National Academies Press: OpenBook
« Previous: Chapter 2 - Background and Literature Search
Page 18
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 18
Page 19
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 19
Page 20
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 20
Page 21
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 21
Page 22
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 22
Page 23
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 23
Page 24
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 24
Page 25
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 25
Page 26
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 26
Page 27
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 27
Page 28
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 28
Page 29
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 29
Page 30
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 30
Page 31
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 31
Page 32
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 32
Page 33
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 33
Page 34
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 34
Page 35
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 35
Page 36
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 36
Page 37
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 37
Page 38
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 38
Page 39
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 39
Page 40
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 40
Page 41
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 41
Page 42
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 42
Page 43
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 43
Page 44
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 44
Page 45
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 45
Page 46
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 46
Page 47
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 47
Page 48
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 48
Page 49
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 49
Page 50
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 50
Page 51
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 51
Page 52
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 52
Page 53
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 53
Page 54
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 54
Page 55
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 55
Page 56
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 56
Page 57
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 57
Page 58
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 58
Page 59
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 59
Page 60
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 60
Page 61
Suggested Citation:"Chapter 3 - Research Results." National Academies of Sciences, Engineering, and Medicine. 2004. Optimal Timing of Pavement Preventive Maintenance Treatment Applications. Washington, DC: The National Academies Press. doi: 10.17226/13772.
×
Page 61

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 3 RESEARCH RESULTS INTRODUCTION There is a need to identify when it is “best” to apply pre- ventive maintenance treatments. Treatment performance is greatly dependent on the condition of the pavement at the time of treatment application, and different types of treat- ments are likely only to be effective when placed at certain times in a pavement’s life. When placed at the right time, a preventive maintenance treatment becomes a cost-effective means of attaining the desired life and performance of the pavement. Treatments applied too soon add little benefit and treatments applied too late are ineffective; however, there is little guidance available on this topic. There are no studies that have successfully determined how to identify the optimal time to apply preventive mainte- nance treatments; although a number of completed studies have examined this issue and other research continues to study it. These include the studies of maintenance effective- ness under the Specific Pavement Studies (SPS-3 and SPS-4) effort (17, 18), and field studies by the DOTs in Iowa (9), Ari- zona (19), Texas (20, 21), and South Dakota (22). The primary objective of this project was to determine an approach for identifying the optimal timing for the application of preventive maintenance treatments. This chapter describes a methodology for determining the optimal time to apply pre- ventive maintenance by analyzing pavement performance and cost data. The methodology is presented as a Microsoft® Excel-based software designated OPTime. The results of an evaluation of OPTime (and the analysis methodology) with data provided by state highway agencies is also described. INTRODUCTION TO THE METHODOLOGY USED TO DETERMINE OPTIMAL TIMING One of the initial challenges in this project was to attach some physical meaning to “optimal” timing in the context of preventive maintenance treatment applications. It could potentially mean to provide the smoothest ride for the least money, to prolong the need for rehabilitation, or to meet some other objective. While the concept of “optimal” tim- ing seems closely linked to cost-effectiveness, the defini- tion of cost-effectiveness also varies among agencies. Ulti- mately, a methodology very similar to the cost-effectiveness analyses used in pavement management systems was selected. Overview of the Analysis Approach The approach is built on a number of fundamental con- cepts. It assesses the effectiveness of a particular preventive maintenance application in terms of both the benefit it pro- vides and the cost required to obtain that benefit. In this methodology, benefit is defined as the quantitative influence on pavement performance as measured by one or more con- dition indicators. Costs that may be included in the analysis include the following: • The agency cost to construct the treatment, • Work zone-related user delay costs, • The cost of a rehabilitation activity that would be con- sidered at the point when the preventive maintenance treatment is considered failed, and • The cost of scheduled routine maintenance. In the optimal timing methodology, the benefits associated with the use of a preventive maintenance treatment are eval- uated in conjunction with its associated costs. The optimal application of a preventive maintenance treatment occurs at the point at which the benefit per unit cost is greatest. Pavement Performance The computation of the benefit associated with an applied preventive maintenance treatment requires knowledge of the anticipated performance of the pavement. The effect of a treat- ment on performance is determined by the change in condition indicators, such as International Roughness Index (IRI), pres- ent serviceability index (PSI), or other custom-defined mea- sure of performance. Condition Indicators The ability of treatment to preserve pavement condition and retard future deterioration is measured by changes in the con- dition indicators that define pavement performance. Condition

19 indicators used in the optimal timing methodology should have the following characteristics: • Be measurable (able to be tracked over time), • Indicate pavement performance (especially functional performance for preventive maintenance), and • Change value following the application of a preventive maintenance treatment. Condition monitoring data are needed for all condition indi- cators that are used in the analysis; the methodology permits the analysis of multiple condition indicators. Do-Nothing Relationships The benefit associated with the application of a preven- tive maintenance treatment at any given time is based on the improvement in condition compared with that for the “do- nothing” alternative. The do-nothing alternative defines the performance over time (in terms of the condition indicator) that would be expected if only minor routine maintenance were conducted. In a plot of pavement condition versus time, the baseline performance relationship is referred to as a do-nothing curve. If benefit is defined in terms of multiple distress types, a do-nothing performance curve is required for each relevant condition indicator. The best source for this information is existing pavement management systems, although the necessary relationships can also be approxi- mated without the assistance of a pavement management database. Post-Treatment Relationships Determining optimal timing also requires an understanding of how performance is changed once the preventive mainte- nance treatment has been applied. A separate performance relationship (condition versus age) is needed for each unique combination of condition indicator and treatment application age; it is generally assumed that this relationship changes depending on when the treatment is applied. For example, if performance is measured by 3 indicators for a treatment applied at 5 ages, 15 (3 × 5) different performance relation- ships must be defined. Benefit Associated with Individual Condition Indicators Benefit is the quantitative influence on condition indicators resulting from the application of a preventive maintenance treatment. Using this definition, different types of benefit may be associated with an application of a given preventive main- tenance treatment. For example, applying a chip seal could result in benefits in the form of improved friction, retarded oxidation, or reduced rutting. For a specific condition indi- cator, the benefit is determined by the difference in computed areas associated with the post-treatment condition indicator curve and the do-nothing curve. For condition indicators that decrease over time (e.g., serviceability, friction, or a typical composite index) the area under the curve becomes relevant to benefit computations, while for condition indicators that increase over time (e.g., roughness, cracking, rutting, fault- ing, and spalling) the area above the curve becomes relevant. Figure 1 illustrates the benefit resulting from the application of a preventive maintenance treatment. As shown in the fig- ure, a defined lower benefit cutoff value limits the areas. The benefit (difference in areas) is generally positive, as a preventive maintenance treatment should improve condition or extend the time until the pavement needs rehabilitation; however, negative benefits may result (e.g., the decrease in friction that follows the application of a fog seal). As different condition indicators are expressed in different units, the methodology normalizes all individual condition indicator benefit values by dividing the benefit area by the original do-nothing area. The result is that all individual ben- efit values are similarly expressed in units of percent. For example, if the do-nothing and benefit areas in Figure 1 are calculated to be 30 and 12, respectively, the individual ben- efit value associated with the condition indicator would be 12/30 = 0.4, or 40 percent. Benefit Weighting Factors When more than one condition indicator is included in the analysis, a method is needed to combine the individual ben- efit values associated with the different indicators. This is done by using benefit weighting factors and a normalization process. Computation of Overall Benefit Benefit weighting factors are used to differentially weight the computed individual benefits associated with each included Age, years Co nd iti on In di ca to r Lower Benefit Cutoff value Do-Nothing Area Benefit Area Figure 1. Conceptual illustration of the do-nothing and benefit areas.

20 condition indicator. Each condition indicator is assigned an integer weighting factor between 0 and 100, where all the entered weighting factors must total 100 for a given analy- sis. Each chosen weighting factor is then converted to an associated weighting percentage by dividing each individ- ual weighting factor by 100 (i.e., the total of all assigned benefit weighting factors). The individual contributions to the overall benefit are then determined by multiplying the bene- fit weighting factor percentages by the individual benefit val- ues. This approach is explained with the following example. Assume that a particular preventive maintenance treatment timing results in individual benefit values of 27 percent for rutting, 12 percent for cracking, and 47 percent for friction. That is, the preventive maintenance treatment application increases performance by 27, 12, and 47 percent over the do- nothing benefit area performances for rutting, cracking, and friction, respectively. Next, assume that the agency chooses benefit weighting factors of 60, 30, and 10 for rutting, crack- ing, and friction, respectively (note that these factors add up to 100). Overall benefit contributions are then determined by multiplying the benefit weighting factor percentages by the individual benefit values (e.g., for rutting 27 percent × 60 = 16.2 percent). The total overall benefit contribution is then the total of those values calculated for each individual con- dition indicator. In this example, the total overall benefit con- tribution is 24.5 percent (see Table 16). The total benefit val- ues computed for different timing scenarios are then used in combination with costs to compare the effectiveness of the different timing scenarios. Selecting Benefit Weighting Factor Values Selecting benefit weighting factors that correctly represent the relative importance of different condition indicators is a difficult task. Because each condition indicator is expressed in different units, an incremental change in the magnitude of one indicator does not necessarily provide the same effect as an incremental change of equal magnitude in another condi- tion indicator. For example, a 10 percent increase in the area (benefit) associated with roughness is not likely to have the same impact on performance as a 10 percent increase in the friction area. Although the selection of benefit weighting fac- tors is a subjective process that requires engineering judg- ment, an investigation can be conducted to provide feedback on multiple condition indicators used in the analysis. Some general steps that can be followed to gather feedback for use in the factor selection process are described in this section. Initial Selection of Benefit Weighting Factors. Engineering judgment is a good starting point in the process of selecting relative weights associated with each performance measure. The initially selected weights represent attempts to quantify the relative purpose or benefit of applying the treatment. For example, if the use of a slurry seal is proposed to reduce or eliminate a historical problem with raveling and low friction characteristics, and if the agency feels that the problems with raveling and low friction are of equal importance, then initial benefit weighting factors of 50 would be appropriate for both. However, if the preventive pavement program is pri- marily being driven by a desire to improve friction charac- teristics, this difference in purpose may be reflected by assign- ing a much larger factor to friction (e.g., 80 for friction and 20 for raveling). Analyze Each Condition Indicator Separately. The initial selection of benefit weighting can be improved by investi- gating the sensitivity of the results. This can be accomplished by analyzing the effects of one condition indicator at a time (set the associated benefit weighting factor for one of the condition indicators to 100 and all other benefit weighting factors to 0). The effects on treatment timing can then be interpreted to identify the condition indicators that are rela- tively more important than others. To demonstrate the importance of the benefit weighting process, assume an individual analysis of three different con- dition indicators (rutting, cracking, and friction). When the optimal timing results are relatively similar (e.g., 3, 4, and 4 years of age, respectively), the weighting process is less important than if the optimal treatment times are substan- tially different. The weighting process will be completed by considering a weighted average of the benefits associated with each condition indicator (the overall optimal timing will still be in a range from 3 to 4 years). However, if the individ- ual analysis results show a wider range of optimal timings (e.g., 4, 2, and 7 years, respectively), the effect of assigned weighting factors on the final optimal timing cannot be easily assessed. In such cases, investigations similar to that described Condition Indicator Individual Benefit Values, % Assigned Benefit Weighting Factor Benefit Weighting Factor Percentage Overall Benefit Contribution, % Rutting 27 60 60/100 = 0.6 16.2 Cracking 12 30 30/100 = 0.3 3.6 Friction 47 10 10/100 = 0.1 4.7 TOTAL — 100 1.0 24.5 TABLE 16 Example computation of overall benefit

21 in the following subsection are needed to determine appro- priate weighting factors. Trials of Different Combinations of Benefit Weighting Factors. Useful feedback may be obtained by conducting a series of analyses in which different combinations of weight- ing factors are investigated. For example, the selection of ini- tial weighting factors as the baseline (e.g., 60 for rutting, 30 for cracking, and 10 for friction) indicates that controlling rutting appears to be the most important purpose of the treatment, and the overall optimal timing is thus likely to be closer to the age associated with the individual analysis in which only rutting was considered (i.e., 4 years) than to the ages associated with the other condition indicators (i.e., 2 or 7 years). Conducting a simplified sensitivity analysis of different combinations of benefit weighting factors should provide information to support the initial choices for weighting fac- tors or to help make appropriate changes to the individual factors. Continuing with the example above, assume that the results summarized in Table 17 are obtained by conducting a series of targeted analyses. A simultaneous interpretation of such “what if” scenario results—combined with the optimal timings estimated by conducting the separate condition indi- cator analyses—should provide a good indication of the ben- efit weighting factors that are best for a specific analysis. Because the process for determining appropriate benefit weighting factors is very similar to that used by agencies to develop composite distress indices, many agencies may already have processes that can be adapted. Regardless of the method used to select the weighting factors, it is recommended that an agency regularly review the selected factors whenever additional (or more accurate) performance data become avail- able (i.e., whenever performance relationships are updated). This review will greatly increase the chances of obtaining more accurate analysis results. Cost Considerations The second fundamental aspect of the proposed method- ology is the inclusion of costs that are impacted by the appli- cation of preventive maintenance activities. The current meth- odology allows the user to consider preventive maintenance treatment costs (agency costs), rehabilitation costs, work zone- related user delay costs, and other routine maintenance costs. The user can select one or more of these available cost types to include in an analysis. The details associated with each of these cost types are described as follows. Treatment Costs Treatment costs include all agency costs associated with the placement of a preventive maintenance treatment. These include design, mobilization, materials, construction, and traf- fic control costs. Although the analysis methodology allows these costs to be omitted, it is highly recommended that treat- ment costs be included in any analysis. Rehabilitation Costs Because the application of preventive maintenance is expected to prolong the need for major rehabilitation, the inclusion of rehabilitation costs is an option in the analysis approach. As the cost of a required rehabilitation activity can be large in relation to the cost of a preventive maintenance treatment, the timing of the expected rehabilitation activity can have a significant impact on a pavement’s overall life- cycle cost (LCC). Work Zone-Related User Delay Costs The methodology considers only user costs associated with work zone delays (i.e., the cumulative delay cost recognized by all users subjected to the work zone during construction of the treatment). This approach favors treatments that pro- vide some benefit but can be placed comparatively quickly with little disruption to the traveling public. The methodology does not include other common types of condition-sensitive user costs (e.g., vehicle operating costs, discomfort, and acci- dent costs) because the difference in pavement condition for preventive maintenance candidates is expected to be rela- tively small. The cumulative delay cost is computed as a function of the average number of vehicles per day (AADT), work zone dura- tion, average vehicle delay time, and cost per delay time per vehicle. Rutting Weighting Factor Cracking Weighting Factor Friction Weighting Factor Resulting Estimated Optimal Timing, age 60 30 10 5 60 25 15 6 60 35 5 4 70 20 10 4 50 40 10 3 TABLE 17 Example of optimal timings resulting from conducting a series of analysis sessions with different benefit weighting factors

