Click for next page ( 90


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 89
CHAPTER 5. INVESTIGATION OF HIGHWAY MAINTENANCE QA AND LONG-TERM PERFORMANCE OF HIGHWAY FACILITIES Introduction Establishing and operating a maintenance QA program can be a substantial inves~anent for a highway agency. As such, Me QA program must show a lasting beneficial effect on the overall condition of highway facilities and We perception of Cat condition by He traveling public. The program must result in a higher quality highway system Bat in He long term provides benefits greater Han He costs of He program. Such benefits will only occur if substantial improvements in long-term performance of highway facilities are made. The long-term performance of highway facilities can be judged In terms of He following four characteristics, which, in essence, are the four main considerations of maintenance: Safety. Comfort and convenience. Aesthetics. Preservation of investment (service life). The maintenance QA program developed under NCE]RP Project ItI2 was specifically designed to evaluate He characteristics most apparent to He traveling public: safety, comfort and convenience, and aesthetics. An implementing agency, however, has He additional responsibility of optimizing He preservation of Heir investments, be Hey pavements, bridges, or over major highway features. This aspect of facility performance, which is not immediately discernable to He vast majority of highway users, is a major focus in every transportation agency, as evidenced by the development of BMS's and PMS's. Highway agencies, therefore, are interested in implementing a QA program Hat will not only satisfy He user's needs by achieving and maintaining He desired LOS, but one Cat shows a positive influence on He preservation of ~nveshnent. In over words, Heir desire is for a QA program that has He dual effect of ensuring quality to He traveling public (~rough safe, comfortable, and visually pleasing roadway facilities) and to He highway agency (through longer lasting and more cost~ffective roadway facilities). How an agency can determine whether He QA program effectively and economically preserves key transportation inves~anents is not an easy task, as there are numerous factors Hat can confound He {ong-term performance of highway facilities. Perhaps He most significant factor in long-term performance is the aDocation of highway funds among He various transportation infrastructure components (design, 89

OCR for page 89
construction, maintenance, rehabilitation). Over major factors include roadway design (type of materials, Sicknesses, and so on), construction (quality of materials and workmanship, conditions during construction, and so on), maintenance, Me amount of traffic and heavy loads carried by We roadway facility, and We climatic conditions in which We facility functions. These factors affect We rate of deterioration of a roadway facility, and the identification of a method mat substantively links We quality of maintenance with long-term performance would be a valuable tool to maintenance managers. This chapter discusses We investigative work effort undertaken to identify a process for tracking We relationship between maintenance LOS ratings and We long-term performance of highway features. Several tracking methods were examined and are briefly described In We section below, titled "Work Approach." Two methods were considered to have We most potential for relating maintenance quality and facility performance and were, therefore, evaluated in greater detail using actual maintenance LOS rating data and highway performance data collected from two States. These two methods are formally presented in We QA program Implementation Manual, but We results of We detailed evaluations regarding their effectiveness and usefulness are provided In this chapter. The last section in this chapter provides a discourse on We merits and ramifications of Me ideas brought to light in Me investigation of QA Output-long-term performance tracking relationships. Work Approach In the search for an appropriate method for tracking Me relationship between maintenance LOS ratings and Me long-term performance of highway features, a number of objectives were used to define a successful methodology. First, it was desired Mat Me methodology make use of readily available performance data contained in existing agency management information systems (PMS's, BMS's, SMS's). Second, Me chosen method was to minimize Me effects of confounding factors (funding, design, traffic) on performance. Third, Me chosen method would be developed by focusing solely on pavement facilities. The decision to focus on pavements was based on Me following factors: Pavement performance is most difficult to assess. That is, Me effects of pavement maintenance on pavement performance are not as easy to assess as Me effects of maintenance on the performance on some of Me other highway features, such as pavement striping, signs, or guardrails. A large amount of Me maintenance budget goes Into maintaining pavements. Highway users are greatly affecter! by Me maintenance applied to pavements. PMS's are an excellent external source for performance clata. Even Cough Me melons presented herein relate specifically to pavement facilities, Me concepts are believed to be applicable to over maintenance features. This is because 90

