First, the rules you make in Plesk are written as blobs to the MySQL DB.
When you hit "Apply" in Plesk it writes out the rules into a bash script in the Plesk folder.
When the Plesk firewall service is restarted it applies the rules from this script to the live firewall (this is the bit that is overwriting the rules you're trying to manually add to the firewall yourself).

I've managed to write MySQL queries to deal with part A, but there appears to be no command line way of doing the second stage (and modifying the bash script it creates is a little dangerous as there's no set structure to it or markers in place, your rules are likely to end up in the wrong place unless you enter them by hand).

In short - pain in the ass. As is always the way with Plesk.

The issue I have is that some clients use the Plesk firewall interface but I still need to add custom rules.

Thus far, I've not found a good way to handle this. I peaked into the plesk database and I've little clue how they right their rules.

The only method I've come up with was to create a custom chain and have a script to assure it is in as rule #1.