Fuel Usage Factors in Highway and Bridge Construction (2013)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

18 Chapter 2 of this report details the initial research and the research approach undertaken by the study team. Specifically, this chapter provides an overview of the initial efforts conducted using the three study methodologies: statistical analysis, engineering estimation analysis, and DOT and contractor surveys of fuel usage. The initial efforts under these methodologies were then evaluated and modified to best ensure that the later data collection phase would be best suited to accurately measure highway construction fuel use. As an initial step, the study team surveyed all 50 state DOTs and a select number of construction contractors. These responses informed the study team of the current state of fuel usage factor and price adjustment clause implementation, perceptions of high fuel use activities, the analytical needs of the targeted end users, and the features and information to include in the full contractor fuel usage survey. The first two sections of this chapter describe the DOT and contractor surveys, respectively. The third section provides a brief overview of the initial engineering estimation effort. The fourth section briefly describes the initial statistical evaluation of fuel intensity. The fifth section provides an overview of the initial test efforts undertaken for each of the three research methodologies. This section includes a discussion of the work performed for each of the three potential methodologies, the results, any unanticipated occurrences, and any modifications made to the three approaches in preparation for the full data collection phase of the project. 2.1 DOT Needs and Perceptions The study team conducted a nationwide survey of state DOTs. The purpose of the survey was to ascertain the current implementation of fuel usage factors, the states’ satisfaction with their current programs, and their perceptions on how to upgrade them. This report section presents the team’s findings. The first subsection describes the survey methodology. Subsections 2.1.2 through 2.1.8 report on individual sections of the survey. Subsection 2.1.9 summarizes the DOT survey and offers conclusions. 2.1.1 Survey Methodology and Response The study team provided the NCHRP project panel with a draft copy of the state DOT survey on August 24, 2010. On September 1, 2010, invitations were sent out to officials from all 50 state DOTs to participate in the online version of the survey available on SurveyMonkey. The survey requested a response by September 10, 2010, and 28 state DOTs responded by the initial deadline. An additional request was sent out to non-responding DOT officials on September 10, 2010. Additional requests were sent out to non-responding DOTs. The study team received the 50th state response on December 10, 2010. This chapter incorporates survey results from all 50 state DOTs. C h a p t e r 2 Initial Research and Research Approach

Initial research and research approach 19 2.1.2 Extent of Fuel Factor Implementation The first survey question asked the respondents whether or not their state uses fuel usage factors to determine price adjustments for fuel. Among the 50 state DOTs, 38 states employ fuel usage factors while 12 do not. Exhibit 2-1 presents these results. This question shows that most states utilize some form of fuel factor. The scope of their use is not uniform, however. Nebraska only adjusts for fuel on grading projects. Alabama does not use the term “fuel factor” but adjusts for fuel price fluctuations in the case of hot mix asphalt (HMA) and other bituminous mixes. It also employs a bid item for lump sum construction fuel. Several states amplified their response regarding the future of their programs in response to other survey questions. Oklahoma is in the process of composing a provision for fuel price adjustment but plans have not yet been finalized. California currently does not employ fuel factors but is working with the contracting industry to study the feasibility of adding them. Michigan does not utilize fuel factors and currently has no plans to adapt them, citing a lack of interest from local contractors. 2.1.3 Current Program This section of the survey asked respondents to comment on the origin of their fuel usage factors as well as to describe their current systems. The first question in this section provided respondents with the opportunity to select the sources for their fuel use factors. The most popular answer choices among state DOTs were Attachment 1 in the original FHWA Technical Advisory T5080.3 and contractor/industry data, which 16 and 15 of the 37 states selected, respectively. Internal DOT data followed with 14 selections. Of the 37 affirmative respondents, 18 states selected a single option. Exhibit 2-2 displays a tabulation of the sources of DOT fuel usage factors. In addition, 18 states further explained the source of their factors or entered an option not included above. Note that collaboration with the contracting industry was mentioned six times, while four states commented that their factors were quite old and/or had not been updated for some time. Exhibit 2-3 displays additional comments from state DOTs regarding the source(s) of their factors. Exhibit 2-1. Implementation of fuel usage factors. Yes 38 76.0% No 12 24.0% Does your DOT use Fuel Use Factors in DOT contracts to determine price adjustments for fuel price changes? n = 50

20 Fuel Usage Factors in highway and Bridge Construction In a related question, the survey queried respondents as to when the states’ factors were last updated. The survey provided a set of responses ranging from within the past year to over 10 years ago. The survey also included an option for unknown. Exhibit 2-4 provides a summary of the 37 responses provided to this question. The responses indicate that the fuel factors used by state DOTs are often quite dated. Only 16 of the 37 respondents identified their factors as having been updated within the last 5 years. In contrast, 11 respondents replied that their factors had not been updated within the last 10 years and another 8 did not know when they were last updated. Several respondents indicated that their factors were very old and that they did not know when, or if, they had ever been updated. Exhibit 2-2. Sources of fuel factors. What are the sources of your current Fuel Use Factors (check all that apply)? Answer Options ResponsePercent FHWA Technical Advisory T5080.3 (Original 1980 Data in Attachment 1) 42.1% 16 FHWA Technical Advisory T5080.3 (Other State Data in Attachment 2) 13.5% 5 Internal DOT Data 37.8% 14 Contractor-Supplied Data 13.5% 5 Contractor Organization/Industry Data 40.5% 15 Other Data Collected or Developed by Your State 35.1% 13 n=37 Total Selected=68 Response Count Exhibit 2-3. Additional comments on sources of fuel factors. State Response Alabama Fuel factor for HMA production was developed jointly with industry reps Connecticut Oil Price Information Service (OPIS) Georgia The state construction engineer (now retired) developed the factors we started to use in 2007 Idaho Factors were developed by the department in consultation with industry Illinois Meetings with industry and equipment manufacturers Kentucky Unsure on what is used. This was done prior to my appointment to this position. There were no records concerning the item. Louisiana Factors developed in the early 80s by materials and testing lab Massachusetts Our fuel usage factors were developed by the Highway Research Board in Circular 158, dated July 1974 Minnesota OPIS daily rax fax New Jersey The basic factors were unchanged – we added newer items using comparable factors as part of an update of NJDOT’s standard specification New York Unknown, very old - historical data has been lost Ohio ODOT used fuel usage factor information from other states. Productivity Rates and Equipment Watch operating cost information was also considered. Oregon Adjusted to Oregon. I don't know the process used to make the adjustments. Rhode Island They are only for bituminous items and are fixed at 2.5 gallons of fuel per ton of bituminous South Carolina Developed in coordination with contracting industry Tennessee Originally from T5080.3, but survey of industry personnel and DOT calculations updated to current rates Washington Developed by following FHWA Technical Advisory West Virginia A task force was formed from the DOH, industry (contractors and suppliers), and local FHWA

Initial research and research approach 21 The age of the fuel factors is highly correlated with data source. For example, only 25 percent of DOTs using the 1980 Attachment 1 data and 36 percent of those using internal DOT data stated that their factors had been updated within the past 6 years (the North Carolina DOT conducted a review of the 1980 data and found that they were still acceptable for contracting in North Carolina). Conversely, 10 of 16 DOTs that utilize contractor-based data have updated their fuel factors within the last 6 years. The DOTs were then asked to identify which organizations were involved in the formulation of their state’s fuel factor policy. The DOTs themselves had the highest rates of participation, with 28 out of 37 state DOTs (76 percent) being involved in the creation of their fuel factors. Contractor organizations contributed slightly more than half the time. FHWA divisional offices and contractors followed shortly thereafter with approximately 43 and 38 percent participation, respectively. Maryland and Vermont reported that they took the fuel factor policies of neighboring states into consideration, although they did not identify the states with which they consulted. Exhibit 2-5 displays a tabulation of the organizations involved in the creation of state fuel usage factors. The vast majority of DOTs who responded (29 out of 36) have shared, or are willing to share, their fuel factors with outside organizations such as municipal or local agencies. Several commented that their fuel factors were available online, while others stated that the factors are public information. Some respondents provided more specific situations in which they would share the factors, such as if requested by county governments and consultant engineers (Minnesota) or on projects where the state DOT writes the specifications (Louisiana). The survey also ascertained the current abilities of state DOTs, through software or other means, to develop and/or calculate fuel factors and price adjustment clause payment. Exhibit 2-6 provides the results. Only three states have software programs used to develop the fuel factors themselves: Minnesota and Nevada, which use an Excel spreadsheet program, and Arizona, with a custom Web application. Each of these states has updated their fuel factors within the last 3 years. Exhibit 2-4. Dates of last fuel factor updates. Within the last year 10.7% Within the last 2-3 years 21.4% Within the last 4-5 years 14.3% Within the last 6-10 years 7.1% Over 10 years 21.4% Unknown 25.0% When were your Fuel Use Factors last updated?

