Abstract

This work deals with the determination of a kinetic model in an anaerobic digestion process. It is applied to the organic stabilization of the solid fraction of piggery slurry. The substratum results after filtering the farm wastewaters through a 1 mm mesh hydraulic sieve. Two pilot scale reactors were used, one digester was working in layered way — without agitation — and other was mixed with a stirrer. The experiments were carried out in a discontinuous process, 60 days hydraulic retention time, in the mesophilic interval (35°C). Average influent concentration was 68 and 97 g l⁻¹ COD respectively for the non-stirred and stirred reactor, whereas the achieved reduction was 61 and 65%. Organic loading rates were 0.80 and 1.45 K VS m⁻³ d⁻¹ . Specific biogas output was more efficient in the stirred reactor than in the stratified one. Several kinetic models were tried out in order to represent the methane production. A first order kinetic model applied in two stages was finally adapted for both reactors. The first stage presented the microbial growing as the limiting step, whereas the second stage was limited by the substratum availability. The effect of mixing on the kinetic parameters was analyzed. Significant differences were attained in the coefficients, thus K was 0.048 and 0.75 d⁻¹ respectively for the non-stirred and stirred reactor.

1. Introduction

A previous separation of the liquid and solid fractions is a physical operation commonly used to deal with the pig slurries. The main aim is to facilitate their transportation through the digesters, tanks or lagoons. Moreover, an elimination of most of the suspended particles is required in several kinds of anaerobic reactors. These biological systems usually operate under limitations in total solids concentration, in order to process the organic matter. As a consequence, the solid fraction is often scattered directly on the ground, without a proper treatment, which gives rise to contamination. So, more improvements should be done about organic stabilization of this substratum before the final disposal.

The standard separation of the solid fraction consists of specific hydraulic sieves or centrifuges. Although the solids content related to the total effluent volume is small (approximately 5%), disposal of this waste is a problem in intensive farms. Studies on the degradation of this fraction of the slurry are scarce. Some research was made by Górecki et al. [1], with previous centrifugation of a piggery slurry before the anaerobic digestion. Several works related to the digestion of the solid fraction were presented by Cechi et al. [2], Iniguez et al. [3], Pescod et al. [4], and Zadrazil and Puniya et al. [5]. Rodríguez [6] presented studies on the effect of the particle size for pig slurry in an anaerobic process, often difficult to be degraded.

Anaerobic digestion is commonly applied in organic stabilization of piggery manure. Hydrolysis and liquefaction of insoluble organic complexes are key steps in the process. Parkin and Owen [7] proposed that this phase allows the organic small size particles to go through the cell membrane, supplying energy resources to the bacteria. The hydrolysis stage is considered the limitating step of the anaerobic process for this substratum.

Besides temperature, hydraulic retention time, etc., reactor mixing is a main factor in the anaerobic process. Mixing improves the efficiency of the reactor, avoiding stratification of the substratum and temperature gradients. It also disperses the products of the bacteria metabolism and the eventual toxics in the residual liquor.

The objective of the present pilot plant study is to determine a reliable kinetic model for the digestion of the mentioned substrata. The study of the rate of reaction is fundamental in order to generalize the results. Mixing has notable influence in the determination of the kinetic models and the value of the constants. So, two similar digesters, stirred and non-stirred, are considered here. Moreover, the kinetic model of the anaerobic process is a previous step to design a digestion plant at commercial scale, which is the final aim of this project.

2. Materials and methods

2.1. Equipment

The experiments were accomplished in a pilot plant, placed at Artxa Farm estate, located in Villarreal (Basque Country, Spain). Average farming operations consisted of feeding about 6000 animals, mostly swine and suckling piglets. Common volume of effluents from the farm were around 30 m³ d⁻¹. The wastewaters were poured in a discontinuous way to an adjacent 12000 m³ lagoon, where a spontaneous anaerobic digestion was achieved, i.e., COD elimination was almost 20%. Final dumping was through a pipe to a nearby river stream, with an obvious harmful effect on the downstream waters.