OCR for page 89
We concepts are based on deterioration, which is a phenomenon experienced by all man-made items. SK different methods for evaluating the effectiveness of the maintenance QA program were identified for investigation. These methods covered a wide array of approaches and complexities meant to accommodate the diverse natures, goals, and management styles of many highway agencies. The six methods are listed below, with a brief description given for each method. Method ~-QA/COS! InJeX. This method is a relatively simple procedure Mat allows an agency to make yearly comparisons of its LOS ratings and Me associated maintenance costs. It allows an agency to first determine whether its maintenance activities are actuary providing Me desired LOS and then evaluate the annual cost of performing Me maintenance activities. The relationship is expressed as a QA/Cost Index and is a function of QA {LOS) rating, maintenance . ~. ~ cost expenditures, and the total lane mileage tor a chosen set of highway pavement sections (e.g., interstate concrete pavements In district 1, AC overlays on primary routes agency wide). The QA/Cost Index is calculated using Me following equation: QA/Cost Index = (QA Rating x Total Mileage)/(F~rst Year $) where: First Year $ Current Year $ n = Eq. 8 Current year's maintenance dollars converted to present-worth dollars corresponding to Me QA program's first year. (Current Year $)/~1 ~ i) Current year's maintenance dollars. Number of years since Me start of the OA do. . program (current year - HA program first years. i = Discount rate. 4 percent fi=0.04)is recommended, based on historical data. QA Rating = QA program rating data representing Me chosen set of pavement projects (i.e., the chosen analysis group). Total Mileage = Total lane mileage of the highway sections ~ ~ ~ .~ ~ ~ - Included in the chosen analysts group. The Inclusion of Total Mileage has a normalizing effect on Me index that allows for year-to-year comparisons for Me same analysis group. Plotted on a yearly basis (as illustrated in figure 6), Me QAICost Index could be examined for meaningful trends. For instance, if Me index increased over time, Me Intuitive interpretation would be Mat Me OA program is becoming more efficient at providing Me desired LOS. 91 _ , ~

OCR for page 89
x - us o 1 1 ~ - . - First Year of QA Program 1 i 1 1 1 1 1 1 1 > Time Figure 6. Example plot of QA/Cost Index versus time. Method 2 Backlog Analysis. A common practice in pavement management is to define a critical pavement condition threshold Cat is used to signal We need for rehabilitation. The critical threshold is generally expressed in terms of roughness (e.g., IRI, MRN), serviceability (e.g., PST, PSR), or visual condition (e.g., PCI), and Me group of pavements win conditions below Me critical threshold are commonly referred to as Me pavement backlog. Since a continuing goal of any agency is to niininiize, or possibly eliminate, its backlog, Me Backlog Analysis method was designed to track not only Me percentage of pavements In Me backlog, but also Me rate Mat pavements deteriorate from one condition range, or performance category, to Me next (i.e., Me accrual rate). The Backlog Analysis method attempts to identify the relationship between maintenance quality and pavement performance by focusing on pavement backlog-related trends. Because higher LOS ratings signify more effective and efficient maintenance, it is logical Mat an increase in pavement LOS should be accompanied by a reduction in the pavement deterioration rate (i.e., Increased pavement performance), all over factors being equal. A reduction in the rate of pavement deterioration would subsequently lessen Me backlog accrual rate, as well as the rates at which pavements move from a higher performance category to a lower one (e.g., Very Good to Good, Good to Fair). Me~od 3-Change in Condition Indicator vs. Time. This method compares Me annual pavement ICES rating for a chosen analysis group to Me average annual change in pavement condition indicator (e.g., IRI, PST, PCI) representative of that group. The memos first entails plotting Me 2- or 3-year condition trends of all pavements in a chosen analysis group, and then developing a best-fit linear trend by computing the average slope of deterioration for Mat group. This is 92

OCR for page 89
illustrated in figure 7, where the change in condition indicator fin this case, condition rating survey [CRS]) for many projects is plotted versus age. The resulting change In CRS exhibited by the linear trend line over a I-year penod (also referred to as Me yearly condition deduct value) is Men cletermined, as shown in figure 8. With deduct values established for multiple years, Me annual pavement LOS ratings for the corresponding years are linked together, so Cat Me resulting trend between pavement maintenance quality and pavement performance can be examined. Figure 9 provides a simple illustration of this concept using performance curves associated win two levels of routine maintenance good and poor. All over factors being equal, a well maintained pavement (one win a high LOS) is expected to have a flatter deterioration slope Man a poorly maintained pavement (one with a low LOS). Subsequently, given a critical condition threshold for rehabilitating pavements, Me flatter slope would result In a longer pavement service life. 9 ~ . . . . , 1~ r3 ~ \ \ 3 z i ~ ~3 Al to 20 3e Age Figure 7. illustration of average rate of deterioration for a group of similar highway pavement sections (ERES, 1995~. 93

