R EPRINT A RTICLE Reprinted from the Oct.

01 Chemical Engineering journal

maturity, and a Class 1 estimate, with the
highest. Five characteristics are used to
Sharpen Your distinguish one class of estimate from
another: degree of project definition, end
Cost Estimating Skills use of the estimate, estimating
methodology, estimating accuracy, and
effort to prepare the estimate. Degree of
Larry R. Dysert, CCC, Chair, AACE International Technical Board project definition is the primary
characteristic used to identify an estimate
ABSTRACT: This classic article reviews the current AACE International Classification System. It class.
discusses various estimation methodologies, from conceptual to definitive, for calculating the The CPI rely on process flow diagrams
cost of capital projects in the CPI, and identifies the technical deliverables required to prepare (PFDs) and piping and instrument
each class of estimate. The techniques used for each type of estimate are discussed. These diagrams (P&IDs) as primary scope-
estimating methodologies, and the engineering information required to support them, should defining documents. These documents are
be understood by all engineers.
key engineering deliverables in
determining the level of project definition,
KEY WORDS: Capital budgets, projects, cost estimates, and estimation methodologies the maturity of the information used to
perform the estimate, and subsequently,
ncreasingly limited capital budgets in The Association for the Advancement of the estimate class. Incorporated into AACE

I the chemical process industries (CPI)

force management to make effective,
early decisions regarding investments in
Cost Engineering International (AACE
International;
aacei.org)
Morgantown,
recently
W.Va.;
developed
Internationals 18R-97 is also an estimate-
input checklist that identifies the
engineering deliverables used to prepare a
strategic assets, often with little or no recommended practices for cost-estimate project estimate, such as PFDs, process-
engineering input. As potential projects are classification [1] for the process industries. and utility-equipment lists, and
considered, these individuals often find This document, known as 18R-97, is a instrumentation-and-control system
themselves in situations where they must reference document that describes and drawings.
decide whether a specific project should be differentiate various types of project Estimating methodologies generally fall
continued. At each stage of the funding, estimates. into two broad categories: stochastic and
management requires costs estimates of AACE Internationals 18R-97 identifies deterministic. With stochastic methods, the
increasing accuracy. Determining which five classes of estimates, which it designates independent variables used in the
estimation method to use at each stage as Class 1, 2, 3, 4, and 5 (See table 1). A algorithm involve modeling (or factoring)
depends on the information available at the Class 5 estimate is associated with the based on inferred or statistical relationships
time of preparation, the end use of the lowest level of project definition or between costs and other design-related
estimate, and its desired accuracy.
This article discusses various estimation
methodologies, from conceptual to Capacity-Factored EstimateChanging Places
definitive, for calculating the cost of capital

A
projects in the CPI, and identifies the 100,000-bbl/d hydrogen-peroxide unit is to be built in Philadelphia and
technical deliverables required to prepare completed in 2002. In Malaysia, a similar plant, with a capacity of 150,000
each class of estimate. The techniques used bbl/d and a final cost of $50 million, was completed in 2000. Recent history
for each type of estimate are discussed. shows a capacity factor of 0.75 to be appropriate. The simple approach is to use the
These estimating methodologies, and the capacity-factor algorithm:
engineering information required to
support them, should be understood by all $B = $A(CapB/CapA)e
engineers. $B = $50M(100/150)(0.75) = $36.9M

Estimate Classifications However, a better estimate is obtained when adjustments for differences in scope,
Most organizations use some form of location and time are made. The plant in Malaysia includes piling, tankage and
classification system to identify the various owner costs that are not included in the plant to be built in Philadelphia. However,
types of estimates that may be prepared construction in Philadelphia is expected to cost 1.25 times that in Malaysia.
during the lifecycle of a project, as well as Escalation will be included as a 1.06 multiplier from 2000 to 2002. Additional costs
to indicate the overall maturity and for the Philadelphia plant (not included in the cost estimate of the Malaysian plant)
quality of the estimates being prepared. involve pollution control. The revised estimate appears below:
Unfortunately, there is often a lack of
Plant in Malaysia = $50M Escalate to 2002
consistency and understanding of the Deduct $10M for piling, tankage and $50M 3 1.06 = $53M
terminology used during classification, owner costs = $40M Capacity factor estimate
both across the process industries and Malaysia to Philadelphia adjustment $53M 3 (100/150)0.75 = $39M
within individual companies or $40M 3 1.25 = $50M Add $5M for pollution requirements
organizations. = $44M

22 Cost Engineering Vol. 45/No. 6 JUNE 2003

Table 1Cost Estimate Classification Matrix

Estimate Project Definition Purpose of Estimating Accuracy Range Preparation Effort

