Elsevier

Energy

Volume 81, 1 March 2015, Pages 462-470
Energy

Mechanistic modeling, numerical simulation and validation of slag-layer growth in a coal-fired boiler

https://doi.org/10.1016/j.energy.2014.12.058Get rights and content

Highlights

  • The phenomenon of recoiling of a molten ash droplet after impaction was adopted to simulate slag-layer growth.

  • Bouncing potential model was incorporated in a computational fluid dynamic package.

  • The simulation results of two Indian coals were compared with field data and it was inferred that the model qualitatively predicts growth profiles.

  • Smaller particles, and hence the thermophoretic transport mechanism, are found to dominate deposit formation and its growth.

Abstract

In a tangentially coal-fired boiler, for locations inside and near the combustor, heat-transfer by radiation is significant, and hence, ash particles arrive in molten state. The aim of the present study is to adopt a mechanistic modeling approach which incorporates energy-conservation principles to address slag-layer growth. In order to determine the outcome of molten ash impaction, a mechanistic bouncing potential model, incorporating the phenomenon of recoiling of molten ash droplets after impaction, is employed. The bouncing potential is a representation of the excess energy possessed by the recoiling splat, and is used to determine the outcome of molten ash impaction – to stick or to bounce. Computational fluid dynamics techniques, incorporating the effect of thermophoresis, are adopted to estimate the arrival rate of ash particles, and the bouncing potential model, as a user-defined function, is incorporated in the simulation package to determine the status of the droplets after impaction. Two coals of Indian origin are simulated for slag-layer growth for a period of 100 min. The simulation results, when compared with field data provided by BHEL-Trichy, indicate that the model qualitatively predicts the growth of slag-layers. It has been further inferred that smaller particles dominate deposit formation and its growth.

Introduction

Deposition of ash particles on heat-transfer surfaces in a coal-fired boiler – termed as fouling and slagging – is one of the main reasons for unscheduled boiler shut-downs. Fouling and slagging deposits offer more resistance to heat-transfer, interfere with the aerodynamic flow of flue gas, reduce the available flow area, and enhance corrosion and erosion of boiler tubes [1]. Hence, developing an understanding of and predictive ability for fly ash deposition are essential to avoid unscheduled power outages and to schedule boiler maintenance. Deposition of molten ash particles – termed as slagging - results in fused deposits, and is identified in the radiant sections of the boiler, where locations are directly exposed to flame radiation and are the hottest parts of the boilers. Though many studies have been published [2], [3], a recent study by Weber et al. [3] made two significant observations regarding a lack of reliable predictions for ash deposition. First, they proposed that ash deposition needs to be addressed in a more scientific manner than employing a variable particle sticking probability whose value is typically tuned to match model predictions with measured values. Second, they noted that not enough publications are available in literature that are based on computational fluid dynamic modeling to address slagging in a boiler. Such suggestions, and an urgent need for effective utilization of high-ash Indian coals, have motivated this study of the slagging propensity of Indian coals.

In order to calculate sticking probability, conventionally, viscosity of arriving ash particles has been used as a relevant parameter [4]. Typically, the actual viscosity is compared with a critical viscosity to arrive at a conclusion regarding fraction of ash particles that sticks [5]. The sticking probability in terms of viscosities is expressed as:ηsticking=μcriticalμactualμactual>μcritical=1μactual<μcritical

The above expression implies that for viscosities lower than the critical viscosity, the sticking probability is 1; for viscosities greater than critical viscosity, the sticking probability is the ratio of critical to actual viscosities.

Richards et al. [6] followed the critical viscosity-based approach to estimate the sticking co-efficient of impacting molten ash particles. With the aim of studying the effect of boiler operating conditions and coal composition, the viscosity-based model was incorporated into the combustion model. Simulations were performed for critical viscosity values of 103, 104, and 105 Pa.s, and no significant difference in deposition profiles were observed. Fan et al. [7] adopted the critical viscosity model along with flow and combustion models to simulate the flow field, temperature field, particle transport, and deposition in a 300 MW W-shaped pulverized coal-fired boiler. They employed a critical viscosity of 1 × 105 Pa.s and observed that the deposit growth rate was high where impact probability and surface temperature were high. While reducing the critical viscosity from 108 to 106 Pa.S, Rushdi et al. [8] observed a decrease in deposition rate on a pilot-scale furnace of capacity 200 kW; hence, it was decided to use a critical viscosity of 108 Pa.s. Walsh et al. [5] reported that for iron-lean ash particles, the particles were completely sticky for a viscosity of 8 Pa.s. In another investigation of coal ash deposition on a superheater tube, it was inferred that the deposition starts when the ash viscosity was less than 6.7 × 109 Pa.s [9]. Apart from viscosity, surface properties and static contact angle were employed to estimate the sticking co-efficient of impacting ash particles [10]. Based on this study, a numerical slagging index was derived by relating ash viscosity, ash fusibility, and ash loading to predict the slagging potential. The authors reported that the predictions in this study assuming a critical viscosity of 108 Pa.s were in good agreement with the experimental results. The mechanisms of both viscosity and surface tension-driven growth were considered by Erickson et al. [11], and a slagging algorithm was proposed with a critical viscosity of 105 Pa.s. Computational fluid dynamics techniques were also adopted to understand slagging deposits [12]. The authors performed numerical simulations on slagging characteristics under different operating conditions in a 300 MW Foster Wheeler arch-fired boiler. They equated sticking co-efficient to the ratio of critical to actual viscosities, and integrated this with the combustion model.

