U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-HRT-04-094
Date: November 2004
Evaluation of LS-DYNA Soil Material Model 147
PDF Version (2.87 MB)
PDF files can be viewed with the Acrobat® Reader®
CHAPTER 8. USER'S CONCLUSIONS AND RECOMMENDATIONS
Significant progress was made examining the effects of material input parameters for the FHWA soil material model developed for LS-DYNA. Appropriate values for some material parameters were found and are presented in this chapter.
From in situ density testing performed in the user's facility, soil densities between 2.082E-6 kg/m3 and 2.242E-6 kg/m3 are reasonable. Soil density is critical when examining the dynamic effects of soil behavior.
For the NCHRP 350 strong soil, reasonable bulk and shear moduli values for a "sand and gravel" soil would be K = 80.5 MPa (0.0805 GPa) and G = 48.3 MPa (0.0480 GPa), respectively. However, these values yield an overly stiff response with the FHWA soil model. Suggested values are significantly lower. Values that appear to yield the appropriate stiffness are on the order of 3.25 MPa and 1.3 MPa for the bulk and shear moduli, respectively.
The baseline value of 63 degrees (1.1 radians), as determined through physical testing performed by the user, is recommended for Phimax, the angle of internal friction. This value is reasonable for the NCHRP 350 strong soil of crushed limestone as used at the user's facility.
An appropriate value for the hyperbolic coefficient, Ahyp, is cot EQ \F(c,20)(). This value was shown to provide stable performance both theoretically and through parameter studies. Unfortunately, this value is dependent on cohesion, Coh, and the angle of internal friction, Phimax. This requires manual calculation to determine the value that adds a human error factor to the simulation.
Therefore, it is recommended that the default value of cot EQ \F(c,20)() be used when the Ahyp field is left blank. This will allow for fewer parameters that may need to be adjusted, while still giving the option of altering the value, if necessary. Additionally, a check should be performed to ensure that the Ahyp parameter is not less than zero. This creates a yield surface that is outside that of the original Mohr-Coulomb yield surface.
In order to maintain similarity to the original Mohr-Coulomb failure envelope, values on the order of 1.0E-7 are recommended for Ahyp. For an internal angle of friction, Phimax, equal to 63 degrees (1.1 radians) and a cohesion of 6.2E-6 GPa, the recommended criteria yields a value of Ahyp = 1.58E-7 GPa.
It is recommended that the values of cohesion for "cohesionless" soil be placed at approximately 6.2E-6 GPa. This value appears close enough to zero, but still allows the plasticity routines to converge relatively rapidly. It should also be determined whether the value of cohesion should be increased to compensate for the soil dilation caused by Taylor's aggregate interlock.
The parameter Itermax is a difficult parameter to specify. This parameter determines the number of iterations for the plasticity routines. This would be much better as a tolerance. As it currently stands, a parameter study of Itermax must be performed for every simulation that is run. Specifying a value for convergence would be more straightforward and would not require the end user to perform comparisons of varying numbers of iterations to determine the appropriate quantity.
For various reasons, several features of the new soil model were not available for complete analysis, including strain-rate effects, moisture content, and pore-water pressure.Additionally, researchers were unable to find physical testing methods or analytical methods for determining the appropriate values for several of the soil parameters.
Although extensive progress has been made on the soil material model, there is considerably more to be accomplished before the model would be effective in most roadside safety applications. The current implementation of the soil material model appears to be applicable for only small displacement problems (on the order of 25 to 50 mm). Techniques for modeling large soil displacement problems were discussed; however, there was not enough time to apply these concepts to the new soil material model.