sponsoring agencies.)

DEPARTMENT OF HIGHWAYS AND TRANSPORTATION VIRGINIA TESTING EQUIPMENT CORRELATION RESULTS SKID 1974, 1975, and 1978 N. Runkle Stephen Analyst Research opinions, findings, and conclusions expressed in this (The are those of author and not necessarily those of report Highway and Transportation Research Council Virginia Cooperative Organization Sponsored Jointly by Virginia (A of Highways & Transportation and Department University of Virginia) September 1978 by sponsoring agencies.) Charlottesville, Virginia VHTRC 79-R17

MAINTENANCE RESEARCH ADVISORY COMMITTEE MR. C. M. CLARKE, Chairman, Asst. District Engineer, VDH&T MR. D. R. COLLINS, Resident Engineer, VDHST MR. R. H. CONNOCK, JR., Asst. District Engineer, VDH&T MR. J. A. C0PP, Residency Maintenance Supervisor, VDH&T MR. W. L. HAYDEN, Asst. Materials Engineer, VDHgT MR. C. 0. LEIGH, Maintenance Engineer, VDHgT MR. J. C. MCCABE, Area Engineer, FHWA MR. J Ko MCEWEN, Resident Engineer, VDHST MR. D. C. MAHONE, Senior Research Scientist, VHSTRC MR. J. R. MILLER, Equipment Engineer, VDHST MR. R.. L. MOORE, Resident Engineer, VDH&T MR. C. B. PERRY, Resident Engineer, VDHgT MR. Do S. ROOSEVELT, Resident Engineer, VDH$T MR. M. B. VANN, Asst. Construction Engineer, VDHgT MR. J. E. WILLIAMS, Asst. District Engineer, VDHgT ii

purpose of this study was to modify, as required, The previously established relationships between VHTRC trailer, VHTRC stopping distance car, and first skid skid trailer, and to develop appropriate relationships VDHT above three devices for second, recently acquired with trailer. VDHT ion equations indicating relationships between Regress trailers are given below. In general se equations indicate relationships remain same with changes i. test speed, and from year to year; and in relationships are essentially one to one, 2. slopes generally are not significantly i.e., from 1.0 with some difference in average different as follows. results expected, values obtained with trailers were As than SDN values obtained with car at same initial less speed and relationships differ with test speed. How- test it is important to note that for se relationships ever, generally do differ significantly from 1.0; i.e., slopes in measured values between car and trailers is not difference from low to high levels of skid resistance, with constant being larger on low skid resistance sites. difference most survey skid data have been collected with VDHT Since #I, it seems reasonable to standardize results in Trailer terms this test unit. Thus, survey data collected with of VDHT Trailer newer should be corrected eir in terms of adding #2 or by using equation 4 in Table 3 of report. However, 2.5 should be pointed out that it correction of this type would no in a conservative (i.e. lower) interpretation of result data. survey of SDN values, PrediCtions stopping distances, should be or based on equations 17-21, 25, or 30-34 from Table 2 of made report, depending on test trailer used. ABSTRACT that VDHT Trailer #i : VDHT Trailer #2 + 2.5 VDHT Trailer #i VHTRC Trailer 3.0 VDHT Trailer #2 VHTRC Trailer- 7.0 iii

