Speed Secrets: Big Boy Driving Simulators
By Ross Bentley
February 25, 2014
A few issues ago in Speed Secrets Weekly, Ryan Hieronymus talked about the value of iRacing and Playstation type simulators for drivers and how these affordable devices can help drivers practice and program their brains to new tracks and behaviors. What about us, the engineers? What can we use to hone our craft when we're not at the track? Simulators, but with a twist. And a much bigger price tag.
The simulators I get to play with are $5 million to $20 million each and can take up an entire building. Think modern flight simulator where pilots train in commercial jets, but for racing instead. However, it's much more difficult to make a simulator replicate a road vehicle than an airplane (for many reasons that I won't get into here). Every F1 team, and major sports car, Indy Car and NASCAR team has a simulator available to them like this. In the past they used wind tunnels, shaker rigs, and shock dynos. Now simulators are advancing the state of race car development at a mind-blowing pace. These DIL (Driver In Loop) simulators allow engineers to climb the development curve very quickly.
There are two primary types of DIL simulators in use today: the hexapod and the table type. The hexapod looks like the flight simulators we have seen, with a pod atop six hydraulic actuators. The table type uses a table-on-a-table to allow the driver pod to move in pitch, roll, and yaw. Each one has its advantages and limitations and both are used very successfully in all forms of racing. The driver sits in a cockpit which replicates the real car. Most teams use the actual steering wheel and data system in the DIL sim. Our hero driver sits in the cockpit with giant screens surrounding him and a massive sound system, including subwoofers in the seat to simulate the exact vibrations the actual car produces. Yes, we measure these vibrations on the actual car so we can replicate them in the DIL. In the control room, the engineers and operators sit with a com link to the driver. The sim engineer can program the sim to replicate the exact feel and response of any car he likes, with any set up. He loads the "model" into the computers and these computers move the simulator just like the actual car would respond to the driver's input. These computer models (nothing but many, many lines of code) are very expensive to produce and come from many laps of actual car-running and data-sampling. Now, the engineers can produce models of cars that don't even exist - but we're getting ahead of ourselves!
The hard part of simulating a race car is producing the lateral G-force of a corner. A centrifuge can produce G-forces, but not for short duration events, like a corner. So the trick we use is called "motion cueing." This is a fancy name for tricking the driver into thinking that something is happening even though it's not. The seat belts can be pulled down to help simulate braking; one belt can be pulled more than the other to help the feel of a corner; we can raise the visual view, so he thinks the nose is dropping under braking; we'll push the pod forward fast and he'll feel acceleration. The cockpit pod will move many feet side-to-side and fore-and-aft, giving the driver all the feel of the real car. Watching the unit from the outside, you would think it's a Disney ride gone crazy and that any human would throw up after ten seconds in it. But when the driver sits in the DIL, the screens and motion are perfectly coordinated and the driver feels like he is driving the real car. If it's off by the slightest amount, we know right away and the clean-up crew has to be called in. Every driver is sensitive to different stimuli, and we do some individual tuning for each driver the first time. As a sidelight, I love seeing what each driver needs to have adjusted in the motion-cueing strategy, because it shows me exactly what that driver is most and least sensitive to. Does he get braking cues from the force or the visual cues? Does he feel understeer from the force in his arms or the amount he turns the wheel? Good stuff for the engineer to know and it often explains why we get different feedback on the same car from two different drivers.
Our driver is in the DIL, motion-cueing is all sorted, car set-up is programmed in. Now we need a track. We use a computer model of each track and it's very expensive to produce. Interestingly enough, some of the best track models come from the game companies like iRacing. They laser-scan the tracks down to millimeter accuracy, so the simulator has every bump, curb, seam, etc. The surrounding visuals are very important as well. Most good drivers use these for reference points. If the guardrail is not positioned exactly correct, the driver will know it. A good laser-scanned track file can cost $25K or more to produce.
