Historical pipeline cost analysis
Upcoming SlideShare
Loading in...5

Historical pipeline cost analysis



This study aims to provide a reference for the pipeline construction ...

This study aims to provide a reference for the pipeline construction
cost, by analysing individual pipeline cost components with historical pipeline
cost data. Cost data of 412 pipelines recorded between 1992 and 2008 in the
Oil and Gas Journal are collected and adjusted to 2008 dollars with the
chemical engineering plant cost index (CEPCI). The distribution and share of
these 412 pipeline cost components are assessed based on pipeline diameter,
pipeline length, pipeline capacity, the year of completion, locations of
pipelines. The share of material and labour cost dominates the pipeline
construction cost, which is about 71% of the total cost. In addition, the learning
curve analysis is conducted to attain learning rate with respect to pipeline
material and labour costs for different groups. Results show that learning rate
and construction cost are varied by pipeline diameters, pipeline lengths,
locations of pipelines and other factors. This study also investigates the
causes of pipeline construction cost differences among different groups.



Total Views
Views on SlideShare
Embed Views



2 Embeds 8

http://www.slashdocs.com 6
https://twitter.com 2


Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
Post Comment
Edit your comment

    Historical pipeline cost analysis Historical pipeline cost analysis Document Transcript

    • 244 Int. J. Oil, Gas and Coal Technology, Vol. 4, No. 3, 2011Historical pipeline construction cost analysis Zhenhua Rui* Department of Mining and Geological Engineering, University of Alaska Fairbanks, Duckering Building 418, P.O. Box 750708, Fairbanks, Alaska 99775, USA Fax: +1-907-474-6635 E-mail: zhenhuarui@gmail.com *Corresponding author Paul A. Metz Department of Mining and Geological Engineering, University of Alaska Fairbanks, Duckering Building 313, P.O. Box 755800, Fairbanks, Alaska 99775, USA Fax: +1-907-474-6635 E-mail: pametz@alaska.edu Doug B. Reynolds School of Management, University of Alaska Fairbanks, P.O. Box 756080, Fairbanks, Alaska 99775, USA Fax: +1-907-474-5219 E-mail: dbreynolds@alaska.edu Gang Chen Department of Mining and Geological Engineering, University of Alaska Fairbanks, Duckering Building 315, P.O. Box 755880, Fairbanks, Alaska 99775, USA Fax: +1-907-474-6635 E-mail: gchen@alaska.edu Xiyu Zhou School of Management, University of Alaska Fairbanks, P.O. Box 756080, Fairbanks, Alaska 99775, USA Fax: +1-907-474-5219 E-mail: xzhou2@alaska.eduCopyright © 2011 Inderscience Enterprises Ltd.
    • Historical pipeline construction cost analysis 245Abstract: This study aims to provide a reference for the pipeline constructioncost, by analysing individual pipeline cost components with historical pipelinecost data. Cost data of 412 pipelines recorded between 1992 and 2008 in theOil and Gas Journal are collected and adjusted to 2008 dollars with thechemical engineering plant cost index (CEPCI). The distribution and share ofthese 412 pipeline cost components are assessed based on pipeline diameter,pipeline length, pipeline capacity, the year of completion, locations ofpipelines. The share of material and labour cost dominates the pipelineconstruction cost, which is about 71% of the total cost. In addition, the learningcurve analysis is conducted to attain learning rate with respect to pipelinematerial and labour costs for different groups. Results show that learning rateand construction cost are varied by pipeline diameters, pipeline lengths,locations of pipelines and other factors. This study also investigates thecauses of pipeline construction cost differences among different groups.[Received: October 13, 2010; Accepted: December 20, 2010]Keywords: pipeline cost; cost analysis; distribution; learning curve; costdifference.Reference to this paper should be made as follows: Rui, Z., Metz, P.A.,Reynolds, D., Chen, G. and Zhou, X. (2011) ‘Historical pipeline constructioncost analysis’, Int. J. Oil, Gas and Coal Technology, Vol. 4, No. 3,pp.244–263.Biographical notes: Zhenhua Rui is a PhD student in Energy EngineeringManagement at the University of Alaska Fairbanks. He also received hisMasters in Petroleum Engineering from the same university, in addition to aMasters in Geophysics from China University of Petroleum (Beijing). Hiscurrent research is the engineering economics of the Alaska in-state natural gaspipeline.Paul A. Metz is a Professor of Department of Mining and GeologicalEngineering at the University of Alaska Fairbanks. He received his PhD fromImperial College of Science Technology and Medicine, and MS in EconomicGeology and MBA from the University of Alaska. His research interestsinclude: market and transportation analysis of mineral resources; analysis oftransport systems; engineering geological mapping and site investigation; andmineral and energy resource evaluation.Doug B. Reynolds is a Professor of School of Management at the University ofAlaska Fairbanks. He received his PhD from the University of New Mexico.His research interests include oil production and energy economics. Some ofhis papers include an explanation of how one energy resource can subsidise thecost of an alternative energy resource and how an energy theory of value can beapproximated by defining energy grades for energy resources.Gang Chen is a Professor of Department of Mining and Geological Engineeringat the University of Alaska Fairbanks. He received his PhD in MiningEngineering from Virginia Polytechnic Institute and State University; Hereceived his MS in Mining Engineering from the Colorado School of Mines.His research interests include: rock mechanics in mining and civil engineering;mine ground engineering; frozen ground engineering and GIS application inmining industry.Xiyu Zhou is an Associate Professor of Finance at the School of Managementof the University of Alaska Fairbanks. He received his PhD of BusinessAdministration (Finance) from the University of North Carolina. He also
