By Dave Hamer:
I joined Benetton in 1988 when they decided to create an R&D department to support their Formula One active suspension ambitions.
Both Lotus and Williams had already seized the moment, pushing well ahead with similar programs. Though Benetton was also running an Active test car, they employed a staff of only 80, considerably fewer than the other leading teams.
The initial system was developed using various Citroen hydraulic parts, including accumulators and a pressure control valve. The hydraulic power was generated by a Sunstrand gear pump, similar to the high pressure oil pump of a central heating boiler, which Lotus also used at the time. However, these pumps were barely adequate for F1 use; operating at 2,000psi-plus, they required constant attention.
The events that led to Benetton’s active suspension and the system’s major components
The advent of aerodynamics and particularly ground-effect brought new challenges to suspension design. Ground-effect relied on the car running at a low consistent ride height and, as such, generated huge downforce and cornering power. Notably, spring rates had increased to 3,000lb/in., making the suspension almost solid. Yet due to tyre squash, the car’s ride height still changed with the car’s speed and its mechanical grip was poor.
The solution came in the form of new technology known as Active suspension, which replaced the springs and dampers (mostly coil-over-shocks) and anti-roll bars with hydraulic actuators (rams) controlled by a computer. The system employed a pump to provide the pressurised fluid which was piped to a computer-driven valve (a moog servo valve) at each corner of the car. This either directed fluid to the actuator or discharged fluid from it to the return line and back to the tank. So now the car’s ride height, including the effects of weight transfer, could be kept constant and true, thus compensating for tyre squash at high downforce loads.
This was the format used for that first Benetton system. The software in the controller (chassis control unit or computer in a black box) used an algorithm (mathematical model) to decide what signal to send to each servo valve depending on input signals from various sensors. It’s important to add that luck intervened on our side when we encountered Dave Bulman of Shrivenham Military College; Dave had developed active control of the guns on tanks and was a keen motor sports fan.
We also used the Shrivenham lab facilities to test a single corner of the car on what was known as a single-post rig. In this way, we could tune the active system to improve the dynamic response and compare it to the performance of a conventional spring/damper unit. It quickly became clear the active system was far superior. We later used their four-post rig to test the complete car.
The headaches of air in fluid
To achieve optimum and consistent response from the system, it was imperative that the fluid contain no air—like brakes that need bleeding. Also worth noting, we knew others struggled with air in their systems. The design of our oil tank followed conventional race engine dry-sump tank practice, having a swirl pot to expel air from the fluid and a vent to atmosphere. This was the wrong approach.
Unlike an engine oiling system with blow-by gasses to expel, a hydraulic system, if properly bled, would seem to be air-free. Yet this is not the case, as there is air dissolved in the oil and, like a bottle of soda, if the pressure drops a mass of bubbles form as the air comes out of solution. For this reason, the system must operate above atmospheric pressure; hence it must be a sealed system.
And so, this insight formed our second approach. Introducing what is known as a bootstrap tank, it featured a small diameter piston—with system pressure on it—that pushes a large (around 4in) diameter piston that pressurises the fluid in a cylindrical tank/reservoir. The piston areas are such that a 2,000psi system pressure generates 20psi in the tank, which contained about 1 pint of fluid. Current F1 hydraulic systems use a small accumulator as the tank where gas pressure on one side of a bladder keeps the fluid under pressure. Not only simpler and lighter, it probably contains one tenth of a pint.
The sealed pressurised system has the advantage of having no air space above the fluid. Therefore, it cannot slosh around or move away from the pump inlet pipe under high G loads, and as the pump inlet has fluid under pressure it cannot cavitate. This was the essential requirement of the aircraft pumps we were intending to use.
Typically, F1 mechanics are a skeptical lot and, as such, ours dreaded the prospects of their celebrated yet simple cars being covered in hydraulics. They were convinced they would be unreliable and leak fluid everywhere, making a mess of their exquisitely finished racing machines. Management also had concerns, as other teams with Active cars had given drivers quite a fright when the system failed.
Understandably, we in R&D were obliged to develop and test all the components to ensure reliability. We had a hydraulic test room with windows peering out onto the race bays. When something went wrong, groups of curious mechanics would gather, taking much amusement as they witnessed streams of oil running down the windows. High pressure oil goes an awfully long way.
At last, after many months we were satisfied with progress, and it was decided to have a dedicated Active test car. To ensure the system was reliable, it would test almost continually throughout 1992.
Prospects of pump improvements: from exasperation to final success
The system on the test car had many improvements. For a start, the industrial spec hydraulic gear pump was replaced by an aircraft swash plate pump. These have a number of pistons in a cylinder (like Colt 45 bullets in the chamber). The piston stroke is controlled by the angle of the swash plate (like a helicopter rotor), so the pump only pressurises the amount of fluid required to keep the pressure constant. There is no relief valve required–which would heat the fluid and waste energy. The pump consumed around 7hp at full flow demand.
