Great answers Kim, I have a few approaches to add - first let me rank the approaches in preference.
In summary, I have in the past simulated one connection at a time (pump/reservoir) if I did not have enough information for a full or skeletonized utility model and avoided modeling 2 or more connections at the same time due to the points mentioned in 3). I would then confirm each pipe could deliver the necessary fire flow volume independently as a degree of factor of safety. Also don't forget to make sure the fire flow data is current and taken during high demand periods (worst case scenario).
You can only replace a full system model with a pump/reservoir if the hydrant flow test is done under conditions similar to those under which the site will operate. So, if you want to represent two hydrants simultaneously as pumps, then the flow test on which you base the pump curves had to have been performed with both hydrants open simultaneously. As you know, two pumps working in parallel perform differently than either pump operating alone and cannot be simulated simply by adding the two pump curves together. The same is true here: two hydrants (near each other) flowing simultaneously will not flow as the sum of the individual flows the hydrants produce when opened by themselves.