OCR for page 89
l ~ ~/ I Yearly CRS deduct , O' ~: / CRS Value from survey - Future Figure 8. illustration of yearly condition deduct (ERES, 1995). PSI)o i - cr, x (PSI)14 ._ - a) O :> h US CO (1) Pavement Performance Deduct Curve Win Good Routine Maintenance. (2) Pavement Performance Deduct Curve Win Poor Routine Maintenance. - - (2) Time, t Figure 9. Relationship between pavement performance and routine maintenance (Fwa and Sinha, 1986). 94

OCR for page 89
Method 4 Chance in Condition Indicator vs. LOS. A slight variation of method 3, ~ ~ _ this method entails plotting Me average annual change in condition indicator (Acondition index) versus Me measured pavement LOS rating for a sequence of time periods (see figure 10). Ideally, as the LOS raking increases, the /`condition index would decrease, suggesting that a higher maintenance LOS corresponds win increased pavement performance. At some point along the LOS rating scale, Me rate of reduction in the Acondition index will decrease win an increase in LOS rating. Hence, a plot of Acondition index versus LOS for a range of LOS ratings may allow an agency to select Me LOS raking that bow meets the desired LOS and optimizes pavement performance. If Were is no relationship found between the Acondition index and the corresponding LOS raUng, then an agency would need to carefully review Me quality of Me EMS data and reevaluate its LOS rating program. Method 5 Condition Indicator vs. LOS. This method Involves plotting condition indictor values win corresponding pavement LOS ratings for individual sections within a chosen analysis group (figure 11). Such plots can help define Me uncertain relationships between maintenance quality ratings and pavement condition, Hereby allowing an agency to estimate He level of maintenance required to bring He average condition of highway pavements to a specified level. Because He "conditions" evaluated as part of He LOS rating program are often different from He "conditions" measured and recorded In a PMS, and because He LOS rating program is attribute-based whereas He PMS is variable-based, a large amount of variability can be expected in He relationship between pavement LOS and He selected condition indicator. Such variability is depicted In figures 12 and 13. Method WRegression Model. This complex method involves conducting simultaneous pavement condition and maintenance condition field evaluations of several "like" pavement sections (i.e., pavements of similar type and located in similar clunatic regions), and then performing a statistical regression analysis to determine He influence of maintenance quality on pavement condition. In this approach, a sufficient number of random sample units within each pavement section are identified and then surveyed, first for pavement condition (PCI) and Hen for maintenance condition (LOS). Using He PC] and I~OS ratings from each sample unit, an average PC] rating and an average l.OS rating for He pavement section are determined. The age of each pavement section at the time of the evaluation is also required. Traffic data (average daily traffic [ADT], equivalent singl~axle loads [ESALs]) may also be used if a wide range in traffic levels is known to exist among He various "like" sections. 95

OCR for page 89
- o ~ o - - 1 1 1 1 1 1 1 1 1 0 50 100 LOS Rating Figure 10. Example plot of Acondition index versus LOS raking. 100 x I_ ~ 50 ._ o r X: X AX x x X /X X X/ Xp X) 1 1 1 0 50 100 LOS Rating Figure Il. Conceptual illustration of pavement condition-LOS rating relationship. 96

OCR for page 89
200 a ~ 100 ._ ~_ x x x x x x ~ x x x x x x~- x x x x y x x x x O o x x x x x x x x x x x x x x~ x x x x xx x x x x x x x x x x v x ^ x x x x x x x x X A ~ X X 100 LOS Rating Figure 12. Conceptual iDustration of variability in roughness-LOS rating relationship. U 50- P~ o / - x x x x x x x x x ~x x x x x ~x x x x x x ~v Y ~ A V V v A A ~SC X ~X X X X A' X X X X OLOS Rahng 100 Figure 13. Conceptual illustration of variability ~n condition-LOS rating relationship. 97

