Thank you for your response. I spoke with a Phoenix contact "Relay Engineer" today. Very helpful BTW. The problem is not in the valve coil, but the voltage spike when the coil de-energizes. The spike is welding the relay contacts closed. The cables we use for the valves have a built in metal oxide varistor, that should prevent this spike from hitting the relay contact. I am going to try using a cable with a different "faster acting" shunt or "zener diode" and see if it makes a difference.
I am not monitoring the valve current or fusing individual valve coils.
As far as the hydraulics are concerned, I have 3 hydraulic pumps that the combined flow is over 900gpm. The pressure sides of all of the pumps have check valves, and overpressure relief valves. But none of the individual relief valves can handle the combined flow of all the pumps. So, if a pump relay sticks on, it can hold the check open, and when pressure starts to build, it can turn the pump with the stuck relay into a motor, "try to spin it backwards" and break it.
One thing i did not realize is that relay current is rated for a resistive load, and not an inductive load, but the relays are sized large enough "lucky".
I am also going to try a DPDT relay and go from the common of one contact out the NO to the common of another contact with the load connected to the NO. This should create a stop gap that should minimize contact arcing.