    • 246 Z. Rui et al. received MS in Economics from the University of Lausanne and MBA from China Europe International Business School respectively. His current research interests include: merger and acquisition, corporate governance and real estate mutual funds.1 IntroductionPipelining is an important and economical method to transport large quantities of oil andnatural gas in the petroleum industry. The first pipeline in the USA, two-inch in diameterand over 8 km long, was built in 1865 (Scheduble, 2002). By 2008, US had a total of793,285 km of pipelines, among which 244,620 km was for petroleum product and548,685 km was for natural gas (Central Intelligence Agency, 2008). Historical pipelinecost data have been analysed and used to estimate the construction costs for the differenttypes of pipeline cost by various researchers. Parker (2004) used natural gas transmissionpipeline costs to estimate hydrogen pipeline cost with the linear regression method.Zhao (2000) analysed the diffusion, costs and learning curve in the development ofinternational gas transmission lines. Heddle et al. (2003) derived a multiple linearregression model to estimate the CO2 pipeline construction cost. McCoy and Rubin(2008) developed multiple non-linear regression models to forecast CO2 pipeline cost.Pipeline cost was compared to LNG and GTL cost as supply options (Gandoolphe et al.,2003). Zhang et al. (2007) calculated share of material cost using pipeline cost between1993 and 2004 and indicated that share of material cost is constant for the same diameterpipelines. The Oil and Gas Journal annually analysed estimated and actual pipeline costand forecasts trends for the next year (PennWell Corporation, 1992, 2009). Variousstudies on pipeline cost have been conducted by different researchers in differentperspectives. The purpose of this paper is to conduct a comprehensive analysis on pipeline costsfrom 1992 to 2008 with various perspectives: the distribution of pipelines, shares ofpipeline cost components and learning-by-doing in pipeline construction. A number ofdata processing and statistical descriptions are applied to the historical data. Causes ofcost differences and learning rate differences are also investigated.2 Data sources and cost adjusting factors2.1 Data sourcesIn this study, the pipelines are selected on the basis of data availability. Pipeline cost dataare collected from Federal Energy Regulatory Commission filing by gas transmissioncompanies, which are published in the Oil and Gas Journal annual data book (PennWellCorporation, 1992, 2009). Due to limited offshore pipeline data, only onshore pipelinesare collected, and the pipeline cost in this paper does not include compressor station cost. The pipeline dataset includes year of completion, pipeline diameters, pipeline lengths,location of pipelines, and costs of pipeline cost components. Pipelines in the dataset weredistributed in all states in the USA (Alaska and Hawaii are excluded). The dataset alsocontains the cost information of 15 Canadian pipelines. The pipelines were completed
    • Historical pipeline construction cost analysis 247between 1992 and 2008. Unfortunately, the data did not show the construction period.Therefore, cost is defined as real, accounted costs determined at the time of completion.All pipeline construction component cost are reported in US dollar. The entire dataset has412 observations of onshore pipelines. The five pipeline cost components are: material,labour, miscellaneous, right of way (ROW) and total cost. Material cost is the cost of linepipe, pipeline coating and cathodic protection. Labour cost consists of the cost of pipelineconstruction labour. Miscellaneous cost is a composite of the costs of surveying,engineering, supervision, contingencies, telecommunications equipment, freight, taxes,allowances for funds used during construction, administration and overheads, andregulatory filing fees. ROW cost contains the cost of ROW and allowance for damages.The total cost is the sum of material cost, labour cost, miscellaneous cost and ROW cost(PennWell Corporation, 1992, 2009).Figure 1 Chemical engineering plant cost indexes between 1990 and 2008 (see online version for colours)2.2 Cost adjusting factorsAll costs are adjusted with the CEPCI – a widely used index for adjusting process plants’construction cost to 2008 dollars. The CEPCI has 11 sub-indexes and a compositeCEPCI, which is the weighted average of the 11 sub-indexes. The changes in costs overtime can be recorded by the index (Chemical Engineering, 2009). Indexes between 1990and 2008 are showed in Figure 1. Two-stages between 1990 and 2008 can be seen inFigure 1. The index increased slowly between 1990 and 2003, while the index increasedsharply after 2003, except for the construction labour and engineering supervision index.For example, the pipe index annual growth rate was 1.40% from 1990 to 2003, but it was5.49% from 2003 to 2008. The soaring index means pipeline construction costsexperienced high cost escalation after 2003. An indication of this is construction costfrequently overran budget during that period.