OCR for page 89
One year later, Me same random sample units are again evaluated for pavement condition and maintenance condition. The change In PCT for each section is calculated by subtracting We current year's PC] from me previous year's PCI. A mean LOS rating is computed for each section by averaging Me current year's LOS rating and Me previous year's LOS rating. Table 13 gives an example of Me data used in this type of analysis. These data represent 30 individual projects of a given pavement type (e.g., jointed plain concrete DPC] pavement) and located bra one geographical area (e.~. District Il. ~O . Using this type of data set, a statistical regression analysis is then conducted Mat relates Me maintenance LOS win the change in PCT, as expressed in Me following general form: ~lPC] = aO + AL (~+ a:Age + a3Tra~c where: Am as, al, as, as = L(W Age Traffic = Eq. 9 Annual change in PCI. Regression coefficients. Average LOS rating. Age of pavement at time of second survey, years. Estimated traffic, vehicles/day or equivalent s~ngle- axIe loads. For maintenance LOS to have a significant effect on Me change in PCT, Me resuming probability value (p-value) corresponding to Me a, regression coefficient would have to be less Man the chosen significance level (e.g., 5 percent or 10 percent). If LOS is found to have a significant effect on APCI, then Me type and magnitude of Me effect can be determined by examining me resulting al value. IdeaDy, as, as, and as should be negative, whereas al should be positive. ~ Ads way, increased age and Increased traffic result In larger decreases In PCI, and increased LOS results In smaller decreases in PCI. Preliminary Assessment of Tracking Methods Many factors were taken into consideration when determining Me worthiness of Me six me~ods for tracking Me relationship between maintenance LOS ratings and Me Tong-term performance of highway features. Key factors in this initial evaluation Included Me complexity of Me memos, the availability of required data, and Me associatecl confounding variables (e.g., traffic, climate, design, rehabilitation). 98

OCR for page 89
Table 13. Regression mode! data example. Site No. Locabon 1 2 3 4 5 6 7 . 8 9 10 11 12 13 14 15 - 16 17 18 19 20 21 22 23 24 25 26 27 28 29 l 30 Rt S3 hDP 0-7.6 Rt 53 hDP 7.6-11.2 Rt 53 0P 22.5-25.6 Rt 53 hDP 25.6-32.2 Rt 53 hDP 51.3-60.0 I-17~BB hDP 14.7-20.4 I-17~IB hDP 20.4-24.8 I-17bJB hDP 24.8-30.5 I-17~IB hDP 30.5-40.6 I-17SB MP 14.7-20.4 I-17SB MP 20.4-24.8 I-17SB MP 24.8-30.5 I-17SB hDP 30.5-36.2 Rt 3 hDP 94.7-98.3 Rt 3 hDP 104.7-110.0 Rt 3 hDP 110.0-113.6 Rt 26 hIP 0-3.4 Rt 26 hDP 3.4~6.4 Rt 26 MP 24.3-27.9 Rt 26 hDP 27.9-33.4 Rt 26 MP 40.3~42.3 Rt 26 0P 42.3-46.8 Rt 26 MP 46.8-50.5 US 64EB hLP 174.3-180.3 US 64EB 0P 180.3-188.1 US 64EB 0P 204.5-209.8 US 64VVB hIP 168.6-174.3 US 64VVB hIP 174.3-180.3 US 64VVB hIP 180.3-188.1 US 64VVB MP 204.~-209.8 PCI RaUng LOS Radng 1996 1 1997 APCI 1996 ~ 1997 76 73 -3 88 66 62 80 98 96 -2 100 94 93 -1 100 87 87 0 96 89 87 -2 86 - 67 63 -3 78 50 43 -7 56 46 38 -8 50 69 64 -5 78 74 68 -6 79 83 81 -2 93 80 77 -3 94 . 94 ~ 94 0 100 100 98 -2 100 66 62 -4 79 77 73 -4 84 93 90 -3 94 90 88 -2 100 76 73 -3 86 73 69 ~4 84 74 70 -4 83 45 36 9 54 48 41 -7 58 58 51 -7 65 42 35 -7 52 45 37 -8 58 65 60 -5 73 71 66 -5 80 62 56 -6 80 85 75 97 98 - 92 82 72 51 46 75 l 72 90 88 100 100 74 78 90 92 87 78 l 81 46 54 . 60 . 54 53 67 72 75 Mean | Ag~ | ADT LOS 1 1997 1 1997 ~:~ 77.S 1 10 1 ~L: 94 1 6 1 84 1 6 . l ~ 53.5 1 19 l 48 1 19 .. ~76.5 1 15 . 75.5 . 91.5 1 7 1 91 1 7 ~_ 100 1 3 1 76.5 1 13 ~L~ 92T 5 1 1 1 96 1 5 1 1 1 82 1 9 1 1 1 50 1 17 1 . ~LN 62.5 1 17 53 1 18 1 55.5 1 18 70 1 15 . 76 1 15 1 77.5 1 15 4,490 4,760 4,050 1 4,050 3,870 12,540 11 112,5401 11 12,310 12,070 r 116601 11 11,6601 1 11,380 11,240 2,740 2~430 2,430 - 6,880 6,880 7,240 7,240 . 7,310 _ . 7,310 7,260 8~040 - 8,040 7,920 8,180 1 8,180 8,180 99

