乙烷氣提塔濃度預測模式開發

在煉油製程中，利用蒸餾塔將原油分餾出液化石油氣、輕油、煤油、柴油等，而本研究是針對汽油及液化石油氣進入氣提塔後乙烷濃度的控制。本研究之氣提塔的目的為裂解汽油及液化石油氣，希望將乙烷(C2)有效的分離，但因進料為現場經過前端製程分離出來產品，成分不固定，導致現場操作人員無法得知正確的操作點。倘若溫度太低，塔底產品中C2的含量過高，溫度太高的話，則造成C3、C4在系統中循環。而塔底產品的C2濃度需經過一段時間，使得塔底產品進入後端丙烯分析儀才能得知濃度的多寡，進而調整氣提塔的溫度。如此一來，經過一段時間後才濃度不符合標準再去調整溫度，導致一些不必要的損失。研究共分為三個部份，第一部分為設計資料之操作指引，利用Aspen HYSYS化工模擬軟體，建立工廠模型。直接輸入設計資料，而得到模擬數據。以模擬數據與現場數據做比較，對於已建立模型做出修正，達到現場C2的濃度。藉由調整流量得知在各個操作條件下，第41板溫度與C2濃度關係。第二部分則是運用Aspen HYSYS呈現工廠的物性狀態，進料成分的改變增加濃度預測的困難度。第三部分則是soft sensor開發，利用較容易測量的變數，預測製程較為困難測量的主要變數。發展此製程的soft sensor困難點有：(1)進料成分隨原油種類及前端製程影響。(2)輸出變數乙烷的分析值與製程變數測量值時間延遲約1-2小時。因此，本研究運用部分最小平方法搭配移動視窗模型藉此解決上述之困難，發展成為具有時變性soft sensor的關鍵技術。

關鍵字

軟測量；氣提塔；部分最小平方法；移動視窗適應模型； HYSYS

並列摘要

In the refining process, Distillation columns are frequently used in chemical plants to separate one or several feed mixtures of two or more components. This study addresses the optimal operation in C2 stripper column. In reality, however, the feeds may vary in their flow rates, temperature, or compositions, which results in the process operating at non-optimal values. Therefore, we use Aspen HYSYS simulation software to builds up the plant model and get simulated data. In addition, using Aspen HYSYS simulation software presents the situation of the de-ethanizer. However, the challenges of developing the soft sensor are: (1) the compositions of the process input are highly dependent on the types of crude oil, which is varied from the front process. (2) the time delay of inspecting the ethane composition is around 1-2 hours. Therefore, the proposed approach for solving these challengers is using the PLS with moving window adaptive model.

並列關鍵字

Soft sensor ； Stripper ； Partial Least Squares ； Moving Window Adaptive Model

參考文獻

[1] P. Kadlec, B. Gabrys, S. Strandt, "Data-driven soft sensors in the process industry," Comp. Chem. Eng., vol. 33, pp. 795-814, 2009.

[2] A.J. Burnham, R. Viveros, J.F. MacGregor, "Frameworks for latent variable multivariate regression," J. Chemom, vol. 10, pp. 31-45, 1996.

[3] R.M. Filho, A.J. de Assis, "Soft sensors development for online bioreactor state estimation," Comp. Chem. Eng., vol. 24, pp. 1099-1103, 2000.

[4] W. Yan, H. Shao, and X. Wang, "Soft sensing modeling based on support vector machine and Bayesian model selection," Computers & Chemical Engineering, vol. 28, pp. 1489-1498, 2004.

[5] X. Zhang, W. Yan and H. Shao, "Nonlinear Multivariate Quality Estimation and Prediction Based on Kernel Partial Least Squares," Ind. Eng. Chem. Res., vol. 47, pp. 1120-1131, 2008.

國際替代計量

乙烷氣提塔濃度預測模式開發

全文下載

主題瀏覽