Q: What models are supported by Suqaba's engine?

A: Suqaba supports the simulation and verification of models within structural and thermal mechanics, with the following core capabilities:

• Static linear elasticity

• Quasi-static nonlinear elasto-plasticity (isotropic and/or kinematic hardening)

• Steady-state heat transfer

• Thermo-mechanical coupling

These models cover a broad range of engineering use cases.

Q: How do you know that Suqaba's results are better?

A: Each simulation ends with a certification of the results' accuracy. Suqaba's framework is built upon the Constitutive Relation Error (CRE). The CRE gives a mathematically proven upper bound of the error between the approximate solutions and the true solutions. This result is rigorously derived from continuum mechanics theory and dual analysis.

Example: If the solver reports “the certified error is 2.1%,” it is mathematically guaranteed that the true error does not exceed 2.1%.

Q: What is the error that Suqaba quantifies and corrects?

A: Suqaba detects and corrects the numerical discretization error, i.e., the error created by approximating a continuous model with discrete representations.

Q: How does Suqaba's simulation engine differ from current FEA tools?

TL;DR: Current tools require significant time for meshing and still produce results whose accuracy is unverified. Suqaba explicitly estimates the error between the computed approximate solution and the true solution of the model, and then automatically corrects that error to meet a user-defined accuracy requirement. As a result, Suqaba removes meshing from the workflow, replacing it with the specification of an error tolerance and delivering results that are verified with respect to the requested accuracy (see our first blog post for details on why Suqaba is state of the art).

A: Suqaba stands apart from currently available tools in three fundamental aspects.

1. Error Verification with Certification

Current CAE tools approximate solutions without quantifying how wrong the result is. Suqaba integrates certified error estimation techniques that computes a sharp upper bound of the error being produced by numerical approximations with respect to the true solutions (while the latter being unknown).

The measure of an error allows for assessing the reliability of the result or quantifying risk, thus making its mitigation possible.

2. Error-Driven Adaptive Meshing

For a given model (geometry, materials, boundary conditions, loads, etc.), the computed error quantifies how well the mesh captures the governing physics. Local error indicators are then used as a means to drive refinement and generate an optimal mesh automatically. In brief, verification detects the errors and mesh adaptation corrects them.

Consequently, the mesh is refined only where it is needed. Thus, it achieves an effective balance between simulation accuracy and computational efficiency without manual tuning.

3. Automation

The above detection/correction strategy is the base for our automation. In our framework, the mesh is no longer considered as an input of the simulation workflow. Instead, users define an error tolerance, that is, the value of the error they are willing to accept in the result. Then, the solver will both compute an optimal mesh and solution approximates that comply with the requested error tolerance.