I've looked at the large events across long periods.    

Here is a plot of the event size per hour for 8am, May 14th.  Through 8am, May 15th  (The green lines indicate the start of runs).    Events less that 1.1MB are not shown.   The important features are that the aberrant events are present all the time (apart from 2 runs on May 14th).   The issue increases gradually during the first 

For a single fill we see:

There was a tune change to that came into effect after the access on May 15th.   This had a significant effect on the size of the aberrant events.   The following shows the average events size of the abort gap events for may 14th and may 15th:

The most import question is:  what is the effect of these events on the space charge? 

The BBC_MON trigger is our most inclusive collision trigger.   And thus gives a lower limit of the total charge in the TPC.  It currently has an event size of ~100k, and a scaler rate around 70k.

BBC_MON:    ~100k @ 70k -->  7e9 bytes / sec

Any additional background that doesn't make it into the BBC_MON trigger would add to this.

The bunch cleaning noise exists for about 2ms per second:   2ms = 155 * 120 bunch crossings = 155 orbits, so the charge deposited by the bunch cleaning would be about EvtSz * 155.   Here are the average event sizes of the bunch cleaning induced events vs time for May 14th and 15th.    Before May 15th:

Before May 15th,  BUNCH CLEANING size:    ~9.3e8 bytes/sec
After May 15th,  BUNCH CLEANING size:      ~3e8 bytes/sec

These are significantly smaller than the space charge due to collisions and other backgrounds, but still may be an issue due to their 2ms bursts.