See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/282212077

Chemical Processes Equipment Cost Estimation Using Parametric Models

Article · May 2006

CITATIONS READS
10 23,085

1 author:

mohammad reza Shabani

Research Institute of Petroleum Industry (RIPI)
15 PUBLICATIONS 131 CITATIONS

SEE PROFILE

All content following this page was uploaded by mohammad reza Shabani on 27 September 2015.

The user has requested enhancement of the downloaded file.

 T ECHNICAL A RTICLE
in which Ws is the shell weight. The
maximum error is ± this amount.
Chemical Processes Equipment By using such relationships, one can
attain acceptable results fast. Some
Cost Estimation Using engineering cost estimation simulation
software allows costs to be estimated by

Parametric Models using such relationships.

A cost database can be used to derive
such relationships. The database is based
Mohammed Reza Shabani and Reza Behradi Yekta
on the costs of material and labor in
ABSTRACT: Parametric cost estimation of process equipment is introduced in this article as a industrialized countries. These totals
shortcut method in fast estimation of equipment cost. Cost relations are prepared for some differ from the costs in developing
equipment, such as atmospheric and pressurized storage tanks, separation towers, heat exchang- countries. So the main goal of this article
ers and compressors. These relations can be proposed as a sample model to organizations and is to show how one can derive the needed
engineering companies to drive their own cost relations by referring to their past project cost data based upon local costs.
archives. In the preliminary stages of a project,
detailed data about equipment and other
KEY WORDS: Cost estimation, parametric models, and process equipment
capital cost items does not exist for
submission to equipment manufacturers.
enerally, various cost estimation Process Equipment Parametric Cost

G
Still, one can estimate equipment cost
methods are prepared in Euro, Estimation and finally the total plant cost at an
American, or industrialized Generally the main process acceptable error percentage.
countries that enjoy modern technologies equipment cost for a plant consists of 20- To obtain a parametric model, one
in process equipment manufacturing. 40 percent of the fixed capital investment needs to have data on equipment costs
These methods can be used in other of a project [1]. An accurate estimation or from a specified time duration. Then,
countries by adjusting for location factors. determination is very important. This is using statistical methods, one can obtain
Because of the industrial growth in especially true when using equipment the models coefficients and study their
developing countries, especially those that factored estimation (EFE) methods in accuracy and estimation capabilities. The
have oil and gas resources where big oil feasibility stages. The fixed capital cost best reference for reliable cost data is the
projects are being executed, the items (direct and indirect project completed projects of an organization.
equipment manufacturing industries expenses) can be attained by inserting Applying this data, using regression
(EMI) are growing and are now able to some factors on main equipment cost. methods and statistical tests, a final model
build major parts of the needed process If the equipment cost estimation has can be proposed.
equipment. some errors, these errors will propagate in As an example, look at the algorithm
There are legal encouragements to all other items of the capital cost and they used to derive a model as shown in figure
persuade the EMIs to build the needed will be collected in the last estimated 1. This algorithm shows that completed
process equipment. For example, about 50 capital cost. projects in RIPI were referred to in order
percent of equipment must be delivered Parametric cost estimation is one of to obtain the cost models for equipment
by local EMIs. This is emphasized in the most useful methods in feasibility that are listed in this article.
contracts in joint-venture projects. The study stages. Design activities are only 15
international partners can maneuver for percent completed in the feasibility study
Regression and Statistical Tests
the remaining 50 percent of the stage. However, by using a parametric Multivariable regression is a method
equipment. model it is possible to estimate capital cost
for analyzing the collective or individual
In any case, cost estimation in such with acceptable accuracy. The partnership of two or more independent
countries has their own unique conditions parametric model is a mathematical variables in variation of a dependent
and it is necessary to propose some new relationship between the equipment cost
variable.
models according to the social and and it's key parameters. Using SPSS software, some
economic conditions of these countries. As an example, the cost model for
relationships are proposed to estimate the
Use of the parametric model is one horizontal carbon steel vessels, used in
cost for some equipment that is given in
solution to this problem. Its use allows one Aspen software, is shown in equation 1 [2].
the achieved results section of this article.
to prepare a specific model for each The equipment cost is dependent variable
region/country for a more accurate cost Ln(Cb) = 8.411 - 0.16449 (LnWs) + in these relationships and independent
estimation. It can be used in all countries. 0.0433 (LnWs)2 variables are key parameters, such as
Different models can be obtained by
(equation 1) weight, (L/D) or power.
input ting local data. In other words, using To drive the models, one needs to
local data produces local models. suppose that a linear relationship exists
where:
between the cost of the equipment and its
869 ≤ Ws ≤ 415000 (kg) key parameters. See equation 2.

