Skip to contentUnited States Department of Transportation - Federal Highway AdministrationSearch FHWAFeedback
Highways for LIFE

Arrow Asphalt Binder Cracking Device to reduce Low Temperature Asphalt Pavement Cracking

<< Back Content Next >>

Task 4. Ruggedness Test

The purpose of the ruggedness test is to identify sources of variation in ABCD testing. ABCD units and asphalt binder samples were delivered to North Central Superpave Center (NC) and University of Wisconsin, Madison (WI) asphalt laboratories. Sang–Soo Kim of EZ Asphalt provided one day on–site training at each laboratory to the equipment operators (August 25–27, 2008). The ruggedness test was designed following ASTM C 1067 – 00 "Standard Practice for Conducting A Ruggedness or Screening Program for Test Methods for Construction Materials". Of seven factors considered, the effect of lubrication was dropped from the test since conducting the ABCD test without lubrication might seriously damage ABCD rings during cleaning.

4A. Seven factors being studied in Ruggedness Test
  1. Cooling Rate (18C/hr versus 22C/hr)

    Temperature profiles to be used is

    18°C/hr 22°C/hr
    20°C to 0°C in 0.5 hour (30 minutes)
    0°C to –63°C in 3.5 hours (3 hour 30 minutes)
    –63 to 25°C in 0.5 hour (30 minutes)
    Hold at 25°C for 1.0 hour
    20°C to 0°C in 0.5 hour (30 minutes)
    0°C to –66°C in 3.0 hours
    –66 to 25°C in 0.5 hour (30 minutes)
    Hold at 25°C for 1.0 hour
  2. Size of Protrusion (5.85mm diameter vs. 6.35mm diameter [0.23" versus 0.25"])

    The two larger protrusion molds also have a hole at the opposite side.

  3. Sample Trimming (0.8mm over trimming versus even trimming)

    For over trimming, 0.8 mm depression (about 5 cm long) is created at the surface of two silicone molds. Carefully trim the asphalt binder so that the trimmed surface is flush with the mold surface.

  4. Lubrication (No Lubrication versus Lubrication)

    Two molds and 2 ABCD rings are lubricated with glycerin–talc mixture.

    The other 2 molds and rings are kept clean and dry without lubrication.

  5. No Turntable versus Turntable

    For 'no turntable test', carefully conduct pouring, trimming, testing and other handling without the use of turntable. For turntable test, place the molds on the turntable and then perform pouring, trimming, testing and other handling.

  6. No cold joint versus cold joint

    To create cold joints in ABCD sample, fill the mold ½ full with the heated asphalt binder. Place the half filled mold in 0°C chamber for 5 minutes and place the binder container back in the 170° oven for 5 minutes. After 5 minutes, completely fill the mold with the heated asphalt binder.

  7. Conditioning time before starting test (0 minute vs. 30 minutes)

    All 4 samples are placed in 0°C chamber at the same time. For 30 minute conditioning time, remove two samples after 20 minutes from the 0°C chamber. Trim the sample and then place them in the room temperature for 30 minutes. The other two samples are removed after 50 minutes at the 0°C chamber. Trim them and then start the test immediately.

Each lab was instructed to follow test procedure given below.

The laboratories used the following mold identification numbers:

Mold #1: Smaller diameter (0.23") protrusion and no depression on the surface.
Mold #2: Smaller diameter (0.23") protrusion and 0.8mm depression on the surface.
Mold #3: Larger diameter (0.25") protrusion and no depression on the surface.
Mold #4: Larger diameter (0.25") protrusion and 0.8mm depression on the surface.

For Each Binder

Run #1: 18°C/hr Test (4 specimens)
  Mold #1 Mold #2 Mold #3 Mold #4
Cooling Rate 18°C/hr 18°C/hr 18°C/hr 18°C/hr
Lubrication Lubricate Lubricate None None
Turn Table None Turn Table None Turn Table
Cold Joint None Cold Joint Cold Joint None
Conditioning Time 30 minutes 0 minute 0 minute 30 minutes
Data #1 Data #2 Data #3 Data #4
Run #2: 22°C/hr Test (4 specimens)
  Mold #1 Mold #2 Mold #3 Mold #4
Cooling Rate 22°C/hr 22°C/hr 22°C/hr 22°C/hr
Lubrication None None Lubricate Lubricate
Turn Table Turn Table None Turn Table None
Cold Joint None Cold Joint Cold Joint None
Conditioning Time 0 minutes 30 minute 30 minute 0 minutes
Data #5 Data #6 Data #7 Data #8