Class (% of complete Estimate Method (variation in low (Index relative to
definition) and high ranges) project cost)
Class 5 02 Screening Capacity-factored, L: 20 to 50% 1
parametric models H: 30 to 100%
Class 4 115 Feasibility Equipment-factored, L: 15 to 30% 24
parametric models H: 20 to 50%
Class 3 1040 Budget Semi-detailed unit- L: 10 to 20% 310
authorization cost estimation with H: 10 to 30%
or cost control assembly-level line items
Class 2 3070 Control of bid Detailed unit-cost L: 5 to 15% 420
or tender estimation with forced, H: 5 to 20%
detailed takeoff
Class 1 50100 Check estimate, Semi-detailed unit cost L: 3 to 10% 5100
bid or tender estimation with detailed H: 3 to 15%
takeoff
The stage of process technology and the availability of cost data strongly affect the accuracy range of an estimate. Plus-or-minus high (H) and low (L)
values represent the variation in actual costs versus estimated costs, after applying contingency factors. The preparation effort uses an index to describe
the cost required to prepare an estimate, relative to that for preparing a Class 5 estimate. For example, if it costs 0.005 percent of the project cost to
develop a Class 5 estimate, then a Class 1 estimate could require as much as 100 times that, or 0.5 percent of the total project cost.

parameters. For deterministic methods, nonlinear relationship between capacity plant increases costs by approximately 100
the independent variables used in the and cost shown in the following equation: percent.
algorithm are a direct measure of the item In reality, as plant capacities increase,
being estimated, such as straightforward $B/$A = (CapB/CapA)e (equation 1) the exponent tends to increase as
counts or measures of items, multiplied by where $A and $B are the costs of the two illustrated in Figure 1. The capacity factor
known unit costs. Deterministic similar plants, CapA and CapB are the exponent between plants A and B may
estimating methods require the quantities, capacities of the two plants, and e is the have a value of 0.6; between B and C, the
pricing and completeness of scope to be exponent or proration factor. The value of exponent may have a value of 0.65; and
known with relative certainty. As the level the exponent typically lies between 0.5 between C and D, the exponent may have
of project definition increases, the and 0.85, depending on the type of plant, risen to 0.72. As plant capacity increases to
estimating methodology tends to progress and must be analyzed carefully for its the limits of existing technology, the
from stochastic (or factored) methods to applicability to each estimating situation. exponent approaches a value of one. At
deterministic methods. The e used in the capacity factor this point, it becomes as economical to
equation is actually the slope of the log- build two plants of a smaller size, rather
Capacity-Factored Estimates curve that has been drawn to reflect the than one large plant.
Generated during the feasibility stage of change in the cost of a plant as it is made Table 2 shows the capacity factors for
a project, the capacity-factored estimate larger or smaller. These curves are several chemical process plants [4].
(CFE) provides a relatively quick, and typically drawn from the data points of the Unfortunately, most of the data are quite
sufficiently accurate means of known costs of completed plants. With an
determining whether a proposed project exponent less than 1, scales of economy
should be continued. It is a good method are achieved such that as plant capacity
to use when deciding between alternative increases by a percentage (say, by 20
designs or plant sizes. This early screening percent), the costs to build the larger plant
method (Class 5 estimate) is most often increase by less than 20 percent.
used to estimate the cost of battery-limit The methodology of using capacity
process facilities, but may also be applied factors is sometimes referred to as the
to individual equipment items. scale of operations method, or the six-
When estimating via CFE, the cost of a tenths factor method because of the
new plant is derived from the cost of a reliance on an exponent of 0.6 if no other
similar plant of a known capacity, with a information is available [2,3]. With an
similar production route (such as, both are exponent of 0.6, doubling the capacity of a
batch processes), but not necessarily the plant increases costs by approximately 50 Figure 1The capacity-factored relationships
same end products (products should be percent, and tripling the capacity of a shown here are logarithmic. Exponents differ
relatively similar, however). It relies on the across capacity ranges CapA is the capacity of
plant A, and so on.

