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.
APPENDIX C FIELD EVALUATION OF CANDIDATE MEASURES Methodological Approach The field study conducted during Task 5 of the NCHRP Project 20-34 efforts evaluated candidate M.O.E. sensitivity to actual truck weight enforcement operations. This study has defined weight enforcement M.O.E.s as.. "..determinable quantities of what is achieved as the result of truck weight enforcement activity. Their application also quantifies the contribution that activity makes toward achievement of one or more of the enforcement goals. " The following M.O.E.s (See Table I) were based on their suitability to demonstrate truck weight enforcement effects. These measures addressed legal load-lim~t compliance objectives of truck weight enforcement procedures as well as Me potential for overweight trucks to produce pavement wear and tear. Having developed a set of proposed truck weight enforcement M.O.E.s to be evaluated, specific methodological considerations were Implemented web regard to the Task-S evaluation study plan. In order to assess the M.O.E.s, it is essential to include Free fundamental measures-evaluation related concepts: reliability, validity, and sensitivity. An urlderstarlding and application ofthese concepts are necessary for an M.O.E. assessment. Reliability This concept addresses measurement repeatability, i.e., confidence that repli- cated applications wall yield consistent results. In order that a measurement technique be 1 Appendix C
Table I. Designated Measures of Effectiveness (M.O.E.s) and their Definitions CO-ED :: i| ~ ~ ~ ii= ~ ~ ^; ~ ~; ;= At f:f f; my. its i:~: :~ The fraction (or percentage) of the total ob Gross Weight Violation, Proportion served truck sample which exceeds the le gal gross weight limit. The extent to which average measured Gross Weight Violation, Severity gross weights for the observed sub-sample | of gross weight violators exceeds the legal gross weight limit. The fraction (or percentage) of the total ob Single-axle Weight Violation, Proportion served truck sample with one or more axles which exceeds the legal single-axle weight limit. | The extent to which average measured sin Single-axleWeightViolation, Severity ale-axle weights for Me observed sub sample of single-axle weight violators ex ceeds the applicable legal limit. The fraction (or percentage) of the total ob Tandem-axIe Weight Violation, Proportion served truck sample with one or more tarl dems which exceeds the legal tandem-axIe weight limit. | The extent to which average measured tan Tar~dem-axIe Weight Violation, Severity dem-axIe weights for the observed sub sample of tandem-axle weight violators exceeds the applicable legal limit The fraction (or percentage) of the total ob Bridge Formula Violation, Proportion served truck sample which exceeds the le gal Bridge Formula weight. The extent to which average measured Bridge Formula Violation, Severity Bridge Formula weights for the observed sub-sample of Bridge Fonnula violators exceeds the legal weight. The fraction (or percentage) of the total ob Excess ESALs, Proportion served truck sample exhibiting Excess ESALs; i.e., ESALs attributable to Me ille gal portion the individual single or tandem axle group. _ . . The average value of Excess ESALs ob Excess ESALs, Severity served for the truck sub-sample exhibiting Excess ESALs. Appendix C 2
dependable, it must be reliable. Reliability refers to the degree of stability exhibited when a measurement is repeated under identical conditions. Concern for reliability comes from the necessity for dependability in measurement. Synonyms for reliability are: dependability, stability, consistency, predictability, accuracy. With regard to Suck weight enforcement M.O.E.s, reliability is necessary for the uniform application of enforcement procedures across regions of the country or within a state. In order to ensure the reliability of recommended M.O.E.s, the Task 5 evaluation uniformly applied the M.O.E. sensitivity analysis to WIM truck weight data collected in four states representing norm, south, east, and western regions of the United States. Validity The validity of a measure refers to the degree to which it actually measures what it is designed to measure. Validity is a complex subject, but it is particularly important in behavioral research. It is possible to study reliability without inqu~nng into the meaning of variables. It is not possible to study validity, however, without inquiring into the nature and meaning of one's variables. The validity of the tested measures in this study was established based on weir relevance to truck weight enforcement objectives, i.e., examine compliance with legal weight limits (e.g., axle, axle-grouping, and gross weights) and infrastructure considerations. Therefore, no further consideration was necessary to ensure validity in We field study. Sensitivity A key element of the Task 5 Field Studies was to assess the sensitivity of candidate M.O.E.s to actual truck weight enforcement operations. In behavioral studies, the concept of sensitivity is as follows. When an instrument, e.g., measure, is used to classify individuals as having or not having a specific attribute, the sensitivity of the measure is the proportion of correct results among people who actually 3 Appendix C
