Co-digestion of municipal sewage sludge and solid waste: Modelling of carbohydrate,lipid and protein content influence

Abstract

Solid wastes from industrial, commercial and community activities are of growing concern as the total volume of waste produced continues to increase. The knowledge of the specific composition and characteristics of the waste is an important tool in the correct development of the anaerobic digestion process. The problems derived from the anaerobic digestion of sole substrates with high lipid, carbohydrate or protein content lead to the co-digestion of these substrates with another disposed waste, such as sewage sludge. The kinetic of the anaerobic digestion is especially difficult to explain adequately, although some mathematical models are able to represent the main aspects of a biological system, thus improving understanding of the parameters involved in the process. The aim of this work is to evaluate the experimental biochemical methane potential on the co-digestion of sewage sludge with different solid wastes (grease; spent grain and cow manure) through the implementation of four kinetic models. The co-digestion of grease waste and mixed sludge obtained the best improvements from the sole substrates, with additional positive synergistic effects. The Gompertz model fits the experimental biochemical methane potential to an accuracy of 99%, showing a correlation between the percentage of lipid in the substrates and co-digestions and the period of lag phase.

Keywords

Anaerobic co-digestion lipids carbohydrates proteins sewage sludge model

Introduction

Solid waste covers all wastes from garbage, refuse, manure, sludge from waste water treatment plants (WWTP) and any other material resulting from industrial, commercial and community activities (Puopiel, 2010). Solid waste presents an important problem for policy makers in most of the developing countries owing to the large quantity produced. Around 531 kg of solid waste was produced in Spain in 2011 (INE, 2011).

It is known that the anaerobic biodegradation of organic matter is related to its composition. It is therefore essential to find out the exact characteristics of the substrate that is to be digested (Llabrés-Luengo and Mata-Alvarez, 1988). Lipids exhibit a much higher biogas potential (1 m³ per kg of volatile solids (VS)) than carbohydrates, proteins, cellulose and starch, and are considered to be a promising substrate for anaerobic treatment and a potential energy source (Noutsopoulos et al., 2013; Wagner et al., 2013). On the other hand, carbohydrate-rich and protein-rich substrates are also feasible as they are good producers of volatile fatty acids and demonstrate a good buffering capacity owing to the production of ammonia (Elbeshbishy and Nakhla, 2012). Despite their high methane potential, lipids and lipid-rich wastes are not commonly used as a sole substrate in anaerobic digestion (AD). This is owing to their effect on anaerobic biocenosis along with the development of other operational problems like clogging, foaming and biomass flotation (Noutsopoulos et al., 2013). In particulate material that is difficult to degrade, such as animal by-products rich in carbohydrates, hydrolysis must be coupled with the growth of hydrolytic bacteria and this factor can limit the overall degradation rate (Palatsi et al., 2011). Furthermore, high amounts of easily biodegradable carbohydrates and amino acids might reveal problems owing to the fast accumulation of intermediates like volatile fatty acids (VFA) (Wagner et al., 2013). Ammonia is also generated during protein degradation, and its inhibitory effects have been reported (Ashekuzzaman and Poulsen, 2011).

Owing to the difficulties in the digestion of these wastes as unique substrates, the anaerobic co-digestion is considered a better option for enhancing biogas production and organic matter degradation. Co-digestion techniques could improve the dilution of inhibitory compounds, increase the degradation of the treated material owing to synergistic effects, optimise the moisture and nutrient content, and balance the carbon to nitrogen ratio (Molinuevo-Salces et al., 2012; Wagner et al., 2013).

The co-digestion of sewage sludge with other substrates could be beneficial owing to both the increased biogas production of one of the most important AD feed materials, and to increased income through gate fees (Luostarinen et al., 2099; Zhu et al., 2011). Sewage sludge is mainly composed of primary and biological sludge, the latter being more difficult to digest and demonstrating lower productivities (Gavala et al., 2003; Wan et al. 2011). The co-digestion of certain substrates can produce synergistic effects owing to the contribution of trace elements, nutrients, enzymes or any other substance that the substrate may lack. Even so, antagonistic effects may also be produced for pH inhibition, ammonia toxicity or high volatile acid concentration (Labatut et al., 2011).

Previous works have marked the use of diverse co-substrates as a good option for enhancing sewage sludge productivity; Luostarinen et al. (2009) studied the co-digestion of sewage sludge with grease trap sludge, thus enhancing the biogas production; and Pitk et al. (2013) considered the co-digestion of sewage sludge with sterilised solid slaughterhouse waste, obtaining optimum results. Nonetheless there are also other substrates that are particularly interesting for their co-digestion with sludge. This is demonstrated in the works of Bolzonella et al. (2006) studying the co-digestion of waste activated sludge with organic fraction of municipal solid waste (MSW), or in the investigations of primary sludge with fruit and vegetable fraction of MSW by Gómez et al. (2006), both of whom obtained increases in the sludge productivity.