22 Fuel Usage Factors in highway and Bridge Construction Approximately half of the states (23) use Excel spreadsheets and custom applications to calculate price adjustment clause payments. The other systems option yielded 11 responses. Five states employ Site Manager, Arizona uses another Web application, Maine uses Transport—CAS, West Virginia uses the Project Record System (PRS), New Jersey uses the Automatic Construction Estimate System, and North Carolina uses HiCAMS, an in-house calculation index. Alabama, Illinois, and Utah do not currently have applications to calculate price adjustment clause payments. 2.1.4 Bridges/Structures and Design/Build An important goal of the survey is to gain insight on the current state of fuel factor application in bridge/structure contracting. The study team composed questions designed to determine current Exhibit 2-5. Organizations involved in fuel factor creation. 15.8 43.2 75.7 5.3 37.8 52.6 10.5 21.1 0 FH W A - N ati on al FH W A - D ivi sio n DO T - St ate wi de DO T - Di str ict s Co nt ra cto rs Co nt ra cto rs Or ga niz ati on s Eq uip m en t S up pli er s Ot he r 10 20 30 40 50 60 70 80 90 100 Pe rc en t I nv ol ve d n=37 Total selected: 97 Exhibit 2-6. DOT fuel factor development and calculation methodology. 34 2 1 3 16 7 11 0 5 10 15 20 25 30 35 40 None Excel Spreadsheet Custom Application Other System Develop Factors Calculate Factors n=37

Initial research and research approach 23 practices, perceptions, and future improvements. The first question in this section asked state DOTs if they use fuel factors in bridge/structure contracting. Of the 37 responding states that employ fuel factors, 20 have factors applicable to bridges/structures (including decking). These states are concentrated in the East and West. Exhibit 2-7 provides the results of a question as to how the DOTs developed fuel factors for structures/bridges. Ten state DOTs responded that they use fuel factors for appropriate items. Eight states use other methods. Arizona and Georgia are the two states that employ percentage of cost. States that selected the “We do not develop Fuel Use Factors for structures/bridges” option have been omitted. The survey form provided the 12 states that chose “Other” with an opportunity to describe their fuel factor development methodology. Six out of 12 states chose to elaborate; Exhibit 2-8 displays their comments. Exhibit 2-7. Methods of developing bridge/structure fuel factors. Percentage of Cost 2 Fuel Use Factors 10 Other 8 n=20 Exhibit 2-8. Details of other fuel factor development systems. State Response Connecticut Based on contract value, which is currently set for contracts greater than $50 million. Please refer to specification for information on how factor derived. Idaho Developed in consultation with industry New York Unknown Oregon The factors are in gallons per $1,000 of work Pennsylvania Diesel Fuel Use Factor of 4 gallons per $1,000 of work performed applied to applicable component items only. South Dakota Data submitted by contractor

24 Fuel Usage Factors in highway and Bridge Construction Exhibit 2-9 provides the responses concerning perceived flaws associated with fuel factors for bridges/structures. The two largest areas of concern are changes in construction methods and fuel intensity and inaccuracies due to differing structure types, sizes, and complexities. Each of these options received 12 selections. Changing construction is an understandable response for states that have not updated their factors for some time, as seven of the respondents who chose this option either have not updated their factors within the last 6 years or do not know when the factors were last updated. This may be explained by changes in construction technology, improved fuel efficiency, and other factors that may have changed over the last 6 years or more. The differences in construction methods option and the lump sum option followed with 11 and 10 selections, respectively. Exhibit 2-10 provides substantive responses from the other category. Half of the six responses cited low fuel intensity for these items. The survey also queried respondents as to their methods for fuel factor adjustment in the circumstance where the state used design/build contracts or lump sum items for bridges/structures. Of the 37 states who responded, 20 chose one of the following options: gallons per contract dollar, percent of cost, contractor estimated quantity, DOT estimated quantity, supplied invoice, Exhibit 2-9. Perceived problems. What problems do you perceive with Fuel Use Factors for structures/bridges? (Check all that apply.) Answer Options ResponsePercent Response Count Inaccuracies due to inflation 13.5% 5 Inaccuracies due to differing structure types, sizes, and complexities 32.4% 12 Many items are bid lump sum 27.0% 10 Changes in construction methods and fuel intensity over time 32.4% 12 Differences in construction methods 29.7% 11 None 24.3% 9 Other (please specify) 29.7% 11 n=37 Total selected =70 Exhibit 2-10. Other bridge/structure fuel factor concerns. State Response Colorado Fuel use factors are for items of work, i.e., pile driving or caissons, not by structure type, size, or complexity. Any item of work that fuel usage can be measured can have a factor calculated for it based on unit of measurement, i.e., gals/ft. for pile or caissons. But fuel PACs should only be used for high fuel consumption items. They are not for every item. The idea is try to reduce the risk on contractors, not totally eliminate it. Iowa Comparatively smaller amounts of fuel used Massachusetts All of the above, except “None” Mississippi Typically, the fuel usage in this area of construction is not as high as other areas such as excavation, and fuel adjustments may not be necessary. Some contractors may include fuel usage in their unit prices so the adjustment may not be necessary. Pennsylvania Since Fuel Use Factor is applied to applicable component items only, portions of the monthly lump sum payment amount must first be reduced to discount non-applicable component items before factor can be applied. West Virginia Small quantities on the items we adjust

Initial research and research approach 25 or not important. The methods varied widely across states. For example, five states use gallons per contract dollar, three use percent of cost, three use a contractor-estimated quantity, four use a DOT estimated-quantity, and four use a supplied invoice. Of the 37 responses, 17 were under the “Other” category. Ohio stated that they use, “contractor-provided quantity based on calculated plan line verified through in-place measurements by the DOT,” while Maryland applies fuel factors after receiving a lump sum breakdown from the design builder. Seven states responded that they do not adjust on design/build contracts and lump sum items. The remaining states either have not considered the issue or do not use fuel factors for bridges/structures. Exhibit 2-11 provides a tabulation of the selected methods for design/build and lump sum adjustments. The final question of this section asked the DOTs to volunteer any ideas they had for improving fuel adjustments for bridges/structure design/build contracts and lump sum items. Four states responded. The most detailed suggestion came from Pennsylvania, which responded, “for lump sum structures and design/build projects, contractor must submit component item breakdown for use in determining payments, as well as computing price adjustments.” 2.1.5 Perceptions—Fuel Intensity/Volatility The objective of this section of the DOT survey was to determine the fuel intensity of various construction activities with the goal of pinpointing certain types of pay items that could benefit from updated fuel factors. DOTs were asked to rank the following construction activities in terms of fuel intensity (defined in terms of gallons of fuel used per contract dollar): grading/ excavation, drainage, asphalt paving, concrete paving, base stone/aggregates, and structures. Ties were not allowed. A ranking system was devised that gave each activity an average hierarchical rating. A “most” rating is worth one point, a “second most” rating is worth two points, and so on. The total assigned points for each activity were then added together and divided by 34 (the number of respondents) to determine each activity’s average ranking. Exhibit 2-12 provides the responses to this question and each activity’s average ranking. Exhibit 2-11. Methods for design/build and lump sum adjustments. Gallons per contract dollar 13.2% Percent of cost 7.9% Contractor estimated quantity 7.9% DOT estimated quantity 10.5% Supplied invoice 10.5% Not important 2.6% Other 47.4% There are contracts or items within contracts that are not unit based. For example, some states use design/build contracts or lump sum items for bridges. What method is best used for these contracts or items? n = 37

26 Fuel Usage Factors in highway and Bridge Construction Grading/excavation and asphalt paving dominate the top of the list, securing 100 percent of the first-place selections and slightly less than 70 percent of the second-place selections between them. Both activities require heavy machinery and equipment with high fuel consumption. Asphalt paving also necessitates the inclusion of petroleum-intensive asphalt cement. Strictly speaking for the purpose of fuel factors, asphalt is considered a material and is not fuel. However, it is not clear if this convention was assumed by the respondents. Structures came in as the least fuel-intensive activity. Overall, fuel consumption on bridge/structure projects is significant but additional costs such as the purchase of materials and staff salaries lower the percent contribu- tion. Vermont suggested adding cold planing and reclaiming to the list of fuel intensive activities, New Jersey suggested milling, and Connecticut suggested environmental excavation and disposal of hazardous or contaminated materials and site work on building construction for rail yards, airports, and train stations. The states were asked if recent fuel price fluctuations (such as during the summer of 2008) had altered their data/analytical needs when conducting fuel price adjustments. A large majority, 39 of the 48 states that responded to the question, replied that their existing methods remained sufficient. Exhibit 2-13 provides the responses of the other nine states. In general, these nine states made incremental changes to their programs. Exhibit 2-13 provides state responses regarding the effects of fuel price fluctuations. 2.1.6 Perceptions—Pay Item Selection The survey posed several questions related to pay item selection. Responses from state DOTs were useful in developing a list of pay items for which the study would develop fuel factors. Exhibit 2-12. Fuel intensity by construction activity. Please rank these activities in terms of fuel intensity, in gallons per contract dollar, in highway construction in your state (use your best judgment for ties). Answer Options Most 2nd Most 3rd Most 4th Most 5th Most 6th Most(Least) Rating Average Asphalt Paving 25 16 6 0 0 1 1.69 Grading/Excavation 23 17 6 1 1 0 1.75 Base Stone/Aggregates 0 8 20 12 7 1 3.44 Concrete Paving 0 2 10 18 9 9 4.27 Drainage 0 3 5 9 16 15 4.73 Structures 0 2 1 8 15 22 5.13 n=48 State Response South Carolina Expanded catalog of items eligible for fuel adjustments. Arizona We updated the formula. Pennsylvania During 2008 construction season, DOT did attempt to develop projections of potential price adjustment expenditures for planning purposes. This was primarily due to fluctuations in the cost of asphalt cement; however, diesel fuel was also included. Vermont Difficult to account for market volatility and time to respond for project budgets. Minnesota Changed from a 50% change from the base to 25%. West Virginia We had to watch the adjustment levels and budget appropriately. New Jersey Construction industry has appealed to NJDOT to revise its fuel usage factors. Connecticut Adjustments have been made to the formula for determining fuel cost and adjustment to better represent use and costs. Ohio Ohio has had a fuel price adjustment in place since 2005. Obtaining and maintaining index information has necessitated data collection and adjustment processing. Exhibit 2-13. Effects of fuel price fluctuations.

