UBC Theses and Dissertations
Experimental study of the effect of particle size distribution on heat transfer within the bed of a rotary kiln Dhanjal, Sanjiv K.
Rotary kilns are some of the largest pieces of equipment found in industry. Their resilience and versatility makes them suitable for a wide variety of applications, such as the calcination of lime and alumina, reduction of iron ore, cement clinkering and solid waste incineration. Common to all of their uses is the need to thermally process a material, typically particulate, by directly firing a fossil fuel. A kiln consists of a refractory lined cylindrical furnace with its axis at a slight angle to the horizontal. Raw material is fed at the upper end and, by virtue of the kiln's rotation, it travels along its length. The burner is located at the lower end. As such, the kiln is a counter current heat exchanger, with hot combustion gases flowing upwards over a bed of particulate material traveling down. To ensure efficient and quality processing, all of the material in a kiln needs to be heated up to temperature, without any of it overheating. This requires efficient heat transfer from the combustion gases to the bed and efficient heat transfer within the bed, to ensure all of the material is evenly heated. However, this is not always achieved. For example, it has been found in industry that fine material often remains unreacted. The reason has been attributed to the phenomenon of segregation, whereby the fines collect towards the center of the bed and so are shielded from the hot combustion gases. Experiments on laboratory kilns have confirmed that segregation in the cross section of a kiln occurs rapidly. Within a few revolutions, the fines collect in the middle of the bed forming a core, often referred to as a "kidney". Since a kiln typically rotates at 1-2 rpm and material residence time is 1-4 hours, a bed that contains fines will be segregated for almost all of its time in the kiln. The fines will not be exposed to the hot gases and so may not react. On the other hand, a bed that contains mono-sized particles has been found to intimately mix. In such a bed all of the particles have an equal chance of being exposed to the hot gases. Since heat transfer within the bed is dominated by material advection, the bed is likely to be isothermal. From this it has been presumed that segregation is the cause of temperature non-uniformity in the bed cross-section. While there has been no experimental data published confirming this, modeling work has concurred with the premise that segregation is the cause of temperature non-uniformity in the bed cross section. The aim of this study is to investigate this through experimentation and to determine what level of fines cause temperature non-uniformity in the cross section of the bed. To meet the study objectives a batch rotary kiln was designed. The furnace was lined with a dense refractory that, when hot, was used as the heat source for the bed. This would allow the heat rate to be calculated using radiation heat transfer theory. The furnace was first heated using a gas flame. When it was hot, material was inserted and ,the furnace rotated. The temperatures of the walls and in the cross section of the bed were measured using thermocouples and a data logger. Sand was selected as the bed material since it is inert. A number of sand mixes with various levels of fines were made by using a sieve shaker. To quantify the level of fines the particle size standard deviation was calculated. This allowed the results to be extrapolated for other applications such as in industry where wide ranges of particle size distributions are common. The design of the furnace was carried out by the use of models written specifically for this project. A one-dimensional model of the heat transfer between combustion gases and the furnace walls was written and used to ensure that the furnace could be heated using a gas flame. It was then modified to ensure the walls could in turn heat up the bed material. A two-dimensional model of the bed cross section was also written and used output data from the one-dimensional model to determine whether the temperature profile in the bed cross section could be measured practically. The models were used to design the pertinent furnace specifications. These were furnace I.D. = 400mm, sand bed depth = 100mm, lining = 70mm thick castable and insulation = 20mm thick blanket. These specifications formed the basis of the furnace mechanical design and were not altered once the experiments were started. Therefore, the models fulfilled their purpose. The furnace was heated until the refractory hot face was 1100°C before the flame was extinguished. Sand was then inserted into it and the furnace rotated. Each test lasted for about 20 minutes and the bed was heated up to about 800°C. A temperature profile was subsequently obtained at select locations in the bed cross section. The number of bed thermocouples was limited to four, since they tended to agitate the bed, affecting heat transfer. Since the locations were the same for each test, their range of temperatures was used to quantify the temperature non-uniformity in the bed. Using these temperatures, together with temperatures of the walls, the heat rate to the bed and corresponding heat transfer coefficient were calculated for each run. In all runs it was found that the spread of the temperatures in the bed was between 100-200°C initially. The range dropped to less than 10°C after 20 minutes. The temperature data was used to calculate heat transfer coefficients and these were in the range 85-175 W / m 2K. A further interesting finding was that the rate of heat transfer across the covered bed surface was 3-6 times the rate across the exposed bed surface. This highlights the importance of conduction between the covered wall and particles adjacent to it. These data represent a significant contribution to the literature since very few studies are available at realistic kiln temperatures and for kilns with diameters above 0.25m. To determine the effect of size distribution, the average range of temperature, heat rate to the bed and heat transfer coefficient were calculated for one minute time periods during each test. The values were then plotted with the standard deviation of particle size. Much scatter was observed in these plots and it appeared that size distribution had little effect on temperature uniformity in the bed and no effect on heat rate and heat transfer coefficient. The results were somewhat surprising, since segregation was evident during the some of the runs. Finer particles could be seen underneath the layer of larger particles, below the bed surface. Therefore, for the size ranges studied segregation did not hinder heat transfer across the bed. Perhaps heat transfer is a stronger function of overall material flow and effective thermal diffusivity, as opposed to segregation. This study provides a substantial amount of data at realistic kiln temperatures, something that is rare in the general literature. The data can be used for validating models of the heat transfer in the cross section of the bed, and so further understanding in this area.
Item Citations and Data