Research on The Dynamic Model and Transient Operation Optimization Control of Coal Mill Pulverizing System

The coal mill pulverizing system plays a crucial role in the power plant’s boiler operation. Its primary function is to convert raw coal into pulverized coal, which ensures efficient combustion in the boiler. A typical pulverizing system comprises several key components: the raw coal bunker, coal feeder, coal mill, separator, pulverized coal bunker, dust exhauster, and exhaust ductwork.

Initially, raw coal is uniformly fed from the raw coal bunker to the coal feeder via a vibrating feeder. The coal feeder regulates the flow of raw coal according to the boiler’s combustion needs, maintaining a steady and consistent supply to the coal mill. Once in the coal mill, the raw coal undergoes grinding. Various types of mills, such as ball mills, vertical mills, and air-swept mills, are commonly used for this purpose. The fundamental principle behind these mills is the mechanical grinding of coal chunks into finer particles.

During the grinding process, grinding rollers (or steel balls) within the mill exert pressure on the coal, breaking it down into smaller pieces as they rotate against a grinding disc or cylinder. The goal is not only to achieve thorough grinding but also to minimize energy consumption while maximizing grinding efficiency. Once the coal is pulverized, it is mixed with hot air from the boiler’s primary fan, exiting the mill through the air ring or the dust exhauster.

The air-coal mixture then enters a separator, where the coal particles are classified based on size. Coarser particles are directed back to the coal mill for further grinding, while finer particles are sent to the pulverized coal storage bin via the powder discharge pipeline.

Inside the separator, centrifugal force generated by a cyclone separator helps separate the fine and coarse particles. The coarse particles return to the mill by gravity, and the fine particles are carried by the airflow to the discharge pipeline. The finely pulverized coal is then stored in the pulverized coal bunker and, when needed, transported to the boiler for combustion via the coal discharge mechanism.

Figure 1.

The schematic of working process of the coal mill pulverizing system

The efficiency of the pulverizing system has a direct impact on both the combustion performance of the boiler and the overall operational economics of the power plant. As a result, investigating and optimizing the dynamic model of the coal pulverizing system is crucial for enhancing operational efficiency and minimizing energy consumption in power plants.

MPS coal mill is a medium-speed coal mill with compact structure and high grinding efficiency, which is widely used in electric power, metallurgy and chemical industry. MPS coal mill is mainly composed of grinding disc, grinding roller, pressurizing device, separator and transmission device. its Composition is shown in Figure 2.

Figure 2.

The schematic diagram of MPs coal mill structure

The grinding disc serves as the central component of the MPS coal mill and is located at the lower section of the machine body. Raw coal is fed into the millstone via a coal inlet positioned centrally on the disc, where it is spread outward by centrifugal force. The grinding disc is surrounded by three grinding rollers that are symmetrically placed around its perimeter. These rollers apply pressure, facilitated by a pressurizing system, to crush and grind the coal between the rollers and the disc. This process reduces the coal into fine pulverized particles. Typically, the pressurizing mechanism for the grinding rollers utilizes a hydraulic system, which automatically adjusts the pressure in response to load variations, ensuring stable coal grinding performance.

Above the grinding disc is the separator, which plays a key role in separating finely ground coal from larger, unprocessed coal chunks. The gas-coal mixture is directed into the separator via nozzles located at the outer edge of the grinding disc. Here, centrifugal force and gravity act to return the coarse particles to the grinding disc for further processing, while the fine coal particles are carried away by the airflow and exit through the powder discharge system.

The drive mechanism of the MPS coal mill comprises a motor, gearbox, and coupling, among other components. The motor transmits power to the grinding disc through the gearbox, enabling it to rotate at the required speed. The design and construction of the drive system must meet high standards to ensure transmission efficiency and reliability, which are crucial for the long-term continuous operation of the coal mill. The grinding process can be represented by the following equation: (1) Pm=Qchcηc

n this formula, P_m represents the power consumption of the coal mill, Q_c is the coal mass flow rate, H_c denotes the specific grinding energy of the coal, and η_c refers to the mill's efficiency. By optimizing these parameters, the coal mill’s operational performance and grinding efficiency can be enhanced, leading to overall system improvements.

The mechanism model is built upon a thorough understanding of the physical properties and operational parameters of the pulverizing system. A set of mathematical equations is formulated to characterize the dynamic behavior within the system. Typically, the mechanism model of a pulverizing system incorporates fundamental equations such as mass, energy, and momentum balances.

To begin with, the mass balance equation for pulverized coal illustrates the variation in coal mass within the mill. Its basic form is expressed as follows: (2) dMdt=Win −Wout

Where M represents the mass of pulverized coal within the mill, W_in is the mass flow of raw coal entering the mill, and W_out is the mass flow of pulverized coal exiting the mill. This equation captures the accumulation and discharge of pulverized coal within the system.

Next, the energy balance equation models the heat transfer and conversion processes within the system. The energy balance in the coal mill can be described as: (3) dEdt=Qin −Qout +Qgen

Here, E denotes the total energy in the system, Q_in and Q_out represent the heat entering and leaving the system, respectively, while Q_gen is the heat produced during the coal grinding process. This equation reflects the heat dynamics and equilibrium in the coal mill.

Finally, the momentum balance equation describes the movement of the airflow and pulverized coal within the mill. Its basic formulation is as follows: (4) d(mv)dt=∑ F

In this equation, m is the mass of pulverized coal, v is its velocity, and ∑F is the total force acting on the pulverized coal. This equation is used to describe the distribution and flow behavior of the pulverized coal in the mill.

Neural Networks (NN) are widely used and effective data-driven methods in the dynamic modeling of milling systems. With their powerful self-learning and self-adaptive capabilities, they can accurately describe the complex nonlinear relationships and dynamic processes in milling systems.

