Use of epoxy for levelling

[Jonathan]

I'm happy to go through how I did my calculations if asked, but for the moment I'll skip directly to my conclusions.

I think your theory is worth exploring and (you guessed it) I'd like to confirm that your measurement method is sound since we don't want to be victims of correlation not implying causation.

Have you measured the shrinkage of your resin directly, since clearly if the resin shrinkage is far from 10% the rest of your analysis is potentially invalid (though not necessarily invalid, since it could still exist as a small effect masked by something else)? If you've not measured or got the data elsewhere, then perhaps pour some into a fairly thin tube, mark the level and when it sets measure the change, or think of a better method as I expect there will be one.

The west system resin datasheet says it 'does not shrink after curing', but does that mean it doesn't shrink during curing? I guess not else they'd surely specify. Adding thinners does increase the shrinkage, so one should be careful when selecting the resin to ensure it doesn't contain them as that option might seem attractive to reduce the viscosity.

Please could you elaborate on your measurement method to get the "height error from horizontal plane"? Did you add the feeler gauge under one end of the level until the level read zero, then note down the thickness of the gauge, or have I completely missed the point? If it's the former then the measurement is an angle from the horizontal plane (or strictly speaking the Earth's center of mass, but lets not worry about the earth curvature error), not height deviation without more careful interpretation of the readings.

Jonathan's measured error curves are not quite in agreement with this principle, but his measurement technique is dependent both on rail height variation and rail twist, and once we are getting down to small numbers the measurement technique becomes important.

I could have indicated from the surface plate to the resin, or from the rail to surface plate at two distances to separate the readings of linear and angular error ... but time was not on my side.

However, it is interesting (but possibly coincidental) that his initial error was reduced from about +-0.3mm to about +-0.03mm. That's the kind of reduction that my thinking would have suggested although it doesn't prove anything due to, as I say, the different measurement technique.

I'm not sure where you got the +-0.3mm from, but from the graph I posted in the thread it was more like +-0.5mm, then 0.3mm after the failed compensation method using a machined strip. However the graph I posted is a combination of the height and twist errors, so we can't actually conclude from this what the height error was.

We could gain confidence in your theory about the resin shrinkage by measuring if the height error follows the original profile, just with a smaller magnitude. My readings do not seem to follow the original profile, but as mentioned already the readings are a combination of the rotational and linear error, so we can't draw any conclusion from this.

Maybe I should try a simple test using my surface plate - create three channels, one directly on the surface plate with the surface plate horizontal (the 'control' experiment) and one with the surface plate tilted so the resin is significantly deeper at one end. Measure the linearity of both and see if there's a discernible difference between the samples.