DEPARTMENT OF HIGHWAYS AND TRANSPORTATION VIRGINIA TESTING EQUIPMENT CORRELATION RESULTS SKID 1974, 1975, and 1978 N. Runkle Stephen Analyst Research newly acquired skid trailer to be utilized for Trailer), skid testing (VDHT Trailer #i), and Research survey were reported in "The Evaluation of New VDHT Skid studies Trailer" and "Methodology for Utilizing Survey Skid Testing three years had elapsed since last correlation Because and a second skid trailer (VDHT Trailer #2) had recently study Trailer #i had been used for several years to collect VDHT data and will be used as primary testing device on survey equations relating testing units were developed with regression VDHT Trailer #i as dependent variable, although this purpose of this study was to modify previously The relationships between VHTRC Trailer, VDHT established #i, and VHTRC Car as necessary based on new test results, Trailer to develop appropriate relationships for recently and by INTRODUCTION summer of 1974 a correlation study was performed During relate Council's skid trailer used in research (VHTRC to stopping distance car (VHTRC Car); and a similar study Council's performed in summer of 1975. The results of both was Data". 1,2) obtained for survey testing, anor correlation study been made in spring of 1978. Additionally, because was wet accident reduction program, it seemed appropri- Virginia's to designate this unit as standard unit. Thus, ate was not always done in 1974 and 1975 studies. PURPOSE AND SCOPE

to ASTM locked-wheel method for skid trailers conforming stopping distance method. and test data were collected on Lynchburg test loop All sites), which is routinely used for control testing by (six bald tires to provide as wide a range in measured skid and as possible and, for most part, were run at resistance mph, speed used for most testing in Virginia. Tests were 40 at multiple speeds with VDHT trailers to verify that run 1978 testing was performed during late March and early All 0nly limited results were obtained with car because April. mechanical problems, and data collected by VHTRC Trailer of first week of testing were later determined to be during because of malfunctions related to brakes in skid faulty Later, when this p oblem had been corrected, t sts trailer. obtained with three trailers. The results of all were are discussed below. tests results of all tests made during 1978 correlation The are shown in Table i. Each value.shown is average study value for five repeat tests at site for speed and tire indicated. Correlation results in previous studies test customarily been based on average values (s mple size have five) since for most purposes in Virginia reported test of are average of at least five tests. As indicated results test data obtained by VHTRC Trailer for previously, 3/20/78 3/22/78 were determined to be faulty because period brake problems in skid trailer, and thus are not shown of 2 contains all regression equations developed Table on 1978 data, 1974 and 1975 data, and combined 1974, 1975 based 1978 data. The equations as shown in Table 2 are numbered and reference purposes in discussion below. for VDHT Trailer #2. The scope of study was limited acquired primary test modes used in Virginia; namely, test to two VDHT trailers and was used in two previous corre- studies. Tests were performed with both treaded tires lation gradients were approximately same for se units, speed some repeat testing being performed at selected sites by with two trailers. Additionally, 1974 and 1975 data were se in all analyses performed. utilized RESULTS in Table i.

3-21-78 3-21-78 3-21-78 4-3-78 3-20-78 3-21-78 3-21-78 3-21-78 4-3-78 4-3-78 3-22-78 3-22-78 4-3-78 4-3-78 3-22-78 3-22-78 3-21-78, 3-21-78 4-3-78 4-3-78 3-22-78 3-22-78 4-3-78 4-3-78 Test Results i$ c Correlation at 20, 40, 60 mph) (Tests 50 New* 35 Bald New Bald New Bald New Bald New Bald New Bald New Bald New Bald New Bald New Bald New Bald. New Bald New Bald New Bald 67 46 28 36 9 16 45 15 42 20 52 64 31 50 44 55 15 24 57 22 58 28 52 36 51 34 42 35 60 18 I0 35 44 17 46 24 51 41 64 24 18 43 54 33 5O 30 29 37 7 14 42 12 39 16 57 44 68 21 14 42 53 19 55 22 49 62 27 44 51 30 47 31 42 32 57 19 13 32 41 15 42 2O 50 39 61 21 15 37 52 27 45 25 64 37 58 39 59 42 44 25 61 43 54 29 61 40 53 38 Table i VDHT #I VDHT #2 VHTRC Trailer VHTRC Car 20 40 60 20 6'(J '20 40 60 Date Site 4O 60 Tire 20 4-3-78 Treaded Tire *New

