If you’ve heard anything about ANSYS 17.0, it’s that it is faster than ever. Faster solvers, faster processing, greater core counts — it all sounds great, doesn’t it? Everyone wants to get their work done faster, and faster is better than slower, isn’t it? But what exactly does “faster” mean to engineers performing structural analysis simulations today?
In this example from Herbert Güttler, MicroConsult Engineering GmbH, Bernstadt, Germany. he’s comparing some runs from big models, I’m not just talking millions of degrees of freedom, but billions of degrees of freedom.
In 2014, we solved this complex turbine blade model with one billion degrees of freedom using the PCG solver in ANSYS 15.0 on a system of 64 cores. A loose solver tolerance was set because of the size of the model. It took 15 hours to write the database and 13 hours to solve the model. In 2015, we repeated the test using an early version of ANSYS 17.0. This time we created a two billion degree model based on the same geometry — a size that exceeds commonly used models by 1-2 orders of magnitude — and even tightened the tolerances. Creating the database took just four hours, while solving the model took 12, this time using 88 cores. By upgrading the hardware and ANSYS software, we were able to generate and solve a model that is twice the size in half the time, at a greater level of accuracy.
“That’s great,” you’re thinking. “Faster solutions mean my job gets done quicker.” But that’s only part of the story.
Faster also means that as computer technology advances, we improve our software as well. With the release of the Intel’s Haswell architecture, Intel has significantly increased the performance of their chips for solver related operations. Additionally, the availability of easily affordable solid state drives (SSDs) has removed one of the bottlenecks routinely seen historically with Mechanical HPC, so that the whole system can deliver sustainable HPC performance. ANSYS solvers leverage these advances, taking greater advantage of more advanced processor technologies and high performance computing (HPC) architectures. This in itself allows faster insight.
We’re conducting a free webinar on the 3rd of March where you can learn more about how faster solve times in ANSYS Mechanical using HPC can help you achieve results quicker and run more models to gain even faster insight into your designs. Make sure you REGISTER.
Now, we’re not just changing how quickly the solvers complete their tasks, but we’re changing the way they solve the problems themselves.
Take, for example, the analysis of this excavator as shown in this animation. ANSYS Mechanical 17.0 allows rigid bodies and bodies modeled using component mode synthesis (CMS) to be combined into one analysis. You can expand the results from the CMS part to 3-D results easily and quickly.
Solver technologies like Rigid Body Dynamics (RBD) and Component Mode Synthesis (CMS) enable you to model this complex hardware, previously an intensely time-consuming task, in next to no time at all. Other advances, such as parallelization of meshing and automatic contact detection reduce the times a CPU sits idle and boosts processing speed by multiple orders of magnitude.
Finally, faster means fewer assumptions and better solutions. Increased core counts let you solve larger, more complex models. And because it takes less time to run a structural analysis, you have time to perform more simulations, analyze a greater number of different caseloads and test more designs, thereby gaining greater insights into your product designs.
So what does “faster” mean to engineers performing simulation today? It means more opportunities to design and manufacture great products with confidence. And maybe take a long lunch once in awhile.