22 In general terms, user costs are defined as non-agency costs that are borne by the users of a pavement facility, and typically consist of the following: • Vehicle operating costs (VOC)—Costs induced because of increased wear and tear on a vehicle when using a pavement (because of stopping/starting or excessive pavement roughness) during normal operations. Normal operations are periods in which a pavement facility is free of construction, maintenance, or rehabilitation activ- ities that would otherwise affect the capacity. Costs in this category are generally related to pavement rough- ness, and therefore do not begin to accrue until after the pavement has reached a higher level of roughness (e.g., an IRI of about 2.7 m/km [170 in./mi], or a present ser- viceability rating [PSR] of about 2.5). • User delay costs—Additional costs caused by time delays in traveling over a pavement facility as a result of the following: – Reduced speed to enter the work zone (or even a com- plete stop, if there is queuing), – Reduced speed through the work zone; and – Use of alternate routes to avoid the work zone. • Crash costs—Costs associated with fatalities or injuries that result from crashes on a pavement facility. The inclusion of user costs as part of a life-cycle cost analy- sis (LCCA) is a controversial issue. While there is general agreement that traffic delays increase user costs, the actual costs are difficult to quantify and they are not costs borne directly by the highway agency. When user costs are included in an analysis, they often overwhelm the direct agency costs, particularly for high-volume facilities. Some highway agen- cies choose not to include user costs in an LCCA while oth- ers choose to compute direct costs and user costs separately and include user costs as an additional evaluation criteria when evaluating competing construction bids (often referred to as A + B contracts). For most pavement facilities in fair or good condition (e.g., pavements with a PSR of 2.5 or greater), user costs dur- ing normal operations are minimal; consequently, the user costs associated with the placement of work zones for pave- ment maintenance or rehabilitation activities are of the great- est concern in this project. Of the three types of user costs, only user delay costs are incorporated into the optimal tim- ing methodology because they are generally significantly larger than the vehicle operating costs or the crash costs; there is a dearth of statistical data to support crash rates in work zones, and there is controversy associated with crash cost rates. Estimating User Delay Costs. The 1998 FHWA report, Life- Cycle Cost Analysis in Pavement Design (23), outlines the steps for estimating work zone user delay costs. The process requires at least the following information: • General project inputs (e.g., project length, number of lanes); • Traffic data (e.g., 2-way average daily traffic [ADT], directional split, hourly traffic distribution); • Work zone closure data (e.g., time period(s) in which the closure is in place, duration of the work zone closure, number of available lanes, posted and work zone speed limits, vehicle capacity, queue dissipation rates); and • Value of time delay costs (for passenger, single unit, and commercial vehicles). With these inputs, the movement of vehicles through the work zone can be analyzed, yielding information on user delay times (including delays because of the possible development of queues) which can be converted to user delay costs. The analysis can be conducted by several computer programs (e.g., MicroBENCOST, QueWZ). It is somewhat complex and requires inputs that may not be readily available for the analysis of most projects. Consequently, a more simplified procedure is needed for this methodology. To include user costs in the analysis, an easy method is needed to compute the time delays for vehicles traveling through the work zone. If this value is estimated, then the entire calculation process becomes straightforward. Table 18 lays out the calculation routine for the incorporation of user delay costs, with each of the columns in that table defined following the table. • Column A: The classification of vehicles using the facility (the 1998 FHWA report recommends just three classes: passenger cars, single-unit trucks, combination trucks) (23). • Column B: The approximate number of vehicles in each of three categories that are affected by the work zone; this is largely influenced by the length of time that the work zone is in place (if the work zone is periodic, then it is the vehicles that pass through the zone only during that period). • Column C: The delay cost rate for each vehicle classifica- tion (ranges are provided in the 1998 FHWA report) (23). • Column D: The average additional delay time for each vehicle that is affected by the work zone. This value is estimated for each project, and is strongly related to the physical length of the work zone, the number of lanes that are closed (and the capacity of those that remain open), and whether or not a queue is expected to form. • Column E: The total cost for each vehicle classification, which is the product of column B, column C, and col- umn D (making sure all units are consistent). This simplified process introduces several sources of error. One obvious source of error is the accuracy of the user’s esti- mates. Although these errors may be significant, these esti- mates can be used to make meaningful comparisons of the relative effects. Another source of error arises if the work zone

23 produces a queue that generates considerable delay costs. These costs may not be accurately accounted for in the user’s estimate of the number of vehicles affected by the work zone. However, because work zones associated with most preven- tive maintenance treatments are of relatively short duration and short length, queues are less likely to form and the error associated with this item is reduced. Additional Routine Maintenance Costs Different pavement structures and surfacing approaches require different needs for routine maintenance. These needs are addressed in the methodology as a recurring cost for which the timing is not optimized. An example of such an activity is pothole patching that may influence long-term perfor- mance but does not fit the preventive maintenance model because it is only done once the distress appears (i.e., its tim- ing cannot be optimized). When choosing to include the costs of routine/reactive maintenance activities in an analysis, the do-nothing perfor- mance curves must account for the expected effect of this maintenance on performance. The routine maintenance sched- ule (and costs) must be estimated and included in the analysis. Determination of Optimal Timing The optimal time to apply a treatment is based on an analy- sis of benefit and costs. That application timing that maximizes benefit while minimizing costs (i.e., that with the largest B/C ratio is the most effective timing scenario. To make the actual values of the B/C ratios more meaning- ful, the concept of an Effectiveness Index (EI) is introduced. The EI normalizes all individually computed B/C ratios to a 0 to 100 scale by comparing all B/C ratios with the maxi- mum individual B/C ratio (i.e., the ratio associated with the optimal timing scenario). The maximum individual B/C ratio is assigned an EI of 100, and all other B/C ratios are repre- sented as a fraction of the maximum EI. The EI is computed for each timing scenario using equation 1. (Eq. 1) where: EIi = EI associated with the ith timing scenario (dimensionless). (B/C)i = B/C ratio associated with the ith timing scenario. (B/C)max = Maximum of all of the B/C ratios associated with the different timing scenarios. i = Index associated with the current timing scenario. DETAILED CALCULATION PROCEDURES OF THE ANALYSIS APPROACH This section describes a step-by-step procedure for (1) com- puting benefit and costs within the methodology and (2) using the results to determine the most effective treatment timing. An example is also presented to illustrate the concepts. Step 1: Analysis Session Setup The first step in the optimal timing analysis process is to select the particular treatment and the specific treatment appli- cation ages that will be used in the analysis. EI B C B Ci i = ( ) ( )     ×max 100 Column A Column B Column C Column D Column E Vehicle Classification Total Number of Vehicles Affected By Work Zone1 Delay Cost Rate, $/hr2 Average Additional Delay Time, hour/vehicle3 Total Cost Passenger Cars 15 $10 to $13 Col. B* Col. C* Col. D Single-Unit Trucks 6 $17 to $20 Col. B* Col. C* Col. D Combination Trucks 1 $21 to $24 Col. B* Col. C* Col. D TOTAL (sum column B) (sum column E) 1 Only the number of vehicles in each category affected by the placement of the work zone over its entire duration. 2 From 1998 FHWA LCC report, p. 23 (note: 1996 values) (23). 3 The delay time for each vehicle is estimated on a project by project basis (it is the same for each vehicle category). This includes all delay times associated with the work zone, including speed change delay (going from posted speed limit to work zone speed limit), work zone speed delay (delay associated with slowing down to the work zone speed limit to traverse the work zone), stopping delay (time delay if a queue forms), and queue speed delay (time delay it takes to traverse the queue). TABLE 18 Calculation of user delay cost

24 Treatment Selection Agencies use a broad range of treatments in pavement preservation and rehabilitation programs. However, preven- tive maintenance, which is a subset of these two pavement activity categories, considers treatments that can be applied to a pavement in good condition to preserve condition and pre- vent or delay future deterioration. The treatments shown pre- viously in Table 2 fit this definition of preventive maintenance and will provide benefits when used in the appropriate condi- tions. The user should, however, carefully consider whether these benefits can be measured using available performance evaluation procedures. For example, crack sealing or main- taining drainage features may be a cost-effective means of maintaining or improving pavement condition, but perfor- mance measures such as IRI, cracking indices, rutting, and faulting may reveal only subtle (or even no) differences when compared with control sections. Furthermore, while the analysis method permits users to analyze performance results from any specified treatment, the approach does not work well for treatments applied when the pavement is deteriorated and rehabilitation is required. The current approach allows the analysis of a single applica- tion of one preventive maintenance treatment but not that of a series of preventive maintenance treatments. Selection of Application Ages The optimal time to apply a selected preventive mainte- nance treatment is estimated by conducting analysis for dif- ferent timing scenarios in which the treatment is applied at different pavement ages. Step 2: Selection of Benefit Cutoff Values The concept of optimal timing stipulates that treatments applied too soon or too late do not necessarily provide added benefit. Benefit cutoff values are defined as the y-axis (con- dition indicator) boundary conditions for the performance curves that define the upper and lower limits for the benefit area calculations. The specific definitions of the upper and lower benefit cutoff values are as follows: • Upper benefit cutoff value—The upper benefit cutoff value is the upper limit to the benefit area computations (i.e., no area above the upper benefit cutoff level is included in the benefit computation). For a condition indicator relationship that increases over time (e.g., IRI), this value also serves as the benefit cutoff value that is used in determining the analysis period (i.e., the age at which the performance curve reaches the benefit cut- off value). For a condition indicator relationship that decreases over time (e.g., friction number), the upper benefit cutoff value defines a “ceiling” that limits the benefit credited to the application of the treatment. For example, assume that an agency associates excellent roadway friction with a friction number of 60 (i.e., FN40R = 60). If the application of a particular treatment is found to result in friction numbers greater than 60, the area above the upper benefit cutoff value of 60 would not be included in the benefit calculations. • Lower benefit cutoff value—The lower benefit cutoff value is the lower limit to the benefit area computations (i.e., no area below the lower benefit cutoff level is counted as a benefit). For a condition indicator relation- ship that decreases over time this value also serves as the benefit cutoff value in determining the analysis period (in the same manner as described for the upper benefit cutoff). Figure 2 illustrates how upper and lower benefit cutoff values limit the area calculations for both decreasing and increasing do-nothing condition indicator curves. In this example, the decreasing relationship is limited by both the upper and lower benefit cutoff values and the increasing relationship is limited only by the upper benefit cutoff value. Benefit cutoff values are unique to an agency, and perhaps even to a given project, and their determination is not straight- forward. In general, agencies should consider benefit cutoff values that relate to the following identifiable condition levels: Age, years Co nd iti on In di ca to r Upper benefit cutoff value Lower benefit cutoff value Area limited by benefit cutoff values DECREASING RELATIONSHIP Age, years Co nd iti on In di ca to r Upper benefit cutoff value Lower benefit cutoff value Area limited by benefit cutoff values INCREASING RELATIONSHIP Figure 2. Illustration of the application of upper and lower benefit cutoff values on both decreasing and increasing condition indicators for the do-nothing case.

25 • Pavement failure (rehabilitation trigger)—the condition level at which a major rehabilitation is required. • Treatment failure—the condition level at which a treat- ment is considered failed (i.e., the benefits of the preven- tive maintenance treatment are no longer being realized). If agencies are unsure about how to select these benefit cut- off values, it is recommended that values be set to closely reflect current maintenance and rehabilitation policies. For example, if an agency typically applies only one preventive maintenance treatment in the life of a pavement, then the benefit cutoff value should be equal to the pavement failure level because a major rehabilitation will most likely be the next pavement-related activity. In contrast, if an agency typ- ically applies a second preventive maintenance treatment after the first application reaches a known condition failure level, the benefit cutoff value can be set equal to the treat- ment failure level. Step 3: Computation of Areas Associated with the Do-Nothing Case The third step in the benefit calculation process involves determining the total do-nothing condition curve areas. The individual condition indicator areas are computed by taking integrals of the specific performance equations that define the do-nothing performance curves. The important benefit-related areas are those below condition indicator curves that decrease over time and above condition indicator curves that increase over time. The final do-nothing condition area for a given condition indicator is determined by applying the following area boundary conditions: • Y-axis limits—in the y direction, the pertinent area is bounded by the defined upper and lower benefit cutoff values. • X-axis limits—in the x direction, the pertinent area is bounded by zero on the lower end and the age at which the performance curve intersects the benefit cutoff value on the upper end. The area calculation details differ slightly depending on whether the performance equation is decreasing or increas- ing. Equations 2 and 3 are used to compute the do-nothing benefit-related areas associated with decreasing and increas- ing equations, respectively. These equations are functions of the actual do-nothing mathematical equation and the upper and lower benefit cutoff values. Figure 3 illustrates the total Age, years UBC = Upper benefit cutoff value LBC = Lower benefit cutoff value X0 X2X1 Co nd iti on In di ca to r AREADN-TOT(-) DECREASING RELATIONSHIP INCREASING RELATIONSHIP Co nd iti on In di ca to r Age, years X0 X2X1 UBC = Upper benefit cutoff value LBC = Lower benefit cutoff value AREADN-TOT(+) Figure 3. Total areas associated with decreasing and increasing condition indicators for do-nothing options.

26 do-nothing areas associated with both decreasing and increas- ing do-nothing curves and the different intersection points used to define the x-axis boundary conditions. The total do-nothing condition area associated with a decreasing condition indicator relationship [AREADN-TOT(−)] is computed from the following equation: (Eq. 2) where: EQDN = Equation defining the do-nothing condition indi- cator relationship. UBC = Upper benefit cutoff value associated with the con- dition indicator. LBC = Lower benefit cutoff value associated with the con- dition indicator. X0 = Lower limit to the age range (set to zero). X1 = Age at which the do-nothing curve intersects the upper benefit cutoff value (X1 = 0 if there is no intersection with the UBC). X2 = Age at which the do-nothing curve intersects the lower benefit cutoff value (X2 = 0 if there is no intersection with the LBC). The total do-nothing condition area associated with an increas- ing condition relationship [AREADN-TOT(+)] is computed from the following equation: (Eq. 3) where: EQDN = Equation defining the do-nothing condition indi- cator relationship. UBC = Upper benefit cutoff value associated with the condition indicator. LBC = Lower benefit cutoff value associated with the con- dition indicator. X0 = This lower limit to the age range is set to zero. X1 = Age at which the do-nothing curve intersects the lower benefit cutoff value (X1 = 0 if there is no intersection with the LBC). X2 = Age at which the do-nothing curve intersects the upper benefit cutoff value (X2 = 0 if there is no intersection with the UBC). Figure 4 illustrates the application of these area boundary con- ditions and the resulting bounded areas (i.e., AREADN(FRICTION), AREA UBC EQ LBC EQ DN TOT( ) DN DN − + = −( ) − −( ) ∫ ∫ X X X X 0 2 0 1 AREA EQ LBC EQ UBC DN TOT( ) DN DN − − = −( ) − −( ) ∫ ∫ X X X X 0 2 0 1 AREADN(RUTTING), and AREADN(ROUGHNESS)) for the previously presented example. Step 4: Computation of the Overall Expected Service Life of the Do-Nothing Case While the computed overall expected service life does not influence the do-nothing area or benefit computations, it serves as a baseline for determining the expected extension of life. As the extension of service life is often used as a mea- sure of the success of a preventive maintenance treatment, this computed value is included as part of the analysis out- put. The expected overall service life for the do-nothing con- dition is selected as the earliest age at which one of the con- sidered condition indicator do-nothing relationships reaches its benefit cutoff value (i.e., the upper benefit cutoff value for increasing relationships or the lower benefit cutoff value for decreasing relationships). This definition is based on the assumption that, in practice, a second treatment would most likely be applied when the first of the considered condition indicator performance curves reaches its benefit cutoff value as illustrated in the following example. The first assumption in this example is that the benefit from applying preventive maintenance lies in its improvement in friction, rutting, and IRI. Next is that the indicators reach their respective trigger- ing benefit cutoff values at 15, 14, and 17 years. Therefore, the overall do-nothing curve expected service life for the analysis session is 14 years—the earliest age of all of the triggering conditions. Figure 5 illustrates the process for this determination. Step 5: Computation of Expected Service Life of the Post-Treatment Case The next step in the benefit calculation process is to plot the post-treatment performance relationships for each condition indicator. The expected service life for the post-treatment case (for a given timing scenario) is then determined as the earliest age at which any of the post-treatment condition indicators reaches its benefit cutoff value. Unlike the do-nothing case where the area computations are unbounded in the x-direction, the area computations for the post-treatment case are bounded at this expected post-treatment service life which is also used as the analysis period for the LCC computations. In the previous example, if a preventive maintenance treat- ment is applied at a pavement age of 10 years, the performance curves for friction, rutting, and roughness, as shown in Fig- ure 6, reach their triggering benefit cutoff values at 20, 22, and 24 years, respectively. Therefore, the expected service life (and analysis period) for this timing scenario is 20 years—the earliest age of all the triggering conditions. Thus, areas would only be computed for the x-range between 0 and 20 years.