jx @

Relationships Between Trailers would expect relationship between trailers to be One one to one at all test speeds. As will be discussed essentially it is true that relationship between trailers does below, differ significantly from one test speed to anor, nor not slope of regression relationship differ significantly does a slope of one. However, it appears trailers do differ from slightly on average, as will be indicated. test as described by Volk.(3) In this case it was determined F no significant basis exists for using separate slopes, thus that equation inclusive of all tests speeds as shown in Figure 2 4 in Table 2) is appropriate for use. It can also (equation shown that on basis of a t test slope of combined be (0.99) is not significantly different from 1.0; thus equation would be appropriate eir to predict VDHT Trailer #i from it equation or to simply add 2.5 to VDHT Trailer #2 results is approximate average difference between two (2.5 trailers). 5 in Table 2 describes relationship between Equation two VDHT trailers with Trailer #2 as dependent variable. Trailer #2 from a VDHT Trailer #i value. in 1974, 1975, and 1978. Equation 6 (1974) is a performed equation for test speeds 30, 40, 50, 60, and 70 composite with no significant difference being found between mph, for each of se test speeds. Similarly, equation equations is a composite equation for test speeds 30, 40, and 60 mph 7 again, no significant difference being found between with, for each test speed during 1975. Equation 8 is equations derived from 1978 data with testing being performed at equation mph only (VHTRC Trailer) as shown in Table i. These three 40 (6-8) are shown toger in Figure 3. equations VDHT Trailer i and VDHT Trailer 2 I through 5 in Table 2 describe relationship Equations two VDHT trailers. Equations i through 3 describe between for test speeds of 20, 40 and 60 mph, and relationships shown graphically in Figure i. It is possible to test are significance of using a single slope for all test speeds using separate slopes for each speed by means of versus with equation 4, one may use equation 5 for prediction As or simply subtract 2.5 to obtain an estimated VDHT purposes VDHT Trailer i and VHTRC Trailer 6 through i0 in Table 2 describe relationships Equations VHTRC Trailer and VDHT Trailer #i for correlations between

Y20 mph : 8.75 + 0.88 x 20 mph Y40 mph : 3.93 +0.96 x 40 x //> Y60 mph : 0.68,,,L 20 30 40 50 60 10 I 6O 5O 4O 3O + 1.01 x 60 mph 2O i0 / VDHT Trailer #2 at i. Relationship between VDHT Trailers #I and #2 Figure 40, and 60 mph. 20,

20 mph Treaded Tire 40 mph Treaded Tire o 60 mph Treaded Tire 20 mph Bald Tire 40 mph Bald Tire 60 mph Bald Tire A y- 2.94 + 0.99 x Confidence Limits 95% any Predicted for,i 20 30 40 50 60 10 LEGEND 6O 5O 30 /o / y Value 2O / / 10 VDHT Trailer #2 S { 2. Relationship between VDHT Trailers #i and #2 at Figure test speeds combined. all

1974 y -2.98 +i.00 x Relationships between VHTRC Trailer and Figure Trailer #I for 1974, 1975, and 1978. VDHT 6O 5O 40 3O 0.84 20 10 i0 20 30 40 50 60 VHTRC Trailer