Repeat Run #1 and Run #2

Repeat Run #1: 18°C/hr Test (4 specimens)
  Mold #1 Mold #2 Mold #3 Mold #4
Cooling Rate 18°C/hr 18°C/hr 18°C/hr 18°C/hr
Lubrication Lubricate Lubricate None None
Turn Table None Turn Table None Turn Table
Cold Joint None Cold Joint Cold Joint None
Conditioning Time 30 minutes 0 minute 0 minute 30 minutes
Data #9 Data #10 Data #11 Data #12
Repeat Run #2: 22°C/hr Test (4 specimens)
  Mold #1 Mold #2 Mold #3 Mold #4
Cooling Rate 22°C/hr 22°C/hr 22°C/hr 22°C/hr
Lubrication None None Lubricate Lubricate
Turn Table Turn Table None Turn Table None
Cold Joint None Cold Joint Cold Joint None
Conditioning Time 0 minutes 30 minute 30 minute 0 minutes
Data #13 Data #14 Data #15 Data #16

'No Lubrication' was dropped due to the possibility of damaging ABCD rings.

The four binders used in the study are as follows

EZ 1 EZ 2 EZ 3 EZ 4
PG 64–16 (AAM–1) PG 70–22M (SBS) PG 58–28 (AAA–1) PG 64–34M (SBS)

All binders were RTFO aged followed by PAV aging and degassing prior to sending them to the participating laboratories. The specimen cooling rate is determined by the slope of the best fit line of ten data points (about 0.56°C change during 100 seconds) prior to cracking. The average and the standard deviation of the specimen cooling rate of test specimens for each lab are as follows:

Laboratory WI NC EZ
Cooling Rate Low High Low High Low High
Average, C/hr –18.6 –22.8 –18.7 –22.9 –18.8 –23.2
st. dev., C/hr 0.25 0.55 0.32 0.38 0.30 0.93

When the slope of a longer time period is used, the specimen cooling rate approaches the intended rates (18 and 22°C/hr).

4B. Ruggedness Test Results
4B.1. ABCD Cracking Temperature

On November 15, 2008, the last set of ruggedness test results were received. Following ASTM C 1067, ABCD cracking temperature data (Table 13) were analyzed and summarized in Table 14. Following the ASTM ruggedness analysis procedure, any factors with F Statistics greater than 5.59 are considered to have significant effects on ABCD cracking temperature with 5% significance level. Standard deviation is the Root–Mean–Square–Error (RMSE). When reviewed for each combination of four asphalt binder types and three laboratories, there are 12 cases as shown in Table 14. The size of protrusion and the over–trimming were the factors affecting test results most significantly (four significant cases each). Formation of a cold joint during sample preparation had two significant cases. Cooling rate and conditioning time had one significant case each.

The required ASTM ruggedness analyses presented in the previous paragraph addressed the significant variables at a given laboratory and for a given asphalt binder. The required analyses do not directly compare the laboratories statistically. Therefore, in addition to the required analyses, EZ Asphalt investigated the overall statistical differences using Analysis of Variance (ANOVA) with the results shown in Table 15. As expected, ABCD cracking temperature clearly differentiates between asphalt types (F value = 436, p–value < 0.001). However, there are significant variations in cracking temperature between laboratories. Among the ruggedness factors, only 'Over Trim' is statistically significant in affecting ABCD cracking temperature (significance level less than 1%).

