The putrescible component of municipal solid waste (MSW) can be treated in a sequential anaerobic leach-bed process. The process consists of two reactors, one containing a fresh waste bed and the other an already anaerobically stabilised waste bed. initially leachate from the fresh waste bed is recirculated after flushing it through the stabilised waste bed. This operation is called sequencing and it enables rapid initiation of degradation in the fresh waste bed. Once a balanced microbial population has been established in the fresh waste bed, the sequencing operation is terminated and the leachate is recirculated directly upon the fresh waste bed. It has been demonstrated that this process is capable of exhausting MSW of its methane potential within 60 days. This thesis develops a criteria to operate the process and examines the means by which the sequencing operation can enhance the initiation of degradation of the fresh waste. This thesis also formulates and validates a mathematical model for the degradation process by incorporating the dynamic interactions between the microbial process, porous media and trickle flow and physico-chemical equilibrium relationships.
Specific methanogenic activities in the leachate from the fresh waste bed after spiking with cellulose, acetate and formate were measured during the duration of the process. It was found that only additions of formate and acetate showed a positive response when comparing with the control assay. Formate addition showed a greater response than that of acetate addition. Formate induced methanogenic activity was shown to provide a more reliable guideline for termination of sequencing than the criteria on pH and methane content.
The effect of supplying inoculum to initiate the degradation process was examine by filtering the leachate from the stabilised waste reactor with a 0.2 µm membrane filter before flushing it through the fresh waste bed. The experimental results indicated that initially, the microbial activity was reduced due to a lack of inoculum addition. However the rate of degradation was improved after twenty days and the fresh waste bed was exhausted of its methane potential within 60 days. This shows that the microorganisms initially present in the fresh waste are sufficient to initiate rapid degradation within this bed, and the predominant effect of sequencing is to flush out the initial inhibitory products of fermentation and supply nutrients from the stabilised waste bed.
The dynamic model consists of three packages; a hydraulic model, a physicochemical equilibrium model and a model that describes the kinetics of anaerobic biological reactions. The flow through a bed of MS W is a combination of trickle and porous media flow and therefore it cannot be modelled by conventional porous media formulations. Instead, the moisture flow through the bed was modelled as flow through a two stage stirred tank cascade system. The physico-chemical model simulates the concentration of dissociated species in the liquid. The model uses a total mass balance on H+, including biological production and consumption terms, to calculate pH. This was preferred to the conventional charge balance equation, where the derived value of H+ is subject to numerical round off error in the presence of more concentrated species.
The biological model consists of three microbial groups; which can be classified as acidogenic organisms; acetoclastic methanogens and hydrogenophilic methanogens. MSW was considered to consist of an insoluble and a readily soluble fraction. Each fraction was considered to be made up of both biodegradable and refractory components. The Contois model was used to describe the kinetics of hydrolysis. Monod kinetics was used to describe the acetic acid and methane production steps. A pH inhibition term was included in all microbial growth expressions.
The biological reaction and physico-chemical models were first used to describe methane production and COD accumulation in a stirred batch reactor. The growth parameters for acidogenic bacteria were selected as fitting parameters because the substrate for the acidogens in this study is MSW, which does not have a universal composition. The simulations showed that one set of parameters was able to predict methane production and residual COD concentrations from repeated spiked of MSW derived soluble COD. Similarly, the parameters in the Contois model were calibrated in stirred batch experiments on MSW derived insoluble COD. The biological and physico-chemical model were validated by simulating the digestion of unwashed MSW.
The hydraulic model was calibrated to fit tritium residence time distributions for a leachate recirculation rate of 10%. This model was then applied, along with the biological reaction and physico-chemical models, to predict methane generation rates and COD accumulation profiles for sequential leach-bed experiments with a recirculation rate of 10%. Close agreement was achieved between the model and the data by using the initial concentration of biomass as a fitting parameter.
The simulation of methane and COD production in an experiment with a 30% recirculation rate successfully was achieved by using the same initial microbial concentrations and the same flow parameters as that used in the simulation of the 10% recirculation rate experiment. However when these initial conditions and flow parameters were used for the simulation of a 5% recirculation experiment, the rate of COD production was under-estimated. This is because the flow model was originally calibrated for a 10% recirculation rate and as a result, the rate that solutes were flushed for lower flow volumes was under-estimated. Consequentially, the residence time in the fresh waste was over-estimated by the model which then inflated the methane production rate, since the microbial activity of the fresh bed was predicted to be high.
The model was then compared with the experiments that assessed the effect of inoculation from the mature leachate to the fresh waste bed. As observed experimentally, the model predicted that the degradation process can be initiated without the addition of methanogenic bacteria. The initial generation of methane was delayed in the simulation. This may be because the model assumed that the residence time of bacteria in the reactor are equal to the hydraulic residence time, ignoring bacterial adhesion on solid surfaces. Despite this lag, the methane predicted that the waste was exhausted of its methane potential within 60 days, the same duration as the unfiltered experiments.