Originally Posted by bruce.augenstein@comcast.
I guess that's true. If you're satisfied with hideous, systematically incorrect results, then I suppose I should be, as well. My systematically documented results over the years are then obviously without meaning.
Talk about hyperbole, pot meet kettle...
Bruce, ever heard of apples to apples? That is simply what this is about. What you (or your buddy as it sounds like) did and how you did it, although using basically the same method (physics/numerics) is very different that what I am doing and how I am doing it. Here is why:
You are using a very controlled and small set of data points. You are controlling the set of cars, the driver, the tires, the technique, the track, as well as likely controlling more variables than that. Then, as I understand it, you tweak the simulation to match the measured results. You even admit some subjective factors such as quality of hook up and traction, etc. CarTest by the way offers no such qualitative inputs, all of them have a close or directly quanifiable meaning/interpretation and tests are then possible to determine these input parameters. How does that quality of hookup adjustment work exactly? Is it just a lookup table that adds or subtracts some number of tenths to the simulation based on asking the driver was it "excellent", "good", "fair" or "poor". Or maybe you have an "exotic" curve fit for corrections for these inputs that use langauge to translate to numbers. Bruce this is called HEURISTICS, not simulation
. Once you have enough repeatable runs this type of "tweaking" of a simulation (be it with quantitative or qualitative factors) is what most in the business call at best simply curve fitting, or at worst cheating.
How good would your infamous method work if all of a sudden you changed the trans, tires, driver, removed 300lb from the vehicle and hopped it up with 15% more power and a 1k rpm higher redline ON THE FIRST RUN? This is a KEY question as the the capability of the tool and of the operator of such a tool.
So on top of how you did this, which I addressed above, there is also what you did. I still contend that it is not possible to match within hundredths of a second, time to distance nor time to speeds ALL the way along the strip. I think you are simply losing sight of reality and orders of magnitude. I'd be suprised if you got within tenths the entire way down, let alone TEN TIMES BETTER than that. Do you realize how much a simple factor like a transmission loss affects 1/4 mi times? A mere 1% error in this crucial input parameter (from either a bad estimate or test with not enough fidelity) results in about FIVE hundredths difference in the 1/4 mi time (in a MT car in something like the E92 M3). Perhaps you can explain how you obtained your transmission losses to less than a 0.2%
. What about friction as a function of temperature, tire growth with rpm, or any of the dozen inputs I have shown that CarTest uses: YOU WOULD HAVE TO KNOW THESE FACTORS MORE PRECISELY THAN YOU CAN ACTUALLY MEASURE MOST OF THEM. Christ we did not even mention rounding. To be withing a hundredths (or was it only hundredths) you really have to be within five thousandths. Starting to see the point ... yet...? Doubtful, but many others here will.
Now back again to what I am doing. Your serious misunderstaning of my last reply, which you dismiss as simply "defensive hyperbole", again shows just how ignorant and confused you are about this whole simulation thing. What I am doing is predicting tests that are known to vary widely from a variety of uncontrolled factors including, driver, tires, track, temperature, age of car, car to car variations, etc. I am using purely analytical techniques and some combination of tuning or better yet using an enveloping process to determine drive train loss parameters. What is the bloody spread of 0-60 times or 1/4 mi times reported by a group of magazines for a typical car?
Multiple tenths of seconds is the answer. So what is THE real time? THERE IS NO REAL TIME nor "best" time or whatever you want to call it, all of the times are valid data points and predicting somewhere in the middle is again the hallmark of a REASONABLE and accurate simulation
Do you see yet
why you are making an apples to oranges comparison?
Why did I bring up old car vs. new car results? Because you brought it up first as evidence that my simulations were so incapable of accuracy. Again, measure the hp, input it to the sim, voila you can easily show the difference between the two cars. Better yet look at some dyno results for a similar car, new vs. well broken in and then you could probably establish a reasonable hp gain.
With your harsh criticisms of my lack of ability, my hideous and systematically (do you even know what that means...) incorrect results, I think there is only one way to settle this. Dust of that old TRS-80
, all of those valuable old time slips and track down your buddy (who probably did most of the work) and BRING THE EVIDENCE
. It is so easy for you to sit in front of your computer now, reference the magic of the glory days and how terrible of a job someone else is doing. First you can prove your claim on the hundredths of a second issue. Then you can show me how you can better hit a constantly variable and moving target such as results from magazines. You can't really do it, almost by definition. I am definitely not one to say I am a leading expert on this topic, I also do not claim my results are perfect and I readily admit that a lot of the input parameters need to be known more accurately. But again given what we all have access to, these simulations are in my opinion, and in the opinion of quite a few folks here, reasonable and accurate.
Really, Bruce, put up or shut up.
Finally, I do think I know our differences about the importance of hp, tq, gearing and redline as well and how they relate to acceleration. Folks like lucid and myself take a sort of reductionist view of the systems of a vehicle where everything can change independently. You look at vehicles as they actually exist where they are mostly already optimized in that the transmission, redline and torque curve all are chosed to maximize acceleration. If you took a vehicle so optimized and drastically altered its gearing or redline of course it's performance would change in a correspondingly predictable way. So your POV is when a vehicle is totally optimized for its engine, redline and gearing do not predict how fast that car is in the 1/4 (right?). My POV is that power to weight is indeed the critical factor but that all systems must be matched carefully to an engines characterisitics. Then of course to the extent that you can adjust and engines tq/hp/redline independently, if you made such changes the gearing would then be critical to get right. Sound about right?