Initial research and research approach 27 When asked how to account for fuel use on pay items not already included in their state’s adjustment programs, a majority of 33 out of 48 states (68 percent) stated that additional factors were not necessary because the fuel use on additional items is limited. Only respondents from Idaho, Montana, Nebraska, New Jersey, Virginia, and West Virginia indicated that they would add fuel factors for additional items. New Jersey is considering using a contract-wide gallons per construction contract dollar application. Several respondents intimated that some new factors would be beneficial, but only for large projects or items. The next question queried respondents on the reasons for limiting the number of pay items. Insignificant fuel use for such items was cited by 31 states, 16 chose administrative cost/time, and 10 stated a lack of contractor interest. The respondent from Colorado indicated that fuel factors should only apply to high fuel use items, mentioning that the goal of a fuel adjustment program should be to lessen contractor risk without eliminating it completely. In a similar vein, the respondent from South Carolina stated that fuel adjustments and other indexes can lead to reductions in payments, something that contractors would like to avoid on low impact pay items. The respondent from New Jersey replied that NJDOT has been unable to quantify a fuel factor for these items. The last question in this section aimed to gauge opinions on how subcontractors should be compensated for fuel cost changes. Eight, six, and five states selected “not applicable, little fuel used,” “add additional fuel use factors,” and “use a percentage of cost method,” respectively. The “Other” section netted 28 responses. The general theme of “that’s between the prime and the sub” made up 15 of these 28 responses. Tennessee, Nevada, and Mississippi responded that the fuel adjustment is applied to the item of work and no distinction exists between primes and subs. 2.1.7 System Design This section of the survey was designed to provide the study team with feedback on the particular elements that may be included in future deliverables such as a software tool. FHWA Technical Advisory T5080.3 provides low, medium, and high values so adjustments can be applied to specific project conditions. These may include grade, terrain, altitude, soil type, and other variables. The majority of DOTs, 30 out of 46, said that this range of factors should not be included in the new system. Exhibit 2-14 provides the relevant comments provided on this question. State Response Arizona It is helpful to the DOT to know the range of applicable fuel factors. Oklahoma Could be helpful. Would need to review the results of the research to see if there would be any benefit. Vermont Keep it simple, if possible. Nevada Does not address regional issues. Oregon I would prefer one number with methods to adjust for individual states, regions, etc. New York Don't need them for unit priced work. Folks will tend to pick the middle number almost always. Utah The study should recommend ranges and each state or agency should have the freedom to take the recommendations and implement that which best fits their specific needs. Idaho Provided some guidance was given in how to apply the ranges. Mississippi Fuel factors need to either apply or not apply to specific items of work; the ranges would add too much subjectivity to the process. Colorado Maybe, CDOT would need to see the new system and test it to see how it applies to CDOT projects. Exhibit 2-14. Comments regarding low, medium, and high values.

28 Fuel Usage Factors in highway and Bridge Construction Units of measure for pay items often vary from state to state. When asked if the ability to convert units would be a helpful component of the new system, 26 states said it would, while 20 said it would not. Only Pennsylvania and New Jersey mentioned having this capability with their existing systems. Nearly 30 percent of the responding states (13) would like both high- medium-low variables and conversion ability, while 17 would want neither. The survey included a question that presented an assortment of variables that affect fuel factors for which a software system could account. The states were asked to select the ones they wanted to be included and were allowed to select multiple features. Exhibit 2-15 tabulates the 74 total options that were selected. Urban/rural and hauling distance each received the support of 20 and 19 DOTs, respectively, more than 40 percent of the total participating. Differing hauling distances to and from construc- tion sites will alter the relative contribution of fuel use in a project budget, making the effects of price fluctuations more noticeable. Variance in urban and rural construction characteristics, such as hauling distance, storage capability, and other factors, can likewise influence project fuel costs. Of the 46 responding states, 15 selected three or more variables. At the same time, 16 states did not want any variables to be included. Three respondents commented that such variables would incur added administrative and contractual burdens that might not justify their inclusion. The survey also queried respondents as to whether fuel factor calculation should be responsive to varying levels of geography. Exhibit 2-16 provides the tabulation of the states’ preferences for geographic level. More than 75 percent of respondents prefer a state or regional level for their fuel factors. The two “extreme” options garnered only about 24 percent combined support. This might be due to concerns that national factors would not be attuned to local conditions, while a project-specific system might be too cumbersome to manage efficiently. The survey also asked the DOTs how often fuel factors should be updated as well as how often they actually are updated. A cross-tabulation of these responses, which is presented in Exhibit 2-17, indicates consistency between how often factors are updated and how often DOTs want them updated. Only 6 of the 29 DOTs chose responses that were separated by more than one spot. This 6 3 7 20 7 19 15 12 16 8 0 5 10 15 20 25 n=46 Total selected: 113 So il T yp e Al tit ud e To po gra ph y/G rad e Ur ba n/ Ru ra l Tra ffi c L ev el Ha uli ng D ist an ce Fu el typ e - Eq uip m en t Fu el typ e - Pl an t No ne Ot he r Exhibit 2-15. Preferences for variable inclusion.

Initial research and research approach 29 may signal satisfaction with the current timing of fuel factor updates among the states, although it does not measure the qualitative aspects of their fuel factor programs. For those DOTs who answered “Unknown,” three wanted the factors updated every 2–3 years, three preferred every 4–5 years, and two chose no less than every 10 years. 2.1.8 Future Plans The state DOTs were asked what actions they would take if they had access to updated fuel factors and a software tool to reduce implementation costs. Exhibit 2-18 tabulates the 76 total responses selected by the 46 responding states. Almost two thirds of the states surveyed (30 out of 46) would either create a fuel factor adjustment program or institute changes to their existing programs. This includes 4 of the 12 states without fuel factor adjustments. The other five non-factor states who responded (Michigan, Texas, Wyoming, Montana, and Hawaii) would not make any changes. A total of 12 states with existing programs would retain their current systems as well. Approximately 45 percent of those surveyed (21 out of 46) would update their factors with the revised factors. Under the “Other” category, 11 states commented that they would evaluate the delivered products and determine how they would be used based on their effectiveness. Of those 11, 3 said they would collaborate with the contracting industry in their evaluation process. How Often Should Fuel Usage Factors Be Updated? How Often Fuel Usage Factors Are Updated Within the Last Year Within the Last 2-3 Years Within the Last 4-5 Years Within the Last 6-10 Years Over 10 Years Every Year 2 Every 2-3 Years 1 3 3 1 Every 4-5 Years 1 4 2 4 Every 6-10 Years 1 1 4 No Less than 10 Years 1 Never 1 Exhibit 2-17. Last updates and timing of future updates. Exhibit 2-16. Geographic preference. At what geographic level should the system attempt to develop Fuel Use Factors? National level only 6 Regional level only 12 State 23 Project 5 n = 46

30 Fuel Usage Factors in highway and Bridge Construction Several DOTs offered recommendations on items for which they would be interested in new or updated fuel factors. Exhibit 2-19 provides these responses. Of note was interest in factors for guiderail and non-standard HMA mixes. The survey concluded by querying respondents as to any concluding comments they might have. Exhibit 2-20 provides these comments. Responses were quite varied, recommending allowances for geographic adjustments, periodic updating, simplicity, and gathering input from contractor organizations. 2.1.9 Summary and Conclusions A large majority of the states surveyed employ fuel factors in some form, but many states rely on antiquated sources for their factors and/or do not update their factors regularly. Of the 50 states that responded to the survey, 38 employ some form of fuel price adjustment using fuel factors on construction contracts. The 1980 data in FHWA Technical Advisory T5080.3 is the source of fuel factors for 21 of these states. A majority of 21 states employing fuel factors have not updated their factors within the last 6 years or do not know when their factors were last updated. 0 5 10 15 20 25 Ad d PA C Ad op tN ew Sy st em U se Re vi se d Fa ct or s U se Ad di ti on al Fa ct or s U se So ft w ar e To ol U se Ex is ti ng M et ho ds O th er Yes NoUse Fuel Factors? n=46 Total =76 Exhibit 2-18. Anticipated actions with updated fuel factors and software tool. State Response Pennsylvania Bituminous pavement milling Vermont The equipment is changing quickly and there needs to be some way to address new equipment, classes of equipment, or ranges of power plants. New York Guiderail Colorado Non-standard HMA mixes like warm mix, shingle mixes, etc. New Jersey Consider a fuel usage factor based on total contract cost and based on contract type. Exhibit 2-19. Preferred additions.

Initial research and research approach 31 The use of fuel factors in bridge/structure contracting is common, but several flaws act as a hindrance to their effectiveness. Fuels factors for bridges/structures were present in 20 states, 40 percent of the total surveyed. When asked to divulge perceived flaws in bridge/structure fuel factors, 28 out of 37 states responded with at least one criticism. Changes in construction methods and fuel intensity and differences in structure type, size, and complexity were perceived as the largest flaws, receiving 12 selections each. Respondents had similar perceptions of the activities that were most fuel intensive. Asphalt paving and grading/excavation were the decisive top choices when ranking construction activities by fuel intensity, and shared all 48 first-place rankings between them. Recent fluctuations in fuel price affected the data/analytical needs of nine states. Respondents had mixed opinions on whether they desired fuel factors for a broader spectrum of items. A total of 33 out of 47 states believe it is unnecessary to include fuel factors for additional pay items due to limited fuel use. Administrative burden was cited as justification for limiting the number of fuel factors by 16 states. State DOTs had definite, although sometimes conflicting, ideas on the form fuel factors should take and how often they should be updated. The ability to convert units of measure in the new system received support from 26 states, while the inclusion of high-medium-low factor ranges would be useful for 16 states. Urban/rural and hauling distance were the most popular options when selecting additional variables for the system, receiving 20 and 19 selections, respectively, although 16 states would not want any additional variables. Seventy percent of states would like the system to be configured at the state (23) or regional (12) level. A majority of 34 states would like the factors to be updated every 5 years or less. State DOTs shared a high level of interest in new research on fuel factors. For example, almost two-thirds of those responding (30 out of 46) would begin a fuel factor program or implement changes to their fuel factor adjustments if presented with revised factors and a software tool. Only 12 states with fuel factor programs would retain their existing methods, while 5 states that do not implement fuel factors would continue to refrain from utilizing them. Several states said they would evaluate the delivered products and consult with the contracting industry before moving forward. 2.2 Contractor Needs and Perceptions The study team conducted a nationwide survey of contractors. Designed as a precursor to the more detailed Contractor Fuel Usage Survey, this survey explored the basic components of the fuel usage experiences and methodologies of construction firms. An additional goal State Response Oregon The fuel factors that come from this study should have the capability to be adjusted for different areas of the country. They should also provide tools to update the fuel factors on a periodic basis. Louisiana Keep it as simple as possible: the more variables, the more mistakes. From an audit standpoint we spend more time recalculating and correcting adjustment errors in our fuel and asphalt adjustments than on the rest of the construction items in a contract. Colorado Highly recommend that this survey be provided to contractor associations for their feedback. New Jersey It is likely that fuel usage varies based on contractor. Analysis should look to normalize average usage among efficient contractors and not simply average in inefficient contractors. Exhibit 2-20. Final comments.