Previously, we had tried to obtain aircraft pumps, but the manufacturers approached showed little interest in supplying F1 teams; we were considered small-time—persona non grata. Luckily, an economic downturn intervened, which meant cutbacks on aircraft production and the number of pumps being sold. Now, our pursuits piqued their interests. In fact, I negotiated with Vickers for the pumps and found them most helpful. Ultimately, we adopted a modified pump made for a Tucano two-seat turbo-prop high performance trainer.
Unforgettably, a very objectionable, unhelpful chap at Cosworth (our engine supplier) prevented us driving the pump from his precious engine. So, for this year, we powered it via the transmission. But when the car first ran, vibrations occurred activated by the pump’s sideways mounting on the gearbox. They caused the swash plate to flutter and lose pressure at certain engine speeds. The pump had been designed for mounting on a sophisticated turbo-prop aircraft not a rough old F1 car! Pleasingly, Vickers quickly remedied the difficulty by using a stronger return spring and modifying the swash plate mechanism.
Learning how to bleed and maintain oil cleanliness
Much thought was devoted as to how we would bleed, keep the oil clean, and generally look after the system. Fortunately, my colleague, Bill Millar, had a contact in the Royal Air Force. Bill had worked for Short Brothers, the aerospace company based in Belfast, Northern Ireland and smooth-talked his way into our arranging a visit to RAF Cottesmore, which had an active Tornado fighter squadron. Consequently, we were able to pick the brains of their hydraulic servicing department. If we could learn how they kept the hydraulic systems reliable on a front-line fighter, we might be able to apply those lessons to advance our own Active suspension aspirations. Helpful in every way, they gave us chapter and verse on how they serviced, bled, tested, and achieved the level of cleanliness required.
Flushing rigs and the painstaking process of expelling air from fluid
From our visit, the conclusions were clear: we needed to build flushing rigs similar in principle to those of the RAF. The servo valves that control the oil in and out of the actuators operate with very close clearances and some extremely small orifices. So, the system needs to be scrupulously clean. And even when meticulous care is taken during the assembly of the system, every time a hose fitting is tightened, it generates thousands of microscopic particles, which can spell disaster.
Therefore the system must be flushed using as much flow as possible from the flushing rig. Ours consisted of a tank holding approximately 20 litres of OM15 oil, which is the oil we initially used in the car’s hydraulics. This is the same fluid used in military aircraft and is very thin, like paraffin, and colored with a red dye and a smell so distinctive it remains with you for days.
It had a gear pump driven by an electric motor, which drove the fluid through a very fine filter and a dry-break fitting into the car pressure line. There was also a dry-break on the car’s reservoir to which a return line to the flushing rig tank was attached.
The oil then flowed over an inverted cone in the tank before returning to the bulk fluid. The tank was kept under as much depression as possible by a vacuum pump which expelled any air. The tank also had a heating element, as hot oil flushes better.
One of the first jobs of the day was to switch on the heater and circulate the fluid within the rig. It was not connected to the car at this stage. As the fluid had been sitting overnight, it absorbed air, so you couldn’t apply much depression (vacuum) on the tank without the pump cavitating, which was audible. Fluid that is free of air flows silently; if you can hear it, it contains air out of solution. After approximately 20 minutes you could increase the depression a little before the onset of cavitation, as the fluid now had less air in solution. After an hour or so, higher depression could be applied without cavitation. The rig could then be connected to flush and fill the car. When flushing the car to clean the system, the moog servo valves are removed and a flushing plate fitted to allow flow through the system without the risk of introducing debris to the valves. Cleanliness was checked by drawing some fluid and passing it through a filter membrane. This is then put on a slide and examined under a microscope and compared to GO-NO-GO reference slides. A sample would be taken at lunchtime and at the end of day. When satisfied the system is clean, the valves are refitted and the car exercised by extending and compressing the actuators to expel any remaining air.
Each actuator had a pressure transducer and also a potentiometer inside to give position. The front of the chassis had an accelerometer on axle centre line to sense vertical accelerations. There was also one on the rear axle centre line, which was mounted on a heavy base that was then rubber-mounted to the gearbox. This mounting was tuned such that high-frequency accelerations were filtered out, including engine vibrations that would destroy the accelerometer; we were interested only in the vertical movement of the car.
Initial track-testing encounters
Meticulous care rewarded us with a reliable air-free system. Even so, throughout the development period of 1992, we wouldn’t permit Michael Schumacher to drive the car for fear of him losing confidence in it. Instead, Alex Zanardi performed all the testing. After his F1 career, Alex went on to race Indy cars and, sadly, lost both legs in a serious accident in East Germany. Conspicuously, it didn’t stop the irrepressible Italian from wheelchair racing.
Despite our best efforts, though, things still went wrong. While testing at Silverstone once, the hydraulic pressure dropped, so Alex pitted. We found a pinhole in one of the high-pressure hoses which had discharged all the fluid. Despite using the best hoses money could buy, the defective one lacked the carbon trace that prevents static build-up. Without this, the static electricity generated by the high fluid flow had sparked through the hose and grounded on the gearbox casing, causing the pinhole.
