How Tightly Can Strip Verification Assure an Almen Intensity? Detection Limits, Tolerance Windows, Monitoring with Strips and a Worked Case of Nozzle Wear

An Almen strip verifies intensity, but the limits of what it can show are rarely stated. The ±0.0015 in tolerance of SAE J443 tests that a strip repeats its reference reading, not that the intensity is correct. This paper derives, in closed form and by simulation, the smallest change that strip verification can reveal, how long a fault goes undetected, and what tolerance on intensity that supports. For a process at 0.01265 in A, with four strips and a calibrated lot response, charted strips at the most sensitive exposure detect a 9.3% loss of exposure rate and a 4.4% change of intensity nine times in ten in one check (10.3% and 4.8% with the fitting uncertainty of the reference curve). Without lot calibration the intensity limit is 13%, no better than the tolerance. Under a screening rule introduced here, a justified tolerance lies between the smallest half-band that verification can assure and the largest change that the claim can tolerate. Under current practice ±0.002 in lies below the first (±0.0028 in) and above the second (±0.0017 in), so no tolerance can be both verified and protective. Under recommended practice, on one machine, ±0.0015 in lies inside the window. When the J443 test signals a slow drift, in 63% of runs the process has already passed a limit 0.00132 in away. The documents read give no derivation of any tolerance. At the common nozzle replacement limit of 1/16 in the modelled loss of intensity is 0% to 19%: the J443 test signals after the margin has gone, and a chart on four strips signals before. All results are model values for one process, and no prospective test has been run.