Biochemical methane potential (BMP) tests are a high quality and valuable way to evaluate the anaerobic biodegradability of the co-digestion of different substrates. Its methodology allows for changes in the procedure, obtaining reliable results for the configuration of different wastes (De la Rubia et al., 2011). Owing to the complexity of BMP curves, the use of several models could be helpful in order to explain the most important variables affecting the process (Garcia-Ochoa et al., 1999). Modelling approaches have been used to study the fitting of experimental BMP to a specific equation. Lo et al. (2010) studied the process modelling of MSW and its co-digestion with ashes derived from its incineration and Cecchi et al. (1991) used different models to compare the anaerobic degradation of complex substrates. However, few works are focused on studying the relation between organic composition and specific models.

The aim of this work is to study the effects of the co-digestion of several organic solid wastes with municipal sewage sludge, and to evaluate the influence of composition in the anaerobic co-digestion process using different BMP model approaches.

Materials and methods

Substrates

Municipal sewage sludge was used as the main co-substrate in order to study different co-digestions with diverse wastes. Mixed sludge is composed of thickened primary and biological sludge (50% weight) sampled from a WWTP in Spain (Table 1). Co-digestions were carried out with biological sludge and mixed sludge, also known as waste activated sludge and waste mixed sludge, respectively, since the former is less productive than primary sludge, and the latter is originally disposed of in the WWTP.

Table 1.

Mixture percentage and characterisation of the sole substrates and of the co-digestion mixtures.

	Mixture ratio			Characterisation
	Weight (%)	VS (%)	CODt (%)	TS (g kg⁻¹)	VS (g kg⁻¹)	CODt (g kg⁻¹)	Lipid (%CODt)	Carbohydrate (%CODt)	Protein (%CODt)
Primary sludge	100	100	100	198	132	188	36	41	23
Biological sludge	100	100	100	72	57	82	4	38	58
Grease	100	100	100	364	371	381	94	1	5
Spent grain	100	100	100	244	233	303	5	79	16
Manure	100	100	100	222	209	259	5	25	70
Grease/biological sludge	26/74	76/24	64/36	201	180	160	61	13	26
Grease/mixed sludge	15/85	38/62	34/66	172	142	146	52	23	25
Spent grain/biological sludge	31/69	64/36	62/38	117	92	169	5	65	30
Manure/biological sludge	26/74	57/43	52/48	99	84	139	4	31	65
Manure/mixed sludge	15/85	28/72	23/77	114	87	133	17	35	48

CODt: total chemical oxygen demand; TS: total solids; VS: volatile solids.

Several wastes from industrial facilities were selected as co-substrates to evaluate their co-digestion with biological and mixed sludge. The co-substrates were classified by their chemical composition (carbohydrates, lipids and protein content) (Table 1).

Grease: substrate with high lipid content taken from the dissolved air flotation unit, as part of the primary treatment of an urban WWTP.

Spent grain: substrate with high carbohydrate content collected in a brewery plant from the filtration process and dried.

Manure: substrate with high protein content collected in the same slaughterhouse (cow dung).

Batch assays were performed studying the methane potential of the sole substrates and their co-digestion with biological and mixed sludge as indicated in Table 1. The percentages of mixture used correspond to optimum ratios from previous internal studies (Bouchy et al., 2010; Cano et al., 2012; Pérez et al., 2012) in which the higher productivities were obtained.

Experimental BMP tests

The BMP assays were performed following an internal method from the University of Valladolid based on standardised assays for research purposes (Angelidaki et al., 2009). The length of the experiments depend on the kind of substrate, taking 26, 39 and 45 days for the substrates with spent grain, manure and grease, respectively. Glass bottles of 2 L capacity were used to carry out the tests. The substrate and the inoculum were introduced following a substrate/inoculum ratio in terms of VS (gVS substrate per gVS inoculum), and some macronutrients and micronutrients were added to ensure the activity of the inoculum (Field et al., 1988). Once the bottles were closed they were placed in a rotational stirrer.

Tests were carried out in triplicate, including a blank to evaluate the final production of the substrate, and a control with cellulose to evaluate the correct operation of the inoculums. Periodical monitoring analyses were performed, both of biogas production, by pressure meter, and of biogas composition, by gas chromatography. The end of the biodegradability assays was determined subject to their productivity, where a production of less than 1% would indicate the end of the experiment. Once the assays were finished, the main parameters were analysed in order to evaluate the effectiveness of the process, taking into account the removal results.

Methane potential or BMP is expressed as the net volume of methane per gram of initial substrate VS content (mlCH₄gVS⁻¹ added). These results were obtained from the triplicate average for each assay. A standard deviation was also calculated in order to identify the error among triplicates.

