Rolling roads and the dyno lottery
A question we are asked on a daily basis is how we measure our published figures, and how do we know that they are correct? Why are our figures lower than some competitors?
The reason for this is twofold:
Firstly, we don't believe in selling based upon inflated figures in order to get a sale. In the tuning industry, as with many other industries, bigger is often seen as better. The more discerning customer may not agree entirely, and ask further questions, and that's where we come in. We believe our figures to be correct, if anything slightly lower than we quote, as we would much rather under-quote and over-achieve. If we told a customer figure "x" and an independent dyno test comes in lower, we would likely have some questions to answer, and potentially an awkward conversation to have. However if we tell a customer figure "x" and we match or beat that figure, it's generally smiles all-round. And we like smiles.
Secondly, it depends on the way the power was measured. And as with just about anything in tuning, dynos aren't dynos in the same way that apples aren't always apples.
Dynamometers, or dynos as they're commonly known, basically consist of a weighing scale (load cell), a torque arm which is connected to the rollers where the road wheels will sit, the aforementioned rollers and a frame for it to all sit in. Most dynos have eddy-current retarder brakes attached to at least one set of rollers in order to control the rate of acceleration.
The thing is, all of these things are changeable and vary in accuracy and/or quality. How accurate is the load cell? Have you ever used a set of scales at someone else's house or a hotel and noticed that your weight is higher or lower than you thought? What about really the unforgiving ones at the gym? Hopefully you're starting to understand where I'm coming from with this. So outside of the accuracy of the load cell, what about the construction of the rollers? The rollers on our Bapro BPA4-R rolling road weigh 250kgs each, are knurled for grip and the front and rear rollers for each axle are belt-linked to ensure that no slip is possible, which can give wildly inaccurate and inconsistent figures. This means that before the retarders are even activated, the car has to rotate a tonne of mass in just the rollers themselves. This means higher accuracy and a more representative simulation of road conditions.
Next come the brakes. Our dyno uses a very large Frenelsa retarder on each axle to control the rate of acceleration very accurately. It's very important to do that, given that we need a fixed rate of acceleration and measured time along with weight (that's where the load cell comes in) to calculate power. Frenelsa very kindly posted this picture to briefly explain how an eddy current brake works.
So without going into the mathematical theory on how power is actually worked out from acceleration and weight, once we know the power that has been transmitted to the rollers at all the RPM sites (our Bapro measures at 10Hz - 10 times per second), we can graph that figure as wheel horse power. Now if all that wasn't enough to be potentially inaccurate for you, wait until you hear this. In order to calculate the power at the crankshaft of the engine, which is what we're all interested in and how the manufacturer quotes their figures, we need to measure the amount of drag created by the mechanicals of the drivetrain. This is normally done via a coast-down procedure, where the clutch is disengaged and the car is allowed to slow down to an almost stop naturally. This calculates, as accurately as possible, the amount of drag at each RPM site, again at 10Hz.
Some dynos GUESS this figure! Almost unbelievable, but true. Others normalise the curve, some even pick one RPM site (normally the peak power RPM) and add that to the whole graph interpolated back to 0. This is clearly never going to be an accurate way to measure a coast-down, and therefore not an accurate way to measure crankshaft horse power either.
The coast-down losses measure everything that mechanically drags the power of your car at the wheels, including the type of tyres that you use, the tyre pressures (even if they change as they get hot), the weight of the wheels and brake discs, the wheel and gearbox bearings and even the weight of the shafts.
Once the coast down has been measured, it is added to the measured wheel horse power to show our crankshaft power. This is calculated from the engine torque in Newton Metres (Nm) vs RPM.
There are some correction factors used in order to ensure readings are comparable between different temperatures, barometric readings and humidity. Some dyno operators have to put those figures in manually, whereas our dyno reads them directly from a control box in the dyno room, which of course protects from the chance of manipulation.
The last but not least piece of dyno variance comes down to the actual rate of acceleration. Each dyno gives the operator a finite adjustment of "ramp rate" that they would like to use. Of course, it's important to closely represent road conditions but not all dynos and not all operators do. If you spend your time watching some dyno videos online, you'll notice that an alleged 1000hp Polo (other dyno videos are available) does a full dyno run in about 4 seconds before spitting flames out of the exhaust and a figure is proclaimed. No. Just no. A dyno run should be representative to how a car actually performs on the road and there's absolutely no way any car would get through 4th gear in a few small seconds. Watch a few of our videos and you'll see that a normal 300-500hp dyno run takes around 25 seconds to perform and a bit longer to coast down. In these precious seconds, a car is constantly working out how much fuel to add, monitoring its own sensors, checking boost levels and trimming them up and down, calculating the maximum best ignition angle and trying to use it whilst listening for knock. All of this takes a little bit of time, nanoseconds but time nonetheless. And for a car to understand what it's doing, and perform correctly, and create heat like it would on the road, it needs to have the opportunity to do it, like it would on the road.
One final thing that people don't often attribute to a dyno itself, and that is air management. A car consumes air, at quite a surprising rate, and all that air has to be supplied and expelled correctly. The size (and effectiveness) of dyno cooling fans, the supply of fresh air to the building, and again the size and effectiveness of the extraction system cannot be overlooked. Exhaust gas doesn't see an open door and think to itself "Ah ha, that's where I've got to go", it has to be collected, accelerated and thrown out with a decent extraction system. You'll see many dyno videos that use small axial fans in order to blow exhaust gas away, but think to yourself, at over 100mph which almost any car will be doing on a dyno, how much air would be passing your window at that speed? How quickly would the exhaust gas dissipate, and how easy is it for the engine to suck replacement air in ready to used? Stick your hand out of the window at 100mph (on a closed track, of course) to give yourself an idea. Then use a dyno with demonstrably decent fans, because a car starved of oxygen will perform badly, just the same as us humans.
So, dynos are not dynos. Apples are not apples. Some apples taste sour, some dyno setups aren't much good. We reckon our apples taste pretty good, and as they taste good enough for vehicle manufacturers to use for calibration work, and to be used as the MSA regulation power test dyno for two UK race series, then that'll do for us.