    • 248 Z. Rui et al. The annual average growth rate between 1990 and 2008 is shown in Table 1. Thestructure support index has the highest average annual growth rate of 4.09%. Engineeringsupervision index is almost constant with the lowest average annual growth rate of–0.04%. Pipe index average annual growth rate is 3.02% which is higher than the CEindex average annual growth rate of 2.54%. The index is a useful tool to adjust pipelinecost data. To make cost data comparable to each other at the same base, different pipelinecost components are adjusted by different indexes to 2008 dollars. Pipe index andconstruction labour index is used to adjust pipeline material and labour cost. CE index isapplied to pipeline miscellaneous and ROW costs.Table 1 Annual average growth rate of the chemical engineering plant cost index Index type Annual growth rate Index type Annual growth rate CE index 2.54% Heat exchange and tanks 3.30% Pipe 3.02% Process instruments 1.10% Construction labour 0.90% Equipment 3.07% Pump and compressor 2.94% Electrical equipment 2.31% Engineering supervision –0.04% Buildings 2.29% Process machinery 3.01% Structural supports 4.09%3 Data descriptive statisticsIn order to better understand pipeline cost, the cost data of pipelines are analysed andsummarised in terms of pipeline diameters, pipeline length, pipeline capacity, year ofcompletion and location.3.1 The distribution analysis of pipelines on year of completion, pipeline diameters and pipeline lengthsThe histogram of pipelines in different years is shown in Figure 2. 56 (13.6% of the total)constructed pipelines were reported in 2002, and only 6 (1.5% of the total) were reportedin 1998. Figure 3 shows the histogram of pipelines in different diameters. Eighteendifferent diameter pipelines were reported. The pipeline’s diameters range from fourinches to 48 inches, and value of all diameters is even number. There are 103 (25% of thetotal) 36-inch diameter pipelines, 63 (15.3% of the total) 30-inch diameter pipelines and62 (15.1% of the total) 24-inch diameter pipelines. These three types of diameterpipelines add up to 228 (55.3% of total). However, there are only two each of 10-inch,14-inch, 18-inch and 34-inch diameter pipelines. Further, there are only 24 (5.8% of thetotal) pipelines with diameters between 4 inches and 10 inches, while 218 (52.9% ofthe total) pipelines with diameter between 30-inch and 48-inch. It indicates thatsome specific diameter’ pipelines are constructed more than other diameters and morelarge-diameter pipelines are constructed than small-diameter pipelines in the lasttwo-decades. Figure 4 displays the histogram of pipelines grouped in pipeline lengths.The distribution of pipeline length is right-skewed. The pipeline length ranges from
    • Historical pipeline construction cost analysis 2490.01 mile to 713 miles. There are 258 (62.6% of the total) pipelines in the 0 to 10 milegroup, and 65 pipelines in the 10 to 20 mile group, but only 30 (7.3% of the total) ofpipelines are longer than 60 miles. It indicates that majority of the reported pipelines areshort pipelines.Figure 2 Histogram of pipelines between 1992 and 2008 (see online version for colours)Figure 3 Histogram of pipelines in diameters (see online version for colours)
    • 250 Z. Rui et al.Figure 4 Histogram of pipelines grouped in lengths (see online version for colours)3.2 The distribution of pipelines regarding pipeline capacity (pipeline volume)Pipeline capacity is calculated with the following formula (Zhao, 2000) V = S∗L 2 ⎛D⎞where S = π ⎜ ⎟ ; V is the pipeline capacity (ft3); S is the pipeline cross-sectional area ⎝2⎠ 2(ft ); L is the pipeline length (ft); D is the pipeline diameter (ft).Figure 5 Histogram of pipeline capacity (see online version for colours)The histogram of pipeline capacity is shown in Figure 5. The distribution of pipelinecapacity is right-skewed. Average pipeline capacity is 86,511,969 ft3 with standarddeviation (SD) of 15,840,088 ft3. The pipeline capacity ranges from 13,270 ft3 to
    • Historical pipeline construction cost analysis 2515,215,691,727 ft3. 58.29% of pipelines’ capacity is less than 30,000,000 ft3, and only3.64% of pipelines’ capacity is larger than 400,000,000 ft3.Figure 6 US natural gas pipeline network region map (see online version for colours)Note: Alaska and Hawaii are not included. Source: EIA (2010)3.3 The distribution analysis of pipeline locationsThe location information for US pipelines is provided in a state format. A total of48 states were referred to, excepting Alaska and Hawaii. Energy InformationAdministration (EIA) breaks down the USA natural gas pipelines network into sixregions: Northeast, Southeast, Midwest, Southwest, Central and Western. The stategrouping is defined based on ten federal regions of the USA Bureau of Labor Statistics(EIA, 2010). These regional definitions are used to analyse geographic difference. Themap of regional definitions is shown in Figure 6. In this paper, US pipeline data aresummarised according to these six-regions (McCoy and Rubin, 2008). Based on theregional definition, region distribution of pipelines are summarised and shown in Table 2.157 (40% of US pipelines) pipelines are located in the Northeast region. Furthermore,46% of these Northeast region pipelines are located in the State of Pennsylvania. Thirty(7.5% of US pipeline) pipelines are located in the Southwest region. The number ofpipelines in other regions is between 48 and 55. In addition, there are 15 Canadianpipelines, but the data did not show a specific province in Canada.Table 2 Number of pipelines in regions and states Region Number of pipelines State* Number of pipelines Central 52 Colorado 15 Northeast 157 Pennsylvania 72.5 Southeast 55 Alabama 20.5 Midwest 55 Ohio 18.5 Southwest 30 Louisiana 9.5 Western 48 Washington 11.5 Canada 15Note: *State has the highest number of pipelines in its region.