of large sample sizes involved, particularly Because 1974 and 1975, differences in slopes of se curves for Figure 3 it is clear that while differences in at curves were judged to be statistically significantly individual (principally 1975 curve), re is only about a different maximum difference over range from 20 to 40. Thus, 2.0 use. In comparing composite curve (equation 9) with for curves (Figure 3) difference in predicted individual Trailer #i value by using composite curve for any VDHT value of VHTRC Trailer is generally no more than 1.0. composite slope of 0.96 was found to be significant at The 90% level; so, while a slightly better estimate may be by using regression equation, it is judged by author made be satisfactory to predict VDHT Trailer #i values from to i0 describes relationship with VHTRC Equation as dependent variable. For this equation slope trailer significantly different from 1.0 at a 99% confidence level, is indicates desirability of using equation for pre- which purposes. (As above, however, maximum difference diction predicted VHTRC Trailer value when using between twelve data points shown in Table i (1978 data) 0nly available for developing relationship between VDHT were relationship remains same for various test speeds. How- based on trailer relationships discussed above, it is ever, equations previously discussed, slopes for equations ii 12 are not significantly different from 1.0. Thus it would and that one may predict VDHT Trailer #2 values from VHTRC appear values by subtracting about 9.0. However, true Trailer differences may be less as suggested by relationships between found to be significant at 95%, but not 99% (again by were of test in Volk referenced above). However, in looking method some specific reasons were known for slight year-to- unless changes, it seems that a composite curve would be appropriate year VHTRC Trailer values simply by subtracting about 3.0. or when simply adding 3.0 to VDHT Trailer #i equation value is less than 2.0.) VDHT Trailer #2 and VHTRC Trailer #2 and VHTRC Trailer. All data shown in Table 1 Trailer obtained at 40 mph, so it is not possible to determine if were probable 40 mph relationship would hold for or test speeds. ii and 12 in Table 2 describe relationship Equations VDHT Trailer #2 and VHTRC Trailer. As for most of between

VHTRC Trailer and VDHT Trailer #i (equations 6-I0). For two trailers average difference in 1978 was 5.4, se was reduced to 3.4 when 1974 and 1975 data were also which This reduction suggests that difference between considered. trailers during 1978 test period may have been relatively Thus, a more appropriate correction factor may be 7.0, high. this value should be verified through additional correlation but testing. be expected that values obtained by a locked-wheel skid should at a constant speed would differ from values obtained trailer a stopping distance car from initial speed to a speed with zero. In fact, Giles has shown that levels obtained with of = SDN45, 30 = SDN60, 40 30 = SDN30 40 2 40 = SDN A and 50 3 50 = SDN -A or,between Tr,ailers and St,oppin$ Distance Ca r Relationships is well known, changes with speed, which means it As stopping distance car should be equivalent to those obtg.i. ed a a trailer at two-thirds initial car test speeds.< ; That with where is trailer skid number and SDN stopping is, distance skid number, = SDN and 75' 50 = SDN or 90' 60 = SDN 2O 30 = SDN 26.7 40 = and SDN 33.3 50 = SDN 40 60 It follows that A, i,. = SDN A 60 60 4 i0

: + A SDN 40 2 40 = + A SDN 50 3 50 decrease in SDN from 75 mph to 50 mph, or increase in A 3 1975, and 1978 average trailer gradients were computed 1974 shown in Table 3 with corresponding values for A I through A 4. as Thus, equations above would become implied for car would be as shown in Table 4 The gradients values of A I through A (also 4 shown are estimated values for relationships between trailers and car will be The relative to last four equations above, with considered = + A SDN 30 i 30 = + A where SDN 60 4 60 A I = decrease in SDN from 45 mph to 30 mph, or increase in from 30 to 20 mph; A 2 decrease in SDN from 60 mph to 30 mph, or increase in from 40 to 26.7 mph; from 50" to 33.3 mph; and decrease in SDN from 90" mph to 60 mph, or increase in A 4 from 60 to 40 mph. values A Obviously, through Aa are dependent on or I speed gradient values. From multiple speed data in SDN = SDN 7.0, 30 30 = SDN 9.2, 40 40 = SDN 10.9, and 50 50 = SDN Ii.0. 6O 6O for A I and A 2 gradients, based on 1974 and 1975 data). that car should yield higher values for a given expectation speed. test ii