Synergistic effects

To evaluate the influence of each substrate in the different mixtures and calculate the possible synergistic effects that could be produced during the biodegradation process, the subsequent equation (1) was followed (Arribas et al., 2012):

\propto = \frac{E x p e r i m e n t a l p r o d u c t i o n}{T h e o r e t i c a l p r o d u c t i o n}

(1)

The ‘experimental production’ is the result of the BMP tests on the different mixtures, while the ‘theoretical production’ refers to the final values of the mixtures obtained from the BMP of each raw substrate. The results of α indicate:

α > 1; the mixture has a synergistic effect in the final production;

α = 1; the substrates work independently from the mixture;

α < 1; the mixture has a competitive effect in the final production.

Prediction models for BMP

Models based on kinetic parameters enable the effect of the most important process variables on system performance to be predicted (Fdez.-Güelfo et al., 2011). The fitting of model equations to experimental results obtained in specific assays is a very important tool in determining the optimum substrate and operational conditions. Models based on first-order kinetics, transference function and Gompertz equation are used to study the co-digestion of sewage sludge with different solid wastes. Table 2 shows all models and their characteristics where P(mlCH₄gVS⁻¹) is the biogas production rate at time t(d) over the digestion period; γ(mlCH₄gVS⁻¹) is the maximum volume accumulated at an infinite digestion time; µ₁(d⁻¹) is the specific micro-organism growing speed for FO I, which, together with µ₂(d⁻¹) represents the two rates of micro-organism growing speeds for FO II; K(mlCH₄gVS⁻¹d⁻¹) is the specific rate constant for Gompertz model (GM) and transference function (TF) and λ(d) is the lag phase.

Table 2.

Models applied.

Reference	Equation		Description
De Gioannis et al., 2009	$P = γ (1 - \exp (- μ t))$	(2)	First-order model (FO I), in which biogas accumulation was simulated using exponential rise to a maximum.
Strömberg et al., 2013	$P = γ (x (1 - \exp (- μ_{1} t)) + (1 - x) (1 - \exp (- μ_{2} t)))$	(3)	First-order model (FO II), considers that the biomass has slow and fast degradable parts represented by two rates.
Lay et al., 1997	$P = γ \exp (- \exp (\frac{K (- t) e}{γ} + 1)$	(4)	Modified Gompertz model (GM), assumes that the rate of gas production is proportional to the microbial activity.
Donoso-Bravo et al., 2010	$P = γ (1 - \exp (- \frac{K (t -)}{γ}))$	(5)	Transference function (TF), considers that any process might be analysed as a system receiving inputs and generating outputs.

The implementation of the four models was carried out using the solver method from excel as a linear programing tool. For each model, a theoretical production curve was generated, calculating the regression coefficient (r²) for each equation through the comparison of the experimental and theoretical results. In a first step, the theoretical production was calculated for each specific time when the pressure and methane composition were measured. Following this, the deviation from the experimental production was obtained for each point and the totality of all the deviation was calculated. A final r² was obtained from the individual deviation of the experimental and theoretical results. In a final step the solver tool was applied to the results, indicating the minimum total deviation as the objective, therefore the rest of the parameters were calculated (λ, µ, K and γ) following the previous objective.

Analytical methods

For the analytical and characterisation methodologies, an internal protocol based on standard methods (Apha, 2005) was used to determine the most significant characteristics of the inoculum and the substrate (total solids (TS) and volatile solids (VS); total chemical oxygen demand (CODt), soluble chemical oxygen demand (CODs); total Kjeldahl nitrogen (TKN)).

In order to study the influence of the composition in terms of proteins, carbohydrates and lipids, an extended characterisation was carried out using gravimetric techniques (EPA method 1664, 1999; CE regulation 152/2009, 2009) for lipids, volumetric procedures (CE regulation 152/2009, 2009) for carbohydrates and elemental analyses (IT-MA-014 AOAC, OAC, 2005) for protein determination.

Experimental procedure

The productivity of the sole substrates and the co-digestion mixtures previously explained in Table 1 were studied using BMP tests. Once the assays were finished, the experimental results were submitted to a modelling process in order to confirm the accuracy of the data obtained, thus enabling the prediction of methane potentials and the study of the methane curves through kinetic parameters. Finally it was possible to determine the needs of a model equation in order to fit with this kind of substrates.

Statistical analysis

The deviation of each assay from the average was considered and represented with the methane curves to indicate the consistency of the BMP experiments. Besides, to determine and compare the accuracy of the experimental results with different models, a regression coefficient and the relative error were calculated by comparing the experimental and prediction productivities for each model, in order to select the optimum model equation. Statistical methods were performed using excel software.