    • 252 Z. Rui et al.3.4 The distribution analysis of pipeline individual cost componentsThe histogram of cost of pipeline cost components are shown in Figure 7 to Figure 11.These figures illustrate that all distributions of pipeline cost components areright-skewed. The majority of cost distribution is concentrated on the left of the figure,indicating more cases of low cost and few relative high cost. Similar trend exists in thehistogram of length group (Figure 4) and the histogram of pipeline capacity group(Figure 5). It seems that pipeline length or pipeline capacity may play a significant role indetermining pipeline construction costs.Figure 7 Histogram of material cost (see online version for colours)Figure 8 Histogram of labour cost (see online version for colours)
    • Historical pipeline construction cost analysis 253Figure 9 Histogram of miscellaneous cost (see online version for colours)Figure 10 Histogram of ROW cost (see online version for colours)3.5 The trend of pipeline capacity over timeThe size of pipeline capacity was analysed in the above section. This section investigatesthe trend of annual pipeline capacity. The annual pipeline volume constructed is shown inFigure 12. There are three major peak years in term of pipeline volume constructed:2000, 2003 and 2008. The year 1998 has the lowest volume of pipeline constructed.Before 1998, the constructed pipeline volume changed slightly. After that, however, thevolume increased sharply from 1,700,168 ft3 to 31,773,396 ft3 between 1998 and 2003.Then there was a dramatic fall to 7,917,393 ft3 from 2003 to 2006. 2006 to 2008 saw thebiggest increase, from 7,917,393 ft3 to 48,262,884 ft3. The annual constructed pipelinevolume exhibited a cyclic characteristic, with a general trend of growing.
    • 254 Z. Rui et al.Figure 11 Histogram of total cost (see online version for colours)Figure 12 Annual constructed pipeline volumes (see online version for colours)3.6 The trend of average unit cost over timeThe unit component costs of pipeline are an important parameter for estimating pipelinecosts. In this section, the trend of unit component costs of pipeline over time is analysed.Unit cost is calculated by dividing cost by volume. For all 412 pipelines, the average unitcost in material, labour, miscellaneous, ROW and total cost were $18/ft3, $24/ft3, $14/ft3,$5/ft3 and $61/ft3 respectively. Figure 13 shows the annual average unit cost of pipelinecost components. Unit costs of labour, miscellaneous and total cost are in a similarpattern, which fluctuates widely. But material and ROW unit costs changed moregradually, and were more stable compared to the other cost components. All costcomponents changed slowly before 1998, similar to the change in constructed pipelinevolume. After 1998, the change was dramatic. The years of 1999, 2002 and 2007 were
    • Historical pipeline construction cost analysis 255the three-major peak years in unit total cost. The highest unit total cost was reached$109/ft3 in 1999, which was almost three-times as high as the bottom point of $39/ft3 in1998. By contrasting Figure 12 and Figure 13, one can find that these three-peak years inunit total cost occurred all one year before the peak years in constructed volume. Thisevidence indicates that expectation of increased pipeline construction induced an increasein the current unit cost. Material suppliers would raise prices with expectation for moredemand the next year. The higher expected demand in labour would cause labourshortage, and the competitive salary and benefits had to be paid in order to hire or keepmore skilled labourers. Miscellaneous cost also increased due to more demand. All thesefactors together resulted in high cost one year before the peak year in constructed pipelinevolume.Figure 13 Annual average unit cost of pipeline cost components (see online version for colours)4 The share of cost components for different pipeline groupsAs mentioned above, the average pipeline unit cost of total cost is $ 61/ft3, but this costincludes material cost, labour cost, miscellaneous cost and ROW cost. In order to betterunderstand the influence of individual cost component for different pipeline groups, theshare of each component cost of pipeline diameters, pipeline lengths and location ofpipelines are analysed in this section. Results are shown in Table 3. For all onshorepipelines, the labour cost has the highest share of 40% of total cost. Material cost has thesecond highest share of 31% of the total cost. The sum of material and labour cost cansometime reach up to 80% of the total cost. Miscellaneous cost was about 23% of thetotal cost. ROW cost accounts for an average of 7% of the total cost. Generally, labourand material costs dominate the pipeline cost, and the labour cost is still the highest costfor all groups except for the Central region group. Table 3 shows that the share of cost components varied under different situations. Interm of pipeline diameters, the share of material cost increased from 19% forsmall-diameter pipelines to 34% for large-diameter pipelines, while the share of othercost components decreased. It indicates that share of cost components related to pipeline