have the attribute. That is, sensitivity is an indication that the applied measure produces a true Indication of the sought attribute or condition. With regard to truck weight M.O.E.s, it was necessary to seek assurance that application of the M.O.E. provided a true indication of truck weight enforcement effects. - The sensitivity of the candidate M.O.E.s to actual truck weight enforcement activities was expenmentally determined in this field study through the controlled (matched day-of-week, tune-of-day, and seasonal) comparison of measures between "baseline", i.e., no enforcement activity, and "enforcement" conditions, i.e., on-going truck-weighing operations. Both permanent weigh scale operations and portable roadside truck weighing procedures were observed as enforcement conditions. Results Candidate M.O.E.s were evaluated In this field study on the basis of matched WIM data sets representing controlled enforcement and nor-enforcement the periods. Data collection periods were controlled so as to avoid time-of-day, day-of-week, and seasonal confounding effects. Applied WIM data were gathered in California, Georgia, Idaho, and Minnesota. Enforcement procedures and results are discussed as follows for each of the four study states. California M.O.E. validation results, based on field observations in California, are dis- cussed in this section. The analyses compare WIM data collected dunng baseline (non- enforcement) and enforcement conditions. The Califorrna Depardnent of Transportation provided output from a WIM scale located approximately Free miles norm of He Santa NelIa weigh scale on I-5. The data sample consisted of a 24-hour observation period containing 3,678 semi- and fi~-trailer truck combinations. The permanent buck-weight enforcement scale was open during the observation penod for seven consecutive hours (4 a.m. to ~ ~ a.m.~. This data set afforded Appendix C
an adequate sample size of uniform truck weights (e.g., no day-of-week or seasonal effect contamination) to support a one-shot determination of enforcement ejects. The analysis depicted in Table 2 below is based on sample of 2,370 Type 9 (Tractor with semi-trailer) trucks. Given observed samples of 416 and 1,954 trucks, re spectively, during the enforcement and baseline conditions, lower gross weights (i.e., 55,948 versus 59,547 pounds) were observed during times when the weigh station was open. A furler examination of axIe-specific weight differences between conditions re vealed that rear-tandem weights were lighter during enforcement conditions. While lower average ESALs were observed during the enforcement period, the difference was not statistically significant. However, it is worth noting that while no trucks exhibiting Excess ESALs were observed during the period when the scale was open, a small (.36) Excess ESAL average was observed when the scale was closed. Table 2. California Measures Sensitivity Experiment Gross Weight Violation, Proportion Gross Weight Violation, Severity Single-axI~n Lion Single-axle Weight Violation, Severity Tandem-axle Weight Violation, Proportion Tandem-axle Weight Violation, Seventy Bndge Formula Violation, Proportion Bridge Formula Violation, Seventy Excess ESALs, Proportion Excess ESALs, Severity Note: Weight units are pounds 5.9% 2,567 2.9% 438 6.8% 2,016 44.3% 7,400 . 1 % - .36 10.1% 2,266 31 % 879 6.5% 1,607 40.2% 9,780 o o No No No No No* Yes Yes No No* Yes * = Non-signiJ cant tendengy An examination of Type 9 truck weights exceeding 80,000 pounds revealed only slightly lower average gross weights (i.e., 82,266 versus 82,567 pounds) during periods ~ The designation, "Non-significant tendency", indicates a numerical difference which suggests a possible observed enforcement effect; however, the difference is not suffi- ciently strong to be statistically significant. 5 Appendix C
when the scale was open. The proportion of trucks exceeding 80,000 pounds was higher (i.e., 1 0.1 percent versus 5.9 percent) when the scale was open. The presence of Bridge Formula violations was examined for both the baseline and enforcement conditions. A smaller proportion of the truck sample (i.e., 40.2 versus 44.3 percent) was seen to exhibit Bridge Formula violations during periods when the weigh scale was open implying a favorable enforcement effect. Nevertheless, the degree of violation was more severe (i.e., 9,780 versus 7,400 pounds) during this enforcement period. An examination of single-axle violations first examined steering axle weights and compared these to the California legal limit of 12,500 pounds. During the baseline con- dition, 2.9 percent of the weighed steering axles exceeded the legal limit by an average of 438 pounds. During the enforcement period, 3.1 percent exceeded the limit by an in- creased average weight of 879 pounds. Second, an examination was conducted for indi- vidual axles throughout the tandems. During the baseline period, 1.2 percent exceeded the California legal limit by art average of 1, 279 pounds, and during the enforcement pe- riod, 4.8 percent exceeded the limit by an average of 1,664 pounds. Thus, no enforce- ment was observed with respect to single axle weights. Tandem weight distributions were compared between baseline and enforcement conditions. A favorable enforcement effect was noted with regard to rear tandem viola- tions. The severity of violation was significantly decreased (i.e., from 2,016 pounds to 1,607 pounds, and the proportion of violations fell slight from 6.8 percent to 6.5 percent during the enforcement period. In addition, we applied one candidate measure that had been considered as an M.O.E., i.e., the 95th-percentile gross weight, to the data set. A slightly lower 95th- percentile, i.e., 80.1 Lips versus 81.2 hips, was found for the "scale closed" Buck sample. This difference was not statistically significant. Appendix C 6
California M.O.E. Validation Summary The California Department of Transportation provided output from a WTM scale located on I-5. An analysis of 3,678 truck combinations exhibited lower gross weights with a smaller proportion of overweight axles dunog the time when the weigh station was open. Data on a sub-sample of 2,370 tractor-sem~-trailer combinations was further analyzed to examine M.O.E. sensitivity to the enforcement activity. The results confirmed the following M.O.E.s: Tandem-axIe Weight Violation Severity, Bridge Formula Violation Proportion, and Excess ESAL Severity. Georgia Mobile truck-weight enforcement operations, utilizing an obtrusive portable roadside weigh scale, were conducted at two locations In Georgia: a rural arterial, State Route 300 in Crisp County; and a rural interstate, Interstate 20 in Taliferio County. WIM equipment problems, i.e., failure to generate data for a representative truck sample, precluded use of data gathered at Me rural arterial location. Data gathered at the interstate location did produce a suitable vehicle sample; however, the