Cost Engineering Vol. 45/No. 6 JUNE 2003 23

 old. Nowadays, however, companies are size. Deduct costs from the known base capacity-factor algorithm to adjust for plant
less likely to make these data available, and case that are not applicable in the new size. Cost indices are used to accommodate
recent studies have been sparse. This data plant. Apply location and escalation the inflationary impact of time (See the
should be used for reference only, and with adjustments to normalize costs and use the How to Use a Cost Index box on page
caution regarding its applicability to any
particular situation. Actual Capacity-Increase Multiplier (CapB/CapA)
If the capacity factor used in the Exponent
1.5 2 2.5 3 3.5 4 4.5 5
estimating algorithm is relatively close to
0.20 23% 41% 58% 73% 88% 100% 113% 124%
the actual value, and if the plant being 0.25 20% 36% 51% 64% 75% 87% 97% 106%
estimated is relatively close in size to the 0.30 18% 32% 44% 55% 64% 74% 83% 91%
similar plant of known cost, then the 0.35 16% 28% 38% 47% 55% 63% 70% 76%
potential error from a CFE is certainly well 0.40 13% 23% 32% 39% 46% 52% 57% 63%
within the level of accuracy that would be 0.45 11% 18% 26% 32% 36% 41% 46% 50%
expected from a stochastic method. 0.50 9% 15% 20% 25% 28% 32% 35% 38%
However, one must also account for 0.55 6% 11% 15% 18% 21% 23% 25% 28%
differences in scope, location, and time. 0.60 4% 7% 10% 12% 13% 15% 16% 18%
Keep in mind that each of these 0.65 2% 3% 5% 6% 6% 7% 8% 8%
0.70 0% 0% 0% 0% 0% 0% 0% 0%
adjustments adds additional uncertainty
0.75 -2% -4% -5% -5% -6% -7% -7% -8%
and potential error to the estimate. Table 3
0.80 -4% -7% -9% -10% -12% -13% -14% -15%
shows the percent error that may occur if 0.85 -6% -10% -13% -15% -17% -19% -20% -21%
an assumed capacity factor of 0.7 is used, 0.90 -8% -13% -17% -20% 22% -24% -26% -28%
and the actual value is different [5]. For 0.95 -10% -16% -21% -24% -27% -29% -31% -33%
example, if the new plant is triple the size 1.00 -11% -19% -24% -28% -31% -34% -36% -38%
of an existing plant, and the actual capacity 1.05 -13% -22% -28% -32% -36% -39% -41% -43%
factor is 0.80 instead of the assumed 0.70, 1.10 -15% -24% -31% -36% -40% -43% -45% -47%
one will have underestimated the cost of 1.15 -16% -27% -34% -39% -43% -46% -49% -52%
the new plant by only 10 percent. 1.20 -18% -30% -37% -42% -47% -50% -53% -55%
Similarly, for the same threefold scaleup in Table 3Percentage Error Factor of 0.7 is Used for Estimate
plant size, if the capacity factor should be
0.60 instead of the assumed 0.70, one will
have overestimated the plant cost by only
12 percent. These data were generated
How to Use a Cost Index
with equation 1. The capacity-increase cost index is table of values that relates the costs of specific items at various
multiplier is CapA/CapB and in the base, e
is 0.7. The error occurs as e varies from 0.7.
To use the CFE method prudently, make
A dates to a specific time in the past. Cost indices are useful to adjust costs for
inflation over time. Chemical Engineering (CE) publishes several useful
cost indices at the back of the magazine each month. Of particular importance to
sure the new and existing known plant are CPI are the CE Plant Cost Index and the Marshall & Swift Equipment Cost
near-duplicates, and are reasonably close in Index.
The CE Cost Index provides values for several plant-related costs, including
various types of equipment, buildings, construction labor and engineering fees.
Product Factor
These values relate costs of overall plants over time, using the 19571959 timeframe
Acrolynitrile 0.60
as the base period (value = 100). The Marshall & Swift indices provide equipment-
Butadiene 0.68
Chlorine 0.45
cost-index values arranged in accordance to the process industry in which the unit is
Ethanol 0.73 used. This index uses the year 1926 as the base period.
Ethylene oxide 0.78 To use one of these indices to account for cost escalation, multiply the cost to be
Hydrochloric acid 0.68 escalated by the ratio of the index values for the years in question. For example,
Hydrogen peroxide 0.75 suppose you want to determine the cost of a new chlorine plant using capacity-factored
Methanol 0.60 estimates. You discover that a similar chlorine plant built in 1994 cost $25M. Before
Nitric acid 0.60 applying the capacity-factor equation (equation1), the cost of the 1994 must be
Phenol 0.75 normalized for 2001.
Polymerization 0.58 The CE index value for 1994 is 368.1. The February 2001 value is 395.1. The
Polypropylene 0.70
escalated cost of the chlorine plant is therefore:
Polyvinyl chloride 0.60
Sulfuric acid 0.65
Styrene 0.60 $25M 3 (395.1/368.1) = $25M 3 1.073 = $26.8M
Thermal cracking 0.70
Urea 0.70 Readers should note that the CE Plant Cost Index is being revised to account
Vinyl acetate 0.65 for changes in many of the individual indexes and reports upon which it is based.
Vinyl chloride 0.80 Look for a future article in CE that will introduce and describe the revised CE Plant
Cost Index.
Table 2Capacity Factors for Process Units [4]

24 Cost Engineering Vol. 45/No. 6 JUNE 2003

24). Finally, add any additional costs that Equipment factored estimates (EFEs, reason to pursue the project. If so, then
are required for the new plant, but were Class 4) are typically prepared during the use this estimate to justify the funding
not included in the known plant. feasibility stage of a project, when required to complete additional
engineering is approximately 115 engineering and design for a Class 3 or
Equipment-Factored Estimates percent complete, and are used to budget estimate.
determine whether there is sufficient

Equipment Factors:
Understanding Them Is As Important As Learning How to Use Them
t is extremely important to understand the basis behind

