Generating Values from a Graph Plot

I looked at your data a little bit.

I'm no expert on combustion, but, off the top of my head, I would hypothesize the curve should have kind of an S shape to it. The simplest equation I thought of that readily generates an S shape is a cubic, in this case writing X (CA) as a function of Y (%comb) (x=a+by+cy^2+dy^3).

LINEST does this kind of thing very readily. =LINEST(CA's,%comb^{1,2,3}) This equation seemed to fit the data reasonably well, but not as good as one might like. I expect there are better possible equations out there (using exponentials, maybe??), but I didn't take the time to look into them.

One potential issue with this equation is that you may not know how to get %comb given CA. Solver or Goal Seek can do it, but these are a little tricky to automate. If you know numerical methods and VBA programming, it would be real easy to write a UDF that solves for %comb given CA using the Newton Raphson method.

As I said, the best first step with curve fitting is deciding on the equation to be used. The cubic equation I've shown here as an example might be adequate for your needs. If not, the concept readily extends to any equation that can be expressed in the form g(y)=a1*f1(x)+a2*f2(x)+a3*f3(x)+...

I put your data into a chart and put a simple cubic trendline on it. It comes up as y = -0.0033x3 + 0.1016x2 + 2.5892x + 14.286 with R^2=1.

Just fill the "A" column with the values you want there and put "=(-0.0033*A1^3)+(0.1016*A1^2+2.5892*A1+14.286) in B1. Then fill the "B" column with that equation. It will generate a nice cubic curve. If it were me, I would be very, very hesitant to extrapolate that much data from these few points, but as long as it's clearly labeled as a projection or some such, I guess it will work.

Thanks for all your help

MrShorty, your right, the true graph should be an "S" shape in theory but in measured results, it never completely fits the theory and for it to be complete a 100% burn would need to be measured which is never possible, also a combustion start point needs to be measured which is again very difficult to do.

I have ended up going back to what i totally overlooked in the first place and used a Trend line and it seems to working ok for me now.

Thanks again,

Chris