Assigning a maximum TPD for these types of calculations poses an interesting challenge. 

Changes in the grind ability, PSD of feed and product, etc. must be dealt with. Also, the nameplate from design holds less meaning. Use of top day of production, top days of production, and other approaches are occasionally used.

Any comments on approaches people have used to determine this? I've used one approach from operating data, but would to hear what others are doing?

Recovery vs. a maximum recovery poses another challenge. Recovery vs. absolute available tells less than recovery vs. a number from a geomet approach.

Wrong metrics in my view. This thinking leads us along the mindless treadmill of more volumes and higher utilizations which are guaranteed to maximize (total) costs, even though they create an illusion of efficiency by reducing unit costs. The objective of a mining company mining is not to move rock, process ore, or to produce metal - it is to produce value for shareholders through cash flow, and other stakeholders through cash flow and other means. Whilst we focus on the physical activities regardless of whether those activities create value or not, then we are missing the point. Mining companies will never be productive in an economic sense whilst they (very efficiently) mine the wrong rock or mine it too soon, process or over-process marginally or non-economic material and create metal at a total cost that is more than or at best fractionally less than the market price. Let's use metrics such as Net Value $'s per man-hour, truck-hour or plant-hour. Focus on economic productivity, measured by the rate of net cash production - i.e. cash throughput. That brings a focus on the economic validity of the work they are doing, not just how much work they are doing.