32 Fuel Usage Factors in highway and Bridge Construction was to determine methods to maximize the visibility and effectiveness of the later survey. The findings of the initial survey are presented in Section 2.2. Subsection 2.2.1 explains the survey methodology and response. The following five subsections, delineated by responses in each survey category, enumerate the survey findings. Subsection 2.2.7 summarizes the survey and offers conclusions. 2.2.1 Survey Methodology and Response The study team provided the NCHRP study panel with a draft copy of the contractor survey on September 29, 2010. The study team reviewed the comments and suggestions of the panel members and made appropriate changes. On October 11, 2010, the study team distributed invitations to 500 contractors to participate in the online survey. Contractors were selected through a random sample of bids in order to ensure a representative sample. This invitation requested that surveys be completed by October 25, 2010. Additional invitations were sent on October 25 and November 1, 2010. Additionally, the study team contacted several additional randomly selected contractors by phone in an effort to amplify participation. These requests resulted in 63 survey responses. The response rate of 13 percent equals the response rate of the original 1980 fuel factor survey disseminated by the American Road and Transportation Builders Association (ARTBA) and the Associated General Contractors of America (AGC). 2.2.2 General Company Information The 63 survey respondents include firm owners and presidents, vice presidents, chief estimators and engineers, and other high-ranking company officials. The responding firms vary widely in size and specialization. Firms with 100 to 200 employees represent the largest group of respondents, with 20 responses, nearly a third of the total. Firms with 200 to 500 employees followed with 18 responses. Small firms with 100 employees or less garnered 20 responses. Very large firms of over 500 employees accounted for five responses. Exhibit 2-21 displays the number of respondent firms by employment size class. Exhibit 2-21. Firm size of survey respondents. 0-50 employees 9 (14.3%) 50-100 employees 11 (17.5%) 100-200 employees 20 (31.7%) 200-500 employees 18 (28.6%) Over 500 employees 5 (7.9%) n=63

Initial research and research approach 33 The questionnaire queried respondents as to the primary and secondary types of work their firms perform. The responding firms conduct varied operations. At least one response was registered for each of the 11 provided work categories. Asphalt paving received the most responses and is the primary area of operations for nearly half of the responding firms. More than half of the contractors selected the bridge and grading categories as well. Additional contracting areas enumerated under the “Other” category include general, civil, marine, and industrial contracting, building construction, base stones/aggregates, and research. Exhibit 2-22 displays the areas of work of responding firms. The final question in this section asked contractors if the DOT in their primary state of operation uses fuel factors to determine price adjustments in construction contracts. More than 75 percent of contractors (50 total) replied in the affirmative. Exhibit 2-23 displays these results. Exhibit 2-22. Areas of work.* Primary 2nd Most 3rd Most Response Count 32 3 3 38 13 13 6 32 7 18 10 35 0 9 13 22 3 4 1 8 3 1 5 9 0 6 3 9 2 0 1 3 0 1 1 2 0 0 2 2 0 0 1 1 0 0 1 1 13 n=175 Other (please specify) Storm Sewer/Drainage Guardrail Landscaping What type of work does your company primarily perform? Answer Options Asphalt Paving Bridge Grading Concrete Paving Water/Sanitary Sewer Misc. Concrete Electrical/Signalization Pavement Marking Clearing/Demolition *Arranged by weighted ranking. “Primary” responses are worth three points, “2nd Most” responses are worth two points, and “3rd Most” responses are worth one point. Exhibit 2-23. Presence of fuel factors in primary state. Yes 50 (79.4%) No 13 (20.6%) n=63

34 Fuel Usage Factors in highway and Bridge Construction 2.2.3 Estimating Methods This section of the contractor survey investigated the estimating methods used by contractors. The responding contractors utilize several different methods to calculate fuel cost. Fuel consump- tion rates by equipment type proved to be the most popular, garnering 24 of 46 responses, or slightly more than half of the total. Smaller numbers of contractors selected percentage of equip- ment cost (nine), percentage of total cost (six), and DOT supplied fuel factors (six). Exhibit 2-24 provides the estimation methods of responding firms. The contractors surveyed employ a variety of sources for their fuel consumption rates. Internally developed rates received 24 responses. Equipment manufacturer’s rates received 16 responses, while historical rates received 15 responses. One contractor uses gallons per second (GPS), while another respondent did not know the source of his firm’s fuel consumption rates. Exhibit 2-25 displays the sources of contractors’ fuel consumptions rates. Exhibit 2-24. Fuel cost estimation methods. Fuel Consumption Rates (by equipment) 24 (52.2%) Percentage of Equipment Cost 9 (19.6%) Percentage of Total Cost 6 (13.0%) Use DOT Fuel Use Factors 6 (13.0%) Other (please specify) 1 (2.2%) n=46 Other (2) Equipment Manufacturer (16) Internally Developed Rate (24) Historical Rate (15) Blue Book Rental Rate Guide (6) Not Applicable (2) 0 5 10 15 20 25 30 n=46 Exhibit 2-25. Sources of fuel consumption rates.

Initial research and research approach 35 When asked when their fuel consumption rates or factors had last been updated, 23 out of 46 contractors indicated within the last year. Overall, 39 out of 46 respondents (more than 84 per- cent) have updated their factors within the last 3 years. Based on the relatively rapid updating of their factors, it appears that contractors have a strong incentive to keep their factors as current as possible in order to facilitate accurate estimation. Two contractors commented that they update the factors when new equipment is purchased. Of the 46 responding contractors, 40 said that they use a tool or software application to prepare their estimates. The most popular method is using a commercial estimating system, which was selected by 22 respondents. An Excel spreadsheet application is used by 10 contractors, or slightly more than a fifth of those responding. Two contractors use programs developed by a construction estimating firm. Exhibit 2-26 displays contractor use of estimating tools and software applications. Sixty percent of contractors (28 out of 46) report that their applications have the capacity to calculate the quantity of fuel needed for a project, while 17 out of 46 do not. One contractor, presumably one of the six without a tool or application, chose “Not Applicable.” Only three contractors using commercial systems cannot calculate fuel quantities. Exhibit 2-27 displays contractor ability to calculate a project’s fuel quantity. A cross-tabulation of the data by firm size reveals that smaller firms are less likely to have the ability to calculate fuel quantities used. Five of the six responding firms with less than 50 employees do not currently have this capability. More than 80 percent of firms with 200 to 500 employees and both firms with over 500 employees possess the capability to calculate fuel quantities. 2.2.4 Fuel Consumption Items The contractors were asked to rank various construction activities in terms of fuel intensity. As was the case with the DOT survey, asphalt paving and grading/excavation were the clear top two choices, combining for 52 of 55 selections for the most fuel intensive activity. Grading/excavation is viewed as more fuel intensive than asphalt paving, receiving 34 top selections compared to 18 for asphalt paving. In comparison, asphalt paving received 25 first-place selections in the DOT survey, No 6 (13.0%) Excel Spreadsheet 10 (21.7%) Commercial Estimating System 22 (47.8%) Internally Developed System 6 (13.0%) Manual System 0 (0%) Other System 2 (4.3%) n=46 Exhibit 2-26. Use of estimating tools and software applications.

36 Fuel Usage Factors in highway and Bridge Construction with grading/excavation receiving 23 first-place selections. Exhibit 2-28 displays relative fuel intensity perceptions. The contractors identified eight other fuel intensive activities. Five of these activities involve asphalt and/or aggregates, and three relate to the handling and transportation of construction equipment and materials. Exhibit 2-29 provides additional comments regarding fuel intensive activities. Nearly two-thirds of the responding contractors (35 out of 55) believe that recent fuel price fluctuations have not altered their analytical needs. Nineteen of the remaining contractors said that their needs had changed and offered explanations (one contractor responded “Yes (please explain)” to this question but put only a dash in the text box). Four respondents have increased their bids or included escalators/hedging as an attempt to control cost. Exhibit 2-30 provides these responses. 2.2.5 Perceptions The responding contractors had varying opinions on how to account for fuel use in excluded pay items. Sixteen would prefer to add additional items through industry consultation and thirteen would add fuel factors to other items of work. However, 14 said that the fuel use for Exhibit 2-27. Ability of estimating tools/software to calculate fuel quantities. No 17 (37.8%) Yes 28 (60.9%) Not Applicable 1 (2.2%) n=46 Exhibit 2-28. Fuel intensity by construction activity. Answer Options Most 2nd Most 3rd Most 4th Most 5th Most 6th Most (Least) Rating Average DOT Rating Average Grading/Excavation 34 14 7 0 0 0 1.51 1.75 Asphalt Paving 18 21 10 2 1 3 2.20 1.69 Base Stone/Aggregates 1 13 21 11 6 3 3.31 3.44 Concrete Paving 0 1 10 23 14 7 4.29 4.27 Drainage 1 4 3 18 22 7 4.40 4.73 Structures 1 2 4 1 12 35 5.29 5.13 n=55