27 Step 6: Computation of Areas Associated with the Post-Treatment Case The sixth step in the benefit calculation process is deter- mining the important post-treatment condition curve areas that are used to compute benefit. As with the do-nothing condition area calculations, the individual condition indicator areas are computed by taking the integrals of the specific per- formance equations that define the post-treatment performance curves. As mentioned previously, the important benefit-related area is the area below condition indicator curves that decrease over time or above condition indicator curves that increase over time (as shown in Figure 7). The final post-treatment condition area for a given condition indicator is only deter- mined after applying the following area boundary conditions: • Y-axis limits—in the y direction, the pertinent area is bounded by the defined upper and lower benefit cutoff values. • X-axis limits—in the x direction, the pertinent area is bounded by an age of zero on the lower end and the overall determined post-treatment case expected service life (from step 5) on the upper end. Age, years Fr ic tio n nu m be r Upper benefit cutoff value Lower benefit cutoff value 5 10 15 20 R ut tin g Upper benefit cutoff value Lower benefit cutoff value set to zero 5 10 15 20 In te rn at io na l R ou gh ne ss In de x (IR I) Upper benefit cutoff value Lower benefit cutoff value 5 10 15 20 Age, years Age, years AREADN-TOT(FRICTION) AREADN-TOT(RUTTING) AREADN-TOT(ROUGHNESS) Figure 4. Total areas associated with individual condition indicators for the do-nothing options.

28 Figure 7 illustrates the total benefit-related areas associated with both decreasing and increasing post-treatment curves. Also illustrated in Figure 7 are the different intersection points used to define the x-axis boundary conditions required for the different parts of the area-calculation equations. The area calculation details are different, depending on whether the post-treatment performance equation is decreas- ing or increasing. Equations 4 and 5 are used to compute these post-treatment benefit-related areas associated with decreasing and increasing equations, respectively. Both of these equations are functions of the actual post-treatment curve equation and the upper and lower benefit cutoff values. It is important to note that the post-treatment performance equations are expressed in terms of the treatment age rather than the pavement age. For example, for a linear treatment performance equation such as y = mx + b, the x values are treatment age values (i.e., time after treatment application) rather than pavement age values (i.e., time since original con- struction). Therefore, some of the x-axis values associated with computing the area after the treatment application age are adjusted to account for this difference in age (e.g., X4 − XA and X3 − XA in equation 4). Age, years Fr ic tio n nu m be r Upper benefit cutoff value Lower benefit cutoff value 5 10 15 20 R ut tin g Upper benefit cutoff value Lower benefit cutoff value set to zero 5 10 15 20 In te rn at io na l R ou gh ne ss In de x (IR I) Upper benefit cutoff value Lower benefit cutoff value Do-nothing curve intersects benefit cutoff value at 17 years 5 10 15 20 Age, years Age, years Do-nothing curve intersects benefit cutoff value at 14 years Do-nothing curve intersects benefit cutoff value at 15 years This 14-year life is is the shortest of all expected condition indicator lives. The expected service life of the do-nothing condition is, therefore, set to 14 years. Figure 5. Determination of the overall do-nothing condition expected service life.

29 (Eq. 4) AREA EQ LBC EQ UBC EQ LBC EQ UBC PT( ) DN DN PT PT − −( ) −( ) = −( ) − −( ) + −( ) − −( ) ∫ ∫ ∫ ∫ X X X X X X X X A A 0 2 0 1 4 3 0 0 where: AREAPT(−) = Computed post-treatment area associated with a decreasing condition indicator rela- tionship (i.e., area from time zero to the end of the post-treatment analysis period). EQDN = Equation defining the do-nothing condition indicator relationship. EQPT = Equation defining the post-treatment condi- tion indicator relationship (i.e., treatment per- formance curve). Note that the post-treatment equation is a function of the treatment age Age, years 5 10 15 20 25 Fr ic tio n nu m be r Upper benefit cutoff value Lower benefit cutoff value Ru tti ng Upper benefit cutoff value Lower benefit cutoff value set to zero In te rn at io na l R ou gh ne ss In de x (IR I) Upper benefit cutoff value Lower benefit cutoff value Post-treatment curve intersects governing benefit cutoff value at 24 years This 20-year life is is the shortest of all expected condition indicator lives. The post-treatment condition analysis period is, therefore, set to 20 years. Age, years 5 10 15 20 25 Age, years 5 10 15 20 25 Do-nothing curve Do-nothing curve Do-nothing curve Post-treatment curve Post-treatment curve Figure 6. Determination of the overall post-treatment condition expected service life (analysis period).

30 (i.e., time since application age, expressed in years) rather than the overall pavement age. UBC = Upper benefit cutoff value associated with the condition indicator. LBC = Lower benefit cutoff value associated with the condition indicator. X0 = Lower age boundary (equal to zero). X1 = One of the following: (1) pavement age (in years) at which the do-nothing curve inter- sects the UBC, or (2) zero if the do-nothing condition at pavement age zero is less than the UBC, or (3) the pavement age at treat- ment application (XA) if the do-nothing con- dition is greater than the UBC at the treat- ment application age. X2 = Minimum of (1) the pavement age at treat- ment application and (2) the pavement age at which the do-nothing curve intersects the lower benefit cutoff value. XA = Pavement age at treatment application. X3 = One of the following: (1) overall pavement age at which the treatment performance curve intersects the UBC value, or (2) XA if the ini- tial treatment condition is less than the UBC, or (3) X4 if the treatment condition is greater than the UBC at the determined X4 age. X4 = The overall post-treatment analysis period (in terms of pavement age). (Eq. 5) AREA UBC EQ LBC EQ UBC EQ LBC EQ PT( ) DN DN PT PT + −( ) −( ) = −( ) − −( ) + −( ) − −( ) ∫ ∫ ∫ ∫ X X X X X X X X A A 0 2 0 1 4 3 0 0 Age, years UBC = Upper benefit cutoff value LBC = Lower benefit cutoff value X0 XAX1 Do-nothing performance curve D ec re as in g Co nd iti on In di ca to r AREAPT(–) X3 X4 Treatment performance curve Note: for this case, X2 = XA as XA is less than the projected intersection of the do-nothing curve and the LBC. Note: for this case, X2 = XA as XA is less than the projected intersection of the do-nothing curve and the LBC. INCREASING RELATIONSHIP In cr ea sin g Co nd iti on In di ca to r Age, years X0 XAX1 UBC = Upper benefit cutoff value LBC = Lower benefit cutoff value X3 X4 AREA PT(+) Do-nothing performance curve Treatment performance curve DECREASING RELATIONSHIP Figure 7. Determination of total areas associated with decreasing and increasing individual condition indicators.

31 where: AREAPT(+) = Computed post-treatment area associated with an increasing condition indicator rela- tionship (i.e., area from time zero to the end of the post-treatment analysis period). EQDN = Equation defining the do-nothing condition indicator relationship. EQPT = Equation defining the post-treatment con- dition indicator relationship (i.e., treatment performance curve). Note that the post- treatment equation is a function of treatment age (i.e., time since application age) rather than the overall pavement age. UBC = Upper benefit cutoff value associated with the condition indicator. LBC = Lower benefit cutoff value associated with the condition indicator. X0 = Lower age boundary (equal to zero). X1 = One of the following: (1) pavement age at which the do-nothing curve intersects the LBC value, or (2) zero if the do-nothing condition at pavement age zero is greater than the LBC, or (3) the pavement age at treatment application (XA) if the do-nothing condition is less than the LBC at the treat- ment application age. X2 = Minimum of (1) the pavement age at treat- ment application and (2) the pavement age at which the do-nothing curve intersects the UBC value. Note: X2 is often equal to XA. XA = Pavement age at treatment application. X3 = One of the following: (1) overall pavement age at which the treatment performance curve intersects the LBC value, or (2) XA if the ini- tial treatment condition is greater than the LBC, or (3) X4 if the treatment condition is less than the LBC at the determined X4 age. X4 = The overall post-treatment analysis period (in terms of pavement age). Figure 8 illustrates the application of these area boundary conditions and the resulting bounded post-treatment areas (i.e., AREAPT(FRICTION), AREAPT(RUTTING), andAREAPT(ROUGHNESS)) for the previously presented example. Step 7: Computation of Benefit Associated with Each Individual Condition Indicator When only one condition indicator is included, the individ- ual benefit is determined by comparing the post-treatment area computed in step 6 with the total area computed in step 3 for the do-nothing case. That is, the benefit is quantified as the difference in area between the overall post-treatment area and the associated do-nothing area (see Figure 9). When more than one condition indicator is included in an analysis, the computations are slightly more complex in that all post- treatment and do-nothing benefit areas are truncated at the expected service life of the post-treatment case computed in step 5. By truncating these areas, it is ensured that all com- puted benefit areas for the included condition indicators use the same analysis period. Figure 10 illustrates the benefit areas associated with friction, rutting, and roughness in the previously presented example. When multiple condition indicators are analyzed simulta- neously, converting individual condition indicator benefit areas into one overall benefit value becomes difficult because different condition indicators are expressed in different units. To solve this problem, each individual benefit area (i.e., the difference between the post-treatment and associated do- nothing areas) is normalized by dividing each computed ben- efit area by its associated total do-nothing area computed in step 3. The total do-nothing area is used as the basis for this comparison so computed benefit areas may be fairly com- pared between different timing scenarios. This normalization process results in all individual benefit values being expressed as a percentage. Equation 6 is used for the individual benefit computations. (Eq. 6) where: %BENEFITi = Individual benefit associated with a given condition indicator (benefit area expressed as a percentage of the associ- ated total do-nothing area). i = One of i = 1 to n condition indicators included in the analysis. AREABENEFIT(i) = Computed benefit area associated with the jth condition indicator in an analysis = (AREAPT(i) − AREADN(i)) AREAPT(i) = Computed post-treatment area between time = 0 and the computed post-treatment analysis period (computed in step 6). AREADN(i) = Do-nothing area between time = 0 and the computed post-treatment analysis period (computed in step 5). Note: if the analysis period is greater than or equal to the age at which the current condition indicator curve intersects the benefit cutoff value, this area will be the total do-nothing area (i.e., AREADN(i) = AREADN-TOT(i)). AREADN-TOT(i) = Total do-nothing area associated with the jth condition indicator in an analysis (i.e., that computed under step 3). Step 8: Computation of Overall Benefit When more than one condition indicator is included in an analysis, individual condition indicator benefit values are combined using defined benefit weighting factors. Continuing with the example, assume that individual benefit values for %BENEFIT AREA AREABENEFIT( ) DN TOT( )i i i= ( ) ( )−

32 friction, rutting, and IRI are 10, 16, and 20 percent, respec- tively (i.e., when compared with the respective areas associ- ated with the do-nothing option, the preventive maintenance treatment application results in increases of 10, 16, and 20 per- cent in the friction, rutting, and IRI areas, respectively). Fur- ther assume benefit weighting factors of 50, 25, and 25 are chosen for friction, rutting, and IRI, respectively (note that these factors add up to 100). The overall benefit contributions are then determined by multiplying the benefit weighting fac- tor percentages by the individual benefit values (e.g., for fric- tion 10 percent × 50/100 = 5.0 percent). The total overall benefit contribution is then computed as the sum of the values calcu- lated for each individual condition indicator. In this example, the total overall benefit contribution is 14.0 percent. While by itself this actual total benefit value is essentially meaningless, total benefit values computed for different timing scenarios can be used to compare the effectiveness of the different timing scenarios. Results of this example are presented in Table 19. Step 9: Cost Computations A simple two-step LCCA is conducted to compare the dif- ferent cost streams associated with each preventive mainte- Age, years 5 10 15 20 25 Fr ic tio n nu m be r Upper benefit cutoff value Lower benefit cutoff value R ut tin g Upper benefit cutoff value Lower benefit cutoff value set to zero In te rn at io na l R ou gh ne ss In de x (IR I) Upper benefit cutoff value Lower benefit cutoff value Age, years 5 10 15 20 25 Age, years 5 10 15 20 25 Do nothing curve Do nothing curve Do nothing curve Overall post-treatment analysis life of 20 years Overall post-treatment analysis life of 20 years (associated with friction) Overall post-treatment analysis life of 20 years (associated with friction) AREAPT(FRICTION) AREAPT(RUTTING) AREAPT(ROUGHNESS) Figure 8. Illustration of the total areas associated with individual condition indicators.