Results and discussion

BMP co-digestion results

The results obtained from the BMP indicated in Figure 1 were successfully submitted to the test modelling. The results of the four models previously explained are presented in Table 3, where the maximum production (γ), the regression coefficient and the kinetic parameters are indicated for each model. Generally, the BMP experimental results could be explained by a complex model based on the Gompertz equation, which explained 60% of the experimental data with values of r² over 0.99 and 40% over 0.98. Nevertheless, some sole substrates such as manure, and several mixtures such as spent grain or manure co-digested with biological sludge, could also be explained with simplified models based on first-order kinetics FO I and FO II with values for r² also over 0.99. The transference function only fits with biological sludge or manure (r² of 0.998 and 0.996, respectively); these substrates could also be represented by a first-order equation.

Table 3.

Parameter estimation for the models in application. Values in bold and italic indicate optimum and poorest data, respectively.

Co-substrate	Model	γ mlCH₄gVS⁻¹	Kinetic parameters	λ	r²	Co-digestion	Model	γ mlCH₄gVS⁻¹	Kinetic parameters	λ	r²
Primary sludge	FO I	408	µ₁ = 0.07d⁻¹	–	0.978	Grease+ biological sludge	FO I	551	µ₁ = 0.02d⁻¹	–	0.940
Primary sludge	FO II	393	µ₁/µ₂ = 0.00/0.07d⁻¹	–	0.978		FO II	484	µ₁/µ₂ = 0.00/0.2d-¹	–	0.940
	GM	336	K = 28mlCH₄gVS⁻¹d⁻¹	2.1	0.998		GM	388	K = 17mlCH₄gVS⁻¹d⁻¹	7.0	0.992
	TF	384	K = 32mlCH₄gVS⁻¹d⁻¹	1.0	0.981		TF	648	K = 15mlCH₄g⁻¹VS⁻¹	2.8	0.937
Biological sludge	FO I	200	µ₁ = 0.11d⁻¹	–	0.994	Grease+ mixed sludge	FO I	872	µ₁ = 0.02d⁻¹	–	0.950
Biological sludge	FO II	194	µ₁/µ₂ = 0.12/0.09d⁻¹	–	0.998		FO II	520	µ₁/µ₂ = 0.11/0.12d⁻¹	–	0.994
	GM	181	K = 15mlCH₄gVS⁻¹d⁻¹	0.0	0.988		GM	526	K = 23mlCH₄gVS⁻¹d⁻¹	4.5	0.996
	TF	193	K = 22mlCH₄gVS⁻¹d⁻¹	0.1	0.998		TF	717	K = 23mlCH₄gVS⁻¹d⁻¹	2.0	0.964
Grease	FO I	43896	µ₁ = 0.0d⁻¹	–	0.869	Spent grain+ biological sludge	FO I	359	µ₁ = 0.26d⁻¹	–	0.991
	FO II	642	µ₁/µ₂ = 0.06/0.07d⁻¹	–	0.954		FO II	364	µ₁/µ₂ = 0.10/0.29d⁻¹	–	0.997
	GM	511	K = 39mlCH₄gVS⁻¹d⁻¹	17.9	0.992		GM	350	K = 62mlCH₄gVS⁻¹d⁻¹	0.0	0.985
	TF	3440	K = 12mlCH₄gVS⁻¹d⁻¹	6.4	0.899		TF	358	K = 97mlCH₄gVS⁻¹d⁻¹	0.1	0.991
Spent grain	FO I	278	µ₁ = 0.09d⁻¹	–	0.991	Manure+ biological sludge	FO I	282	µ₁ = 0.15d⁻¹	–	0.982
Spent grain	FO II	278	µ₁/µ₂ = 0.09/0.09d⁻¹	–	0.991		FO II	282	µ₁/µ₂ = 0.00/0.00d⁻¹	–	0.976
	GM	251	K = 19mlCH₄gVS⁻¹d⁻¹	0.8	0.994		GM	272	K = 18mlCH₄gVS⁻¹d⁻¹	0.0	0.958
	TF	272	K(mlCH₄/gVS/d): 26	0.5	0.992		TF	282	K = 41mlCH₄gVS⁻¹d⁻¹	0.0	0.976
Manure	FO I	345	µ₁ = 0.08d⁻¹	–	0.996	Manure+ mixed sludge	FO I	311	µ₁ = 0.07d⁻¹	–	0.979
	FO II	373	µ₁/µ₂ = 0.03/0.17 d⁻¹	–	0.997		FO II	294	µ₁/µ₂ = 1.40/0.00d⁻¹	–	0.986
	GM	317	K = 20mlCH₄gVS⁻¹d⁻¹	0.0	0.981		GM	273	K = 20mlCH₄gVS⁻¹d⁻¹	1.4	0.991
	TF	345	K = 28mlCH₄gVS⁻¹d⁻¹	0.0	0.996		TF	301	K = 25mlCH₄gVS⁻¹d⁻¹	0.9	0.982

FO: first-order; GM: Gompertz model; TF: transference function.

