Jennifer Sinclair Curtis, PhD , University of Florida
We have shown that the addition of bulking additives to the biomass initially charged to the fermentor can significantly enhance the rate of methane production. We hypothesize that this increase in methane production is due to an increase in bed permeability during decomposition. We are performing experiments in a laboratory scale fermentor to test this hypothesis. We have also developed a complementary discrete element method (DEM) code and are performing simulations to study the effect of particle shape on particle packing, with an aim to explore the effect of bulking particles on the overall packing characteristics. Below gives a summary of our progress on both of these fronts:
In the first year of the project, we have varied the concentration of the bulking additives in the digester. The feedstock was sugar beet tailings shipped in cooled buckets of 20 L volume from American Crystal Sugar. On receiving, the sugar beet tailings were taken out from the cooler, washed and subjected to storage in a cold storage room maintaining negative 20 degrees C. An anaerobic digester is made from inverting a 20 L pyrex glass carboy jar with the lid made of aluminum and sealed with a rubber gasket. A perforated plate is placed in the bottom of the inverted bottle to prevent solids from settling down in the reactor. Heating is provided by coiling heating tape around the glass bottle. A temperature of 55 degrees C was controlled by software in conjunction with the data logger and control module. The sugar beet tailings were thawed before loading them into the digester. For each digestion run, 3 kg of sugar beet tailings were used. The inoculum was added to the digester to generate a working volume density of 250 kg/m3. Buffer (sodium bicarbonate) of 5 g/L is added to the digester to maintain the pH range of 7-8.5. A recirculation rate of 100 ml/min, from bottom to top, is also maintained during the anaerobic digestion.
Three runs were performed - one run with no bulking agents (lava rocks) and two with varying layers of lava rocks. For the run involving three layers, three layers of bulking agent and 3 layers of sugar beet tailings were arranged alternatively over the solid holding mesh of the digester. The first layer over the mesh was that of the bulking agent. On top of it, sugar beet tailings were spread, such that the bulking agent was visible. The sugar beet tailings were then covered with bulking agent, with sugar beet tailings slightly visible. For the run involving six layers, the same pattern was used. The runs with bulking agent showed increased production of methane over the run with no bulking agent. In addition, increased concentration of bulking agent showed an increase in methane yield. The runs are currently being duplicated for reproducibility and error analysis.
In this first year of the project, we have developed a discrete element method code to simulate particle packing for non-spherical, elongated particles matching the approximate physical dimensions of the biomass. To validate the packing simulation predictions, model particles of soda lime glass beads packed in cylindrical containers of varying diameter and height. Image analysis of the height of the measured and predicted particle bed, taken over four views, is used to determine the particle packing fraction. Two types of representations of elongated rod-like particles were generated in the DEM modeling. One of them is referred to as a glued-spheres particle, which is formed by gluing identical spheres in a line; the other one is a perfect cylindrical particle, which is a representation of true cylinder with a cylindrical band and two flat end surfaces. In the DEM simulations, the translational and rotational motion of an individual glued-spheres particle or cylindrical particle is governed by Newton’s second law of motion. For both glued-spheres and true cylindrical particles, the Hertz theory is used to model the normal forces, and the Coulombic sliding friction model is adopted for the tangential forces. Particles are always generated and dropped from the same point with random orientations. Particles drop under the influence of gravity with zero initial translational velocity and zero angular velocity.
Thus far, we have investigated the effects of drop height, fill height and container size on the packing fraction. For example, an increase in drop height results in higher solid volume fraction due to higher impact force, and particles pack more densely. We have also investigated the effect of parameters in the contact model and the number of glued-spheres necessary to represent an exact cylinder. As the number of constituent glued spheres increases, the predicted packing fraction more closely matches that of the true cylinder and the experimental value. As the Young’s modulus increases, the contact stiffness increases and particle overlap decreases, resulting in a decrease of the packing fraction. Now that we have a validated code, we will be exploring the effect of particle deformation by allowing for flexibility in the spring that connects the constituent glued-sphere particles which describe the non-spherical biomass particle.