OCR for page 89
Table 18. 1991 pavement condition range probability matrix. 1991 Distribution 1992 Year Distribution | Miles of Very ~ Very Condition Pavements Good Good Fair PoorPoor Very Good 3.099 0.00 1.00 0.00 0.00 _ Good 147.043 0.10 0.31 0.43 0.09 Fair 96.590 0.00 0.13 0.53 0.34 Poor 23.904 0.00 0.00 0.10 0.45 Very Poor 77.993 ~0.00 0.00 0.00 0.02 Total ~348.629 ~0.041 0.181 0.33' 0.171 0.25 0.00 0.00 0.00 0.45 0.98 Table 19. 1995 pavement condition range probability matrix. Rehabilitated .oo 0.07 o.oo o.oo o.oo 0.03 1995 Distribution 1996 Year Distribution | Miles of ~Very ~ Very Condition Pavements Good Good Fair PoorPoor Very Good 54.594 0.53 0.45 0.00 0.02 Good! 278.609 0.07 0.75 0.12 0.00 Fair ~26.494 ~0.00 ~0.00 ~0. ~ 7 ~0.00 Poor 10.872 0.00 0.00 0.02 0.59 Very Poor ~ 17P`O~] r ooo ! ! Cal! 026! ~ | Total 1 388.3691 0.131 0.601 0.121 0.031 0.07 Rehabilitated 0.00 0.00 0.01 0.05 0.00 0.00 0.74 Table 20. Average LOS ratings by year. ~.. | 11 Mean LOS I Year || Rating . 11 1 1991 11 70 . . 995 Il -90 106 0.53 0.39 l 0.00 .51

OCR for page 89
1.o l QGj It Q7 ~ QB Z 05 o Q. - ~ 0.3 C] no Q1 ~0 0.1 QO . ~ Q7 8 os rams ~ - - 1 Mean = 7.59 e,Q. S~Dev=~.16 `~ Q3 Jilt,,,/ 02 :"// . ;~ Very Good l 1991-92, QA = 70 c . Mean = 8.00 , Std Dev= 0 1 1 1995~96~ ~ = . Mean = &52 Std Den= Q84 / ' / ~ ; Far ; Poor ; Very Poor , Rehabilitated New Category Figure 14. Probability plot for pavements win a previous condition of Very Good. , ; 1991-92, QA = 70 . - 7~ Mean = 7.47 \j _r/ Std Dev= 0~74 , IT -a m ~ 0 ~' Fair , Poor ' Very Poor Rehabilitated New Category Figure 15. Probability plot for pavements win a previous condition of Good. 107

OCR for page 89
1.0, Q. 0.8 0.7 QB 0 05 g _~ 0.~ ~2 `~, Q3 C] Q2 Q1 1.0 Qua Q8 0~7 ~- O ~ I: 3 Q4 ~ Q3 C] 02 . Q1 . An . . ~ ~1 1 # 1 ~ ~1 1 1 ~1 # 8 8 1 ,' ~199~96, QA = 90 ; ' I / Mean = 660 ~ I / Std D3v= Q38 ', / `' tat 1 1991-92, QA- 70 ~; ~Mean = &63 ~j~/ Std Dev= 0.72 t\ ~I t\ 1 \\ ~1 ~' r _ Ve y Good I Good I Fair I Poor , Very Poor New Category Figure 16. Probabilibr plot for pavements w~ a previous condition of Fair. VeJy Good ' Good I Fair I New Category ,,~,. , Rehabilitated 199~;96, QA- 90 Mean = ~12 ~/ Std Dev= 0.66 . \ 1991-92, QA = 70 ~ Mean = ~50 / Std D~= 1.35 I ' ~ ' Poor ' V~v Poor , Rehabditated Figure 17. Probability plot for pavements wi~ a previous condition of Poor. 108

OCR for page 89
1.0, 0.~ oft ~ ~7 EN o.e Cal o 05 I: o 3 0.4 O 3 - 02 Q1 TO ' Very Good; Good 199~961 QA = 90 Mean = 0.94 Std Dev= 1.61 ; 1 L'.~' 111~ Poor Vey Poor New Categoly Figure IS. Probability plot for pavements win a previous condition of Very Poor. . . \ 199i-92, QA= 70 ~ Mean = 3.20 / Std Dev= 1.37 Rehabilitated Figure 14 (previous condition Very Good)- This figure shows Cat Me 1995-96 pavements are actually increasing in condition, whereas the 1991-92 pavements are remaining We same. Although logic and experience tell us Cat pavements do not Increase In condition win dine, proactive maintenance practices can have a positive effect on certain pavement management condition indicators In the higher pavement condition categories. For example, a sealed crack is often not rated as severe as an unsealed crack, or a surface treahnent may hide or mask over small distresses. No comparison of Me foDow~ng year's pavement distribution can be made for this category. The 1991-92 category contains only one pavement section of 3.! mi (5.0 km). Therefore, Me distribution for the following year exactly mirrors Cat section's performance. More pavements in this category are required to support Me results. Figure 15 (previous condition Good) This figure shows Cat Me 1995-96 pavements have a higher probability of remaining in the Good category, and Me 1991-92 pavements have a higher probability of moving to a lower category. This is further supported by Me distribution analysis. The distribution of Me 1995-96 pavements shows a higher mean condition rating value and a larger area between the two curves in the higher pavement category, although Me difference in the area between Me distribution curves i09

