According to the documentation: "Hardware control options (AMD cards only; use comma to specify different values for each GPU):"
We might therefore assume that single value stands for all cards.
I have observed that "-tt 60" option does not work and got it working with setting an individual value for each card like: "-tt 60,60,60,60,60,60,60,60" on the 8 AMD rig.
Not sure if this is a "bug" or intended behavior; the supplied readme.txt (the documentation) is not clear on the details.

My understanding of the temperature management in PhoenixMiner is that you need to setup a target temperature (-tt) before you can utilise other temperature parameters such as -fanmin, -fanmax, -tmax...

So give next options a try (taking into the account you have 12 amd GPUs):
-tt 60,60,60,60,60,60,60,60,60,60,60,60,
-fanmin 90,90,90,90,90,90,90,90,90,90,90,90

BTW: Using -fanmin 90 is in my opinion not the best approach in keeping the temperature as steady as possible. It will introduce higher than optimal temperature fluctuation which is bad for the hardware. The "default" is in my opinion optimal -fanmin 0 and -fanmax 100 leaving the miner a full flexibility on managing the target temperature of the GPUs using the build in temperature/fan curve. PH does not document how it does the temperature management so we do not know what exactly will the "-fanmin 90" do in terms of temperature management.