The problem of mass failure of 450G boards?

Question

Mikrotik

There are 50 electrical distribution substations, they are equipped with RouterBOARD 450G boards for transmitting data from telemetry data from sensors and data from video cameras on duty. The boards are installed in housings on a DIN rail inside a separate large box. Somewhere in April 2019, they began to freeze en masse, and some of them failed. If you believe the person who monitors them, the reason is overheating. Following the readings of the sensors inside the cabinet, the temperature did not rise above +40, in the TX of the board the temperature is +70. What could be the reason? Power is supplied to the boards from a stabilized Delta 24V power supply
UPD: according to the latest information, in response to my proposal to barbarously perforate the cases (simply drill holes), it was said that, judging by the readings of the Mikrotik itself, its sensors show an increased voltage (instead of the incoming 27.5 volts 32.4) and an increased temperature (even in winter). At the same time, on the 450Gx4 board, the same parameters correspond to reality.

Answer 1

What does the manufacturer say?

Answer 2

It is worth starting to dig with the power supply. Delta is what it is. See what's at the output, incl. under load, look at the ripple again. The pulsations alone can shake anything.
Next, it’s worth diagnosing the 450G boards, to see what crashes on them.
It is also possible that the relative humidity is not suitable.
The situation is so strange. If massively began to die. Again, if we assume that it was installed and put into operation at about the same time, then the reason should be similar, the manufacturer answered you that everything is OK, which means that the PSU remains again.
You can try to bring down the temperatures by lowering the frequency in the settings.