Before we apply the "Andersen" Cone Crusher Estimation method in METSIM, lets first think about what we would use it for.
First, we should check what it can be used for (how good is it) and then we should think about what we should use it for.
Andersen wrote his masters thesis on this, so he gave it some thought over a few years. So the Andersen thesis is what you should read to get a well considered opinion on how good it is - or how much data and work is required to get a useful fit - and also to see the raw data and the scatter plots with the r-squared fit from all the work that was done for the masters thesis (and previous surveys).
First, are we going to use it to specify the installed power for a new crusher? Well I hope not. Researcher's with power estimation algorithms based on different ore through different crushers should be kept some distance from the writing of specifications for equipment supply. Equipment suppliers have a much more pragmatic approach to installed power, such as being able to start the machine under less than ideal conditions, and short term peak power in the typical abuse situations that allow operations to continue through recoverable "process excursions".
Can we use it in design to impress the client with our knowldege about these things and as a basis for checking a vendor's offer? Yes, that will probably work. Any time an engineer shows a calculated approach rather than "opinioneering" that can only improve client / engineer relations on an EPCM contract.
Can we go down to the maintenance office and tell them they really need to reduce the no load power draw so your model fits better? No, that approach may not be good for your career in operations.
Can you suggest a full on plant survey, shutting down the crusher circuit every Monday, and with full particle size distributions on feed and product with changing operating conditions? Well no, again that may not be well received (but is a whole lot better than suggesting you do that with the really big rotating crusher called the SAG Mill).
So be careful about being too academic with steady state simulation.
But I like the idea behind the algorithm (factor x theoretical + no load value) for real time operations decision support system using dynamic modelling, that can be compared with measured data (and therefore can be self calibrating) and providing one of the many inputs to realtime operating cost analysis.