    • 256 Z. Rui et al.size, which agrees with Zhao’s (2000) finding. It also indicates that the share of materialcost increased when pipeline diameter increased. In term of pipeline lengths, the share ofmaterial cost rose from 28% for short pipelines to 35% for long pipelines, with share ofthe other cost components decreasing except ROW, which was constant at 7% regardlessof the total pipeline length. Therefore, the share of material cost increased when pipelinediameter and length increased, but the labour cost maintained as the no. 1 cost componentfor all diameters and lengths, averaging 40% of total cost. Furthermore, the shares of costcomponents were different for different regions. The material cost in the Central regionmade up around 41% of the total cost, while it was only 24% of the total cost in theNortheast and Southeast regions. The share of labour cost is between 34% and 48% indifferent regions. Miscellaneous cost was often a small part of the total cost, but the shareof miscellaneous cost in the Southeast region reached to 30% of the total cost, evenhigher than share of material cost. The share of ROW cost of US pipelines ranged from4% to 12% of total cost, while the share of ROW cost in Canada share was only 1% oftotal cost. The lower share of ROW cost for Canada pipelines allows us to conclude thatCanada has less ROW issues than the US does. The share of material cost and labour costwere approximately the same for Canadian pipelines, about 40%. The results agree withthe conclusion that the shares of labour and material costs varied by countries (Zhao,2000). It also support that the shares of cost components vary in different regions of USlocal regions or countries with no pipeline producing capacity may have high materialcost, and the pipeline cost can be reduced by developing technology to produce pipelinematerials (Zhao, 2000). The high share of labour cost was possibly caused by local highcost of living. For example, the Northeast region had the highest labour cost compared tothe other regions. Hence, studies on share of cost components will provide usefulinformation for pipeline companies to estimate pipeline cost and reduce the total cost bysome actions, such as improving pipeline production capacity.Table 3 The shares of pipeline cost components for different pipeline groups Material Labour Miscellaneous ROW All data Average 31% 40% 23% 7% Diameter 4–20 inches 19% 43% 28% 9% 22–30 inches 28% 38% 26% 8% 34–48 inches 34% 40% 20% 6% Length 0–60 miles 28% 41% 24% 7% 60–160 miles 31% 39% 23% 7% 160–713 miles 35% 39% 20% 7% Region Central 41% 38% 18% 4% Northeast 24% 43% 27% 6% Southeast 24% 34% 30% 12% Midwest 26% 37% 27% 11% Southwest 31% 41% 23% 5% Western 32% 48% 13% 8% Canada 39% 40% 19% 1%
    • Historical pipeline construction cost analysis 2575 Learning curve (learning-by-doing) in pipeline construction5.1 Introduction to learning curveThe productivity of technology and labour normally increases as workers engage inrepetitive tasks. The unit costs typically decline with cumulative production. The learningcurve is derived from historical observation to measure learning by doing, and it ishelpful for cost estimators and analysts. The learning curve theory is based on theseassumptions:1 the unit cost required to perform a task decreases as the task is repeated2 the unit cost reduces at a decreasing rate3 the rate of improvement has sufficient consistency to allow its use as a prediction tool (Federal Aviation Administration, 2005).The consistence in improvement is expressed as the percentage reduction in cost withdoubled quantities of product. The constant percentage is called the learning rate. Forexample, a 20% learning rate implies the cost is reduced to 80% of its previous level aftera doubling of cumulative capacity. The learning curve is normally exhibited in power function form and linear functionform. The power function form is shown below (Federal Aviation Administration, 2005): Yx = T1 i X bwhere Yx is the average cost of the first X units; T1 is the theoretical cost of the firstproduction unit; X is the sequential number of the last unit in the quantity for which theaverage to be computed; b is a constant reflecting the rate costs decrease from unit tounit; 2b and 1–2b are called progress ratio and learning rate respectively (Federal AviationAdministration, 2005; International Energy Agency, 2000). Learning curve function is normally expressed in log-log paper as a string line.Straight lines are more easily for analysts to extend beyond the range of data (FederalAviation Administration, 2005). Take the logarithms of the both sides to get a straightline equation, Y = bX + Cwhere Y = log Yx , X = log X , C = log (T↓ 1) . The learning curve effect is a complicated process. Some of major reasons forlearning-by-doing effect are: intensive use of skilled labour, a high degree of capital,research and development intensity, fast market growth and interaction between supplyand demand (Wilkinson, 2005). In addition, accumulated learning has a start-up and asteady period. The cost reduction is significant in the start-up period and modest in thesteady period (Grubler, 1998). It is the same for technology development. There aresignificant cost improvements during R&D phase followed by more modest improvementafter commercialisation. The longer technology has been in operation, the smaller thecost decreases (Zhao, 2000). It is possible that no further improvement in cost reductionoccurs for existing and mature technology (Grubler, 1998). The commercialisation oftechnology in the oil and gas market is costly and time intensive with an average 16 yearsfrom concepts to widespread commercial adoption (National Petroleum Council, 2007).
