Developing an online predictor to predict product sulfur concentration for HDS unit

Sharma, Prafull (2018) Developing an online predictor to predict product sulfur concentration for HDS unit. Masters thesis, Memorial University of Newfoundland.

[img] [English] PDF - Accepted Version
Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.

Download (1MB)


Hydrodesulfurization (HDS) is an important process in refining industries. Advanced control system (e.g. model predictive controller) requires on-line measurement of the product sulfur at the reactor outlet. However, most HDS processes do not have a sulfur analyzer at the reactor outlet. In order to predict product sulfur concentration usually a data based sulfur predictor is developed. Performance of data based predictor is usually poor since some of the input parameters (e.g. feed sulfur concentration) are unknown. The objective of this thesis is to overcome these limitations of data based predictors and develop an online product sulfur predictor for HDS unit. In this thesis, a hybrid model is proposed, developed and validated (using industrial data), which could predict product sulfur concentration for online HDS system. The proposed hybrid structure is a combination of a reaction kinetics based HDS reactor model and an empirical model based on support vector regression (SVR). The mechanistic model runs in off-line mode to estimate the feed sulfur concentration while the data based model uses the estimated feed sulfur concentration and other process variables to predict the product sulfur concentration. The predicted sulfur concentration can be compared with the lab measurements or sulfur analyzer located further downstream of the process at the tankage. In case there is a large discrepancy, the predictor goes to a calibration mode and uses the mechanistic model to re-estimate the feed sulfur concentration. The detailed logic for the online prediction is also developed. Finally a Matlab based Graphical User Interface (GUI) has been developed for the hybrid sulfur predictor for easy implementation to any HDS process.

Item Type: Thesis (Masters)
Item ID: 12966
Additional Information: Includes bibliographical references (pages 89-95).
Keywords: Hydrodesulfurization, Hybrid model, SVR, Online prediction, Design of Experiments, Graphical User Interface (GUI)
Department(s): Engineering and Applied Science, Faculty of
Date: May 2018
Date Type: Submission
Library of Congress Subject Heading: Desulfurization -- Equipment and supplies

Actions (login required)

View Item View Item


Downloads per month over the past year

View more statistics