tenblade2001

I was thinking about performing a study. I and others, I'm sure, would like some real data on mods and relativer HP gain. I could create a database with Microsoft XL and enter in data points like yr model, cai or no cai, boost and no boost, exhaust (type, etc), dyno numbers, and type of dyno. From there, one could use a log rank regression to see which mods made a statistically significant impact on HP and which ones didn't. This won't be quick. It will take some time but in the end, it could have some really interesting results. What do guys think? I would need dyno info. If I get busy, some help with entering the data would be nice. But, once it is in XL, there is soooo much we could do with it and potentially answer many questions.
Thanks,
Tenblade

hatchback

I think it's a great idea. Remember that dyno measurements are relative to the specific dyno, not just the dyno brand. So at the very least you need to include a categorical variable for the specific dyno and try to get a large number of results for the same dyno on the same day.

ReDevil

I agee that this is a great idea. I just put on a cai w/ nothing else done to the car. I would dyno it but theres nowhere around where I live to do that. If you need help w/ data entry let me know

SECRET AP1

The problem is that not all engines are created equal. Dyno three brand new S2000s and you're bound to see at least a 5hp variance. An intake that works well with a stock setup may not do so well once you throw on a certain exhaust. But that exhaust together with the stock setup or even a different intake may show gains. Then throw the whole setup on another car and the gains change.

There's too many variables which is why this type of study is not usually not effective.

cdelena

And not all dyno's, operators, and test conditions are equal. I have seen a 4% variation in my car without change. I don't think you can prove anything.

thereisnospoon

i think you can prove a relative difference. if you require a baseline run and then a post-mod run - you can record percent change. If you get enough entries that averaged out show a 5HP intake increase or an 8HP exhaust increase for specific models - I think it woudl be meaningful. You can probably lump the 4ig, AEM, and Injen together because they are almost identical - but exhausts would have to be separate items.

go for it!

beroznikmal

I'm sure if everyone was honest and true about their mods.this could be very useful information.personally, I have changed around my mods setup various times....
I would include the following fields:

1.1 MY
1.2 Existing mods
1. mod tested
2. baseline dyno
3. after install dyno
4. type of dyno
5. what gear it was dynoed
6. Ambient temperature
7. A/f ratio at 1K, 2K, 3K, 4K,5K,6K,7K,8 K,9K
8. Type of tuning mechanism (VAFC, MUGEN ECU, AEM,ETC.)
9. Any timing tuning?????

IMO the variance between variables like various engines at stock give diff. Numbers can be eliminated/ fine tuned with enough accumulated data.

Beroz

tenblade2001

these are all good points. I plan to take the year s2000 into account as well as the dyno. By doing a log regression, you should be able to see which variable impacted dyno results i.e. HP>210 and which ones didn't.

SECRET AP1

the problem isn't so much a variance due to randomness, but the number of possible combinations of mods. an intake that performs well on a stock car combined with an exhaust that performs well on a stock car may not perform well together. but, im all for getting data together and developing a possible pattern for power.

tenblade2001

Right API, that is why you can use a log regression. That will allow me to take many variables into account and see which ones were significant i.e a VAFC may prove significiant, but only on s2000 before 2002. I won't include boosted cars.