WIM equipment proved to be "over-calibrated", i.e., generating higher than expected truck weights, thus requiring a "Qu~ity-Control" analytic procedure. A brief explanation of the applied Quality Control procedure follows. Chaparral Systems of Santa Fe, New Mexico has developed a software package to analyze 'raw WIM data and apply a series of quality control corrections, e.g., factors to compensate for WIM-calibration error. The software examines truck weight distributions and notes distributional 'peaks' due to Me presence of emptr and loaded trucks in the traffic stream. The QC software then applies correction factors based on expected peaks for loaded and empty trucks. This public-doma~n software is ready available2, and can be operated using Windows and SAS software. Truck weight data collected on I-20 in Talifero County demonstrated problems due to apparent WIM equipment over-calibration. Therefore, two follow-on steps were taken 2 Interested users should contact Statistician, Cindy Cornell, at Chaparral Systems, 649 Harkle Road, Santa Fe, New Mexico 87501 7 Appendix C
with regard to the Georgia data. First, data files were supplied to Chaparral Systems so they could conduct a Quality Check analysis. Second, the M.O.E. sensitivity analysis was conducted with regard to the existing data to examine for enforcement effects. This analysis tested the sensitivity of 'uncorrected' data to enforcement effects. This step is Important in the M.O.E. assessment process in order to address the requirement to prelimi- nanly conduct quality control steps prior to applying any M.O.E. evaluation procedure. The applied Quality Check (QC) analysis generated the plot, Figure Ion the next page, comparing baseline (06JUN95, broken line) with enforcement (13JUN95, solid line) conditions. It is important to note that plotted data have been corrected for the calibration error through the application of correction factors. Results follow for M.O.E. computations based on data generated directly from the SHRP WIM equipment and not subjected to the QC analysis. We fee! that it is important to exarn~ne calculations based on these data, as these are the form of data Initially generated as the result of WIM data-collection procedures. Any M.O.E. analysis too! that can be validated with "raw" data will be more easily applied by states than if a QC analysis is necessary. Analysis of uncorrected WIN data An analysis of "uncorrected" Georgia data, i.e., Quality Control correction factors were not applied, yielded a number of results which supported M.O.E. development. Although no promising difference was observed for average gross weight difference (e.g., gross weights exhibited a larger variance In the enforcement condition), lower rear tandem weights were evident during the enforcement period. The most significant M.O.E.-developmental effects were noted for the proportion of overweight trucks and associated axles. The proportion of overweight Ducks was significantly lower (at the .05 significance level) during the enforcement period. This finding also held for Me examination of compliance proportion associated with each axle companson. Appendix ~ 8
- - to a' ~ - a) Q IMP ~ ~1m Hi , ~ ~ UJ a 39 o in Cal o _ a) Q X 8 to i` in a, 1 A Cal ~ . U) ~ ~ .! ,, ~ o , CO to - Figure 1. Plot of QC-corrected data versus uncorrected data. ·9 u, a) z lo 1 a, z _ ~ Appendix C
Two critical M.O.E.s, the proportion of Single Trailer trucks exhibiting Excess ESALs (i.e., .72 and .67, respectfully, for the baseline arid enforcement conditions) and the proportion of trucks exhibiting Bridge Formula violations (i.e., .69 arid .63) did not differ between baseline arid enforcement conditions. Analysis of QC<orrected data Results of We M.O.E. Sensitivity Expernment data analysis based on data to which the Quality Conho} factor was applied are summarized In Table 3 below. Table 3. Georgia Measures Sensitivity Experiment Gross Weight Violation Proportion t _ Gross Weight Violation, Severity Steering Axle Weight Violation, Proportion Steering Axle Weight Violation, Severity Tandem Weight Violation, Proportion Tandem Weight Violation, Severity Bridge Formula Violation, Proportion Bridge Formula Violation, Severity Excess ESALs, Proportion . Excess ESALs, Severity Note: Weight units are pounds B-:- ease ins :: ..... ~ . ... ....... . ~.... -I ~ . ~ _ . ... .. I. . ~ ~ _ =,,,n..o~do..n....,:.-:..:~ 61 onto 2,370 9.4 onto 695 54 onto 1,675 1.0 % 390 9.4 onto . 1.0 6 7 onto 4,709 7.5 onto 545 3.7 onto 1,545 3.8 % 223 8.8 % .65 No . No Yes No* Yes No* No Yes No* Yes * = Non-signif cant tendency Observed gross weight differences failed to provide M.O.E. field validation. More severe average gross weight violations (4,709 versus 2,370 pounds) were noted dunng the enforcement penod. However, it is also noteworthy that proportions of trucks exhibiting gross weight violations did not significantly differ between enforcement arid baseline conditions. The analysis of specific axle and tandem weights revealed enforcement-effect differences as follows. An analysis of single axle weights indicated a reduction in steering and rear- axle violations. For example, lower overall steering-axle weights were evident via a smaller proportion (7.5 versus 9.4 percent) of violations dunng the enforcement condition. Also, given the sample of trucks exhibiting steering axle violations, there was a statistically non-sign~ficant tendency toward less severe violations. While there was an observed Appendix C IQ
