How to improve runtime accuracy for different models of APC

[Anonymous user]

In our production system, we have over 100s of APC UPS's all varying models some SUA ,some smart UPS, 1kva, 2.2kva, and 5kva. They range in 1 year old to some over 10 years. All of them show different runtimes, and during actual power outages the APC UPS did not hold up to the displayed runtimes by an factor of over 60%. To improve the accuracy of the runtime ( i can tolerate within 10 % of stated runtime)  and by reading through various APC literature, and advice from engineers, I have come to the following conclusions:

1. Replace UPSes every 6 to 8 years.

2. Replace Batteries every 3 to 5 years,  more sooner for closets running hotter.

3. Do runtime calibration every 6 to 12 months. ( Doing this more often, will reduce the life of the battery, was what i heard.  Not sure if doing this "runtime calibration" can knock out power to production systems, when we do this, and if certain models can handle this better than other. Plus if there an automated way to do this for the 100s of UPS, and schedule them in groups - that would be awesome).

4. Buy only smart UPS models or Symmetras ( As they can do calibration better, and are more accurate).

5.  Upgrade to new NMC 30 or 31 model - and with version 6.0.6.

6. Run at 30 to 40% load or less. ( we normally run at around 20 to 25% load).

Do these conclusions hold good ? Are others replacing Batteries and UPSes at this rate ?

Finally, do all these make good standards for standard IT closets - which we try to keep at room temperature of around 70 F, but some run hot at 85 to 90 , due to numerous running equipment and not having sufficient cooling in the closets.

Thanks for shedding light on this.