OCR for page 89
is rawer small. A more complete comparison could be made if Me actual conditions of the rehabilitated pavements were known prior to We application of the rehabilitation activity. Even Cough the rehabilitated pavements represent only a small percentage (7 percent) in this category, Hey may still affect He results of He analysis. Figure 16 (previous condition Fair) The results of this figure are inconclusive. Over half (53 percent) of He 1995-96 pavements were rehabilitatecl. When such a significant percentage of an individual category Is rehabilitated In He same year, it is Impossible to determine which condition category He majority of He pavements would progress to if He rehabilitation was not appliecl. The only way to make a complete comparison would be to record the actual condition of He rehabilitated pavements prior to the rehabilitation activity. Figure 17 (previous condition Poor) Even Cough a substantial percentage (39 percent) of He 1995-96 pavements were rehabilitated, He pavements win He higher LOS rating have a higher probability of remaining In He same category and not deteriorating to a lower category, even if ah of He rehabilitated pavements fell to He Very Poor category. However, just like He Fair category, the only way to make a complete and accurate comparison would be to know He actual condition of the rehabilitated pavements prior to He application of He rehabilitation activity. Figure 18 (previous condition Very Poor - This figure seems to show He greatest difference of all of the figures. Ahnost one-third (29 percent) pavements win the higher LOS raking moved from He Very Poor category to He Poor category, compared to 2 percent of the pavements win the lower LOS. This shows that a higher LOS can keen pavements in He backlog at a ~ . ~ higher condition until Hey can be rehabilitated. This figure also shows He effects of reactive maintenance activities performed on the pavement sections in this example. Discussion of Resulls Upon reviewing the results of He Florida example, it was difficult to determine if He difference In He 1991 and 1995 LOS ratings substantially affected He rate at which He pavements in He analysis group progressed toward a lower pavement condition and ultimately became part of He backlog. A number of factors confounded He analysis, He most obvious and important of which are described below. Analvsis croup size The analysis croup used In He example represented only a ~~~~~~~ o--~r ~ a--- Cal - r ~ ~ - -a -- .. ~ . ~ . ,^ ~^ r-,r, ~ ~ ~ ~~~ ~ ^^r~ ~ r,m, ~ ~ small sunset or pavements t~ m1 LocU Km~ In Hi and J~o nu Lozo KmJ In 1995~. When a small analysis group is used, the amount of rehabilitation and He 110

OCR for page 89
total amount of pavement in each individual category can affect Me analysis results. In the example, the amount of rehabilitation greatly affected the analysis. A substantial percentage of me pavements in certain pavement categories were rehabilitated in 1995. When a substantial percentage of pavements in a category are rehabilitated, it becomes virtually impossible to analyze the pavement distribution of that category for the next year, unless the conditions of these pavements are known immediately before rehabilitation. Likewise, when Me analysis group is small, Me amount of pavements In an individual category can be very small, or a given category may not even contain pavements. In this example, certain categories contained a very small amount of pavement miles. In these cases, unusual performance of one pavement can skew Me results of Me entire category analysis. Level of maintenance during the years between 1991 and 1995-The LOS ratings used n this example were for 1991 and 1995. These years were chosen because Hey exhibited a large difference in Me LOS ratings. However, Me pavement condition and He backlog accrual rate is not only affected by He current year's maintenance, but the maintenance during the past years as well. By selecting non-consecutive years, no consideration was given to He level of maintenance during He years between 1991 and 1995. To asses He Hue accuracy of He backlog analysis, a year-to-year review of the LOS ratings should be performed and incorporated into He results. Missing or inaccurate pavement condition data The analysis group for this example was not only chosen because of He large difference In LOS, but also because almost ad highway segments in He EMS database contained pavement condition data. However, some sections did not have pavement condition data for 1991 or 1995, or He pavement condition ratings were not updated to reflect recent rehabilitation projects. In an attempt to mistune He effects of these factors, sections having these characteristics were eliminated from the analysis. This resulted in a decrease in the size of the analysis group and, consequently, magnified He problems Cat result from using a small analysis group. Reactive or proactive maintenance approach The maintenance approach selected by an agency will greatly affect the categories in which an increase in pavement performance characteristics may occur. If an agency performs proactive maintenance and is performing He maintenance activities before substantial visual distress occurs, the results may be seen in aD condition categories, but He most noticeable affects win be in He higher pavement ~. - ~ AL condition categories. lt an agency performs reactive maintenance, the results win not be seen unto He middle to lower pavement condition categories, where visual distress is evident.

