Racing Developments: Actually, It Is Rocket Science

Whether it’s winning Le Mans, or helping to develop U.S. Navy missiles, engineers from Pratt & Miller Engineering are involved.

Whether it’s winning Le Mans, or helping to develop U.S. Navy missiles, engineers from Pratt & Miller Engineering are involved.

What do these vehicles have in common: GM’s factory GT1 Corvettes and Cadillac CTS-Vs, Pontiac’s GTO-R and its replacement GXP-R, and missiles fired from Navy submarines? Answer: Pratt & Miller Engineering and Fabrication (New Hudson, MI; www.prattmiller.com). More than just a collection of old hands fabricating race cars, Pratt & Miller has expertise in simulation, computational fluid dynamics (CFD), relational databases, modeling, and software creation and development. These guys are more than racers, they’re engineers.

“We do quite a bit of software and application development to support our race and engineering activities,” says Lynn Bishop (who, frustratingly, like most Pratt & Miller employees, has many responsibilities, but no title). “It’s a large part of what we do here.” Not to mention that it’s also the primary focus of the company’s Aletheon Div. in Mooresville, NC, where 45 engineers are engaged in the development of proprietary CFD software (sold under the name “Raven CFD”) licensed from the U.S. Navy. (The software was created to model submarine-launched missiles through all phases of flight.) “About five years ago, we wanted to get an in-house CFD capability, and were fortunate to find some PhD aerodynamicists that were developing CFD code and Linux clusters to efficiently run the CFD solvers,” says Bishop. In time, the Navy decided to exit the software development business, and gave Aletheon the rights to continue the CFD solver’s development. According to Bishop, a vehicle dynamics guy by trade, it’s the closest things he’s ever seen to rocket science. “Aletheon has a 460-processor Linux cluster with 460 Gigabytes of RAM that allows its engineers to model 75 million elements, which means they can do an entire vehicle in a single model. They also have the ability to take a geometric shape and dynamically move it during the CFD solver’s run.” In most CFD programs, analyzing the aerodynamic effects of different vehicle ride heights or wheel attitudes means it’s necessary to regenerate the surface, re-grid it, re-run it, and repeat that process for every position. “Our aerodynamicists can see the interactions and changes in real time,” claims Bishop. 

Embedded in the software are unique post-processing tools that make each surface of the vehicle discrete so it can assess the contribution of each part of the bodywork to the total downforce of the vehicle. “We can determine what changes will give the greatest effect, and how this will affect its drag, side-force moment, and yaw moment,” says Bishop. This information is fed into an aerodynamic optimization program to produce the best configuration for a specific track. It is an increasingly important piece of the development puzzle as race sanctioning bodies limit the aerodynamic changes that can be made to a vehicle during a season.

Dynamics analysis extends to the vehicle’s tires as well. With more series running common, or “spec.” tires, provided by a single tire company, getting the vehicle’s dynamic balance in sync with those tires gives that team an edge over its competitors. In the past, it was common to take the information provided by the tire companies and design the suspension, weight distribution, and related systems around those numbers. Only the procedures used by the various tire companies—which encompasses tires from Michelin, Hoosier, Goodyear, Pirelli, or Toyo—often were developed from static tests, and produced values that didn’t correlate. “From company to company the tire data wasn’t the same, so we decided to develop our own tire test methods and models for all of our programs and test them at Calspan,” says Bishop. (Calspan—Cheektowaga, NY, www.calspan.com—has five operating units: Crash Data Research Center, Flight Research, Systems Engineering, Transonic Wind Tunnel, and Transportation Science Center. It is famous for its work in aircraft and automotive research.) “Because we use our own tire models for simulation, we are in a better position than just about anybody to capitalize on a change to the tire construction or compound because we can take that tire and test it.” The proof of that statement came when the Grand Am series specified a tire that produced 15% greater grip at the front than at the rear, a detriment for a rear-drive car like the Pontiac GTO-R. Pratt & Miller engineers discovered this change during testing at Calspan, and ported this information to the team’s vehicle simulation environment to create a setup that compensated. According to Bishop, this ADAMS-based dynamic simulation capability also can look at the dynamic spring rate of the tire, and correlate that with the damper curves and vehicle spring rate. (ADAMS is developed by MSC Software (www.mscsoftware.com; Santa Ana, CA). Not surprisingly, Pratt & Miller has the ability to analytically predict a damper curve before the damper is ever built, and can choose components from a list of attributes to meet a certain need. This same program recently was used to develop a semi-active damper for a Tier 1 supplier. “Everything happens at the tire contact patch,” he says, “and we have data available on the cornering stiffness, damping and spring rate of the tire, how that tire generates heat, and how the suspension design and setup can influence these things.”