tendency for lower tandem-axIe violations during the enforcement period, the observed difference was not statistically significant. Two additional M.O.E.s were validated on the basis ofthe field observations. First, less severe Bndge Formula violations were observed during the enforcement penod. Second, while a non-sign~ficantly smaller percentage (~.8 versus 9.4) of Excess ESAL violations was observed during the enforcement period, the level of severity was reduced by a small (.65 versus I.0 ESALs) but statistically significant level. M.O.E. differences between QC-corrected and non-corrected data sets The comparison between Georgia enforcement and baseline conditions involved two examining data sets, one on which WIM-calibration corrections (i.e., Quality Control or QC analyses) had been applied and another on which corrections had not been applied. This comparison was conducted to determine the suitability of non-corrected data sets for M.O.E. evaluation. The QC-corrected set indicated significantly lower steenng-axIe weights during the enforcement condition, i.e., indicating a valid enforcement effect. This effect was not evident In the non-corrected data set. Over less consequential difference were as follows: the non-corrected data set indicated greater variability on steering-axIe ESALs during the baseline condition, less vanability on second-axIe ESALs during Me baseline condition, and greater variability in third tandem weights during Me enforcement condition. In general, the QC-corrected data set was more sensitive to ESAL variability differences between the baseline and enforcement conditions, and it detected lower third tandem weights during the enforcement condition. Moreover, the QC-corrected data set discerned Excess ESAL differences between baseline and enforcement conditions. Agreement between the two data sets was noted wad regard to certain M.O.E. differences, e.g., lower rear-axIe weights dunng the enforcement condition. The most ~ .! Appendix C
serious divergence in detected differences between the data sets related to a critical measure, total weight violation severity. The non-corrected data set failed to statistically distinguish between baseline and enforcement samples on the basis of this measure. The QC-corrected data set did discriminate between conditions, unfortunately noting larger gross weight violations during the enforcement period. The foregoing analysis concludes that it is generally recommended to conduct a QC check of WIM databases prior to application of an M.O.E. evaluation. However, we must note that this finding is based on "poor case" scenario, i.e., the required correction factor was .5 due the over-calibration severity of the applicable WIM system. Obviously, data collected by a recently-calibrated WIM system does not require application of a QC check prior to an M.O.E. analysis. It is also noteworthy that during the surprise enforcement period, a number of overweight trucks were observed to either park alongside the Interstate or divert to alternate routes. Alert enforcement personnel were able to administer citations to a number of Divers exhibiting this behavior. Georgia M.O.E. VaIiciation Summary Mobile truck-weight enforcement operations, utilizing an obtrusive portable roadside weigh scale, were conducted at a rural interstate location. An analysis of 483 combination trucks revealed a number of M.O.E. validation effects associated with observed axle and tandem weights. Under conditions of visible (and unexpected) mobile enforcement operations, the observed truck sample exhibited lower steering-axle weights, Tower rear-axle weights, and lower rear-tandem weights. During the surprise enforcement operation, a number of overweight trucks were observed to either park alongside the roadway or divert to alternate routes. The results confirmed the following M.O.E.s: Single-axIe Weight Violation Proportion, Tandem-axle Weight Violation, and Excess ESAL Severity. Appendix C ~2
Idaho A large volume of WIM data was provided by the Idaho Transportation Department. Data were collected at a WIM site located two miles north of the Cotteral Port of Entry, near milepost 232 on ]:-84, approximately 40 miles north of Me Idaho-Utah border. WIM data represented traffic flow in both directions dunng 24-hour observation periods for approximately two weeks in March and April 1996. The available 29,000-truck database was divided into sub-samples in order to achieve necessary experimental control required to examine specific day-of-week effects and separate weekend from weekday violation trends. Table 4 illustrates the enforcement schedule, i.e., days/hours of operation of the Port of Entry dunug dates for which data were supplied. Data analysis periods were designated to permit observation of enforcement ejects while controlling for time-of-day and day-of-week. Seasonal variation was not a consideration as all data were collected within a six-week time frame. The large amount of data provided by the Idaho Transportation Department supported a determination of enforcement effects for various times during the week. This consideration is important due to enforcement agencies' manpower-scheduling requirements. The data analysis below contains separate enforcement analyses for enforcement activity conducted on a Sunday, Tuesday, and Thursday. Sunday enforcement effects A baseline condition, i.e., no enforcement, was established for data collected on Sunday, Apn! 7, 1996 in the northbound direction. Data were compared for an enforcement condition In effect on March 3. On that diate, Me Port of Entry was operated during both the Day and Swing shifts, thus allowing data comparison for Me corresponding period and traffic direction on April 7. The applied database for this enforcement-effects comparison consisted of 393 trucks for the baseline condition and 957 trucks for the enforcement condition, which were observed during equal time periods. Due to the fact that the Measures Sensitivity analysis required a large sample of equivalent truck-Ioading axIe-conf~guration characteristics, this ~3 Appendix C
o - ~ ~ o ~ o A ~ ~ ~ o ~ ~ ~ o o ~ ~7 ~ o ~ ~7 ~ - ~ - - ~ (V e_ Let ~ 0 ~ O _ ° ~ _ r ~ V _ C'7 o - ·_ 00 Appendix C I on Cat - s W- ~ ~ ~ ~ X `.bO ~ _ _ aid Cat ~ ~ ~1: 1 ~ .; ~C ~ .E ~ ~, - c~-." - ~ a~ ,,, ~_ _ _ ~8' ~ O · :__ ~ ~ 4= _ =.~=1=,~, =~. ~1 1 ~0 : ;^ Fa cn O O o o0 ~ e" ~: .; CD 00 .~ cn _ C" . c: ._ o0 ; - c: ~ ._ CD ';~ O tn O O ) (_) O O L) `_) O C" O'; .~ ~0 c: ._ ~ ~ a U) ; - C C\S .= ~0 ;^ C e`5 ·= U' 08 ~ a CD ~D ·a s: c ~ g: ~ ~ m c: ._ u' oo ·; - o o e ~ ~ Table 4. Idaho Port of En~y Operai~on and ~M Data Collection Schedule 14 ~ 0 0 ~ _ . . ~ 0 0 ~ _ ~ o 0 0 . . ~D 4; 5: ~ cn ~o ~ S ~ c', 1 ~ a cn