OCR for page 89
Change in Condition Indicator vs. Time The following section describes the data collection and analysis work conducted to evaluate We proficiency and usefulness of this method as a way of tracking Me relationship between QA program outputs and long-term performance of pavement facilities. The formal methodology for the Change in Condition Indicator vs. Time method is presented in appendix C of Me QA program Implementation Manual. Data Collection and Analysis To evaluate Me Change in Condition Indicator vs. Time method, comprehensive pavement condition data and maintenance quality retina data were obtained from the Iowa DOT. The pavement condition data were downloaded from Me Iowa EMS database and included annual PCI ratings and surface roughness measurements for all interstate and primary route pavement sections for Me years 1988 through 1993. The LOS data were obtained from Me 1993 maintenance quality evaluation report (Iowa DOT, 1993) and consisted of annual pavement surface element LOS ratings for all interstate and primary highways in Iowa for fiscal years 1983 through 1993. The Change in Condition Indicator vs. Time method was tested using the data collected from Iowa and the seven procedural steps described in the Implementation Manual. The application of each step and its results are given below. Step 1. Determine the Analysis Group. The analysis group for this example was limited to all primary roads and nterstates, of all pavement types, In Me State of Iowa. Step 2. Compile Table of Yearly Condition Indicator Data Representing the Chosen Analysis Group. Yearly PCI data (1988 Trough 1993) were obtained from Me State's PMS database. The data were matched (year-to-year) based on knowing each section's pavement type, cons~uchon year, and section length. Step 3. Delennine the Yearly Deduct Values. Yearly deduct values were obtained for Me analysis group using Me PMlool computer software. PMTool is a proprietary software Cat first calculates yearly deduct values for individual sections based on consecutive years of data, Men determines the overall deduct value (representing Me data) as an average of Me individual section deduct values. ~2

OCR for page 89
Both the two-point and three-point analysis approaches (described in appendix C of QA program Implementation Manual) were used to calculate yearly deduct values. The results of both approaches are su~runar~zed in table 21. Step 4. Plot of Condition Indicator Data Versus Age (optional). A plot of condition indicator vs. age was not constructed for this example because Me analysis group contained many different pavement types. Step 5. Summarize LOS Rating Data Representing the Chosen Analysis Group. The annual roadway element LOS rating data for all of We highway sections included in Me analysis group are summarized In table 22. Step 6. Plots of Yearly Deduct Curves for Different Average LOS Rahugs (optional). The step 6 option was not exercised in this example. Step 7. Plots of LOS Rating and Yearly Deduct Values Versus Time. Figure 19 contains the resulting plots of LOS ratings and yearly deduct values vs. time for Me Iowa highway sections. As can be seen, Me trend of LOS over time generally mimics the trend of PCI deduct values over time. Beginning in 1989 and ending in 1992, maintenance ratings dropped from 82.2 to 69.8, which corresponded with steady increases in the PCI deduct values (i.e., higher negative values). Then, a slight increase in maintenance quality between 1992 and 1993 was accompanied by a much sharper decrease In Me deduct values (i.e., lower negative values). Table 21. Yearly deduct values calculated using the two- and three-point analyses. Analysis Type Tw+Point Analysis 1989 1990 Three-Point Analysis 1992 1993 1990 1991 1992 1993 Years Included in Calculated Deduct ~Villa 1991 1989-1990 1990-1991 -1.51 ~991-1992 1992-1993 1988-1990 1989-1991 1990-1992 1991-1993

