Many of us are doing some kind of changes to your vehicles have tried to find best ways which is the best setup of what you are testing. Usually the problem is that the changes/effects are so minimal that it is really difficult to be measured on open road as wind, others cars, weather in general, cars warmup times and how you drive the tests effect the results more than the results.
How to Minimize error
Choose a time when there is no wind
Preferably whole day sunny or cloudy as tarmac temp effect RR
If you dont have long level roads you for constant speed testing, I recommend using roads which are in the middle of forests. This minimizes wind effect and usually there is less cars to mess your testing. Also hilly road means you testing more speeds than just one constant speed meaning you get both aero and RR tested at the same place with just one test per change
warm your car fully before you start tests. I recommend 100km of driving to get everything warm. You will learn whats needed by testing.
Try to finish your testing as fast as possible so results are comparable
Biggest error in my tests have always bee pedal movement. That`s why I made custom test program for my ecu. When driven easily car works normally. But when I want to begin my efficiency test I have selected start speed where I floor the gas pedal --> This is programmed to give contant fuel injection levels throughout the whole rev range--> This means in uphills speed gets slower, In downhills higher. Then you just measure the time it took to drive your test distance. I normally use 3-5km distances. From the times you can calculate AVG speeds--> From AVG speeds you can calculate how much more power you are making or how much less drag you are producing depending what you are testing.
Do A-B-A testing to verify that you get constant results. If I dont get reliable difference and I feel the other setup is better I just double or triple the test distance. You can see from your memos if some setup works better on low speeds vs high speeds as during testing you need to select 4-6 measuring points where you measure the speeds you have at certain points of the route. Specially in engine tuning some setups works really well uphill parts and some at higher speeds. Luckily in most ecu program changes the better setup works best on both low and higher revs. So best setup is easier to find.
You can see that my #audi #a4 #1.9TDI afr gen2 #ecutuning setup is 14.5% better than cars stock program between 75-103 km/h speeds. Top speed in this test was 5 km/h higher that with stock cars engine control program which is a big difference as amount of fuel injected stayed the same during whole testing.
Results
With this kind of testing method I have been able to find even smaller changes effect reliability to my car. Previous method was to test 400km road test with each modification and that is not easy, takes time and had many error possibilities due to other traffic users and weather. Biggest improvement has been contant injection quantity values. If you just build constant throttle your and your rpm change you will have diffenrent settings that engine will use. which is not good.
With custom ecu program you can test even on not so flat routes which makes testing easier. My flat tests grounds in finland are minimal and usually there is some traffic which is not good for reliable testing.