Table 13. Results of Ruggedness Test.
Lab Binder ID Data: ABCD Cracking Temperature, C
#1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13 #14 #15 #16
WI EZ 1 –31.2 –26.8 –30.1 –29.4 –34.8 –32.0 –32.9 –30.4 –33.3 –33.7 –33.8 –30.8 –30.4 –29.5 –30.8 –31.4
EZ 2 –41.0 –37.7 –39.3 –42.2 –38.6 –40.1 –42.9 –37.5 –41.5 –36.9 –38.4 –39.2 –39.0 –40.3 –42.1 –38.3
EZ 3 –35.6 –40.0 –37.9 –35.7 –37.4 –39.5 –38.4 –37.9 –41.9 –37.4 –40.1 –37.3 –38.6 –39.9 –40.8 –35.5
EZ 4 –48.3 –47.1 –47.3 –44.8 –50.8 –48.5 –48.8 –46.0 –45.3 –42.4 –47.0 –44.9 –47.4 –45.4 –46.5 –45.4
NC EZ 1 –30.2 –30.6 –30.3 –28.3 –31.2 –32.9 –27.4 –28.5 –32.4 –32.6 –28.7 –30.3 –33.2 –31.4 –32.9 –31.8
EZ 2 –32.1 –32.2 –35.8 –32.6 –36.1 –33.6 –33.5 –35.9 –34.9 –33.4 –34.9 –38.9 –33.2 –32.7 –35.2 –37.6
EZ 3 –39.1 –36.3 –36.5 –36.1 –35.6 –37.1 –35.8 –36.7 –38.1 –37.6 –31.8 –34.7 –37.7 –36.6 –35.1 –36.8
EZ 4 –46.3 –40.8 –45.3 –41.0 –44.9 –46.6 –39.6 –41.9 –41.3 –40.3 –41.9 –41.2 –43.9 –39.4 –43.1 –43.7
EZ EZ 1 –31.0 –29.2 –29.9 –28.3 –30.0 –30.9 –31.1 –29.1 –29.8 –33.3 –32.1 –28.3 –29.8 –28.3 –29.5 –28.9
EZ 2 –36.3 –36.3 –36.8 –34.8 –36.4 –36.9 –36.4 –33.2 –36.3 –37.2 –35.5 –34.4 –36.4 –37.0 –37.2 –34.8
EZ 3 –35.2 –32.9 –36.3 –32.8 –33.1 –33.2 –33.4 –35.0 –34.4 –35.5 –36.3 –34.9 –36.2 –37.0 –36.9 –34.5
EZ 4 –44.9 –43.7 –45.0 –42.8 –43.3 –43.0 –44.0 –41.0 –43.9 –42.8 –44.5 –43.0 –44.3 –44.0 –43.9 –42.2
Lab Binder Data: ABCD Strain Jump at Fracture, ìå
#1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13 #14 #15 #16
WI EZ 1 39.7 17.0 97.5 33.8 47.9 26.0 17.0 19.9 32.3 40.4 27.2 21.0 46.3 97.7 64.3 65.0
EZ 2 54.0 36.7 38.6 36.7 42.1 37.8 44.7 31.4 58.8 20.4 26.3 33.4 36.6 47.2 32.6 49.9
EZ 3 18.5 45.6 25.4 21.3 30.6 41.0 24.9 26.0 45.6 25.1 27.7 38.8 31.2 47.8 29.5 18.4
EZ 4 62.8 29.3 28.6 47.9 74.9 65.5 43.1 55.4 59.8 35.9 35.4 28.2 54.9 40.4 31.2 35.8
NC EZ 1 32.6 25.2 18.9 12.0 25.4 29.6 12.8 8.8 31.4 14.0 8.1 23.6 33.8 29.3 25.7 40.9
EZ 2 7.1 14.7 8.9 9.8 21.5 16.6 13.4 28.1 9.9 9.1 28.4 33.7 8.6 9.6 13.4 43.0
EZ 3 12.0 17.7 22.8 22.0 18.9 34.4 28.7 34.9 32.0 30.0 8.4 23.8 39.1 18.7 23.8 29.2
EZ 4 31.4 11.6 28.4 17.1 20.7 38.4 18.1 18.0 14.3 16.1 21.1 12.7 21.1 15.6 10.5 16.5
EZ EZ 1 37.0 39.7 29.4 31.2 39.4 47.9 23.9 35.2 30.6 37.5 35.2 29.8 25.5 27.8 20.8 26.8
EZ 2 42.2 28.6 34.8 34.8 40.2 49.7 31.4 32.9 38.7 44.8 25.9 35.5 40.5 41.8 31.2 40.5
EZ 3 30.5 16.3 20.4 13.6 26.2 11.5 9.1 26.4 24.3 27.4 23.2 23.3 34.0 33.1 18.9 24.6
EZ 4 64.8 49.9 48.3 38.5 48.7 52.3 34.5 33.9 46.6 46.4 41.4 44.4 52.4 51.6 35.7 41.8
Table 14. Summary of F Values for All Laboratories, All Materials, and All Factors (ABCD Cracking Temperature)
Lab Asphalt Std. Dev. (RMSE) Cooling Rate Protrusion Over Trim Lubrication Turn Table Cold Joint Condition Time
WI EZ 1 2.34 NS NS NS NT NS NS NS
EZ 2 0.83 NS NS 58.27 NT NS NS 1432.82
EZ 3 1.93 NS NS NS NT NS NS NS
EZ 4 1.87 8.36 NS 36.40 NT NS NS NS
NC EZ 1 1.87 NS 16.51 NS NT NS NS NS
EZ 2 1.91 NS 13.75 NS NT NS NS NS
EZ 3 1.54 NS 27.57 NS NT NS NS NS
EZ 4 2.49 NS NS NS NT NS NS NS
EZ EZ 1 1.39 NS NS NS NT NS NS NS
EZ 2 0.57 NS 82.70 18.82 NT NS 122.44 NS
EZ 3 1.65 NS NS NS NT NS NS NS
EZ 4 0.63 NS NS 133.54 NT NS 7.44 NS

