The engineering community is a tough cookie. And, if anyone tells you otherwise, don’t believe it. The hard truth is that engineers like to pose questions. And, rightfully so: this is how they find solutions to the most sophisticated problems. Without this mindset, there won’t be a breakthrough in industrial development, disruptive changes in the existing processes, or simply better solutions. I am an engineer myself. And I sincerely understand the necessity of these questions. So, today I would like to answer one of them.
Can our software, ReveaL, compete with existing commercial solutions? We heard many times that our solution solves a lot of engineering pain for cases where the available model is non-watertight or exists only in a form of a computed tomography recording. But, can ReveaL deliver the same results for a standard use case while keeping up with the performance? Well, let’s give it a go. Let’s challenge ReveaL in accuracy, time and memory.
In a frame of a pilot project with Siemens, we were confronted with the following challenge. We were offered to take a simple use case in industrial 3D printing: a sandwich structure with lattice between two plates. When we increase the size of this structure, a conventional FEM solver starts being more expensive in terms of meshing and solving times.
Our partner at Siemens simulated these lattices under torque using their conventional commercial software while we provided them with the results of ReveaL. We agreed on comparing four main quantities: average stress arising in one diagonal strut, average stress in the whole lattice, simulation time and peak memory. Similar workstations, same setup – same challenge.
“Eh?! How is it possible?” – would you ask, – “The software is so young!” Yes, we are a young startup, but our technology is mature.
Let’s look at the numbers. For a medium size of the lattice Reveal showed the following results:
|Case 3, quadratic elements||Commercial software||ReveaL|
|Degrees of freedom||4.9 Million||4.84 Million|
|Wall clock time (run time)||532 s||583 s|
|Maximum memory||41.8 Gb||64.8 Gb|
|Corner distortion||86 µm||92 µm|
|Average stress in a diagonal strut||107 MPa||109.5 MPa|
|Average stress in the lattice||141 MPa||144.9 MPa|
Looks promising? Yes! And this is just the tip of the iceberg, you can read more details about this use case in our whitepaper:
Of course, we have still a lot to improve, a lot of features to build. But we can say already now that ReveaL is a simulation tool which is fast, reliable and accessible. The outstanding advantage of using ReveaL to perform simulation is that it works for any type of geometrical models: non-watertight CAD, point clouds or CT scans. It is agnostic to the input geometry and its quality. This allows to reduce computational costs, remove manual effort and explore designs that were never feasible for simulations before.
We at DirectFEM do not provide yet another STL repair tool, we offer a standalone software, ReveaL, that allows to simulate directly on the provided geometry removing the step of geometry reconstruction and healing. And, as you see, for a standard case, ReveaL performs at least as good as commercial tools or even better.
Do you want to try?