Dynamometer runs of an engine fitted with two "identical" sets of hand-ground cylinder heads can vary by 5 or 10 hp. NC-machined cylinder heads, in contrast, keep that number within 1 or 2 hp. Even the most-gifted craftsman can't match a machine's accuracy and repeatability.
NASCAR mandates all teams begin with the same manifold castings. GM and Edelbrock Corp., Torrance, Calif., supply several designs in three basic versions: short runner for power (high speed); long runner for high torque (short track); and a special restrictor plate design (super speedways). The manifold runners come "roughcast" for modification by teams. NASCAR rules permit unlimited modification to internal passages on short and long-runner manifolds, but limits runner height, opening taper, and other features on restrictor plate designs.
GM farms out machining of rough head castings. Heads arrive at team shops with finished intake and deck faces, valve guide bores, stud holes, rocker pads and holes, and counterbores for the valve seats. Teams choose valve sizes, and shape the combustion chamber, intake, and exhaust ports.
THROWING IT IN REVERSE
A 3D point cloud is the first step to build a digital model of a physical object. RCR engineers use a CMM touch probe to digitize internal port surfaces of intake manifolds and cylinder heads without having to cut the parts in two. The raw data, or point cloud, in ASCII format goes to Geomagic Studio software from Geomagic, Research Triangle Park, N.C. The software converts the raw data to a Nurbs surface in IGES format for import to Pro/Engineer. RCR engineers first tried taking a point cloud directly to CAD, though connecting the points with CAD tools proved cumbersome.
In the case of an intake manifold, engineers reference the files containing runner and plenum data to the same coordinate system and merge them into a single point cloud. The software automatically converts the point-cloud model into a polygon model. It also fills holes in the polygon model and validates intersections.Next, an autosurfacing routine turns the polygon model into a watertight Nurbs surface.
The resulting surface patches are close to the desired final quality. Minor edits and interpolation finish areas of missing point-cloud data. Grids are constructed on the revised patch layout, and the software creates the final surface model for half of the manifold. An IGES file of the model goes to Pro/Engineer where engineers copy and rotate it to generate the other half. The complete model goes to a five-axis surfacing program, which generates the tool paths for an NC machine. The machine does the porting net shape, necessitating only minor hand tweaks. For cylinder heads, a separate, dedicated machine cuts the valve seats.
EXPERIENCE TRUMPS THEORY
Reverse engineering saves a lot of time and toil, but the resulting parts are only as good as the one-off designs on which they are based. Prior experience in this case trumps fancy theories because the latter often border on black art. "Flow in a V8 intake manifold is highly complex," explains RCR Race Engineer, Chad Zimmer. "All eight cylinders feed from a common plenum and a single four-barrel carburetor. The arrangement tends to interrupt the low-pressure signal responsible for drawing fuel from the carburetor jets."
The goal, of course, is equal fuel distribution to all cylinders. But so-called "cross-talk" between individual cylinders and the manifold prevents realizing this ideal. Each cylinder runs a little different because each sees a slightly different fuel-air mixture. Factors such as bore spacing, bank offset, port shape, and the finish of port surfaces, significantly influence flow and power output.
"To my knowledge, nobody has successfully done an accurate, full-blown CFD analysis on a V8 manifold with mixed air-fuel flow and heat transfer," says Zimmer. "We do CFD analyses on pure airflow, which helps point designs in the right direction. A flow bench lets us compare different heads and intakes."
RCR engineers just recently began building a library of digitally designed manifold runner and port shapes. The surface models are unparameterized features and not as well defined as, say, a prismatic shape in which dimensions are easily changed. But they can serve as a reference for free-form modeling.
Ultimately, any legal modification that makes more horsepower is considered fair game. Teams primarily rely on an engine dynamometer to sort out what works. R&D engineers will run a head-matrix test of, say, six sets of heads with slightly different port and chamber shapes. Such optimization efforts may yield a few tenths percent of total horsepower. That's huge considering that the typical Nextel Cup engine makes north of 750 hp.
"We try to give drivers as much horsepower as they can use," Zimmer says. "If we find during track tests that a driver routinely runs the throttle wide-open, he needs more horsepower. With cars finishing inches apart, it matters."
Circling in on spiraling costs
It goes without saying NASCAR is big business. What fans see on race day, however, is just the beginning. The big money is in spare parts. Building trick, one-off parts in mass quantities is a way NASCAR teams shave costs. Teams also get a leg up from recent NASCAR regs that ban the use of separate engines for practice, qualifying, and racing. A few years ago, teams would prepare a qualifying engine that made huge horsepower but lasted just a few laps. "It was a ‘time bomb,'" RCR's David Hart explains. "The engine oil was thinner than water to cut friction."
The trend now is toward standardization. "Essentially the only differences between the Busch, Nextel Cup, and Supertruck engines are the cylinder heads, intake manifold, and carburetor," says Hart. "The lower end, including the pistons and crankshaft, are nearly identical." Nextel Cup engines run at higher rpm than the others and thus need a bigger carburetor and intake valves, a different camshaft and rocker arms.
Still, one engine costs $75,000. The restrictor-plate engines each cost a jaw-dropping $250,000 because they need significantly more preparation and R&D to overcome the horsepower loss induced by the plate. All engines need a complete rebuild after one race.
An RCR technician prepares a prototype Car of Tomorrow. The ground-up design debuts this year and will compete in 17 races. Reasons cited for the change: parity and safety. All teams must use the new body style, which takes parity to a new level. Car bodies already must match 40 separate templates before they are certified to race. Half of the templates are specific to the manufacturers Chevy, Ford, Dodge, and this year, Toyota. This is so no one manufacturer dominates because of superior aerodynamics. A shorter nose and hence shorter moment arm on the new design necessitates a larger splitter to generate greater down force. A wing will replace the current rear spoiler.
Regarding safety, a taller roof and side windows improve visibility. The driver compartment sits more toward the car center to create a larger crush zone. Safety features seen in race cars today often are the result of driver injuries, or worse. The center roll-cage member bisecting the front window, for example, is dubbed the Earnhardt bar after the late driver, Dale Earnhardt. His accident at Talladega crushed the car top, prompting the design change. Likewise, Earnhardt's fatal crash at Daytona in 2001 triggered a redesign of the driver's seat. The new seats better support a driver's head, while a Han's device limits forces during a crash, factors cited in his death.