https://www.mdu.se/

Petroleum products are widely used as an energy supply, and the total production capacity of petroleum refineries is quite high. In this thesis, an energy intensive refinery process, hydroprocessing, is selected and evaluated in terms of its energy loss contributors. Digital solutions are discussed and demonstrated to reduce losses. Both hydrotreatment and hydrocracking processes are included in the evaluation since both require elevated temperatures due to the relevant reactions. While the former is the removal of undesired atoms, the latter is the production of short chain hydrocarbons from heavy oil. Both these processes contribute to cleaner fuel production.

When these processes are carried out in fixed bed reactors, the catalyst ages over time, slowing the reactions. Understanding the changes in system dynamics enables the control system to calculate the necessary temperature adjustments to facilitate stable product quality. The usual response is increasing the temperature, which adds to the heat load. If reaction rates are known, the temperature increase can be kept to a minimum. Obtaining real-time feed quality information can aid flexible feed processing refineries intensely. With real-time feed characterization, it is possible to use a feed forward model predictive control system to optimize reactor temperatures. Therefore, for varying crude oil quality, the control system can estimate the minimum temperature requirements for the product to be in the desired quality interval. Additional notice should be given to the temperature sensors as they supply data to the suggested control architecture. Wrong measurements threaten the optimality of the estimated control response. Faulty sensors should be detected and replaced to minimize the risk and collect correct data.

Observations made in this thesis show the possible energy gain for hydroprocessing by understanding the aging catalyst, soft sensor installation, feed forward model predictive control, and sensor fault detection. Hydroprocessing is a relevant topic for biorefineries. Although the demonstrations in this work are only for petroleum refineries, the suggested methods can be used in biorefineries as well as integrated co-processing petroleum and biorefineries.