The monoethylene glycol (MEG)-based natural gas (NG) dehydration process often faces significant glycol losses at the stripper column because of vaporization and poor NG liquid recovery because of the inefficient refrigeration system. A proportional integral derivative (PID)-based regulatory control strategy can partially mitigate this problem under certain upset conditions. However, to run the plant more efficiently and economically, it is necessary to apply advanced control technologies to reduce MEG losses. Although there are several examples of industrial implementation of advanced process control technologies in the refining industry, there are very few realistic examples that demonstrate the advantages of modern control technology in NG production. In this study, a plant-wide dynamic simulation model for an NG dehydration plant is developed and examined. A hierarchical control system comprising dynamic matrix control (DMC) and basic regulatory control loops is constructed to optimize the plant operation in terms of reducing MEG losses and minimize the operating costs with ensured product qualities under various process upsets. Both standard DMC and adaptive DMC controller models are developed based on the subspace identification model in the Aspen manufacturing platform. Although adaptive DMC performs a little better than the standard DMC, both controller models are very robust and can handle +/- 25% feed and 31% chiller temperature disturbances and demonstrate significant improvement over traditional PID control methodology.