Model concentricity of gears in gearbox to simulate variance in sensor ride height

I have a stepper controlled gearbox, on which, a linear encoder read head rides. If the concentricity of specific gears vary, it's possible that the distance between the encoder read head and the encoder magnetic strip will vary due to increased distance between the gearbox and rack (due to difference in gear diameter from concentricity issues). Variation in read head height can cause an increase in encoder sub-divisional error.

I'd like to model the gear box with different concentricity parameters to predict variation in encoder sensor error. I haven't been able to locate any examples which show how I might approach this.