This comes up time and time again....
The specification calls for the pump with the lowest run-hours to be started first, and so on....
Balancing the "run-time" in this way will ultimately mean that either the pumps will require planned servicing at roughly the same time, or they will fail at roughly the same time.
If this were my call, I would attempt to spread the maintenance....
example :
4 pumps require servicing (inspection, new seals, or whatever) after 20,000 run hours.
I would run 1 pump constantly until it had achieved 5,000 hours run-time, then bring pump 2 into the mix, alternating the usage of the two pumps. After 10,000 hours I would include pump number 3, and after 15,000 hours pump number 4.
Now my 20,000 hour maintenance means I'm not down more than 1 pump at a time, and the pressure on the maintenance schedule is removed.
If the run-time hours were clocked by counters, I would do a one-time only pre-configuration of the counter accumulators to 0, -5000, -10000, and -15000, and maintenance required flagged up at +5000 for each pump. Reset each counter when it is "serviced". This offset in "run-hours" would perpetuate through the life-time of the system.