Generating the power

When BMW decided to tackle Formula One, they started with a humble four-cylinder saloon car engine - and quickly extracted power of a level we might never see again. Keith Howard asked Paul Rosche, the man in charge of engine development, how they did it

One of Paul Roschen favourite photographs (It was on the wall of his office at BMW Motoisport for years) was taken at the first, secret test of the M12/13 turbocharged Fl engine at Paul Ricard in 1981. From under the back wing of the Brabharn BT49T a huge sheet of flame erupts on the overrun, perhaps a metre long. Looking more like the efflux of a rocket than the exhaust of an internal combustion engine, the image was an appropriate portent. Within a couple of years, BMW's four-pot turbo would be delivering something over 1400bhp in qualifying trim — vastly more power than any Formula One driver has had available before or since.

BMW's decision to enter F1 was no rush ofblood to the corporate head. The Munich firm had already made impressive turbo engines for the German touring car championship and IMSA in the US, and a very successful normally-aspirated engine for F2. Formula One was the obvious next step.

To maintain at least a nominal link with the company's showroom products, they decided to base the engine on the cylinder block from a road car. That meant the in-line four from the 316/318/518, designed by Alex von Falkenhausen as long ago as 1959. It was less than ideal, and looked a little daft alongside the bespoke turbo V6s of Renault and Ferrari, but basic strength wasn't one of its problems.

Stories are legion about the measures taken to prepare this humble casting for F1 duty, but the truth is that, at least initially, very little needed doing to it. When the same block was used for the F2 engine, Rosche's team soon discovered that second-hand, high-mileage examples were preferable because they had been stress-relieved by service, improving their dimensional stability. For the F1 engine a heat treatment process was used to achieve the same end, with blocks lifted straight from the production line.

The only other modifications were to machine internal surfaces to improve oil return and remove stress raisers, and to remove redundant ribs and mounting points. Together these shaved about 7kg from the weight Only when the power rose above 750/800bhp did strengthening prove necessary. Everything else on the engine was purpose designed. The DOHC four-valve head, based on the F2 design, was cast in aluminium, the camshaft carriers and inlet manifold of magnesium. To cope with the extreme combustion pressures, Mahle forged aluminium pistons passed the baton first to H-section titanium con rods, and thence to a forged steel crankshaft running in five main bearings. Although oilways were widened to improve lubrication, the bearings themselves remained the same size as on production engines. A shallow cast magnesium oil pan fed the dry-sump lubrication system.

Fuelling was by Kugelfischer mechanical injection pump, to which a stepper motor was added, controlled by Bosch electronics, to provide the threedimensional fuelling map (engine speed, engine load, boost pressure) required by a turbo engine. Initially the control electronics were analogue, and they proved a major headache during 1981 and the engine's first fitful race appearances in 1982. Able now to laugh at problems which, at the time, nearly resulted in a tenninal split with Brabham, Rosche jokes: "With the analogue system, I think it was only luck that gave us the right mixture. We had no idea how the engine behaved on transients — we had no dynamic dyno at BMW then — and therefore we had to find this out on the race circuit

"For the whole of 1981, we ran on circuits — mainly Donington, Silverstone and Paul Ricard — for this reason. We were the first to use a telemetry system [bought from aircraft manufacturer MBB] and it was an important step to developing the engine. The acquisition system fitted in a small truck, and in those days the computers were so limited we had to write all the data down on sheets of paper. But it was very helpful."

Unfortunately human error was to undo much early development work. "We found a set-up where Nelson [Piquet] said the engine was fantastic, don't touch a thing. The Bosch technician then took the controller back to the factory to note down the settings, but he plugged it into the mains rather than 12 volts and the whole box burnt out So Bosch made some new boxes and we went to Ricard for more testing — and during that week we broke 11 engines. I don't think we ever found that ideal setting again." Only when the analogue control box was superseded by a digital controller for 1983 were the engine's reliability problems finally put behind it. All up, complete with heat exchangers, the M12/13 weighed about 170kg, some 30kg more than the Cosworth DFV. But that was a small penalty to pay fora whole new world of power. BMW's 14litre touring car turbo engine had developed around 540bhp at a time when the Cosworth was worth about 470bhp, so it was plain that the Fl engine, with 100cc greater capacity, would easily see off the 3-litre normally-aspirated competition. It began life with around 560bhp (557bhp was quoted) and a boost pressure of 2.9 bar absolute from the single KKK turbo, after which things went slightly mad.

In the search for greater power from their own turbo engines, Renault and Ferrari both started using water injection to cool the inlet charge. The idea was that this would increase the amount of boost pressure that could be used before detonation set in (this being the era before HA-mandated boostlimiting pop-off valves). BMW had tried the same but to no useful effect. Water concentrations of up to seven per cent had been tested, beyond which the power began to tail off. All they discovered, in Paul Rosche's own ironic words, "was that water doesn't bum." So instead, Rosche telephoned a contact at chemicals giant BASF and asked if a different fuel formulation might do the trick. After a little research, a fuel mix was unearthed that had been developed for Luftwaffe fighters during World War II, when Germany had been short of lead. Rosche asked fora 200-hire drum of the fuel for testing and, when it arrived, he took it straight to the dyno.

"Suddenly the detonation was gone. We could increase the boost pressure, and the power, without problems. The maximum boost pressure we saw on the dyno was 5.6 bar absolute, at which the engine was developing more than 1400 horsepower. It was maybe 1420 or 1450 horsepower, we really don't know because we couldn't measure it — our dyno only went up to 1400."

This fuel formulation was used from 1983. "Maximum power was strictly for qualifying, though, only for one or two laps. In the race we ran about 1000 horsepower. This was one of the reasons we won the championship in 1983." And when an engine did fail under the strain of such enormous outputs, was it the block that wilted? Not usually. The engine's weakest point was actually the piston, around the gudgeon pin. That low-born street lump was still game for more.