Abstract

Sour water are aqueous waste streams from oil refining operations, heavily contaminated with hydrogen sulfide and ammonia, which need to be stripped before reuse or disposal, avoiding damages to process and environment. Two-stage sour water stripper units are the most common technology to treat sour water for hydrogen sulfide and ammonia separation to produce reusable water and send these species respectively to Claus and ammonia plants. The first stage of a two-stage sour water unit is responsible for properly splitting hydrogen sulfide and ammonia. This work uses surrogate models to predict the limiting point of hydrogen sulfide separation in the first stage of a sour water unit, allowing more efficient heat duty control strategies to achieve the difficult split of hydrogen sulfide and ammonia and simultaneously lowering heat consumption. Failure of compliance to this limit results in unspecified stripped gas from the first stage, impeding it to directed to Claus plant, entailing loss of sulfur production and higher load of pollutant emissions from flared gases. Therefore, a precise surrogate predictor was developed to dynamically define a quasi-optimum set-point to the controller of the first stage reboiler duty based on dynamic disturbances – the first stage input factors to the surrogate model, such as hydrogen sulfide and ammonia contents of the sour water. The new control policy outperformed the traditional first stage ratio control in terms of stripped gas composition and plant stability.