The combustion of pellets is an essential source for room heating appliances and sustainable and renewable energy deployment. This is the main motivation that makes combustion technologies subject, in recent years, to constant developments and investigation. However, the formation of pollutant emissions accompanies the thermal conversion of pellets and, some environmental considerations and restrictions are becoming more restrictive, requiring preventing emissions to the atmosphere. To achieve these demands, stable and efficient combustion is imperative. Consequently, to improve biomass combustion in automated small-scale pellet boilers, this works presents a numerical model to predict the combustion behavior inside the equipment and the implementation of a Design of Experiments based on the Taguchi method to perform an experimental campaign and understand the influence of several parameters. The CFD results were able to predict the arrangement of the particles on the grate, the temperature profile of the particles, and the main gas species concentration inside of the boiler with acceptable accuracy compared with the experimental measurements. In combination with the experimental work, recommendations to reduce pollutant emissions and obtain stable combustion were indicated.
Biomass combustion is one of the most thoroughly investigated topics nowadays, and
it is a promising research area. The main reason is its complexity in terms of
conversion since it involves simultaneous multiphase fluid flow, chemical reactions, heat
(convection and radiation), and mass transfer. Within the scope of combustion in
small-scale reactors, concerning pellet boilers, represents an evolution of traditional
residential biomass systems considered inefficient and pollutants [
Copyright © 2021 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
Hence, in the last decade, some authors conducted experiments to assess the design
strategy [
Computational Fluid Dynamics (CFD) tools have been increasingly used in
optimizing the combustion process, and they have become essential in boiler design and
operation troubleshooting. Furthermore, CFD modeling can also be helpful to analyze
different working conditions and estimate a multitude of variables inside the whole
domain, particularly pollutant emissions [
In contrast, the second approach fully integrates the bed region in the CFD
simulations [
Nevertheless, the accuracy of these models is committed by their assumptions and
the physical models' capability to predict the phenomena. It is in this sense that
another approach appears. Recent developments related to machine learning algorithms and
big data provide platforms to develop advanced process data analytics to predict, for
instance, boiler performance and emissions. In this way, artificial neural networks are
becoming used as an alternative tool to simulate complex problems. The application
of this approach to small-scale boilers is still limited. Böhler et al. [
In this context, the present study aimed to develop a numerical model able to simulate the combustion process in a 20 kW pellet boiler. This study was developed using the ANSYS Fluent software, which is often used to analyze the combustion processes. However, as previously mentioned, modeling the combustion of the solid particles is not allowed, and only the combustion of the flue gases after the combustion above the bed zone is possible. Consequently, ANSYS Fluent was combined with a fully integrated packed-bed model. An experimental test was also performed to validate the results and to study the influence of the grate dimensions, excess air, the thermal load, and the split ratio of primary and secondary air on the fuel bed temperature, gas emission, boiler efficiency, and agglomeration on the grate. 2 2.1
Even before the flue gases are released into the exhaust, a gas sample is extracted from the final section of the stack using the gas analyzer apparatus. This equipment includes a sample gas pipe system from the stack, vacuum pump, cooling system, filters, and calibration gases. Multi-Gas Analyzer, model SIGNAL9000MGA, measures the gas emissions including O2, CO2, and CO. Before the experiments, the gas analyzers are calibrated in a nitrogen diluent with concentrations for CO=5000 ppm, O2=20%, and CO2=10%. Then, the vacuum pump is used to extract the sample and take it to the gas analyzer. Before entering the vacuum pump, the sample is cooled and filtered to remove moisture and particles.
In addition to all equipment used to characterize the combustion behavior inside the grate there are 4 thermocouples (K-type) installed to measure the temperature inside the fuel bed. To determine the boiler efficiency, also 2 thermocouples (K-type) are installed at the inlet and outlet of the heat exchanger to measure the hot and cold water temperature. Table 1 presents the baseline operating conditions of the pellet boiler. The boiler operates at 20 kW of thermal power with 50% excess air, and the split ratio of primary to secondary air was 37/63%.
