How do you handle a PhD student whose experiments keep failing?

I have a PhD student in environmental engineering whose bench-scale experiments have failed to produce usable results for several months now.

The methods look sound on paper, but something in the setup or the samples keeps going wrong.

I want to support them without either dismissing the problem or letting them spiral.

How do you tell the difference between a genuine technical issue and a student who needs to rethink their approach?