13 0

results for all tests run at 30, 40, 50, and 60 mph Combined 1974, 1975, and 1978 were used to develop equations 13 through in in Table 2. Of four equations, none have slopes signifi- 16 different from 1.0 at more than an 85% level of significance. cantly SDNs0 equations, while not identical to oretical equations These above, follow same pattern as oretical equations listed that correction factor increases as test speed increases. in in predicted values are generally no more than 2.0 Differences for SDN range 25-60, depending on wher one uses 3.0 equations 13-16 in Table 2 or modified equations regression VHTRC Car as dependent variable (equations 17-20, With 2), slopes do differ significantly from 1.0. In this Table equations 17-20 or four equations above. essence, difference between VHTRC Car and VDHT In #i is greater for low skid resistance pavements than Trailer because gradient is no doubt generally higher for occurrence skid resistance pavements. It does, however, mean that for low additional relationship of interest between se two One devices was evaluated. Since normal trailer survey testing testing is at 40 mph, oretical discussion above would primary and interstate highways. This relationship is shown for equation 21 in Table 2. While average difference is only as VDHT Trailer #I and VHTRC Car setting slope equal to 1.0 and taking average differ- Thus, in values yields ence SDN30 8.2 30 SDN40 12.0 40 50 = and 13.8, = SDN 14.7. 60 60 above with a slope of 1.0. SDN values may differ by as much as 6.0 to 8.0 for case range 20-60, depending on wher one uses regression high skid resistance pavements, as can be seen by plotting for of equations 13-20. This, of course, is a reasonable any purposes it is desirable to use regression equations prediction 13-20. indicate values are equivalent to SDN6^ values, or 40 SDN values that ould be obtained at approximate speed limit w 14

, with trailer yielding lower values, slope 4.3 significantly different from 1.0. As in or equations is car as dependent variable, difference is greater with low skid resistance sites (see Figure 4). In this case, at than at sites with high measured skid resistance (i.e. higher resistance in wheel path is relatively less), and car, skid testing with all four wheels and, at times, tending to slide by of wheel paths, would likely measure relatively higher out shown in Table i, only six data points (all at 40 mph) As obtained during 1978 testing upon which to develop were between VHTRC Car and VDHT Trailer #2. However, relationship would expect relationship to be very similar to that between one car and VDHT Trailer #I, since two VDHT trailers relate well (equations 1-16, Table 2). very 22 and 23 look somewhat different from equations 14 equations 18, differences in slopes are not statistically and at a high level of significance. Also, if only 1978 significant are considered, relationships between each of two data trailers and car are very similar, as shown in Figures VDHT and 6. In fact, for 1978 relationships only difference 5 essentially average difference between two trailers is about 2.5. Thus, it seems appropriate to approximate of and 25 in Table 2, should be verified by additional testing, 24 additional testing also being required for relationships with for this purpose were obtained during 1978 testing). data on data from 1974 and 1975 equations 26-34 in Table 2 Based developed, and y exhibit same basic trends as were for relationships between VDHT Trailer #i and equations difference at lower skid resistance sites is probably greater at se sites lateral change in skid resistance is because values than would trailer. VDHT Trailer #2 and VHTRC Car 22 and 23 show relationships between car Equations VDHT Trailer #2 based on limited 1978 data. While and between VDHT Trailer #i and VHTRC Car based relationships equations 14 and 18. These approximations, shown as equations on at speeds or than 40 mph. VHTRC Trailer and Car data from 1974 and 1975 were available for determining 0nly between VHTRC Trailer and VHTRC Car (no additional relationships VHTRC Car (equations 13-21). 15

=16.7+ SDII 6O.7O 4O 20 30 40 50 60 1974 Test Phase i 6O 1974 Test Phase 2 1975 o 5O 4O - 3O Line of Equality 2O i0 VDHT Trailer #i 40 mph 4. Relationship of VHTRC Car at 60 mph Figure VDHT Trailer #I at 40 mph. and 16

19 4, 75, and Combined Trailer #i) (VDHT y- 20.0 +.78 x Results" 1978 #I" y 23.5 + 0.67 x Trailer Trailer #2- y 24.9 + 0.69 x between Trailer #i and VHTRC Car with VHTRC ship as dependent variable. Car 6O 5O 4O 3O 1978 2O 10 i0 20 30 40 50 60 VDHT Trailer #i and #2 40 mph 5. 1978 regression equations for VDHT Trailers and Figure Car, and combined 1974, 75, and 78 relation- VHTRC 17

