Introduction: The Concepts of SOC and SOH in Thermal Energy Storage
In the battery field, "State of Charge" (SOC) and "State of Health" (SOH) are important metrics for measuring a battery's remaining energy and degradation. Similarly, the concepts of SOC and SOH can be applied to solid-state sensible heat storage devices: SOC represents the percentage of thermal energy remaining in the storage medium relative to its full charge capacity, reflecting the remaining steam output of the device; SOH indicates the thermal health of the storage medium, indicating whether its thermal capacity has decreased or its thermal conductivity has deteriorated relative to new equipment. Accurate SOC estimation helps dynamically monitor steam supply capacity and optimize charging and discharging scheduling; assessing SOH can guide maintenance and replacement, ensuring long-term reliable operation.
However, unlike battery charge, the energy content of thermal energy storage cannot be directly inferred from easily measured voltage and current. Instead, the heat content must be calculated based on temperature distribution, physical properties, and other factors. Furthermore, thermal storage materials may gradually age (e.g., repeated thermal cycling leads to a decrease in heat capacity and thermal conductivity), which manifests as a decrease in SOH. Therefore, developing SOC estimation and SOH assessment technologies is a challenging task for solid-state thermal storage control systems. Henan Rentai is exploring how to achieve online monitoring of SOC and SOH for solid-state thermal storage devices, using measurement methods, modeling algorithms, and calibration approaches.
Temperature Field Measurement and Heat Capacity Model
Temperature field measurement is fundamental to SOC estimation. The energy stored in solid thermal storage media (such as specialized bricks, concrete, and molten salt solids) primarily manifests as sensible heat stored as their temperature rises. Knowing the temperature at each location in the medium allows us to calculate the total heat content based on its heat capacity. As shown in the diagram above, multiple temperature sensors are embedded within the thermal storage block. These sensor readings provide temperature information at different locations within the thermal storage block. To infer the entire temperature field from a limited number of measurement points, interpolation or partitioning models can be used. For example, the thermal storage block can be divided into several regions, with the temperature in each region being treated as the value of a neighboring sensor or calculated using a heat transfer model. This is similar to discretizing a heat storage body into "temperature units" and combining them to approximate the actual temperature distribution.
With the temperature distribution, a heat capacity model is needed to calculate the heat. Heat capacity models are based on the material's thermal properties. However, many solid materials exhibit little variation within a common range, so the average specific heat capacity can be used for simplified calculations. Materials with phase changes require special consideration of the latent heat component, but most solid heat storage uses sensible heat, which does not involve the complexity of phase changes.
One approach to improving SOC calculation accuracy is multi-point temperature field reconstruction: Barz et al. have shown that by using multiple temperature sensors combined with modeling algorithms, the temperature field can be reconstructed and the average phase change or heat storage degree calculated, resulting in a more accurate SOC. Modern control systems can even embed simplified finite element models or neural networks to calculate the heat storage distribution using real-time temperature data, significantly improving SOC estimation accuracy.
SOC Estimation Algorithm and Online Calibration
Based on the above measurements and models, the control system software needs to execute the SOC estimation algorithm in real time. Common implementations include:
Table lookup or simple formula: For specific materials and calibrated devices, a table or empirical formula can be established that relates temperature to residual heat. The controller reads in various temperature values ??and outputs the SOC through interpolation or calculation. This approach requires minimal computation and is easy to implement in real time. However, when there are limited temperature sensors and the temperature field is non-uniform, simple formulas cannot capture complex distributions.
Layered Partitioning Algorithm: The heat storage body is divided into layers based on the heat transfer path, with decreasing temperature from the inside to the outside. It can be assumed that the temperature gradient is primarily distributed along the thickness, from the high core to the low surface. Several sensors in the inner layer are used to infer the average temperature of the central region, while sensors on the surface represent the boundary temperature. A multi-layered heat capacity superposition model is constructed to estimate the total heat. This is similar to treating the heat storage brick as several concentric shells, each with known or inferred temperature, and then calculating the total heat.