    • 258 Z. Rui et al.The range of progress ratio for technology is between 65% and 95%, and between 70%and 90% for energy technology (Christiansson, 1995).5.2 Selecting pipeline cost data for calculating learning rateThe cost data for learning curve analysis has to be recurring cost, because non-recurringcosts will not experience the learning effect (Federal Aviation Administration, 2005).Zhao (2000) calculated the learning curve of the total cost without considering thisrequirement and her results may be less accurate. The miscellaneous and ROW costs aswell as the total cost are not qualified for the learning curve analysis due to inclusion ofnon-recurring costs. The learning curve analysis is, therefore, only conducted for materialand labour costs. The pipeline data provide the cost data from 1992 to 2008. However,the 1999 data are considered an outlier due to extremely high cost. Hence, the 1999 datais not used for learning curve analysis. The learning curve of the material and labour costof pipelines constructed from 1992 to 2008 is presented in Figure 14. Figure 14 showsthat there was an attractive cost reduction in unit cost before 100 million ft3. After100 million ft3, the unit cost did not show cost reduction even increases. It indicates therewas not cost reduction after 100 million ft3, which was considered as a more matureperiod. In the standard experience curve theory, it is assume that learning rates do notchange over time, but the technology or labour learning are going to a more maturephase. However, the learning curve analysis does not always strictly agree with thisassumption (Schaeffer and de Moor, 2004). In order to better fit the learning curve, thelearning rate is calculated with data from 1992 to 2000. The learning curves of thematerial and labour costs from 1992 to 2000 are shown in Figure 15, and the learningcurve equations are expressed below: Material cost equation : Y = 103.2 X −0.09 or Y = −0.09 X + 2.01 R 2 = 0.93 Labour cost equation : Y = 722.8 x −0.19 or Y = −0.19 X + 2.86 R 2 = 0.91Figure 14 Learning curves of material and labour costs between 1992 and 2008 (see online version for the colours)
    • Historical pipeline construction cost analysis 259Figure 15 Learning curves of material and labour costs between 1992 and 2000 (see online version for colours)Both R2 (coefficient of determination) are higher than 0.9, which indicates a very goodfit. The learning rates of labour and material cost are 12.4% and 6.1%, respectively. Thatis, doubling the construction of pipeline volume, the labour cost and material cost will bereduced by 12.4% and 6.1% respectively. But it can be noted that the cost reductionbecomes smaller with increasing volume, same as the finding of Zhao (2000).5.3 Learning rate for different pipeline groupsThe learning rates for different pipeline diameters, lengths and locations are calculatedand shown in Table 4. In general, the learning rate of material cost was lower than thelearning rate of labour cost in all subgroups except in the Southeast region. For allsubgroups, the range of the learning rate of material cost was between 1.40% and14.60%, and the range of the learning rate of labour cost was between 6.10% and23.00%. For different diameters, learning rates of labour cost is between 13.60% and14.20%, but learning rates of material cost ranges from 4.10% to 8.00%. For differentpipeline lengths, the learning rate of labour cost showed a significant difference about6.70%. As expected, the results indicate that longer pipelines can achieve a higherlearning rate in labour cost. However, the results also show that longer pipelines have adisadvantage on learning rate of material cost, 6.10% for zero to 20 miles pipeline and4.80% for 20 to 713 miles pipelines. In terms of regions, the results show that thelearning rate varied widely in different regions. The Northeast region had the lowestlearning rate of material and labour cost. A plausible explanation for this finding wouldbe that a large amount of pipeline built in the Northeast region makes Northeast regionreach a more mature stage earlier and faster than other regions. Pipelines in the Southeastand Western region showed higher learning rate of material and labour costs than otherregions. In summary, the above analysis reveal that learning rates varied by differentpipeline diameters, pipeline lengths and the location of pipelines at different degree.
