Natural gas dehydration unit is employed to eliminate water from natural gas liquids and natural gas, and it is needed to avoid
condensation of free water and creation of hydrates in transportation and processing facilities, prevent corrosion, and meet
a water content condition. In this paper, a least-square support vector machine (LSSVM) coupled with genetic algorithm
(GA) was employed to estimate the water dew point of a natural gas stream in equilibrium with a triethylene glycol (TEG)
solution at different TEG concentrations and temperatures. Results showed that GA–LSSVM accomplishes more reliable
outputs compared with real recorded data in terms of statistical criteria.