33 nance scenario. First, the present worth (at year zero) of each included treatment, rehabilitation, user-delay, or routine main- tenance cost is determined using equation 7. (Eq. 7) where: PW$ = Present worth value of an included cost (in year zero dollars). C = Individual maintenance or rehabilitation cost (in actual dollars). d = Discount rate expressed as a percentage (e.g., a discount rate of 4 percent translates to d = 0.04). n = Year (since construction) in which the individual cost is realized. Second, the computed total present worth cost is converted into an equivalent uniform annual cost (EUAC) using equa- tion 8. (Eq. 8) where: EUAC PWi i p p d d d i i = × +( ) +( ) −    ∑ $ 1 1 1 PW C$ = × +( )−1 d n EUACi = Computed equivalent uniform annual cost asso- ciated with the ith timing scenario. ∑PW$i = Sum of present worth values of all agency main- tenance or rehabilitation costs included in the cost stream associated with the ith timing scenario. d = Discount rate expressed as a percentage (e.g., a discount rate of 4 percent translates to d = 0.04 in the equation). i = Index associated with the current timing scenario. pi = Analysis period associated with the ith timing scenario (time from construction until year at which the first included condition indicator per- formance curve reaches the benefit cutoff value [from step 5]). Step 10: Determining the Most Cost-Effective Timing Scenario The final step of the analysis procedure is to analyze the benefits and costs computed for each application age to deter- mine the timing scenario that provides the largest B/C ratio. To normalize these computed B/C ratios, EIs are computed for each timing scenario by dividing each individual B/C ratio by the largest observed B/C ratio from all the different timing Age, years UBC = Upper benefit cutoff value LBC = Lower benefit cutoff value Do-nothing area D ec re as in g Co nd iti on In di ca to r AREABENEFIT(–) DECREASING RELATIONSHIP INCREASING RELATIONSHIP In cr ea sin g Co nd iti on In di ca to r Age, years UBC = Upper benefit cutoff value LBC = Lower benefit cutoff value Do-nothing area AREABENEFIT(+) Figure 9. Illustration of the benefit area (AREABENEFIT) for a single decreasing or increasing individual post-treatment condition indicator.

34 scenarios investigated. The most cost-effective timing scenario is that with the largest B/C ratio (i.e., that associated with an EI of 100). This process is best illustrated using an example. For an analysis to investigate six timing scenarios for a treat- ment applied on an HMA pavement 1, 2, 3, 4, 5, and 6 years after construction, benefit, cost, and B/C ratios are com- puted for each scenario using the previously outlined pro- cedures. The computed values for this example are presented in Table 20. These values show that timing scenario 4 (i.e., application age at 4 years) provides the largest B/C ratio. Using equation 1, EIs are computed for each scenario by dividing each individually computed B/C ratio by the largest observed B/C ratio (i.e., 0.01123 computed for an application age of 4 years after construction). Thus, the EI for applica- tion age 1, for example, is 0.00527/0.01123 × 100 = 47). The EI results for this example are illustrated in Figure 11. These results indicate that the optimal time to apply this preventive maintenance treatment is in year 4, although an application in year 3 produces very similar results. In such cases, other output results such as total benefit, EUAC, or extension of life may help identity the most appropriate tim- ing scenario. Although the optimal timing methodology is based on comparing B/C ratios, an agency may select treatment based Age, years 5 10 15 20 25 Fr ic tio n nu m be r Upper benefit cutoff value Lower benefit cutoff value R ut tin g Upper benefit cutoff value Lower benefit cutoff value set to zero In te rn at io na l R ou gh ne ss In de x (IR I) Upper benefit cutoff value Lower benefit cutoff value Age, years 5 10 15 20 25 Age, years 5 10 15 20 25 Overall post-treatment analysis life of 20 years Overall post-treatment analysis life of 20 years (associated with friction) Overall post-treatment analysis life of 20 years (associated with friction) AREABENEFIT(FRICTION) AREABENEFIT(RUTTING) AREABENEFIT(ROUGHNESS) Figure 10. Illustration of benefit areas associated with individual post-treatment condition indicators (applicable for investigating multiple condition indicators).

35 on other criteria. For the previously presented example, if maximizing benefit is the most important overall goal of the agency, an application age of 3 may be chosen because it pro- vides the highest benefit value (i.e., 102.4) in Table 20. If however, adequacy of the performance prediction equations is in question, cost may become the most important decision factor, and an application age of 4 with an EUAC of $8,890 would be favored over an application age of 3 with an EUAC of $9,246. ANALYSIS TOOL DEVELOPMENT An important component of this project was the develop- ment of a flexible, easy-to-use analytical tool that agencies could use to apply the proposed methodology. The result was OPTime—a macro-driven Microsoft® Excel Visual Basic Application (VBA)—that can be used to analyze actual data or evaluate hypothetical situations. Details of its develop- ment and make up are described. Built-In Flexibility Creating a tool that can be readily used by all agencies is difficult due to variations in agency practices, such as condi- tion rating systems, data availability, and data quality. How- ever, flexibility is intentionally built into the analysis tool to facilitate use by different users. Choice of Detailed or Simple Analysis Types The primary purpose of the OPTime tool is to allow engi- neers to analyze actual historical preventive maintenance- related condition data in order to determine the optimal tim- ing of a specific preventive maintenance treatment. However, many agencies are still in the early stages of collecting the performance data needed for such analysis. Therefore, addi- tional flexibility is built into the analysis tool through the inclusion of two distinct analysis methods (referred to as “detailed” and “simple” analysis methods) that may be used to compare preventive maintenance timing scenarios. The detailed analysis method is used to analyze actual (or estimated) condition versus age data. When actual field data are used for this analysis, expected condition versus age rela- tionships (before and after preventive maintenance treatment applications) must be defined by either selecting an equation type and entering known equation coefficients, or fitting a regression equation using condition versus age data points. Condition Indicator Individual Benefit Values, % Condition Indicator Benefit Factor Overall Benefit Contribution, % Friction 10 0.50 5.0 Rutting 16 0.25 4.0 Roughness (IRI) 20 0.25 5.0 TOTAL — 1.00 14.0 Year of Application BENEFIT (B) Overall Benefit, % COST (C) EUAC, $ BENEFIT-TO- COST RATIO (B/C), %/$ Effectiveness Index (EI) 1 52.7 $10,000 0.00527 47 2 65.5 $9,615 0.00681 61 3 102.4 $9,246 0.01108 99 4 99.8 $8,890 0.01123* 100 5 72.5 $8,548 0.00848 76 6 65.4 $8,219 0.00796 71 * Largest B/C ratio. TABLE 19 Example computation of overall benefit TABLE 20 Example computation of overall benefit (BENEFITOVERALL) 0 20 40 60 80 100 1 2 3 4 5 6 Age at PM Application Ef fe ct iv en es s I nd ex Figure 11. Example of Effectiveness Index versus timing of preventive maintenance application.

36 If actual data are not available, or if the user is concerned about the adequacy of specific mathematical relationships (i.e., choosing equation types and coefficients), the simple analysis method is used to compare many “what if” timing scenarios. Performance relationships are more easily defined by choosing a starting condition level, a condition versus age point for the curve to pass through, and the expected exten- sion of life at the benefit cutoff value. Essentially, the condi- tion indicator versus age relationships are defined visually rather than through a specific mathematical relationship. Flexibility in Defining Pavement Performance Because there is no universal goal for a preventive mainte- nance program, potential users of the optimal timing method- ology may approach the problem in unique manners. The agency may seek improved friction, reduced roughness, better overall pavement condition, or reduced user delay costs, for example. The analytical tool allows the evaluation of optimal timing in terms of any desired condition or criteria. In addition to the typical condition indicators associated with both HMA and portland cement concrete (PCC) pavements, two user- definable fields are provided to customize the analysis. The units associated with all condition indicators (both standard and user definable) are also completely customizable. Cost Type Options Four different cost types may be included in the analysis— treatment costs, work zone user delay costs, rehabilitation costs, and routine maintenance costs; the user decides which of these costs to include in the analysis. This flexibility allows a user to conduct typical, as well as specialized, analysis. A typical analysis primarily consists of determining an optimal timing scenario based only on treatment costs. An example of a specialized analysis is one in which the user wants to choose the optimal timing of a treatment while only consid- ering user costs (i.e., all the other cost types, including the cost of the treatment, are ignored). While such an approach would be considered unconventional, the analysis method permits such investigations. Analysis Setup The determination of the optimal timing requires many different inputs during the analysis setup phase. The general steps required to setup an analysis are presented in Figure 12. Brief descriptions of each step are included below. • Analysis type selection—OPTime includes a choice of two distinct analysis types. The detailed analysis is pri- marily used to analyze actual historical performance data; the simple analysis approach is generally used to easily conduct hypothetical “what if” scenarios in the absence of actual data. • Select condition indicators to be included—The meth- odology allows the user to select the one or more condi- tion indicators that will be tracked/predicted over time. The influence of the selected condition indicators’ ben- efit and costs will be estimated to determine optimal preventive maintenance timing. • Define the preventive maintenance treatment to be analyzed—The methodology requires that the user spec- ify a particular preventive maintenance treatment to be analyzed. The methodology only analyzes one treatment at a time (i.e., it does not compare preventive mainte- nance treatments). • Define all timing scenarios that will be investigated— The methodology evaluates treatment timings that are specified by the user (i.e., all possible treatment timings Analysis Type Selection Selection and Definition of Condition Indicators Selection of Preventive Maintenance Treatment Simplified Definition of "Do- Nothing" Performance Curves Simplified Definition of Post PM Treatment Performance Relationships Definition of Costs Definition of Benefit Ranking Factors Conduct Analysis Detailed Definition of Post PM Treatment Performance Relationships Detailed Definition of "Do- Nothing" Performance Curves Simple Detailed Definition of Application Ages (Timing Scenarios) Figure 12. Outline of the data flow through the methodology.

37 are not considered). Therefore, one of the important steps in the analysis setup process is the definition of the specific treatment application ages that will be consid- ered in the analysis. The primary result of the analysis is identifying the most effective treatment application age from among those considered. • Define do-nothing curves for each included condition indicator—The user must define, for the do-nothing option, the expected performance curves for each con- dition indicator included in the analysis. These relation- ships represent the expected pavement performance in the absence of any preventive maintenance or rehabili- tation activities. The relationships should, however, con- sider any routine or reactive maintenance that is typical for a given pavement type. • Define post-preventive maintenance performance relationships—In order to compute benefit for a given performance indicator, the user defines how the pave- ment will perform in the prediction mode (i.e., condition indicator versus time relationships) after a preventive maintenance treatment is applied. Because these perfor- mance relationships depend on pavement age (or pave- ment) condition at treatment application, a separate per- formance relationship is required for each application age (timing scenario) included within the analysis. • Define cost types and values—The user has the option to include any or all cost types (i.e., preventive mainte- nance treatment costs, user delay-related costs, rehabili- tation costs, and the cost of additional routine or reactive maintenance activities) in the LCCA. Included costs are used to make up the cost streams associated with each individual preventive maintenance timing scenario. • Define benefit weighting factors—Each considered condition indicator is assigned an integer weighting fac- tor between 0 and 100, where all the entered weighting factors must add up to 100. The selected weighting fac- tors are used to combine the individual benefit values into an overall benefit value for each timing scenario. Data Interpretation The analysis tool includes a number of summary tables and charts that illustrate the results of the analysis. Specifi- cally, the results summarize the benefit values, EUACs, B/C ratios, and effectiveness indices determined for all timing scenarios. The timing scenario with the largest B/C ratio is identified as the scenario representing optimal timing. VALIDATION OF THE ANALYSIS METHODOLOGY An evaluation of the analysis methodology and the analysis tool was undertaken using actual data provided by four state agencies—Arizona, Kansas, Michigan, and North Carolina. These evaluations are referred to as Case Studies 1 through 4. Also, the LTPP data for the maintenance effectiveness experiments—available on the DataPave 3.0 software—are presented as Case Study 5. This section describes the details of the validation process. The purpose of the evaluation was not to identify the opti- mal time to perform preventive maintenance using an agency’s data, but to demonstrate the methodology’s and the OPTime analytical tool’s ability to use actual data. The following case studies indicate how actual data are handled in the analytical approach, the types of assumptions that are made to use the methodology, and how the absence of certain types of infor- mation preclude the successful use of the methodology. Case Study 1—Arizona Introduction The Arizona Department of Transportation’s (ADOT) data for seal coat treatments on flexible pavements were analyzed. Data summaries and performance models were obtained from a 1999 report (24, 25). Condition Indicators The effectiveness of seal coats was measured by examin- ing their effect on the following three condition indicators: • Roughness (IRI) • Friction (measured with a Mu-meter) • Cracking (measured in terms of percent area) An earlier study (24, 25) determined that roughness was by far the most useful performance indicator to distinguish between different materials and different circumstances. Table 21 sum- marizes subjective ratings associated with ranges of each of these condition indicators. Condition Indicator Value/Range Condition Category <1.47 m/km (93 in./mi) Low 1.47 to 2.26 m/km (93 to 143 in./mi) Medium Roughness (IRI) >2.26 m/km (143 in./mi) High <35 Low 35 to 42 Medium Friction (Mu-meter reading) 43 to 99 High <10 Low 10 to 30 Medium Cracking (percent of area) >30 High TABLE 21 Condition indicator ranges and their associated subjective ratings

38 Do-Nothing Performance Curves A series of linear do-nothing performance curves are avail- able for the selected roughness, friction, and cracking condi- tion indicators for pavements in all roadway classes, envi- ronmental regions, and traffic levels (24, 25). Table 22 lists the do-nothing condition indicator versus age relationships selected for this demonstration; these relationships are shown in Figures 13, 14, and 15, respectively. Benefit Cutoff Values Benefit cutoff values are determined by analyzing the expected regression equations over the condition indicator ranges listed in Table 21. Details of this analysis are presented as follows: • Roughness—Because IRI increases with time, an upper IRI benefit cutoff value is required. A value of 1.47 m/km (93 in./mi) was chosen because it indicates the transition from a low roughness level to a tolerable roughness level. According to the roughness regression equation, this value is predicted at an age of 27.8 years. The lower ben- efit cutoff value was set to a value of 0 m/km (0 in./mi). • Friction—As Mu-meter values typically decrease over time, a lower benefit cutoff value is required for the analysis. A comparison of the regression equation to the condition ranges listed in Table 21 finds that friction val- ues of 35 and 43 correlate to ages of 117.1 and 80.7 years, respectively. Because these are extremely high ages, a friction value of 55 was chosen, corresponding to a pre- dicted age of 26.2 years. An upper benefit cutoff value was conservatively set to a value of 100. Condition Indicator Regression Equation Roughness (IRI), m/km* IRI = 0.0207 × AGE + 0.89 Friction (Mu-meter results) Friction = –0.22 × AGE + 60.76 Cracking (% of 1,000 sf area at each milepost), % Cracking = 0.33 × AGE + 0.6 *1 m/km = 63.4 in./mi. TABLE 22 Do-nothing performance equations IRI = 0.0207* Age + 0.89 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 0 5 10 15 20 25 30 Pavement Age, years Ro u gh ne ss (IR I), m /k m y = -0.22x + 60.76 53 54 55 56 57 58 59 60 61 62 0 5 10 15 20 25 30 Pavement Age, years Fr ic tio n (M u- me ter re su lts ) y = 0.33x + 0.6 0 2 4 6 8 10 12 0 5 10 15 20 25 30 Pavement Age, years Cr ac ki ng , p er ce nt a re a Figure 13. Assumed roughness do-nothing curve for Case Study 1—Arizona. Figure 14. Assumed friction do-nothing curve for Case Study 1—Arizona. Figure 15. Assumed cracking do-nothing curve for Case Study 1—Arizona. Condition Indicator Regression Equation Roughness (IRI), m/km IRI = 0.0273 × TAGE + 1.52 Friction (Mu-meter results) Friction = –0.54 × TAGE + 69.2 Cracking (% of 1000 sf area at each milepost), % Cracking = 0.7 × TAGE + 3.2 TABLE 23 Post-treatment performance equations associated with a seal coat treatment • Cracking—Since cracking increases with time, an upper cracking benefit cutoff value is required for the analy- sis. A value of 10% was selected, as it indicates the tran- sition from low to medium cracking. According to the cracking regression equation, this value corresponds to an age of 28.5 years. The lower benefit cutoff value was conservatively set to a value of 0% cracking. Post-Preventive Maintenance Performance Relationships A series of linear post-treatment performance curves for the selected roughness, friction, and cracking condition indicators are available (24, 25); these relationships are listed in Table 23 and plotted for roughness, friction, and cracking in Figures 16, 17, and 18, respectively. However, these relationships are pre- sented as functions of the treatment age (TAGE), not as func- tions of the pavement’s age when the treatment was applied.