Kalman Filter Estimation: A dynamic model of the heat storage process (such as a thermal network model) is established, treating the temperature measurement points as observations, and using an Extended Kalman Filter (EKF) to continuously correct the SOC estimate. For example, the state equation is the change in heat storage, with the input being the heating power minus the heat release, and the output being the temperature of each sensor. The EKF corrects the internal SOC based on real-time temperature deviations, ensuring that the calculated value is more consistent with the measured value. This method integrates information from multiple sources and filters out noise. It is commonly used in aerospace battery management and can also be applied to thermal energy storage SOC estimation.
Neural Network/Machine Learning: A model is trained using large amounts of simulation or experimental data. The network takes temperature at each point and historical charging and discharging data as input and directly outputs an estimated SOC value. This data-driven approach can capture complex nonlinear relationships without requiring a precise physical model. Once trained, the neural network is deployed on the controller for inference. Each input of multiple current temperature points produces an SOC value. Information such as pressure and power can also be integrated to improve accuracy. It is important to note that the training data must cover a wide range of operating conditions; otherwise, generalization may be limited.
Regardless of the algorithm, online calibration is crucial. Initial factory testing can accurately calibrate the model parameters for thermal storage devices, but in actual operation, environmental factors and aging can introduce deviations, necessitating continuous calibration. Common online calibration methods include:
Regular calibration at known operating conditions: For example, after each complete heat release (steam is completely discharged and cooled to ambient), the SOC can be assumed to be 0%, while after each full charge (reaching the highest temperature equilibrium), the SOC can be assumed to be 100%. When the control system detects these conditions (such as the lowest temperatures at all points being equal or the highest temperature reaching the design value with no gradient), it automatically corrects the SOC calculation offset.
Calibration using the principle of energy conservation: The electrical energy input during the charging process is recorded, the heat loss estimate is deducted, and the calculated SOC result is compared to adjust the model parameters to ensure consistency. Similarly, for heat release, the measured steam heat is compared with the SOC reduction to correct for errors in the estimated heat capacity. This is similar to Coulomb integral calibration for batteries, except that heat losses are taken into account.
Sensor recalibration: Sensors may drift over time. For critical thermometers, the sensor coefficients can be corrected by comparing the sensor to a standard temperature during shutdown and cooling to ambient temperature, or by checking the reading error at a known temperature point (such as 100°C steam temperature). This ensures accurate temperature data and improves the reliability of SOC estimation from the source.
By combining multiple calibration measures, the control system can self-learn and gradually improve the accuracy of SOC estimation. In commercial operations, the SOC estimation error is generally required to be within ±5-10%. This allows dispatchers to rely on SOC readings when scheduling heating without incurring insufficient or wasted heat. State of Health (SOH) Assessment and Thermal Decay Diagnosis
Assessing the state of health (SOH) is more challenging than state of charge (SOC) because it involves monitoring and quantifying long-term changes in thermal storage performance. Several SOH metrics can be considered:
Maximum heat storage capacity decay: Corresponding to battery capacity decay, the equivalent of thermal energy storage is the percentage decrease in heat storage capacity from a full charge. For example, if a new device can store 1000 MJ on a full charge, but only 900 MJ at the same maximum temperature after two years of use, the SOH is defined as 90% based on thermal capacity. This requires regular measurement of the actual heat storage capacity. This can be achieved by fully charging and discharging the device multiple times and measuring it based on input and output energy. For example, record the input power from a charge of 0 to 100% SOC and compare it to the theoretical value; or measure the total heat output of steam from a full discharge and compare it to the initial rated value. The difference is the capacity decay. The control system can perform full charge and discharge tests during certain maintenance cycles to automatically calculate the SOH.
