Advanced Control and Monitoring Technologies for Optimizing Wide Load Operation in a 330 MW Boiler

Current temperature measurement technologies for boiler furnaces can be categorized into four main types: (1) laser optical temperature measurement [4], (2) digital image processing-based flame detection [5], (3) radiation temperature measurement [6], and (4) acoustic temperature measurement [7]. The laser optical method requires an additional laser light source, complicating its implementation for furnace temperature measurements. Both digital image-based flame detection and radiation temperature measurement face significant challenges in high-temperature environments, particularly due to dust-induced wear on their measurement windows.

In contrast, acoustic temperature measurement method is well-suited to harsh conditions, such as high temperatures, corrosion, and dust, offering substantial potential for applications in thermal power plants. Since its initial application in 1983 for measuring temperature distribution, this technology has gained increasing attention, with ongoing research focused on its application in boilers. Acoustic temperature measurement can now be used to construct 3D temperature fields within the furnace. Notably, researchers have made significant progress in reconstructing these 3D temperature fields, achieving improvements in reconstruction accuracy, speed, and noise resistance. For instance, a reflected sigmoid function has been employed in [8] for interpolation, successfully reconstructing a two-dimensional temperature field without missing data. Additionally, an ultrasonic simulation front-end TDC1000 combined with a TDC7200 digital conversion chip has been utilized in [9] to create a time difference measurement system, enabling direct acquisition of flight time and significantly enhancing the efficiency of the acoustic temperature measurement system.

Sampling from positive pressure direct-fired coal pulverizing systems presents challenges, as only coal powder fineness and concentration at the outlet of the coal mill can typically be obtained through system adjustment tests. In recent years, some pow-er plants have installed online coal powder sampling and concentration measurement devices to address issues such as pipe wear and blockage. However, these devices of-ten rely on tube suction sampling, which struggles to effectively extract coarse coal powder. Consequently, the fineness analysis may be limited, with analysis result sometimes approaching zero, significantly deviating from true values [10]. Further-more, online sampling devices, operating within the coal powder pipe, are prone to wear and blockage, reducing their reliability.

With advancements in mathematical theories, such as neural networks and sup-port vector machines, researchers have begun applying these techniques to optimize coal powder production systems. For example, neural network modelling has been used in [11] to optimize steel ball coal mills and reduce unit consumption. Similarly, genetic algorithms have been applied in [12] to model and optimize ball mills, aiming to lower coal mill unit consumption. Research on the online optimization of coal milling systems has also been conducted in [13], analyzing the dynamic characteristics of coal mills and optimizing power consumption by efficiently distributing loads among multiple coal mills. As power plants face increasing demands for energy conservation and environmental protection, coal powder fineness, concentration, and unit consumption directly impact the economic operation of boilers and NOx emission levels. Achieving online optimization of coal powder production systems can reduce coal consumption, improve boiler efficiency, and lay the groundwork for online combustion optimization, which is crucial for developing energy-saving and emission reduction technologies.

The results of online optimization control systems for coal powder production directly contribute to energy conservation and emission reduction in thermal power units. On one hand, the combustion efficiency of low-NOx burners is highly sensitive to changes in coal powder fineness and is significantly influenced by the primary air rate, meaning the operational status of the coal mill directly affects boiler efficiency. On the other hand, the uniformity of coal powder fineness and concentration restricts NOx generation and emission levels [14], also impacting high-temperature corrosion of water-cooled walls. Therefore, optimizing the fineness and concentration uniformity of coal powder is essential to ensure both safety and environmental performance in power plant operations.

Boiler combustion optimization is a complex multi-objective problem that significantly impacts the production and operation of thermal power plants. In the context of deep peak shaving, optimizing wide load operation control becomes even more critical. Currently, the optimization and control of coal-fired boiler combustion systems primarily rely on parameters such as changes in steam pressure and flue gas composition. In recent years, NOx emissions have garnered considerable attention, with flue gas NOx levels becoming key parameters for adjusting combustion processes.

However, whether the control parameters are based on gas composition in the flue gas or changes in steam pressure, the time delay relative to the combustion reaction time scale is often too long. Consequently, many researchers have turned to technologies such as acoustic temperature measurement, infrared temperature measurement, and optical temperature measurement to enhance combustion process monitoring. Although significant advancements have been made in temperature field measurement technologies [15-19], there remains a gap in applying these measurements to guide boiler combustion optimization and adjustment. This challenge is particularly pronounced during deep peak shaving, where the need for rapid monitoring and regulation of the combustion process during wide load operation presents a major obstacle for the power industry. Thus, studying acoustic temperature measurement technology for boiler combustion optimization holds great potential for enhancing the wide load operation capability of boilers.

Figure 1 illustrates the 3D reconstruction of the furnace temperature field, generated through temperature detection and numerical simulation. This process comprises three main components: first, the visualization and reconstruction of the furnace’s plane temperature field using the acoustic wave method; second, one-dimensional modeling of the furnace temperature field based on the monitoring of the boiler's metal wall temperature; and finally, the reconstruction of the 3D temperature field in heterogeneous media.

