I would have a few suggestions. First, don't size the pump larger than it needs to be. Determine what the next constraint is to increasing flow. In a refinery unit, there could be a hard limit based on permit considerations, heater, exchanger or reactor capacity. It does no good to buy a pump capable of 500 gpm if the pipeline is only rated for 400 gpm.
Balance the need for a good NPSH margin with the need for a lower suction specific speed. In this analysis, be realistic. The NPSH(r) is sometimes based on a group of worse case assumptions: lowest possible tank level, highest possible product viscosity, highest possible vapor pressure. If the convergence of these unlikely extremes is impossible, then don't design for it. If you are overly conservative about the suction head, it may drive you to buy a pump with a higher Nss which will limit the downturn.
Allow for unusual operating modes. In tank services, we size pumps to transfer the product faster. That is what the pumpers want. But, if they have to drain the tank to change to new RVP requirements seasonally, it can be a problem. When you want to set the floating roof on its legs, or re-float it, you have to go very slowly. A large pump might be incapable of running low enough in flow to meet the regulations for re-floating the roof. You might need a spill-back line to allow for these unusual conditions without destroying the pump. Or it might be worth the price to add a smaller heel pump to handle the low flow condition. I personally, believe that the minimum flow value given by the pump company can often be too low. Determine a more conservative minimum flow based on Nss and specific speed.
Ask what the future might hold. We recently bought a pump that was for a fuel oil desulfurizer. Three weeks after it was started up, they got an idea that they could capture a high profit by switching the unit to hot gas oil. The design of the pump (materials, gaskets, etc.) needs to be able to handle the new service.
Johnny Pellin
