Integrating CAE Into Every Stage of the Product Development Process

PNY Technologies has submitted this post.

Designers need the ability to explore and predict how products will work—or won’t work—in the real world. Computer Aided Engineering (CAE) provides this capability. Engineers need rapid design validation and optimization feedback while developing products in a fiercely competitive landscape.

In most CAD/CAE workflows testing how a design will work usually comes late in the design process resulting in extra cost and rework to resolve any issues found. Engineers often design products without physically accurate simulation in real-time. Engineers utilize CAE specialists to simulate and test product designs which slows project progress.

If CAE simulation late in the design process identifies a failure or problem designers must cover previous ground to remedy them; in extreme instances they may have to start over. This results in cost and schedule issues and increases time-to-market.

CAE software that allows in-silico real-time simulation throughout the design process offers decisive advantages. This would not be possible without the compute and graphics power delivered by GPUs, artificial intelligence (AI), and advanced interactive CAE software intended for use by designers and engineers upfront in the design cycle.

For example, Ansys Discovery combines interactive modeling and multiple simulation capabilities in a revolutionary GPU based solver to enable real-time simulation. The Ansys solution using NVIDIA RTX professional GPUs takes design and engineering creativity to new levels by integrating real-time simulation and design validation into every stage of the design process.

Hardware and Software Required to Simulate CAE Workflows

CAE hardware includes CPUs, GPUs, GPU memory, and networking sufficient to meet the needs of in-silico CAE simulation including:

• Access to massively parallel processing power for computationally intensive tasks
• Networking infrastructure (for remote work and collaborative design reviews)
• High-capacity and performance storage, with RAID 1 for critical design data
• Adequate GPU memory to contain model geometry, textures, and lighting effects, run real-time physically-based simulations, possibly with an AI assist
• Ability for team members to remotely and collaboratively work on the same design, while retaining a ‘single source of truth’ master file

GPUs are required for design simulation because, unlike CPUs, they can handle thousands of parallel processing tasks, support interactive graphics, and deliver AI functionality.

Live Simulation

Ansys Discovery software, which is NVIDIA CUDA-based, empowers designers to run live simulations on NVIDIA GPUs thus making it easier to test products early in the product development process. It allows designers to change physical parameters, make geometry edits or even change materials to show how they can improve a candidate design and test for failures.

The below image shows how Ansys Discovery is used for transient inverter simulation by changing heat values to analyze how temperature changes affect the product.

The Right Graphics Hardware for CAE Software

CAE design software requires CPUs, GPUs, and memory with the proper configurations. An NVIDIA certified hardware distributor such as PNY can help recommend appropriate GPUs, discuss bare-metal versus virtualized deployment, and consult on hardware configurations.

Ansys Discovery provides access to both GPU-accelerated physics simulation and high-fidelity CPU-based simulation. Ansys Discovery uses NVIDIA CUDA and requires a discrete NVIDIA GPU to run Live Simulation.

Running Discovery in Explore mode provides real-time simulation and runs entirely on an NVIDIA RTX GPU. Ansys Discovery uses GPU acceleration for all aspects of Ansys Live Simulation in the Explore stage, including discretization, solving, and post-processing. Refining results through post processing is also GPU accelerated to allow for interactive investigation and manipulation. It is important to have enough GPU memory to allow for interactive investigation and manipulation. Ansys recommends using NVIDIA professional GPUs with at least 8GB of dedicated GPU memory.

It is important to not use consumer or gaming GPUs because Ansys does not test them under CAE processing loads. Ansys recommends using NVIDIA RTX GPUs based on NVIDIA’s Ampere architecture when running simulations with Ansys CAE software.

Ansys CAE software can run across a wide range of professional NVIDIA GPUs including the NVIDIA RTX A6000, A5500, A4500, A4000, and the A2000 6GB or 12GB GPUs. Users can select the NVIDIA RTX product that best fits their needs and budget. Ansys Discovery automatically boosts simulation fidelity based on available GPU memory, so more is better.

The NVIDIA Quadro GV100 board can be used for simulations requiring FP64 double-precision floating point calculations in hardware, dramatically outperforming FP32 boards. NVIDIA has kept the GV100 in the product line for precisely this reason.

Essential Business CAE Software and Workstation Metrics

Organizations should consider the following parameters when evaluating CAE software and workstation requirements:

• Cost of ownership – Does it provide strong ROI and predictable future costs?
• Flexibility – Does it support CAE compute, graphics, and AI-intensive workloads?
• Ecosystem integration – Does it include standards-based equipment and support leading CAE software that fully exploits the capabilities of GPUs?
• Scalability – Is it easy to upgrade and easily scalable with the ability to integrate complex CAE workflows?
• In-silico simulation – Does the CAE software allow in-silico simulations so designs and testing can be done in near real-time?
• Productivity – Does it offer fluid, interactive performance that enables creativity and allows additional or new design possibilities to be fully explored?
• Creativity – Does the system have the performance to ‘get out of the way’ of the creative process to allow users to design or simulate different approaches in-silico before physical prototypes are produced?
• ROI – Can it support GPU workload bandwidth requirements but also scale out in a tiered storage system?

Currently CAE design and testing is done late in the engineering process which can cause cost overruns, schedule slippage or even product redesign. Modern CAE tools require processing performance and speed that only GPUs can deliver. Sophisticated CAE tools using the latest graphics hardware allow users to make simulation changes in near real-time which saves engineering time and prototyping costs, and can help reduce product time-to-market while improving product quality and innovation.