Initial research and research approach 37 other pay items is limited and fuel factors should not be extended. Two contractors recommended eliminating fuel factors altogether, as follows: • Eliminate fuel use factors and use a more equitable and accurate method of accounting for the risk of fuel cost escalations. Fuel use factors are not equitable even on the traditional items to which they are applied. Fuel use is very dependent on type of equipment owned, more so than what work is performed. • Drop the use of fuel factors and use three lump sum items (grading, paving, and structures) to allow the contractor to place the fuel dollars he wishes to be indexed in the program. Due to timing and fuel pricing, the major work disciplines need to each have their own fuel item. On medium- and long-term projects, it is possible for the grading work to experience a significant windfall while the paving work can lose significant fuel dollars, and vice versa, and all due to timing of the work versus actual fuel prices. Respondent Comment 1 Materials transport trucking 2 Line utility work 3 Quarrying aggregates 4 On-highway commuting 5 Materials handling 6 Milling 7 Transportation of aggregates for asphalt by barge 8 Plant fuel use 9 Delivery of asphalt 10 Cold planing asphalt Exhibit 2-29. Additional contractor-identified fuel intensive activities. Respondent Comment 1 We add a "fuel" factor as a lump sum to our bid based on the length of the project (for non- covered items like concrete) 2 Price per ton of asphalt 3 No analytical tool can predict what the fuel price will be when a project has no fuel adjustment 4 Need to be able to quickly analyze cost data 5 Can't rely as much on our historical information, have to make projections based on current info 6 We have to add enough for fuel so that it doesn't kill us to do the job. We overestimate the cost of fuel on purpose. 7 Productivity concerns on minimums needed to be competitive and fuel conscious 8 We track spot prices to purchase prices and purchase futures based on spot prices 9 Plan for worst case 10 When fuel was stable it was more like a fixed cost on the project with little or no variation. When it started having major fluctuations and with projects that extend over multiple months or years, the fuel became a major concern. 11 Seasonal pricing, futures, intensity of work, and timing of major activities, theft control 12 Only price 13 Much more attention is paid to actual fuel unit pricing and how it relates to the rates posted by the DOT 14 Had to attempt to bid in fuel escalators from suppliers 15 We now factor fuel in our bids 16 Petroleum-based material cost fluctuations 17 Fuel consumption and pricing are monitored much closer now than in the past 18 We look at fuel per piece of equipment annually, project drying cost and trucking fuel requirements for large jobs 19 Fuel cost is analyzed with every large bid and protected sometimes by hedging Exhibit 2-30. Additional contractor comments on experiences with price fluctuations.

38 Fuel Usage Factors in highway and Bridge Construction Approximately half of the responding contractors (25 out of 51) pass along fuel price adjust- ments to their subcontractors, while 14 do not. The remaining 12 utilize varying practices. Three contractors said it depends on the subcontractor’s quote, two said that price adjustment clauses are negotiated, and two said that price adjustment inclusion can occur if the subcontractor requests it. When asked to provide opinions on how to compensate subcontractors for increased fuel costs, 23 contractors approve of including additional fuel factors, while using a percentage of cost method and not extending fuel factors due to limited fuel use each received support from 10 contractors. Eight contractors selected the “Other” category and enumerated their preferences. Exhibits 2-31 and 2-32 display contractor preferences for reimbursing subcontractors and additional comments on the topic, respectively. When asked what approximate percent of the contract dollars they received from the state DOTs were subcontracted, 86 percent (44 out of 51) selected an option of 30 percent or less. The selections “11–20 Percent” and “21–30 Percent” were chosen by 20 contractors each. Only two respondents selected “Over 40 Percent.” In a similar vein, contractors were asked what percentage of their DOT contract dollars were performed as a subcontractor. A substantial majority of 94 percent (48 out of 51) operate as Not applicable, little fuel used 10 (19.6%) Add additional Fuel Use Factors 23 (45.1%) Use a percentage of cost method 10 (19.6%) Other 8 (15.7%) n=51 Exhibit 2-31. Preferences for compensating subcontractors. Respondent Comment 1 They should get the increase/decrease on the item they are doing 2 Grading and asphalt subs are compensated 3 Set up terms and conditions under subcontract by type of work and amounts of fuel being used 4 Generally negligible, but depends on which items of work are subcontracted 5 Negotiate adjustment factors with subs based on agreed-upon usage 6 WYDOT's method is a percentage of cost method, which we pass on to any sub that chooses to participate when we choose it on the prime contract 7 Non-factor unless the subcontractor states it in their quote 8 Allow the contractor to manage Exhibit 2-32. Additional contractor methods for subcontractor compensation.

Initial research and research approach 39 subcontractors for 40 percent or less of their DOT project dollars, and 31 out of 51 did so 20 percent of the time or less. The final question of this survey was designed to gauge contractor satisfaction (or lack thereof) with their primary state DOT’s fuel factors. Sixty percent (30 out of 50 respondents) believe that their fuel factors are somewhat accurate. Twenty-six percent (18 out of 50) stated that their factors are somewhat or very inaccurate. Exhibit 2-33 displays perceptions on fuel factor accuracy. Several contractors offered additional comments. Two express general satisfaction with the current factors, one mentions the dualistic nature of price adjustment clauses, and another comments on the difficulty in creating and utilizing a single fuel factor. Exhibit 2-34 displays additional comments regarding fuel factor accuracy. 2.2.6 Future Plans Seven contractors responded with pay items that they would like revised or for which they would favor the development of additional fuel factors. Four of the seven would like transportation 2 (4.0%) 30 (60.0%) 18 (26.0%) 5 (10.0%) 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% Very accurate Somewhat accurate Somewhat inaccurate Very inaccurate n=50 Exhibit 2-33. Accuracy of fuel usage factors. Respondent Comment 1 Concrete paving, asphalt, and grading are the largest pay factors and they seem adequate 2 When the fuel decreases, it takes too much money away from the contractor, and when it increases, it gives too much money to the contractor 3 Fuel consumption rates vary as to haul lengths, method of work, type of materials, and grades—all these factors have major impacts to fuel consumption and make the use of a single fuel factor inaccurate from the beginning 4 Seems close 5 The paving diesel factor only covers about 50% of the fuel used Exhibit 2-34. Contractor comments on fuel usage factor accuracy.

40 Fuel Usage Factors in highway and Bridge Construction and hauling to be added. Two contractors support the addition of bridge/structure pay items. Exhibit 2-35 displays contractor comments regarding additional or updated fuel factors. Two contractors provided comments for improving or refining the use of fuel factors. One recommended the automation of index price adjustments as a means of ensuring accuracy. The other suggested studying the system employed by the Wyoming DOT. Six contractors provided advice on how the study team can maximize participation in the later fuel use survey. Several contractors suggested addressing the survey to estimators, project and equipment managers, and accountants. Another suggested working through the Associated General Contractors of America (AGC) and the National Asphalt Pavement Association (NAPA). Exhibit 2-36 displays these, and other, contractor suggestions. 2.2.7 Summary and Conclusions Nearly 80 percent of the responding contractors operate primarily in states that use fuel factors. A sizable majority (39 out of 46) of responding contractors have updated their fuel consumption rates or factors within the last 3 years, while less than a third of state DOTs have done the same. Individual contractors would seem to have an incentive to update this information regularly as a means of increasing bid accuracy and eliminating uncertainty. The most popular method of fuel cost estimation is fuel consumption rate by equipment type, which is used by 52 percent of responding contractors. Contractors most often utilize internally developed rates, although historical rates and rates supplied by equipment manufacturers are Respondent Comment 1 Bridge items 2 Grading and excavation, hauling (stone/earth/demo) 3 The production of hot mix asphalt mixes 4 Trucking 5 Storm drainage items, box culvert items, box bridge items, bridge items 6 Transportation costs (trucking/barge) 7 Item 502-01: There should be two separate diesel fuel indices calculated. Plant drying fuel should be separate from transportation and paving fuel usage. There should also be a factor for trucking and barging aggregate. Item 501-01: Should be eligible for the asphalt index for square yard patching items. The fuel usage factors should be re-evaluated for all pay items. Exhibit 2-35. Contractor comments on additional or updated fuel usage factors. Respondent Comment 1 Equipment managers and estimators 2 Work through AGC, NAPA, etc. 3 Contact the organization’s estimators and project managers. They are the ones who deal with fuel adjustments on a regular basis and can provide the most input; DOT Road Builder Associations. Keep it short and sweet. 4 This topic has broad applicability to off-road emissions reduction targets. The more accurate and broad the dataset, the better we will be able to respond to U.S. EPA/California emissions mandates. 5 Address inquires to the accountants. 6 Paving contractors are primary users of fuel. This group should be main focus. Exhibit 2-36. Contractor suggestions for the fuel use survey distribution.

Initial research and research approach 41 also popular. Eighty-seven percent of responding contractors employ a software application or estimating tool, with commercial estimating services being the most popular option. More than 60 percent of responding contractors report that their estimating tools are capable of calculating fuel use, although nearly 40 percent are not capable of doing so. The contractors and DOTs are in broad agreement over the fuel intensity of various construction activities. Although the contractors believe that grading/excavation (rather than asphalt paving) is the most fuel intensive work type, the rankings of the remaining activities were the same in both surveys. About two-thirds of contractors replied that recent fuel price fluctuations had not affected their analytical needs. This percentage is lower than the 81 percent of state DOTs who believe similarly. More contractors pass on price adjustments to their subcontractors than not. Slightly less than half of the responding contractors believe that additional fuel factors should be added to cover subcontractors’ increasing costs. Large majorities of contractors subcontract between 11 and 30 percent of their DOT contract dollars and perform less than 40 percent of their DOT contract dollars as a subcontractor. More than 60 percent of contractors believe that their state’s fuel factors are somewhat or very accurate. The remaining 36 percent believe them to be somewhat or very inaccurate. The results for the contractor survey illustrate several trends that may be ameliorated by updated fuel usage factors. Contractors update their fuel consumption rates or factors more often than state DOTs. Fluctuations in commodity pricing have a larger effect on contractors than DOTs, primarily due to smaller operating budgets. Contractors then have an incentive to update and maintain factors. While 60 percent of the responding contractors expressed satisfaction with the accuracy of their primary state’s fuel factors, nearly 40 percent find them to be somewhat inaccurate at best. Inaccuracies can be compounded if a contractor’s estimating tool cannot calculate the amount of fuel used on a project, which nearly 40 percent of respondents indicated. 2.3 Initial Engineering Estimation of Fuel Intensity As part of the study team’s three-pronged approach to addressing the research problem, the project team conducted an initial investigation to determine construction pay items that had high fuel intensity. An expert panel of professional estimators and contractors rated the fuel use of over 1,000 specific pay items. The ratings of individual estimators were averaged to create a composite ranking of fuel use. Reviewer D is a member of the research team and Reviewers A through C performed as consultants for the research team. Each member of the panel possesses at least 25 years of experience in the highway construction and/or cost estimation fields. This analysis consisted of three parts, as follows: • Creating a list of pay items to study by filtering unsuitable pay items, • Creating a ranking system to apply to the pay items, and • Performing the fuel use ranking of each pay item and pay item category. 2.3.1 The Expert Panel The initial engineering estimation was conducted using a four-person expert engineering panel. Each member of the panel estimated the relative fuel intensity of over 1,000 specific pay items and 31 summary categories. The four panel members were • Expert Panel Member A is a civil engineer and former district engineer and contracting officer in the U.S. Army Corps of Engineers. He has over 30 years of experience in the heavy/highway

