Fast Numerical Simulation (CAE)

Numerical simulations have been central in industrial product and process development for decades. In structural mechanics, the Finite Element Method (FEM) has become the established standard.

Due to the computational demands, numerous attempts have been made to accelerate simulation workflows — including high‑performance computing (HPC) approaches and model complexity reduction techniques. However, these often lead to compromises in accuracy or require extensive preparation.

At Fraunhofer IGD, we take a different approach: we develop fast, highly accurate simulation algorithms based on the finite element method (FEM) that run entirely on GPUs, leveraging their massive parallelism for maximum performance.

Our technology delivers results 10–100 times faster — without compromising accuracy. Simulations that once required hours can now be completed within minutes.

Our solutions are particularly tailored to structural mechanics and distortion simulation in metal 3D printing (esp. laser‑based powder bed fusion).

Traditional FEM solutions make limited use of the enormous computational capacity of modern GPUs. Often, only a subset of compute steps is offloaded, which leads to intensive data transfer between the CPU and GPU a significant bottleneck that limits potential speedups.

Even when GPUs are used more consistently, memory architecture constraints remain challenging. Available GPU memory is smaller than CPU memory, and inefficient data structures can hinder performance on large models.

With RISTRA, our GPU‑accelerated structural mechanics solver, the entire simulation pipeline runs directly on the GPU. Through specialized algorithms and memory‑efficient data structures, RISTRA fully exploits GPU parallelism, thus, overcoming limitations of other implementations.

The result: extreme acceleration with consistently high accuracy — ideal for structural mechanics and additive manufacturing.

Contact us — our team at Fraunhofer IGD in Darmstadt will be happy to advise you on the RISTRA product family and its application.