| For reference, I'm not being argumentative for the sake of such - I'm doing my best to sort out your conclusions. You make perfectly valid points, then you overplay your hand (so to speak) and jump to fallacious conclusions. I'll take a set of MS manifolds to the flowbench and test them. Guess what? The flow values I will see are going to be greater than the values provided by the flowtest that was performer earlier. Then, take a set of stock manifolds and perform the same test. Compare that data to the data collected on the MS manifolds. NOW you will have a TRUE comparison that was not obtainable by the previous testing method. In addition, now you have two sets of data that you can compare to any other manifold flow test (given that it isn't flown through a head with 'x' valve opening). A straight up manifold only test would certainly be a great test, and when you're talking single components, it probably has the highest degree of merit. Up until this point, the standard test wasn't to use just the manifold, so everyone followed suit so that test data could be as "apples to apples" as possible. I dont know what your inclinations are in this argument, other than the most obvious being your argumentative nature (which I enjoy ~90% of the time :), but in this one, I figure your argument is based solely on face rather than true, objective comparison. Trying to sort things out, not trying to be argumentative. Although I agree with you that the data provided in the non-standard form of testing they performed does have merit, it was very much likely the worse case of non-apples to apples comparisons that could have been performed given the simplicity of performing the test WITHOUT the cylinder heads as a factor. It is as if they went out of their way to make the comparison as complicated as possible. Nobody purposefully made the test overly complex for the sake of complexity. The first tests were done with the heads, the rest followed suit. The data has merit for partial system flow rates (using standard head geometry). My biggest question is why did they perform their test by flowing through the cylinder head rather than flowtesting the manifold all by itself? Assuming the head flow dynamics are a constant, such a test would give a more accurate partial system flow metric than manifolds alone. There are only two possible conclusions, of which neither are good and only one is the most probable: 1) They didn't realize the effect the cylinder would have on their test data; that being, the fact that the valve/port/runner is going to introduce significantly more of an effect into the flow measurements as compared to the effects of the different manifold designs. Agreed it has an important effect on flow dynamics, that being said this should not be an issue if the heads geometries are similar enough to end up in the margin of error percentile. 2) They were aware of these effects, but performed the test anyway. It's elementary that the heads would effect the final data, of course they knew it would be a factor. My above point still covers this. 3) It was simpler to attach the head to the flowbench as it would require machining of unique adapter plates to perform the test on the manifold itself. Consider this example as well: Say you want to flowtest an intercooler. Would you connect just the single component of the intercooler itself to the flowbench or would you attach the 22-some-odd feet of intake piping to it as well? Both metrics would have merit, there's no reason not to do either. Arguably the partial-system test has more merit than the single component test. As you well know dynamic systems are more than a sum of their parts - the closer to a full assembly you can test, the better imho. The results they provided are specific to that cylinder head and the MS manifolds. It is NOT a representation of what the manifolds themselves flow and is not reproducable. The test was flawed, the data is incorrect, and the data is incomplete. You're back in the deep end. If the head geometry is stock, and there are not enough geometric anamolies to take the data out of the margin of error, then the test is not flawed and it is certainly reproducable. The data itself is merely the results from the flowbench - if it's properly calibrated it's not incorrect. Finally the data is arguably incomplete, they only have one data point when the entire manifold should be tested. Hopefully a more comprehensive test can be conducted (like the one you propose) and we can have a better comparison metric. That being said, a single datapoint is a better comparison metric than no datapoint at all.
Recursively Yours,
Kenny...
PETZ Member #5
You guys rock socks. (Click for pie-chart)
 |