42 Fuel Usage Factors in highway and Bridge Construction construction industry as an estimator, project manager, division manager, operations manager, and vice president. • Expert Panel Member B has nearly 35 years of experience with the Georgia Department of Transportation, has estimated or supervised estimation for over 8,000 DOT projects, is a three- time chairperson of the Transportation Estimators Association, and has been elected to the FHWA’s Peer Team Review. • Expert Panel Member C is a veteran consulting estimator for the heavy construction industry with 30 years of experience. • Expert Panel Member D has over 25 years of experience in the road building industry, is the creator and primary developer of the BidTabs Professional and ProEstimate line of estimating software, and assisted in the development of the FHWA Highway Construction Cost Index. 2.3.2 Pay Item Selection The first part of this analysis was to develop criteria for filtering the list of pay items to eliminate unsuitable pay items. The source of the data was the BidTabs professional database development, which contains all pay item prices for all DOT contracts in 48 states (Alaska and Hawaii are not included). This database also divides items among 31 predefined categories of pay items that are assigned to every standard pay item in the database. The first step in developing the database was to exclude older data. The decision was made to eliminate data prior to 2006. The second step was to eliminate data for bids that were not awarded, leaving the low bid only. The third step was to eliminate lump sum pay items and non- standard pay items. Since each of these bids was for a unique construction item, there is no basis for comparison amongst them. The final step was to eliminate pay items with a bid frequency of less than 100 bids during the selected time period. Items that are purchased so infrequently would not be useful for inclusion in the final fuel factor database. The results of the program generated a list of 1,176 pay items across all states and across all pay item categories. 2.3.3 Fuel Intensity Ranking The project team developed a scale to use in the classification of pay items based on “fuel intensity.” The scale ranges from one to five (1–5) with items marked as a 5 being “heavy use” and items marked as a 1 being “light use.” To have a better understanding of the actual fuel use as a percentage of cost, the team identified two known very heavy use items: (1) on-road truck haul excavation and (2) off-road truck haul excavation. These tasks include only labor and equipment cost and heavy fuel consumption equipment. The team priced these items to determine the fuel cost as a percentage of the total cost and this value allowed the team to establish an upper end value for “high use” items. From this analysis, the team then established ranges to use in the fuel ranking. The fuel cost strictly as a percentage of the pay item cost (equipment, labor, etc.) was 22 percent and 28 percent. Adding 10 percent overhead and 10 percent profit to this pay item yielded a fuel cost percentage of 18 percent and 23 percent of the estimated bid price. Using this range of values as the high end due to hauling being a very fuel intense activity, the project team used a value of over 15 percent as the top-end fuel ranking. Breaking this down into five categories, the project team set the fuel ranking system as follows: 1. Less than 1 percent, 2. From 1 to 5 percent, 3. From 6 to 10 percent, 4. From 11 to 15 percent, and 5. More than 15 percent. These values provided a guide to the expert panel.

Initial research and research approach 43 2.3.4 Ranking of Fuel Intensity Once the pay item list was created and the ranking method determined, each member of the expert estimating panel assigned a value to each pay item. In addition, each team member assigned a ranking to each of the 31 summary pay items. Exhibit 2-37 provides fuel use rankings at the 31 summary pay item level. The first four columns provide the ranking selected by the four reviewers at the 31 summary level. The fifth provides the average of the four rankings. The sixth column shows the range of the rankings as a measure of the variation. The final column provides the average of the values for the detailed pay items within each category. 2.3.5 Initial Recommendations Historically, the most common categories of pay items used for fuel use factors are grading, asphalt, base stone, and concrete pavement. All four of these categories ranked high in both the summary and detailed analysis. Exhibit 2-38 breaks down the categories into three sections of high, medium, and low fuel use based on the rankings. The pay items are listed from highest to lowest fuel use within each column. Exhibit 2-37. Fuel use rankings by category.* *Fuel intensity is estimated on a 1 to 5 scale with 1 being the least intense. A B C D GRADING/EXCAVATION 5 5 5 5 5.00 - 4.67 CLEARING 5 3 4 5 4.25 2.00 3.24 MOBILIZATION 5 4 4 3 4.00 2.00 2.41 BASE STONE 3 4 4 4 3.75 1.00 2.85 MISC STONE/RIPRAP 5 3 3 3 3.50 2.00 3.00 CONCRETE-PAVEMENT 3 3 4 4 3.50 1.00 2.99 ASPHALT 2 4 4 4 3.50 2.00 2.83 EQUIPMENT/LABOR 3 4 3 3 3.25 1.00 4.25 UNDERDRAIN 5 1 3 3 3.00 4.00 3.08 BRIDGE 3 3 3 3 3.00 - 2.32 DRAINAGE-PIPE 3 2 3 3 2.75 1.00 3.01 DRAINAGE-INLETS/CATCH BASINS 3 1 4 3 2.75 3.00 2.30 CONCRETE-MISC 3 1 4 3 2.75 3.00 2.12 EROSION CONTROL 4 1 4 2 2.75 3.00 2.01 UTILITY-WATER 3 2 3 2 2.50 1.00 2.63 UTILITY-GAS 3 2 3 2 2.50 1.00 2.63 UTILITY-SEWER 3 2 3 2 2.50 1.00 2.63 RETAINING WALL 3 2 3 2 2.50 1.00 2.50 CONCRETE-CULVERTS 3 1 3 3 2.50 2.00 2.30 TRAFFIC CONTROL 4 2 2 2 2.50 2.00 2.02 GRASSING 3 2 2 1 2.00 2.00 2.51 GUARD RAIL 3 1 2 2 2.00 2.00 2.20 FENCING 3 1 2 2 2.00 2.00 2.17 MISC ELECTRICAL 3 1 3 1 2.00 2.00 1.77 ROADWAY LIGHTING/ELECTICAL 3 1 3 1 2.00 2.00 1.77 STRIPING/PAVEMENT MARKING 3 1 2 2 2.00 2.00 1.75 SIGNALIZATION 3 1 2 2 2.00 2.00 1.57 SIGNS-PERMANENT 3 1 2 2 2.00 2.00 1.50 BUILDINGS/MISC STRUCTURES 3 1 3 1 2.00 2.00 1.31 PAINTING STRUCTURES 2 1 2 1 1.50 1.00 1.75 ALTERNATES/BONUS/TIME 1 1 1 1 1.00 - 1.63 Category Detail Average Reviewer Average High-Low

44 Fuel Usage Factors in highway and Bridge Construction Several pay item categories in the high group have been removed from this list. These categories and the reasons for their exclusion are presented in Exhibit 2-39. 2.4 Initial Statistical Analysis of Fuel Intensity This report section documents the development of the BidTabs database that the study team analyzed as part of this project. In this initial effort, the objective was to examine which pay item prices are sensitive to changes in fuel prices in order to develop a list of items for which to develop fuel use factors. The thesis was that if there is no association between fuel prices and pay item prices, it would not be necessary to provide a price adjustment clause for those pay items. The initial statistical analysis consisted of three steps. The first step was to tabulate unit prices for pay items over time. The second step was to develop price indices for fuel. The third step was to conduct the initial BidTabs statistical analysis. 2.4.1 Selecting Pay Items for the Development of New Fuel Usage Factors The study team designed the database so that it would contain prices over 3 to 5 years. The study team selected a start date of January 1, 2006, and an end date of September 1, 2010. In total, data are available in the Oman Systems BidTabs Database for 335,564 separate pay items. For these pay items, there are almost 3.6 million low bids. Note that low bids are the unit price bid for the pay item in the winning low bid as opposed to the lowest bid for that pay item. Exhibit 2-40 summarizes the process of filtering the pay items used for analysis. To prepare the database, the study team excluded records that were not suitable for the analysis. The first step was to exclude non-standard pay items. Non-standard pay items are items that do High Medium Low Grading/Excavation Drainage – Pipe Grassing Clearing Drainage – Inlet Guardrail Mobilization Concrete – Misc Fencing Base Stone Erosion Control Misc. Electrical Misc Stone/Riprap Utility – Water Roadway Lighting Concrete – Pavement Utility – Gas Striping/Pavement Mark Asphalt Utility – Sewer Signalization Equipment/Labor Retaining Walls Signs – Perm. Underdrain Concrete – Culverts Buildings/Misc. Structures Bridge Traffic Control Painting Alternates/Time Exhibit 2-38. Ranking of pay item categories by fuel use. Pay Item Category Justification for Exclusion Equipment/Labor This category consists of equipment rental or labor hour pay items and is used only by a very limited number of states and is rarely used by those states. Clearing This category is typically bid utilizing lump sum pay items. Mobilization This category is typically bid utilizing lump sum pay items. Exhibit 2-39. Excluded pay item categories.