Regarding the fuel, commercial pine wood pellets, with 6 mm diameter, certified according to the European Standard 14961-2 and ENPlus® A1, were used during the experiments. Their composition is summarized in Table 2. The ultimate and proximate analysis values and the calorific value of the pine wood samples correspond to the samples obtained after the milling process. The technical specifications of European standards were followed to perform the analyses, and a Leco TruSpec Series measured the sulfur content of the samples. To measure the heating value, a calorimeter, Leco AC500 was used. The pine wood lower heating value estimated was 17.10 MJ/kg. A set of experimental tests was planned according to the Taguchi method to analyze the influence of the most important parameters on the combustion performance, including excess air, power, primary to secondary air ratio, and grate area. These parameters were considered to be the most dominant in the combustion of the pellets. In this plan, three different levels for each parameter were selected, as presented in Table 3. If Taguchi's method was not applied in this study, the number of experiments would reach a total of 81 (34). However, with the selection of an appropriate orthogonal array, this number is highly reduced to 27.
From the experiments, several parameters are recorded, including the gas emissions (CO, NOx, O2, and CO2) and the temperature of the fuel bed measured at four different positions (5, 15, 25, and 60 mm from the bottom) in the center of the fuel bed. Furthermore, the CO emission is the parameter that better describes the combustion quality during the combustion process. Meanwhile, during long operation periods, as the CO value could fluctuate and increase significantly with the fuel bed rise, the CO emission is considered in this study as the factor for the occurrence of those instabilities in the fuel bed.
Regarding the procedure of the experiments, the operating conditions are previously determined for each run and configurated in the LabVIEW program. Then, the pellets are introduced automatically in the combustion chamber, and the ignition starts after around 95 seconds. After that, the combustion process tends to stabilize after about 8 minutes, and the experiments are carried out for more approximately 1 to 4 hours. From this period, the average values of gas emissions and temperatures are calculated. CFD simulations in combination with detailed sub-models for the solid fuel conversion have been developed regarding numerical studies. This methodology allows clear visualization of the bed arrangement and the solid particles' conversion on the grate. Consequently, the treatment of the interactions between the gas and solid phases on the surface of individual particles and the movement of the bed particles are allowed. These simulations are very useful to predict information about the pellets conversion, which is very difficult with the available experimental instruments. Consequently, to predict the arrangement and temperature profile of the particles and the main gas species concentration inside the boiler. The simulation and procedure developed in the numerical study are described in detail in [46].
Figs. 2 and 3 present an example of the temperature and gas emissions values recorded during one of the experiments where poor combustion with high CO emission and a stable combustion process occurred, respectively. The experiments were performed with the same grate area, a power of 10 kW, and the difference was the air split ratio and excess air. For the first case, the air split ratio and excess air were 37/63 and 50% and for the second case was 30/70 and 110%, respectively. This difference in the air supply, where mainly more secondary air exists than air supplied below the grate, results in a reduction of around 4.5 and 1.8 times in the average CO and NOx emissions, respectively. With this reduction, more oxygen is available inside the combustion chamber, one of the most critical factors in completely oxidizing the carbon monoxide emissions. In addition to this example regarding the flue gas emissions and temperatures, it was observed that instabilities could occur over long runs during the experiments. These were identified with a sudden rise in the fuel bed height that would lead to an increase in emissions and, ultimately, to a collapse of the combustion, which may be a consequence of the lower combustion rate (accumulation of unburned pellets on the grate). This is a reason why boiler manufacturers introduced a control strategy that periodically cleans the fuel bed. As an example of the occurrence of this problem, Fig. 2 b), after 1 hour of the experiment, presents an increase in the CO emissions that were observed due to the fuel rising in the grate.
Furthermore, as an overview of the Taguchi method, it was observed that the air split ratio contributes 21.45% to CO reduction, followed by power with 14.59%. The medium power (13 kW) results were the ones with the highest efficiency or lowest CO emissions. The excess air, grate area, and split ratio have the same tendency on the efficiency and CO emissions, where higher thermal efficiency at lower and middle values of those parameters was achieved. The average temperature values in the fuel bed indicated that the highest temperature was observed at 15 mm, followed by 25, 5, and 60 mm in height. The most important parameter contributing to the fuel temperature is the air split ratio and power. 3.2
The numerical study was performed considering the operating conditions presented in Table 1. Figs. 4 and 5 show an example of the results obtained with the numerical model developed. As can be observed by the carbon monoxide emissions, the CFD results showed that the operation of the boiler allows an effective mixing between the combustion air and the flue gases inside the combustion chamber. The mixing between the flue gases and the air inside the boiler is always significant to improve the combustion efficiency and reduce pollutant emissions. As shown in Fig. 4, due to the penetration of the secondary air nozzles, there is a sudden increase in the temperature due to the mixture of the flue gases with the air supplied by the nozzles. Therefore, the configuration of the air jets is vital to improve the combustion process. However, the release of the flue gases is more evident in the region near the middle of the grate because, in this place, there are fewer particles and, therefore, the flow resistance is lower.
