Capital Project Management Support

Tom Wolf                                                    (281)565-4038

Capital Cost Estimates - Understanding, Analyzing, and Evaluating

List Price:  $60.00

 8" x 10", 232 pages

Available at and other retailers

One of a series for the person requiring more than just an answer.
Refining and petrochemical project management personnel, most without estimating background, are required to understand, analyze, and determine validity of complex cost estimates prepared by others. 

This book is specifically designed and written as a desktop resource. In addition to presenting an organized approach to analyzing cost estimates, results of decades long historical cost data is provided, and  needed assistance to answer the crucial question of “is this estimate reasonable?” is presented.

List Price: $56.00

AVAILABLE at and other fine retailers.

Lang Factors, an Update5


T. E. Wolf, Sugar Land, Texas


(Please note this is copyrighted material, it is provided for your information.  Please feel free to use it, but please don't abuse the copyright.  Thank you.)     



Seventy years ago, Hans J. Lang, an engineer for Day & Zimmermann, Inc. in Philadelphia, PA, wrote a series of articles suggesting a quick, simplified estimating method to approximate capital projects’ final cost.  The simplified estimating method was not intended to replace more detailed estimates but to provide a quick, rule of thumb type method to replace the ‘guess’.  His method was to present factors which when multiplied by the cost of the plant equipment (including delivery cost) would yield the total installed plant cost.  

Using the cost data available to him, fourteen (14) plant estimates of various sizes and types, ranging from $100,000 to $15,000,000 (late 1940’s USD), he identified three process plant types.  Mr. Lang defined these as follows: “a distillation unit would be classified as a typical fluid process plant, a coal briquetting plant as a typical solid processing plant, and a solvent extraction plant complete with bean preparation and meal processing facilities as a typical solid-fluid processing plant.”1,2,3  Thus by estimating the plant equipment cost (abbreviated below, Equip) and choosing one of three factors (f), a rough approximation of the total installed cost (abbreviated, TOT) could be determined quickly.  The three factors for each type of plant being:

3.10  -  Solid process plants

3.63  -  Solid-fluid process plants 

4.74  -  Fluid process plants                        


In the decades that followed their introduction however, there have been many changes in capital projects’ execution, some very significant.  There are now governmental rules and regulations in-place, which just did not exist in the 1940s and 1950s.  There are materials and construction methods that are different.  There are digital process controls instead of pneumatic controls.  The computer is used in lieu of the slide rule and there is three dimensional computer design.  And that is just to mention a few.

Still, these factors are cited frequently and used within the industry; with the ensuing changes one must ask, ‘Do Lang Factors still present a realistic approximation of capital cost today?’

Data Collection and Results

Over my forty years in the industry, I collected essentially every bit of cost data with which I came in contact.  Ten years ago I decided to summarize that information in a spreadsheet database to identify metrics to assist in evaluating estimates.  All the raw data was first input, then certain ratios were calculated.  When I had finished, the spreadsheet consisted of over 200 columns of data and over 200 rows of projects.  However, not all of the projects’ data columns were completely populated; some projects were engineering only, some engineering and procurement only, etc., some were only fragmentary data sets, and I was interested in EPC (engineering, procurement, and construction) projects.

I settled on some thirty ratios along with other percentage benchmarks to evaluate.   To ensure consistency among the ratios and percentages, I focused on only those projects with essentially complete EPC data sets, that is, those containing essentially all of the 200 columns of data.  This reduced the project population to just under sixty projects, somewhat disappointing, but still a fairly decent sample.  This reduced set comprised about 55% refining projects and 45% petrochemical and chemical, with 25% being under $30 million and 60% under $100 million in total cost.  The projects had approximately 15% to 25% revamp component, some more, some less, but during the collection period none could be characterized as truly grassroots.4 

Subsequent to virtually completing the metrics exercise I realized one of the benchmarks summarized was the Lang Factor, TOT/Equip.  I had asked myself how my results compared with Lang’s results, fully expecting the decades of changes to result in different answers.  To my surprise and amazement that really was not the case, see Figures 1, 2, and 3 below.

As a brief refresher for those who haven’t thought about statistics in a while, Figure 1 is a plot of the 58 sample (n = no. in sample population) projects’ total installed cost (TOT) versus equipment cost (Equip).  The straight line shown is the least squares straight line curve fit, and its formula is shown just above the data plot along with its associated R2, which equals 0.809.  R2 is the Determination Coefficient of the least squares curve fit and is a measure of how good the curve represents the relationship between the variables; an R2 close to zero indicates ‘no correlation among the data’ while an R2 close to 1.000 would indicate ‘very good correlation of data’.

In this case, the R2 of 0.809 is not too bad, but for comparison other curves were calculated to determine the Best Fit.  For these data, a least squares power curve produced an R2 of 0.909, a very good correlation.  But for simplification, the straight line fit is the only one actually plotted.

