Introduction
Biomass power plants play a critical role in sustainable energy production, utilising organic materials to generate heat and electricity. Within these facilities boilers convert heat energy (liberated from combusting biomass) into steam to drive turbines and produce power.
Diagnosing faults within biomass boilers can be challenging, as issues may not always be readily apparent or easily detectable through traditional means. To address this challenge, modern industrial water-tube boilers are equipped with an array of sensors capable of capturing vast amounts of operational data. While these sensors provide a wealth of information, extracting actionable insights requires sophisticated data processing and analysis techniques.
A biomass power plant located in Mpumalanga was dealing with a sudden drop in generation capacity with no clear signs of the root cause. This prompted an investigation to determine what may be causing this deficiency within the plant.
In this study, engineers leveraged historical sensor readings and control parameters from a biomass power plant to pinpoint defects impacting boiler efficiency and power generation at maximum continuous rating (MCR). Using exploratory data analysis [1], the aim of the study was to identify boiler components exhibiting abnormal behaviour which contribute to the reduction in power generation capacity. Moreover, by comparing data across multiple seasons, we enhance the reliability of the findings and gain insights into the long-term performance trends of the equipment.
Use of data mining as investigative tool
The approach required the extraction of useful information and patterns from boiler instrumentation data can be done in a logical and thoughtful manner. The process, known as data mining, is to analyse large amounts of raw data to search for patterns and extract useful information [1-2].
The most significant pattern extracted came from the exhaust steam pressure of the turbine into the condenser, which is presented below in Figure 1.
Based on previous performance tests, the heat and mass balance indicated an exhaust pressure of 0.09 bar (a). Figure 1 a) and b) show that for April and May 2022, the average turbine exhaust pressure readings was around 0.125 bara. While during April 2023 (Figure 1a), the exhaust pressure was up at 0.175 bara on average and during May 2023, the exhaust pressure had an upwards trend on average that varied between 0.15 to 0.17.5 bara.
Based on the analysis, it appears that the condenser performance was already out of specification in 2022. Possible issues may be fouling of the condenser, issues with vacuum control or underperformance of the cooling towers.
Figure 2 considers the effect the condenser pressure has on the generated power. The data was generated using a validated numerical thermal model of the plant. It can be seen that for a higher exhaust pressure a drop in the generated power is observed, thus indicating the condenser operation is currently sub-optimal.
This study has demonstrated the effectiveness of employing data analysis techniques to analyse historical boiler data for diagnosing faults and evaluating the performance of a biomass power plant, by following a systematic approach outlined in the data mining process, including problem definition, data processing and exploratory data analysis. Not all patterns extracted have been described within this article but still bear significance in providing insight into the plant’s operation. They are, however, unrelated to the study.
Key factors behind power generation reduction
Through data analysis, engineers identified several key factors contributing to the reduction in power generation performance at the plant. These factors include variations in auxiliary steam demand, steam turbine performance, fuel feed consistency, combustion stability, and condenser operation. By examining trends in exported power, generated power, parasitic load, reclaimer speed, economiser temperature differences, and condenser exhaust pressure, it was shown that the exported power dropped significantly over the period between 2022 and 2023 when operating at 100% MCR. Furthermore, analysis of the various data trends illustrated minimal increases in the parasitic loading and the higher economiser outlet temperature seen in 2023 point towards combustion instability.
A substantial increase in the turbine exhaust pressure/condenser inlet pressure was observed, which was likely the main contributing cause for the power island underperforming. Subsequently, the effects of the condenser pressure were modelled numerically to understand the relation between the exhaust pressure and generated power, which showed that for an increase in the turbine exhaust pressure results in a decrease in the generated power.
By Christopher van den Berg
Cyber-Physical Systems Engineer
Industrial Watertube Boilers
John Thompson