Auto Changing transformer tap

hi, I have a computer controlled power supply and I have problems with power dissipation. power supply voltage is +/-30V and max current is 2A. in the worst case I have 60W(30 x 2) power dissipation on output transistor. actually before building it I had no Idea how difficult 60W power dissipation can be but after that I ended up with a huge heat sink and a fan and many dead transistors. another problem is I had to accomodate 8 of such power supplies in a box(relatively big) and I can have 480W power dissipation in worst case. so now I'm thinking about changing transformer tap according to potential difference between collector and emmiter of the transistor, I don't want to use relays for many reasons:
1-out voltage may change continuesly and relatively fast (every sec for many hours)
2-output should be stable within a fraction of millivolt so noise sources should be minimized
3-I don't like relays
...
so I decided to build it with some sort of SCR. my idea is to use OpAmps with hysteresis for comparison of collector-emmiter voltage with a constant voltage and accordingly switch on or off a few triacs that each connected to one tap of the transformer. the selection is managed so that only one triac be in on state any time. output of all triacs are connected to one bridge. I thought maybe someone have another idea so I can study that option too. I will appreciate giving any idea and discuss about possible pitfalls of different options. thank you very much!

The solution with several taps and triacs will work but you will have to make sure that the switching is synchronized with the zero-crossing ..

Here is another option: have you heard of saturable reactor?

A saturable reactor capitalizes on this effect by forcing the core into a state of saturation with a strong magnetic field generated by current through another winding. The reactor's "power" winding is the one carrying the AC load current, and the "control" winding is one carrying a DC current strong enough to drive the core into saturation.

Click to expand...

See picture below ..
For more details goto the bottom section of: https://www.allaboutcircuits.com/vol_2/chpt_9/7.html
This method is non- linear, but all what you need is D/A converter and PA for DC current and saturable reactor ..

Regards,
IanP