Web Exclusive Articles
« back to listing
Case study: CFD at ECR Engines
The powertrain division of Richard Childress Racing is using an innovative CFD meshing tool to improve results in NASCAR and IMSA sportscar racing
Above: RCR driver Ty Dillon won at Indianapolis in 2014
Brian Kurn is in charge of valvetrain development and all virtual prototyping technologies, including engine and valvetrain simulation and computational fliud dynamics (CFD), at ECR Engines, a division of Richard Childress Racing (RCR). In a career spanning 25 years he has worked for some of the biggest names in stock car racing, including Roush and Hendrick, and with Toyota team Bill Davis Racing.
Kurn started his craft building and improving V8 engines for the small dirt tracks and worked his way up to the elite series in NASCAR. He was highly regarded for his work on cylinder heads and his ability to extract more power whilst retaining crucial reliability. Today, he’s a highly respected engine developer who has worked successfully across many racing formulae – IMSA, NHRA Pro Stock, touring cars in Brazil and Argentina, truck-pulling, supercross and, of course, NASCAR.
“In the good old days, so-called tuners determined the biggest valves that could be used and then they simply began to hand-port the head, believing that the more air that would flow, the more power it should make,” says Kurn.
“After spending a lot of time doing this, you took the parts to the dyno and only then did you find out if you had found a solution or just scrapped another cylinder head! It was an expensive and time-consuming way to see if your idea gained a few horsepower or not. And each time you got a new head design, you really had to start again!” It was this inefficiency which drove the forward-thinking Kurn to investigate simulation technologies.
As an experienced CFD user, primarily to analyze internal flows in the engine, both standalone and coupled with engine simulation, Kurn is at home with state-of-the-art technology. But in those early days, just over a decade ago CFD posed problems. “The run times to do the simulations took too long and when we had to create our own mesh, we really suffered with the variability between users,” claims Kurn. “It can affect your results, introducing inconsistency, and ultimately your trust in the data can go out of the window.”
Even today, creating a good mesh is crucial to resolve the flow. But the quest for an automatically generated mesh never quite delivered the accuracy needed to move away from user-generated data. For years, engineers simply accepted the challenges and did what little they could to minimize the variations.
For Kurn, it was a frustration. “I never believed that an effective automatic meshing tool would happen in my lifetime. I thought we would be stuck with the longer run times forever,” he explains. Then, through engine simulation provider Gamma Technologies, Kurn first encountered an innovative automatic meshing solution called Converge CFD Software. Developed by Madison, Wisconsin, firm Convergent Science, it automates the meshing at run time with a perfectly orthogonal Cartesian mesh that eliminates the need for a user-defined mesh.
“To be honest, I’d heard it all before and I was sceptical,” says Kurn. “Automatic meshing had been around for long time but none of the solutions I tried lived up to their claims of a completely automatic mesh that produced accurate results. I longed for the day when I wouldn’t’t have to predict the outcome in order to define the mesh, and I didn’t want my meshing to affect the result. But I ended up taking a look and the rest is history.”
Above: Converge automates the mesh at run time and automatically refines when and where it is needed through Adaptive Mesh Refinement (AMR)
Written by engine simulation experts to address what they perceive as the deficiencies of other CFD codes, Converge offers run-time grid generation and refinement so users such as Kurn no longer need to spend their time creating meshes.
Instead, the user supplies a triangulated surface and a series of guidelines from which the Converge proprietary code creates the grid at run time. “They had hit on my objective – reduce the run time whilst retaining the accuracy of the simulation, removing assumptions,” says Kurn. “Testing would become more fruitful as with consistent meshing and we could test more solutions in the same timeframe. Once you have the model you can keep re-running the job without recreating the case.”
For race teams, the software achieves the one thing that is hard to buy – more time – and Kurn has been astounded with the amount he’s saved. “We gained literally weeks on some developments in 2014,” he claims. “On our Daytona prototype engine we got ahead of the development schedule and we were able to start testing different trumpet lengths before the engine was even ready to run on the dyno, or got anywhere near the car. This saved not only time but also the number of prototype parts produced. Knowing the exact parameters of key items such as combustion chamber, intake and exhaust ports means when we make changes, we can accurately measure just those changes and have complete control over them. We now run a number of simulations and with the accurate data generated can confidently pick the best one or two to try on the car.”
Left: With specific expertise in liquid spray, combustion, flame propagation and emissions formation, Convergent Science produced a tool suited to engine simulation
Trumpet design is just one of the areas that Kurn is trusting to Converge. Others include the very challenging modeling of combustion and he sees the potential for using Converge for optimizing future fuel efficiency.
Rob Kaczmarek, marketing director at Convergent Science, explains: “Our genetic algorithim optimization can run cases depending on design parameters such as fuel efficiency or power and think outside the box,” he says.
“Our creators came from engine simulation and struggled with CFD meshing in the early years. They focused on creating a tool that would simplify meshing and increase accuracy. To achieve this they allowed the program to automate the mesh at run-time and automatically refine when and where it is needed through adaptive mesh refinement (AMR).”
A common area of interest where this approach works particularly well is in-cylinder flame propagation. Kaczmarek believes non-Converge users really struggle with hard-to-define areas like this.
“It leads them to either go to a larger-sized mesh, maybe up to 1mm in order to save time, but they lose accuracy, or they go to a smaller mesh increasing the accuracy but also increasing run times,” claims Kaczmarek.
Converge can take care of this, allowing the program to refine the mesh when and where it is needed at run time for more accuracy, while keeping run times manageable. In addition, the software also comes equipped with detailed chemistry and physical models to help engineers make the gains they want.
Right: CFD helps to develop an insight into flow patterns from liquids or gases that are difficult or expensive to study using traditional techniques
“For example, measuring turbulence of a flame in microseconds and how it changes is very hard to do but crucial for efficiency,” adds Kaczmarek. “Converge can help. It’s great for transients and we saw, for example, with the use of direct injection in Daytona prototypes and other high pressure scenarios, that it is very effective. Even though NASCAR engines have been around for a long time, well over 40 years, some of the best tuners who think they have understood them can now really see and truly understand what is actually happening for the first time.”
Doing that work on track rather than in the garage is another area where fast and accurate simulation is helping Kurn and ECR. He believes that a NASCAR pushrod engine is probably one of the worst case scenarios for different behaviors when fired up. “The actual exhaust valve opening can be delayed as much as 10-15 crank degrees as a result of the cylinder pressure acting on the valves,” he explains. “Using trusted valvetrain simulation, we can now optimize the design of the valvetrain components to work in the real world, even in a NASCAR application.”
Virtual testing is currently unrestricted in NASCAR and is becoming more and more popular as teams look for the edge over the others. Converge was used effectively to make significant gains in power in 2014. These leaps in performance helped RCR secure second overall in the Sprint Cup Series standings with Ryan Newman. Over 23 top 10 finishes resulted in Brian Scott snatching 4th in the second-tier Xfinity championship with Ty Dillon gaining a rookie victory at Indianapolis and 5th overall. Team mate Brendan Gaughan showed the versatility of the team’s development, winning on the traditional Road America circuit.
Above: Brendan Gaughan takes the checkered flag at Road America
ECR’s engines proved crucial in IMSA’s Tudor Sportscar Series, too, helping Chevrolet to win the Tudor and Endurance Cup titles. Action Express’s DP car with an ECR engine scored eight podiums including three victories, most notably the Daytona 24 Hours. The ECR engine had reliability too, with the car completing every lap of competition throughout the season.
“It was a special year for Kurn and the ECR team,” adds Kaczmarek. “We are so proud to have been involved with the team and Brian.”