Related Experiment Video
Updated: Mar 8, 2026
A Method to Estimate Cadaveric Femur Cortical Strains During Fracture Testing Using Digital Image Correlation
Published on: September 14, 2017
Validation of a parametric finite element human femur model
Katelyn F Klein1, Jingwen Hu1,2, Matthew P Reed1,3
1a University of Michigan Transportation Research Institute , Ann Arbor , Michigan.
Objective:
Finite element (FE) models with geometry and material properties that are parametric with subject descriptors, such as age and body shape/size, are being developed to incorporate population variability into crash simulations. However, the validation methods currently being used with these parametric models do not assess whether model predictions are reasonable in the space over which the model is intended to be used. This study presents a parametric model of the femur and applies a unique validation paradigm to this parametric femur model that characterizes whether model predictions reproduce experimentally observed trends.
Methods:
FE models of male and female femurs with geometries that are parametric with age, femur length, and body mass index (BMI) were developed based on existing statistical models that predict femur geometry. These parametric FE femur models were validated by comparing responses from combined loading tests of femoral shafts to simulation results from FE models of the corresponding femoral shafts whose geometry was predicted using the associated age, femur length, and BMI. The effects of subject variables on model responses were also compared with trends in the experimental data set by fitting similarly parameterized statistical models to both the results of the experimental data and the corresponding FE model results and then comparing fitted model coefficients for the experimental and predicted data sets.
Results:
The average error in impact force at experimental failure for the parametric models was 5%. The coefficients of a statistical model fit to simulation data were within one standard error of the coefficients of a similarly parameterized model of the experimental data except for the age parameter, likely because material properties used in simulations were not varied with specimen age. In simulations to explore the effects of femur length, BMI, and age on impact response, only BMI significantly affected response for both men and women, with increasing BMI producing higher impact forces.
Conclusions:
Impactor forces from simulations, on average, matched experimental values at the time of failure. In addition, the simulations were able to match the trends in the experimental data set.