NS: not significant at 95% confidence level
NT: not tested
Critical F Statistics at 95% confidence level = 5.59

Table 15. Analysis of Variance of Ruggedness Test including Laboratories and Binder Types (Cracking Temperature)
Dependent Variable: ABCD Cracking Temp
Source Type III SS df Mean Square F Sig.
Corrected Model 4860.693(a) 11 441.881 130.313 .000
Intercept 263551.470 1 263551.470 77722.475 .000
Lab 365.832 2 182.916 53.943 .000
Asphalt 4434.398 3 1478.133 435.908 .000
Cool Rate 3.825 1 3.825 1.128 .290
Protrusion 10.407 1 10.407 3.069 .082
Over Trim 39.513 1 39.513 11.652 .001
Turn Table 3.440 1 3.440 1.014 .315
Cold Joint 1.860 1 1.860 .549 .460
Condition Time 1.418 1 1.418 .418 .519
Error 610.367 180 3.391    
Total 269022.530 192      
Corrected Total 5471.060 191      

a R Squared = .888 (Adjusted R Squared = .882)

Linear regression can provide quantified measures of significant variables, in terms of factor adjusted mean differences as shown in Table 16. In comparison to EZ Asphalt lab data, the average ABCD cracking temperatures determined by NC lab is 0.019°C warmer and those WI is 2.9°C lower. In comparison to EZ 1 binder (PG 64–16), EZ 2 (PG 70–22M) showed 6.0°C lower ABCD cracking temperatures on average; 5.8°C and 13.5°C lower for EZ 3 (PG 58–28) and EZ 4 (PG 63–34M), respectively. Figure 7 shows the average effects of laboratory and binder type. 'Over Trim' by 0.8mm resulted in 0.9°C warmer ABCD cracking temperatures on average. All other factors' influence in ABCD cracking temperature was less than 0.50°C and was not statistically significant.

Table 16. Linear Regression of Ruggedness Test Results (Cracking Temperature)
Model Unstandardized Coefficients Standardized Coefficients t Sig.
B (°C) Std. Error (°C) Beta
1 (Constant) –30.226 .460   –65.657 .000
NC .019 .326 .002 .058 .954
WI –2.919 .326 –.258 –8.966 .000
EZ2 –6.042 .376 –.490 –16.073 .000
EZ3 –5.825 .376 –.473 –15.497 .000
EZ4 –13.540 .376 –1.098 –36.021 .000
CoolRate –.282 .266 –.026 –1.062 .290
Protrusion .466 .266 .044 1.752 .082
OverTrim .907 .266 .085 3.414 .001
TurnTable .268 .266 .025 1.007 .315
ColdJoint –.197 .266 –.018 –.741 .460
Condition –.172 .266 –.016 –.647 .519

This figure illustrates the average effects of laboratory and binder type on cracking temperature in ABCD Ruggedness Test Results.  There were variations among laboratories. University of Wisconsin laboratory showed lowest cracking temperatures in general.
Figure 7. ABCD Ruggedness Test Results (Cracking Temperature); Comparison of Laboratory and Binder Type

4B.2 Strain Jump at Fracture

ABCD strain jump (å) at fracture can be converted into the fracture stress of the asphalt binder (óf ) as follows:

óf = (K) å EABCD AABCD/Abinder (or óf = 147 å kPa)

where,

K = stress concentration factor, 2.02
EABCD = Young's modulus of ABCD ring, 140 GPa
AABCD = Cross sectional area of ABCD ring, 21.0 x 10–6 mm2
Abinder = Cross sectional area of asphalt binder, 40.3 x 10–6 mm2

Following ASTM C 1067, data (Table 13) were analyzed and summarized in Table 17. When reviewed for each combination of four asphalt binder types and three laboratories, there are 12 cases as shown. The size of protrusion is the factor affecting test results most significantly (six significant cases). For all four binders tested at the EZ Asphalt laboratory, the effect of protrusion size is significant. Lack of significance for other laboratories may be due to larger overall test variability (within lab).