39 Analysis Setup The analysis tool is used to analyze the interpreted perfor- mance data. Specifically, the following inputs define the analy- sis session for this case study: • Analysis Type—A detailed analysis type is selected because actual data are being analyzed. • Condition Indicators—Three condition indicators are used in this analysis: roughness, friction, and cracking. • Preventive Maintenance Treatment Selection—A seal coat applied at 1, 4, 7, 10, and 13 years is investigated. No routine/reactive maintenance costs are included. IRI = 0.0273 * TAGE + 1.52 0.00 0.50 1.00 1.50 2.00 2.50 0 5 10 15 20 25 30 Treatment Age, years Ro u gh ne ss (IR I), m /k m Figure 16. Assumed post-treatment seal coat roughness relationship for Case Study 1—Arizona. Cracking = 0.7 * TAGE + 3.2 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Treatment Age, years Cr ac ki n g, pe rc en t a re a Figure 18. Assumed post-treatment seal coat cracking relationship for Case Study 1—Arizona. Friction = -0.54 * TAGE + 69.2 0 10 20 30 40 50 60 70 80 0 5 10 15 20 25 30 Treatment Age, years Fr ic tio n (M u -m et er re su lts ) Figure 17. Assumed post-treatment seal coat friction relationship for Case Study 1—Arizona. Output Data Pavement Surface Type: HMA Treatment Type: Seal Coat Application Years: 1, 4, 7, 10, 13 Expected Do-Nothing Service Life (yrs): 26.2 Benefit Summary Individual Benefit Summary Benefit Ranking Factors => 25 15 60 Application Age, yrs Total Benefit Nonload- Related Cracking Roughness/ Smoothness Friction 1 0.04 -0.68 -0.84 0.57 4 0.23 -0.49 -0.75 0.77 7 0.39 -0.32 -0.67 0.95 10 0.53 -0.17 -0.59 1.11 13 0.65 -0.05 -0.52 1.24 Cost Summary Application Age, yrs Treatment Cost, PW $ User Cost, PW $ Other Maintenance Cost, PW $ Rehab. Cost, PW $ Total Present Worth, $ EUAC, $ 1 $24,423 n/a n/a n/a $24,423 $2,847 4 $21,712 n/a n/a n/a $21,712 $2,088 7 $19,302 n/a n/a n/a $19,302 $1,606 10 $17,159 n/a n/a n/a $17,159 $1,275 13 $15,255 n/a n/a n/a $15,255 $1,035 Effectiveness Summary Application Age, yrs Effectiveness Index Total Benefit EUAC, $ Expected Life, yrs Expected Extension of Life, yrs 1 2.38 0.04 $2,847 10.7 -15.5 4 17.32 0.23 $2,088 13.7 -12.5 7 38.66 0.39 $1,606 16.7 -9.5 10 66.30 0.53 $1,275 19.7 -6.5 13 100.00 0.65 $1,035 22.7 -3.5 TABLE 24 Analysis results for Case Study 1—Arizona

40 Effectiveness Index vs. Application Timing 0 20 40 60 80 100 0 2 4 6 8 10 12 14 Timing of First PM Application, years Ef fe ct iv en es s In de x Extension of Life vs. Application Timing -20.0 -15.0 -10.0 -5.0 0.0 0 2 4 6 8 10 12 14 Timing of First PM Application, years Ex te ns io n of L ife , ye ar s EUAC ($) vs. Application Timing $ $500 $1,000 $1,500 $2,000 $2,500 $3,000 0 2 4 6 8 10 12 14 Timing of First PM Application, years EU AC ($ ) Figure 19. Results of the analysis for Case Study 1—Arizona. • Performance Relationships—Models are already pro- vided. The do-nothing performance relationships are defined in Table 22 and the post-treatment performance relationships are defined in Table 23. • Project Definition—The project size is defined as 16,723 m2 (20,000 yd2). • Cost Data—Only treatment costs are included in the cost analysis (i.e., rehabilitation, user, and routine maintenance costs are excluded). The in-place unit cost of a seal coat application is $1.52/m2 ($1.27/yd2) as reported by ADOT (i.e., for the entire treatment application. A discount rate of 4.0 percent is used in the analysis. • Benefit Weighting Factors—Benefit weighting factors are needed for three condition indicators; they were arbi- trarily chosen as 15, 60, and 25 percent for roughness, fric- tion, and cracking, respectively. Analysis Results The output results are summarized in Table 24 and Fig- ure 19. These results show that of the five investigated appli- 0 2 4 6 8 10 12 0 5 10 15 20 25 Age N on lo ad -R el at ed C ra ck in g Do Nothing The Best Treatment Application Lower Benefit Cutoff Value Upper Benefit Cutoff Value Figure 20. Cracking versus age for the most appropriate application age of 13 years for Case Study 1—Arizona.

41 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 Age Ro ug hn es s Do Nothing The Best Treatment Application Lower Benefit Cutoff Value Upper Benefit Cutoff Value Figure 21. Roughness versus age for the most appropriate application age of 13 years for Case Study 1—Arizona. 0 20 40 60 80 100 120 0 5 10 15 20 25 Age Fr ic tio n Do Nothing The Best Treatment Application Lower Benefit Cutoff Value Upper Benefit Cutoff Value Figure 22. Friction versus age for the most appropriate application age of 13 years for Case Study 1—Arizona. cation ages, the most cost-effective option is applying the treatment at age 13 (indicated by an EI of 100). Because the same equation is used for all application ages, it is not unex- pected that the largest benefit is associated with the latest application. This is because the benefit increases and treat- ment cost decreases with application age. The actual condi- tion versus age plots for the three included condition indica- tors are illustrated in as Figures 20, 21, and 22. Note that all of the individual benefits associated with cracking and roughness are computed as negative values. As illustrated in Figure 20, the post-treatment cracking curve crosses the do-nothing curve and the areas above the do-nothing and post-treatment curves bound by the upper benefit cutoff value of 10 percent appear to be similar. However, the results provided in Table 24 show an indi- vidual benefit value of −0.05 (for an application age of 13 years) that indicates a slightly greater benefit area for the do-nothing case than for the post-treatment case. The neg- ative benefit values associated with roughness occurred because, according to the IRI equations, the application of the treatment resulted in an increased pavement roughness as shown in Figure 21.

Although these results appear to contradict engineering judgment, they reflect the accuracy of the provided condition prediction models. This case study points out the importance of not only obtaining representative datasets, but also focus- ing on compiling separate datasets for different treatment application ages. Case Study 2—Kansas Introduction As part of an ongoing study, the Kansas Department of Transportation (KDOT) is developing condition indicator prediction models based on the historical condition data available in their pavement management database for nearly 11,000 pavement segments. For this case study, only the transverse cracking models developed by KDOT were used to demonstrate the analysis approach. This subsection intro- duces the modeling approach used by KDOT and demon- strates how such agency-developed models may be used within the analysis approach developed under this project. KDOT Modeling Procedure Modeling the performance of a given construction, rehabil- itation, or maintenance activity is a four-step process. Estimate Equivalent Asphalt Thickness (EqThick). In an effort to estimate the expected pavement performance impact associated with a specific paving activity, KDOT has esti- mated the equivalent asphalt thicknesses associated with dif- ferent non-structural, light-structural, and heavy-structural paving actions used in Kansas. Examples of selected equiv- alent thickness values are listed in Table 25. 42 Compute Expected Design Lives. The second step of the KDOT modeling procedure is to compute an expected design life of a selected paving action. Based on the results of a mul- tiple linear regression process, the following equation is used to compute the expected design life for a given paving activ- ity on a flexible pavement: (Eq. 9) where: DL_Flex = Flexible pavement design life, years. FDBit = Full-depth bituminous index (value of 1.0 if the pavement is a full-depth section). EqThick = Equivalent thickness of current paving action (construction, rehabilitation, or maintenance activity), in. EqTCR = Equivalent number of transverse cracks at time of current paving action. Note: EqTCR is the equivalent number of “code 3” (rough or very wide) cracks expected per 30-m (100-ft) segment. D_ADL_t = Design lane average daily 80 kN (18 kip) loads. In the KDOT study, the limits shown in Table 26 are used to “cap” the computed design life if necessary. Compute the Condition Indicator Value for the First Sur- vey Year After a Paving Action. The third step of the KDOT modeling procedure is to compute the condition indicator DL_Flex FDBit EqThick Ln EqTCR Ln D_ADL_tEqThick = + × + × − × +( ) − ×   8 836 1 610 1 201 3 725 1 0 957 . . . . . Action Type Paving Action Description Equivalent Thickness, mm (in). Do nothing 0 (0.00) Modified slurry seal 6 (0.25) Rout and crack seal on flexible pavement 13 (0.50) Non- Structural 25-mm (1.0-in.) asphalt overlay 25 (1.00) 38-mm (1.5-in.) asphalt overlay 38 (1.50) Extensive patching, 38-mm (1.5-in.) asphalt overlay 44 (1.75) Light- Structural 50-mm (2.0-in.) asphalt overlay 50 (2.00) 63-mm (2.5-in.) asphalt overlay 63 (2.50) Cold recycle 100-mm (4-in.), 38-mm (1.5-in.) asphalt overlay 100 (4.00) Heavy- Structural New HMA construction 50, 100, 150, or 200 mm (2, 4, 6, or 8 in.) depending on chosen design 50, 100, 150, or 200 (2, 4, 6, or 8 in.) depending on chosen design TABLE 25 Examples of equivalent thickness values associated with various construction actions in Kansas

values expected at the first survey year after a paving action. Different prediction equations are developed for KDOT’s structural and non-structural paving actions. The following equations are used to compute the EqTCR value at the first year after a structural or non-structural paving action, respectively. Structural Action EqTCRpost = 0.0973 + 0.0845 × EqTCRprior + 0.000394 × D_ADL (Eq. 10) where: EqTCRpost = Equivalent number of transverse cracks at year 1 after a structural paving action. Note: EqTCR is the equivalent number of “code 3” (rough or very wide) cracks expected per 30-m (100-ft) segment. EqTCRprior = Equivalent number of transverse cracks immediately before the paving action. D_ADL = Design lane average daily 80 kN (18 kip) loads at the year of the last structural action. Non-Structural Action EqTCRpost = 0.376 + 0.239 × EqTCRprior − 0.351 × EqThick (Eq. 11) + 0.0943 × FDBit − 0.0190 × DL_Flex where: EqTCRpost = Equivalent number of transverse cracks at year 1 after a non-structural paving action. EqTCRprior = Equivalent number of transverse cracks immediately before the paving action. EqThick = Equivalent thickness of current paving action (construction, rehabilitation, or maintenance activity), in. FDBit = Full-depth bituminous index (value of 1.0 if the pavement is a full-depth section). DL_Flex = Flexible pavement design life (years) based on the design life regression model of the last structural action. 43 Note: the equivalent transverse cracking value is assumed to drop to zero immediately after a rehabilitation action (i.e., EqTCR = 0 at the year of the paving action). Compute the Condition Indicator Values for Subsequent Years. The last step of the KDOT modeling procedure is to compute the condition indicator values for all other years after the first survey year. In this case study, the following equation is used to compute the EqTCR at these subsequent years regardless of the type of the most recent paving action: EqTCRt + 1 = 0.182 + 1.10 × EqTCRt + 0.282 × CTCRt − 0.0218 × FDBit (Eq. 12) − 0.0113 × DL_Flex where: EqTCRt + 1 = Equivalent number of transverse cracks in any year after the first survey year. Note: EqTCRt+1 is the maximum of the predicted value from the regression or EqTCRt + 0.05. EqTCRt = Equivalent number of transverse cracks in the previous year. CTCRt = Change in EqTCR in the previous year (i.e., CTCRt = EqTCRt − EqTCRt − 1). FDBit = Full-depth bituminous index (value of 1.0 if the pavement is a full-depth section). DL_Flex = Flexible pavement design life (years) based on the design life regression model of the last structural action. The following subsections describe an example of how the KDOT modeling equations are used within the analytical tool. Treatment Selection For this case study, routing and sealing cracks on a flexi- ble pavement is chosen as the preventive maintenance treat- ment. “Rout and Crack Seal” is assigned an effective HMA thickness of 13 mm (0.5 in.). Treatment Costs The assumed average crack sealing cost is $1,865 per km ($3,000 per mi). A discount rate of 2.0 percent is used for the analysis (the discount rate typically used by KDOT). Condition Indicators The equivalent number of “code 3” (rough or very wide) cracks expected per 30-m (100-ft) segment (EqTCR) is the sole condition indicator for this treatment. Note that at the time of initial construction, or the time at which all cracks are routed and sealed, this EqTCR condition indicator is set or Equivalent Thickness of Last Paving Action Design Life Projection Limit, yrs < 38 mm (1.50 in.) 10 39 to 75 mm (1.51 to 3.00 in.) 10 76 to 100 mm (3.01 to 4.00 in.) 15 > 100 mm (> 4.01 in.) 20 TABLE 26 Flexible pavement design life limits for equivalent thickness values

