Measuring Real World Engine Performance

This article was first published in 2011.

The approach covered in this article is not for everyone. If you have plenty of money, don’t really want to see what your modifications are doing or have no time, then it’s is definitely not for you. It’s also not for you if you live in the middle of a crowded city and there are no empty roads accessible.  Finally, it doesn’t suit those with mega-powerful cars.

But if you want to measure and examine the results after each modification, it’s ideal. After an initial (low cost) purchase, it costs nothing. It is very accurate and, unlike much automotive performance testing, gives you actual, real-world, on-road results.

So what’s it all about? In short, it’s the measurement of the actual instantaneous acceleration of the car, graphed against engine rpm for the whole engine rev range. From these data you can graph the shape of the torque curve. Furthermore, you can assess percentage gains or losses, so seeing exactly what mods are working and what are not – and importantly, where within the engine rev range their impact is being felt.

But let’s take a step backwards and see what it’s all about.

Tractive Effort

When a car travels down a road, its tyres are pushing back on the pavement. If the push backwards equals the forces the car needs to overcome (on a flat road that’s aerodynamic drag and rolling resistance), then the car will move at a constant speed. If the force pushing backwards (called the tractive effort) is greater than the forces that need to be overcome, the car will accelerate. The greater the surplus of tractive effort, the greater the acceleration.

If the greater the acceleration, the greater the tractive effort; and the greater the tractive effort, the greater the torque being produced by the engine, then by measuring actual on-road acceleration we can see the shape of the engine’s effective torque curve. Taking this approach then automatically takes into account losses that occur at different rpm and gear-train loadings, aerodynamic loadings, losses due to accelerating rotational mass – the whole lot.

There’s no need for a dyno or expensive gear – and in fact, it’s more accurate than a dyno because it is like a dyno with an infinitely big single roller (well, one as big as the diameter of Earth!) and like a dyno mounted in a variable speed, climate-controlled wind tunnel. Furthermore, this incredible dyno is also programmed for the actual, varying ‘ramp rates’ that occur when your car accelerates in each gear!

Taking this approach won’t give you absolute numbers (“My car has 134kW at the wheels at 5000 rpm”) but it will show you changes (“My car has increased in acceleration by 6 per cent at 3000 rpm but it has lost 8 per cent at 6000 rpm”).

For my money, unless you’re into boasting, the absolute power outputs matter not at all: what matters is on-road performance – and, during modification, the on-road changes.

While I have explored the techniques covered in this article for close on 20 years, I have never applied them with such effectiveness as I did when I recently modified my diesel Skoda Roomster. That modification process allowed step by step comparison of which mods were working and which were not, and it also showed some clear advantages over using a dyno. The process made me realise how stunningly effective these techniques are.

The Instrument

So how do you measure instantaneous acceleration? Performance measuring accelerometers are available in two types - electronic and mechanical.

Electronic accelerometers are most often integrated into full digital performance computers and can be tied to data logging. But what if you don't have the money or expertise for an electronic accelerometer? A mechanical accelerometer is cheap and easy to obtain and use. Compared with an electronic accelerometer performance computer, it requires much more work from the user - but there are always going to be some trade-offs!

A mechanical accelerometer can use either a vertical pendulum that is deflected across a scale by acceleration, or a tube shaped in a semi-circle in which a small ball bearing is moved. A US company called Analytical Performance many years ago produced one of the best of the latter type of accelerometers, which was called the G-Curve. Their accelerometer consisted of an engraved alloy plate into which was let a long curved glass tube. The tube was filled with a damping fluid and a small ball bearing was sealed inside. A very good handbook was also provided with the instrument.

Unfortunately Analytical Performance is no longer in business but a substitute accelerometer can easily be assembled. Boat and yatching supply companies sell clinometers that are designed to measure the angle of boat heel.

One such clinometer is the 'Lev-O-Gage', which in construction is very similar to the G-Curve. However, because it is designed to measure heel angles, the scale is calibrated in degrees rather than g units. Like the G-Curve, the glass tube is filled with a damping fluid to prevent the ball overshooting.

Both accelerometers are attached to the car in the same way. To measure longitudinal acceleration, the instrument is mounted level and parallel with the direction the car is moving. This means that the accelerometer is often mounted on the passenger side window. The G-Curve came with suction caps to allow this to be easily done, but the Lev-O-Gage is designed to be fixed in place with double-sided tape and so a suitable bracket should be made and then equipped with suction caps (available from rubber supply shops).

So how does the accelerometer work? When the car accelerates, the ball climbs up one arm of the curved tube, showing how hard the car is accelerating. To convert the degrees reading of the clinometer to g readings, simply use a scientific calculator to find the tangent ("tan") of the number of degrees indicated. This means that if the car is accelerating hard enough to move the ball to the 20 degree marking, the acceleration is about 0.36g (tan 20 = 0.3639). However, note that converting the degree readings into g's isn't necessary for most testing, where you are only trying to see changes rather than measure absolute values.

Mechanical accelerometers must be accurately levelled before performance measurement can take place. The accelerometer will clearly show any road gradient and so levelling is best done on a flat road. If the road is not level (and most aren't), the instrument needs to be adjusted so that the error when the car is parked facing in opposite directions on the same spot is of the same amount but in different directions. For example, the clinometer might show +2 degrees with the car facing in one direction and -2 degrees with it facing in the other. If the road was level, the instrument would therefore show 0 degrees. Using a mechanical accelerometer requires a sharp-eyed assistant armed with a paper and pencil to record the data.

