本研究首先建立填充床反應器及鈀膜反應器來進行甲醇水蒸汽重組反應模擬器以代表程序系統。在填充床反應器與鈀膜反應器的系統中,當程序模式已知時,由已知的物理模式或以一替代模式,藉由隨機搜尋的方法獲得能產生最佳氫氣產量之操作條件與觸煤量。因為物理模式在隨機搜尋的計算過程中,需耗廢相當長的時間,故可選擇一替代模式來縮短計算時間。當程序未知時,我們以本研究室所提出之接續式擬均勻設計的實驗設計方法得到相似之結果。在所有實例研究中,所使用的控制因子有反應壓力、反應溫度、和水對甲醇的進料莫爾流量比。經由填充床反應器及鈀膜反應器設計與操作最適化成功的驗證了接續式擬均勻設計的有效性,並且上述的結果也顯示以本研究室之接續式擬均勻設計的實驗設計方法可在少量實驗過程中建立可靠的模式。
In this research, the simulations of a packed-bed reactor (PBR) and Pd membrane reactor (PMR) for carrying out the methanol steam reforming were performed first. For both PBR and PMR systems, if the process model is well known, direct random search method was applied to the original physical model or the meta-model to obtain the optimal catalyst loading and operating conditions for achieving a desired hydrogen production rate. Depending the computation time in solving the governing equations, the physical model or the meta-model are optional. If the process is not well known, the experimental design method proposed by our laboratory (the sequential pseudo-uniform design (SPUD) method) was applied to obtain similar results using only limited experimental data generated from the simulator. In all the case studies, the control factors are the reaction temperature, the reaction pressure, and the molar feed ratio of water to methanol. The effectiveness of the proposed sequential pseudo-uniform design (SPUD) method was demonstrated through the successful implementations in both a PBR and a PMR. It is expected that one can build a reliable model from limited experiments designed by the SPUD method.