reset to a value of zero. Also, because the number of devel- oping cracks increases over time, the general trend of this condition indicator is increasing. Benefit Cutoff Values Based on recommendations from KDOT personnel, a distress threshold value for EqTCR is identified as 0.62. Because EqTCR is expected to increase over time, 0.62 is set as an upper benefit cutoff; the practical lower limit of EqTCR of zero is used as the lower benefit cutoff value in the analysis. Do-Nothing Performance Curve In this example, the do-nothing performance curve is defined as the EqTCR versus time relationship associated with the initial pavement construction. An equivalent asphalt pavement thickness of 200 mm (8.0 in.) represents the do-nothing condition. The following steps are used to determine the do-nothing performance curve for the equiv- alent transverse cracking condition indicator. Step 1—Compute the Expected Design Life Associated with the Initial Construction Action. The first step in deter- mining the representative EqTCR do-nothing curve is to compute the expected service life for the assumed 200-mm (8.0-in.) equivalent asphalt thickness. The following inputs are used in equation 9 to compute the design life associated for the initial construction. • FDBit = 1.0 for a full-depth bituminous pavement. • EqThick = 200 mm (8.0 in.) for an equivalent asphalt thickness of 200 mm (8.0 in.) associated with new construction paving. • EqTCR = 0 for pavement with no equivalent number of transverse cracks at time zero (initial construction). • D_ADL_t = 250 for assumed 250 average daily 80kN (18 kip) loads (average daily ESALs) in the design lane at time of initial construction. (This is reported by KDOT to be a typical traffic level for a 2-lane high- way in Kansas.) Inserting these values into equation 9 results in the fol- lowing: DL_Flex = 8.836 + 1.610 × (1.0) + 1.201 × (8.0) − 3.725 × Ln(0 + 1) − 0.957 × Ln(250/8.0) = 16.8 years (design life associated with initial construction) Step 2—Compute the EqTCR Value for the First Survey Year (Year 1) After Initial Construction. The next step is to determine the expected EqTCR value for year 1 (i.e., the first survey year after initial construction). Since the initial 44 construction activity is a structural action, the EqTCR value at year 1 is computed using equation 10. The specific inputs used in that equation are the following: • EqTCRprior = 0 at the previous year (initial construction). • D_ADL = 250 for the number of average daily 80kN (18 kip) loads (average daily ESALs) at the time of ini- tial construction (assumed to be 250). Inserting these input values into equation 10 results in the following: EqTCRpost = 0.0973 + 0.0845 × (0) + 0.000394 × (250) = 0.196 (EqTCR at first year after initial construction) Step 3—Compute Subsequent Year EqTCR Values Used to Define the Performance After Initial Construction (Do- Nothing Curve). The final step is to determine the expected EqTCR values for years other than years 0 and 1. EqTCR values for subsequent years are computed using equation 12. The following inputs illustrate the case for computing the EqTCR value at year 2: • EqTCR1 = 0.196 as computed in step 2. • CTCR1 = 0.196 is the computed change in EqTCR in the previous year. (For this example, CTCR1 = EqTCR1 − EqTCR0 = 0.196 − 0 = 0.196.) • FDBit = 1.0 for full-depth bituminous pavement section. • DL_Flex = 16.8 years is the expected initial construc- tion design life as computed in step 1. Inserting these values into equation 12 results in the following: EqTCR2 = 0.182 + 1.10 × (0.196) + 0.282 × (0.196) − 0.0218 × (1.0) − 0.0113 × (16.8) = 0.241 As explained in equation 12, the EqTCR2 value is the higher of this computed value (i.e., 0.241) or EqTCR1 + 0.05 (i.e., 0.196 + 0.05 = 0.246). Therefore, EqTCR2 is redefined as 0.246. Completing this iterative process for subsequent years (up to year 20) results in the expected EqTCR values presented in Table 27. Figure 23 illustrates the plotted EqTCR data and the following second-order polynomial equation that repre- sents the do-nothing condition: EqTCR = 0.0015 × Age2 + 0.0348 × Age + 0.1415 (Eq. 13) Post-Preventive Maintenance Performance Relationships In order to test the sensitivity of the timing of routing and sealing cracks, a wide range of application ages (1, 3, 5, 7, 9,

11, and 13 years) were considered. The following two-step process is used to determine post-treatment performance curves for each application age. Compute the EqTCR Value for the First Survey Year After a Treatment Application (for All Application Ages). The first step in determining the representative post-treatment performance relationships is to estimate the expected EqTCR value for the first survey year after each treatment application. Since the rout and seal activity is a non-structural action, these year 1 EqTCR values are computed using equation 11. As 45 indicated previously, the expected treatment design life is a function of four different variables: • EqTCRprior—the computed values given in Table 27. • EqThick—the equivalent asphalt thickness of 13 mm (0.5 in.) associated with the rout and crack seal preven- tive maintenance treatment. • FDBit = 1.0—for full-depth bituminous section. • DL_Flex—the expected design life of the last structural treatment application. Since the last structural applica- tion is initial construction, this value is held constant in the analysis at the calculated 16.8 years. Table 28 lists all the required inputs and the resulting expected first survey year EqTCRpost values (computed using equation 11) for each application age. The following exam- ple illustrates the computation of the EqTCRpost value for the application age of 3 years using equation 11. EqTCRpost = 0.376 + 0.239 × (0.239) − 0.351 × (0.5) + 0.0943 × (1.0) − 0.0190 × (16.8) = 0.047 (equivalent number of cracks at the first survey year after routing and sealing cracks at a pavement age of 3 years) Compute Subsequent Year EqTCR Values Used to Define the Performance After Applying the Rout and Crack Seal Treatment. The second step involves defining the post-treatment performance curves for each application age by computing EqTCR values for subsequent years (i.e., all years after the first survey year after treatment application) using equation 12. Table 29 lists all the computed EqTCR val- ues that define the post-treatment performance for the differ- ent application ages. Table 30 lists the EqTCR versus age second-order polynomial regression equations that are fit through the data for each application age. Also shown in Table 30 are the computed times at which each regression equation crosses the previously determined upper benefit Pavement Age Computed Equivalent Number of Rough or Very Wide Transverse Cracks per 30 m (100 ft) (EqTCR) 0 0.000 1 0.196 2 0.246 3 0.296 4 0.346 5 0.396 6 0.446 7 0.496 8 0.546 9 0.596 10 0.646 11 0.696 12 0.750 13 0.812 14 0.881 15 0.959 16 1.048 17 1.149 18 1.263 19 1.392 20 1.538 TABLE 27 Computed yearly EqTCR values used to define the do-nothing curve EqTCR = 0.0015x2 + 0.0348x + 0.1415 R2 = 0.9857 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 0 5 10 15 20 Pavement Age, years Eq ui v al en t N o . o f R o u gh o r Ve ry W id e Tr an sv er se C ra ck s pe r 30 m (1 00 ft) (E qT CR ) Figure 23. Estimated do-nothing curve for Case Study 2—Kansas.

46 Inputs Application Age FDBit Index (FDBit) Equivalent Asphalt Thickness (EqThick) Equivalent No. of Cracks Before Paving Action (EqTCRprior) Expected Design Life, years Equivalent No. of Cracks at First Year After Paving Action (EqTCRpost) 1 1 0.196 0.023 3 0.296 0.047 5 0.396 0.071 7 0.496 0.095 9 0.596 0.119 11 0.696 0.143 13 1.0 (the FDBit variable is held constant for all application ages) 13 mm (0.5 in.) (the EqThick variable is held constant for all application ages) 0.812 16.8 (the DL_Flex value is held constant for all application ages) 0.170 1 Equivalent number of cracks at year 1 after paving action (EqTCRpost) are computed using equation 11. TABLE 28 Required inputs and computed equivalent cracking values at the first survey year after a treatment application (at different chosen application ages) Application Age, years Treatment Age, years 1 3 5 7 9 11 13 0 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1 0.023 0.047 0.071 0.095 0.119 0.143 0.170 2 0.073 0.097 0.121 0.145 0.169 0.193 0.220 3 0.123 0.147 0.171 0.195 0.219 0.243 0.270 4 0.173 0.197 0.221 0.245 0.269 0.293 0.320 5 0.223 0.247 0.271 0.295 0.319 0.343 0.370 6 0.273 0.297 0.321 0.345 0.369 0.393 0.420 7 0.323 0.347 0.371 0.395 0.419 0.443 0.470 8 0.373 0.397 0.421 0.445 0.469 0.493 0.520 9 0.423 0.447 0.471 0.495 0.519 0.543 0.570 10 0.473 0.497 0.521 0.545 0.569 0.593 0.620 Application Age Regression Equation Computed Life Until Equation Reaches Upper Benefit Cutoff Level (EqTCR = 0.62) 1 EqTCR = 0.0492 * AGE - 0.0199 13.0 3 EqTCR = 0.0499 * AGE - 0.0022 12.5 5 EqTCR = 0.0506 * AGE + 0.0156 11.9 7 EqTCR = 0.0515 * AGE + 0.0325 11.4 9 EqTCR = 0.0523 * AGE + 0.0499 10.9 11 EqTCR = 0.0536 * AGE + 0.0653 10.3 13 EqTCR = 0.0555 * AGE + 0.0820 9.7 TABLE 29 Computed post-treatment EqTCR values associated with the different chosen application ages TABLE 30 Determined post-preventive maintenance performance relationships for Case Study 2—Kansas

cutoff value of EqTCR = 0.62. These computed times repre- sent the expected ages at treatment failure. Finally, the deter- mined post-treatment performance curves associated with different application ages are plotted in Figure 24. Analysis Setup The analysis tool is used to evaluate the estimated perfor- mance data described. The following inputs are used for ana- lyzing the data obtained for this project: • Analysis Type—A detailed analysis type is selected because actual data are being analyzed. • Condition Indicators—A custom condition indicator for equivalent transverse cracking is defined and labeled as EqTCR. • Preventive Maintenance Treatment Selection—A custom treatment, Rout and Seal Cracks, is used. Appli- cation ages of 1, 3, 5, 7, 9, 11, and 13 years are investi- gated. • Performance Relationships—The do-nothing perfor- mance curve from Figure 23 and the post-treatment per- formance relationships defined in Table 30 are entered directly. • Project Definition—The sample project is assumed to be a 1.6-km (1-mi) segment of a 2-lane (7.3-m [24-ft] wide) rural highway. Therefore, for this particular con- dition indicator, the project is defined by setting the proj- ect length to 1.6 km (1 mi). 47 • Cost Data—Only the cost of routing and sealing cracks is included in the analysis (i.e., rehabilitation, user, and routine maintenance costs are excluded). The assumed average crack sealing cost is $1,865 per km ($3,000 per mi). A discount rate of 2.0 percent is also chosen for the analysis based on KDOT’s typical practice. • Benefit Weighting Factors—Since only one condi- tion indicator is used in the analysis session, the bene- fit weighting factor associated with the equivalent cracking value (EqTCR) is set to 100 percent. Analysis Results The results obtained from this analysis are summarized in Table 31. These results indicate that out of the seven investi- gated application ages, application of the treatment at age 11 is the most cost-effective option as indicated by an EI of 100. Also, the application treatment at this age is expected to extend pavement life by 11.6 years (i.e., 11.6 more years than the expected do-nothing service life of 9.7 years) and an EUAC of $140. To help illustrate the results of this analysis, plots of EI, total benefit, extension of life, and EUAC versus treatment application age are shown in Figure 25. It is interesting to note that the highest EI is obtained for an application age of 11 years while an application age of 7 years provides the largest total benefit. Therefore, if an agency regards the differences in EUAC as insignificant, the most appropriate option would be the application age of 7 years. Regression Trends Through Selected Post-Treatment Data 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0 2 4 6 8 10 12 14 Age after treatment application, years Eq u iv al en t N o . o f r o u gh o r v er y w id e tr an sv er se cr ac ks pe r 30 m (10 0 ft) (E qT CR ) App Age = 1 App Age = 3 App Age = 5 App Age = 7 App Age = 9 App Age = 11 App Age = 13 Linear (App Age = 1) Linear (App Age = 3) Linear (App Age = 5) Linear (App Age = 7) Linear (App Age = 9) Linear (App Age = 11) Linear (App Age = 13) Upper Benefit Cutoff Value Figure 24. Illustration of post-treatment performance relationships based on KDOT models.

48 Output Data Pavement Surface Type: HMA Treatment Type: Rout and Seal Cracks Application Years: 1, 3, 5, 7, 9, 11, 13 Expected Do-Nothing Service Life (yrs): 9.7 Benefit Summary Individual Benefit Summary Benefit Ranking Factors => 100 Application Age, yrs Total Benefit EqTCR 1 0.81 0.81 3 1.01 1.01 5 1.15 1.15 7 1.22 1.22 9 1.21 1.21 11 1.12 1.12 13 1.02 1.02 Cost Summary Application Age, yrs Treatment Cost, PW $ User Cost, PW $ Other Maintenance Cost, PW $ Rehab. Cost, PW $ Total Present Worth, $ EUAC, $ 1 $2,941 n/a n/a n/a $2,941 $243 3 $2,827 n/a n/a n/a $2,827 $214 5 $2,717 n/a n/a n/a $2,717 $191 7 $2,612 n/a n/a n/a $2,612 $171 9 $2,510 n/a n/a n/a $2,510 $154 11 $2,413 n/a n/a n/a $2,413 $140 13 $2,319 n/a n/a n/a $2,319 $128 Effectiveness Summary Application Age, yrs Effectiveness Index Total Benefit EUAC, $ Expected Life, yrs Expected Extension of Life, yrs 1 41.38 0.81 $243 14.0 4.3 3 58.89 1.01 $214 15.5 5.7 5 75.45 1.15 $191 16.9 7.2 7 89.09 1.22 $171 18.4 8.7 9 97.77 1.21 $154 19.9 10.2 11 100.00 1.12 $140 21.3 11.6 13 99.24 1.02 $128 22.7 13.0 TABLE 31 Analysis results for Case Study 2—Kansas

49 Effectiveness Index vs. Application Timing 0 20 40 60 80 100 0 2 4 6 8 10 12 14 Timing of First PM Application, years Ef fe ct iv en es s In de x Extension of Life vs. Application Timing 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 0 2 4 6 8 10 12 14 Timing of First PM Application, years Ex te ns io n of L ife , ye ar s EUAC ($) vs. Application Timing $ $50 $100 $150 $200 $250 $300 0 2 4 6 8 10 12 14 Timing of First PM Application, years EU AC ($ ) Total Benefit vs. Application Timing 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 0 2 4 6 8 10 12 14 Timing of First PM Application, years To ta l B en ef it, % Figure 25. Summary charts for Case Study 2—Kansas.

