Burst disks are the safety valve that will blow (theoretically) before the tank. That way you get a mad rush of air relieving pressure before hitting the tank rupture pressure.
So if you swap out the burst disks for higher rated ones you run the risk of a tank rupture if you over pressurize past the rating of the tank as the burst disk will not fail by design. Overriding the built in safeties is not a very safe thing to do in my book.
A tank can still fail even with a burst disk if you pressurize the tank to hydrostatic pressure multiple times. The concept of a hydrostatic pressure test is to take the material to near yield conditions TO SEE IF IT EXPLODES. This generally creates a hysteresis effect where the material does not go back to its original shape as soon as the pressure is relieved. Even getting close to yield can create a similar condition. While this in itself is not necessarily bad, bringing it back to yield afterwards can be since it can create a plastic condition (permanent deformation). Once the plastic condition sets in then any further filling to exactly the same pressure will not create any further permanent (plastic) set BUT any overfill beyond that will. That means if the tanks even heats up in the sun, it will overpressurize and plastically set even more. This cycle will continue until the ultimate stress is reached in the material and the tank ruptures. The strain required to go from yield to ultimate failure is very short for these materials. If this material state is ever reached then the tank will be very sensitive to the upper pressure. Badly calibrated fill gauges can be enough to do more harm to the material. Eventually, the tank can explode if it ever reaches this plastic condition even with a burst disk. This is why tanks have ratings. 10% overfill does not create a problem since significant margin remains but routinely taking it to hydrostatic pressures mostly likely will. Whether it lasts one year or ten years depends on the quality of the material, however, we have found through nondestructive testing and thousands of empirical tests that small voids may not be a problem until plastic set begins and microcracks reach that void. At that point it becomes unstable and goes all at once. The crack might take hundreds of hours to reach a microvoid but it when it does the whole tank will unzip at one time. So in short, you can get away with doing cave fills many times but that doesn't make it safe. It's kind of like driving drunk, first 10 times nothing happens so you think its fine but then that last time doesn't go so well.
Granted this will happen much sooner with aluminum tanks due to the lower yield strength but it can still happen with steel tanks.
We have this problem on airplane structures where we often times design parts to plastically deform at limit load (maximum load ever expected). This is to save weight and we know the part will never see a load greater than limit so we can get away with it but if the load does go slightly above then the part typically catastrophically fails within very few cycles. I run these models quite often and see it first hand. It doesn't matter if its an aluminum bulkhead or a steel fastener. They fail quickly when they exceed the load that made them go plastic. This is highly dependent on material quality and the maintenance. We don't even parts to rub on each other since even small scratches are crack starters at stresses well below yield. I am not sure how much control tank manufacturers have over materials and also any dents and scratches will greatly reduce life if the tanks are routinely filled to hydrostatic pressure.
Routinely taking a tank to hydrostatic pressure is insane IMHO, however 10% overfills are generally safe since there is significant margin left and you should never be approaching a yield state in the material with a mere 10% overfill.