Figure 2 is the histogram showing the relative frequency distribution of the 58 TOT/Equip ratios; in this case two maximums are indicated.  By reviewing the types of projects falling in the range between 7 and 10 the prevalent, common trait are processes requiring very large steel structures and vertical pipeways.  The mean, or average value, for the data set is 5.12 with 68% (+/- 1σ) of the projects’ ratios being between 3.38 and 6.86; sigma (σ) being the calculated standard deviation.

Figure 3 overlays Lang’s results with the Fluids (a distillation unit) Lang Factor of 4.74 being quite close to the calculated mean of 5.12, the Solid-Fluids factor within the minus one sigma limit, and the Solids factor falling just outside minus one sigma.  Which is not bad considering the Lang data is 70 years old, contains a very small sample population, and with all the project execution changes mentioned above.

Scroll down to bottom of webpage for:

Figure 1.4 

Figure 2.4 

Figure 3

Recently a senior colleague, Anthony Bryda of Proven Solution, a D.C area project benchmarking consulting service, asked how the larger database TOT/Equip ratios compared to the Lang Factors; a very good question considering the information was readily available. 

Going back to the over 200 projects’ data, eliminating only non-EPC projects the data set totaled 127 projects.  Of those, several were offsites only projects (projects with relatively little equipment compared to civil, structural, piping, etc. yielding large and varied TOT/Equip ratios).  Eliminating these outliers yielded a data set of 116 projects.  This larger set comprised 68% refining projects, 25% petrochemical and chemical, and 7% cogen, with 25% being under $30 million and 60% under $100 million in total cost.  And again, the projects had approximately 15% to 25% revamp component, some more, some less, and during the collection period none could be characterized as grassroots.  Plotting the same three curves, see Figures 4, 5, and 6, provided additional interesting information.

First the straight line curve fit produced a relatively disappointing R2 of 0.659, but the power curve fit’s R2 was very good at 0.875 indicating high data correlation (Figure 4).  The frequency graph with the calculated mean and plus and minus one sigma plotted are shown on Figure 5, mean equaling 4.76, with 68% of the calculated TOT/Equip factors falling within the range of 2.90 and 6.63.  The frequency plot indicates a third maximum on the lower end of the scale; reviewing the types of projects with these low ratios indicated these to be Cogen plants (that is, projects with very expensive third party equipment packages making up a significant part of the total installed cost).   However, the most surprising result was the comparison with the Lang data, Figure 6, that is, the 116 projects’ mean was for all practical purposes equal to the Lang Fluids factor.

Scroll down to bottom of webpage for:

Figure 4

Figure 5

Figure 6


The industry doesn’t design and build many “coal briquetting plant(s)” as typical solid processing plants, nor many “solvent extraction plant(s) complete with bean preparation and meal processing facilities” as typical solid-fluid processing plants.  However, Lang’s fluid process plant factor, 4.74, and the reduced data set’s average ratio, 5.12, are extremely close, within 8%. 


5.12 – 4.74 / 4.47 = 0.0802


And the much larger database’s mean ratio is virtually the same as Mr. Lang’s Fluids ratio.  His database of 14 estimates may not have included the breadth of plants contained within the cited databases, but certainly the results do show close correspondence.  If you need a quick, simplified estimating method and are using the Fluids Factor to obtain a rough order of magnitude estimate of a refinery or petrochemical unit, the results will be within the estimate accuracy expected, say ±50%.  Another practical use would be the Capital Cost comparison of one revamp scheme against another or one type unit against a similar, competing type unit; just don’t forget to include Owner Cost, Operating Cost, etc, if appropriate.


On the other hand, care should be taken when the unit under consideration differs significantly from a fluid processing unit, particularly as mentioned with units with significant steel structures and vertical pipeways, and with units containing extremely large third party supplied equipment such as cogen units.  Even so, based on the results of a larger database, over a longer period, the results would indicate within certain limits the use of Lang type factors are still valid and provide a realistic approximation of capital cost.


Project management is as much art as science.  The key is to continue to learn and never stop learning.

Literature Cited.

  1. Lang, Hans J.  “Engineering Approach to Preliminary Cost Estimates,” Chemical Engineering, September 1947, New York: Chemical Week Associates.
  2. Lang, Hans J.  “Cost Relationships in Preliminary Cost Estimation,” Chemical Engineering, October 1947, New York: Chemical Week Associates.
  3. Lang, Hans J.  “Simplified Approach to Preliminary Cost Estimates,” Chemical Engineering, June 1948, New York: Chemical Week Associates.
  4. Wolf, Thomas E.  Capital Cost Estimates – Understanding, Analyzing, and Evaluating, Chapter 6,, 2011.
  5. Wolf, T.E.  "An Update on Lang factors," Hydrocarbon Processing, August 2015, Volume 94, Number 8, Page 21, Houston: Gulf Publishing Company.


Tom Wolf has over 40 years petroleum and petrochemical experience, including 25 years in project management.  He holds a BS degree in Mechanical Engineering.


© 2013 by Thomas E. Wolf.  All rights reserved.  No part of the information contained in these webpages may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage or retrieval system without written permission of the copyright holder, except for the inclusion of brief quotations in a review.