Page 3 of 3
(continued...) The update command then rebuilds the model parametrically to the calculated value for each controlling dimension.
At this point the modellers interference detection tool can be launched to identify any interference volumes.
Although the dimension values update to reflect the selected tolerance build-up, the full tolerance range is still annotated without the nominal dimension value.
It would be excellent if dimensions were at least colour coded to indicate being at max or min condition.
The post processor window also indicates the name of each toleranced dimension and their percentage contribution to the total variance.
Selecting a dimension highlights it on the model enabling excellent analysis of where tighter tolerances could make the most significant improvement or indeed an alternative dimensioning scheme may be better or alternatively others, that have a low influence in the response function, may be increased in order to reduce the manufacturing costs.
The recompute button enables the effect on the percentage contribution of changing any tolerances to be assessed.
A graph also plots the results from the analysis.
Sliders above and below the graph enable the max and min percentages to be set to reflect target or actual capability.
Results can be output together with the data table to an HTML report for all or selected response functions and analyses.
Unfortunately this does not include a descriptor of the response function or the relevant dimension and percentage contribution information.
TASys is not an instant cure-all for tolerance analysis; careful consideration is still needed in creating the tolerance scheme and the use of the analysis results.
It does however offer an excellent way of visualising and assessing complex tolerance chains.
It behaved unexpectedly on a few occasions running in SolidWorks, particularly with inconsistent geometric tolerances.
Also if TASys detects anything having changed in the model it deletes results for new ones to be calculated.
Since a changed may only have been a part colour change for instance it is somewhat inelegant to instantly lose access to previous results.
Since the process can be processor and time intensive it would be frustrating to need to recalculate essentially identical results.
A sensible future enhancement will be the better management of multiple schema, perhaps with greater integration with model configurations.
Despite these minor criticisms, it is certainly valuable to be able to conduct statistical analysis of tolerances directly on the model data and particularly to associatively rebuild the model to be able to visualise the probable consequences.