I
Three specific variables affect the equipment cost to a
the equipment factors being used, and to account for all greater degree than they affect the cost of the bulk materials
costs that are not covered by the factors themselves. The and installation [10]: the size of the major equipment, the
factors may apply to total installed costs (TICs) or direct field materials of construction and the operating pressure. As the
costs (DFCs) for the facility. Usually, the factors generate costs size of a piece of major equipment gets larger, the amount of
for inside battery limits (ISBL) facilities, and require outside- corresponding bulk materials (foundation, support steel,
battery-limit facilities (OSBL) costs to be estimated separately. piping and instruments) required for installation does not
In some cases, factors are used to estimate the costs of the increase at the same rate. Thus, as the equipment increases in
complete facilities. size, the value of the equipment factor decreases. A similar
Hans Lang introduced the concept of using the total cost tendency exists for metallurgy and operating pressure. If the
of equipment to estimate the total cost of a plant [7,8]: equipment is made from expensive materials (stainless steel,
Total plant cost (TPC) = Total equipment cost (TEC) 3 titanium or Monel), or if the operating pressure increases, the
Equipment factor (equation 2) equipment factor decreases. These three variables could be
summarized into a single attribute known as the average unit
Lang proposed three separate factors based on the type of cost of equipment, or the ratio of total cost of process
process plant. For solids, the factor is 3.10; for combined solids equipment to the number of equipment items [11,12].
and fluids, 3.63; and for fluids alone, 4.74. These factors were
meant to cover all the costs associated with the TIC of a plant, The Discipline Method
including the ISBL costs and OSBL costs. Here is an example Another way to use equipment factors, aside from
of a Lang-factor estimate for a fluid-process plant: calculating DFC or TIC, is to generate separate costs for each
of the disciplines associated with the installation of equipment.
TEC = $1.5M Heres how: Each type of equipment is associated with several
TPC = $1.5M 3 4.74 = $7.11M discipline-specific equipment factors. For example, one
discipline-equipment factor will generate costs for concrete, a
Langs approach was simple, utilizing a factor that varies second factor will generate costs for structural steel, and a third
only by the type of process. Today, many different methods of will generate the costs for piping. An advantage to this
equipment factoring have been proposed. The Lang factor, approach is that it provides the estimator with the capability to
however, is often used generically to refer to all the different adjust the costs for the individual disciplines based on specific
types of equipment factors [9]. knowledge of the project conditions, and improves the
W. E. Hand elaborated on Langs work by proposing using accuracy of the equipment factoring method. It also allows the
different factors for each type of equipment (columns, vessels, costs for each specific discipline to be totaled, and compared
heat exchangers and other units) rather than process type. to those of similar projects [13,14].
Hands equipment factors estimate DFCs, excluding An example of discipline-specific equipment factors is
instrumentation, and range from 2.0 to 3.5, which might shown in Table 5. The total DFC costs for installation of this
correlate to approximately 2.4 to 4.3 if instrumentation were heat exchanger totals $28,600 (including the equipment
included. Hands factors exclude indirect field costs (IFC), purchase cost of $10,000). This equates to an overall DFC
home office costs (HOC), and the costs for OSBL facilities, all equipment factor of 2.86. These costs do not include IFC,
of which must be estimated separately. HOC or OSBL costs.
We will run an estimate prepared for a fluid processing Development of the actual equipment factors used for
plant using Hands equipment factoring techniques in Table 4. process plant estimates is time-consuming. Although some
Each equipment item has its own factor. The ratio of DFC to published data exists, much of these data are old, and some of
TEC to is 2.8, whereas a typical value would range from 2.4 to the assumptions in normalizing the data for time, location and
3.5. The ratio of TFC to TEC is 3.6, while a typical value scope are incomplete or unavailable. For lack of anything
would range from 3.0 to 4.2, and the ratio of total project cost better, this data is an acceptable source of equipment factors.
(including contingency) to TEC is 5.1. A typical value would However the best information will be that which come from a
range from 4.2 to 5.5. This correlates closely with Langs cost database that reflects the companys project history.
original overall equipment factor of 4.74 for fluid plants.

