U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content U.S. Department of Transportation/Federal Highway AdministrationU.S. Department of Transportation/Federal Highway Administration

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-05-063
Date: May 2007

Evaluation of LS-DYNA Concrete Material Model 159

PDF Version (6.84 MB)

PDF files can be viewed with the Acrobat® Reader®

Chapter 10. Summary and Recommendations

Developer Comments

A material model for concrete was developed and implemented into the ls-dyna finite element code. Also implemented was a suite of default material properties as a function of concrete compressive strength and maximum aggregate size. The model and default properties were evaluated for use in roadside safety applications through correlations with test data. Some of the test data-including those related to drop tower and bogie vehicle impact into reinforced concrete beams-as generated specifically for this effort. Some of the test data was generated prior to this effort but made available for this study, including pendulum impact into bridge rails and quasi-static loading of a safety-shaped barrier.

Analyses of these tests indicate that in most cases, damage modes and deflection histories were accurately simulated using default material properties. In a few cases, alternative properties were suggested to simulate additional damage. The properties that seemed to be most critical are the fracture energies (Gft in tension, Gfs in shear, Gfc in compression), the rate effect on fracture energy (repow), and the maximum principal strain at which erosion occurs (ERODE). In general, the developer recommends using default properties of Gfs = Gft with Gfc = 100 Gft with repow = 1. The fracture energy in tension was obtained from values listed in the CEB. If more shear or compression based damage is desired, then use of Gfs = ½ Gft or Gfc = 50 Gf are reasonable reductions. If more overall damage is desired, then repow = 0.5 is a reasonable reduction. Reductions below these values are not recommended.

Erode is a control parameter. If a value is not specified (ERODE = 0.0), then erosion will not occur. The recommendation is to specify erosion between 5 and 10 percent maximum principal strain (1.05 £ ERODE £ 1.10). By taking this step, severe damage is able to accumulate and be maintained before erosion occurs. In numerous calculations, erosion set to 5 and 10 percent strain gave similar results. Erosion at 1 percent strain was evaluated but typically simulates a structural response that is more flexible than at 5 to 10 percent strain and may be more flexible than measured. Once an element erodes, then a 25- to 38-mm (1- to 1.5-inch) gap forms (the dimension of the element eroded). This type of gap is much larger than a hairline crack. Thus, it is reasonable to delay erosion for lightly cracked structures, particularly if the structures are cycling between tension and compression (opening and closing cracks).

During the computational review of the concrete material model, the user recommended the following changes:

  • Adjust fracture energies to 80 percent of the preliminary values used by the user. This change was made, and any reference to default values within this report refers to the updated fracture energies. These updated fracture energies are those documented in the companion to this report's Users Manual and are in agreement with those tabulated in the CEB. The updated fracture energies are, however, 80 percent less than those predicted by the formula given in the CEB. An inconsistency appears to exist between the tabulated values and formula documented in CEB for aggregate sizes greater than 19 mm (0.75 inches).
  • Correction to an unrealistic increase in the concrete internal energy. This correction was made to the material model energy term.
  • Correction to some reporting problems associated with the long input format of the concrete material model. These corrections were made. The companion to this report Users Manual and material model printout routines reflect these changes.

User Comments

Numerical analyses were conducted using the newly developed LS-DYNA concrete material model. The analyses showed very good potential for useful application of the model in analyses of steel-reinforced concrete roadside safety structures. The T4 bridge rail study is a good example of this potential. The analyses of the T4 bridge rail alternatives provided a good benchmark for evaluating the strengths, sensitivity, and usability of the new concrete material model. Use of the pendulum tests provided more controlled impact conditions and eliminated many variables associated with the comparison of simulation to full-scale vehicle crash tests. Two specimens were tested for each design variation, providing some information regarding system variability. Furthermore, the two design variations demonstrated different levels of damage to the concrete parapet, which provided an opportunity to assess the sensitivity of the concrete model to small design changes under similar loading conditions.

After adjusting the fracture energies to 80 percent of their preliminary baseline values, good overall correlation was achieved between the T4 bridge rail simulations and pendulum test data for both design alternatives investigated. The simulation matches reasonably well with respect to the damage profile, concrete fracture (i.e., element erosion in simulation), peak impact force, and acceleration-time history of the pendulum impactor.

After calibrating the fracture energies, the simulation of the T4 bridge rail system with three-bolt anchorage and 254-mm- (10-inch-) wide parapet captured the punching shear failure of the concrete underneath the steel base plate and post. As in the pendulum tests, the concrete failure radiated out from the anchor bolts across the top surface of the concrete parapet and extended down the back side of the parapet wall. When the same material properties were used in a subsequent simulation of the T4 bridge rail with four-bolt anchorage and 317.5-mm- (12.5-inch-) wide parapet, the results again correlated with the test data. No concrete fracture was observed, and the damage fringes matched the pattern of hairline cracks observed in one of the test specimens.

The user recommended that the default fracture energies be adjusted to levels found to provide damage correlation between simulations and tests. The developer followed this recommendation, which is available in subsequent releases of LS-DYNA 971. Results of the parametric analyses also indicate that element erosion should be used in conjunction with a maximum principal strain failure criterion (i.e., the ERODE parameter should be set greater than 1.0). The current default setting (ERODE = 1.1) accomplishes this recommendation.

The model is stable and has a reasonable time step. Some problems were identified with the energy balance associated with use of the concrete material model. As damage accumulates in the concrete structures, an artificial or nonphysical increase in internal energy appears to causes the total energy of the system to increase dramatically. This glitch was reported to the developer, and a fix was implemented in subsequent releases of LS-DYNA 971.

Some reporting problems related to the long input format of the concrete material model were also identified. The parameter listing generated by the output routines does not correspond to the written documentation of the model. Details of this problem were reported to the developer, and corrections were made in subsequent releases of LS-DYNA 971.

The short input format is recommended for a first cut analysis and to generate the long input format values. However, the user is encouraged to perform experiments to determine values that best fit his or her concrete data.

All analyses reported herein were conducted using a single processor on a Linux platform because releases for other hardware and multiple processors were not available or were unstable. However, a few analyses were conducted using different platforms and multiple processors as some of the beta binaries became available during the course of the project. The analyses conducted on different platforms showed some differences in the number of eroding elements in the simulations involving concrete fracture. However, the differences are not considered to be significant or indicative of a flaw in the material model. Moreover, the eroding pattern was the same across all calculations for a given case. Note that erosion takes place when the damage value associated with a particular element exceeds a prescribed threshold (if desired by the user). The numerical analysis has established that different hardware platforms, roundoff errors, and number of processors have an influence on results consistency. This influence was observed in other research efforts where a sudden collapse, element erosion, and any other discontinuity phenomenon were numerically analyzed. Therefore, the researchers believe that these differences for the cases studied are not attributed to the material model but rather to other factors.

Although the results of the evaluation of the concrete material model are encouraging, resource limitations precluded a more indepth evaluation of the model. Further study is recommended to more fully quantify the applications and limitations of the material model. The investigation of the New Jersey safety-shape bridge rail is incomplete due to lack of time and resources. Further investigation of the quasi-static load tests would be helpful. Additionally, an analysis of the pickup truck impact into this barrier and/or a portable concrete barrier is recommended.

Previous | Table of Contents | Next

United States Department of Transportation - Federal Highway Administration