In presence of market prices, energy utilities have to evaluate and hedge the impact of energy prices fluctuations on their Profit and Loss. This requires, in particular, to develop efficient computational methods to price, optimize and hedge physical assets (such as thermal power plants). Mathematically, the problem of pricing a thermal asset can be stated in terms of a stochastic control problem and can be reduced in a very specific case to the thoroughly studied optimal stopping problem or American option pricing. We present some original variance reduction techniques for American option pricing and thermal power plants valuation.