analysis was conducted for Type 9 trucks, i.e., five-axle semi-trailer combinations. The baseline and enforcement condition samples consisted of 348 and 846 Type 9 trucks, respectively. Table 5 summarizes results of the M.O.E. analysis for this experiment. A smaller percentage of the sample, 4.1 versus 1 1.2 percent, was observed to exceed the 80,000- pounds gross weight limit during the enforcement period. However, gross weight violation severity, average truck weights in excess of 80,000 pounds, was greater during the enforcement period. This effect may be due in part to the presence of permitted vehicles in excess of 80,000 pounds. Nevertheless, the enforcement sample demonstrated lower gross weights on average' i.e., 57~379 pounds versus 63,953 pounds, with the difference being distributed over all axles. Table 5. Idaho Measures Sensitivity Experiment (Sunday Enforcement Effects) .- - ~.(,~ -. ., Em.- . - - -- ~- I--- ~. .- .. Gross Weight Violation, Proportion Gross Weight Violation Severity S'ngle-axle Weight Violation, Proportion Single-axle Weight Violation, Severity Tandem-axle Weight Violation, Proportion Tandem-axle Weight Violation, Severity Bridge Formula Violation, Proportion Bridge Formula Violation, Severity Excess ESALs, Proportion Excess ESALs, Severity Note: Weight units are pounds 1 1 .2 °/0 ; 3,744 Ibs. 13.5 NO 1,019 1bs. 7.2 onto . 1,767 2.8 onto 3,512 Ibs. 14.9 NO ~- 1.18 4 1 onto 8,089 Ibs. 2.1 onto 2,306 Ibs. 3 8 onto 2,534 2.5 onto 3.435 Ibs. 5.0 O/o 0.61 l Yes No Yes No Ages No l No * l No Yes Yes , $ = Non-signif cant tendency An examination of axle-specific violations revealed slightly reduced violation proportions for all axles during the enforcement condition. The largest violation reduction, from 13.5 to 2.1 percent, was observed for rear axles. However, the average severity of rear-axle violations increased from 1,019 pounds to 2,306 pounds during the enforcement period. A number of tandem-weight differences were observed between the baseline and enforcement conditions. For example, lower average drive arid rear tandem weights were 15 Appendix C
observed during the enforcement condition. While average observed drive-tandem violations were greater during the enforcement period, this trend was not evident for rear tandems. However, significantly lower proportions of trucks with overweight tandems (drives and rear combined) were observed during Be enforcement condition. Thus, the observed proportion of trucks with tandem violations was determined to be an enforcement- related M.O.E. During the enforcement period, observed average ESALs per hock were significantly lower than those observed during the baseline condition. Also, lower average axIe-specific ESALs were observed for the second through fifth axles. A significantly smaller proportion of the Duck sample (5.0 versus 14.9 percent) exhibited Excess ESALs during the enforcement period. Moreover, this sub-sample exhibited lower average Excess ESALs than were observed for the baseline condition. Tuesday enforcement effect; This data comparison is based on WIM data collected in the southbound direction on consecutive Tuesdays, i.e., March 5 and March 12. A baseline condition, i.e., no enforcement, was established for data collected on Tuesday, March 12. Data were compared for an enforcement condition In effect on March 5. On that date, the Port of Entry was operated during the Day shift, thus allowing data comparison for the corresponding tune period one week later. The database using for this enforcement-e~ects comparison consisted of 539 trucks for the baseline condition and 546 trucks for the enforcement condition. The analyzed baseline and enforcement condition samples consisted of 492 and 500 Type 9 trucks, respectively. During the enforcement period, observed trucks did demonstrate significantly lower gross weights on average, i.e., 61,684 pounds versus 63,563 pounds. However, the sample Appendix C 16
proportion exceeding gross-weights of 80,000 remained essentially Me same as during the baseline condition. Table 6 summarizes results of the M.O.E. finding for this experiment. By comparison with the Sunday analysis, a lesser number of findings was available to support the development of buck weight enforcement M.O.E.s. as follows. Table 6. Idaho Measures Sensitivity Experiment (Tuesday Enforcement Effects) Gross Weigh! Violation, Proportion Gross Weigh! Violation Severity . . Single-axle Weight Violation, Proportion Single-axle Weigh! Violation, Severity Tandem-axle Weight Violation, Proportion . Tandem-axle Weight Violation, Severity Bridge Formula Violation, Proportion Bridge Formula Violation, Severity ~ ~ Excess ESALs Proportion . At. . Excess ESALs, Severity Note: Weight units are pounds 10.1 % 2,262 9.4o/o 9,324 9 2 o/o 2,622 4.1 onto 1 ,91 8 12.6 % 1 23 9.8 onto 2,240 4.0% . 7,896 6.8% 2,699 3.2 onto 2,538 12.2 % 1.31 No * No * Yes No Yes No* No* No No* No - * = Non-signif cant tendency An examination of axIe-specific violations revealed slightly reduced violation proportions for all axles during the enforcement condition. The largest violation reduc tion, from 9~4 to 4.0 percent, was observed for the forth axle (lead axle on rear tandems. However, Me severity of axle violations increased during Me enforcement period, aver aging 9,324 pounds per axle as opposed to 7,896 pounds per axle in the baseline condi tion. A similar analysis of tandem-axIe violations revealed reduced proportions of viola- tors in the enforcement period, characterized by a reduction of rear tandem from 9.2 per- cent to 6.8 percent, and slightly increased average tandem violation severity during the enforcement period. 17 Appendix C
