Your approach is good but not your math. The spread for a target of 15C could be ± 2 points and that would be 13-17C. This would be a good way to set the process limits if you had determined that the optimum intensity (and the resulting depth of compression) was obtained at 15C.
To insure process consistency you will come back on a frequent basis (your own plan which could be daily or weekly depending upon circumstances) and discover if a strip arc height can repeat ± 1.5C compared to the 15C. This assumes that the exposure time for the consistency check is the time T1 from the saturation curve.
Keep in mind that sometimes it is not possible to run a consistency check at exactly T1 time. This might be due to using the number of table rotations or other methods instead of adjustable exposure time. In those cases you would perform the 4 strip exposure for the saturation curve and using your example you would obtain an intensity of 15C. But, if your machine used a rotating table you might notice that T1 occurred at 8 1/2 revolutions. You cannot run a consistency check at 8 1/2 revolutions but you can run 9 revolutions.
Right after the saturation curve is generated expose one strip for 9 revolutions. Its arc height, when superimposed onto the saturation curve, should fall closely on to the curve. That arc height now becomes a "Target" arc height for your periodic checks. The value at 9 revolutions, for example might be 16.0C. Therefore your consistency band would require arc heights in between 14.5 - 17.5.
In a well maintained machine it should be easy to get well withing the consistency band. Notice that the band for consistency checks uses the SAE J443 criteria of ±1.5, not the ±2 for your tolerance band. The consistency check is testing the repeat ability of the process. There will be natural variation due to all of the parameters involved. Demonstrating that you can repeat the process at a given exposure time within ±1.5 indicates that the process is in control.