Those substrates with higher lag phase periods are better explained with the Gompertz model. The kinetic curves of these experiments are slightly slower in the beginning of the test, as happens with the substrate of grease or primary sludge and their co-digestions, or spent grain as sole substrate. Therefore, the curves produced from substrates as biological sludge, manure and its co-digestions, lacking of lag phase periods, fit better with first-order models. It is also observed that the Gompertz model adjusts better to those substrates with the highest kinetics, while first-order model (FOII) fit with slower kinetic substrates. The co-digestion of spent grain and biological sludge fits with first-order model (FOII) and obtained the highest values for K (62 mlCH₄gVS⁻¹d⁻¹), however it presents a low r² for the Gompertz model of 0.985.

Similar values of µ indicating the growing rate of micro-organisms expressed in d⁻¹ were obtained for first-order models, although FO II was better able to explain the fast and slow biodegradable parts of each assay. For complex models, the value that indicates the specific rate is the constant K in mlCH₄gVS⁻¹d⁻¹, which gave comparable results for the Gompertz model and transference function, showing the fastest kinetics for spent grain co-digested with biological sludge in both cases. Values of λ are also equivalent, showing that substrates and co-digestions with a period of lag phase fit better with the Gompertz equation than simplified or transference models.

Given the results achieved from the test modelling, it was concluded that the BMP assays could be explained using the Gompertz equation which had been successfully adapted to the experimental results. This model offers some additional kinetic parameters to further explain the data and methane curves obtained, which are represented in Figure 1.

Comparing the BMP of the three raw substrates, grease obtained the best productivities even though the kinetics were slower for this kind of waste than for manure and spent grain, which achieved 60% of their final production in the first 10 days, while grease maintained a period of adaptability until the 20th day.

In all the co-digestion experiments there is an increment or diminution in productivity owing to the synergistic or competitive effects produced by the mixture of different wastes. This data is presented in Table 4 with the values of α indicating the synergy of each co-digestion.

Table 4.

Increase (+) and decrease (–) productivities based on the raw substrates.

	Raw substrate (%)	Biological sludge (%)	Primary sludge (%)	Synergy (α)
Grease + biological sludge	−27	+35	–	0.92
Grease + mixed sludge	+1	+88	+50	1.31
Spent grain + biological sludge	+49	+97	–	1.65
Manure + biological sludge	−11	+47	–	1.10
Manure + mixed sludge	−15	+41	−6	0.94

Methane production of grease increases in the co-digestion with mixed sludge (526 mlCH₄gVS⁻¹). Although the rise is only by 1%, and the content of grease is smaller than in the raw substrate (94% and 52%, respectively), the result of the synergistic effect is positive, indicating that the substrates play a different role within the mixture than they do independently. However, biological sludge did not support the mixture with grease, and the value of α (0.92) points to the competitive effect between co-substrates in the mixture. Looking at the kinetics of the sole substrate and the two mixtures, faster productivities are observed for the mixtures during the first 10 days than for the raw substrate, and the slope increases with the decrease in the lipid content. The modelling results (Table 3) are also a sign of the improvement of the co-digestion of grease and mixed sludge. Although the addition of mixed sludge decreased the kinetic rate (K) from values of 39 mlCH₄gVS⁻¹ for grease to 23 mlCH₄gVS⁻¹ for the co-digestion of grease and mixed sludge, the lag phase period represented by λ was reduced in time from 17.9 to 4.5 days as its biodegradation was quicker.

The BMP of spent grain increases when co-digested with biological sludge up to 49% more than in the sole substrate. Given the results obtained with biological sludge, there is no need to blend this substrate with mixed sludge. The experimental and theoretical results indicate a high synergistic effect within the mixture (value of 1.65 for α). However, the methane production of the raw substrate is less than in grease or manure owing to the high percentage of carbohydrate, which decreases in the mixture with the addition of biological sludge. Values of K confirm the improvement of the co-digestion, these being three times faster than in spent grain alone, presenting the highest rates of all the assays (62 mlCH₄gVS⁻¹d⁻¹)

Manure co-digestion productivity did not increase with any of the mixtures, even though there was a positive synergistic effect (1.10) with biological sludge, but not with mixed sludge (0.94) where a competitive effect between substrates was observed. The content of lipids (17%) in the configuration of manure and mixed sludge could explain its slower kinetics, while the high-protein content in the sole substrate (70%) and the diminution in VS and CODt in the co-digestion make this mixture less biodegradable. The kinetics of the methane curves indicate that even though the mixture with biological sludge did not improve the productivity, the slope was more pronounced during the first 10 days, indicating quick biodegradability as shown by the value of K 18 mlCH₄gVS⁻¹ for the Gompertz model.

