CD-adapco users validate, automate to success

In part 1 of my report on the CD-adapco Global User Conference, I wrote about plans for STAR-CCM+, a new iPad app and my adventures in a STAR-CCM+ hands-on session. There were so many excellent user presentations, that I’d like to give a quick recap of what I learned there as well. Validate, validate, validate It always amazes me that, at CAE conferences, so much is made of comparing analytical results to physical tests. I honestly would have thought, 40 years after the invention of the art form, that we are beyond this point. Why do we need to keep proving the validity of our results? First, because each enterprise needs to assure itself of the credibility of its methodology and tool application — as I learned, products like STAR are so complicated, with so many possible variations in settings, that two users could solve the same problem very differently, possibly leading to different results. Too, many engineers are pushing the boundaries of the codes and need to be certain that the new usages match physical results. Since the real power in CAE lies in simulating what cannot effectively be physically prototyped, so, at some point, we are going to need to be confident enough in our models to simply trust the simulation. Most enterprises are nowhere near this today. There is no standard process It also became clear that CAE fits into each company’s product design process in a unique way. Some companies elect to simulate often, at many point in the design; others do it only at the end. To some extent, this is tied to the correlation question and how confident the engineering team is in its ability to model physical behaviors. But it’s also tied to usability, how long simulations take to converge on a result and the complexity of the system being simulated — more components seems to lead to simulation at the end, if only because that ensures that the latest version of everything is being simulated. There’s definite room for improvement here, for a better a and more efficient use of resource, but the good news is that the use of simulation is growing. CD-adapco Global User Conference attendees are clearly a biased lot, but all indicated that their companies are using more simulation now than ever before, and that the potential exists for even broader usage. Process automation is key to broader usage From Ford Australia, I learned just how complex many companies’ CAE process really is. Thorsten Maertens described how the company revamped its aerodynamic analysis of the 2011 Ranger Global Compact Pickup. The Ranger was designed to meet the needs of 2 brands (Ford Ranger and Mazda BT50), in 3 cabin styles (double, crew, single), 2 ride heights (highway and off-road) and 2 drivelines (4WD and 2WD) — altogether, 24 variants. Mr. Maertens says that the goal of the program was to “chase down every last count of drag to make the vehicle as good as possible.” His background in Formula 1 racing enabled him to create a process that “would enable us to turn around a large amount of data within a short time. We therefore scripted our process in a similar way we did in motorsports to reduce the engineers’ time to work on the geometry, the simulation setup and the post processing as well as to avoid errors in the set up. From our point of view this was the only way to deliver the project in time and on target.” Ford Australia uses a system of macros and Microsoft Excel to automate its process and to enable non-experts to take advantage of CFD. I’ll be writing more about this, since it also came up often at the Altair Hyperworks Technology conference I attended last week, but Ford Australia’s methods team devised a mechanism in which an engineer inputs case details into Excel spreadsheet that then defines how best to set up the simulation. This enables the engineer to evaluate may more possible designs with “minor” amounts of user input and streamlines job submission. The macros read the Excel case data, define properties, run meshing, invoke simulation and defines how the run will be post-processed, automatically creating a report of the relevant data. — But have experts check Mr. Maertens says that it is not Ford Australia’s intention to rely on engineers without CFD experience but rather to enable “someone with no or less experience [to] help us out in peak times to get the simulations out. A person to do this still needs CAD experience to clean up the geometries and get out an STL file and needs to know how to fill out an Excel sheet. The risk to get out an invalid simulation is pretty low as long as the STL file has the correct labeling and the geometry is clean enough so STAR-CCM can handle it. As most of the errors, based on my experience, happen during the setup we minimize the risk of meaningless simulations as much as possible.” He also points out that all results are checked by an aerodynamicist to “make sure we do not give out misleading results”. The Ford Ranger program was a success, but Mr. Maertens’ team isn’t done: “We still see a huge potential to improve the drag coefficient of the new vehicle but, as always in vehicle aerodynamics, might have to compromise with design and the costs that might come with additional aero parts or surface changes. Nevertheless we are convinced that we still have enough options available to release the next generation Ranger again with class leading aerodynamics and unsurpassed fuel efficiency.” Ford Australia’s process automation enable its design teams to (1) run many more cases than would otherwise be possible, (2) with greater consistency across engineers of varying levels of expertise and (3) create consistent output to help evaluate alternatives. All three of these are common desires, no matter what type of simulation or problem, and are key to spreading adoption of CAE to new users. Note: Many thanks to Thorsten Maertens for answering follow-on questions. CD-adapco graciously covered expenses and registration for the event but did not in any way influence the content of this post.