What triggers Essbase to alter the commit block interval during a calculation script?

Essbase 11.2.4

When examining the duration that shell script ran and comparing the results in a staging, quality assurance and production environment, I observed a wide swing in the time the same script took to execute. The production environment ran the quickest. The log showed that while executing a calculation script triggered by a MAXL command, the Essbase agent modified the commit block setting from 3000 to 100000. This did not happen in the staging or quality assurance environment.

In the ESSBASE DBA guide, there is a note about the commit threshold stating that Essbase automatically increases the threshold to 10 MB if the setting requires less that 10 MB be written before an internal commit. What I am trying to determine is why this happens in one environment, but not the others despite similar servers.