Case Study 3—Michigan Introduction Michigan DOT has a well-documented preventive main- tenance program with many years of experience. Much of MDOT’s preventive maintenance is applied through a capi- tal preventive maintenance (CPM) program aimed at pro- tecting the pavement structure, slowing the rate of pavement deterioration, and correcting pavement surface deficiencies, mostly through the use of surface treatments. The CPM guide- lines indicate that preventive maintenance projects should be relatively simple and should focus on pavement structures with more than 2 years of remaining service life. Severely distressed pavement structures or pavements with a severely distorted cross section are generally not candidate projects for the CPM program (11). MDOT provided data for 56 preventive maintenance proj- ects of HMA pavements. Much of the data were from a report documenting a 3-year evaluation of MDOT’s Capital Pre- ventive Maintenance Projects (26). Table 32 presents evalu- ation details of four treatment types that were initially con- sidered for use in this project. Specific types of data available for each project consist of the following: • Project location data (route number, project number, MDOT region, beginning and ending mileposts, project length), • Construction history (pavement type, initial construc- tion type and year, rehabilitation and treatment history), • Traffic information (1993/1994 and 1997 ADT), • Distress data, and • Computed remaining service life (RSL). Conventional chip seal and crack sealing data were selected for evaluation. Descriptions of these two activities (as pre- sented in MDOT’s CPM Manual) are included below (11). Conventional (Single) Chip Seals. A single chip seal is defined as an application of a polymer modified asphalt emul- sion with a cover aggregate. The purpose of a chip seal is to 50 • Seal and retard the oxidation of an existing pavement surface, • Improve skid resistance, • Seal fine surface cracks in the pavement, thus reducing the intrusion of water into the pavement structure, and • Retard the raveling of aggregate from a weathered pave- ment surface. The existing pavement should exhibit a good cross section and a good base. The visible distress may include (1) slight ravel- ing and surface wear, (2) longitudinal and transverse cracks with a minor amount of secondary cracking and a slight ravel- ing along the crack face, (3) first signs of block cracking, or (4) slight to moderate flushing or polishing and/or an occa- sional patch in good condition. MDOT reports an expected life extension of 3 to 6 years from a chip seal application on a flexible pavement. Crack Sealing of Bituminous Surfaces. MDOT specifies a “cut and seal” technique to seal cracks on bituminous pave- ments. This method consists of cutting the desired reservoir shape at the working crack in the existing bituminous sur- face, cleaning the cut surfaces, and placing the specified sealant into the cavity to prevent the intrusion of water and incompressible material. The existing bituminous surface should be a relatively newly placed surface on a good base with a good cross sec- tion. On a flexible base, the bituminous surface should be 2 to 4 years old, and 1 to 2 years old on a composite pave- ment. The visible surface distress may include fairly straight, open longitudinal and transverse cracks with slight secondary cracking and slight raveling at the crack face, and no patch- ing or very few patches in excellent condition. MDOT reports an expected life extension of up to 3 years on a flexible pave- ment as a result of crack sealing. However, it is noted that in order to remain effective, this treatment should be followed by routine maintenance crack sealing operations when addi- tional cracks develop. Treatment Costs Average cost data for the two chosen treatment types are listed in Table 33 (26). Treatment Year of Evaluation Number of Projects Evaluated Construction Years of Selected Projects Conventional (single) chip seals 1999 17 1994 to 1995 Crack sealing of HMA surfaces 1999 12 1994 to 1995 Non-structural HMA overlays without milling 2000 13 1995 to 1997 Double chip seals 2001 14 1995 to 2000 TABLE 32 Summary of projects (for selected treatment types) included in a recent evaluation of MDOT’s Capital Preventive Maintenance Program

Condition Indicators MDOT performance data are expressed in terms of a dis- tress index (DI) and ride quality index (RQI). DI is a mea- sure of the extent of surface distress and is expressed on a 0 to 100 scale, where a value of 0 represents a pavement with no distress. MDOT uses DI to determine the RSL of a pave- ment, that is, the number of years left to reach a threshold DI value of 50 (27). RQI is an objective measure of ride quality computed from the power spectral density (PSD) of the road surface profile. Table 34 summarizes Michigan’s RQI ranges and associated subjective ride quality rating. For the projects included in this analysis, 1999 RSL val- ues are provided for the conventional chip seal and crack sealing projects and 2000 RSL values are provided for the non-structural overlay projects. RSL values are not available for the double chip seals, making it very difficult to deter- mine meaningful performance relationships without addi- tional monitoring data. The RSL data are used to complete estimated linear perfor- mance trends by defining the pavement age at which the pave- ment is expected to reach a terminal DI value of 50. Because RSL is only a function of DI, RQI could not be used as a con- dition indicator. RSL data were not available for some sections. MDOT has also investigated the practice of sealing cracks prior to the placement of conventional chip seals on bituminous surfaced pavements; the database includes sections both with and without presealing. Because of the limited number of sec- tions for which data are available, the current conventional chip seal data groups all projects together regardless of whether they received presealing. Benefit Cutoff Values To remain consistent with the DI threshold used for RSL, an upper benefit cutoff value of 50 is chosen for use in the 51 analysis. However, the benefit calculations are not limited on the lower end (i.e., a lower benefit cutoff value of 0 is used for the analysis). Do-Nothing Performance Curves A linear do-nothing curve is assumed for this analysis because no data were available to support the use of an alter- native. This relationship is defined by the line that passes through DI = 0 at an age of zero, and DI = 50 (terminal DI value) at an assumed age of 13 years (see Figure 26). Thus, the linear equation representing the do-nothing DI versus age relationship is as follows: DI = 3.8462 × Age (Eq. 14) Post-Preventive Maintenance Performance Relationships Available data are listed in Table 35. These data are used to determine performance equations for all observed appli- cation ages for conventional chip seals and crack sealing. Based on general observations of the time series perfor- mance data, engineering judgment is used to choose linear regression equations to fit the monitoring data associated with each application age. Since the initial DI rating is always zero, the linear model equation will take the form DI = m × (TAGE), where m is the slope of the line and TAGE is the age of the treatment (i.e., years since placement). The determined regression equations are listed in Table 36. Charts showing the post–treatment performance trends for Treatment Type Average Cost, $/lane-km ($/lane-mi) Year of Cost Data Conventional (single) chip seals $7,603 ($12,240) 1998 Crack sealing of bituminous surfaces (bituminous crack treatment) $4,288 ($6,900) 1998 TABLE 33 Average treatment cost data RQI Range Subjective Ride Quality 0 to 30 Excellent 31 to 54 Good 55 to 70 Fair > 70 Poor TABLE 34 RQI ranges and their subjective ride quality ratings DI = 3.8462 * Age 0 10 20 30 40 50 60 70 80 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Pavement Age, years Di st re ss In de x (D I) Figure 26. Assumed distress index do-nothing curve for Case Study 3—Michigan.

52 Treatment Type Application Age or Age Range Regression Equation Computed Treatment Life Until Equation Reaches Upper Benefit Cutoff Level (DI = 50), years 10 DI = 10.05 × TAGE 5.0 11 DI = 7.4447 × TAGE 6.7 Conventional (single) chip seals 12 DI = 8.26685 × TAGE 6.0 3 DI = 4.825 × TAGE 10.4 4 DI = 3.3814 × TAGE 14.8 5 DI = 3.2394 × TAGE 15.4 7 DI = 6.6536 × TAGE 7.5 Crack sealing of bituminous surfaces 8 DI = 5.0327 × TAGE 9.9 TABLE 36 Treatment performance relationships 0 10 20 30 40 50 60 70 80 0 5 10 15 20 Treatment Age, yrs Di st re ss In de x, DI App Age = 10 yrs App Age = 11 yrs App Age = 12 yrs Linear (App Age = 10 yrs) Linear (App Age = 11 yrs) Linear (App Age = 12 yrs) Upper Benefit Cutoff Value Figure 27. Post-treatment performance trends for chip seals applied at different ages. Treatment Type No. of Sections with Meaningful RSL Values Construction Years of Selected Projects Application Ages of Data Conventional (single) chip seals 17 1994 to 1995 10, 11, 12 Crack sealing of bituminous surfaces 12 1994 to 1995 3, 4, 5, 7, 8 TABLE 35 Construction history analysis for the preventive maintenance sections conventional chip seals and bituminous crack sealing are shown in Figures 27 and 28, respectively. Analysis Setup Because two different treatments are considered, two sep- arate analyses are conducted. Specifically, the analyses are performed using the following inputs and assumptions: • Analysis Type—A detailed analysis type is selected for both analyses since actual data are being analyzed. • Condition Indicators—A custom condition indicator is defined and labeled Distress Index for both analysis sessions. • Preventive Maintenance Treatment Selection—The treatments defined for the two different analyses are Chip Seals and Crack Sealing, respectively.

53 0 10 20 30 40 50 60 70 80 0 5 10 15 20 Treatment Age, yrs Di st re ss In de x, DI App Age = 3 yrs App Age = 4 yrs App Age = 5 yrs App Age = 7 yrs App Age = 8 yrs Linear (App Age = 3 yrs) Linear (App Age = 4 yrs) Linear (App Age = 5 yrs) Linear (App Age = 7 yrs) Linear (App Age = 8 yrs) Upper Benefit Cutoff Value Figure 28. Post-treatment performance trends for bituminous crack sealing applied at different ages. Output Data Pavement Surface Type: HMA Treatment Type: Chip seal Application Years: 10, 11, 12 Expected Do-Nothing Service Life (yrs): 13.00 Benefit Summary Individual Benefit Summary Benefit Ranking Factors => 100 Application Age, yrs Total Benefit Distress Index (DI) 10 0.33 0.33 11 0.49 0.49 12 0.46 0.46 Cost Summary Application Age, yrs Treatment Cost, PW $ User Cost, PW $ Other Maintenance Cost, PW $ Rehab. Cost, PW $ Total Present Worth, $ EUAC, $ 10 $8,268.91 n/a n/a n/a $8,268.91 $744.62 11 $7,950.87 n/a n/a n/a $7,950.87 $634.99 12 $7,645.07 n/a n/a n/a $7,645.07 $602.80 Effectiveness Summary Application Age, yrs Effectiveness Index Total Benefit EUAC, $ Expected Life, yrs Expected Extension of Life, yrs 10 56.99 0.33 $744.62 15.0 2.0 11 100.00 0.49 $634.99 17.7 4.7 12 98.16 0.46 $602.80 18.0 5.0 TABLE 37 Analysis results of chip seal for Case Study 3—Michigan

• Performance Relationships—For both analyses, the do-nothing performance relationship shown in Figure 26 and the respective post-treatment performance relation- ships defined in Table 36 are used. • Project Definition—A typical project size is defined as 1.6 km (1 mi) long. • Cost Data—Only treatment costs are included in the cost analysis (i.e., rehabilitation, user, and routine main- tenance costs are excluded). The unit costs per mile are listed in Table 33; a discount rate of 4.0 percent is used. • Benefit Weighting Factors—Since only one condi- tion indicator is used in each analysis session, the ben- efit weighting factor associated with the DI is set to 100 percent. Analysis Results The analysis results for the chip seal and crack sealing treatments are presented separately. 54 Chip Seal Example. The results of the chip seal analysis are listed in Table 37. These results indicate that of the three investigated application ages, applying the treatment at age 11 is the most cost-effective option as indicated by an EI of 100, although the application age of 12 produced a greater life extension (5.0 years) and a smaller EUAC ($603) than the year 11 timing option. Note that the EI for the year 12 timing scenario is 98.16, which is very close to 100. Therefore, for all practical purposes, a chip seal applied at year 12 is likely to be as effective as a chip seal applied at year 11. Crack Sealing Example. The results of the crack sealing analysis are listed in Table 38. These results indicate that of the five investigated application ages, applying the treatment at age 5 is the most cost-effective option as indicated by an EI of 100. This timing scenario not only produces the largest total benefit value (0.81) and the largest extension of life (7.4 years), it also has the second lowest EUAC at $2,427. The second most effec- tive timing scenario is the year 8 application with an EI of 78.55. The large difference between the first and second timing Output Data Pavement Surface Type: HMA Treatment Type: Crack sealing Application Years: 3, 4, 5, 7, 8 Expected Do-Nothing Service Life (yrs): 13.00 Benefit Summary Individual Benefit Summary Benefit Ranking Factors => 100 Application Age, yrs Total Benefit Composite Index 3 0.21 0.21 4 0.66 0.66 5 0.81 0.81 7 0.37 0.37 8 0.62 0.62 Cost Summary Application Age, yrs Treatment Cost, PW $ User Cost, PW $ Other Maintenance Cost, PW $ Rehab. Cost, PW $ Total Present Worth, $ EUAC, $ 3 $6,134.08 n/a n/a n/a $6,134.08 $601.51 4 $5,898.15 n/a n/a n/a $5,898.15 $452.51 5 $5,671.30 n/a n/a n/a $5,671.30 $411.46 7 $5,243.43 n/a n/a n/a $5,243.43 $483.19 8 $5,041.76 n/a n/a n/a $5,041.76 $399.26 Effectiveness Summary Application Age, yrs Effectiveness Index Total Benefit EUAC, $ Expected Life, yrs Expected Extension of Life, yrs 3 17.38 0.21 $601.51 13.4 0.4 4 74.01 0.66 $452.51 18.8 5.8 5 100.00 0.81 $411.46 20.4 7.4 7 38.44 0.37 $483.19 14.5 1.5 8 78.55 0.62 $399.26 17.9 4.9 TABLE 38 Analysis results of crack sealing for Case Study 3—Michigan

scenario choices suggests that the year 5 application is the far more cost-effective choice for applying crack sealing. Fig- ure 29 shows plots of EI, extension of life, and EUAC versus treatment application age for this analysis. While an age of 5 years is the suggested application age based on the default analysis approach (i.e., analyzing bene- fit and cost simultaneously), this may not represent the phi- losophy of all agencies. For example, if the benefit differences in the crack sealing example were considered insignificant, an application age of 8 years would become most appropriate as it provides the lowest EUAC value. Therefore, it is always important for an agency to consider the analysis results in conjunction with other established goals. 55 Case Study 4—North Carolina Introduction North Carolina Department of Transportation (NCDOT) provided project data for 10 HMA sections, including pave- ment condition rating (PCR) history, treatment type, year of treatment application, the (estimated) year of the previous maintenance treatment, and pavement structure (from cor- ing) for 5 of the 10 sections. Treatments organized by type and DOT division were obtained from NCDOT’s “2001 Road Oil Summary (28).” Cost information was obtained from NCDOT pavement management unit staff. Treatment Selection Two different asphalt seal coats (Triple Seal and Split Seal) were used on the 10 projects as preventive maintenance treatments. Split Seal treatment was used on 8 projects and Triple Seal treatment was used on 2 projects. Construction details described in Section 660 of North Carolina’s State Construction Handbook (29) are summarized. Split Seal. A split seal consists of two applications of asphalt binder and aggregate. Total binder and aggregate application rates are approximately 2.04 to 2.26 L/m2 (0.45 to 0.50 gal/yd2) and 16 to 19 kg/m2 (30 to 35 lb/yd2), respectively. In the first application, approximately 0.91 to 1.13 L/m2 (0.20 to 0.25 gal/yd2) of asphalt material is applied to the existing surface, followed immediately by the application of approx- imately 11 to 12 kg/m2 (20 to 22 lb/yd2) of seal coat aggre- gate spread uniformly over the treated surface. Immediately after the first application of seal aggregate has been made uniform, the remainder of the required amount of asphalt material and seal coat aggregate are applied and the seal coat is rolled; specific rolling instructions are provided in Section 660 of the handbook (28). Triple Seal. To construct a triple seal, approximately 0.91 to 1.13 L/m2 (0.20 to 0.25 gal/yd2) of liquid asphalt is applied to the existing surface followed immediately by the appli- cation of approximately 8 to 9 kg/m2 (15 to 17 lb/yd2) of seal coat aggregate spread uniformly over the treated surface. The operation is performed three times; aggregate applied in the final application is then rolled as described in the handbook (29). Treatment Costs Treatment costs are summarized in NCDOT’s “2001 Road Oil Summary (28).” Relevant details are provided in Table 39 for both treatment types. Effectiveness Index vs. Application Timing 0 20 40 60 80 100 0 2 4 6 8 10 Timing of First PM Application, years Ef fe ct iv en es s In de x EUAC ($) vs. Application Timing $ $100 $200 $300 $400 $500 $600 $700 0 2 4 6 8 10 Timing of First PM Application, years EU AC ($ ) Extension of Life vs. Application Timing 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 0 2 4 6 8 10 Timing of First PM Application, years Ex te n si on o f L ife , ye ar s Figure 29. Summary charts of crack sealing for Case Study 3—Michigan.