OCR for page 89
Table 22. Yearly maintenance LOS ratings for Iowa pavements. 1989 1990 90.0 80.0 70.0 m.0 - 50.0- cn o 40.0 30~0 20.0 10.0 1989 1990 1991 Year Year ! Roadway LOS Rating 100.0 1 ' . ~... . t ~ i ~ t s ; -___ . ~ ~ . . . i j ~ . , ~ . . . .. . . . _-,. . ~ i a t . ., , . . : r ., . . , . . . . . . ~ . . . .................. - ~_ _ ~_ ~. ~. ~............. I I ~ Tw>Point Analysis | - ~ Three Point Analysis . ~ -* - LOB Ratings _. - i 1992 1993 6.00 . 400 In 0.00 ~ en ~5 - 400 - ~.00 Figure 19. Plot of LOS rating and PCI deduct values vs. time for Iowa Interstate and primary roads, ad pavement herpes. ~4

OCR for page 89
Discussion of Resulls Although the maintenance quality and PCI deduct value trends are fairly similar, Me effects of over factors, such as highway funding allocations, traffic, pavement design, and materials, can be sensed in this evaluation. For instance, in 1990, maintenance quality increased slightly, whereas the PCI deduct increased considerably. Also, in 1993, a small increase in maintenance quality was accompanied by a large decrease in the PCI deduct. Although engineers largely involved win Me performance of pavements in Iowa during this period will have a much better understanding for the noted occurrences, one possible scenario is that a large number of pavements reached Me steepest part of Me pavement deterioration curve from 1989 to 1992, and Mat several pavements were rehabilitated in 1992 as a result of increased funding or reallocation of funds. Maintenance may have only been able to keep quality up until 1990, at which time We same level of maintenance funding, or even slightly increased funding, was insufficient to keep conditions up. Maintenance quality may have then rebounded In 1993 due to Me infusion of "new" pavements into Me network. Interpretation, Appraisal, and Applications Presented in this chapter were sac possible me~ods for tracking the relationship between maintenance LOS and the long-term performance of highway facilities. Two of these six me~ods, Me Backlog Analysis memos and the Change in Condition Indicator vs. Time method, were selected for detailed evaluation to determine Weir overall effectiveness and practicality as tracking melons. The evaluations were performed in accordance with the procedures set forth for each memos, using actual maintenance LOS and pavement condition data collected from two SHAs. The evaluation results were, at best, fairly supportive of Me effectiveness of each methodology. In bow cases, it was quite clear Mat not being able to eliminate . ~. - confounding factors Is a serious drawback. Significant changes or differences in ~ v funding, design, materials, tragic, climate, and rehabilitation criteria can cloud the relationship between maintenance quality and long-term performance. Perhaps more importantly, these factors can offset one another, leading one to wrongly interpret the maintenance quality-performance relationship. From a practicality standpoint, bow methods were relatively simple and straightforward. It is believed that infrastructure management personnel at both State and local highway agencies would be very capable of applying Me Backlog Analysis method to the condition data Hey have collected and stored. Likewise, it Is believed Cat infrastructure management personnel at more advanced State and local agencies would have very few problems following He Change in Condition Indicator vs. Time methodology. Many pavement management groups at He State level have developed and routinely update pavement performance models, and are therefore very familiar win He rates of pavement deterioration unique to Heir network. ~5 1

OCR for page 89
In light of Me above appraisals, it is recomanended that bow We Backlog Analysis and Change in Condition Indicator vs. Time me~ods be viewed as precursory evaluation me~ods. Each is capable of providing an Axial indication of the maintenance quality-performance relationship, but each should be supported by a more reliable methodology (one similar to the Regression Mode! approach discussed earlier in this chapters that accurately accounts for Me effects of funding allocation, design variables, traffic, climate, materials, and over factors. Interested agencies must be advised to recogruze me differences between Me purposes for and Me inferences to be derived from information collected for PMS and LOS rating systems. These differences are as follows: PMS data generally tends to quantify the remaining life of specific pavement sections or projects. The impacts of age, design parameters, materials, traffic, weaver, geography, construction and maintenance techniques each contribute to Me total life expectancy of pavements. Although Me purpose of most maintenance treatments is to prolong Me life or overwise slow deterioration, it cannot cost-effectively hak the eventual need for rehabilitation. LOS rating clata generally provide an indication of how well the maintenance operation within an agency is performing those activities under its span of control. Although it Is reasonable to assume a good maintenance program can indeed slow the deterioration of a properly designed and constructed pavement, factors beyond Me control of most maintenance orgaruzations Innit the ultimate time before a pavement reaches Me point of requiring rehabilitation. ~ closing, comparisons of PMS and LOS data were conducted in this study with Me idea of identifying possible linkages between highway maintenance quality programs and over management information systems. Each agency, depending on its circumstances, should carefully consider Me value to be received when attempting to use over management information system data in lieu of a program specifically designed to identify Me LOS being provided by its maintenance operation. T%% ran ~T ~ ~ ~, 116