Primary and biological sludge obtained experimental BMP results of 336 and 181 ml CH₄gVS⁻¹, respectively, increasing in productivity when mixed with the three wastes, even though the major increases occurred in the co-digestion of mixed sludge and grease, with positive increments of 88% for biological sludge and 50% for primary sludge. Kinetic rates were improved for biological sludge in all cases, especially with spent grain. On the contrary, primary sludge kinetics decreased in the co-digestion with biological sludge and solid wastes. Comparing the results with other works with sewage sludge, similar results were obtained with increases of 31% with the addition of grease trap sludge (46% VS) (Luostarinen et al., 2009), although higher increments were found of at least 176% with the addition of sterilised solid slaughter house sludge (7.5% weight) (Pitk et al., 2013). Other co-substrates have also shown increases in the production of biological sludge. An increase of 37.5% was observed in its co-digestion with the organic fraction MSW (Bolzonella et al., 2006) and a 20% increase with the addition of the fruit and vegetable fraction of MSW (Gómez et al., 2006).

Influence of organic composition on the methane potential

The results follow the standards based on the organic composition. Grease waste obtained the highest productivities owing to the high content in lipids (94%), then manure with high protein content (70%) and finally spent grain rich in carbohydrates (78%). Dissimilar behaviour occurred with co-digestions, as in the case of grease and mixed sludge, which obtained the best productivities even though the grease content was smaller than in the mixture with biological sludge. The synergistic effects play an important role in the co-digestions, also obtaining better productivities for spent grain and biological sludge, probably owing to the increase in protein content from the sole substrate, but also thanks to the synergism. On the other hand, the competitive effect between substrates and the decrease in the protein content from the sole manure produces lower productivities. In this case, the grease percentages are higher, affecting only the initial kinetics of the biodegradation process without increasing the methane productivity.

The lag phase parameter (in days) appears only in substrates with a lipid content of at least 17%, otherwise substrates with a lipid percentage content below 6% did not show any signs of slow biodegradation periods. This behaviour occurs in the case of substrates, such as biological sludge, spent grain or manure, where all those with less than 6% lipid content showed no lag phase periods (λ) for the Gompertz equation, or in the co-digestions of spent grain and manure with biological sludge. The connection between the lipid content (%) and the lag phase of the Gompertz model follows the next polynomial equation (Figure 2), increasing the lag phase with the rise in the lipids content. Despite the adaptation period, these substrates and co-digestion combinations also obtained higher productivities.

Similar references, in which the excess of organic components were studied, showed slower methane kinetics for lipid excess, but higher methane productions and faster kinetics for excess in proteins and carbohydrates (Neves et al., 2008). Besides, Zhu et al. (2011) studied the co-digestion of grease trap waste and municipal waste sludge, confirming that the use of grease trap waste at proper loadings could improve the process efficiency of the AD of municipal waste sludge, enhancing methane at up to 65%.

Statistical analysis

The four models were compared in order to analyse the most significant parameters in each case, and to determine which models are the best at predicting the BMP of wastes with similar characterisation. The prediction capacity and the ability to reproduce the methane curves were studied from the experimental BMP results (Table 5). For the prediction capacity, the relative error was calculated using the experimental and model values for the final productions, while the regression coefficient of each method was considered in order to evaluate their reliability.

Table 5.

Statistical results for the models in application.

	γ exp (mlCH₄gVS⁻¹)	γ model (mlCH₄gVS⁻¹)				Error %				r²
	γ exp (mlCH₄gVS⁻¹)	FO I	FO II	GM	TF	FO I	FO II	GM	TF	FO I	FO II	GM	TF
Primary sludge	338	408	393	336	384	−17	−14	0	−12	0.978	0.978	0.998	0.981
Biological sludge	188	200	194	181	193	−6	−3	4	–2	0.994	0.998	0.988	0.998
Grease	503	43896	642	511	3440	−99	−22	–2	−85	0.869	0.954	0.992	0.899
Spent grain	249	278	278	251	272	−10	−10	–1	−8	0.991	0.991	0.994	0.992
Manure	337	345	373	317	345	–2	−10	6	–2	0.996	0.997	0.981	0.996
Grease + biological sludge	365	551	484	388	648	−34	−25	–6	−44	0.940	0.940	0.992	0.937
Grease + mixed sludge	508	872	520	526	717	−42	–2	−3	−29	0.950	0.994	0.996	0.964
Spent grain + biological sludge	370	359	364	350	358	3	2	6	3	0.991	0.997	0.985	0.991
Manure + biological sludge	300	282	282	272	282	6	6	10	6	0.982	0.976	0.958	0.976
Manure + mixed sludge	288	311	294	273	301	−7	–2	5	−4	0.979	0.986	0.991	0.982

FO: first-order; GM: Gompertz model; TF: transference function. Best results for each susbtrate is indicated in bold.