Initial research and research approach 45 not have the same definition or units from one project/bid to another. Therefore, there is no price per unit of work. There is no ability for the analysis to compare unit price across projects or over time. There is no ability for the analysis to regress unit price against fuel prices to assess the existence of a relationship or correlation. Note from Exhibit 2-40 that the exclusion of non- standard pay items from the sample does not have a large impact on the total number of records included in the study. Although the number of pay items excluded is a large percentage of the total number of pay items, these items were bid much less frequently than standard pay items, resulting in a much smaller percentage drop in the number of records included in the study. The second step was to exclude lump sum pay items. Lump sum pay items are items for which the bid quantity is essentially equal to one. For example, build one bridge or pave one section of road. In this case, there is once again no price per unit of work and therefore no ability for the analysis to compare unit price across projects or over time. There is no ability for the analysis to regress unit price against fuel prices to assess the existence of a relationship or correlation. The exclusion of non-standard items only reduces the number of pay items by about 14,000, but again reduces the number of bids by only about 10 percent. The third step was to exclude pay items that the issuing state DOT did not put out for bid with much frequency. In this case, the analysis excluded pay items if there were less than 100 lettings of that item over the examination period, or approximately 22 bids per year. The research team determined that this average of 22 observations per year provided sufficient data to determine both means and variability in each year. The purpose of excluding pay items with very few bids is that the small sample size may hamper the ability to accurately assess the existence of a relation- ship or correlation. The exclusion of these pay items reduces the number of pay items drastically to about 2 percent of the original number of pay items, but reduces the number of bids to only about half of the original number of pay items. Note that the average number of records for the pay items excluded from the analysis was only 7.8 over the examination period. The fourth step was to exclude pay items that were not used during the critical 2008 time period when fuel prices were fluctuating. The analysis examined the first bid date and the last bid date for each pay item to remove all pay items with less than a 3-year range of data. This action removed less than 400 pay items. The final database contained approximately 1.8 million records providing data on 5,965 pay items. The mean number of bids per pay item was approximately 284. The database included state, pay item number, pay item description, unit, quantity, amount (in dollars per unit), a category identifier developed by Oman Systems, and the bid date. 2.4.2 Tabulation of Diesel Fuel Price Index The second step was to tabulate price indices over the same period. Highway construction projects are known to use large amounts of diesel fuel for equipment use. Diesel fuel prices also serve as a surrogate for the price of other petroleum-product-based inputs to highway construction Options Number of Pay Items Number of Bids Low Bids Only 335,564 3,597,517 Also Exclude Non-Standard Pay Items 171,381 3,289,606 Also Exclude Lump Sum Pay Items 157,407 2,973,784 Also Exclude Pay Items Bid Less than 100 Times 6,338 1,799,740 Also Exclude Pay Items with Less than 3 Years of Data 5,965 1,723,059 Exhibit 2-40. Number of pay items and bid lettings from 1/1/2006 to 9/1/2010.

46 Fuel Usage Factors in highway and Bridge Construction such as asphalt, paint, and sealers. In addition, many other inputs to highway construction such as concrete and steel have high fuel-use input requirements and high transportation costs to the work site. The available data indicate that fuel costs have become a more important component of construction costs in general. For example, in 1998, the Bureau of Economic Analysis (BEA) national input-output matrix had a total input requirement coefficient of 0.029 (2.9 cents per dollar output) for petroleum and coal-based products used by the construction industry as a whole. By 2008, total input requirement of petroleum and coal-based products had increased by almost a factor of three to 0.083 (8.3 cents per dollar output). Given the reduction in fuel costs since 2008, it is likely that the 2010 BEA benchmark revision will have a smaller coefficient. Fuel is an important input to highway construction activities, and petroleum products represent a relatively larger proportion of the total costs of production than for the construction industry in general. Although fuel and petroleum-based products are important components of production costs for the construction industry, when viewed on a total requirements basis, the direct input requirement of fuel and petroleum-based products is somewhat smaller (2.2 cents per dollar output in 1998 and 6.1 cents per dollar output in 2008). The lower level of direct costs is due to the relatively large embodied energy content of other input materials such as concrete and steel. Given these figures, and applying an extra factor of two to account for the fact that highway construction is more fuel-intensive than construction in general, suggests that the direct cost of fuel and petroleum-based products represented somewhere between 2 percent and 5 percent of production costs for highway construction in 1998, rising perhaps to a range from 6 percent to 12 percent in 2008, and falling since then to somewhere below 10 percent of production costs. To put these costs in perspective, employee compensation costs in the con- struction industry rose from 30 cents per dollar output to 35 cents per dollar output between 1998 and 2008. Despite the relatively small (10 percent) share of direct fuel costs per dollar output, diesel fuel prices have ranged both up and down by over a factor of three from 2004 to 2010. It is reasonable to expect that this large variation in diesel fuel price and other petroleum-based product prices has had a measurable impact on the bid prices received for highway construction projects. The statistical analysis attempted to find empirical evidence of this relationship. The daily U.S. No. 2 diesel fuel price (cents per gallon) was calculated as the arithmetic average of three regional price indices published by the U.S. Department of Energy. The three regional price series represent daily market closing prices in the New York area (New York Harbor No. 2 Diesel Low Sulfur Spot Price FOB), the Gulf Coast (U.S. Gulf Coast No. 2 Diesel Low Sulfur Spot Price FOB), and Los Angeles (Los Angeles, California, No. 2 Diesel Spot Price FOB). The published data do not include prices for weekends or U.S. oil market trading holidays. The research team imputed the last available price in these instances. For example, prices on a Monday holiday would match the last available spot prices from the previous Friday. 2.4.3 Assessment of BidTabs Data The highway construction bid database includes over 3 million records with information on bids submitted in the 48 states. Each pay item has a unique definition within each state and is provided in terms of specific units. Variables in the bid price database include the quantity and unit bid prices, and the bid date for each pay item. Dates range from January 1, 2006, to September 1, 2010. Since the project was interested in the impact of the large diesel fuel price swings, the first bid date and the last bid date for each pay item were examined to remove all pay items with less than a 3-year range of data. The 3-year requirement ensures that the range of

Initial research and research approach 47 bid dates for the pay item includes the critical 2008 time period. Pay items receiving less than 100 bids were also removed. The resulting database included 5,965 pay items. The pay items were categorized into 29 summary categories used in the BidTabs database. A partial correlation analysis was run for each pay item within each category. The partial correlation of the bid price with the diesel fuel price and the significance level of the estimated partial correlation coefficient were then summarized by category. The categories with the 10 largest mean partial correlations coefficients are displayed in Exhibit 2-41. The overall conclusion of the initial statistical analysis is that there is a positive relationship between fuel prices and bid prices. The positive relationship is strongest where the significance of the correlation is strongest. However, there is a large amount of variation in the results for individual pay items within the categories of construction. The negative coefficients indicate the fuel price is not always an important factor for determining bid prices for many types of purchases. Further analysis is needed to determine why this is the case. It may be concluded that fuel con- sumption is significant in most types of highway construction, but perhaps is not limited to only certain construction activities, as previous studies have suggested. A major goal of the initial analysis was to identify construction tasks that consume large amounts of fuel and are fuel intensive. These items would be obvious candidates for newly calculated fuel factors. The initial statistical analysis indicated that a larger number of activities than previously envisioned are heavy users of fuel and/or are fuel intensive. Many heavy construction tasks, such as asphalt paving and grading, were confirmed as being heavy users of fuel. However, additional items appear to be more fuel intensive than anticipated. For example, the roadway lighting/ electrical and signalization categories ranked second and third in the initial statistical analysis. Those categories did not rank within the top 10 of the other initial methodologies. 2.5 The Three-Pronged Research Methodology The first phase of the study examined three strategies for developing fuel usage factors. The study team examined the strengths and weaknesses of each approach in preparation for the second data collection phase of the project. This report section describes observations and lessons learned during the first phase, assesses the strengths and weaknesses of each approach, and outlines the research approach that was ultimately used in collecting the data and developing the fuel usage factors. Rank Category Mean Correlation 1 Concrete - Culverts 0.099 2 Roadway Lighting/Electrical 0.092 3 Signalization 0.078 4 Retaining Wall 0.069 5 Bridge 0.062 6 Guard Rail 0.058 7 Drainage - Pipe 0.052 8 Underdrain 0.050 9 Concrete – Misc. 0.045 10 Buildings/Misc. Structures 0.045 Exhibit 2-41. Ten largest mean partial correlation coefficients by category and significance level.

48 Fuel Usage Factors in highway and Bridge Construction 2.5.1 Issues in Developing Fuel Usage Factors This subsection discusses several issues that the study team encountered during the first phase of the project. The first is a discussion of the economic production function of construction activities and how it relates to the observed fuel intensity rankings. The following section addresses the number of updated and additional fuel factors. The last section discusses fuel factors for bridges and structures. Understanding the Production Function An underlying assumption in the literature and in the application of fuel factors by state DOTs is that certain construction activities, such as grading and paving, are more fuel intensive than many other activities. Both state DOT officials and contractors share this perception, as indicated by the mutual preference for selecting grading/excavation and asphalt pavement as the most fuel intensive activities. However, the statistical analysis performed during the analysis of the BidTabs database indicated that a larger variety of activities might have significant fuel use. In fact, the statistical analysis found significant correlations between bid price and fuel prices for a large variety of construction activities. Exhibit 2-42 displays the fuel intensity rankings determined by each research method. It also displays the fuel percent of cost rankings from Attachment 3 of the original Technical Advisory T5080.3. Note that several work categories that rank in the top 10 for 3 of the research efforts and the Attachment 3 rankings (notably grading/excavation, asphalt paving, and base stone/aggregates) do not appear in the BidTabs statistical analysis rankings, while the BidTabs statistical analysis contains items that have historically been thought of as less fuel intensive. A potential reason for this apparent contradiction is that the focus is often on total fuel use and the dichotomy between heavy and light construction activities. An alternative is to focus on the full economic production function. In economics, factors of production (or productive inputs or resources) are any commodities or services used to produce goods and services. “Factors of production” may also refer specifically to the primary factors, which are stocks including land, labor (the ability to work), and capital goods applied to production. For example, in productivity analysis, the U.S. Bureau of Labor Statistics defines the production function as the combination of capital (K), labor (L), energy (E), materials (M), and purchased business services (S) inputs, or KLEMS inputs. DOT Survey Contractor Survey Estimating Analysis BidTabs Statistical Analysis 1 Grading/Excavation Asphalt Paving Grading/Excavation Grading/Excavation Concrete - Culverts 2 Asphalt Paving Grading/Excavation Asphalt Paving Clearing Roadway Lighting/Electrical 3 Base Stone/Aggregates Base Stone/Aggregates Base Stone/Aggregates Mobilization Signalization 4 Concrete Paving Concrete Paving Concrete Paving Base Stone/Aggregates Retaining Wall 5 Bridges/Structures Drainage - General Drainage - General Misc. Stone/Riprap Bridges/Structures 6 Landscaping Bridges/Structures Bridges/Structures Concrete Paving Guard Rail 7 Roadway Lighting/Electrical Asphalt Paving Drainage - Pipe 8 Deck Repair/Minor Widening Equipment/Labor Underdrain 9 Striping/Pavement Mark Underdrain Concrete - Misc. 10 Bridges/Structures Buildings/Misc. Structures Attachment 3 from TA5080.3Rank Research Method Exhibit 2-42. Rankings of fuel use by activity.

