these are the only metrics JMeter offers (as you know).

timeStamp, elapsed,label, responseCode, responseMessage, threadName, dataType, success, failureMessage, bytes, grpThreads, allThreads, URL, Filename, Latency, Encoding, SampleCount, ErrorCount.

MY guess would be to divide # of users by ramp up time.

[# users] / [ramp-up time in seconds] = users(i.e. requests) per second.

The only thing is what if response time is +/-. That would mean # concurrent users would vary at certain times. (More if response time is increased, less if response time is lessened).

Anyone know how to obtain this?