Cost Engineering Vol. 45/No. 6 JUNE 2003 25

 An EFE can be quite precise if the
equipment factors are appropriate, if the
correct adjustments have been applied, Item Description Equipment Equipment Total, $ Derived
Cost, $ Factor Multiplier
and if the list of process equipment is
Columns 650,000 2.1 1,365,000
complete and accurate. These estimation
techniques have an advantage over CFEs Vertical vessels 540,000 3.2 1,728,000
in that they are based upon the specific Horizontal vessels 110,000 2.4 264,000
process design. Typically, EFEs rely on Shell-and-tube
the existence of a ratio between the cost of heat exchangers 630,000 2.5 1,575,000
an equipment item and costs for the Plate heat exchangers 110,000 2.0 220,000
associated non-equipment items, such as Pumps, motors 765,000 3.4 2,601,000
foundations, piping, and electrical
Raw equipment costs
components needed when building a
(TEC) 2,805,000
plant.
The first step when preparing an EFE is Direct field cost (DFC) = 2,805,000 3 2.8 7,754,000 2.8
to estimate the cost for each piece of Direct field labor
process equipment [6]. Examine the (DFL) cost = DFC 3 25% 1,938,000
equipment list carefully for completeness, Indirect field costs (IFC): Temporary construction facilities; construction services,
and compare it against the PFDs and supplies and consumables; field staff and subsistence expenses; payroll, benefits,
P&IDs. However, there is a problem insurance; construction equipment and tools
when an EFE is prepared, the equipment IFC = DFL 3 115% 2,229,000
list is often in a preliminary stage. Total field costs (TFC) = DFC + IFC 9,982,000 3.6
Although the major equipment is
Home-office costs (HOC): Project management,controls and estimating criteria,
identified, it may be necessary to assume a
procurement, construction management, engineering and design, and home-
cost percentage for auxiliary equipment
office expenses
that has not yet been defined.
This is the time to verify equipment HOC = DFC 3 30% 2,326,000
sizing. Equipment is often sized at 100 Subtotal project cost = TFC + HOC 12,308,000 4.4
percent of normal operating duty, but by Other project costs (OTC), including project commissioning costs
the time the purchase orders have been Commissioning = DFC 3 3% 233,000
issued, some percentage of oversizing has Contingency = (TFC + HOC) 3 15% 1,846,000
been added to the design specifications.
Total OTC 2,079,000
The percentage of oversizing varies with
the type of equipment, as well as with the Total installed project
organizations procedures and guidelines. cost (TIPC) = 14,387,000 5.1
It is prudent to check with the process Note: In the table above, the multiplier is the ratio of DFC, TFC, TIPC and other
engineers and determine if an allowance costs to the raw-total equipment cost of $2,805,000.
for oversizing the equipment, as listed on In this table, the cost of each type of equipment was multiplied by a factor to derive
the preliminary equipment list, should be the installed DFC for that unit. For instance, the total cost of all vertical vessels
added before pricing the equipment. ($540,000) was multiplied by an equipment factor of 3.2 to obtain an installed DFC
The purchase cost of the equipment is of $1,728,000. The total installed cost (TIC) for this project is $14,387,000.
often obtained from: purchase orders,
published equipment-cost data, and Table 4Equipment-Factored Estimating Example
vendor quotations. Since the material cost
of equipment can represent 2040 percent
of the total-project costs for process plants, Factor Cost, $
it is extremely important to estimate the Equipment cost 1.0 10,000
equipment costs as accurately as possible. Installation labor 0.05 500 This table showcases the
If historical purchase information is used, equipment factors for a Type 316
Concrete 0.11 1,100
make sure that the costs are escalated stainless steel heat exchanger
Structural steel 0.11 1,100
appropriately, and that adjustments are with a surface area of 2,400 ft2.
Piping 1.18 11,800 The purchase cost of $10,000 is
made for location and market conditions.
Once the equipment cost is established, Electrical parts 0.05 500 multiplied by each factor to
the appropriate equipment factors must be Instrumentation 0.24 2,400 generate the DFC for that
generated and applied. (See the Painting 0.01 100 discipline
Equipment Factors box on the previous Insulation 0.11 1,100
page. Tables 4 and 5 and [714] are Total DFC 2.86 $28,600
referenced in this sidebar). In doing so,
one must make the necessary adjustments Table 5Heat Exchanger Discipline-Equipment Factors

26 Cost Engineering Vol. 45/No. 6 JUNE 2003

for equipment size, metallurgy, and Data collection and development for a that is the best fit for the data will typically
operating conditions. parametric estimating model require a have the highest r-squared (R2) value. R2
Specific project or process conditions significant effort. Both cost and scope provides a measure of how well the
must be evaluated. For example, if the plot information must be identified and algorithm predicts the calculated costs.
layout of the project requires much closer collected. It is best to collect cost data at a However, a high R2 value by itself does not
equipment placement than is typical, one fairly low level of detail [15]. The cost data imply that the relationships between the
may want to make adjustments for the can always be summarized later if an data input and the resulting cost are
shorter piping, conduit, and wiring than aggregate level of cost information statistically significant. One still needs to
would be accommodated by the the provides a better model. It is obviously examine the algorithm to ensure that it
equipment factors. Or, if a project is important to include the year for the cost makes sense.
situated in an active seismic zone, one data in order to normalize costs later. The A cursory examination of the model can
may need to adjust the factors for type of data to be collected is usually help identify the obvious relationships that
foundations and support steel. After decided upon in cooperation with the are expected. If the relationships from the
developing equipment factored costs, one engineering and project personnel. It is model appear to be reasonable, then
must account for project costs that are not best to create a formal data-collection additional tests (such as the t-test and f-
covered by the equipment factors, such as form that can be consistently used, and test) can be run to determine statistical
by generating indirect field costs (IFCs) revised if necessary. significance and to verify that the model is
and home-office costs (HOCs). After the data have been collected, it providing results with an acceptable range
must be normalized. By doing this, we of error. A quick check can be performed
Parametric-Cost Estimation make adjustments to account for by running the regression results directly
A parametric-cost model is an extremely escalation, location, site conditions, against the input data to see the percent
useful tool for preparing early conceptual system specifications and cost scope. Data error for each of the inputs. This allows
estimates when there are little technical analysis, the next step in the development the estimator to determine problems and
data or engineering deliverables to provide of a parametric model, is achieved by a refine the algorithms. After the individual
a basis for using more-detailed estimating wide variety of techniques that are too algorithms have been developed and
methods. A parametric model is a complex to delve into in this article [16]. assembled into a complete parametric cost
mathematical representation of cost Typically, data analysis entails model, it is important to test the model as
relationships that provide a logical and performing regression of cost versus a whole against new data (data not used in
predictable correlation between the selected design parameters to determine the development of the model) for
physical or functional characteristics of a the key drivers for the model. It is verification.
plant and its resultant cost. Capacity- and understood that regression involves During the data-application stage, a user
equipment-factored estimates are simple iterative experiments to find the best-fit interface and a presentation form for the
parametric models. Sophisticated algorithms or mathematical relationships parametric cost model is established.
parametric models involve several that describe how data behave. The result Electronic spreadsheets provide an
independent variables or cost drivers. is a parametric model. Most spreadsheet excellent means of accepting estimator
The first step in developing a parametric applications provide regression analysis input, calculating costs based upon
model is to establish its scope. This and simulation functions that are algorithms and displaying output.
includes defining the end use, physical reasonably simple to use. Perhaps the most important effort in
characteristics, critical components, and As an algorithm is discovered that developing a parametric (or any other)
cost drivers of the model. The end use of appears to provide good results, it must be cost model is making sure the application
the model is typically to prepare tested to ensure that it properly explains is thoroughly documented. Record the
conceptual estimates for a process plant or the data. Advanced statistical tools can actual data used to create the model, the
system and takes into consideration the quicken the process, but can be more resulting regression equations, test results
type of process to be covered, the type of difficult to use. Sometimes, erratic or and a discussion on how the data was
costs to be estimated (such as TIC and outlying data points will need to be adjusted or normalized for use in the data-
TFC) and the accuracy range. removed from the input data in order to analysis stage. Any assumptions and
The model should be based on actual avoid distortions in the results. The allowances designed into the cost model
costs from completed projects and reflect algorithms will usually take one of the should be documented, as should any
the organizations engineering practices following forms [17]: exclusions. The range of applicable input
and technology. It should use key design A linear relationship, such as, values, and the limitations of the models
parameters that can be defined with algorithms should also be noted. Write a
reasonable accuracy early in the project Cost = a + bV1 + cV2 + ... (equation 3) user manual to show the steps involved in
scope development, and provide the or a nonlinear relationship, such as, preparing an estimate using the cost
capability for the estimator to easily adjust model, and to describe the required inputs
the derived costs for specific factors Cost = a + bV1x + cV2y + (equation 4) to the cost model. An example of
affecting a particular project. Finally, the where V1 and V2 are input variables; a, b, developing a parametric estimating model
model should generate current year costs is described in the Parametric Equations
and c are constants derived from
or have the ability to escalate to current and Data Manipulation box. (Table 6 and
regression; and x and y are exponents
year costs. figure 2 are referenced in this sidebar).
derived from regression. The equation