A neural network model typically consists of an input layer, one or more hidden layers, and an output layer. The input layer receives the input signals from the milling system, while the hidden layer uses nonlinear algorithms to process these input signals. The output layer generates the system's predicted results. The training of the neural network model often requires a large amount of historical and experimental data to optimize the weights and biases in the hidden layer's nonlinear algorithms, ensuring that the model's predicted output accurately reflects the system's dynamic response.

As showed in Figure 3, the neural network model's output is represented as: (5) y=f(Wx+b)

Figure 3.

The schematic diagram of typical BP neural network.

In equation (5), y represents the output vector, x represents the input vector, W represents the weight matrix, b represents the bias vector, and f is the nonlinear activation function. The network's weights and biases are optimized continuously through the backpropagation algorithm with the goal of minimizing the prediction error. In the milling system modeling process, neural networks can be used to predict key parameters such as the mill outlet temperature, the fineness of the coal powder, and other critical indicators.

In the dynamic modeling of pulverizing systems, both mechanistic models and data-driven models present their own unique advantages and limitations. Mechanistic models, rooted in the physical and chemical properties of the system, provide a solid theoretical foundation and explanatory power. However, they may struggle when addressing complex nonlinear relationships or unaccounted-for factors. In contrast, data-driven models, which rely on historical data, excel in capturing the nonlinearities inherent in complex systems through machine learning techniques. Yet, their accuracy and reliability can suffer if data is incomplete or noisy. To leverage the strengths of both approaches, combining mechanistic and data-driven models can significantly enhance the model's prediction accuracy and robustness.

This paper adopts a mechanism model updated with a data-driven approach. The data-driven model is used to adjust the mechanism model, compensating for the nonlinear and dynamic aspects that are either challenging to describe or omitted in the mechanistic framework.

The process can be outlined in the following steps: 1)

Development of the Mechanistic Model: The first step involves constructing the mechanistic model of the coal mill, based on its physical and chemical characteristics. This model includes mass and energy balance equations, as well as dynamic equations. Key parameters and boundary conditions are identified through experimental data and parameter calibration. The model’s accuracy and stability are then verified through simulations.

Where △P represents power fluctuations, P_peak is the transient peak power, and P_steady denotes the steady-state power. Power fluctuations not only increase system energy consumption but may also cause frequent equipment start-ups and shutdowns, leading to accelerated wear and tear.

Additionally, changes in pulverized coal fineness during transient operations present another significant issue. The fineness, which directly impacts boiler combustion efficiency and operational stability, can be quantified by the particle size distribution of the pulverized coal. The standard deviation σd of the particle size distribution is calculated as follows: (7) σd=1N∑i=1N(di−d¯)2

Where d_i is the diameter of the i−th particle, and d is the average particle size. Instabilities in coal fineness can result in incomplete combustion and reduced boiler efficiency.

Finally, system stability during transient operation is crucial. To assess system stability, gain margin (GM) and phase margin (PM) are commonly used. These can be defined as: (8) GM=1| G(jωc) |PM=180∘+∠G(jωc)

Where G(jw_c) represents the amplitude of the open-loop transfer function of the system at the crossover frequency wc, and <G(jw_c) denotes the phase angle. Maintaining appropriate gain and phase margins is essential to ensure system stability during transient states and prevent instability. Therefore, designing effective transient operation control strategies is necessary to improve both the stability and efficiency of the system.

At time t=1000 s, the moisture content of the raw coal (Mar) increases abruptly by 25%, with all other operating conditions remaining constant. Figure 6 illustrates the transient changes in the outlet air temperature (tm) under both the original and optimized control logics. Figure 7 displays the energy consumption variations in the pulverizing system when dealing with the sudden moisture fluctuation using the original and optimized logics.

Figure 6.

Transient change in outlet temperature of the coal pulverizer when the raw coal moisture increases suddenly.

Figure 7.

Energy consumption changes under different control strategies when raw coal moisture suddenly increases.

From the comparison of the curves in the figures, it is evident that the optimized control logic enhances the temperature regulation of the outlet air powder from the coal mill, reducing the fluctuation in the air powder temperature by approximately 48%. Additionally, it leads to a more than 2% reduction in pulverizing energy consumption. The primary reasons for these improvements are as follows: under the optimized logic, the increased moisture in the pulverized coal is fed forward to the hot air inlet, causing a rapid increase in the hot air volume. This adjustment prevents a significant drop in the outlet air temperature. Furthermore, the increase in hot air volume reduces the overall primary air volume required to maintain temperature, thereby decreasing electrical and mechanical consumption in the primary air system. As a result, there is a reduction in the unit energy consumption for pulverizing.

Initially, the raw coal flow rate from the feeder is 10 kg/s. At time τ=1000 s, this flow rate increases by 50% instantly, with other operating conditions unchanged. Figure 8 shows the transient trend of the outlet air temperature (tm) when the original and optimized control logics are applied. Figure 9 illustrates the variation in energy consumption (Wt) when the coal feed rate is increased using the two control strategies.

Figure 8.

Transient variation in outlet temperature of the coal mill when the coal feed rate is increased suddenly.

Figure 9.

Operational energy consumption under different control strategies when the coal feed rate is increased suddenly.

The optimized control logic leverages the feed-forward capability of the primary air mixing temperature, enhancing the control response to the increased coal feed. By directly adjusting the primary air mixing temperature setpoint in response to the higher coal feed rate, the coal mill's cold and hot air regulation is more effectively managed, leading to an improved system response. This results in reduced fluctuations in the outlet air temperature (tm) and a reduction in operational energy consumption.