(59y) Optimal Sensor Network Design for Maximizing Net Present Value and Its Application to Corrosion Monitoring in a Power Plant

During the past ten years, the total share of U.S. electricity generation from renewables has increased significantly. Though renewables are projected to be the primary source, fossil energy sources are still expected to be used to ensure grid stability. Coal is one of the prominent non-renewable sources that is expected to be continued to be used in many parts of the world for next several decades. Load-following operations of existing supercritical pulverized coal-fired (SCPC) power plants are causing significant damage to the health of the key components, as they are generally designed to operate at base-loaded conditions. One key boiler component that undergoes damage is the waterwall. The high metal temperature in the waterwall section of SCPC power plants and the presence of corrosive gases result in equipment damage primarily by corrosion. The accelerated damage can cause premature failure of equipment items leading to forced outages. Potential revenue is lost due to the plant’s inability to generate electricity during this downtime, and unexpected maintenance costs are incurred. Monitoring corrosion real time in the harsh environment is difficult. A novel electrochemical sensor for in situ monitoring of hot corrosion was developed in house.[1] Since the measurement is localized, it is required to deploy these sensors at every possible location, which, obviously, is not feasible. Furthermore, these novel corrosion sensors are likely costlier than traditional sensors like temperature and composition sensors. It is feasible to estimate corrosion at a given location by placing temperature and appropriate composition sensors at that location or some other location while placing the corrosion sensor at a different location. Thus, a cost-effective solution to place these sensors at optimal locations is essential.

In the open literature, optimal sensor networks have been designed for various objectives such as maximizing process information, minimizing error covariance, etc., subject to constraints like the number and cost of sensors.[2]–[4] An algorithm has been developed by some authors of this work for optimal sensor placement for multi-scale time-varying system.[5] However, for corrosion monitoring, time scale differences are much longer. Furthermore, to the best of our knowledge no sensor placement algorithm has been developed for maximizing net present value (NPV) by considering the incremental revenue due to placement of sensor integrated over the plant lifetime.

A quantitate model-based approach is proposed for optimal sensor placement and corrosion monitoring. For the model-based approach, a model of the hot corrosion mechanism is developed and validated using the data obtained from an industrial boiler. The proposed sensor placement algorithm considers an estimator for state estimation using the measurement for placed sensors. For state estimation, the Unscented Kalman Filter (UKF) is used for its ability to handle highly non-linear processes and produce accurate estimates.

A novel sensor placement algorithm is developed to incorporate the posterior error covariance of UKF and estimate the potential revenue increase from an energy market forecasting software. Increase in the availability due to corrosion monitoring is estimated using the energy forecasting software and posterior error covariance matrix from the UKF is developed as part of this work. The sensor network design problem leads to a mixed integer nonlinear programming problem. The algorithm yields the optimal number, location, and type of sensors. The optimal sensor network is used to estimate the corrosion rate along the waterwall. A number of operating scenarios are simulated by changing the temperature in the waterwall and concentration of the combustion gas. Performance of the estimator with the optimal network in terms of computational expense and estimation accuracy is evaluated. Impact of process mdoel and measurement uncertainties is evaluated.

References

[1] N. N. Aung and X. Liu, “High temperature electrochemical sensor for in situ monitoring of hot corrosion,” Corros. Sci., vol. 65, pp. 1–4, 2012, doi: 10.1016/j.corsci.2012.08.010.

[2] C. Sumana and C. Venkateswarlu, “Optimal selection of sensors for state estimation in a reactive distillation process,” J. Process Control, vol. 19, no. 6, pp. 1024–1035, 2009, doi: 10.1016/j.jprocont.2009.01.003.

[3] H. Y. Guo, L. Zhang, L. L. Zhang, and J. X. Zhou, “Optimal placement of sensors for structural health monitoring using improved genetic algorithms,” Smart Mater. Struct., vol. 13, no. 3, pp. 528–534, 2004, doi: 10.1088/0964-1726/13/3/011.

[4] A. K. Singh and J. Hahn, “Determining optimal sensor locations for state and parameter estimation for stable nonlinear systems,” Ind. Eng. Chem. Res., vol. 44, no. 15, pp. 5645–5659, 2005, doi: 10.1021/ie040212v.

[5] Q. Huang and D. Bhattacharyya, “Optimal sensor network design for multi-scale, time-varying differential algebraic equation systems: Application to an entrained-flow gasifier refractory brick,” Comput. Chem. Eng., vol. 141, 2020, doi: 10.1016/j.compchemeng.2020.106985.