(483h) Development of a Health Monitoring Framework Under Uncertainty for Supercritical Coal Power Plants Considering Material and Operational Uncertainties

With the increased penetration of renewable sources of power in the electric grid, fossil-fired power plants are adapting their operational strategy accordingly changing their load faster and more frequently, which has greatly impacted the equipment health in these power plants. In plants that operate at extreme conditions, like supercritical pulverized coal-fired (SCPC) power plants, this impact is particularly profound, especially in the boilers of SCPC plants where the highest temperatures and pressures are reached. Therefore, as SCPC plants operate under load-following conditions more frequently, it is vital that the health of their equipment be monitored to avoid significant damage that would lead to costly unplanned shutdowns for maintenance or replacement.

To predict the damage done under load-following conditions, a high-fidelity model of the SCPC boiler is essential, since the thermal and pressure profiles of the water / steam and coal / flue gas, as well as the thermal profiles of the boiler tubes and steam headers, which are the components that tend to be susceptible to cycling damage.[1], [2] Variables such as temperature can be difficult to measure due to the harsh operating conditions in the boiler, the through-wall temperature profiles in the tubes are not practical to measure, and the spatial profiles of thermal and mechanical stress are also practically impossible to measure. Therefore, it is necessary to estimate these variables using few measurements that are available[3], [4] and a detailed model. Since the exact locations of the highest stress and most damage are not known and shift as the load changes, a monitoring approach based solely on measurements may fail to identify the most damaged component.

Thus, a model-based approach was developed in this work where a dynamic, distributed-parameter model of an SCPC boiler was developed to obtain the relevant thermal and pressure profiles within the boiler, which were then used to calculate the thermomechanical stresses at all locations in the boiler that can then be used to determine indicators of damage. In contrast to available boiler models that tend to either neglect essential variables for health monitoring, such as the thermal profiles of the tubes and headers,[5] or are too complex to run real- or near real-time,[6] the boiler model that was developed here retains enough complexity to capture the key variables for health monitoring while still being able to simulate a load-following boiler in real-time. The first-principles, dynamic model developed here is validated against load-following, transient industrial data for not only the main steam conditions, as has been done in the open literature,[7] but also for multiple points along the boiler where measurements were available. As part of this validation, data reconciliation was performed using the detailed, dynamic boiler model, which has full consideration of the thermohydraulics and heat transfer limitations within the boiler, rather than a simple model that only ensures satisfaction of mass and energy balances, as is common in the literature.[8], [9]

The validated boiler model with consideration of the tri-axial stresses on the tubes and steam headers formed the basis of the health monitoring framework. The framework was designed to account for uncertainty in both the operation of SCPC power plants and the material properties of the boiler. The operational uncertainty was characterized based on three years’ worth of power demand data from CA ISO’s OASIS,[10] where the past power demand for non-renewable power sources was used to define “typical” operating days for each season, as the power demand is highly seasonal, as well as the 95% confidence ranges for power demand at each hour for the “typical” operating days. Material uncertainties were quantified similarly where data was available to do such analysis, defining confidence intervals for the material properties, such as the rupture time, based on data for various temperatures and pressures.

Following the quantification of the uncertainty, Monte Carlo analysis, simulating over expected trajectories, was used to define the distribution of the expected Remaining Useful Life (RUL) of the boiler with regard to damage indicators, such as the rupture time and number of allowable cycles,[11], [12] under load-following conditions. This framework allows for the use of historical data to update the RUL such that the uncertainty in the estimate reduces with time. Using the framework, results of a case study will be presented wherein the RUL for an SCPC boiler is projected for ten years, then historical data for the first year is added to the framework as an update. The framework exhibits how the uncertainty in prediction changes as more data becomes available. Multiple scenarios for that first year are presented (e.g., one in which the plant cycled aggressively, one where the plant cycled normally, and one where the plant cycled conservatively) to show how the RUL evolves with different operational strategies.

References

[1] A. Shibli and J. Ford, “Damage to coal power plants due to cyclic operation,” in Coal Power Plant Materials and Life Assessment: Developments and Applications, Elsevier Inc., 2014, pp. 333–357. doi: 10.1533/9780857097323.2.333.

[2] R. K. Smith, “Analysis of hourly generation patterns at large coal-fired units and implications of transitioning from baseload to load-following electricity supplier,” Journal of Modern Power Systems and Clean Energy 2018 7:3, vol. 7, no. 3, pp. 468–474, Dec. 2018, doi: 10.1007/S40565-018-0470-9.

[3] M. Jaremkiewicz and J. Taler, “Online determining heat transfer coefficient for monitoring transient thermal stresses,” Energies (Basel), vol. 13, no. 3, 2020, doi: 10.3390/en13030704.

[4] J. Taler et al., “Thermal stress monitoring in thick walled pressure components of steam boilers,” Energy, vol. 175, pp. 645–666, May 2019, doi: 10.1016/j.energy.2019.03.087.

[5] M. Trojan, “Modeling of a steam boiler operation using the boiler nonlinear mathematical model,” Energy, vol. 175, pp. 1194–1208, May 2019, doi: 10.1016/j.energy.2019.03.160.

[6] M. Granda, M. Trojan, and D. Taler, “CFD analysis of steam superheater operation in steady and transient state,” Energy, vol. 199, p. 117423, Mar. 2020, doi: 10.1016/j.energy.2020.117423.

[7] A. K. Olaleye, “Modelling and Operational Analysis of Coal Fired Supercritical Power Plant Integrated with Post Combustion Carbon Capture,” 2015.

[8] X. Jiang, P. Liu, and Z. Li, “A data reconciliation based framework for integrated sensor and equipment performance monitoring in power plants,” Applied Energy, vol. 134, pp. 270–282, Dec. 2014, doi: 10.1016/j.apenergy.2014.08.040.

[9] S. Guo, P. Liu, and Z. Li, “Data processing of thermal power plants based on dynamic data reconciliation,” Chemical Engineering Transactions, vol. 61, pp. 1327–1332, 2017, doi: 10.3303/CET1761219.

[10] California ISO, “Open Access Same-time Information System (OASIS).” http://oasis.caiso.com/mrioasis/logon.do (accessed Apr. 07, 2022).

[11] ECCC, “ECCC Data Sheets 2005.” D G Robertson & S R Holdsworth ETD Ltd., Sep. 2005.

[12] “Water-tube boilers and auxiliary installations - Part 3: Design and calculation of pressure parts,” London, UK, May 2002.