One of the prime rules of troubleshooting is not to trust a single thing until proven. You've checked the breaker's readings of the phase currents but not that of the supposed ground current. Without actually measuring that, you're trusting an unproven device which could very well be wrong.
There are a couple of options- measure the actual current on the grounding conductor (difficult), measure the "zero-sequence" current on all the phase leads (mentioned earlier), or have the breaker properly tested. Only then will you know if it's a real problem or an imagined one inside the breaker. You could also swap the breakers between machines but that takes both out of service for a while.
One other thought- is the breaker actually measuring that ground current or calculating it? I'd be extremely surprised if there was only one grounding path (a wire) from a machine with lots of pipe connected, so the breaker is probably (wrongly) calculating that current.
What's model # of the breaker? Folks here like to look up the data sheets and manuals.