Figure 1.

Schematic diagram of 3D furnace temperature field reconstruction.

In the first component, the square of the sound wave propagation speed in gas is proportional to the gas temperature along the transmission path. Utilizing sound wave transmitter/receiver devices, the average temperature along the sound wave's path inside the furnace is measured. A temperature field reconstruction algorithm, combined with tomographic imaging technology based on Gaussian functions and regularization methods, is employed to reconstruct the temperature field from limited temperature projections. This approach facilitates real-time online visualization of the temperature distribution and related parameters within the furnace.

The second component involves segmental thermal calculations to model the furnace temperature field. The furnace is divided into several sections, each exhibiting distinct combustion reactions. Fuel input varies across these sections, with some areas transferring unburned coal downstream. Additionally, the air supply differs in each section, necessitating separate calculations for combustion efficiency. Factors such as inter-sectional heat transfer and surface fouling are also considered to refine the model. Drawing from previous research, a new soot-blowing scheme has been introduced to monitor ash accumulation within the furnace. By installing a multi-point temperature measurement system on the rear fins of the water-cooled wall tubes at varying heights, real-time temperature readings are obtained, yielding accurate results under relatively stable load conditions.

Finally, by capturing the furnace's planar temperature field using the acoustic wave method, and applying furnace wall temperatures as boundary conditions, computational fluid dynamics can be effectively utilized for 3D furnace temperature field reconstruction. The combustion process is a highly complex interplay of physical and chemical phenomena, involving fluid flow, heat transfer, and chemical reactions, often accompanied by reflection and heat release. In addition to adhering to the principles of mass conservation, Newton's second law, and the first law of thermodynamics, the process also complies with essential laws of species transformation and equilibrium, ensuring the stability and accuracy of combustion.

This paper employs a multi-point array temperature measurement system, which involves installing temperature sensors on the back of the furnace wall. This system integrates monitoring of the furnace outlet flue gas temperature as a comprehensive indicator of performance. By detecting changes in the heat transfer efficiency of the water-cooled wall pipes due to slagging and contamination within the furnace, the system monitors temperatures at the fire-side fins of the water-cooled wall pipes, as well as the temperature between the rear of the pipes and the insulation layer. The temperature difference recorded before and after soot blowing is used to assess the slagging and contamination status at specific points, across different furnace zones, and on single-sided walls.

Utilizing a furnace outlet flue gas temperature algorithm, the system assumes an ideal, clean interior of the furnace to infer the expected outlet flue gas temperature. Infrared temperature measurement equipment is installed on the opposite wall of the furnace outlet section to compare the calculated clean-state outlet temperature with real-time measured values. This comparison provides insights into the overall slagging condition of the furnace, as illustrated in Figure 2.

Figure 2.

Schematic diagram of furnace wall temperature measurement point layout.

Research on the online optimization of the pulverizing system is based on multi-source information fusion, necessitating a comprehensive assessment of both boiler efficiency and the energy consumption of the coal mill. The optimization process begins by determining the optimal coal powder fineness. Subsequently, adjustments are made to the loading force of the coal mill and the rotational speed of the dynamic separator to ensure that the pulverizing system operates at peak efficiency.

3.3.1

Determination of Economic Coal Powder Fineness

Excessively low coal powder fineness significantly increases power consumption during production. Therefore, it is essential to identify an optimal balance referred to as the economic coal powder fineness. This optimal fineness is influenced by factors such as boiler load and the volatile matter content of the coal, as the complete combustion and burnout of coal powder are closely linked to furnace temperature and secondary air distribution. For a given coal powder fineness, higher furnace temperatures and more efficient secondary air distribution lead to more complete combustion, resulting in lower carbon content in fly ash and slag. These factors must be considered when determining the economic coal powder fineness. The calculation process for determining this fineness is illustrated in Figure 3.

Figure 3.

Calculation of economic coal powder fineness.

3.3.2

Boiler Intelligent Combustion Optimization Control

In response to the operational characteristics of the powder-making system, which exhibits significant delays and strong coupling, this study focuses on the hydraulic loading force, dynamic separator speed, and coal mill outlet temperature as regulated variables. The generalized predictive control algorithm is employed to implement multivariate automatic control of the powder-making system, thereby enhancing the operational efficiency of the medium-speed mill.

During boiler operation, the radiant energy produced by combustion is influenced by furnace temperature, establishing a direct relationship among coal feed, furnace temperature, radiant energy, and boiler load. When the unit operates steadily, this relationship remains stable, and the radiant energy in the furnace corresponds to the unit load, referred to as normal radiant energy. However, during actual operation, particularly during rapid changes in unit load, discrepancies may arise between radiant energy and boiler load. These discrepancies manifest as follows: 1)

Load-Raising Stage: If the combustion rate in the furnace increases too rapidly, it may lead to overheating of the boiler furnace tubes, superheated steam, and fuel wastage. In this scenario, the actual radiant energy significantly exceeds the normal radiant energy.