A 3 m³ sewer provided with a sludge pump and a stirrer was built in order to collect the slurries for the reactors from the farm sewage. It provided a certain homogenization prior to the treatment. The slurry was conditioned before the introduction to the digester through a solids separator. The filter was designed for this kind of residue, built of stainless steel and operated under gravity. It consisted of a 1 mm mesh inclined sieve, removing most of the suspended solids (Fig. 1).

The liquid fraction was treated in a Contact and DSFF anaerobic digesters, related by Lomas et al. [8], whereas the “solid” or concentrated fraction of the manure was the substratum for the present experiment. Two pilot scale anaerobic reactors were designed and built for this particular substratum. They were made in an ovoid shape, with different volume and characteristics ( Fig. 2). The net volume of the mixed digester was 245 l, and was provided with a helicoid stirrer to mix the substratum. The net volume of the stratified reactor was 565 l, and the substratum load was arranged in layers. Both digesters were built of galvanized iron. Walls were thermally isolated with 80 mm thick Vitrofilm fibre glass and covered with a 0.8 mm aluminium sheet. The inner surfaces were treated with Derekane 411 for chemical protection. An internal heating system with a copper coil, by means of hot water flow, was installed.

The long duration of the fermentation in discontinuous culturing facilitated the register of many parameter data. The experiments were performed into the mesophilic range (around 35°C), and the temperature was usually kept into the selected interval. The process was regulated analyzing the characteristic physico-chemical parameters of influents and effluents. Automatic devices for the operational control were installed, both at the plant and the farm sewage. Volume of the biogas was monitored with an industrial gas meter, Sacofgas brand, mod. E2/160. Composition was recorded with a gas chromatograph mod. Shimadzu 6C-8A equipped with a graphic integrator model Chromatopac C-R6A. Electronic probes for the temperature and pH measurement with continuous recording were used. An ultrasound flow meter — Flowgef brand — was installed to monitor load rate. Density, volatile acidity, total alkalinity, total and soluble COD, total and volatile solids have also been determined, using conventional methods [9].

3. Results

3.1. Substratum characterization

Table 1 displays the average values of the characteristic parameters of the substrata. The disparity of the influent data for both reactors was related with the different periods of experimentation. Some characteristic parameters presented great alterations, i.e., total COD, total solids or volatile solids. Moreover, the acid–base features of the influents displayed scattered values, affecting the digestion. This diversity in load characteristics led to a knowledge of the digester efficiency in actual feeding conditions, quite different from the previous experiences at the laboratory, when a uniform substratum was utilized [6].

Considering the nutrients factor in the aforementioned characterization, the phosphorus:nitrogen coefficients were around 1:15 and 1:18 for the stirred and stratified reactors respectively. Results from literature are between 1:5 and 1:7, in order to supply adequate nutrients for the microorganisms, since Lema et al. [10]. The ratios indicated that the effluent left the plant containing a high content of nitrogen. In connection with the coefficients COD/N/P, related to the cellular output of the bacteria, they were about 323/15/1 and 251/18/1 for the stirred and stratified reactors respectively. According to the same authors, these relations vary widely with the kind of substratum: For a sour influent, capable only to carry out the methanegenic step, relations are around 2660/7/1. Meanwhile, proportions are about 443/7/1 for a complex substratum, which is closer to the present case.

3.2. Biogas production and composition

The hydraulic retention time was kept around 60 days. Biogas flow was measured daily during a period of working stability (Fig. 3). Great differences were found comparing biogas yield in both reactors. So, production increased notably after one week in the stirred digester. The low output during the starting period can be attributed to the adaptation of the biomass to the reactor conditions. Afterwards gas production rate increased, reaching a maximum at around 30 days, and from then on clearly decreased. On the other hand, the biogas output in the stratified digester was smaller and more uniform along the experiment. These results can be partially assigned to the mixing, because the substrata conditions were different, although the nutrients removal led to a more efficient process.

