The best operating chiller depends on the KW/Ton for each chiller running at full and part-load. This should be from actual instantaneous measured compressor and pump current, and delta temperature/flow across the evaporator, on each chiller.
From past experience it is more efficient to load one chiller fully before adding the next chiller instead of running all chillers at part load which also depends on the type of chilled water system proposed for a building, constant speed or variable speed, and the performance of theses chillers at various load points and a bunch of other things.
Another approach is referenced as “Running Redundancy” where multiple chillers and towers are running at part-load because if accidentally a single chiller failed there should still be chillers running to handle the load, avoiding the start-up time required to get the “standby” chiller up running into operation. If the load could be handled by two chillers, three chillers are running at reduced part-load and if one chiller failed the remaining two chillers load up and carry the required load on and satisfy the system demand.
On the other hand splitting the maximum load between two chillers “in series” reduces the lift on each compressor and maximizes operating effiencies.
The trick is merging the chiller performance with the weather, building load, tower performance, pump performance, pumping senario (P/S or VPF) etc. to get a valid representation of how the plant as a whole will work, which is really what matters. The combination of the industry rating standards, ASHRAE 90.1, etc are pushing manufacturers and the technology to increase part-load efficiencies.
HVACing 