煉油產業的穩健生產規劃問題--以台灣中油股份有限公司為例

Robust Production Planning of the Refinery Industry --With the Case of CPC Corporation, Taiwan

鄭凱文

指導教授 ： 王小璠

國立清華大學/工學院/工業工程與工程管理學系/博士(2011年)

https://doi.org/10.6843/NTHU.2011.00075

摘要

台灣油品市場在開放私營企業之後，煉油產業不僅已經進入一個完全自由市場，而且也面臨來自於全世界的競爭與挑戰。在這樣的一種環境裡，存在著許許多多不確定性的因素，然而這些不確定性的因素，如果不加以妥適規劃，將造成台灣整體能源供給之危機。因此，為了增加煉油產業的協調性與流暢性，如何節省生產成本、提升產能、減少存貨、促進運輸的流暢和利潤的提升是當前必然且重要的課題。
原油從油輪運送到港口，卸油至儲槽，再由儲槽輸送到煉油廠提煉，在一連串過程裡，產生了購油成本、運輸成本、儲存成本與生產成本等等，直到成品油輸送到市場販售時，才開始產生收益。本研究旨在探討煉油產業的生產成本、存貨成本，以及在一個連續的生產週期系統裡，油輪提早到達或延遲到達港口卸油時，對煉油廠連續生產流程所造成的影響。因此，如何妥善管控這些生產因素，一套兼具有敏感度與彈性的穩健生產規劃系統，是煉油產業所必須的。基於上述原因，本研究發展一個數學規劃模型，具有最小總生產、存貨和啟動成本的最佳生產架構，並以容忍度分析來處理這些不確定性的環境因素。另一方面，本研究亦基於模糊集合理論，提出一套模糊線性規劃(FLP)模型，將不確定性直接考慮在模式中，而非作求解後的分析，以對於不確定性的需求和成本提出一個最大化的利潤生產策略。
本研究將以台灣中油股份有限公司為例，當模型的規模很大時，將以柔性演算法來尋求最佳近似解。在眾多演算法中，特別是粒子群聚最佳化演算法(PSO)在搜尋最佳解時有相當好的表現。因此，本研究將分別以精確解與近似解就演算效率與精確度加以比較，提出最適解法以驗證本研究所提出的模型，俾能提供更有用、更精確的資訊，朝向在一個不確定因素的環境裡，發展有效控制成本和利潤極大化的煉油產業策略。

關鍵字

煉油產業 ； 連續生產週期系統 ； 容忍度分析 ； 模糊線性規劃 ； 粒子群聚演算法 ； 台灣中油

並列摘要

After the oil market in Taiwan has been opened to the private enterprise, the oil refinery industries in Taiwan have not only entered a completely free market, but also faced the challenge from world competition. In such an environment, there exist varieties of uncertain factors which have caused the failure of traditional production planning model. Therefore, to increase compatibility, it is essential to increase efficiency in the inventory, production, transportation and profit.
Refineries begin to gain revenues when their products are sold in the market. This research examines the possible costs of additional inventories or recurrent times of refineries derived from early or late shipment in a continuous production system. Meanwhile, we shall develop a responsive and flexible production planning system to cope with uncertain manufacturing factors. Then, based on the developed mathematical programming model, an optimal production structure with the minimum total cost of production, inventory and start-up is proposed. Tolerance analysis is conducted to cope with the uncertain environment. Furthermore, in order to clarify the uncertain demand and cost, Fuzzy Set Theory is adopted to develop a fuzzy linear programming (FLP) model, so that a maximal profit production strategy is proposed with highest degree of satisfaction.
Then, we would conduct a case study on CPC Corporation, Taiwan (CPC). For the purpose of more general application, the issues of large-scale cases have been taken into account so that if the model scale is too large, the corporation can use a heuristic approach to find the near optimal solutions. In this research, we shall adopt the soft computing technique, in particular, the Particle Swarm Optimization (PSO) algorithm, to propose time-effective solution procedure for the proposed models.
It is expected that the results of this research will be able to provide useful information towards developing cost and profit-effective oil refinery strategies in an uncertain environment.

並列關鍵字

Refinery industries ； Continuous production system ； Tolerance analysis ； Fuzzy linear programming(FLP) ； Particle Swarm Optimization ； CPC Corporation, Taiwan

參考文獻

Wangb, H. F., Zhengb, K. W., 2010. An alternative dynamic local and global conjoint particle swarm optimization algorithm. Submitted to Applied Mathematics and Computation.

Abilov, A. G., Zeybek, Z., Tuzunalp. O., Telatar, Z., 2002. Fuzzy temperature control of industrial refineries furnaces through combined feedforward/feedback multivariable cascade systems. Chemical Engineering and Processing 41, 87–98.

Alatas, B., Akin, E., 2009. Chaotically encoded particle swarm optimization algorithm and its applications. Chaos, Solitons & Fractals 41(2), 939–950.

Amos, F., Ronnqvist, M., Gill, G., 1997. Modeling the pooling problem at the New Zealand refining company. Journal of the Operational Research Society 48, 767-778.

Angeline, P. J., 1998. Using selection to improve particle swarm optimization, in: IEEE International Conference on Evolutionary Computation, Anchorage, Alaska, 231–238.