26 Cost Engineering Vol. 48/No. 5 MAY 2006

Figure 1— Cost Model Determining Algorithm

Figure 2— Comparison of Estimated Cost and Cost Data for BEU

Cost Engineering Vol. 48/No. 5 MAY 2006 27

 Ln(CE) = A + Bln(KP) + C(Ln(KP)2 For additional information, the reader can number of projects and in the type of
(equation 2) refer to statistical texts listed in the equipment, the defined models are in
reference section of this article. specified limited domains.
Where CE is equipment cost and KP is To increase these domains, additional
a key parameter. Then by using the linear Achieved Results cost data in broader ranges was needed.
regression method, one can find the Parametric models for some The achieved results in this article can be
coefficients A, B and C, which are equipment in this article were prepared used as initial data to develop more
tabulated in table 1. using a provided data bank, including the complete models.
Statistical tests such as F, t, and cost and some specifications of This methodology is general and was
confidence internal tests are done on the equipment. The data used is shown in used by others at Aspen software and at
proposed models, and finally the models appendix 1. other companies [2,3,4,5]. Everyone has
that passed these tests are selected. The As previously noted, the used data was their own models because of the
details of the tests are not included with gathered from projects archived in RIPI. dependency of models to the input cost
this article because of space limitations. Because of limitations, both in the

Table 1— Obtained Models for Some Equipments

Table 2— Confidence Intervals

28 Cost Engineering Vol. 48/No. 5 MAY 2006

Appendix Table 1 — Cost Data on Pressure Vessel (Carbon Steel)

Cost Engineering Vol. 48/No. 5 MAY 2006 29

 Appendix Table 2 — Cost Data on Pressure Vessel (Stainless Steel) Appendix Table 3 — Cost Data on Atmospheric Stproage Tanks
(Carbon Steel)

Appendix Table 4 — Cost Data on Seperation Tower (Carbon Steel) Appendix Table 5 — Cost Data on Seperation Tower (Carbon Steel)

30 Cost Engineering Vol. 48/No. 5 MAY 2006

data. So nobody can claim that someone else is copying his or ∧
her model. Where B is estimated coefficients, t is t-student from the
In table 1, the obtained models are shown, as well as the distribution table depends on degree of freedom and statistical
applicable ranges and absolute average deviation percentages significance. SE is the standard error for coefficients.
which are listed as (%AAD). The (%AAD) can be defined as In this article, the confidence interval was determined at 95
shown in equation 3. percent statistical significance for coefficients of six first models
and 90 percent statistical significance for the last model for a
compressor.
In table 2, the related confidence intervals and standard
errors are shown for coefficients in the proposed models. As in
(equation 3) table 2, none of the intervals straddle zero, so none of the
∧ coefficients are zero, so they are acceptable.
Where Yi is the estimated value Yi is the cost value from data The goodness of fit is explained by R square in regression.
R 2 = 1 is a perfect score. The R square for models is also shown
bank and n is the number of data.
2
For example, the cost of a BEU type heat exchanger (carbon in table 2. R = 0.99 is a very good score that shows the goodness
steel) with a weight of 10000 Kg can be calculated as follows. of fit.