1974, Combined Trailer y -15.5 + = -31.8 + 1.42 y #2: Trailer y = -33.4 + 1.39 I, I I, 30 40 50 60 20 1' 6C 50 X 40 30 20 Results" Trai r -# i:- 1978 i0 VHTRC Car 40 mph 6. 1978 regression equations for VDHT Trailers Figure VHTRC Car, and combined 1974, 75, and 78 and between Trailer #i and VHTRC Car relationship VDHT Trailers as dependent variable. with 18

40 40 = SDN 11.3, and 50 50 = SDN 8.0. 60 60 above equations are not too different from oretical The discussed at beginning of this section, but are equations desirable for use because, in fact, slopes of regression not variable. The probable reasons for differences dependent slopes from 1.0 were discussed previously. of discussion would indicate, average test result is previous same (44.6 for trailer versus 43.8 for about That is, setting, slope of curves equal to 1.0 would yield SDN30 5.3, 30 = SDN 8.5, frequently do differ significantly from 1.0 at high equations of significance, particularly with. VHTRC Car as levels relationship of car at 60 mph to trailer at The mph was again determined as shown in equation 34. As 40 but again slope is significantly different from 1.0, car), car getting relatively higher values on low skid with resistance pavements (Figure 7). 19

SDN 6O O.65 40 o o 6O 5O o 14.97 + o = 4O Line of Equality 3O 2O i0 I0 20 30 40 50 60 VHTRC Trailer- 40 mph Relationship of VHTRC Car at 60 mph Figure VHTRC Trailer of 40 mph. and 2O

are conclusions and recommendations based on results Below this study as discussed above. of equations indicating relationships between Regression trailers are shown in Table 2. In general se equations relationships remain same with changes in i. speed, and from year to year; and test relationships are essentially one to one, 2. slopes generally are not significantly i.e., from 1.0, with some difference in average different as follows- results speed and relationships differ with test speed. (Regression test describing relationships between car and trailers equations also shown in Table 2.) However, it is important to note are for se relationships slopes generally do differ that from 1.0; i.e., difference in measured values significantly car and trailers is not constant from low to high between of skid resistance with difference being larger on levels skid resistance sites. low this test unit. Thus, survey data collected with newer of Trailer #2 should be corrected eir in terms of adding VDHT or by using equation 4 in Table 3. However, it should be 2.5 out that no correction of type just mentioned would pointed in a conservative (i.e. lower) interpretation of survey result data. of SDN values, or stopping distances, should be Predictions based on equations 17-21, 25, or 30-34 from Table 2, depending made CONCLUSIONS AND RECOMMENDATIONS Conclusions indicate that- VDHT Trailer #I VDHT Trailer #2 + 2.5 VDHT Trailer #i VHTRC Trailer 3.0 VDHT Trailer #2 VHTRC Trailer- 7.0. expected values obtained with trailers were As than SDN values obtained with car at same initial less Recommendations most survey skid data have been collecte.d with VDHT Since #i it seems reasonable to standardize results in terms Trailer on test trailer used. 21

Council, February 1975. for Utilizing Survey Skid Data" "Methodology Highway and Transportation Research Council, V' rginia Skid Prevention dbhference, Virginia Highway and national Research Council, 1959. Transportation REFERENCES S. N., "Evaluation of New VDHT Skid Testing Runkle, Virginia Highway and Transportation Research Trailer", i 0ctober 1975. William, Applied Statistic...s...f..qr. Engineers, pp. 241-252. Volk, Hill Book Company, Inc., 1958. McGraw C. G., "Some European Methods for Measurement Giles, Skidding Resistance" Proceedin..@s, Part I, First Inter of 22