56 Treatment Type No. of Divisions with Data Length of Preservation Projects, km Area of Preservation Projects, m2 Total Cost of Preservation Projects, $ Average Unit Cost for Data from All Divisions, $/m2 Range of Average Unit Costs Determined for Each Division, $/m2 Split Seal 12 1,166 6,154,018 $2,346,429 $0.84 $0.75 to 1.05 Triple Seal 9 88 475,022 $643,093 $1.24 $0.97 to 1.65 Note: 1 mi = 1.61 km; 1 yd2=0.84 m2 TABLE 39 Summary of 2001 treatment cost data PCR = -1.6506 * Age + 100 R2 = 0.2605 0 20 40 60 80 100 120 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Pavement Age, years Pa v em en t C o n di tio n Ra tin g (P CR ) Figure 30. Assumed representative do-nothing curve for the North Carolina projects. Condition Indicator Time series PCR data are provided for the 10 sections. The pavement condition rating is a composite index that reflects the extent of surface distress, expressed on a 0 to 100 scale (a value of 100 represents a pavement with no distress). Benefit Cutoff Values Based on the pavement condition time-series data, a lower benefit cutoff value of 70 is selected, suggesting that when the condition falls below 70, a second asphalt seal coat is trig- gered. The benefit calculations are not subjected to any limit on the upper end (i.e., an upper benefit cutoff value of 100 is used for the analysis). Do-Nothing Performance Curves All analyzed preventive asphalt seal coats were placed on pavements that already had a Mat and Seal treatment applied. To simplify the analysis, the performance of the existing Mat and Seal layers is defined as the do-nothing performance. Thus the do-nothing performance curve is defined by the time series performance data from the Mat and Seal layer application year (defined as year 0) to the application year of the first preven- tive asphalt seal coat. A representative do-nothing curve is then assumed for the analysis by fitting a linear equation through this time series data (through a value of 100 at time zero) as shown in Figure 30. To check the reasonableness of this approach, the age at which the resulting regression equation (PCR = −1.6506 × Age + 100) crosses the assumed condition trigger level of 70 is determined. The expected age at this trigger value is 18.2 years, which is reasonable. Post-Preventive Maintenance Performance Relationships Construction and maintenance history of the 10 sections is summarized in Table 40. It appears there is a definitive rela- tionship between the timing of the first and second preventive maintenance treatments (see Figure 31). The trend indicates that the life of the first preventive maintenance treatment is longer when applied sooner rather than later after initial construction. As shown in Table 40, monitoring data associated with first treatment application ages of 4, 5, 8, 9, 11, 13, and 14 years are available (two sections with unknown construction history were ignored). Three additional sections were elimi- nated from the analysis. One of these sections with an appli- cation age of 8 years was eliminated because the monitoring data for the section did not appear to be representative; treat- ment condition deteriorated at a much more rapid rate than any other sections. Two other sections with application ages of 11 years were eliminated because the data showed com-

57 State Route Assumed Original Construction Year First Preventive Maintenance Treatment Year Last Preventive Maintenance Treatment Year Age at Timing of First Preventive Maintenance Treatment, yrs Time from First Treatment Application to Last Treatment Application, yrs SR 1125 1984 1992 1999 8 7 SR 1226 1982 1986 2000 4 14 SR 2249 1980 1991 2000 11 9 SR 2018 1982 1993 2000 11 7 SR 2028 1980 1989 2000 9 11 SR 1828 1982 1996 2002 14 6 SR 1722 Unknown SR 1721 1983 1988 2002 5 14 SR 2245 1982 1995 2002 13 7 SR 1719 Unknown TABLE 40 Construction history analysis for the 10 asphalt seal coat sections 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 Timing of First Asphalt Seal Coat (time since original construction) Ti m e fro m fi rs t P M a pp lic at io n un til s ec on d PM ap pl ic at io n, y ea rs Figure 31. General trend between the life of the first preventive maintenance treatment and its application timing (time after initial construction). Application Age Regression Equation Computed Life Until Equation Reaches Lower Benefit Cutoff Level (PCR = 70) 4 PCR = 100 – 0.007535 * AGE 2.747556 20.4 5 PCR = 100 – 0.333097 * AGE 1.511787 19.6 9 PCR = 100 – 5.618E-12 * AGE 11.474446 12.9 13 PCR = 100 – 0.000138 * AGE 6.191617 7.3 14 PCR = 100 – 0.020379 * AGE 3.969362 6.3 TABLE 41 Post-treatment performance relationships for Case Study 4—North Carolina

Output Data Pavement Surface Type: HMA Treatment Type: Asphalt Seal Coat Application Years: 4, 5, 9, 13, 14 Expected Do-Nothing Service Life (yrs): 18.18 Benefit Summary Individual Benefit Summary Benefit Ranking Factors => 100 Application Age, yrs Total Benefit Pavement Condition Rating 4 1.04 1.04 5 0.77 0.77 9 1.05 1.05 13 0.61 0.61 14 0.50 0.50 Cost Summary Application Age, yrs Treatment Cost, PW $ User Cost, PW $ Other Maintenance Cost, PW $ Rehab. Cost, PW $ Total Present Worth, $ EUAC, $ 4 $14,531.67 n/a n/a n/a $14,531.67 $943.00 5 $13,972.76 n/a n/a n/a $13,972.76 $902.39 9 $11,943.97 n/a n/a n/a $11,943.97 $829.87 13 $10,209.76 n/a n/a n/a $10,209.76 $744.37 14 $9,817.08 n/a n/a n/a $9,817.08 $715.75 Results Application Age, yrs Effectiveness Index Total Benefit EUAC, $ Expected Life, yrs Expected Extension of Life, yrs 4 87.43 1.04 $943.00 24.4 6.3 5 68.04 0.77 $902.39 24.6 6.5 9 100.00 1.05 $829.87 21.9 3.7 13 64.84 0.61 $744.37 20.3 2.1 14 55.32 0.50 $715.75 20.3 2.1 58 60 65 70 75 80 85 90 95 100 105 0 2 4 6 8 10 12 14 16 18 20 22 Age After Treatment Application, yrs Pa ve m en t C o n di tio n R a tin g (P CR ) AppAge = 4 (regression) AppAge = 5 (regression) AppAge = 9 (regression) AppAge = 13 (regression) AppAge = 14 (regression) AppAge = 4 (raw data) AppAge = 5 (raw data) AppAge = 9 (raw data) AppAge = 13 (raw data) App Age = 14 (raw data)Lower Benefit Cutoff Value Figure 32. Determined post-treatment performance relationships for Case Study 4— North Carolina. TABLE 42 Analysis results for Case Study 4—North Carolina

59 pletely different rates of deterioration, without clarification. The analysis compares the expected post-preventive mainte- nance trends associated with assumed application ages of 4, 5, 9, 13, and 14 years after initial construction. Based on general observations of the time series perfor- mance data, engineering judgment was used to select an exponential regression equation to fit the monitoring data for each of the application ages. The exponential model type is a good choice for a decreasing trend that has a known start- ing condition value. In this case, the initial PCR is always 100; therefore, each of the individual post-preventive main- tenance relationships must yield a value of 100 at an age of zero. The equation form PCR = C − m × (Age)P was selected; the specific regression equations are listed in Table 41 and plotted (along with the reported data) in Figure 32. Analysis Setup The following inputs are used in the analysis: • Analysis Type—A detailed analysis type is selected since actual data are being analyzed. • Condition Indicators—A custom condition indicator is defined and labeled Pavement Condition Rating. • Preventive Maintenance Treatment Selection—A cus- tom treatment named Asphalt Seal Coat applied at ages of 4, 5, 9, 13, and 14 years is investigated. • Performance Relationships—The do-nothing perfor- mance curve shown in Figure 30 and the post-preventive maintenance performance relationships defined in Table 41 are used. • Project Definition—A typical project size is defined to be 15,290 m2 (20,000 yd2). • Cost Data—Only treatment costs are included in the cost analysis (i.e., rehabilitation, user, and routine main- tenance costs are excluded). Because data for three split seal and two triple seal projects are used, a treatment unit cost of $1.02/m2 ($0.85/yd2) is chosen for this analy- sis, because it is within the observed cost ranges for both treatment types. The selected project size and unit cost would yield a total cost of $17,000 for each treatment application; a discount rate of 4.0 percent was also cho- sen for the analysis. • Benefit Weighting Factors—Because only one con- dition indicator is used in the analysis session, the ben- efit weighting factor associated with the PCR is set to 100 percent. Analysis Results Results of the analysis are summarized in Table 42. These results indicate that out of the five investigated application Effectiveness Index vs. Application Timing 0 10 20 30 40 50 60 70 80 90 100 0 5 10 15 Timing of First PM Application, years Ef fe ct iv en es s In de x Extension of Life vs. Application Timing 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 0 5 10 15 Timing of First PM Application, years Ex te ns io n of L ife , y ea rs EUAC ($) vs. Application Timing $ $100 $200 $300 $400 $500 $600 $700 $800 $900 $1,000 0 5 10 15 Timing of First PM Application, years EU A C ($) Figure 33. Summary charts for Case Study 4— North Carolina. ages, applying the treatment at age 9 is the most cost-effective choice as indicated by an EI of 100. At this application age 9, a life extension of 3.7 years is expected (i.e., the pavement will last 3.7 more years than the 18.2 years expected if no treat- ment is applied), with an EUAC of approximately $830. The largest expected extension of life (6.5 years) is with a treatment applied at age 5 that provides the second highest EUAC and the

60 • Oxidation—viscosity and penetration of asphalt from recovered cores (available for a very limited number of sections) For rigid pavements, 43 SPS-4 sections were found to have data for crack and joint sealing treatments. However, these maintenance activities were combined in these sec- tions, making it difficult to isolate the separate effect of each treatment. For these sections, the following types of condi- tion indicator data are available: • Cracking • Joint spalling • Faulting • IRI • Friction An initial review of the collected data was conducted to determine their usefulness in evaluating the optimal timing methodology. Specifically, it was considered essential to have data available on the performance of the pavement after application of a specific preventive maintenance treatment to compare with the performance of a control section that did not receive the treatment (i.e., the do-nothing trends). This review concluded that the LTPP data could not be used to con- duct a meaningful analysis for several reasons. The perfor- mance trends for a large number of sections revealed counter- intuitive trends (e.g., untreated control sections performing better than adjacent sections that received a preventive main- tenance treatment [see Figure 34]). Because these sections did not show an improvement in performance as a result of treatment application, they were not studied further. Also, sections that were not in good condition when treatment was applied were excluded because they do not meet the defini- tion of preventive maintenance. With this, there were not enough remaining sections with treatments applied at differ- ent ages that exhibited the expected trends to support a mean- ingful analysis. SUMMARY A product of this research was the development of a methodology that can be used to determine the optimal time to apply preventive maintenance treatments. The methodology is based on an understanding of how pavements perform over time and how preventive maintenance affects its performance. By analyzing appropriate performance data from pavements treated at a variety of times, it is possible to identify the “right” time to apply preventive maintenance. That “right” time, iden- tified through the optimal timing methodology, is defined as the time when the treatment’s application provides the great- est ratio of improvement in condition (benefit) to cost (i.e., that time with the largest associated B/C ratio). third highest EI at 68.04; the second largest effectiveness (87.43) is with an application age of 4 years. The results of this analysis session, shown in Figure 33, illustrate EI, extension of life, and EUAC versus treatment application age relationships. Case Study 5—LTPP Data This example involves the use of data from the LTPP SPS-3 and SPS-4 experiments. In these experiments, main- tenance treatments were applied to both HMA and PCC pavements and performance of these pavements and nearby control sections was monitored over time. Using the LTPP DataPave 3.0 software program, LTPP data were examined to identify maintenance effectiveness test sections meeting the following requirements: • Have “adequate” time series data—in order to establish accurate condition indicator trends over time, only sec- tions with three or more time series data points were included. • Be applied on pavements in “Good” condition—since preventive maintenance treatments are applied to pave- ment in “Good” condition, only sections with treatment applied to a pavement in “good” condition were included. • Have condition data before first preventive mainte- nance treatment application—in order to determine the initial impact of a preventive maintenance treatment on condition, sections that had condition information in the year immediately prior to the preventive mainte- nance treatment applications were included. • Use of control section—in order to assess the impact of preventive maintenance on pavement performance its expected service life, all sections suitable for this eval- uation must have data associated with a “control” sec- tion to define the do-nothing performance trend. For flexible pavements, the initial search of the database identified the following SPS-3 sections as meeting these criteria: • 80 sections with chip seal coats • 80 sections with slurry seal coats • 69 sections with crack sealing • 79 sections with (thin) overlays The following types of condition indicator data are available for each of these sections: • Nonload-related and load-related cracking • Average rut depth • IRI • Friction

61 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 12/23/1988 5/7/1990 9/19/1991 1/31/1993 6/15/1994 10/28/1995 3/11/1997 7/24/1998 12/6/1999 Survey Date IR I (m /km ) 48 1310 (w/Thin Overlay) 48 1340 (Control) Figure 34. Example of LTPP data exhibiting a counterintuitive trend for a thin overlay application. To assist in the implementation of the methodology, OPTime, a Microsoft® Excel-based analysis tool capable of analyzing actual preventive maintenance-related performance data, was developed. The analysis tool greatly facilitates the application of the methodology through a logical, step-by- step, input sequence. Further explanation of the optimal tim- ing approach and a detailed user’s guide is provided in Appen- dix C, which is available to users by accessing the NCHRP website (http://trb.org/news/blurb_detail.asp?id=4306). Data were collected from four SHAs and from the LTPP SPS-3 and SPS-4 experiments for possible use in an analy- sis to validate the optimal timing approach and to demon- strate the use of the OPTime tool. These data were analyzed using OPTime although the results of the analyses did not always match expectations. Data from the LTPP experi- ments were not analyzed because the data did not support the premise that the maintenance treatments improved per- formance compared with the do-nothing case. A holistic approach to identifying the optimal time of preventive main- tenance application is needed. Such an approach should address project selection, treatment selection, pavement per- formance monitoring, and data analysis and reporting. These observations are further described in Chapter 4 as part of Suggested Research.

Next: Chapter 4 - Conclusions and Suggested Research »
Optimal Timing of Pavement Preventive Maintenance Treatment Applications Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB's National Cooperative Highway Research Program (NCHRP) Report 523: Optimal Timing of Pavement Preventive Maintenance Treatment Applications describes a methodology for determining the optimal timing for the application of preventive maintenance treatments to flexible and rigid pavements. NCHRP Report 523 also presents the methodology in the form of a macro-driven Microsoft Excel Visual Basic Application--designated OPTime.

OPTime User’s Guide

Download the OPTime CD-ROM (.ISO) Image

Help on burning a .ISO CD Image

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!