We take for granted that new products, even simple ones, are designed on a computer first. This has been true in nearly every engineering-centric industry for years. But today, these designs aren’t merely models, or flat schematics. With software from companies like ANSYS, these models are sent to inhabit stunningly complete simulations of the real world, where they are virtually subjected to, well, virtually every physically force known to man.
The possibilities are nearly endless: cars can be sent to a digital wind tunnel to see how they perform at high speeds; buildings can be outfitted with a new air circulation system to see how well they can be ventilated; garments can be subjected to stretching, friction or liquid before the first stitch is sewn; laptops can be dropped from their digital bags, onto a modeled hardwood floor; future military aircraft can sustain fire from nonexistent enemies; cellphones can be tested for heat dissipation, electromagnetic interference, or water resistance. For a list of the types of companies that use simulation software like this, check here.
I recently spoke with Ahmad Haidari of ANSYS about this eerily comprehensive virtual proving ground--including its uses, limits, past and future.
What is ANSYS, and more broadly, what is simulation-driven design?
What our software programs do is allow you to answer design and engineering questions with a computer. Effectively, a concept is designed and digitized, and then tested. You’ll learn something about its performance in this test, which will direct your thinking, and where you should go to modify the product, or pass it forward to the next level.
What our company does is it allow you to do all this in a computer space.
So it’s a way to do early product testing without having to build a product?
Precisely. Or look at potential concepts you might be considering.
The scope is massive--ANSYS software can be used for something as modest as a garment or a pair of shoes, all the way up to an airliner. How does your software deal with such disparate things?
Basically, whatever you design and are looking at, you’re interested in knowing how it responds to some forces generated--either electrical, mechanical, or in fluid dynamics. There may be flow, or stresses, or pressures, or electronic response. The commonality in these applications is to look at applied forces and the responses of your design.
I guess what I mean is, people expect this kind of simulation to be carried out with a car or an airplane--engineers design these things on computers, so subjecting them to tests in extensively detailed simulations is an obvious next step. But how does something like that apply to a smaller consumer product like, say, a laptop?
Well, let’s take that even further, and talk about children’s diapers. With diapers, you want to know how much a material you use puffs up as it gets wet. And then you want to be able to look at the absorption rate, and the distribution of liquid. With that example, you’re looking at wicking.
Or, to look at another example from P&G (Procter and Gamble), their potato chips are actually designed to have a [specific] airflow, for drying during manufacturing. They wanted to look at the airflow of the potato chip to make it aerodynamic.
Or designing inhalers, companies need to see how heavy or light the particles need to be in order to go as deeply into the lungs as possible.
In laptops, the simple answer would be cooling in components, or power consumption, or electromagnetic interference, or signal integrity, or even crash tests--when you drop you laptop or cellphone, what impact would that have on the device. You could also look at how you might electrochemically finish a product, or electroplate it, or solder it. It’s everything from design to packaging to usage analysis.
So, what can’t you model?
There are sometimes very complex questions about interdependencies, which can make problems very computationally difficult. These would take a long time to run, so we might not be able to do them.
In addition, if we know the physical properties of a certain materials, we can calculate their behavior. But there are some material properties that have not been fully characterized. For example, there are properties of tissues and arteries; if we wanted to study an artery’s expansion due to some stresses of blood pumping, it would be difficult.
But those limits don’t sound structural. There’s lack of data about certain materials, or a lack of computational power and/or payoff for spending computer time on something. To take the question further, are there any theoretical limitations when it comes to using software like ANSYS?
When we get to molecular response, and atomic response, our solutions aren’t really designed for that. The space we operate in is in an engineering scale from sub-micron to kilometers. The real answer to your question is that the continuum has to hold.
So, within the limitations of Newtonian physics, you cover pretty much everything.
Heh, well, as far as the ability of the software to model it, yes. But sometimes the physics are missing, and certain details are missing, and materials might not be fully understood, and so on. Looking at a boiler in a nuclear reactor, for example, sometimes people won’t quite know the models that close the equations for boiling applications. If there’s no physical understanding of a situation, there’s no way to model it.
What are some of the most ambitous projects our software has been used for?
Most of our clients use the software for components, but other use it for complex systems. About two years ago, we introduced a way to model a car with a coupling of our electromechanical solutions and our fluid solutions. This interdisciplinary approach was able to be applied to the Chevy Volt by GM, who was already a customer of ours. [Ed. note: for more on that, click here.]
That’s where simulation software is now. But what about the early days? ANSYS has been around for about 40 years, right?
ANSYS came from a need where certain engineers thought that simulation could do more than it was being used for, and decided to make it into a commercial entity. [Founder] John Swanson started with simulation tools for nuclear components, so ANSYS grew from mechanical modeling. Elsewhere, in about 1983 people started doing fluid mechanical analysis, then electromechanical and electronic simulations. Eventually, all of these business solutions have been acquired by or merged with to create ANSYS’ current solutions.
I like to tell people, 15 years ago we were barely doing 2D, cartoon-like fluid simulations. Now we model entire jet engines.
Image credit: ANSYS