The CFD results, compared with the experimental data, present a generally good agreement since they are close and within the maximum and minimum values. However, the CO concentration predicted is lower than the experimental observation, probably due to the simplicity of the global combustion reaction scheme employed in CFD simulation. More details about the CFD model and results can be found in [46]. a) b) This paper presented an analysis of the combustion behavior of a 20 kW prototype pellet boiler through an experimental campaign and a computational model. The main results and key findings led to the conclusion that: The combination of the experimental and numerical work allowed a deeper understanding of pellets combustion and provided a tool for improving the efficient use of biomass and decrease pollutant emissions. The set of experiments were very useful to better understand the different operating conditions of a commercial pellet boiler and obtain information to validate the numerical model developed to evaluate the combustion performance of the equipment. During the experiments, it was observed that instabilities could occur over long runs. These were identified with a sudden rise in the fuel bed height that would lead to an increase in emissions and, ultimately, to a collapse of the combustion. From the Taguchi plan of experiments, it was possible to understand with a reduced number of experiments the influence of the four parameters identified with the most impact in the combustion performance. From the numerical modeling perspective, the main assumptions, geometric model and its discretization, methodology, and physical models employed seem correct. The prediction of the main variables of the solid and gas phase and the temperature of the flue gases above the grate where the pellets conversion takes place has been made with a reasonable level of accuracy.
The combination of the experimental research and the development of the numerical model is expected to reduce pollutant emissions and costs, leading to the improvement of the process used in commercial pellet boilers. Also, as future work, the data obtained during the experiments will be applied to machine learning algorithms to analyze the conditions inside the prototype pellet boiler and anticipate the appearance of specific phenomenon that results in maintenance actions or high emissions.
This work has been supported by the Portuguese Foundation for Science and Technology (FCT) within the R&D Units Project Scope UIDB/00319/2020 (ALGORITMI) and UIDP/04077/2020 (METRICS). The first author would like to express his gratitude for the support given by the FCT through the PhD Grant SFRH/BD/130588/2017. 41. Bacci di Capaci, R., and Scali, C., 2020, “A Cloud-Based Monitoring System for Performance Assessment of Industrial Plants,” Ind. Eng. Chem. Res., 59(6), pp. 2341– 2352. 42. Muhammad Ashraf, W., Moeen Uddin, G., Muhammad Arafat, S., Afghan, S., Hassan Kamal, A., Asim, M., Haider Khan, M., Waqas Rafique, M., Naumann, U., Niazi, S. G., Jamil, H., Jamil, A., Hayat, N., Ahmad, A., Changkai, S., Bin Xiang, L., Ahmad Chaudhary, I., and Krzywanski, J., 2020, “Optimization of a 660 MWe Supercritical Power Plant Performance—A Case of Industry 4.0 in the Data-Driven Operational Management Part 1. Thermal Efficiency,” Energies, 13(21). 43. Strušnik, D., Golob, M., and Avsec, J., 2015, “Artificial Neural Networking Model for the Prediction of High Efficiency Boiler Steam Generation and Distribution,” Simul. Model.
Pract. Theory, 57, pp. 58–70. 44. Ilamathi, P., Selladurai, V., Balamurugan, K., and Sathyanathan, V. T., 2013, “ANN–GA Approach for Predictive Modeling and Optimization of NOx Emission in a Tangentially Fired Boiler,” Clean Technol. Environ. Policy, 15(1), pp. 125–131. 45. Korpela, T., Kumpulainen, P., Majanne, Y., and Häyrinen, A., 2015, “Model Based NOx Emission Monitoring in Natural Gas Fired Hot Water Boilers,” IFAC-PapersOnLine, 48(30), pp. 385–390. 46. Silva, J., Fraga, L., Ferreira, M. E., Chapela, S., Porteiro, J., Teixeira, S. F. C. F., and Teixeira, J., 2018, “Combustion Modelling of a 20 KW Pellet Boiler,” Volume 6B: Energy, ASME, p. V06BT08A036.