100222-1054 EST
powerplay:
mivey has given you an easy out.
But I want add some broad general information.
If you have computers that are critical, then these should be on UPSs. Nothing to do with your current question, just basic philosophy. This means only the computer and display need to be on the UPS. Thus, high current loads, such as a laser printer, that are not critical to data should not be on the UPS in order to keep the UPS size down. This is not because in general the loads you are talking about will cause a problem, but actual loss of line voltage for more than about 1 cycle.
If you monitor the line voltage at your customer's location and find that typical voltage is 120 to 125 V, as it seems to be in our area, and dips do not drop below 105, this would be a very large drop on a system that is normally in the 120 to 125 range, then it is unlikely there would be any problem with computers. The regulated power supplies in most computers probably hold correct output voltage down to 95 V input or less. Just tried one of my computers and it functioned to 70 V input and failed about 65 V. You would monitor line voltage with a Fluke or equivalent with a MIN-MAX capability for maybe a day.
The average load you described on a 100 KVA transformer is moderately small. Probably the biggest problem is when any one device with a high inrush requirement turns on. Apparently there are a number of different, about 5 HP, loads. Assume 5 KVA for a motor. At 240 this about 21 A. Assume inrush is 10 times this, or about 200 A. The 100 KVA at 240 is 426 A.
I am going to use some wild assumptions because it is easy and I am only trying to present a broad concept. mivey with available resource information on power system components may come back with refinements that provide a more accurate picture. To continue suppose that there is a 5% drop in voltage from no load to 426 A. Then the inrush of one of your motors would produce about 2.5% voltage change. At 120 V this is a drop of 3 V, nothing to be concerned about. I think current power system transformers have less drop than this.
An important point is that the probability of more than one of your motor loads turning on at the same instant, meaning within about 5 seconds of each other, is very unlikely. Thus, this system would look relatively stiff most of the time. If some of the motors are related to each other, would turn on at the same time, then the voltage change would be greater. But this can be solved by a staged turn on of these loads.
A separate subject is high voltage transients generated by turning off inductive loads while current is flowing. Transient voltage limiters should be at the input of all critical equipment.
.