Properly seen, an active system needs to do more than just control the ride height; it needs to absorb bumps, and if setup properly can do a much better job than springs and dampers. Ours was adequate for 30Hz (30 cycles-per-second). For higher frequencies and for massive bumps (like kerb strikes), a small stack of Belleville spring washers resided in the base of each actuator behind a floating piston. Had we aimed for higher frequency response, the power requirement would have exceeded the 7hp available. Also, to reduce drag, a button was fitted to the steering wheel to lower the rear of the car on the straights. I mimicked this some years later with the Pro Pitch System (PPS), though only as a passive system of course.
Off to the races!
Our active suspension system contended the F1 World Championship of 1993. The first race was set for Kyalami in South Africa. Our drivers were Michael Schumacher and Riccardo Patrese. To ensure the cars were fit for competition, we arrived a week early and transported Michael’s car about 100 miles north of Johannesburg to a small Kart track known to *Rory Byrne. Rory, a South African, was the chief designer at Benetton and later at Scuderia Ferrari. Michael tested the car around this very narrow twisty course, and having satisfied ourselves that all was well, we returned to Kyalami. As to our successes in 1993 with Schumacher, the year culminated in one race victory in Portugal, 3 second places, 4 third places and retired 7 times. Every time he finished he was on the podium. As for the mechanics, the system met with their full approval. It required significantly less setup checking such as ride height and spring changes; they just plugged in the computer.
Racing people love the big moment and the daring act, and F1, like most esteemed sports, is infused with politics, so dull moments were rare. On one occasion during the 1993 season, I was asked to bring my passport to work the following day. A trip somewhere with destination unspecified. But nothing happened the next day. It was some time later I learned my outing had been planned for Italy, to Maranello, 30 miles west of Bologna. Seemingly, the Ferrari F1 team had encountered difficulties with their active suspension, and, apparently, their image was so inextricably linked to that of F1, assistance must be provided! My mission was to ascertain if our system could be installed on their title contender – our rival! But the proposal was not well-received by John Barnard, Ferrari’s then Technical Director, and his rebuff was entirely plausible. Barnard, accomplished and strong-willed, had served Benetton as Technical Director from around 1990 to ’92. During his tenure, however, eleven of us quit, including Rory Bryne and Pat Symonds.
We joined Adrian Reynard’s fledgling F1 program, and the year spent at his Bicester factory was probably the most enjoyable of my career. We never built a car, as our meagre funds evaporated. But during this brief interval, our small team, approximately 15 in number, produced and tested a wind tunnel model, many Active parts, and succeeded in completing a fully designed race car. It was a fun time for I felt I played an essential role, and recognised as such, and not just another small cog in a big wheel.
Luckily, John Barnard departed Benetton following a dispute with team principal, Flavio Briatore just before the Reynard F1 project folded. Adrian tried to sell our design to Ligier and a few of us were flown over to Magny Cours for the day and wined and dined by Guy Ligier. But he would only consider the project if he could recruit the people to go with it, so we were all offered employment! Meanwhile Benetton didn’t have a car for the 1992 season. So, Rory negotiated our return. At the time the politics were intense, and when we rejoined we were accompanied by half the TWR staff, which included Ross Brawn as Technical Director.
And there’s a sequel
Inevitably, Active suspension systems were banned with other controls for 1994 in an attempt to reduce costs. So to maximize the opportunities available in 1993, we fitted the car with four-wheel steering. This was a computer-controlled hydraulic system at the rear adopted for the last two races of the season.
Moreover, our Active campaign had resulted in reliable race car hydraulics. And soon after, we introduced hydraulically assisted power steering and clutch and gear-change operations. The power steering was well received by Gerhard Berger who stated “now I can cancel my gym membership”, whereas Jean Alesi insisted “it was for pansies” and wanted nothing to do with it. Nonetheless, power steering was fitted to his car with very low gain/assistance initially and through the year the gain was steadily and surreptitiously increased, though he would never acknowledge it.
Crucially, the knowledge gained from developing the Active car also led to our widespread use of seven-post rigs, where bumps and forces are introduced to the car on the rig that simulate track conditions. Seven-post rig testing remains essential tuning technology today.
*Rory Byrne, now in his seventies and still active, continues to work six months of the year at Ferrari. He maintains a home at Maranello and helped develop the carbon chassis design of the La Ferrari. He tests their road cars… and continues to drive like a maniac!
NB, When I joined the team in 1988, Benetton’s machine shop comprised three Harrison lathes and two Bridgeport mills. Today, you would scarcely recognise it as a machine shop. It doesn’t even smell of cutting fluid! How times have changed.
To read Dave’s article on how F1 cars achieved consistent ride height click here.
Or, about tuning with mass dampers, click here.