predict- Criterion Average as Average Average
able, but currently undefined and unknown, extra costs. Complexity (12 areas) 2.5 5.5 4.0
It is a measure of the uncertainty in an estimate and, Constructability (8 areas) 5.1 1.9 1.0
to some degree, the entropy of the universe. Typical
contingencies for scoping estimates range from 20% (materials of construction, flowrates, pressures, tempera-
to 100%. Contingency needs to cover the uncertainty — it tures, etc.) available at the scoping stage. The experience
is not an allowance for poor design, scope or basis changes, of the estimator, in both pilot plant design and construction
program additions, desirable features, schedule changes, or and in cost estimating, is also a critical and often difficult-
similar changes to the basis for the estimate. to-determine factor. Identifying areas of major overrun or
To properly estimate contingency, start by analyzing the underrun and correcting the next estimate always produces
historical performance of previous estimates, and identify better results than trying to guess the correct contingency.
the major variables that affected the uncertainty in the (Uncertainties associated with estimates must also be clearly
estimate. Common variables include novelty (in the process, understood and communicated in order to be helpful when
equipment, chemistry, science, etc.), familiarity (or more major decisions are made. This was discussed in “Communi-
importantly, lack of familiarity) with the process, operation, cating the Cost of Product and Process Development,” CEP,
equipment, location, etc., and the degree of unit definition Feb. 2010, pp. 46–51.)
Article continues on next page

Copyright © 2011 American Institute of Chemical Engineers (AIChE) CEP  July 2011  www.aiche.org/cep  43
Process Development

Table 6. The example pilot plant consists of Table 7. Combine the assessments of the modules’ complexity
35 modules that were assessed as shown here. and constructability (top) with information from the cost
database (middle) to obtain the scoping estimate (bottom).
Process Complexity Constructability
Step Modules Factor Factor Complexity
1 2 Average Average Low Average High
2 2 Average Average Low 1
3 2 Average Average Constructability Average 4 10 4
4 2 High Average High 2 14
5 2 High High
Low $35,000 $65,000 $100,000
6 2 Average High
Constructability Average $100,000 $155,000 $225,000
7 4 Average Average
High $130,000 $250,000 $300,000
8 10 High 8 High
2 Average Low $35,000
9 4 High High Constructability Average $400,000 $1,550,000 $900,000
10 2 Low Average High $500,000 $4,200,000
11 2 Low Average Subtotal $8,000,000
12 1 Low Low Contingency 30% = $2,400,000
35 Scoping Estimate $10,400,000

Example updating and analysis to remain useful and to account for

A recent pilot plant estimate was developed using this evolution in the ways units are built.
methodology. When the estimate was made, the available Finally, it is worth emphasizing that the system cannot
data included the preliminary overall process flow steps, account for scope or basis changes.
a preliminary design basis for about 15% of the unit, the If an organization has little or no experience with
approximate module sizes/scope, and some global concerns. constructing pilot plants and has no historical data to draw
The pilot plant consisted of 35 modules with the complexity upon, developing scoping estimates is more difficult.
and constructability shown in Table 6. Table 7 shows how Another company may be able to provide cost data if their
these assessments were combined with data from the cost work is sufficiently similar. However, this is often difficult,
database to arrive at the scoping estimate of $10,400,000. as they are usually competitors; even if they are not, they
The definitive estimate prepared when the design was may be reluctant to release proprietary information. Consul-
completed was $10,800,000, or within 3.8% of the scoping tants or a pilot plant design firm can also help in develop-
estimate. Similar results have been obtained consistently ing a fast budgetary estimate. The quality of the estimate
over the last 20 years. will, in both cases, be strongly dependent on the skill of the
person or company making the estimate, their understand-
Lessons learned ing of your specific needs, and the accuracy of their concept
The better the data, the better the results will be. Hence, of what the final design will be. All are fraught with the
care taken in obtaining, organizing, and analyzing the data is potential to be sources of errors. In this case, the organiza-
effort well spent. tion may decide to expend the additional time and effort
The primary assessment is critical. Attempting to revise to produce a more-detailed estimate or accept that a much
it later is often less successful due to lingering prejudice larger contingency is required. CEP

from the initial assessment or to inflationary/deflationary

tendencies arising from the decision to reassess. Trained Richard P. Palluzi, P.E., CSP, is a Distinguished Engineering Associate
and experienced pilot plant estimators are also critical, and at ExxonMobil Research and Engineering, where he is involved in the
design, construction, and support of pilot plants and laboratories for
the process is much more effective if the assessor is also the ExxonMobil’s research site in Clinton, NJ, as well as affiliates world-
cost estimator. The estimator needs to fully understand the wide. He is the author of two books, 30 articles, and 40 presentations.
He is a past chair of the AIChE Pilot Plant Committee and he currently
system and the proposed unit. chairs ExxonMobil’s Pilot Plant and Laboratory Safety Standards
Care also needs to be taken to identify concerns about Committee. He is a member of AIChE, the International Society of
Automation, and the American Society of Safety Engineers. He is also
specialty equipment that are not addressed in the database. a member of the National Fire Protection Association (NFPA) and the
Typical and often-overlooked areas include structures, con- NFPA-45 Fire Protection for Laboratories Using Chemicals and NFPA-55
Industrial and Medical Gases committees. He has BE and ME degrees in
trol system issues, safety or fire protection, ventilation, and chemical engineering from Stevens Institute of Technology.
unique or novel conditions. The database also needs routine

44  www.aiche.org/cep  July 2011  CEP Copyright © 2011 American Institute of Chemical Engineers (AIChE)