No significant differences were observed with regard to Excess ESAL or Bridge Formula violations. Thursday enforcement effects This data comparison is based on WIM data collected in the southbound direction on consecutive Tuesdays, i.e., March 7 and March 14. A baseline condition, i.e., no enforcement, was established for data collected on Tuesday, March 14. Data were compared for an enforcement condition in effect on March 7. On that date, the Port of Entry was operated during the Day shift, thus allowing data comparison for the corresponding time period one week later. The applied database for this enforcement effects comparison consisted of 474 trucks for the baseline condition and 512 trucks for the enforcement condition. The analyzed baseline and enforcement condition samples consisted of 439 alla 473 Type 9 trucks, respectively. Table 7 summarizes results of the M.O.E. finding for this expenment. A large num her of the M.O.E.s were validated in this data set. During the enforcement penod, a smaller percentage of trucks violated the 80,000-pound gross weigh limit; and given Me sub-sample of violators, Me sevens of Me violations was decreased. Table 7. Iciaho Measures Sensitivity Experiment (Thursday Enforcement Effects) Gross Weight Violation, Proportion v _ A_ ~_< ~ _ ~ v_! 7~m Single-axle Weight Violation' Severity Tandem-axle Weight Violation, Proportion n - . ,. ~ I _ .. Bridge Formula Violation, Proportion Bridge Formula Violation, Severity Excess ESALs, Proportion E~E~ ~ _ Note: Weight units are pounds Appendix C 1 2.5 °/0 2,493 10.3 SO 6874 . , 14.8 °/0 1,621 3.4 onto 1 ,854 16.2 °/0 . .42 8 5 5 onto 1 ,765 5.o onto 5,188 8.5 onto 834 3 8 onto 3,361 8.2 onto . .33 Yes Yes Yes Yes Yes Yes . No No . Yes l No* * = Non-signif cant tendency
An examination of axle-specific violations revealed reduced violation proportions for all axles during the enforcement condition. The largest violation reduction, from ~ 0.3 to 5.0 percent, was observed for the forth axle (lead axle on rear tandem). Moreover, the average severity of axle violations decreased during the enforcement period, i.e., from 5,188 pounds per axle, as opposed to 6,874 pounds per axle in the baseline condition. This experiment produced M.O.E. validation with respect to the Excess ESAL measure. Lower average ESALs were observed for the enforcement truck sample and a smaller proportion of this sample exhibited Excess ESALs. IcIaho M.O.E. Validation Summary A large volume of WIM data, i.e., gathered on ap- proximately 29,000 commercial vehicles, was provided by the Idaho Transportation De- partment. A comparison of baseline versus enforcement conditions during three different weekdays produced a number of significant findings. While no day-of-week effects were readily evident to indicate on which days enforcement effort would more likely be effective, all of Me tested operational measures were shown to be sensitive to enforcement activity. M.O.E.s most consistently demonstrating sensitivity to enforcement activity were: (~) Gross Weight Violation Proportion, (2) Single-axle Weight Proportion, (3) Tandem-axte Weight. Proportion, and (4) Excess ESAL Proportion. While less frequently associated with en- forcement activity, the following measures were also validated in the Idaho data: (~) Gross Weight Violation Seventy, (2) Single-axle Weight Violation Severity, (3) Tandem-axle Weight Violation Seventy, and (4) Excess ESAL Seventy. Minnesota Data sets provided by the Minnesota Deparunent of Transportation were applied in this measures sensitivity experiment. Bending plate WIM data were gathered approximately five miles from a permanent truck-weight enforcement scale during times when the scale was open and closed. Data collection periods were designated to conform to the weigh station-operating schedule. The weigh station is routinely closed one weekend Appendix C
per month, thus providing opportunity to obtain data sets for open versus closed periods while controlling for day-of-week and season of year effects. During the research team's November 1995 meeting with Minnesota D.O.T. Officials, we requested data sets representing designated study days during November, December, and January in order to support the Measures Sensitivity experimental design. However, November data were not provided due to problems with the WIM scale. Data collected over twelve days were provided for designated days in December 1995 and January 1996. Due to truck shipping trends affected by the holidays, we were limited regarding the applicability of certain of the data sets. However, a number of non-confounded data comparisons were possible to support the Task 5 Measures Sensitivity experiments. Two comparisons are discussed herein which compare candidate M.O.E.s between periods of enforcement versus non-enforcement. The first comparison, carefully controlling day-of-week effects, is based on data sets collected on consecutive Tuesdays. In general, no enforcement eject was found wad regard to candidate M.O.E. differences. The second comparison based on consecutive business days of operation, contrasting on the last two days in 1995 wad the first business day in 1996, thereby emanating New Year's Day traffic, did reveal some differences. Each of these two comparisons is now separately discussed. Consecutive-Tuesdays Comparison The first comparison was based on two samples of Type 9 (Five-axIe, Semi-tra~ler combination) trucks. Data were controlled for bow seasonal arid day-of-week ejects. Each data set was collected on a Tuesday during late December 1995 and early January 1996. The "Enforcement" condition, i.e., scale-open, sample contained 1,915 trucks, arid the "Non-enforcement" condition, i.e. scale-closed, sample contained 1,357 trucks. Observed differences between the data sets did not reveal Appendix C 20