With regard to the biogas composition, only significant concentration of three components was found in the analysis: methane, carbon dioxide and nitrogen. Nevertheless little hydrogen and sulphur were detected, which are usual gasses from the fermentation of this substance. Gas composition was quite similar in both reactors. Methane concentration rose after 10 days, stabilized between 45 and 50% around 40 days, and decreased later. These values were smaller compared to the literature references, but similar to some previous experiences for this substratum, as Lapp et al. [11] and Montalvo [12]. On the other hand, the high nitrogen concentration in the beginning was attributed to the initial air into the reactors.

4. Kinetic study

4.1. Approach

Several common kinetic models were tried to adapt to the global anaerobic process in discontinuous load, i.e., Monod or Chen and Hashimoto [13]. Nevertheless, the constants obtained from those models were not reliable, because those systems didn’t fit the process of this particular substratum. The results from the first order model were acceptable in some cases, whereas difficult to be adapted in other stages of the production. This led to divide the range of the experiments in two stages, which were adjusted to simple empirical models, with a minimum amount of parameters. In consequence, a two-stage model was tried. The first model was applied from the early methane production until reaching the maximum rate. The second model was applied along the period from that point until the ending of the fermentation.

Evaluation of the constant B₀, specific methane production at infinite hydraulic retention time, was required to evaluate both phases, due to the difficulty to know the temporal variation of S. Fig. 4 displays B (m³ CH₄/kg VS_0.) vs the HRT inverse. B trends to B₀ when HRT trends to infinite, which is equivalent to a batch feeding system.

Ordinate B₀ values resulted as follows:

4.2. Two stages model

4.2.1. Development of the model for the first stage

The organic matter, coming from the solid substratum, is usually easily to digest. Then, the hypothesis for the first stage model considers that the limitation factor for the gas production is the microbial growing rate, because of the low initial concentration of the microorganisms. The endogenic respiration is negligible relative to the growing factor, due to excessive nutritional conditions. Thus, the variation of the cellular mass vs time in a discontinuous reactor can be expressed as follows:

Integrating Eq. (1), the next expression is achieved for the microorganisms concentration:

Combining (1), (2) and (3), and integrating, the result is:

Other reference used in this model is the methane to substratum equation of proportion from Chen and Hashimoto [13).

According to other studies quoted by Luengo [14], the value of μ for similar waste is about 0.4–0.8 days⁻¹ or shorter. When methane production is high, the 1 term is negligible in Eq. (6). In consequence, the simplified equation is:

4.2.2. Development of the model for the second stage

A first order kinetic model was applied for the adjustment of the second stage, which yields the greater amount of biogas. The next equation is the expression of the model applied to the substratum S:

Using the Eq. (5), and defining B′ variable as:

4.3. Calculation of the kinetic constants

Table 2 displays the values of the kinetic coefficients calculated for each stage. The specific growing rate (μ) determines the cellular output or the microbial mass generation rate. Referring to Y constant, substratum affinity by the microbial mass, the attained value is lower than the mentioned data by Luengo [14] and Martinez [15]. This result can be due to the diversity of the slurry characteristics. As a consequence, an appropriate selection of HRT for the anaerobic digestion is fundamental in this system. The reactor is more influenced by the accumulation of acids when there is a low affinity for the substratum, due to overloads. These results lead to small sludge yield compared to other related anaerobic processes.

4.4. Adjustment of the two stages model

Once attained, the kinetic model was adjusted to the reactor data obtained. Fig. 5 and Fig. 6 display the experimental values and the adjustment of B, according to (8) and (17). As a result, a coincidence of both series of values for the digesters was found.

Concord can be evaluated for the stirred reactor as follows:

5. Conclusions

The specific production of methane in the anaerobic digestion of the solid fraction of screened piggery waste can be adjusted dividing the process in two stages. The separation of those stages is considered at the time of maximum specific methane production rate. Then a first order kinetic model is applied in each period. The microbial growing is the limiting factor in the first stage, whereas the nutrient availability is the limiting factor in the second stage. This two stage model suitably fits the presented experimental results.