Right about here, it might be worthwhile to note something. Let's say we have two engines of approximately equal power. That is to say their respective torque and horsepower curves are quite similar. It happens with some frequency. However, one exhibits a BSFC curve with values lower than the other engine. It turns out that the one with the lower BSFC curve will accelerate more quickly than the other, once on the track—all else being equal. It will also be more responsive to changes affecting combustion efficiency as well and will very likely require less total spark timing to perform at its best. On-track fuel economy will also be superior. But we digress.
Overall, there are some truths and there seems to have been a bit if misunderstanding about what BSFC data are actually indicating. For example, if an engine is experiencing an inordinate (maybe even some) combustion contamination from residue (exhaust gas) left in the combustion space during succeeding combustion events, BSFC data generally increase numerically.
In such cases, exhaust gas temperatures (EGT) tends to trend downward. This is often accompanied by the necessity for additional fuel and more spark timing, in an attempt to resolve the problem. Conditions like this simply mean the contamination problem requires correcting.
One approach I observed a number of years ago and found to be helpful is the following method of using BSFC as a tool. Since an engine tends to be the most combustion efficient at or near peak torque, it will save wear and tear on an expensive engine (or even one not as expensive) to perform initial spark and fuel calibration tests at peak torque. There is a reason why BSFC tends to be numerically the lowest at this point, but we'll get to that in a moment.
By using this method, it's possible to minimize the engine's test burden but make possible the pinpointing of the best (lowest) BSFC by adjusting spark and fuel…separately. First, find the best spark setting. Then you can adjust fuel flow until the lowest BSFC is determined without an attending loss in power. Simple enough.
Ideally, you'd like to create a "flat" BSFC curve, but that's not entirely possible. What you can do is work toward establishing that condition, using the value of BSFC at peak torque as your target. However, a couple of conditions play into any attempt at "flattening" a BSFC curve. Among them is the fact that at engine speeds below peak torque, there's an increased amount of time for heat losses to the cooling system and related parts and passages. And especially if the engine is using a carburetor, air/fuel charge quality tends suffer from less efficient atomization (mixing) at these engine speeds than higher rpm.
Beyond peak torque, there is the issue of mechanical separation of air and fuel as signified by a corresponding rise in BSFC values. In other words, combustion efficiency tends to deteriorate, accordingly. Plus, in the higher rpm ranges, while there is an increase in the amount of otherwise useable heat from combustion, high rpm shortens the time available to deliver that heat, resulting in a loss in power.
So there you have it. Given contemporary technology and equipment availble to qualify and quantify the combustion process, using BSFC analysis has become somewhat of a "poor man's" method of tuning or modifying a racing engine, but it works and is far more economical than some of the other methods. Just keep in mind that it can also be useful when evaluating parts of modifications that relate the combustion process. Although that was previously mentioned, the technique can be a valuable approach to determining the best package of engine components and level of tune before you head to the track.