Extended Grey-Box Modeling of Real-Time Hydrocyclone Separation Efficiency

Deoiling hydrocyclones are used offshore to reduce the amount of oil in produced water that is discharged to the surrounding marine environment. Stricter regulations and environmental concerns incentify optimization of the hydrocyclones' deoiling performance. These requirements enforce the pursuit of better models and better controllers. This paper presents suggestions and benchmarking of a modified hydrocyclone grey-box model. The proposed model modifications relax flow conditions, such that the model errors are reduced when the system deviates from normal operation. With the aim of a more realistic model, the cylindrical underflow apex was included in the oil droplet trajectory space. An estimation of the flow that recirculates inside in the hydrocyclone was introduced as this phenomenon has observed in multiple CFD models and research studies. The reduced computational load enabled the proposed model to be executed below 10ms, thus enabling the model to be used for model-based control. The original and the modified model are experimentally validated using a pilot plant. The modified model performs better during all tests and provides a reasonable prediction of deoiling separation efficiency compared with the separation efficiency obtained through real-time oil-in-water measurements.