    • 260 Z. Rui et al.Table 4 Learning rates of material and labour cost in different groups Material Labour All data Average 6.10% 12.40% Diameter 4–20 inches 7.40% 13.60% 22–30 inches 4.10% 13.60% 34–48 inches 8.00% 14.20% Length 0–20 miles 6.10% 8.70% 20–713 miles 4.80% 15.40% Region Northeast 1.40% 6.10% Southeast 14.60% 11.80% Midwest 4.80% 8.00% Western 7.40% 23.00%6 Factors causing pipeline construction cost differenceSpecial geographic and surrounding environmental conditions may induce morecomplexities in pipeline construction, and have various degrees of impact on theconstruction costs. In some cold regions, pipelines need to be insulated or built aboveground when they pass the permafrost area resulting in additional construction cost. Inpopulated regions, thicker pipeline wall has to be selected to mitigate societal andenvironmental risk concern (Sanderson et al., 1999). Although some argued thatpopulation density has less impact on cost than type of pipelines (Zhao, 2000). Roads,highways, rivers or channel crossings and marshy or rocky terrains, all these factors,strongly affect pipeline unit cost (PennWell Corporation, 1992, 2009). For example, theperformance of all trenching units is largely dependent on soil type and amount of debrisencountered. Heavy, clay soils or soils littered with rock or construction debris willrequire more horse power and larger machines to lay pipes (Houx, 2010). There are alsomany other geographic and environmental factors influencing pipeline cost and costreduction which need to be identified in specific circumstances. Someone may argue gas price or oil price possible influences pipeline constructioncost. In order to discover relation between gas price or oil price and pipeline constructioncost, the correlation between gas price or oil price and lag zero year to four-years averageunit costs from 1992 to 2008 are analysed and shown in Table 5 and Table 6,respectively. The values of all correlation coefficients in Table 5 are between –0.41 to0.3. It indicates that linear relationship between gas price and pipeline construction cost isvery weak. The values of coefficients in Table 6 indicate the same conclusion for oilprice and pipeline construction cost. Some non-linear transformations (power,exponential, reciprocal, square root) are also used to deal with oil/gas price and unit costdata. However, these typical non-linear relationships between gas price or oil price andunit cost are also very low. Therefore, there is no sufficient evidence that gas or oil pricechange causes pipeline construction cost change with available data.
    • Historical pipeline construction cost analysis 261Table 5 Correlation coefficient between gas price and average unit cost Material Labour Miscellaneous ROW Total Lag 0 year –0.01 –0.14 –0.28 –0.23 –0.20 Lag 1 year 0.17 0.02 –0.12 –0.19 –0.03 Lag 2 years 0.29 0.23 0.10 –0.05 0.18 Lag 3 years 0.26 0.15 –0.06 –0.41 –0.19Table 6 Correlation coefficient between oil price and average unit cost Material Labour Miscellaneous ROW Total Lag 0 year 0.24 0.10 –0.08 –0.21 0.03 Lag 1 year 0.34 0.16 –0.11 –0.27 0.05 Lag 2 years 0.49 0.34 0.06 –0.17 0.24 Lag 3 years 0.33 0.25 –0.03 –0.51 –0.28From technology perspective, pipeline transportation has not seen a major technologicalbreakthrough over the last few decades (Roland, 1998). However, gradual cost reductionis possible by optimising project design and construction, inspection activities, laying andwelding methods, steel quality and weigh and the period of construction and increasingcompetition among inspection service companies (Gandoolphe et al., 2003). The costreduction through improved technology for laying, inspection and welding can becounterbalanced by other factors, such as, high strength and thick pipe used to reducepotential risk (Zhao, 2000). Compared to other technologies, such as LNG process, thecost reduction in pipeline transportation is smaller due to less complicated process.However, offshore pipeline technology has made possible deep-water projects andcontributed to lower unit cost. S-lay method and J-lay methods were used to installmarine pipeline (Gandoolphe et al., 2003). The average learning rate of offshore pipelinebetween 1985 and 1998 was 24% (Zhao, 2000). For example, the pipeline installing costin Norwegian part of North Sea in 1998 was 44% lower than the corresponding cost forStatpipe in 1985 (Roland, 1998). The history of onshore pipeline was 100 years earlierthan the offshore pipeline in the USA. Therefore, onshore pipeline construction is in amore mature stage, and has less learning effect (Zhao, 2000). US Department of Energy(DOE, 2007) has funded many new projects to develop advanced technologies, such asrobotic platforms, pipeline diameter reductions and expansions and variables types ofpipeline bends. These technologies may be progressively applied to onshore pipeline tocreate significant cost reduction. Besides geographic, environment and technological factors, potential market demandalso influence learning rate of pipelines. As mentioned in unit cost section, potentialdemand will cause increasing current unit cost of pipelines. Therefore, expected demandof pipelines will indirectly influence learning rate of pipelines. In order to fully explain pipeline construction cost difference, there are more factorsthat need to be investigated. Due to limited information, the discussions in this sectionfocus on a few identified factors affecting pipeline construction cost difference:development stage of technology, geographic and environmental condition as well asmarket situation.