In most of the assays, the regression coefficient is associated with the error, which means the model is reproducible and is also able to predict the final productivity without significant error. However, in some cases it is possible for the model to be reproducible but not predictable, as is the example of the model FO II for manure, whose final production indicates an error of 10% although it is the most reproducible. On the contrary, there are some models that predict the final production with errors of less than 7%, but are not as easily reproducible as other models, as happened in the co-digestion of manure and mixed sludge for FO II and TF or biological sludge for GM.

Finally, when looking at the general composition of the substrates analysed (Table 1), it can be observed that the fitting of a substrate to a specific model could be defined by its specific content. Lipid-rich material curves are suitable for the Gompertz model, while wastes with a high content of protein or carbohydrates adjust better to the simplified models FO I or FO II.

Removal efficiency

At the end of the experiment each substrate and co-digestion was analysed in order to obtain the final composition for TS, VS and CODt. These analyses were compared with the initial values obtaining the removal efficiency for each experiment.

High removal values were obtained for both types of sludge with efficiencies of 64% TS, 47% VS and 68% CODt for primary sludge. Spent grain and its corresponding mixture with biological sludge achieved removal percentages of over 40% for TS, 20% for VS and 50% CODt, while the rest of the substrates and mixtures did not surpass these values.

Conclusions

The use of primary and biological sludge for co-digestion is a good way to enhance the productivity and take advantage of these two available substrates. The co-digestion of grease and mixed sludge obtained the best BMP, while improving grease, biological and primary sludge productivities from the sole substrates. Synergistic effects were found, not only in this mixture, but also in the mixtures of biological sludge with manure and spent grain. Models based on Gompertz kinetics could explain the experimental BMP with high consistency (99%), especially for lipid-rich material, presenting a correlation between the lag phase parameter and the lipid content in the substrates.

Footnotes

Declaration of conflicting interest

The author declares that there is no conflict of interest.

Funding

The work presented is carried out under the financial support of R+i Alliance with the Consolider group and FEDER funds.

References

Angelidaki

Alves

Bolzonella

. (2009) Defining the biomethane potential (BMP) of solid organic wastes and energy crops: A proposed protocol for batch assays. Water Science and Technology 59: 927–934.

Apha (2005) Standard Methods for Examination of Water and Wastewater. 21st ed. New York: APHA, AWWA, WPCR. 1.194.

Arribas

Montañez

Milhau

. (2012) Digestión de RSU: influencia de un pretratamiento térmico y de su mezcla con lodos de depuradora. Salamanca: International Congress Water, Waste and Energy Management.

Ashekuzzaman

Poulsen

(2011) Optimizing feed composition for improved methane yield during anaerobic digestion of cow manure based waste mixtures. Bioresource Technology 102: 2213–2218.

Bolzonella

Battistoni

Susini

. (2006) Anaerobic codigestión of waste activated sludge and OFMSW: The experiences of Viareggio and Treviso plants (Italy). Water Science & Technology 53: 203–211.

Bouchy

Rouez

Pérez

. (2010) Sludge co-digestion: a viable path towards energy self-sufficient WWTP. In: Proceedings of ORBIT, Heraklion Crete, 29 June–3 July 2010.

Cano

Nielfa

Fdz-Polanco

. (2012) Optimization of cow manure and sewage sludge co-digestion using disintegration technologies. In: Proceedings of ORBIT, Rennes, 12–15 June 2012.

Cecchi

Mata-Alvarez

Marcomini

. (1991) First order and step-diffusional kinetic models in simulating the mesophilic anaerobic digestion of complex substrates. Bioresource Technology 36: 261–269.

CE 152/2009 (2009) Commission Regulation 27 January 2009. Laying down the methods of sampling and analysis for the official control of feed. Anex III, C. Official Journal of the European Union. L54/1.

10.

De Gioannis

Muntoni

Cappai

. (2009) Landﬁll gas generation after mechanical biological treatment of municipal solid waste. Estimation of gas generation rate constants. Waste Management 29: 1026–1034.

11.

De la Rubia

Fernández-Cegrí

. (2011) Influence of particle size and chemical composition on the performance and kinetics of anaerobic digestion process of sunflower oil cake in batch mode. Biochemical Engineering Journal 58–59: 162–167.

12.

Donoso-Bravo

Pérez-Elvira

Fdz-Polanco

(2010) Application of simplified models for anaerobic biodegradability tests. Evaluation of pre-treatment processes. Chemical Engineering Journal 160: 607–614.

13.

Elbeshbishy

Nakhla

(2012) Batch anaerobic co-digestion of proteins and carbohydrates. Bioresource Technology 116: 170–178.

14.

