Fuel Economy, Hypermiling, EcoModding News and Forum - EcoModder.com - View Single Post

JulianEdgar · 05-16-2020, 04:23 AM

In all on-road testing (other than long-term mileage) it's very important that you're absolutely rigorous in the testing.

For example, in my throttle stop testing I:

Use the same section of straight road each time
Pass a marker at the entrance to the straight at exactly the same speed with exactly the same run-up distance each time
Hold a constant throttle position
Measure the maximum speed at the same point each time
Repeat the run

I then make a quick aerodynamic change (eg to the angle of the Edgarwit) and repeat the above process.

Before I drive to my test location, I make a list of the tests I am going to do, and draw up a table so the results can be easily written-in as each test is done. That lists of tests includes returning the car to its standard form several times during the test session. Obviously all of those 'standard form' tests should give the same results. With throttle stop testing, I usually also do a windows up / windows down comparison and check these results match what I've previously recorded on days of the same weather.

It's quite easy to lose concentration when doing dozens of tests. To avoid that, I say out aloud in the car what I need to do just as each run starts. So I might say, "Fifth gear, 100 kays entrance, top-right speed." Or, "Reset pressure chamber, 80 kays, 4th gear, early reading."

When trying new modifications, I have also found it much more preferable to make major changes. So if trialling an undertray, trial a big one. If trialling a rear spoiler, trial a big one. These mock-ups don't need to be anything special - cardboard, plywood, corfboard, tape.

What you are looking for is a major, measurable change. Things might be worse, or things might be better - but they need to have measurably changed.

Once it's established that you are measuring the change you've made, then you can finesse the modification from that point. The alternative (making tiny changes) is that you may end up chasing testing scatter (ie normal variations in tests) rather than the results of your modification. This can be very confusing and waste a lot of time.

I've found it's really easy in testing to let your hopes and desires over-ride good testing approaches. To avoid this, I deliberately don't try to memorise results as I am doing tests - I complete the series of tests and only then look at the results more closely.

It's also really important that, once you've established you are measuring real changes, you believe your tests. I've lost count of the number of times I have done a series of tests and disbelieved the results - only to find through further testing that the original tests were correct. Here's also where so much of the theory and judgements that you read on discussion groups (including this one) can lead you astray. What works on one car may well not work on another, and vice versa. The tech literature is full of examples where the results of the same approach differs from car to car, and when you are testing your own car, you'll soon find that out for yourself. And that's especially the case with pressure testing (both coarse and panel measurement).

For example, in that original underhood intercooler development, where I was improving flow across the intercooler core, my first test found that the air was coming in through the front grille, going through the radiator, passing up through the intercooler, and then flowing out through the forward-facing scoop! I am quite certain that wouldn't have been the judgement of any discussion group...

Finally, I am not wedded to the idea that each of my modifications will work. So, when I test them and I don't think the results are satisfactory, I don't try to rationalise the outcome. I don't say, "The test results were nothing special but I am sure the modification is working." Instead I say, "Damn, it didn't work. Oh well." The likelihood that you will become wedded to a mod is highest when you spend the most time and energy on making it. Therefore, always initially trial the quickest and simplest mock-up you can make. Overall, it saves far more time by making (maybe) three iterations of a successful modification, rather than making an elaborate first version of every modification - including those that turn out not to work.

05-16-2020, 04:23 AM	#3 (permalink)
JulianEdgar Banned Join Date: Nov 2017 Location: Australia Posts: 2,060 Thanks: 107 Thanked 1,605 Times in 1,136 Posts	In all on-road testing (other than long-term mileage) it's very important that you're absolutely rigorous in the testing. For example, in my throttle stop testing I: Use the same section of straight road each time Pass a marker at the entrance to the straight at exactly the same speed with exactly the same run-up distance each time Hold a constant throttle position Measure the maximum speed at the same point each time Repeat the run I then make a quick aerodynamic change (eg to the angle of the Edgarwit) and repeat the above process. Before I drive to my test location, I make a list of the tests I am going to do, and draw up a table so the results can be easily written-in as each test is done. That lists of tests includes returning the car to its standard form several times during the test session. Obviously all of those 'standard form' tests should give the same results. With throttle stop testing, I usually also do a windows up / windows down comparison and check these results match what I've previously recorded on days of the same weather. It's quite easy to lose concentration when doing dozens of tests. To avoid that, I say out aloud in the car what I need to do just as each run starts. So I might say, "Fifth gear, 100 kays entrance, top-right speed." Or, "Reset pressure chamber, 80 kays, 4th gear, early reading." When trying new modifications, I have also found it much more preferable to make major changes. So if trialling an undertray, trial a big one. If trialling a rear spoiler, trial a big one. These mock-ups don't need to be anything special - cardboard, plywood, corfboard, tape. What you are looking for is a major, measurable change. Things might be worse, or things might be better - but they need to have measurably changed. Once it's established that you are measuring the change you've made, then you can finesse the modification from that point. The alternative (making tiny changes) is that you may end up chasing testing scatter (ie normal variations in tests) rather than the results of your modification. This can be very confusing and waste a lot of time. I've found it's really easy in testing to let your hopes and desires over-ride good testing approaches. To avoid this, I deliberately don't try to memorise results as I am doing tests - I complete the series of tests and only then look at the results more closely. It's also really important that, once you've established you are measuring real changes, you believe your tests. I've lost count of the number of times I have done a series of tests and disbelieved the results - only to find through further testing that the original tests were correct. Here's also where so much of the theory and judgements that you read on discussion groups (including this one) can lead you astray. What works on one car may well not work on another, and vice versa. The tech literature is full of examples where the results of the same approach differs from car to car, and when you are testing your own car, you'll soon find that out for yourself. And that's especially the case with pressure testing (both coarse and panel measurement). For example, in that original underhood intercooler development, where I was improving flow across the intercooler core, my first test found that the air was coming in through the front grille, going through the radiator, passing up through the intercooler, and then flowing out through the forward-facing scoop! I am quite certain that wouldn't have been the judgement of any discussion group... Finally, I am not wedded to the idea that each of my modifications will work. So, when I test them and I don't think the results are satisfactory, I don't try to rationalise the outcome. I don't say, "The test results were nothing special but I am sure the modification is working." Instead I say, "Damn, it didn't work. Oh well." The likelihood that you will become wedded to a mod is highest when you spend the most time and energy on making it. Therefore, always initially trial the quickest and simplest mock-up you can make. Overall, it saves far more time by making (maybe) three iterations of a successful modification, rather than making an elaborate first version of every modification - including those that turn out not to work.