2)

Load Reduction Stage: Conversely, if the combustion rate decreases too quickly, the heat load becomes insufficient, leading to combustion deterioration, which may ultimately result in boiler flameout. During this stage, the actual radiant energy is considerably lower than the normal radiant energy.

Consequently, this paper investigates the impact of the furnace radiant energy signal on unit load, using the radiant energy signal as a critical parameter for the safe and economical operation of the combustion system. This signal is incorporated into the closed-loop control of the combustion system to achieve flexible and rapid alignment of furnace radiant energy with unit load and fuel quantity, thereby significantly enhancing operational efficiency.

Figure 4 illustrates the traditional load control strategy for thermal power units. In this diagram, NE. SP and NE represent the unit load setpoint and the actual unit power output, respectively. PT. SP and PT denote the pre-machine pressure setpoint and actual pre-machine pressure, respectively. PIT and PIB are the main controllers for the steam turbine and the boiler, respectively. Additionally, uT and uB signify the commands for the turbine regulating valve opening and the boiler combustion rate, while △uT and △uB represent external disturbances affecting these parameters. Furthermore, GNT(S) and GPT(S) characterize the impact of variations in the steam turbine valve on unit load and pre-machine pressure, respectively, while GNB(S) and GPB(S) represent the effects of combustion rate variations on unit load and pre-machine pressure.

Figure 4.

Traditional load control strategies for thermal power units.

The thermal unit load control strategy incorporating radiant energy signals is illustrated in Figure 5. In this figure, M0 represents the boiler radiant energy signal, GBM(S) denotes the influence of combustion rate changes on the boiler radiant energy. GPC(S) serves as the boiler radiant energy model prediction controller. Additionally, GNBM(S) and GPBM(S) characterize the influence of variations in boiler radiant energy on unit load and pre-machine pressure, respectively. The introduction of the radiant energy signal M0 optimizes the traditional unit load control strategy, transforming the original single-loop control of pre-machine pressure on the boiler side into a series control system. The radiant energy control loop (inner loop) makes coarse adjustments to the boiler combustion rate, while the pre-machine pressure control loop (main loop) ensures stable alignment between unit energy demand and unit load.

Figure 5.

A load control strategy for thermal power units with the introduction of radiant energy signals.

The acoustic temperature measurement system features several notable characteristics and parameters. It provides real-time online visualization of the temperature distribution within the furnace's two-dimensional plane and includes capabilities for archiving and replaying historical temperature changes over user-defined time spans. A key component is the temperature distribution isotherm chart, which allows users to click on specific points to view temperature values, calculate average and path temperatures within the distribution area, plot temperature trend charts, and perform statistical analyses.

In terms of key parameters, the system guarantees a path temperature measurement accuracy of ±1.5% at full boiler load. It offers a two-dimensional average temperature distribution with real-time isotherm display (refer to Figure 6), providing temperature values for 24 regions and real-time data for any selected point. The sound source utilizes a pneumatic design with white noise across a specified bandwidth, employing 24 sound paths for temperature measurement via cross-correlation. Data is sampled at a frequency of 200 kHz and transmitted to the DCS port for all zones.

Figure 6.

Temperature profile.

In industrial settings, this precise path temperature measurement accuracy en-sures production stability and safety, even at full load, allowing operations to run smoothly with minimal margin for error. The wide temperature range accommodates diverse high-temperature environments, and the inclusion of 24 zones and sound paths facilitates a comprehensive representation of the furnace's internal temperature distribution.

For personnel coordination, the boiler discipline is responsible for drilling the water-cooled walls and installing wall cavities, while electrical professionals provide an AC220V power supply and lay out the electrical circuits. The thermal control disci-pline oversees the installation of communication lines. On-site testing procedures fol-lowing construction are crucial and include sound generation and reception in both cold and hot states, temperature measurements along pathways, communication checks between the local process control unit and the central processing unit in the main control room, and temperature field reconstruction via the central processing unit. These tests must be conducted sequentially to ensure system functionality.

The construction phase is scheduled to coincide with the boiler's maintenance pe-riod, lasting 30 days. Each specialized team plays a vital role: the boiler discipline pre-pares for subsequent equipment installations, the electrical discipline ensures a con-tinuous power supply and circuitry, and the thermal control discipline maintains sta-ble communication networks. A series of on-site tests will comprehensively assess the system's performance under varying conditions. By scheduling construction during the boiler's maintenance window, disruptions to regular production are minimized. The 30-day timeframe necessitates meticulous planning and efficient execution to guaran-tee timely project completion.

Figure 7 illustrates the wall temperature detection and analysis system. This paper presents a multi-point temperature measurement system for the boiler's water-cooled wall and associated equipment, featuring hundreds of temperature measurement points arranged both horizontally and vertically. This system visually reflects changes in work quality within the water-cooled wall and provides effective data sup-port for monitoring its health status through the acquisition of combustion-related big data.

Figure 7.

Wall temperature detection and analysis system.