My opinion is that a UPS is a waste of money for most applications. If the power is out, what is the PLC going to control? Further, if you have a UPS then you need special logic to "idle" the control during the power failure and implement a start-up sequence when the power comes back on. This is particularly critical if there is a short duration power blip - say a second or so. Further, a UPS only delays the inevitable. It is uncommon to have a UPS able to keep the PLC on for more than an hour. After that you still have the same issues. Can you guarantee that power problems will always be resolved in less than 30 minutes? Finally, a lot of folks install a UPS for transient protection, EMI noise filtering, and surge protection. There are lots cheaper ways to accomplish that objective!
This opinion is contingent on a couple of assumptions:
1 - the PLC has battery backed memory and the programmer included all critical setpoints within it
2 - The controlled process doesn't need PLC intervention to implement a safe shutdown on power failure (and it is a rare process that would need that!)
"Uninteruptible Power Supply" is one of those phrases that tends to give one a warm, fuzzy feeling. It sounds really good, right? However, if you think through the application thouroughly, most of the time it is a false sense of security being provided.
Note: an exception to the above is an unmanned remote location connected by telemetry to a main control station. Then you may need the UPS to keep the remote PLC alive long enough to send a "power failure" message to the main control, allowing an alarm to be logged and alerting operators there.