基於實廠數據之化學合成反應器性能預測模式

A Performance Prediction Model of the Chemical Synthesis Reactor Based on Plant Data

呂宗祐(Zong-Yo Leu)

指導教授 ： 張煖

淡江大學/工學院/化學工程與材料工程學系碩士班/碩士(2022年)

https://doi.org/10.6846/TKU.2022.00108

摘要

數位孿生(Digital Twin, DT）作為現代智慧製造的核心，充分運用感測器的資訊蒐集，並分析導入的數據與整合各種設備的模擬，創造數位孿生的多方面應用與發展，藉由感測器來偵測周遭環境及狀態，並將蒐集的資料處理分析後，將最佳數據直接套用於實物，達成智能製造的成果。本次實習主要基於實廠數據建立醋酸合成反應器預測性能模式，以在未來提供最佳操作指引與優化製程為目標進行醋酸實廠之AI數位轉型專案在實廠與文獻皆沒有反應動力學資料的情況下，利用Aspen Plus的數據調和(Reconciliation)功能，得以決定反應動力學模式參數，並據以探討不同反應器操作條件對產物的影響。此外，為了改善反應器性能，利用實驗設計(Design Of Experiment, DOE)法建立三種方便使用的性能預測模式，分別是醋酸淨生成量預測模式、一氧化碳對醋酸轉化率預測模式與甲醇對醋酸轉化率預測模式。最後優化操作之探討方面，因本研究未涵蓋製程之分離與回流段，無法獲得其操作成本，所以僅提出目標函數之定義。
複線性迴歸之三種性能預測模式的結果顯示皆有很高的R2值，分別為0.986、0.974與0.987，代表所需要被預測的對象具有很高的精確度，且針對迴歸式之95%信賴區間也可以定義出對預測對象具有較顯著影響程度的進料與操作條件。

關鍵字

甲醇羰基化 ； 方程導向求解 ； 反應動力學模式建立 ； 優化操作

並列摘要

As the core of modern smart manufacturing, Digital Twin (DT) makes full use of information collection from sensors, analyzes the imported data and integrates simulations of various devices to create various applications and development of digital twins. The device detects the surrounding environment and state, and after processing and analyzing the collected data, the best data is directly applied to the physical object to achieve the results of intelligent manufacturing. This internship is mainly based on real plant data to establish a predictive performance model of the acetic acid synthesis reactor, aiming to provide the best operation guidelines and optimize the process operation in the future to carry out the AI digital transformation project of the acetic acid plant. There is no reaction kinetics data in the plant or literature. In the study, Aspen Plus (Reconciliation function) is used to reconcile the kinetic parameters of the reaction. The rigorous CSTR reactor model is then used to explore the influence of different reactor operating conditions on the reactor performance. In addition, in order to serve for the improvement of the reactor operation, three easy-to-use prediction models were established using the Design Of Experiment (DOE) method. The models predict the net production rate of acetic acid, the conversion of carbon monoxide to acetic acid, and the conversion of methanol to acetic acid with high R2 values of 0.986, 0.974 and 0.987, respectively. Because only the reactor section of the acetic acid is investigated in the report, for the optimal operation study, only the objective function is discussed, which should include the profit of acetic acid as well as the operation and recycling costs of unreacted carbon monoxide and methanol.

並列關鍵字

Methanol carboxylation ； Equation oriented solver ； Reaction kinetic model establishment ； Optimal operating

參考文獻

Álvarez, V.H., Mattedi, S., Iglesias, M., Gonzalez-Olmos, R., Resa, J.M., Phase Equilibria of Binary Mixtures Containing Methyl Acetate, Water, Methanol or Ethanol at 101.3 kPa. Physics and Chemistry of Liquids. 49. 52-71, 2011.

Aspen Technology, Inc.. Aspen Plus Release 11.2, 2010.

Bidgoli, R.G., Naderifar, A., Mohammadrezaei, A.R., Nasr, M.R.J., Kinetic Study Modeling and Simulation of Homogeneous Rhodium-Catalyzed Methanol Carbonylation to Acetic Acid, Chemistry and Chemical Engeering, 31, 57-73, 2012.

Biegler, L.T., Grossmann, I.E., Westerberg, A.W., Systematic Methods for Chemical Process Design, Prentice Hall PTR, 1997.

Box, G.E.P., Wilson, K.B., On the Experimental Attainment of Optimum Conditions. Journal of the Royal Statistical Society, 13, 1-45, 1951.