|INSIGHT: Avoiding the Pitfalls of General Purpose CFD
The power and accessibility of today's general purpose CFD offerings remains both a blessing and a curse. When used appropriately with good methodology and proactive validation, it can inspire confidence. When used incorrectly, it can wreak havoc. In this month's issue of The Flow, we sit down with Bob Ni to discuss some of the common pitfalls associated with general purpose CFD and how to avoid them. Bob is the Chairman and CTO of ADS. Prior to founding the company, Bob spent 28 years at Pratt & Whitney leading turbomachinery CFD efforts.
FLOW: What's so challenging about general purpose CFD?
BOB: General purpose CFD has the potential to become very challenging because so much analysis power is being put in the hands of less aerodynamically-savvy users. Let me provide some context. Back when I started in the 1970s, industrial turbomachinery CFD was limited to a relatively small group of companies with the financial, development and computing resources to harness it for commercial product design. CFD was a specialized capability targeted for trained aerodynamicists and developed with very specific applications in mind. Against this backdrop, companies like Pratt and GE were able to invest meaningfully in in-house CFD, rigorously validate that CFD against rig data, and successfuly institutionalize a design methodology around it over the course of years if not decades.
The world today is vastly different. The luxury of long term investment in IR&D has given way to cost cuts, billable hourse and outsourcing. General purpose commerical codes have proliferated in this setting and become extremely feature rich and accessible along the way. This has fueled a "democratization" of CFD from specialists to the mainstream designer. This bodes well for the industry long-term, but there are lots of potholes to avoid along the way. Lots of CFD features mixed with less aerodynamically-savvy users is a dangerous cocktail that can create all sorts of opportunities for misuse if not handled with care.
FLOW: Give us an example of a common misuse.
BOB: One common pitfall for smaller companies and nascent users is a tendency to accept CFD predictions at face value. The push button nature of general purpose CFD makes it all too easy to produce impressive 3D visualizations and animations that can create a false sense of confidence in the results.
Beware of the pretty picture! As we all know, CFD is an inexact science, and turbomachinery tends to be one of the more exacting applications of CFD, so it's vital to maintain a healthy skepticism of CFD results. Users must be vigilant about looking beyond the flash and to the substance to ensure the underlying predictions make physical sense, otherwise they may be in for a shock when the rig data becomes available.
FLOW: How can a company assure itself of the integrity of the CFD if it doesn't have rig data or a resident aerodynamicist?
BOB: First, when evaluating a commercial code, be sure to verify that it's been deployed successfully in your application space—the more references and vendor expertise in the domain the better. Second, if a company can't draw on internal turbomachinery aero expertise, they can look to outside consultants for mentorship and counsel. Here at ADS, for example, we've been able to assist our clients in this manner since we are quite familiar with our CFD and its application to turbomachinery aero design.
FLOW: Makes sense. What's another pothole to avoid?
BOB: Another common pitfall is to fall into the "more is better" trap. Here again, the myriad of options available to the user of general purpose CFD makes it all too tempting to add more completeness (and complexity) to an analysis.
Though this make sense conceptually, it also assumes there is plenty of time and computational resource on hand. It also assumes, of course, that the additional complexity will deliver the insights needed.
We'd advocate deferring complexity until it is needed. Avoid turning aero design into a research project by resisting added complexity until it is absolutely needed to resolve a design problem. For early design work, a "less is more" approach is likely to be far more effective in terms of turnaround time and general trend identification. For later stage design work, introduce complexity if it helps to answer a specific design question—for example, running time accurate simulation to identify adverse blade-row interaction effects impacting durability.
FLOW: It seems like the range of configuration and analysis options in general purpose CFD can get you into trouble.
BOB: Yes, without training it can be problematic. In fact, this manifests itself in yet another way—a tendency to "data match" in the name of predictive accuracy. Again, the vast number of configuration options available in general purpose codes makes it very tempting for the nascent user to want to turn knobs and pull levers to get CFD predictions to precisely match rig data.
While it's nice to show that rig data can be precisely matched, does this prove the CFD has good predictive accuracy and can be counted on for design? We think not, at least not until the configuration settings are shown to work equally as well on other cases of import to the business.
This brings us to our next piece of advice, which is to focus more on accurate trend prediction across a body of cases rather than on absolute case accuracy. Don't waste time trying to force match the data with exotic configuration options; focus more on picking the options you understand and that can consistently predict trends across the range of cases you're interested in.
FLOW: How can you build trust in your CFD if the body of work employs vastly different configuration options?
BOB: Not easily. Inconsistent application is the fourth and final challenge of using general purpose CFD. Without a method to the CFD analysis madness, very little organizational learning and trust in CFD can be gained.
Our advice here is to standardize the analysis process. Building confidence in CFD is a journey, not an event, so it's vital to be consistent in the application of CFD to leverage learnings across a body of cases, not cases in isolation. The best way to do this is to standardize on a common analysis process and a baseline set of CFD configuration options that work, validate and iterate.
FLOW: Sounds good, any parting thoughts?
BOB: To avoid the pitfalls of general purpose CFD, I'd recommend the following. First, to assess the integrity of your CFD code, focus on accurate trend prediction over a body of work rather than absolute accuracy for a specific case—it'll avoid a lot of wasted time. Second, keep CFD analysis simple. Resist temptations to introduce complexity until you're at the appropriate stage in the design cycle and only if it sheds light on your specific design question. Third, maintain a healthy skepticism about your CFD predictions--make sure they make physical sesne and be vigilant about experimental validation. Finally, have a method to your CFD analysis madness. Conduct CFD analysis in a consistent fashion to ensure that learnings about your CFD can be gained over a body of work and time.
FLOW: Thanks Bob.
BOB: My pleasure.
CASE STUDY: Conjugate Heat Transfer Assessment of a 3-D Vane with Film Cooling and Comparison to Experiments The High Impact Technologies (HIT) Research Turbine Vane at the U.S. Air Force Research Laboratory is a three-dimensional fully film-cooled modern turbine inlet vane. In this paper, Captain Jamie Johnson and Paul King from the Air Force Insitute of Technology conduct conjugate heat transfer analysis on the vane using Code Leo and compare the predictions to experiments. <more>
TECHTIPS: Creating Inlet Profiles for Unstructured Meshes When converting externally generated unstructured meshes to the ADS restart file format using ADS-GEO, a radially varying inlet profile can be applied using a format similar to that used in the aero geometry file. This tech tip details how to use this feature. <more>
Welcome to The Flow
Welcome to The Flow, a newsletter for monthly insights on turbomachinery CFD published by AeroDynamic Solutions, Inc.
Each month we'll spotlight a topic of interest, discuss a case study and/or provide useful pointers about how to get the most out of the ADS CFD system.
You are receiving this email because you or someone else you know thought you would be interested. To unsubscribe please click here. We value your privacy.