EPA Method 1664 (1999) Revision A: N-Hexane Extractable Material (HEM; Oil and Grease) and Silica Gel Treated N-Hexane Extractable Material (SGT-HEM; Non-polar Material) by Extraction and Gravimetry.

15.

Field

Sierra

Lettinga

(1988) Anaerobic assays. In: Proceedings of 4th symposium on wastewater anaerobic treatment, Valladolid, 23–25 November 1988, pp.52–81.

16.

Fdez.-Güelfo

Álvarez-Gallego

Sales Márquez

. (2011) The effect of different pretreatments on biomethanation kinetics of industrial Organic Fraction of Municipal Solid Wastes (OFMSW). Chemical Engineering Journal 171: 411–417.

17.

Garcia-Ochoa

Santos

Naval

. (1999) Kinetic model for anaerobic digestion of livestock manure. Enzyme and Microbial Technology 25: 55–60.

18.

Gavala

Yenal

Skiadas

. (2003) Mesophilic and thermophilic anaerobic digestion of primary and secondary sludge. Effect of pre-treatment at elevated temperature. Water Research 37: 4561–4572.

19.

Gómez

Cuetos

Cara

. (2006) Conditions for mixing and evaluation of the organic loading rate. Renewable Energy 31: 2017–2024.

20.

INE (2011) Instituto Nacional de Estadistica. Available at: http://www.ine.es/inebmenu/mnu_medioambiente.htm#4

21.

Labatut

Angenent

Scott

(2011) Biochemical methane potential and biodegradability of complex organic substrates. Bioresource Technology 102: 2255–2264.

22.

Lay

Noike

(1997) Influences of pH and moisture content on the methane production in high-solids sludge digestion. Water Research 31: 1518–1524.

23.

Llabrés-Luengo

Mata-Alvarez

(1988) Influence of temperature, buffer, composition and straw particle lenght on the anaerobic digestion of wheat straw-pig manure mixtures. Conservation and Recycling 1: 27–37.

24.

Kurniawan

Sillanpää

MET

. (2010) Modeling biogas production from organic fraction of MSW co-digested with MSWI ashes in anaerobic bioreactors. Bioresource Technology 101: 6329–6335.

25.

Luostarinen

Luste

Sillanpää

(2009) Increased biogas production at wastewater treatment plants through co-digestion of sewage sludge with grease trap sludge from a meat processing plant. Bioresource Technology 100: 79–85.

26.

Molinuevo-Salces

González-Fernández

Gómez

. (2012) Vegetable processing wastes addition to improve swine manure anaerobic digestion: Evaluation in terms of methane yield and SEM characterization. Applied Energy 91: 36–42.

27.

Neves

Gonçalo

Oliveira

. (2008) Influence of composition on the biomethanation potential of restaurant waste at mesophilic temperatures. Waste Management 28: 965–972.

28.

Noutsopoulos

Mamais

Antoniou

. (2013) Anaerobic co-digestion of grease sludge and sewage sludge: The effect of organic loading and grease sludge content. Bioresource Technology 131: 452–459.

29.

OAC Official Method 990.03. (2005) Protein (Crude) in Animal Feed, Combustion Method, in Official Methods of Analysis of AOAC International. 18th ed. Revision 1, Ch 4, pp.30–31. Arlington, VA: AOAC International.

30.

Palatsi

Viñas

Guivernau

. (2011) Anaerobic digestion of slaughterhouse waste: Main process limitations and microbial community interactions. Bioresource Technology 102: 2219–2227.

31.

Pérez

Cano

Nielfa

. (2012) Operation strategy of grease and municipal sewage sludge co-digestion: results of continuous trials. In: Proceedings of ORBIT, Rennes, 12–15 June 2012.

32.

Pitk

Kaparaju

Palatsi

. (2013) Co-digestion of sewage sludge and sterilized solid slaughterhouse waste: methane production efficiency and process limitations. Bioresource Technology 134: 227–232.

33.

Puopiel

(2010) Solid waste management in Ghana: the case of tamale metropolitan area. Thesis, Department of Planning Kwame Nkrumah University of Science and Technology.

34.

Strömberg

Nistor

Liu

(2013) Early prediction of biochemical methane potential (BMP) based on real-time modelling of Automatic Methane Potential Test System II (AMPTS II) data. IWA. Santiago de Compostela Congress, 25–28 June 2013.

35.

Wagner

Lins

Malin

. (2013) Impact of protein-, lipid- and cellulose-containing complex substrates on biogas production and microbial communities in batch experiments. Science of the Total Environment 458–460: 256–266.

36.

Wan

Zhou

. (2011) Semi-continuous anaerobic co-digestion of thickened waste activated sludge and fat oil and grease. Waste Management 31: 1752–1758.

37.

Zhu

Hsueh

(2011) Enhancing biomethanation of municipal waste sludge with grease trap waste as a co-substrate. Renewable Energy 36: 1802–1807.