So how do you go about the test procedure? A gear is selected and the car driven at as low a speed as is possible in that gear. After warning your assistant that you are about to start the run, quickly push the accelerator to the floor. Every 1000 rpm (or on low revving engines, every 500 rpm) yell "now!". Each time you yell, your assistant records the accelerometer reading. In many cars the acceleration will be too quick for the assistant to keep up, so on the first run do for example 2000, 4000 and 6000 rpm, and in the second run do 3000, 5000 and 7000 rpm.

Example

So how does all this look? About a decade ago I modified a 5 cylinder turbo Audi S4. I designed and made a new pneumatic boost control (see The Audi's DIY Boost Control - Part 1) and assessed the performance gains by directly measuring on-road acceleration. Here’s the result.

The measured acceleration before and after the fitting of the new boost control shows clearly how much harder the car came on boost. At 4000 rpm the acceleration in second gear was increased by massive 17 per cent, showing an enormous lift in mid-range torque. Amazingly, this was with identical before/after maximum boost - the boost wasn't "turned up" at all! Instead, the new system brought boost on harder and faster than the electronic factory control.

So in this example you can see that full throttle acceleration increased in the mid-range but was unchanged elsewhere.

The Skoda

The Skoda Roomster modifications provide the best examples of this type of approach. The Skoda runs the generic 1.9 turbo diesel PD engine that’s been fitted to lots of Volkswagen, Seat and Skoda products over the years.

When researching the modifications that I wanted to perform, two points came out frequently. Let’s take them one by one.

The first widely held belief was that the standard engine intake was restrictive and that an immediate performance gain could be made by fitting the Seat PD160 intake (pictured here on right).

Measurements of pressure drop showed that the intake was indeed restrictive so a huge new intake was made (the PD160 one cost too much for my taste).

As this before/after graph shows, the new intake massively dropped the measured pressure drop (ie the flow restriction).

So how was performance? Was there that better outcome that everyone said they’d felt with a new intake? Well, as this graph of on-road measured acceleration shows, no there was not!

This is a really important graph to examine.

It shows that peak tractive effort has decreased – and by about 7 per cent! There is a little more response at low revs (the turbo comes on boost a little quicker) but the overall there is reduced performance.

So, if all that was being done was a new intake, the mod would make the car slower, not faster. (Why? Probably because the EDC15 diesel management system isn’t coping with the changes in airflow.)

The other widely held belief is that the rear muffler should be removed. I tried the car without any rear muffler but it was too noisy for me, so instead I fitted a straight-through resonator (pinched from a supercharged Jaguar).

So what happened to actual on-road performance? Again, as this graph shows, if anything it went backwards! Even taking into account some variations in reading the accelerometer, you can safely say that the muffler change did nothing at all for performance.

So to make the point again, those people with the TDI 1.9 VAG engine who believe that a new intake and exhaust will make their cars faster are simply wrong. Do these mods and the car is slower. Despite doing numerous searches and reading everything I can find on the web relating to this engine (and that’s a lot), I have never seen this stated.

The final step in this series of mods (later more modifications were done) was to have the engine management reflashed. If in fact the better flows were causing the engine management system to be unhappy, then a reflash would solve that – and the free-flow benefits of the intake and exhaust mods could then be realised.

Here it can be seen that after the intake, muffler and reflash (orange line) on-road acceleration is clearly improved – and pretty well through the whole rev range. (The only exception is at around 1200 rpm, where in fact on the dyno fuel was pulled out to reduce the rate of boost increase that was leading to a dyno boost spike.)

Showing just the ‘before’ and ‘after’ curves more clearly shows the overall gains. These are clear and unambiguous… another way of thinking about this is that if you cannot easily measure an acceleration gain on the road, the mods are not being real-world effective.

Dyno Comparisons

Let’s compare the improvement found in on-road acceleration with the change in dyno-measured tractive effort.  Dyno runs were made before and after the reflash (ie when the intake and muffler had already been done). This shows those same conditions for on-road measurement and dyno tractive effort results.

Before / After Reflash Measurements

	Road acceleration gain/loss (%)	Dyno tractive effort gain/loss (%)
1000	-  3	+ 8
1500	+ 5	+ 3
2000	+ 15	+17
2500	+ 19	+12
3000	+ 20	+17
3500	+ 22	+17
4000	+26	+21
4500	+91	+33

The relationship is reasonably close except at the bottom and top of the rev range. At around 1000 rpm the on-road driving experience suggests that the dyno is ‘more correct’ – the car has better off-idle torque. At the top end, where the on-road results are vastly better than the dyno results, I’d back the on-road results. In driving, the top-end performance is in an utterly different class to previously, and on the dyno the intercooler may well have started getting hot at the end of the power run, so limiting the measured gains.

In short, the measured change in on-road acceleration is typically a pretty good guide to the change in tractive effort as measured on the dyno. For all practical purposes, of course the change in on-road acceleration is the more relevant measurement!

Conclusion

Once you get used to the technique, whipping out the accelerometer, grabbing an assistant and hitting the road for some testing becomes second nature. It costs nothing, can be done as many times as you like, and allows you to test both the magnitude of changes and also where in the rev range the changes occur.

It’s ideal for measuring the performance of different boost controls, intercoolers, turbo swaps, the result of changing cam or cam timing – and even of altering the mass of rotating items like wheels and flywheels.