In this lecture, we will cover the following:

• We will compare the custom shell solver with the OpenSeesPy benchmark model,
• We will check whether the deflected shapes agree visually and compare displacement magnitudes from both analysis approaches,
• You will identify a hidden robustness issue that appears after small parameter changes.

We begin by placing the custom solver results and the OpenSeesPy results side by side. Visually, the two models show very close agreement, and the displacement magnitudes are also almost identical. This is a strong validation signal because the two analyses have been built through different routes, so agreement suggests that the custom Reissner-Mindlin shell implementation is broadly working as intended.

We then introduce an important caveat. When a small change is made to the input parameters, the custom solver can suddenly produce unrealistic results, while the OpenSeesPy benchmark remains well behaved. This shows that the custom implementation is not yet robust under all conditions, even though it can agree very closely with the benchmark in the initial case.

Next up

In the next section, we will begin diagnosing this hidden issue by studying drilling degrees of freedom and stiffness matrix singularity.