Industrial sustainability is pursued by people to achieve the long-term sustainable development of a given industrial zone. In practice, decisions and strategies for sustainable development must be made, reviewed, and assessed by industrial planners, business leaders, and involving communities from time to time. However, industrial sustainability problems are always difficult to be fully investigated and further optimized, because of the large size and scope that carries highly complexness, and inevitable uncertainties that are associated with data, information, and knowledge. Therefore, most known studies on sustainability decision-making are scenario based, where the degrees of sustainability of the scenarios as well as the decisions are compared, and then the best scenario associated with decisions are selected [1]. This type of decision-making approach heavily relies on the identified scenarios，and decision making is always a heuristic based. Moreover, no uncertainties are being considered in making decisions, which is inconsistent with the real situation.

In this paper, an approach consisting of both the system optimization and Monte Carlo based simulation is introduced to guide the decision-making process for more effectively identifying solutions of sustainability improvement. The basic algorithm of the proposed approach is structured in the following way. First, an industrial sustainability is described as a system optimization problem, whose objective function is the overall sustainability criteria of the whole system, and constraints are those subjected by the system's characteristic. Second, a genetic algorithm based approach is implemented to solve the optimization problem [2]. The best solutions will be recorded as candidates for further uncertainty analysis in the next step. Third, uncertainties are introduced into the system by changing the key input variables and/or system parameters from constant values to stochastic changing ones with normal distribution. In the next step, Monte Carlo simulation is applied by randomly specifying the values of input variables and/or system parameters with a great amount of times, and the best solution for each time is then selected from all the recorded candidates based on their values of sustainability criteria [3]. Finally, the most desirable solution will be readily identified through a comparison of the count of times for each candidate solution.

The main advantage of this approach is its capability of identifying optimal choice effectively with the consideration of system uncertainties. The proposed approach will be illustrated through analyzing the sustainability issues and developing strategies for enhancing the sustainability of an automotive manufacturing centered industrial zone.

Reference:

[1] Cristina P, Huang YL, and Lou HH, Ecological Input-Output Analysis-Based Sustainability Analysis of Industrial Systems. Ind. & Eng. Chem. Research. 2008: 47(6), 1955-1966.

[2] Tillman FA, Hwang CL, and Kuo W, Optimization techniques for system reliability with redundancy. IEEE Trans. Reliability. 1977: 26, 148-155.

[3] Bernd A. Berg, Markov Chain Monte Carlo Simulations and Their Statistical Analysis. New Jersey: World Scientific, 2004.