The choice of an appropriate thermodynamic model is one of the critical concerns in the chemical process simulation problems. After the selection of the most appropriate thermodynamic model, the uncertainties associated with the parameters of the model should be taken into account and the effects of such uncertainties should be addressed in the chemical plant design. In this work, the assumption of normal probability distributions for thermodynamic model parameters is coupled with unscented transform for efficient uncertainty propagation and obtaining probabilistic characteristics of output variables of the simulation model. The proposed methodology can be used easily when the input-output representation of the system such as process simulators are used as the modeling tool of the chemical processes. The law sampling points required in the unscented transform formulation can considerably reduce the computational burden while an acceptable accuracy compared to Monte Carlo techniques can be obtained. The methodology is illustrated by two case studies. (C) 2018 Elsevier B.V. All rights reserved.