Decreased Thermal Efficiency: SOH can also manifest as heat loss or a deterioration in heat transfer efficiency. For example, if insulation aging causes increased heat loss, more power input is required to achieve the same SOC. Alternatively, decreased thermal conductivity leads to larger internal temperature gradients and reduced available heat energy. The control system can track changes in charging/discharging efficiency. For example, by comparing the amount of energy charged over a certain period of time with the SOC increase, if more energy is required to achieve the same SOC, this indicates increased losses and a decrease in SOH. Another example is monitoring the ratio of effective steam heat output to stored heat during heat dissipation. A gradual decrease also indicates performance degradation.
Changes in temperature response: Material degradation alters thermal conductivity and heat capacity properties, resulting in different dynamic responses to the temperature field. By observing the temperature rise rate at various points during heating or the cooling curve during heat dissipation and comparing them with the original baseline, abnormal changes can be detected. For example, if cracks or sintering occur in the material near a sensor, slowing heat transfer, the temperature rise at that point will lag behind during the initial charging phase. The control system can use this dynamic signal for diagnostic purposes: by slightly adjusting the heating power as a stimulus, analyzing the temperature response (similar to a frequency response test) to determine whether thermal resistance has increased. This is relatively complex, but some research is leveraging model parameter identification to achieve this goal. This involves continuously adjusting model parameters to fit the measured curve, and the parameter evolution reflects changes in the material's SOH.
Cycle count and stress: The system can also simply infer SOH by recording the number of thermal storage cycles and the highest temperature experienced. For example, if a material specification specifies a 5% thermal capacity decay after 500 cycles, the controller can estimate an SOH value based on the cycle count. This value can then be refined based on actual performance monitoring. This approach is similar to estimating the remaining life of a battery based on the number of charge and discharge cycles and can serve as an auxiliary tool.
During SOH assessment, data noise and fluctuating operating conditions can affect judgment, so long-term trend analysis is preferred over single measurements. The control system can store daily SOC, input and output energy, and other data, and periodically execute trend algorithms to assess SOH. For example, full charge capacity estimates could be compared every 30 days to smooth out daily fluctuations.
When the SOH drops to a certain threshold (e.g., 80%), the system should prompt maintenance, replacement of the thermal storage module, or insulation. SOH warnings are particularly helpful in preemptively scheduling maintenance shutdowns in applications where highly reliable steam supply is crucial. Through the coordinated monitoring of SOC and SOH, operators can clearly understand not only how long the existing heat storage can provide steam, but also how many years of service life the equipment can maintain, providing a scientific basis for production scheduling and asset management.
Conclusion
The intelligence of a solid thermal storage steam generator is reflected not only in its control of current temperature and pressure, but also in its insight into its own "energy balance" and "health status." By deploying multi-point temperature sensing, building a heat capacity model, and using algorithms to estimate SOC, the device achieves real-time quantification of "residual heat," essentially adding a meter to the thermal storage "battery." Simultaneously, through thermal efficiency monitoring and performance curve analysis, the control system gradually identifies the degradation trends of the thermal storage medium and completes SOH assessments. This makes the operation of thermal storage equipment more transparent and predictable.
It is foreseeable that with algorithm optimization and data accumulation, SOC/SOH estimation will become even more accurate and reliable. In the future, it will be possible to integrate the Internet of Things to aggregate and analyze data from large quantities of similar devices, further improving the model and achieving swarm intelligence. At the same time, integrating SOC information into scheduling optimizes charging and discharging schedules for optimal economics; incorporating SOH information into operations and maintenance plans ensures on-demand maintenance and extends lifespan. All of this is built on the technical foundation discussed in this article. By accurately calculating the charge and health status of the "thermal bricks," we ensure the equipment is fully aware of both heat and cold conditions, ensuring efficient utilization of thermal energy storage.
Henan Rentai Electrical Equipment Co., Ltd. specializes in industrial automation control system solutions, providing one-stop services including PLC+HMI, DCS, and remote monitoring. For inquiries, please call 0371-56520104 / 13526433367 or email info@hnrentai.com.