Walsh et al. [13] considered 8 Pa.s as the critical viscosity. Senior and Srinivasachar [4] suggested that the critical viscosity was between 104 and 108 Pa.s. Huang et al. [14] employed a value of 104 Pa.s in their study. Fan et al. [7], Fang et al. [12], Lee and Lockwood [15], Richards et al. [6], Costen et al. [16], and Erickson et al. [11] utilized 105 Pa.s in their simulations. Rushdi et al. [8] and Degereji et al. [10] adopted a critical viscosity of 108 Pa.s while Yilmaz and Cliffe [9] reported ∼109 Pa.s. It may be inferred that there is a lack of consistency in choosing one critical viscosity value to simulate slagging deposits. The reported critical viscosity values vary from as low as 8 Pa.s to as high as 109 Pa.s. Hence, the viscosity-based approach requires further study to determine one critical viscosity value acceptable to all.

Due to the inconsistency, scientists have focused their attention on developing models that incorporate the physical phenomena involved during molten ash deposition. To align with that, Mao et al. [17] studied the drop dynamics – spreading and recoiling – of sucrose in water samples in the context of molten ash deposition encountered in the Kraft recovery boiler used in the pulp and paper industry. They coined the term “bouncing potential” and defined it as the excess energy possessed by the recoiling droplet. With such a definition, it was easy to set a criterion for a droplet to bounce, i.e., when the bouncing potential is positive. In a subsequent study, Mao et al. [18] suggested that the effect of rebounding was significant in the Kraft recovery boiler. With the aim of developing a computational sub-model, Mueller et al. [19] studied the deposition behavior of alkali-rich molten ash. The physical phenomena considered by them were: particle melting, impaction, sticking, and rebounding. From their experimental studies, it was understood that for small-sized particles, a molten fraction of 15% and above was sufficient to stick the particles directed toward the surface; however, the physical phenomenon of rebounding determined the sticking probability of molten particles. Ni et al. [20] analyzed the bouncing potential model developed by Mao et al. [17] for predicting slag deposit formation in coal gasification systems. They inferred that the maximum spread diameter of the slag droplet– precursor to recoiling - was the key parameter in modeling the bouncing potential. The temperature effect was reflected through the change in slag viscosity. With the aim of developing a sub-model to incorporate retraction phenomena in understanding slagging deposits, Balakrishnan and Nagarajan [21], developed a bouncing potential model theoretically by considering the shape change from cylindrical-rimmed splat to sessile drop at equilibrium, and viscous dissipation while recoiling.

A paradigm shift from viscosity-based sticking probability to the rebounding phenomenon in terms of bouncing potential has taken place over the years. An approach based on the rebound phenomenon incorporates all the physical parameters involved during drop deposition or slagging – viscosity, impact velocity, impact angle, particle diameter, density, surface tension, contact angle – in contrast to the viscosity-based approach. In addition to that, considering the retraction phenomenon is a reliable procedure compared to the highly variable viscosity-based approach. Hence, in this study, the rebound phenomenon modeled mechanistically in terms of the bouncing potential developed by Balakrishnan and Nagarajan [21] is incorporated with computational fluid dynamics to simulate slagging deposits. Instead of a sticking co-efficient, the model employs a rational, phenomenological criterion to decide the status of the droplet after impaction. The simulated results are validated with field data provided by BHEL engineers.

Section snippets

Deposition mechanism

Deposition of molten ash particles is termed as slagging. Studies have demonstrated that the retraction phenomenon of a recoiling droplet after impaction plays a vital role in determining the outcome – to stick or to bounce – of molten ash impaction [17], [19]. Impact-driven spreading occurs when a molten ash particle collides with the heat-transfer surface. Fig. 1 schematically depicts the mechanism of molten ash deposition with the assumption of no fingering and splashing. A molten ash

Model description

In a tangentially coal-fired boiler, coal along with air is introduced into the furnace tangentially to an imaginary circle in the center of the furnace. The circle consists of flue gas and the fly ash particles, and expands along the geometric axis of the boiler. The present computational fluid dynamics model focuses on tubes placed on the wall of the boiler. Fig. 3 shows the imaginary circle at the center of the furnace, the computational domain considered in the study, and the meshing

Experimental details

The field data of heat flux were measured on a water-wall tube by a heat-flux sensor placed on the rear wall at a location 3 m from the burner zone. Experiments were conducted on the boiler (Unit 3) of a 210 MW thermal power plant located in Raichur, India by BHEL engineers. Heat flux data were recorded for two Indian coals for a period of 100 min. The boiler and burner zone dimensions, and location of the heat-flux sensor on the rear wall, are shown in Fig. 5. The performance parameters of the

Results and discussions

The algorithm adopted for simulating molten ash deposition is schematically represented in Fig. 6. By solving momentum balance, energy balance and discrete second phase equations, the arrival rate of molten ash droplets was estimated. In addition to rate of arrival, the diameter, temperature, impact velocity, and impact angle of the impacting particle can be obtained. The user-defined function will be invoked when a molten ash particle touches the surface, and sticking probability of the

Conclusions

A mechanistic bouncing potential parameter to determine the outcome of molten ash impaction is incorporated in a computational fluid dynamic model in order to simulate molten ash deposition in the radiant section of a tangentially coal-fired boiler. The field data provided by BHEL Trichy, with experiments conducted on a 210 MW thermal power plant, were compared with the simulation results. Two Indian coals from the central part of India having different ash content were fired and simulated. It

References (38)

Cited by (0)

View full text