Surrogate articular contact models for computationally efficient multibody dynamic simulations.

Area of Science:

• Multibody Dynamics
• Computational Mechanics
• Contact Mechanics

Background:

• Contact analysis in multibody dynamic systems is computationally intensive, limiting sensitivity and optimization studies.
• Existing methods struggle with the high cost of repeated contact simulations, especially in complex systems like the human musculoskeletal system.

Purpose of the Study:

• To present a novel surrogate modeling approach for computationally efficient three-dimensional elastic contact analysis in multibody dynamics.
• To address challenges in applying surrogate modeling to elastic contact problems and demonstrate its effectiveness.

Main Methods:

• Developed a computationally cheap surrogate contact model by fitting it to data from an expensive elastic contact model (e.g., finite element or elastic foundation).
• Applied the surrogate model to multibody dynamic simulations of a Stanmore wear simulator for a total knee replacement.
• Validated accuracy by comparing wear volumes with elastic foundation (EF) models and assessed computational speed via Monte Carlo analyses.

Main Results:

• Surrogate contact models predicted wear volumes within 1.5% accuracy compared to EF models.
• Computational time was drastically reduced from an estimated 284 hours per analysis to 1.4 hours (17 min vs. 5 s per simulation).
• Higher wear sensitivity was observed for motion variations than for load variations in the simulations.

Conclusions:

• The proposed surrogate modeling approach significantly enhances computational speed for multibody dynamic simulations involving 3D elastic contact.
• This method enables more efficient and feasible sensitivity and optimization studies for systems with complex contact interactions.