differences that support the operational validation of truck weight enforcement M.O.E.s. Table ~ summarizes the observed M.O.E. comparison between conditions. Table B. Minnesota Measures Sensitivity Experiment One -if ^ --DIM. BE.--- ~-~ - -- ~- Gross Weight Violation, Proportion Gross Weight Violation, Severity Single-axle Weight Violation, Proportion Single-axle Weight Violation, Severity Tandem-axle Weight Violation, Proportion Tandem-axie Weight Violation, Severity Bridge Formula Violation, Proportion Bridge Formula Violation, Severity Excess ESALs, Proportion Excess ESALs, Severity 1.55 °/0 2,O43 6.0 onto 1 ,338 7 2 onto 3,566 1 5o/o 2,200 9.1 3% .52 1.90 °/0 2,000 4.4 onto 1 ,231 8.3 onto 5,900 36 onto 1 ,700 10.70 °/0 .55 No No* No No* No No No No* No No Note: Weight units are pounds * = Non-signif cant tendency Truck weights were heavier on average, 48,228 pounds versus 46,166 pounds, with an insignificant increase, 1.90 versus 1.55 percent, in gross-weight overload violations during the enforcement period. The distribution of axle-specific overload violations was consistent with the noted gross-weight violation rate. A slight increase in average ESALs per truck, .99 versus .85, was noted during the enforcement period. The only observed difference, supportive of M.O.E. development was that larger Bridge Formula violations did occur during the nonenforcement period; however the sample sizes were too small to be statistically significant. Of the 1,915 Bucks observed during the enforcement period, a single Bridge Formula violation, i.e., 1,700 pounds, was detected. Of the 1,357 trucks observed during the non-enforcement period, only two Bridge Formula violations' i.e., averaging 2~200 pounds, were detected. An examination of axle-specific violations revealed non-significant differences with the exception of an increased proportion of Axle 3 (rear axle in drive tandem) of 6.0 versus 4.4 percent in the enforcement condition. The average severity of axle violations was slightly reduced' i.e.' 12~306 versus 13~379 pounds during the enforcement period; however' this reduction was not statistically significant. 21; Appendix C
Slight but statistically non-significant increases in the proportion of overweight tandems were associated with enforcement activity. The most pronounced difference was seen for driver tandems, whereby the proportion of violators increased from 7.2 to 8.3 percent. The severity of the associated tandem violations increased from an average of 3,566 pounds during the baseline condition to 5,900 pounds during the enforcement period. In addition to the 3,272 Type 9 (five-axle semi-trailer) trucks noted above, a similar analysis for 260 Type 10, 11, and 12 (multi-trailer) trucks revealed similar results. No differences were observed to support the development of M.O.E.s. Specifically' Bridge Formula violations noted for the Type 9's were not replicated. The likely explanation is that there were significantly fewer of the latter truck types observed. In summary, this data set revealed no statistically significant truck weight effects to support M.O.E. validation. Consecutive Business~ays Comparison The second comparison revealed slightly more promising results in terms of establishing the applicability of candidate truck weight enforcement M.O.E.s. Based on two samples of Type 9 (Five-axIe, Semitrailer combination) trucks, data were controlled for seasonal effects due to the close time proximity between enforcement and non-enforcement conditions, i.e., this sample pair contrasted the last two days in ~ 995 with the first business day in ~ 996, again omitting New Year's Day. The enforcement sample contained 1,915 trucks, and Me non-enforcement sample contained 1,671 trucks. Table 9 on the next page summarizes the observed M.O.E comparison between conditions. Truck weights were marginally lower on average, 48,228 versus 50,646 during the enforcement period. During the enforcement period, the average overload violation was Appendix C 22
Table 9. Minnesota Measures Sensitivity Experiment Two Gross Weight Violation, Proportion I Gross Weight Violation, Severity Single-axle Weight Violation, Proportion Single-axIe Weight Violation, Severity Tandem-axle Weight Violation, Proportion Tandem-axie Weight Violation, Severity Bndge Formula Violation, Proportion Bridge Formula Violation, Severity Excess ESALs, Proportion Excess ESALs, Severity Note: Weight units are pounds 4.19 °/0 2,323 6.9 onto 1 ,054 1 1.6 onto 1,31 1 0.21 °/0 1 ,650 1i.4 / 0.57 .... .. :.:.: .. . :i.t,.i ~,, n, :.,: 1 .9oo/o 2,000 3.9 onto 1,230 8.3 TO ~ ,314 0.43 TO 1 ,700 10.7 TO 0.55 ....... i ~1~..-... Yes No* Yeses No Yes No No No No* No* * = Non-signif cant tendency lower, i.e., 2,000 pounds compared with 2,323 pounds during the non-enforcement period; however this difference was not statistically significant. There was a significant decrease in Me proportion (~.90 versus 4.19 percent) of gross-weight overload violations during the enforcement period. Also, a comparison of axIe-specific violations revealed a greater proportion of overloads on the steering and last axles during the period of non-enforcement. Decreased average ESALs were observed dunng the enforcement period; however, axIe-specific analyses demonstrated Mat the decrease could not be associated with specific axles. The proportion of trucks exhibiting Excess ESALs, and their associated severity, while exhibiting tendencies to demonstrate valid enforcement effects, did not significantly differ between the enforcement and non-enforcement conditions. Generally smaller proportions of single-axle violations were observed during the enforcement condition, with the most pronounced difference being a reduction from 6.9 to 3.9 percent for the form axle. Very small differences (1,054 versus 1,230 pounds) in the severity of average axle violations were observed between the baseline and enforce- ment conditions. 23 Appendix C
Smaller proportions of tandem violations were observed during the enforcement condition, with the most pronounced difference being a reduction from ~ I.6 to 8.3 per- cent for the drive tandem. No statistical effect was associated with severity of the tandem violations, as nearly identical average tandem violations (1,31 ~ and 1,314 pounds) were observed between baseline and enforcement conditions. A very small number of Type 9 trucks were observed to exhibit Bridge Formula violations. Four (of 1,915 trucks) during the enforcement condition and seven (of 1,617 trucks) dunng the non-enforcement condition were in violation. The level of observed violation, i.e. 1,700 and 1,650 pounds was also quite small. These violations did not statistically differ between the enforcement and non-enforcement conditions. In summary, this data set revealed a few truck weight effects (i.e., a lower proportion of overweight trucks and lower average ESALs) which