(700d) Risk-Averse Health-Aware Control of Subsea Plants

Optimal process operation is important in the subsea oil and gas industry, especially for fields with high break-even oil prices. In order to optimize production, it is desirable to maximize the throughput. However, it is at the same time necessary to consider plant reliability and availability, as maintenance costs can have a significant impact on the profitability of the plant. Maximizing throughput while simultaneously maximizing the availability of the plant is often not possible in reality, as the two objectives are conflicting. This is why it is necessary to find an acceptable trade-off between plant revenue and maintenance cost.

This trade-off is usually not defined in a systematic manner. Instead, a sequential approach is commonly employed. First, the safe operating regions are defined and imposed as back-offs from safety-critical constraints. Only after the allowable operating region has been defined it is attempted to optimize the operation. This method may be overly conservative, for at least two reasons:

• The back-off is often defined with a worst-case scenario in mind, which leads to sub-optimal performance under regular operating conditions.
• When the system is complex and many degradation modes are effecting the reliability simultaneously, it might be difficult to specify meaningful back-offs manually.

Our proposed approach is to integrate prognostic and diagnostic models into an advanced, optimization-based control strategy, either by imposing constraints on the system availability, or by penalizing system degradation in the objective function (Verheyleweghen & Jäschke, 2017). We thereby achieve proactive fault prevention rather than retroactive fault-tolerance, which is more in line with industrial needs (Escobet, Puig, & Nejjari, 2012). The problem formulation is made risk-averse by using a CVaR-formulation of the problem, similar to what is commonly done in financial decision making (Rockafellar & Uryasev, 2000).

By doing so, we get improved economic performance compared to traditional control strategies by exploiting the additional flexibility gained by reducing the large back-off from safety-critical constraints. Simultaneously, we ensure safe operation and minimize unexpected breakdowns and maintenance interventions. The degree of conservativeness of the control structure can be adjusted online to reflect the risk aversion of the decision maker. Our proposal is different from current industrial practice, where it is still common to treat reliability and optimal operation in a sequential manner.

References

Escobet, T., Puig, V., & Nejjari, F. (2012). Health aware control and model-based prognosis. 20th Mediterranean Conference on Control & Automation (pp. 691 - 696). Barcelona: IEEE.

Rockafellar, R. T., & Uryasev, S. (2000). Optimization of conditional value-at-risk. Journal of risk, 21-42.

Verheyleweghen, A., & Jäschke, J. (2017). Framework for Combined Diagnostics, Prognostics and Optimal Operation of a Subsea Gas Compression System. IFAC-PapersOnLine, 15916-15921.