TRANSPORTATION RESEARCH RECORD 1311 Comparison of the South Dakota Road Profiler ith Other Rut Measurement Methods ]AMES B. DuBosE During the fall of 1989, the Illinois Department of Transportation ompleted the onstrution of a profile-measuring van that as based on the South Dakota road profiler. One feature of the van is the ability to ollet rut depths at highay speed at 2-ft intervals. The ability to ollet more rut data and to do it more quikly and safely ere of great interest to the department. Hoever, no information as available that desribed ho the data obtained ith the road profiler ould ompare ith manual rut measurements or rut measurements obtained ith other automated systems. In an effort to determine if there as any orrelation beteen the different rut-measuring methods, a number of experiments ere onduted. Three method -(a) South Dakota road profiler, (b) PASCO, and () manual- ere ompared for a 7.5-mi streth of FA 49 (US-5) in St. Clair and Clinton ounties; the road profiler and manual methods ere ompared for all of the Interstate highays in Distrit 5, loated in east-entral Illinois. In addition, the proedures used by eah method ere analyzed to theoretially determine ho ell the methods ould agree and also to help explain any observed differenes in the data. On the basis of the results of these experiments, reommendations ere made desribing the most appropriate use of the road profiler data. During the fall of 1989, the Illinois Department of Transportation ompleted the onstrution of a profile-measuring van based on the South Dakota road profiler. One feature of the van is the ability to ollet rut depths at highay speed at 2-ft intervals. The ability to ollet more rut data and to do it more quikly and safely as of great interest to the department. Hoever, no information as available that desribed ho the data obtained ith the road profiler ould ompare ith manual rut measurements or rut measurements obtained ith other automated systems. In an effort to determine if there as any orrelation beteen the different rut measuring methods, a number of experiments ere onduted. Three methods-(a) South Dakota road profiler, (b) PASCO, and () manual-ere ompared for a 7.5-mi streth of FA 49 (US-5) in St. Clair and Clinton ounties; the Road Profiler and manual methods ere ompared for all of the Interstate highays in Distrit 5, loated in east entral Illinois. In addition, the proedures used by eah method ere analyzed to determine theoretially ho ell the methods ould agree and also to help explain any observed differenes in the data. Illinois Department of Transportation, Bureau of Materials and Physial Researh, 126 East Ash Street, Springfield, Ill. 6274-4766. DESCRIPTION OF METHODS The South Dakota road profiler uses three aousti sensors to measure rut depths. These sensors are loated as follos : (a) left heelpath, (b) enter of lane, () right heelpath. The rut depth is defined as the average of the distanes measured by the to outside sensors minus the distane measured by the enter sensor. This quantity is equal to the height of the hump beteen the heelpaths. Only one measurement is obtained at eah loation and separate readings annot be determined for eah heelpath. PASCO's method is a photographi one. A hair line projetor is used to projet a blak line aross the idth of the pavement at night. The resulting image is then photographed using a pulse amera. For a pavement ith no ruts, the blak line ill be perfetly straight. The line ill be avy for a rutted road. To obtain a quantitative measure of the amount of rutting present, PASCO digitizes the photographi images and orrets for the amera angle. A omputer an then produe orreted images and measure the amount of rutting in eah heelpath. The manual method is one that has been used in Illinois sine 1985. Measurements are made using a 6-ft aluminum beam and a measuring shoe. The measuring shoe looks like a miniature stairase from the side ith eah step representing an inrement of.5 in. The largest inrement that an be measured is 1 in. Rut readings are obtained by plaing the beam aross half of the lane and then randomly sliding the measuring shoe under the beam until the maximum reading is obtained. Separate measurements are made for eah heelpath. THEORETICAL AND PRACTICAL CONSIDERATIONS Even before onduting any field experiments, one an theorize that there ill be instanes here the three methods ill not agree. The PASCO and manual methods measure basially the same thing, but use to different approahes. The fat that the manual readings are only reorded to the nearest.5 in. may affet the degree of orrelation. The road profiler an only report one rut measurement at eah loation. This is supposed to be the average of the to heelpath ruts, but is atually the height of the hump in the enter of the lane. The folloing examples ill illustrate that these to quantities are not the same.