    • 262 Z. Rui et al.7 Concluding summaryBased on historical data collected from Oil and Gas Journal, the distribution of pipelinesin term of year of completion, pipeline diameters, pipeline lengths, pipeline capacity andlocation of pipelines are analysed. Among the data examined, 78.3% of pipelines wereless than 20 miles, 52.9% of them had a diameter of 30 inches or larger and 58% ofpipelines’ capacities was less than 30,000,000 ft3. The pipelines were located across theUSA, but about 40% of them were located in the Northeast region. The distributions ofcost of pipeline cost components were all right-skewed (Figure 7 to Figure 11), and therange of cost of pipeline cost components was very large. The trend of annual constructedpipeline volume and annual average unit cost indicates that expecting of increasedpipeline demand will causes increasing currently unit cost. Shares of cost components aredifferent for various pipeline diameters, pipeline lengths and locations of pipelines. Thematerial and labour cost are major component of pipeline construction (Table 3). Resultsof learning curve analysis show that learning rate also varied by pipeline diameters,pipeline lengths, locations of pipelines (Table 4). Furthermore, development stage ofpipeline technology, site characteristics and market condition are identified as the factorsinfluencing pipeline construction cost difference.ReferencesCentral Intelligence Agency (2008) The World Factbook, available at https://www.cia.gov/library/publications/the-world-factbook (accessed on 9 January 2010).Chemical Engineering (2009) Chemical Engineering’s Plant Cost Index, available at http://www.che.com/pci (accessed on 4 January 2010).Christiansson, L. (1995) ‘Diffusion and learning curves of renewable energy technologies’, pp.95–126, Working paper, International Institute for Applied System Analysis, Austria.DOE (2007) ‘Transmission, distribution and storage’, available at http://www.fe.doe.gov/programs/oilgas/delivery/index.html (accessed on 3 January 2007).Energy Information Administration (EIA) (2010) ‘Natural gas transportation maps’, available at http://www.eia.doe.go (accessed on 9 January 2010).Federal Aviation Administration (2005) FAA Pricing Handbook, available at http://www.fast.faa.gov/pricing/index.htm (accessed on 9 January 2010).Gandoolphe, S.C., Appert, O. and Dickel, R. (2003) ‘The challenges of future cost reductions for new supply options (pipeline, LNG, GTL)’, Paper Presented at the 22nd World Gas Conference, 1–5 June, Tokyo, Japan.Grubler, A. (1998) Technology and Global Change, Cambridge University Press.Heddle, G., Herzog, H. and Klett, M. (2003) The Economics of CO2 Storage, MIT LFEE 2003-003 RP, Laboratory for Energy and Environment, Massachusetts Institute of Technology.Houx, J. (2010) ‘Trench warfare’, Grounds Maintenance, available at http://www.grounds-mag.com/mag/grounds_maintenance_trench_warfare (accessed on 9 January 2010).International Energy Agency (2000) Experience Curves for Energy Technology Policy, Paris, France.McCoy, S.T. and Rubin, E.S. (2008) ‘An engineering-economic model of pipeline transport of CO2 with application to carbon capture and storage’, International Journal of Greenhouse Gas Control, Vol. 2, No. 2, pp.219–229.
    • Historical pipeline construction cost analysis 263National Petroleum Council (2007) ‘Oil and gas technology development’, available at http://www.npc.org (accessed on 9 January 2010).Parker, N.C. (2004) Using Natural Gas Transmission Pipeline Costs to Estimate Hydrogen Pipeline Costs, Research Report UCD-ITS-RR-04-35, Institute of Transportation Studies, University of California, Davis.PennWell Corporation (1992, 2009) Oil & Gas Journal Databook, Tulsa, Oklahoma.Roland, K. (1998) ‘Technology will continue to profoundly affect energy industry’, Oil & Gas Journal, Vol. 96, No. 13, pp.69–74.Sanderson, N., Ohm, R. and Jacobs, M. (1999) ‘Study of X-100 line pipe costs points to potential savings’, Oil & Gas Journal, Vol. 11, pp.54–56.Schaeffer, G.J. and de Moor, H.H.C. (2004) ‘Learning in PV trends and future prospects’, Paper Presented at the 19th European PV Solar Energy Conference and Exhibition, 7–11 June, Paris, France.Scheduble, K.L. (2002) ‘Trenchless technologies in pipeline construction’, Journal for Piping, Engineering, Practice, Special edition, pp.1–17.Wilkinson, N. (2005) Managerial Economics: A Problem-Solving Approach, Cambridge University Press.Zhang, C, Yan, D., and Zhang, Y. (2007) ‘The design procedure model study of gas transmission pipeline’, Oil & Gas Storage and Transportation, Vol. 26, No. 7, pp.18–20.Zhao, J. (2000) ‘Diffusion, costs and learning in the development of international gas transmission lines’, Working paper IR-00-054, International Institute for Applied Systems Analysis, Austria.