[quote=“pyrlix”]
abload.de/img/flatnesswnu13.png
All of these 3 would be around 0.9 to 1.0 Flatness… I can see peak torque 1800rpm (for example) and engine is obviously rubbish, sure. Just the fact that it revs to 5krpm and is surely not for a Mid-Range or Sportscar… thought about that? The flatness only describes how the torque curve will progress - a flat one is easier to drive than a steep (both up and down) one…
What you are talking is plainly not logical at all - when you KNOW how flat the curve runs approx between 2000rpm and peak torque, you KNOW how the torque curve looks like… and what usual customer looks at dyno graphs in a showroom and says “Oh smokes! This car has the best Dyno ever, i need this one.”[/quote]
Your own example is a pretty good illustration of why (Nm@2k/Nm@peak) is inadequate for describing the “flatness” of the torque curve. Three torque curves with quite different levels of actual flatness are, by that metric, quite similar and the least actually flat of the three is the most “flat” of the three. It works fine when comparing engines with peak torques at ~4-6k rpm because in such a case it is describing the general trend of a large section of the total torque curve, but it falls short when peak torques are close to 2k rpm (as in your example, or in the 1975 “entry level” category of the first competition) because in such a case it only describes a very small section of the total torque curve. In such a situation, “torque flatness” is at best useless and at worst misleading.
I do not agree with the call for posting screenshots, though, as I think it is overkill, would create too much additional workload when judging, and would involve too much qualitative assessment. So I suggest an alternative solution, an improved equation that makes use of existing data. The peak power/rpm can be used to create a third data point in the torque curve without requiring any additional inputs by the contestants. That third data point can then be used to create an average of (Nm@2k/Nm@Nm.peak) and (Nm@kW.peak/Nm@Nm.peak) weighted by how much of the torque curve each ratio describes. In spreadsheet-ready format, I propose this:
=(C10/C8ABS(C9-2000)+C69549/C7/C8*(C7-C9))/((C7-C9)+ABS(C9-2000))
Using this equation, the “torque flatness” of your example engines would be (approximately): 1.) 0.79 2.) 0.84 3.) 0.86. Which is not as informative as a dyno chart, obviously, but it’s a heck of a lot more relevant/representative than 1.) 0.99 2.) 0.98 3.) 0.94. It is also much better at evaluating turbocharged engines which, under the current equation, are at a huge disadvantage in “flatness” unless they hit peak boost at or before 2k rpm, which in this game is very difficult to do without crippling the power output at higher rpms.
I think the broadness problem is specifically about the “roadster” category. Mid-Range and Entry-Level are great, they are specific market segments, but roadster is really more of a type and some of the examples listed, while similar in form, were from very different market segments; The SLK230 cost literally twice as much as the Miata in the US. It would be akin to having a category of “Sedan/Saloon” and listing a Corolla and a 5-Series BMW as examples. And the problem is that those different market segments have different requirements for engines; The more expensive/luxurious car needs to have a more powerful engine and can afford to have a higher price tag as a result, but that would be unnecessary and overly expensive in something like a Miata. Calling the category something like “Affordable Roadster” would have been much more in line with the other categories so far.
! But it went so well, we have some usable scores now and you surely to know them 
it seems for the '92 roadster, people voted for the mx5 engines and not the SLK that i designed. Can’t be helped:p