support the M.O.E. validation effort. Minnesota M.O.E. Validation Summary Data sets representing two weeks of continuous traffic monitoring were provided by Me Minnesota Depardnent of Transporta- tion. Ben(ling-plate WIM data were gathered approximately five miles from a permanent truck-weight enforcement scale dunng times when the scale was both open and closed. While generally weak M.O.E. validation findings were seen in Minnesota results, one data set did exhibit a smaller proportion of gross weight and tandem axle violations along with a tendency for less severe excess ESALs, and the other set produced a tendency to lower Bridge Formula violations. The results confirmed the following M.O.E.s: (~) Gross Weight Violation Proportion, and (2) Tandem-axIe Violation Proportion. Summary of Measures-Sensitivi~cy FieIc! Vaticiation Study Candidate M.O.E.s were developed during We course of NCHRP Project 20-34 based on their suitability to demonstrate truck weight enforcement effects: Proportion and Severity of Gross Weight Violations, Proportion and Severity of Single-axie Weight Violations, Proportion and Appendix C 24
Severity of Tandem-axle Weight Violations, Proportion and Severity of Bridge Formula Violations, and Proportion and Severity of Excess ESAL Violations. These measures addressed legal load-limit compliance objectives of truck weight enforcement procedures as well as the potential for overweight trucks to produce pavement wear and tear. However, a field study was necessary to examine the sensitivity of these measures to actual field truck weight enforcement operations. This four-state effort examined WIM data gathered in the presence of enforcement activities and compared it with data collected trader non-enforcement affected flow conditions. Data collection periods controlled for day-of-week, time-of-day, and seasonal effects. Findings for each state are summarized as follows. California The Californua Department of Transportation provided output from a WIM scale located on I-5. An analysis of 3,678 truck combinations exhibited lower gross weights with a smaller proportion of overweight axles dunng the time when the weigh station was open. Data on a su~sample of 2,370 tractor-semi-tra~ler combinations was further analyzed to examine M.O.E. sensitivity to the enforcement activity. The results confirmed Me following M.O.E.s: Tandem-axte Weight Violation Severity, Bndge Formula Violation Proportion, and Excess ESAL Severity. Georgia Mobile truck-weight enforcement operations, utilizing an obtrusive portable roadside weigh scale, were conducted at a rural interstate location. An analysis of 483 combination trucks revealed a number of M.O.E. validation effects associated with observed axle and tandem weights. Under conditions of visible (and unexpected) mobile enforcement operations, the observed truck sample exhibited lower steenng-axie weights, lower rear-axle weights, and lower rear-tandem weights. During the surprise enforcement operation, a number of overweight trucks were observed to either park alongside the roadway or divert to alternate routes. The results confirmed the following M.O.E.s: Single- axIe Weight Violation Proportion, Tandem-axle Weight Violation, and Excess ESAL Severity. 2s Appendix C
Idaho A large volume of WIM data, i.e., gathered on approximately 29,000 com- mercial vehicles, was provided by the Idaho Transportation Department. A comparison of baseline versus enforcement conditions during three different weekdays produced a number of significant findings. While no day-of-week effects were readily evident to indicate on which days enforcement effort would more likely be effective, all of the tested operational measures were shown to be sensitive to enforcement activity. M.O.E.s most consistently demonstrating sensitivity to enforcement activity were: (1) Gross Weight Violation Propor- tion, (2) Single-axle Weight Proportion, (3) Tandem-axle Weight Proportion' and (4) Ex- cess ESAL Proportion. While less frequently associated with enforcement activity, the fol- lowing measures were also validated in the Idaho data: (1) Gross Weight Violation Sever- ity, (2) Single-axle Weight Violation Severity, (3) Tandem-axle Weight Violation Severity, and (4) Excess ESAL Severity. Minnesota Data sets representing two weeks of continuous traffic monitoring were provided by the Minnesota Department of Transportation. Bending-plate WIM data were gathered approximately five miles from a permanent truck-weight enforcement scale during times when the scale was both open and closed. While generally weak M.O.E. validation findings were seen in Minnesota results, one data set did exhibit a smaller proportion of gross weight and tandem axle violations along with a tendency for less severe excess ESALs, and the other set produced a tendency to lower Bridge Formula violations. The results confirmed the following M.O.E.s: (1) Gross Weight Violation Proportion, and (2) Tandem-axle Violation Proportion. Overview A large number of factors were seen to affect M.O.E. sensitivity to enforcement procedures, including actt3;l1 truck weight/configuration characteristics, shipping commodity demands, observed truck sample size, and WIM equipment variables. Table 10 on the next page summarizes which M.O.E.s were shown to be sensitive to actual truck weight enforcement actives in each of the states. It is highly evident that all M.O.E.s will not discriminate between enforcement conditions at every site. Appendix C 26
Table 11. Measure Sensitivity Summary by State t~i-~-~:y~ ~E~MiO.=E.~::~ i: ~ ~1 :~::~ CAN :~ - ~ | Grid :| ~ MA I Gross Weight Violation, Proportion | ~| | ~l I Gross Weight Violation, Severity | ~| | ~| l ingle-axle Weight violation, Proportion I I ~I ~l l ingle-axle WeightViolation, Severity | I ~I l andem-axle Weight violation, Proportion I | ~| ~l . l andem-axle Weight violation, Severity | ~| I ~l ridge Formula Violation, Proportion | ~| :' I Bridge Formula Violation, Severity | | ~I I . Excess ESALs, Proportion | ~| ~I ~ I Excess ESALs, Severity | ~| ~| ~l l Legend: 0= Significant elect; ~ = Non-significant tendency 27 Appendix C