CE = exp(-2.910474 + 1.01655 ln(10000)) = 631.15 MRls = Results Analysis

71253.6 $ Cost estimation accuracy by parametric models, in the
feasibility study stages, ranges between 20-50 percent (upper
(For exchange rate in 2004 use: 8900 Rails = 1 $)

In figure 2, the comparison between the estimated cost for

BEU heat exchanger and the cost data are shown.

Uncertainty Analysis
Another methodology, confidence interval, provides a
means of quantifying uncertainty. For each coefficient (Bi) is as
shown in equation 4.

∧
Bi = B ± t x SE
(equation 4)

Appendix Table 6 — Cost Data on BEU Type Heat Exchanges

(Cargbon Steel) Appendix Table 7 — Cost Data on Oil Injected Screw Compressor

Cost Engineering Vol. 48/No. 5 MAY 2006 31

 limit) and - 15 to -30 percent ( lower limit) ABOUT THE AUTHORS
[1]. These models can be accepted with Mohammad Reza Shabani and Reza WEBSITE LISTS
accuracy ranges between +/- 3 percent to Behradi Yekta are with the Feasibility Study WOMEN IN PROJECT
+/- 37 percent. The obtained models are Department of the Research Institute of CONTROLS
related to a specific year. Because of Petroleum Industry of Iran. Contact can be
inflation, they must be re-evaluated for use made by sending e-mail to:
SURVEY RESULTS
in the following years. shabanimr@ripi.ir. There is a whole lot of shaking going
on with the Women in Project Controls
Task Force.

I
n this article, the authors introduced Technical Articles - Each month, Cost
the parametric cost estimation model Engineering journal publishes one or more peer- This small but mighty group of women
as a fast shortcut method to obtain an reviewed technical articles. Unless noted otherwise, (and a few good men) is a force to be reck-
acceptable estimation during the these articles go through a blind peer review oned with. Since their formation in 2005,
evaluation prior to publication. Experts in the
preliminary design stages. This is a time subject area judge the technical accuracy of the
the Women in Project Controls Task Force
when there is little engineering information articles, advise the authors on the strengths and has set high goals. With this enthusiastic
about needed process equipment. weaknesses of their submissions, and what changes group of dedicated members, no one will be
Using collected cost data from RIPI's can be made to improve the article before surprised when they truly help to shake
publication.
project archive, cost models for some things up in AACE International.
equipment was obtained. The results
represented limited ranges and was
obtained by using regression methods.
Models were selected after passing
statistical tests.
The derived models are related to the
specific ranges (key parameters) because of
a limited data bank. If the authors had a
bigger data bank, the models would be
applicable in wider ranges. But the main
goal in this article was to introduce some
sample models to show how one can derive
such relationships.
Engineering companies who have
experience in doing big projects, enjoy
strong archives for process equipment cost
data. Such archives are the best source to
derive cost relations. Each company can
have its own models for proposal or tender
document preparation. ◆

REFERENCES
1. Dysert L.R. Sharpen Your Cost
Estimation Skills, Cost Engineering,
Vol. 45, No.6, June 2003.
2. Mulet A., A.B. Corripio, and L.B.
Evans, Estimate Costs of Pressure
Vessels Via Correlations, Chemical
Engineering, Oct. 5, 1981, page 145.
3. Turton R., R.C. Bailie, W.B. Whiting,
and J.A. Shaeiwitz. Analysis, Synthesis
and Design of Chemical Processes,
1998, ISBM 0-13-064792-6.
5. Corripio A.B., K.S. Chrisen, and L.B.
Evans, Estimate Costs of Centrifugal
Pumps and Electric Motors, Chemical
Engineering, Feb. 22, 1982, page 115.
6. Corripio A.B., K.S. Chrisen, and L.B.
Evans, Estimating Costs of Heat
Exchangers and Storage Tanks Via
Correlations, Chemical Engineering,
Jan. 22, 1982, page.125.

32 Cost Engineering Vol. 48/No. 5 MAY 2006

View publication stats