As you might expect given the above, the team has created its own parametric vehicle model, or lap time, simulator. Unlike commercially available systems, the Pratt & Miller unit claims to be fully dynamic, rather than a quasi-static model that solves for each instance around a track. This allows the unit to include inertial effects, damping and aerodynamic effects, etc., and gives engineers all of the forces leading up to and through a particular point on the track. “We have quite a few production teams at GM using it to understand the influence of weight distribution, front and rear tire cornering stiffness, center of gravity height, roll couple distribution, front and rear track widths, tires design, and more,” claims Bishop.

Once the parametric model is complete, more detail is gathered through the use of multi-body models based on the ADAMS motorsport toolset with hundreds of degrees of freedom. “Obviously, we have our own customization that we keep for ourselves,” says Bishop with a smile, “but all the good bits are available to the GM teams we support.” (Pratt & Miller have a joint licensing agreement with MSC Software to develop this software.)

Lest you think the folks at Pratt & Miller are analysis junkies with little use for the subjective experience that, say, a ride engineer builds up over the years on what constitutes the proper feel for a particular brand, think again. “We try to download the subjective experience and tie a metric to it,” says Bishop. “That means we can determine, within this subjectivity, what falls outside of the line without excluding subjective or human elements, or excluding traditional test methods.” A similar process is followed with the team’s drivers. At the end of each test session, driving stint, or race, the drivers are expected to fill out a standardized driver survey form on a tablet computer, and rank various aspects of vehicle behavior on a scale from 1 to 5. “The form uses standard terminology for all of the vehicle dynamics metrics,” explains Bishop. “Corner by corner they are asked to rate the car consistently in terms of understeer, oversteer, stability, ride characteristics, braking, etc., and all this information is loaded in to the database so we can do additional physical analysis and correlate the subjective and objective analysis to the model.” That database, it should be noted, also includes all of the lap time information, weather conditions, vehicle data acquisition readings, and myriad other variables. “If you want to know the temperature of the left front brake disc in turn seven of the fifth lap at Le Mans, we’ve got it,” says Bishop proudly.

Using this information effectively comes down to what Bishop refers to as the “target cascade process.” In it, the team starts with parametric attributes—a lap time target, a lateral acceleration target, a top speed target—and determines the system that lets you achieve that full-vehicle target. “So you cascade down to the chassis and the suspension, the aerodynamics, the powertrain, etc. and set targets for each,” explains Bishop. “Then you break those down to the point of saying, ‘The suspension has a control arm, bushing, tire, damper, and spring. Where do those elements have to be so you can meet the vehicle target?’” Not only can this tell you what effect a particular control arm position will have on lap times at Le Mans, it creates what Bishop terms “an algebraic roadmap” of how it affects overall vehicle performance. From this, it is a—relatively—simple step to reading the regulations, determining what attributes the vehicle needs, and creating the car in math. “From this we know what the car looks like,” he says. “We know what the body surface has to look like to achieve a certain aerodynamic performance. We know what the engine architecture has to be to achieve the torque, horsepower, and fuel consumption levels we need.” Without a doubt, these guys are both racers and engineers.

Lynn Bishop graduated from the Rochester Institute of Technology with a BS in Mechanical Engineering in 1992, and went to work directly for Ford. While working in the Vehicle Dynamics CAE Dept., Bishop met Doug Louth, and the two later left Ford to go it alone before joining Pratt & Miller. “Everybody at Ford was talking about how cool it would be to develop a central database all of the engineers from all of the vehicle attributes could access,” says Bishop, “and Doug and I totally bought into it.” Because they were tired of waiting for it to happen, “we went off to do it ourselves.” Louth and Bishop have a pilot program with Ford to implement the database in a vehicle development program. “Ironically,” Bishop remarks, “the supervisor of this program used to supervise us.”

GM is Pratt & Miller‘s only racing customer, the automaker binding the team to an exclusive contract through the 2010 season. Its focus on analytical tools arose out of GM not giving Pratt & Miller an opportunity to bid on the ill-fated Cadillac Le Mans prototype program due to its impression of the company as “racers, fabricators, and a job shop.” However, when Ford wanted to quantify the effects rising weight might have on the Ford GT’s performance, it asked Pratt & Miller to create closed-loop simulations. And while Bishop states, “We want to be leading the technological edge in implementing new CAE technology and tools to help in the product development process, know that two goals are still unmet. “We’d love to take the overall win at both Le Mans and the Indy 500.”