A method for improve the reliability in a gas liquefied plant using Hotelling T2 is showed in this paper. The stationery work in this manufacture facilities during a few moths in a year involve a heavy duty service of gas diesel engines and ammonia gas plant for processing the methane gas and extract the condensate fluid of it. Then, a predictive maintenance plan is necessary to prevent a possible malfunction or shut down of the plant and avoid an operational cost increased. We are just sampling the signals from the plant when its working in optimal condition and then we will compare the next incoming data from the machinery versus the previous historical data set. An statistical process control algorithm Hotelling T2 based for monitoring the condition of gas engines and ammonia gas plant will be implemented.