How it Works
I was involved in a recent DIL simulator-driven project which encompassed every benefit of these new machines. Let's take a look; it may give you an idea of what this tool can do and leave you, I'm sure, thinking about where this technology will go in the years to come.
I was assigned to build a sports racing prototype car for the very specific task of being the first one to run a lap at Road America in an SCCA car under 2:00. The DIL simulator was the key that allowed us to do it.
We started with a current car and developed the simulation model (that project is a full SSW story for another time - shaker rig, wind tunnel, K&C rig, spin rig, tire data at Calspan, etc). So now we have a base line sim model of an existing car, but that car is not fast enough. So the designers, aero experts, and suspension designers go to work to come up with different ideas. They narrow the possible designs to four aero configurations and two suspension designs. The engine guys have three possible engine configurations, producing different power curves. In the "old days" (five years ago), you would have to build all these different parts, go to the track and spend two years sorting out these combinations, at great expense.
We sat our test driver in the DIL and let him try them all. This simulator driving gig is HARD. Because there is never the need to go on the set-up pad, or make a long change, he sits in the car for four- to five- hour stints with only a break for lunch. Most good simulator drivers charge more for sim work than track driving work, and no wonder! Our guy did thousands of laps at Road America on the sim. With the push of a button, we could change the suspension design or complete body shape, or give him a different engine to try in only thirty seconds. We did two years of development in three days, narrowed down the best combination in the virtual world, then went and built the car for real, the way we knew it would be good, right out of the box. When the car hit the track for its first lap, it had already done thousands of miles at Road America.
I remember after we ran the car for the first time at the real track, I asked the test driver, "How did it feel?" I expected that big, excited, emotional reply when you take the first laps in a brand new race car. He said, "Well, it's nothing weird or different. It's just like the simulator." How good was the simulation? We tested for six days at Road America with the real car trying to improve the simulator set up - cambers, ride heights, shock settings, aero, gears - all that stuff. We ended up running our best laps ever on the exact set up derived on the simulator, and that best time was within 0.4 seconds of the simulator best time. Yes, a 750 pound, 380 HP, 670cc turbo sports racing car can do a lap at Road America within two seconds of a modern LMP1 car, and four seconds faster than a LMP2 car, but it would have not happened without the DIL simulator. If we'd done it the "old-fashioned way," we would still be trying today.
I will leave you with a few points to think about. DIL simulators are an amazing tool and we're just getting going.
- Tires never wear out. Each lap you have the same grip, so you never have to ask "Do you think that change was masked by the tires going off?"
- You can test on worn-out tires all day to optimize the end-of-stint performance.
- You can do one hundred qualifying runs and never use a set of new tires.
- You don't have to move one truck, crew member, or use one ounce of fuel or rebuild anything after a simulator test.
- Want to test a new engine? It takes sixty seconds to take the old one out and put the new one in. Diffs, geometry changes, spring changes can be done in thirty seconds, and never go on the set-up pad.
- Want to work on a particular track corner? The driver can use the back button and step back 200 feet at every push, to run one corner ten times in sixty seconds and really feel what it's doing or evaluate a small change.
- Never ever had to repair a DIL car, and we crash a lot!
- Ever wonder if that turn could be taken flat? Go for it and see! I personally know one driver who has used the simulator to improve his lap time at his home track by trying things he would never risk in the real car.
- Simulator time is required by some sanctioning bodies to get a license. Never been to Le Mans before? You must log simulator time to be approved now.
- Got a new driver in the team? Put him with the race engineer in the DIL and turn them loose. They will know each other and be on the same page, long before turning the first real laps.
- Every F1 team employs a simulator driver, sitting in the DIL at home base, running different set- ups the race engineers send back between sessions from the actual track. This way, they test between sessions and give the engineers the best solution for the next track session.
I will leave you with this. One F1 team technical director was asked that if testing budgets had to be cut, where he would cut them? He said, "Track testing! Please don't take the DIL simulator away from me."
- Jeff Braun