Initial research and research approach 49 For example, most observers would characterize asphalt paving as a fuel intensive activity and pavement marking as a non-fuel intensive activity. In the case of asphalt paving, the equipment used is much heavier and has a higher fuel consumption rate. However, when examining the whole production function, asphalt paving also requires more capital, labor, and materials. In particular, although liquid asphalt is a petroleum product, it is not fuel and its consumption is not part of the fuel factor. In the case of pavement marking, the equipment may only be one light vehicle with a low fuel consumption rate. However, if there is only one driver and very little material cost, fuel cost as a percentage of total cost may actually rival or exceed the fuel cost percentage for asphalt paving. Pay Item Flexibility The flexibility to alter the list of items or add new items varies considerably among the three study methodologies. Including a lengthy list of items in the contractor survey would reduce response rates and increase processing time and costs. Once a survey is distributed, it becomes infeasible to add new or additional items. In contrast, changes or additions to the list of items considered in the engineering or statistical analyses can be accomplished with relative ease. For this reason, the pro- posed methodology for the survey included a flexible additional factor survey section. Contractors were allowed to write in non-traditional items that they believed to be fuel intensive. An important consideration was the analytical needs of the ultimate users. Different users have different priorities and preferences. For example, while only five state DOTs stated that they would prefer additional fuel use factors, nearly 57 percent of the contractors in the initial survey recommended either adding additional fuel factors or consulting with the construction industry to select new fuel factors. Accordingly, the methodology envisioned two levels of detail in the final fuel factors. The basic product is a hard copy table containing a limited number of fuel factors, including the items commonly used by state DOTs in price escalation clauses. The more detailed Excel spreadsheet tool allows the user to access additional, and more detailed, fuel factors. Structures Fuel factors for structures presented a particular concern because the current factors are on a unit consumption per $1,000 of work basis. Therefore, as fuel and other input prices vary, the measure can become unreliable, especially over time. The study team envisioned two method- ological alternatives to address this problem. The first option was to include links and information regarding price indices within the software tool. These indices allow the user to update the fuel factors to address the effects of cost inflation. The second option was to tabulate fuel factors on a gallons per unit basis. The study team ultimately included both of these options. 2.5.2 Assessment of the Survey Approach As part of the first phase of the study, the team conducted an initial survey of contractors to assess their ability and willingness to provide data on fuel usage estimates for bid items. In addition to questions concerning fuel factor administration and fuel intensity perceptions, the initial survey invited suggestions for improving the response rate for the more comprehensive survey in the second project phase. Strengths of the Survey Approach The survey approach relied on information gathered directly from the contractors who perform construction activities. The original fuel factors research was also conducted in this manner. Contractor survey results are easily understood among a non-technical audience.

50 Fuel Usage Factors in highway and Bridge Construction Shortcomings of the Survey Approach Similar to the engineering estimation approach, a contractor survey has the potential to be influenced by responder biases. Contractors may allow their own experiences with fuel factors and fuel costs, whether they are positive or negative, to influence their responses. The legitimacy of data obtained using this method is dependent on a satisfactory response rate as well as a rep- resentative sample. A full contractor survey is also an expensive undertaking. Recommendations, Modifications, and Final Methodology The study team estimated fuel usage directly from a survey of contractors. In the first project phase, the team conducted an initial survey of contractors to assess their ability and willingness to provide data on fuel usage estimates for bid items. Based on the findings and lessons learned from that task, the study team developed a survey of fuel usage. Oman Systems maintains a Contact Management System that collects information on all of the firms bidding highway projects. The study team utilized this database to develop a list of firms to survey. A stated goal for this project was to exceed the 13 percent response rate achieved in the original contractor survey, the results of which were published in Technical Advisory T5080.3. The initial Task 6 contractor survey matched this response rate. However, this required substantial follow-up efforts such as phone calls. For the larger second phase survey, the study team made a concerted, multi-pronged effort to maximize the response rate. As suggested by several contractors in the initial survey, the study team targeted estimators and other personnel with knowledge of their firm’s construction costs. Additionally, the study team attempted to design a survey that was as brief as possible while still being able to capture the needed data. The study team also contacted key industry associations to elicit their support. 2.5.3 Assessment of the Engineering Estimating Approach For the initial engineering estimating analysis, an expert panel of four construction profes- sionals rated the fuel consumption of 31 work categories and over 1,000 individual pay items. The data used for this effort came from Oman Systems’ BidTabs database. In order to create a reasonable number of items for analysis, the research team created several parameters for exclusion, including discarding lump-sum and non-standard pay items, pay items that were bid less than 100 times, and pay items lacking sufficient data for the targeted 4.5-year time period. These efforts resulted in a data set comprising 1,176 unique pay items from states nationwide. The expert panel then created a 1 to 5 fuel use scale with a “5” ranking indicating heavy fuel use. Each of the 1,176 pay items and 31 work categories was then issued a fuel use ranking informed by this scale. Strengths of the Engineering Estimating Approach The engineering estimating approach employed a methodology that is transparent to the user. The methods of the ranking of fuel use are clearly described and easily understood. The process of developing estimates of fuel use, which was based on types of equipment, consumption rates and work rates, is also a method that laypeople and engineers will readily understand. Items can be added relatively quickly and inexpensively. Shortcomings of the Engineering Estimating Approach The relative ranking of fuel use can be a subjective exercise. As witnessed in the first phase, equally qualified estimators assigned different rankings for the same pay item. For example, out of the 31 summary work categories, the four members of the estimating team assigned identical rankings for only three categories (grading/excavation, bridge, and alternates/bonus/time). The

Initial research and research approach 51 engineering estimation of fuel use was also subject to this limitation. Updating the entire set of engineering estimates requires new estimates of equipment consumption rates and work rates, a medium cost activity that is also relatively time consuming. Recommendations, Modifications, and Final Methodology In consultation with the NCHRP project panel, it was agreed that the research team would estimate fuel usage using engineering cost estimation software. This would require the collection of fuel consumption factors for specific pieces of equipment, the assigning of crews (composed of labor, equipment, material), and crew production rates. 2.5.4 Assessment of the Statistical Approach The initial statistical analysis of the BidTabs database examined the relationship between fuel prices and bid prices. Pay items whose prices correlate with fluctuating fuel prices would be likely to be fuel intensive and could be considered for inclusion in a price adjustment clause program. The BidTabs analysis used the same parameters for exclusion as the estimating analysis, as well as the additional caveat that price information must have been available for the periods of rapid fuel price fluctuation in 2008. Strengths of the Statistical Approach The BidTabs statistical analysis had several potential strengths. One strength was that the analysis uses an objective assessment of the correlation between fuel prices and bid prices based on historical data. The analysis did not rely on subjective judgment to select items. This method also had the advantage that the analysis could be replicated in future years to update results. Since this method was based on data that is collected on an ongoing basis, it would preclude the need for future surveys or data collection. Shortcomings of the Statistical Approach Statistical analysis is a complex tool that is often difficult to explain to the layperson. Statisti- cal analysis may not always provide the expected result in every case. In some instances, analyses may be subject to problems such as multicollinearity, where several important variables are also correlated so that only one can be included in the analysis. The analyses could also be subject to confounding variables and could produce unexpected results. The initial statistical analysis did not clearly illustrate whether or not a statistical analysis could produce direct estimation of fuel use. Recommendations, Modifications, and Final Methodology In consultation with the study panel, the study team decided to include the statistical analy- sis in the data collection phase. This included a specification of the model, testing of different combinations and forms of the variables, exploration of lagged variables, evaluation of residuals and error terms, and exploration of different combinations of pay items both within and across states. The study team integrated the KLEMS model into the analysis. The final analysis was designed to produce correlation coefficients that would indicate fuel use. 2.5.5 Overview of the Final Research Approach Exhibit 2-43 presents the final methodology utilized in the project from the initial scoping efforts to the development of the final fuel usage factors. This methodology is presented in flow chart form and indicates the sequential steps undertaken for each of the three methodologies,

52 Fuel Usage Factors in highway and Bridge Construction as well as areas where the methodologies intersected with, and complemented, each other. The survey approach provided much of the data used in formulating the new factors. The engineering approach confirmed the survey data and provided additional detail when the survey approach did not garner sufficient observations for particular work items. More particular details regarding the step-by-step process employed for each methodology may be found in Chapter 3, Findings and Applications, and Chapter 4, Conclusions, Recommendations, and Future Research. S It C C A Estimating Pa Ranking of F Use elect BidTab ems/Paramet Exclusion Calculate Cr ompile Equi Fuel Usage R alculate Fue per Unit by T djust Engine Results base Contracto Response nel uel s Pay ers of U ews pment ates l Use ask ering d on r s Ta Compare I se Categorie Query D Cont Secure Coop Deve Disse SurveyM Excel Remin Study T C Tabula and D Inaccura bulate Final nitial Results s and Pay Ite OTs and ractors Industry eration lop and minate onkey and Surveys ders and eam Phone alls te Results iscard te Figures Fuel Usage F ms from Bid Pa U In F actors Tabs Databa Estimate Me rtial Coeffic sing Diesel P Index tegrate KLE Model Develop Die Price Index Calculate Me uel Coefficie se an ients rice EM sel an nts Engineerin Estimatio g n Contra Usage ctor Fuel Surveys BidTabs Statistical Analysis Exhibit 2-43. Project methodology flow chart.