I believe somethings of major importance has been left out here and suspicions of ultra-conservatism have crept in. The major reason I would never employ a verbatim "normal" flow rate of a liquid on the basis of performance for a centrifugal pump are:
1) Until I have a detailed explanation of the process and its flows, I can't have an accurate evaluation of what someone else calls "normal". Few, if any, processes are really and truly at steady-state. In the actual, real industrial processes flows are constantly varying - and it is important to know the process intimately to estimate the variances. Some processes are designed as batch and some as steady-state.
2) No process calculation -or simulation - can be defined as 100% accurate. This is an impossibility in real life. As an example, process raw materials and temperatures/pressures will vary - and so will the resultant flow rates within the process. Process analysis can be done, but even this is subject to human and instrument inaccuracies. To compound the problem, all the other major equipment designed and sized that operate in unison with the given pump will have inherent inaccuracies and capacities built-in. This includes the plant's electric motor drives that, if measured with instruments, will be found to have varying voltage drops and frequencies - contributing to motor speed variances. This, among others, is the reason why the major world-scale process contractors (Fluor, BKR, Bechtel, etc.) will not issue a "guarantee" of expected performance that is less than 15-20% of the design unless you pay them extra monies (a LOT of monies) to obtain such a guarantee. That is what I've been told by them in the past, and I can see their dilemma.
3. The centrifugal pump in question is always subject to "wear and tear" - particularly in the impeller clearances when handling gritty or slurry-like fluids. In fact, the improper installation of the pump can cause it to develop some capacity losses. Another fact is that impeller and "wear ring" clearance losses are a normal and routine check for maintenance crew in process plants - or they should be. That's why it's called a "wear ring" - it wears. If we accept fouling factors in heat exchangers to compensate for what we know is a contingent, capacity-negative effect - what do we accept as a compensation for impeller or pump wear that reduces capacity? The pump's performance curves also contain inherent inaccuracies - especially as one approaches the end-of-curve area.
I will not support the inflation of process flow rates merely for conservatism or "ass-covering" and I applaud persons who also take this stand. However, there are real, and experienced contingencies that exist in the design of a process and these contingencies often cannot be 100% identified and quantified. Like any other contingency that is know to exist, albeit not identified, an engineer is bound to recognize and identify this fact by allowing for them in an experienced and methodical manner. This is an area where experience and process know-how is of primary concern and value. This is also the reason why I have never understood the logic of some engineering contractors who believe that calculations are numbers and numbers are to be believed; therefore, since young graduate engineers can literally swamp us with calculations and numbers, they are the likely and best engineers to calculate the simulations and write the specifications for major equipment. Nothing, in my opinion, could be further from logic than this type of thinking. Young grads surely can grind out numbers; however, they lack the experience and process know-how and this often becomes a point where ultimate plant performance is affected. That's why I'm an advocate of bringing back the engineering mentor method of yester-year and giving young grads the opportunity to gain experience and know-how in fields other than just calculations and simulations.
I would never apply a "pat" or fixed number to the amount of capacity specified to a pump without good knowledge of what it is pumping and how much I expect it to vary in capacity. That "excess" capacity (over the calculated rated figure) is something that is, as I stated, subject to experience and know-how. But I would be forced to study well the process and the application.