Linear correlations are observed between CMV and point-MVs. Moisture content should be considered in correlations for fine-grained soils. Roller results in a composite value in a layered soil condition. CMV is affected by roller speed (higher speeds result in lower CMV)
Dynapac SD; CMV; gravelly sand, silty sand, and fine sand
Sand cone, pressure-meter, PLT, CPT, and DCP
Compaction growth curves showed improvement in CMV and other mechanical properties (i.e., modulus and cone resistance) with increasing pass. Relative percent compaction (or density) was not sensitive to changes in CMV
Correlations with modulus and DCP measurements are generally better than density. CMV measurements are dependent on speed, vibration frequency and amplitude, soil type, gradation, water content, and strength of subsoil
Correlation between CMV and PLT modulus (initial) showed different regression trends for partial uplift and double jump operating conditions. Regressions in partial uplift and double jump conditions yielded and 0.6, respectively
Dry density and CMV increased with increasing roller pass on a calibration test strip. Linear regression relationships with are observed for correlations between dry density and CMV
Correlations between MDP and in situ test measurements using simple and multiple regression analyses are presented. MDP correlated better with dry density () than with DCP () or Clegg impact value (). Including moisture content via multiple regression analysis improved the values for DCP and Clegg impact value (). Results are based on data averaged 20 m long strip per pass
L. Petersen and R. Peterson [29]; TH53, Duluth, Minn, USA
Caterpillar SD; CMV and MDP; fine sand
LWD, DCP, and SSG
Weak correlations are obtained on a point-by-point basis comparison between in situ test measurements and roller measurements, likely due to the depth and stress dependency of soil modulus and the heterogeneity of the soils. Good correlations are obtained between CMV values and DCP measurements for depths between 200 and 400 mm depth
Average MDP values showed a decreasing trend on a log scale, and dry unit weight and DCP index values showed an asymptotic decrease with increasing roller pass. Correlations between MDP and point-MVs showed good correlations ( to 0.9). Incorporating moisture content into analysis is critical to improve correlations for dry unit weight
Correlations between MDP and point-MVs are presented using simple and multiple regression analysis. Averaging the data along the full length of the test strip (per pass) improved the regressions. Multiple regression analysis by incorporating moisture content as a regression parameter further improved the correlations
Caterpillar SD; CMV; poorly graded sand and well-graded sand with silt
LWD, DCP, and NG
Project scale correlations by averaging data from different areas on the project are presented, which showed values ranging from 0.52 for density and 0.79 for DCP index value. Correlations with LWD showed poor correlations due to the effect of loose material at the surface. The variability observed in the CMV data was similar to DCP and LWD measurements but not to density measurements
Based on average measurements over the length of the test strip (~20 m); correlations between MDP and point-MVs showed for density and 0.96 for DCP index values
Correlations were obtained on a test bed with multiple lifts placed on a concrete base and a soft subgrade base. Correlations between MDP and point-MVs yielded to 0.85 for spatially nearest point data, and to 0.92 for averaged data (over the length of concrete or soft subgrade)
Caterpillar PD-MDP80 and SD-CMV; sandy lean clay to lean clay with sand
Heavy test roller, DCP, LWD, and PLT
Correlations are presented from multiple calibration test strips and production areas from the project. MDP80 and LWD modulus correlation showed two different trends ( and 0.65) over the range of measurements as the MDP80 reached an asymptotic value of about 150 which is the maximum value on the calibration hard surface. CMV correlation with LWD modulus produced , and with rut depth produced
White et al. [13], TH36, North St. Paul, Minn, USA
Caterpillar SD; CMV; granular subbase and select granular base
DCP, SSG, Clegg Hammer, LWD, PLT, FWD, and CPT
Correlations between CMV and point-MVs from calibration and production test areas based on spatially nearest point data are presented. Positive trends are generally observed with (for LWD, FWD, PLT, SSG, and Clegg) with exception of one test bed (FWD, LWD, and CPT) with limited/narrow range of measurements
Caterpillar SD; CMV; poorly graded sand with silt to silty sand
LWD, PLT, and DCP
Correlations between CMV and point-MVs from calibration and production test areas based on spatially nearest point data are presented. Correlations between CMV and point-MVs showed value ranging from 0.2 to 0.9. The primary factors contributing to scatter are attributed to differences in measurement influence depths, applied stresses, and the loose surface of the sandy soils on the project. Correlations between CMV and LWD or DCP measurements improved using measurements at about 150-mm below the compaction surface
White et al. [13], CSAH 2, Olmsted County, Minn, USA
Caterpillar PD; MDP80; sandy lean clay
LWD
MDP80 values are influenced by the travel direction of the roller due to localized slope changes and roller speed. Correlations between MDP80 and LWD generally showed (with exception of one case) when regressions are performed by separating data sets with different travel directions and speed. Data was combined by performing multiple regression analysis incorporating travel speed and direction which showed correlations with
Mooney et al. [25]; Minnesota, Colorado, Maryland, North Carolina, Fla, USA
Caterpillar PD-MDP and SD-CMV, Dynapac SD-CMV; two types of cohesive soils, eleven types of granular soils
NG, DCP, LWD, FWD, PLT, Clegg hammer, SSG
Simple and multiple regression analysis results are presented. Simple linear correlations between RICM and point-MVs are possible for a compaction layer underlain by relatively homogenous and a stiff/stable supporting layer. Heterogeneous underlying conditions can adversely affect the correlations. A multiple regression analysis approach is described that includes parameter values to represent underlying layer conditions to improve correlations. Modulus measurements generally capture the variation in RICM values better than dry unit weight measurements. DCP tests are effective in detecting deeper “weak” areas that are commonly identified by RICM values and not by compaction layer point-MVs. High variability in soil properties across the drum width and soil moisture content contribute to scatter in relationships. Averaging measurements across the drum width, and incorporating moisture content into multiple regression analysis, can help mitigate the scatter to some extent. Relatively constant machine operation settings (i.e., amplitude, frequency, and speed) are critical for calibration strips and correlations are generally better for low amplitude settings (e.g., 0.7 to 1.1 mm)
Dynapad SD; CMV; granular base and lime stabilized subgrade
NG, LWD, PLT, FWD, D-SPA
CMV measurements showed good repeatability but are influenced by vibration amplitude. High amplitude (i.e., >1.5 mm) caused drum bouncing and affected the CMV measurements. Increasing amplitude generally showed an increase in CMV. Results showed that FWD modulus point measurements tracked well with variations in CMV in some cases and in some cases it did not. The reason for poor correlations with FWD measurements in some cases is attributed to the possible influence of heterogeneity observed in the material across the drum width due to moisture segregation. The CMV measurements however were well correlated with variations in moisture content as evidenced by a decrease in CMV with increasing moisture content. D-SPA, PLT, and DCP measurements tracked well with the variations in CMV
Nonlinear power, exponential, and logarithmic relationships are observed between RICM and point-MVs. Correlations between RICM values and different point-MVs are generally weak when evaluated independently for each test bed due to narrow range of measurements. When data are combined for site wide correlations with a wide measurement range, the correlations improved. RICM values generally correlated better with modulus/stiffness and CBR point-MVs than with dry density point-MVs. Correlations between RICM values and FWD measurements showed the strongest correlation coefficients