Englishman Keith Duckworth, designer of the highly successful Cosworth DFV Formula 1 engine, said too many builders were mesmerized by airflow but then did nothing to burn all the air they’d stuffed into the cylinder.

When I was a bored high school student, we had a school assembly once for a visiting showman from General Motors. One of the demos he showed us was to put a few drops of gasoline into a plexiglass tube about 3 feet long plugged at one end and with a tennis ball inside. Capping the open end, he stirred that fuel into the air in the tube by rolling a tennis ball back and forth inside it. Then he removed the cap and tennis ball, whipped out his lighter, and held it to the open end.

Borrrrp was the rather drawn-out sound we heard as the flame took 2–3 seconds to burn its visibly slow way to the plugged end of the tube. This tells us that the speed of combustion in perfectly still gasoline-air mixture is about 1 foot per second. If that were the actual speed of combustion in a motorcycle engine, could it even run?

Taking a 4-inch bore as an example and assuming a modern centrally located spark plug, the flame travel is 2 inches so at 1 foot per second the flame will need 1/6th of a second to burn from spark plug to cylinder wall. If we're just chugging along at 3,000 rpm, that implies the crankshaft will complete 8.3 rotations while the flame is on its way. And if this is a World Superbike Ducati V4 R turning 12,000 rpm, the crank will turn 33 times.

That’s not going to work! We need the flame travel to be a fraction of one crank revolution. And that’s just what it is in real-world engines, whose ignition is typically timed to fire somewhere between 30 and 45 degrees before top dead center.

When we do the arithmetic, we find that flame speed in actual engines is 50 to 200 times faster than in the quiescent mixture experiment I witnessed as a teenager. Where is that extra flame speed coming from?

Chaz Davies
Head down, bum up: World Superbike race winner Chaz Davies reacts to Ducati’s practice of using intake velocity to generate fast combustion. Davies and teammate Álvaro Bautista are eighth and second, respectively, in the championship with one round remaining before the summer break.Courtesy of Ducati

It comes from rapid air motion in the fuel-air charge, resulting from the high velocity of mixture entering the cylinder through the intake valve(s). When a spark generates a flame kernel in fast-moving mixture, on the micro-local scale, the flame front moves at the usual 1 foot per second, but because that air is swirling and eddying, it tears the flame kernel into many shreds and quickly disperses them throughout the whole mixture. By hugely increasing the area of flame, the charge is consumed by an apparent flame speed of 50 to 200 feet per second.

Returning to the example of 4-inch pistons in an engine turning 12,000 rpm, if we ignite the mixture at 40 degrees BTDC and combustion ends roughly 40 ATDC, that is (40 + 40)/360 of a revolution, or 0.22, and since one revolution takes 60/12,000 of a second, or 0.005, we see that combustion in our example takes 0.005 x 0.22, or just over 0.001 of a second. Traveling 2 inches in that time—2 inches being half the bore of 4 inches—we get an apparent flame speed of 183 feet per second.

In a conversation with Ducati CEO Claudio Domenicali, I learned that company practices in using intake velocity to generate fast combustion employ mainly two variables:

  1. Intake port size: the smaller the port, the greater the intake velocity that can be applied to the job of making the fuel-air mixture as turbulent as it needs to be to burn rapidly.

  2. Intake port downdraft angle: This acts like the hose we've all played with while filling a bucket. If the water goes straight into the bucket, the water accumulating there doesn't rotate. But as we aim the hose more off-center, the stream drives the water into rotation.

In modern four-valve engines with flat combustion chambers, the goal is not to make the “water in the bucket” rotate around the centerline of the bucket but to “tumble,” to shoot from the intake valves to the far cylinder wall, be deflected downward by it, then curl back across the top of the piston to rise up the near cylinder wall. This creates rotation but around a different axis. As the piston rises to top center on compression, this orderly rotating motion breaks up into smaller and smaller turbulence cells, which will shred and widely distribute the flame from the kernel originating at the spark plug.

Some will say, “Yes, but don’t we also use squish to generate rapid charge motion and so accelerate the flame?”

Fast combustion in the DFV reduced the heat lost to combustion-chamber surfaces, leaving more heat and pressure to push its pistons.

Squish describes what happens as a surface on the piston closely (0.030 of an inch or even closer) approaches a matching surface of the head at TDC, rapidly “squishing out” the mixture between the two. This creates a jet that can stir the main mixture. Duckworth, on his way to discovering the power of tumble motion, tried for two years to use squish to achieve fast combustion in his SCA engine. It wouldn’t cooperate, even at only 70 percent of the rpm of the above Ducati example. That forced Duckworth to think through in detail what was necessary. In the end, a combination of such focused thought plus experiment resulted in the combustion-chamber concept now found in most of the world’s car, light truck, and motorcycle engines.

In 1967, Duckworth’s DFV V-8 Formula 1 engine was able to outpower V-12s with far more valve area and peak rpm. How? Fast combustion in the DFV reduced the heat lost to combustion-chamber surfaces, leaving more heat and pressure to push its pistons. Duckworth’s plain flat-topped pistons gathered less heat than did the extra surface area of high piston domes in the Matra V-12.

Duckworth’s thought and work are all around us.