A dyno is indeed a very useful diagnostic tool but its only a good reference tool if the operator is honest. As I said above I know the guy I used years ago massaged the figures to suit the customers wallet and I trusted him to dyno my engine.
My old race engine produced about 175 bhp when it was in its finest fettle. The local guy I used regularly gave me a figure of 115 bhp at the wheels. Most people would have gone away suicidal with that figure but since I was winning and setting records I knew the engine was as good as any out there. At the shoot out I mentioned above it gave 145 bhp at the wheels.
An engine builder I had been acquainted with for probably 15 years (ex. Blydenstein, Hart etc.) offered me some free development work on a 3D mapped ignition, I pay for the parts (£450 ish) and he paid for the development and fabrication of fitting parts and did the mapping on a friends rolling road, ultimately he was adding the developed parts to his products catalogue and making a nice profit on each sale which he did quite successfully (and is probably still doing). Before the work they did a power run and it produced 155 bhp, after the work 156 bhp (both at the wheels). But that was only part of the story, with the 3D map the mid range power and response was noticeably increased (but not measured on an ancient water braked dyno) which significantly improved lap times.
A couple of years later I need the set up checking after some head work but the rolling road where the work was done had gone bust. Found another local guy who had the software for the ECU and on his rolling road the car gave 170 bhp at the wheels, total nonsense.
So the same engine gave between 115 bhp and 170 bhp at the wheels on various sets of rollers and in truth there should have been no more than a couple of bhp difference at peak power.
My conclusion has always been the figures that the operator give you have to be taken with a pinch of salt. The method I use now of the flat out run in 3rd up a 1 in 8 from 30 mph to the red line is recorded by my data logger and its easy to overlay runs onto each other and see if any changes have made improvements. This not only applies to full power it applies right across the rev band. Even in a race car you spend more time below the red line than on it and on twisty UK circuits very little time on full throttle.
Driveability matters far more than a magic figure to boast about in the pub.