Cost Engineering Vol. 45/No. 6 JUNE 2003 27

 There are various degrees of detail in a detailed estimating methods are used with
Detailed-Cost Estimation detailed cost estimate. In a completely incomplete design information. Typically,
A detailed estimate is one in which each detailed estimate, all costs are considered in a forced-detailed estimate, detailed
component contained a project scope including the direct field cost (DFC), takeoff quantities are generated from
definition is quantitatively surveyed and IFC, HOC and all other miscellaneous preliminary drawings and design
priced using the most realistic unit prices costs for both the inside battery limits information.
available. Detailed estimates are typically (ISBL) and outside battery limits (OSBL) Planning: The detailed estimate is
prepared to support final budget facilities. In a semi-detailed estimate, costs typically used to support cost control
authorization, contractor bid tenders, cost for ISBL process facilities are factored, during execution of the project. The first
control during project execution and and the costs for the OSBL facilities are step is to establish a project-estimate basis
change orders. Detailed estimates can be detailed. In a forced-detailed estimate, and schedule. This may involve several
very accurate (Class 3 through Class 1).
However, completeness of the design Cooling Temperature, Flowrate, Actual Predicted % Error
information is critical. If an engineering Range, F Approach, F gal/min Cost, $ Cost, $
drawing or other information is missing, 30 15 50,000 1,040,200 1,014,000 -2.5%
then the scope items covered by those 30 15 40,000 787,100 843,000 7.1%
documents will not be included in the
estimate, and the results will not be on par 40 15 50,000 1,129,550 1,173,000 3.8%
with a Class 1 or 2 estimate. 40 20 50,000 868,200 830,000 -4.4%
It is not unusual for detailed estimates 25 10 30,000 926,400 914,000 -1.3%
on very-large projects to take several 35 8 35,000 1,332,400 1,314,000 -1.4%
weeks, if not months, to prepare, and to
require thousands of engineering hours to Table 6Actual Costs Versus Predicted Costs With Parametric Equation
generate the technical deliverables (the
engineering and design data). At the very
least, this includes PFDs and utility flow Parametric Equations and Data Manipulation
drawings, P&IDs, equipment data sheets,
motor lists, electrical diagrams, piping nduced-draft cooling towers are typically used in process plants to provide a
isometrics (for alloy and large diameter
piping), equipment and piping layout
drawings, plot plans and engineering
I recycle cooling-water loop. These units are generally prefabricated, and
installed on a subcontract or turnkey basis by the vendor. Key design
parameters that appear to affect the costs of cooling towers are the cooling range, the
specifications. temperature approach and the water flowrate. The cooling range is the temperature
Pricing data should include vendor difference between the water entering the cooling tower and the water leaving it.
quotations, pricing information from The approach is the difference in the cold water leaving the tower and the wet-bulb
recent purchase orders, current labor temperature of the ambient air.
rates, subcontract quotations, project Table 6 provides the actual costs and design parameters of six recently completed
schedule information (to determine units whose costs have been normalized (adjusted for location and time) to a Northeast
escalation requirements) and the US, year-2000 timeframe. These data are the input to a series of regression analyses that
construction plan (to determine labor are run to determine an accurate algorithm for estimating costs. Using a Microsoft Excel
productivity and other adjustments). spreadsheet, the following cost-estimation algorithm was developed:

Predicted Cost = $86,600 +

$84,500 3 (Cooling Range, F)0.65 $68,600 3 (Approach, F) +
$76,700 3 (Flowrate, 1,000 gal/min)0.7 (equation 5)

Equation 5 demonstrates that the cooling range and flowrates affect cost in a non-
linear fashion, while the approach affects cost in a linear manner. Increasing the
approach will result in a less costly cooling tower, since it increases the efficiency of
the heat transfer taking place. These are reasonable assumptions. The regression
analysis resulted in an R2 value of 0.96, which indicates that the equation is a good-
fit for explaining the variability in the data.
In Table 6, the actual costs and the predicted costs from the estimating algorithm
are shown. The percentage of error varies from 4.4 percent to 7.1percent. The
estimating algorithm developed from regression analysis, can be used to develop
cost-versus-design parameters that can be represented graphically (Figure 2). This
information can then be used to prepare estimates for future cooling towers. It is
Figure 2This graph, developed from regression fairly easy to develop a spreadsheet model that will accept the design parameters as
data for tower cost versus design parameters
input variables, and calculate the costs based on the parametric-estimating
(Table 6), such as flowrate, can be used to prepare
estimates for future cooling towers.
algorithm.

28 Cost Engineering Vol. 45/No. 6 JUNE 2003

estimators and extensive support from provide detailed manhour estimates for (for instance, vessel internals). Depending
engineering to review the organizations their project activities, and the appropriate on the information available, the labor
estimating guidelines and procedures. wage rates, as approved by management, hours needed to set and erect a heavy
The schedule indicates when the are applied by the estimating engineers. vessel may not include the hours to erect,
deliverables are to be supplied, when each Home-office overhead factors are used to take down, and dismantle a derrick or
major section of the estimate should be project overhead costs and expenses. other special lifting equipment. Special
completed, and when reviews of the Local sales-tax rates or duties may need consideration may also be required to
estimate will take place. Any exclusions to be included in the estimation. Estimate ensure that costs for calibration, soil
that are known at this time should be escalation costs based on the project settlement procedures, special internal
reviewed and documented. A kickoff schedule. Depending on the contracting coatings, hydrotesting and other tests are
meeting should be scheduled to inform strategy and schedule of delivery, project- included in the estimate. Some
the project team of the roles and fee estimates may be included. Finally, equipment may be erected by
responsibilities of the various participants, perform a risk analysis and include the subcontractors or the vendor, and
and allow them to review the plans for appropriate contingency in the estimate. included in the material purchase costs.
estimate preparation. On very large Pay close attention to pricing the process Care must be taken to identify these
projects, it is helpful to appoint a few key equipment, as it contributes to 20-40 situations.
contact people who will act as liaisons percent of the facilitys TIC. The
between estimating and engineering minimum information required for lthough detailed estimates are
personnel.
Estimation activities: Preparation of the
DFC estimate is the most intensive
pricing equipment includes the PFDs,
equipment lists and data sheets, which are
usually prepared by the process-
A desirable for final budget
authorization, the level of
engineering progress needed and the time
activity in detailed cost estimation. The engineering group. Process and required for preparation will sometimes
project scope should be reviewed and mechanical engineers are in the best prevent them from being used for this
understood, and all technical deliverables position to make accurate estimates of purpose. In todays economy, budgeting
assembled. On large projects, engineering equipment pricing, since they are usually and investment decisions are often needed
drawings and technical information may in close contact with potential equipment sooner than a detailed estimate would
be submitted to the estimating department vendors. Whenever possible, the vendors allow. Semi-detailed and forced-detailed
over time. All information received from should provide the equipment-purchase estimates will often be employed for final
engineering should be logged. costs to the estimator. budget authorizations, and a completely
The estimate takeoff is determined by Slight differences in equipment detailed estimate may be prepared later, to
quantifying all of the various material and specifications can sometimes result in support project control.
labor components of the estimate, to large differences in pricing. Formal When deciding upon potential
ensure that all quantities are accounted vendor quotes are preferred; however, investment opportunities, management
for, but not double-counted. Material time constraints in preparing the estimate must employ a cost-screening process that
pricing is applied using the best cost often do not allow for solicitation of requires various estimates to support key
information available. The labor worker formal vendor quotes. In this case, decision points. At each of these points,
hours are assigned and adjusted for labor equipment pricing may depend on the level of engineering and technical
productivity, and wage rates are applied. informal sources, such as phone information needed to prepare the
Allowances, or costs that are factored from discussions, in-house pricing data, recent estimate will change. Accordingly, the
more-significant equipment, are made for purchase orders, capacity-factored techniques used prepare the estimates will
bolts, gaskets, hangers for pipes and estimates from similar equipment, or from vary. The challenge for the engineer is to
similar items. For example, bolts and parametric pricing models. know what is needed to prepare these
fittings may be factored as 3 percent of the Equipment-installation costs are usually estimates, and to ensure they are well
piping cost. Finally, the DFC estimate is prepared by the estimator, with assistance documented, consistent, reliable,
summarized, formatted and reviewed for from construction personnel. accurate and supportive of the decision-
completeness and accuracy. Construction companies may be called making process.
After the DFC estimate has been upon for heavy lifts, or for help where
prepared, start the IFC estimate. The total special installation methods may be used.
manhours, identified from the DFC, serve The placement of large process Recommended Reading
as the basis for factoring many of the of equipment in an existing facility may also
IFC costs (refer to table 4). Indirect-labor require special consideration. Worker 1. AACE Internationals 18-R97,
wage rates and staff-labor rates are hours for equipment installation are "Recommended Practice for Cost
Estimate Classification - As Applied in
established and applied, and any indirect usually based on weight and equipment
Engineering, Procurement, and
estimate allowances are taken into dimensions, which are obtained from the Construction for the Process
account. The construction manager equipment process-data sheets. Industries," AACE International,
should be involved in the initial review of When referencing the labor worker Morgantown, W.Va., 1997.
the IFC estimate. hour data for equipment, the estimator 2. Williams Jr., R., Six-Tenths Factor Aids in
The HOC estimate follows. Project must be careful to include all labor Approximating Costs, Chem. Eng.,
administrators and engineers should associated with the pieces of equipment December 1947.

