In a Bayesian Network how do I "neutralize" a node?

If I have a node in the network that's no longer functioning or observable what probability do I have to assign to it to neutralize it completely.

Say for example that I have the sprinkler, rain, wetgrass network example from wikipedia. If my sprinklers break how I do I make it so their variable doesn't affect the outcome anymore?

Obvsiouly I could remove the variable from the network but that's quite difficult to do in programming because it alters the structure. Instead I would like to assign it a probability that no longer alters the final result. What's the proper way to deal with this?