Bottom Line--Theory: The pump was cavitating at some level and corrected by slowly spinning down and back up.

Details--Checked the accumulators last Tuesday and found they all looked good.  But you have to bring all the pumps down to check/set the accumulator gas pressure.  I noticed the bubbling/gurggling/rock-rolling-around sound in pump #8 right away after bringing the pumps back up and reported in alog 19699 showing the drop in output from this pump.  The pressure out from this pump was down ~10psi even though the pump was spinning ~20% faster: see first attachment of 10 days showing the pressure out at the pumps for the four CS PS skids.  Also the upper right channel is the PID servo drive to the motor controller increasing the speed in response to the loss of pressure at PS1_PRESS1.  It is also evident that this pressure sensors is noisier by a lot due to this cavitation.

While the UGF of the PID is quite low, ~10mhz, starting the pumps back up with the PID will ramp it up quite quickly.  I suspect this start up had been just a little too fast for PS1, this time, and some noisy air bubbles would not clear the pump.

We increased the VFD Accel and Decel settings from 1 to 300sec to stop the pump spinning over 5 minutes.  After a quick look at the spider coupler, the pump was spun back up over five minutes and the noise was gone and the output efficiency and noise level was recovered.  Sweet!

Take away: Anytime the pumps are restarted, check the trends to confirm pump is operating as usual, or better.  Audible changes at the pumps mean something--investigate.  A slow spin up of the pumps is better.