Cost Engineering Vol. 45/No. 6 JUNE 2003 29

 3. Chilton, C.H., Six Tenths Factor Applied to Complete Plant Costs,
Chem. Eng., New York, April 1950.
4. Guthrie, K.M., Capital and Operating Costs for 54 Chemical
Processes, Chem. Eng., June 1970.
5. Remer, D. and Chai L., Estimate Costs of Scaled-Up Process Plants,
Chem. Eng., April 1990.
6. Hall, Richard S. and W.M. Vatavuk, Estimating Process Equipment
Costs, Chem Eng., New York, November 1988.
7. Lang, H.J., Cost Relationships in Preliminary Cost Estimation,
Chem. Eng., October 1947.
8. Lang, H.J., Simplified Approach to Preliminary Cost Estimates,
Chem. Eng., June 1948.
9. Hand, W.E., "Estimating Capital Costs from Process Flow Sheets,"
Cost Engineer's Notebook, AACE International, Morgantown,
W.Va., January 1964.
10. Miller, C.A., "Capital Cost Estimating - A Science Rather than an
Art," Cost Engineer's Notebook, AACE International,
Morgantown, W.Va., 1978.
11. Rodl, R., others, "Cost Estimating for Chemical Plants," 1985
AACE International Transactions, AACE International,
Morgantown, W.Va.
12. Nishimura, M., "Composite-Factored Estimating," 1995 AACE
International Transactions, AACE International, Morgantown,
W.Va.
13. Miller, C.A., New Cost Factors Give Quick Accurate Estimates,
Chem. Eng., September 1965.
14. Guthrie, K.M., Data and Techniques for Preliminary Capital Cost
Estimating, Chem. Eng., March 1969.
15. Rose, A., "An Organized Approach to Parametric Estimating,"
Transactions of the Seventh International Cost Engineering
Congress, 1982.
16. Black, J.H., "Application of Parametric Estimating to Cost
Engineering," 1984 AACE International Transactions, AACE
International, Morgantown, W.Va.
17. Dysert, L.R., "Developing a Parametric Model for Estimating
Process Control Costs," 1999 AACE International Transactions,
AACE International, Morgantown, W.Va.
18. Enyedy, Gustav, How Accurate Is Your Cost Estimate? Chem. Eng.,
July 1997.
19. Gerrard, Mark, "Capital Cost Estimating," University of
Teesside,October 2000, ISBN# 0-85295-3991.

AACE Internationals 18-R97 Recommended Practice for Cost

Estimate Classification - As Applied in Engineering, Procurement,
and Construction for the Process Industries is available for viewing
on the AACE International website, www.aacei.org.

About the Author

Larry R. Dysert, CCC is lead estimator with Intel Corporation.
Dysert has 21 years of project consulting experience in the chemical
process, petroleum refining and construction industries. He is an active
member of AACE International, the Association for the Advancement of
Cost Engineering International, serves as chairman of the technical
board has spoken at numerous AACE International conferences, and
has taught project estimation at the Rochester Institute of Technology
(N.Y.). Dysert has a B.S. in economics from the Univ. of Calif. in San
Diego and an advanced-graduate degree in economics from Univ. of
Calif. at Santa Barbara. He may be contacted by phone at 971-214-9940
or e-mail at [email protected].

Reprint Articles - This article is reprinted by permission from Chemical

Engineering, Vol. 108, No. 11 (Oct. 2001) Copyright 2001, by Chemical
Week Associates, New York, NY 10038. The Cost Engineering publications
staff and AACE International administrative staff select reprint articles based
on their perceived benefit to AACE International members and other readers
of the Cost Engineering journal.

30 Cost Engineering Vol. 45/No. 6 JUNE 2003