Marc Halpern, a vice president of research at Gartner, tells NAFEMS attendees of major shifts in CAE.
Marc Halpern gave his history in computer-aided engineering (CAE) at the NAFEMS annual North America conference near Seattle. Given that Halpern has been into CAE practically as long as there have been computers, I settled in for a long session.
NAFEMS, the international organization of CAE users and companies, was founded in the UK, but it is in the US that it has found its largest audience. It chose Halpern to deliver its keynote speech.
Halpern is a vice president of research at Gartner, an eminent research group and business think tank that boasts a total of 1,600 research analysts and consultants that monitor a hundred industries.
Not long after graduating from Carnegie Mellon University with a PhD in civil engineering, Halpern was the nineteenth employee at ANSYS. A course in numerical methods inspired him and sent him down the road of engineering analysis. Thirty-some years later, after gigs doing analysis and working with CAE companies, he wound up as an industry analyst at Gartner in the manufacturing segment.
The cloud is central to everything, said the Venn-erable Halpern, and is contributing to the trend toward “freemium” products. (Image courtesy of Gartner.)
Freemium and Democratization
A host of applications are available for free or near free. The most publicized example has been in CAD with Onshape, which lets anyone use its CAD for free, even without a workstation (it runs on a browser and is cloud-based), and OpenFOAM and Simscale have freemium products for CAE.
A growing class of products is appearing and making itself hard to ignore, as users can pay as they go according to their need, use anywhere without regard to the computer or mobile device and have built-in collaboration tools.
In an apparent warning to big CAE, Halpern stressed that democratization is happening and current licensing schemes (meaning perpetual licenses that sell for multiples of thousands of dollars) are in big trouble over the long term.
Componentization
The ability of a group of finite elements to know that they constitute a structural shape, such as an I-beam, rather than behave according to their foundational mathematics, would simplify the input and use of a structural analysis. Similarly, taking a particular type of analysis, for example the thermal analysis of a microprocessor on a board, can be parametrized so that it can be applied to similar configurations and variations, without so much as a care of meshing.
Halpern called this encapsulated behavior. Others call it “appification.” Either way, it spares the engineer from having to be a full-time analyst. The specialized terminology, conventions and peculiarities of simulation can be kept under the hood and the interface is greatly simplified. A complex problem could be solved with a handful of parameters. This would bring simulation to not just engineers, but others who have a stake in the product and its design.
Halpern cited Comet Solutions as a software vendor that has already “appified” a number of simulations to make them usable by more engineers.
Systems Analysis—AKA Model-Based System Modeling (MBSE)
While the design and analysis of parts, even assemblies, is routine for engineers, full-system behavior that can be represented at multiple levels of detail across mechanical parts, electronics and software functions is not modeled routinely. Halpern used the example of an automobile.
The emerging technology allows engineers to model the entire automobile across a range of critical “global” variables that have greatest influence on behavior (e.g. center of gravity, center of force, thrust, braking force, etc.) combined with detailed CAD geometry, CAE models and embedded software logic (e.g.to capture the car’s internal workings, from internal combustion in the engine to the gas through its tailpipes, the stress in the suspension components, safety systems, etc.) to understand vehicle behavior holistically.
A big investment in testing remains essential to understanding how the car reacts on the road—or track. System modeling and analysis, such as that offered by companies like Comet Solutions, Modelon or ESTECO, is not in the mainstream of most engineers’ activities. Not yet.
Knowledge Creation
Getting wiser from sifting through multiple analyses and using analytics to uncover best design practices is something Halpern referred to as knowledge creation. Instead of performing new analyses each time a new design challenge arises and just letting the results reside on individual hard drives unused for eternity, engineers “mine” that data from prior simulations and use analytics to combine the data from those simulations to understand design tradeoffs.
This is being attempted by products such as Clear Vu Analytics by Divis, Phoenix Integration and, in the case of CFD, by TecPlot. Pareto Efficiency, a theory often used to understand economic behavior as well as physical behavior, is being applied in knowledge creation software. BMW is using ClearVu Analytics to study the results of many simulations to glean knowledge.
Multiscale Modeling
The ability to model, in the same software, something as small as molecules to something as big as rocket ships, is being considered by at least a few key providers of software that engineers, designers, and scientists use.
For example, Dassault Systèmes is promoting its molecular modeling software, acquired with Accelrys (now the major part of Dassault Systemes’ BIOVIA business unit) that can model on the molecular level, to Abaqus, its finite element software. See recent coverage of Dassault Systèmes’ “Science in the Age of Experience” coverage. MultiMech for ANSYS also claims to offer “true multiscale analysis.”
Hype Versus Realization
Halpern had so much to say, he couldn’t fit it into the allotted time. No one rose to push him off the stage, so he closed with a dose of reality. Some of the aforementioned technologies are not certain, but early on in what Gartner calls a hype cycle.
For example, multiscale modeling is only now being introduced as a concept. There may be promising research and therefore it would warrant attention and the media may well overplay its promise, but it’s too soon to predict if it will be adopted and in wide use. On the other end of the hype spectrum is system modeling, which has already spawned mature and robust applications that are gaining users.