Conclusion
A computational model was developed to investigate how variation of surface growth stress around the stem effects results of the splitting test and the mean values of various numbers of strain gauges. By modeling multiple populations with differing stress relationships around the stem surface and comparing the results to previous work, realistic bounds were able to be estimated for the relatedness of surface stress points. The repeatability of the splitting test under rotation was investigated and a similar, although lower result as Chapter 5 was found (95% confidence interval on the difference distribution of \(\pm480\ \mu\epsilon\) ). A lower bound (i.e. it is likely to be less accurate than) on the 95% confidence of a splitting test result predicting the real mean surface strain for similar populations was estimated to be \(\pm\ 281\ \mu\epsilon\). Further, it was concluded that the rapid splitting test provides a similar accuracy in predicting mean surface strain on a sample as using four evenly spaced strain gauges (a difference distribution standard deviation of \(262\ \mu\epsilon\) for the splitting test in the full populations set and \(132\ \mu\epsilon\) for a more realistic limited set).