2 The road profiler rut depth is defined as here ht> h 2, and h 3 are defined in Figure 1, hih shos that R equals r, the height of the hump. The road profiler ill agree ell ith manual measurements hen the height of the hump is the same as the average of the to heelpath ruts. This situation ours hen the hump and the to edges an be onneted by a straight line (i.e., they are all in the same plane). Figure 2 illustrates this priniple. In this ase, the left edge, A, the hump, B, and the right edge, C, an be onneted by a straight line. Even if the line ABC is not parallel to the line DE onneting the bottom of the to ruts, r ill still equal (m 1 + m 2 )/2, here m 1 equals the left heelpath rut and m 2 equals the right heelpath rut. Beause the sensors are equidistant, as ABC rotates and inreases either m, or m 2, the other value ill derease by exatly the same amount. Unfortunately, the edges and the hump are often not in the same plane. This presents to additional possibilities. The first ase is hen the hump is higher than the to edges. Figure 3 shos this ondition. In this ase, r ill be greater than the average of m 1 and m 2 In other ords, the road profiler ill measure larger ruts than a manual survey. The opposite situation ours hen the hump is loer than the to edges (see Figure 4). In this ase, the road profiler ill measure smaller ruts than a manual survey beause r is smaller than the average of m 1 and m 2 A pratial onsideration that ill influene the degree of orrelation is hether or not the road profiler driver is able to onsistently drive in the deepest part of the ruts. The vehile ill have a natural tendeny to ant to follo the path of least resistane, hih ill normally be in the deepest ruts. Driver error ill prevent the road profiler from staying in the deepest ruts 1 perent of the time. Driver error is more likely to adversely affet the average obtained on a short FIGURE I rut depth. South Dakota definition of FIGURE 2 Road profiler rut depth equal to average manual rut depth. FIGURE 3 Road profiler rut depth greater than average manual rut depth. FIGURE 4 Road profller rut depth less than average manual rut depth. TRANSPORTATION RESEARCH RECORD 1311 setion than a long one beause of the limited data olleted on short setions. In addition to driver error, the Road Profiler ill also have diffiulty traveling in the deepest ruts hen the distane beteen ruts is greater than the van's heelbase. This situation is possible hen truk traffi reates a double rut ith dual axles rather than the normal single-rut pattern. FA 49-PROJECT LEVEL COMPARISON FA 49 is an experimental highay loated in St. Clair and Clinton ounties. Seven and one-half miles of FA 49 are onstruted of various thiknesses of full-depth asphalt. Figure 5 shos the layout of the experimental setions. All three methods ere used to survey FA 49. The PASCO survey as onduted on June 23, 1989, the manual survey on August 19, 1989, and the road profiler survey on Deember 1, 1989. Figures 6 and 7 sho ho the average rut depths ompare for eah setion on the basis of a measurement interval of 2 ft. Beause the three surveys ere not onduted at the same time, are should be exerised in analyzing these data. In addition to variations in the number of axle loads arried before eah survey, there as also some pathing done during July and August 1989 that ould affet the results. For these reasons it as deided to try to estimate ho the three methods ould ompare if the surveys ere all done on June 23, 1989. These estimates ere made using the ross slopes obtained from PASCO. The next three figures illustrate ho rut depths ere estimated from these ross slopes. Figure 8 shos the PASCO method. The rut depths listed on the figure ere measured hy representatives of PASCO. Note that, in this ase, PASCO defined the rut depth as the distane from an imaginary line onneting the to edges of the pavement to the bottom of the rut. If the hump in the enter of the lane had been higher than this imaginary line, PASCO ould have based the rut depth on the distane from the bottom of the rut to an imaginary line onneting the enter hump to the appropriate edge. l'igure 9 illustrates the manual method. These measurements ere made by draing a straight line six sale feet in length at eah heelpath and then measuring the distane from eah line to the bottom of eah rut. The South Dakota road profiler method is shon in Figure 1. For this method, a sale draing of the road profiler van as superimposed on the ross slope. The rut depth is defined as the differene beteen the