Abstract: Valves are an important tool for flow control in fluid or gas networks. We investigate the possibilities to set up valves on the Lightning Network to improve flow control and reduce expected payment failure rates. We show that good asymmetric choices for `htlc_maximum_msat` on a channel can reduce the expected payment failure rate significantly. We assume that the potential benefits from carefully selecting `htlc_maximum_msat` as a valve seem not to be fully exploited by routing node operators and Lightning Service Providers yet to optimize their flow control and improve the reliability of their channels. We analyze the power of valves on the Lightning Network mathematically by using Markov Chains. These can be used to model the uncertainty about the liquidity in a channel that arises from drain in the depleted channel model. Using these techniques we show theoretically that properly using valves may statistically lead to a more balanced channel and expected payment failure rates drop from two digit numbers below 3%. We provide two experimental algorithm ideas and encourage Lightning node operators to investigate the potentials from utilizing this tool that is already being shipped with the protocol.