Table 17. Summary of F Values for All Laboratories, All Materials, and All Factors
Lab Asphalt Std. Dev. (RMSE) Cooling Rate Protrusion Over Trim Lubrication Turn Table Cold Joint Condition Time
WI EZ 1 30.74 NS NS NS NT NS NS NS
EZ 2 8.13 NS NS NS NT NS NS 5.99
EZ 3 9.97 NS NS NS NT NS NS NS
EZ 4 11.30 7.89 52.80 NS NT NS 40.45 NS
NC EZ 1 10.14 NS NS NS NT NS NS NS
EZ 2 9.45 NS 6.46 NS NT NS NS NS
EZ 3 9.59 NS NS NS NT NS NS NS
EZ 4 7.77 NS NS NS NT 8.09 NS NS
EZ EZ 1 6.89 NS 12.12 NS NT NS NS NS
EZ 2 5.44 NS 24.35 NS NT NS NS NS
EZ 3 7.46 NS 7.77 NS NT NS 7.22 NS
EZ 4 5.61 NS 169.3 NS NT NS NS NS

NS: not significant at 95% confidence level
NT: not tested
Critical F Statistics at 95% confidence level = 5.59

In preparation for the Interlaboratory test, the statistical difference between laboratories was also examined and the results are given Table 18. ABCD strain jump at fracture can differentiate asphalt types (F value = 6.68, p–value < 0.001). Among the ruggedness factors, 'Protrusion' size and use of 'Turn Table' are statistically significant in affecting ABCD strain jump (significance level less than 1%).

Linear regression provides quantified measures of these differences, in terms of factor adjusted mean differences as shown in Table 19. In comparison to EZ Asphalt lab data, the average ABCD strain jump determined by NC lab is 13 ìå lower and that by WI is 5 ìå higher. Increasing the protrusion diameter from 5.84 mm (0.23 in.) to 6.34 mm (0.25 in.) lowered the strain jump at failure by 5 ìå. The use of turntables in sample preparation and handling also lowered the strain jump at failure by 4 ìå. Figure 8 shows the average effects of laboratory and binder type on ABCD strain jump at fracture.

Table 18. Analysis of Variance of Ruggedness Test including Laboratories and Binder Types (Strain Jump)
Source Type III Sum of Squares df Mean Square F Sig.
Corrected Model 17545.530(a) 11 1595.048 11.190 .000
Intercept 194298.025 1 194298.025 1363.055 .000
Lab 11749.674 2 5874.837 41.214 .000
AC 2857.001 3 952.334 6.681 .000
CoolRate 410.670 1 410.670 2.881 .091
Protrusion 1346.201 1 1346.201 9.444 .002
OverTrim 3.521 1 3.521 .025 .875
TurnTable 900.467 1 900.467 6.317 .013
ColdJoint 269.327 1 269.327 1.889 .171
Condition 8.670 1 8.670 .061 .805
Error 25658.284 180 142.546    
Total 237501.840 192      
Corrected Total 43203.815 191      

a R Squared = .406 (Adjusted R Squared = .370)

Table 19. Linear Regression of Ruggedness Test Results (Strain Jump)
Model Unstandardized Coefficients Standardized Coefficients t Sig.
B Std. Error Beta
1 (Constant) 40.078 2.985   13.427 .000
NC –13.270 2.111 –.417 –6.288 .000
WI 5.336 2.111 .168 2.528 .012
EZ2 –1.798 2.437 –.052 –.738 .462
EZ3 –6.796 2.437 –.196 –2.789 .006
EZ4 3.940 2.437 .114 1.617 .108
CoolRate 2.925 1.723 .097 1.697 .091
Protrusion –5.296 1.723 –.177 –3.073 .002
OverTrim –.271 1.723 –.009 –.157 .875
TurnTable –4.331 1.723 –.144 –2.513 .013
ColdJoint –2.369 1.723 –.079 –1.375 .171
Condition .425 1.723 .014 .247 .805

This figure illustrates the average effects of laboratory and binder type on strain jump in ABCD Ruggedness Test Results.  There were large variations among laboratories. NCSC data showed the lowest strain jump in general.
Figure 8. ABCD Ruggedness Test Results (Strain Jump); Comparison of Laboratory and Binder Type

The analysis of the ruggedness test shows that ABCD testing is

  1. Able to differentiate different asphalt binders in terms of their low temperature cracking potential and strength (strain jump at failure).
  2. At each laboratory, the ABCD test results were reproducible.

However, as the statistics above indicate, there was significant variability among laboratories. Therefore, the EZ Asphalt Technology laboratory performed investigative testing which resulted in modifying the ABCD test procedure. Then, additional samples of the binders (EZ1, EZ2, EZ3, and EZ4) were sent to the NC and WI labs, and those labs conducted supplemental tests to validate the updated ABCD test procedure.

<< Back Content Next >>

More Information

Events

Contact

Julie Zirlin
Highways for LIFE
202-366-9105
julie.zirlin@dot.gov

Updated: 04/04/2011
 

FHWA
United States Department of Transportation - Federal Highway Administration