average of the distanes measured by the outside sensors and the distane measured by the enter sensor. This quantity is equal to the height of the hump at the enter point. For expedieny, the rut an be determined by draing three parallel segments of equal length, one at eah sensor loation, and then measuring the amount the enter segment protrudes through the ross slope hen the other to segments are just touhing the ross slope. Notie that the exat same profile as used for eah figure, but different rut readings ere obtained for eah method. The estimates did not alays agree ith the atual field measurement beause of the time differene and pathing mentioned previously and also beause it is muh easier to loate the deepest rut from ross slopes than it is to do so in the field. Figures 11 and 12 sho the estimated averages for eah setion. Setions H, J, and L are very short in length (less
DuBose 3 T "' 285' 1' 167' 1218' UD NUD NUD UD L L L L "I" i: I: AC T 6' UD L E - 2 "' 14' 24' NUD NUD L NL M Ml 9112" I: 14978' UD L p 11" AC 2 "I""" "'..!!=2. :: T + is I :I a CODE: UD = Underdrains NUD No Underdrains L = Lime Modifiation NL = No Lime Modifiation M = Moisture Gauge TB = Tipping Buket T = Thermoouple F= Frost Gauge FIGURE 5 FA 49 experimental setions..4 - :z:. 3 Q, Q :I - MANUAL :::J ROAD PROFILER. 2 1., 1 l r A B C D E H J K L M Ml N P FIGURE 6 Average rut depth by setion, eastbound FA 49, 1989 data.. 4 :z:. 3 Q, Q. 2 1- :::i -MANUAL CJROAD PROFILER ABCDEH J K L M Ml N P FIGURE 7 Average rut depth by setion, estbound FA 49, 1989 data. than 1, ft) and therefore no firm onlusions should be dran from the data olleted from these setions. Setions here the hump in the enter of the lane rose above the to edges, suh as Setion N, had higher average rut depths in the road profiler survey than in the manual survey. The opposite phenomenon ours if the hump in the enter of the lane is loer than the to edges. At first glane at Figure 6, it appears that this is hat has happened in Setions B and C. Hoever, examination of the ross slopes indiates that hat atually happened as that the road profiler got out of the heelpaths and therefore did not travel in the deepest rut.
4 TRANSPOR TATION RESEARCH RECORD 1311 (IN.) 5. 5 4" 8" = = :-:=-- - =-=--====== - - ES T M ANU A L. 4 ::::J EST ROAD PROFILER ::. :. 3 - ""' Cl. Q ::I ""'. 2. 1 l j I 2 3 4 6 6 7 8 9 1 11 12 (FT.) FIGURE 8 PASCO rut depth measurement method. A B D E H K L M M1 N P FIGURE 11 Estimated average rut depth by setion, eastbound FA 49, 6/23/89 data. (IN.) " ". 5 r -:::::::;::::::=-;::;-... =-=::::::=:r:::::?--== I,,.,..- ---. 25. 4 :.3 ""' Cl. Q ""'. 2 ::I. 1 - ES T M A NUA L ::::::::J EST ROAD PROFILER 2 3 4 5 6 7 8 9 1 1 1 12 (FT.) FIGURE 9 Manual rut depth measurement method. A B C D E H K L M M1 N P FIGURE 12 Estimated average rut depth by setion, estbound FA 49, 6/23/89 data. (IN.) 5. 5 2" 2 3 4 5 6 7 8 9 1 11 12 (FT.) FIGURE 1 South Dakota road profiler rut depth measurement method. DISTRICT 5 INTERSTATES-NETWORK LEVEL COMPARISON The Interstate system in Distrit 5 as tested ith the South Dakota road profiler during Deember 1989. The data obtained ere ompared ith data obtained manually during the Summer of 1989 as part of the Illinois Pavement Feedbak System (IPFS). The rut values obtained from the IPFS survey ere based on the average of four measurements in a 5-ft sample unit. The four measurements ere taken in the folloing manner: one in eah heelpath at the beginning of the sample unit and one in eah heelpath at the end of the sample unit. The road profiler data ere summarized at - mi (528-ft) inrements on the basis of the average of 264 readings. For this omparison, the manual survey averages ere ompared ith the road profiler average for the -mi segment that as loated losest to the sample unit. The results of this omparison are plotted separately for eah Interstate route and diretion and are shon in Figures 13-16. The to methods did not ompare exatly, but in most ases appeared to indiate the same trends. A number of fators ould have aff ted the mparison. Comparing the average of four mea urements ith the average of 264 measurements is a rather tenuous proedure for a quantity as variable as rut depth. Also, the time differene beteen surveys probably affeted the results. Hoever, even if all of these fators ou Id be ontrolled, the to methods ould still not agree exatly beause they do not measure ruts in the sa me manner. Some enouragement an at least be found in the fat that the to methods seem to indiate the same general trends.
.8. 7 -.6 D...4.3.8. 7 -.6 D...4.3 17 19 21 23 25 7 72 74 76 a MANUAL ROAD PROFILER 78.8.7.6 -. 5 D...4.3 17 19 2 1 23 25 a MANUAL. ROAD PROFILER FIGURE 13 Rut depth versus milepost, 57, Distrit 5: top, southbound; bottom, northbound..8. 7.6 a.. 4 u..... 3. 2!J-1 68 7 72 74 76 a MANUAL. ROAD PROFILER FIGURE 15 Rut depth versus milepost, 72, Distrit 5: top, eastbound; bottom, estbound. 78.8. 7.6 D...4.3. 2 1 12 14 16 a MANUAL ROAD PROFILER.7. 6 - I D.. 1-.4.3 +-.,---.-r-.,---r--; 15 17 19 21.8. 7.6 D...3 FIGURE 14 Rut depth versus milepost, 7, Distrit 5: top, eastbound; bottom, estbound..7. 6 - D...4.3 O +--.---.-...-.,---r- 15 '17 19 21 FIGURE 16 Rut depth versus milepost, 74, Distrit 5: top, eastbound; bottom, estbound.
6 SUMMARY There are fundamental differenes in the ay the three methods disussed in this paper measure ruts. Fators suh as the shape of the pavement ross slope and driver error an dramatially affet the orrelation beteen ruts obtained using the road profiler and other methods. The stepped inrement of O.OS in. used in the manual method may also affet the orrelation. P ASCO's rut measurement method appears to be an aurate one. Hoever, this option has some other onsiderations that make it less attrative. It requires either buying the equipment, hih is expensive, or ontrating PASCO to provide the servie. Contrating ith PASCO is also expensive and requires advane notie to shedule a projet for testing. After the projet is tested, it may take several months to reeive the data. Despite its limitations, the South Dakota road profiler has some intriguing properties. Its relatively lo ost and ability to ollet rut data at SS mph at 2-ft intervals make its use an attrative option for state highay departments. Illinois' experiene has indiated rut depth to be a highly variable property. Beause of this variability, it may be more useful to the department to have a large quantity of road profiler data, even if it is less aurate, than the amount of manual data it urrently an ollet, hih is limited by manpoer onstraints. In addition, beause of safety onstraints, high traffi areas suh as the Chiago area expressays an only be surveyed using an automated method suh as the South Dakota road profiler. RECOMMENDATIONS TRANSPORTATION RESEARCH RECORD 1311 A reasonable proedure is to use the South Dakota road profiler to ollet rutting data on a netork basis. After the data have been olleted, they an be analyzed by the individual highay distrits. Beause the distrits deal ith these highays on a daily basis, they may have opinions about hether or not the data seem aurate. If a distrit feels that the data are in error, manual measurements an be taken to validate or invalidate the data. FURTHER RESEARCH Researh should be undertaken to determine the proper definition of the rut depth. All three methods differ in the ay they define the rut depth. The folloing questions should be studied: 1. Should rut depth measurements be relative to a level referene in order to predit the depth of ater that an aumulate? 2. Is the hump in the enter of the lane a useful quantity for performane predition? Publiation of this paper sponsored by Committee on Surfae Properties- Vehile Interation.