Get Complete Project Material File(s) Now! »
Information characterizing data origins
In order to perform a statistical analysis, FW characteristics data were sorted considering three categories of possible variation factors: continents of the analysed FW, collection sources and season of collection.
In the category “continent”, three main subcategories were chosen: Asia, North America (NA) and European Union (EU), with a proportion of data of 42%, 15% and 43% respectively. The availability of data from Africa, Oceania and Central and South America was very limited. They were thus not included in this study.
For the category “collection source”, five subcategories were chosen: restaurant food waste (RFW), household food waste (HFW), FW mixed with green waste (FWGW), FW of large producers (FWLP) and organic fraction of municipal solid waste (OFMSW). RFW represents the most important part of the available data with 41% of the total, as shown on Figure 10.
Representativeness of the mean value
A statistical study was performed using Statgraphics Centurion XVI ® software. All data were characterized with the mean ( ̅), standard deviation (σ), coefficient of variation (CV) and a confidence interval (CI), as shown in Table 6. The normality of the distribution was tested by calculating the standardized skewness and the standardized kurtosis (Baillargeon, 2008).
For Martin and Gendron (Martin & Gendron, 2003), a CV between 0% and 16% shows a low variation on the sample. With a CV between 16% and 33%, the sample has an important variation and the mean may contain many errors. Finally with a CV over 33%, the variation in the sample is too high to consider the mean as representative for the whole population. In our case, to determine whether a FW characteristic mean is representative from the data set, this later must have a low CV (close or below 16 %) and follow a normal distribution.
For the case where the mean characteristic value could not be considered as representative, the Kruskal–Wallis test was used to compare the medians of FW characteristics in order to test the variation caused by the three categories (“continent”, “collection source” and “season”). However, interactions between the categories were not studied. With a Kruskal–Wallis probabilities (KWP) lower than 0.05, a link between the variation of the characteristics and the selected categories is probable with certitude of 95%. In such cases, the Bonferroni correction was used to specify the subcategories that may generate these variations. For each characteristic, the subcategories are defined as belonging to a statistical group (1, 2 or 3). When several subcategories belong to the same group, significant statistical difference cannot be observed amongst themselves. To implement the Bonferroni correction, the data set was segregated into each individual subcategory. Only, the subcategories with at least 3 data were taken into account.
Mean characteristics and representativeness
Average data and CV for FW characteristics are presented in Table 6. In this section, a comparison between the FW and others substrates characteristics is proposed and the representativeness of FW characteristics values is discussed.
With a mean pH of 5.1, FW is more acid than other organic waste as green waste with a pH around 7.3 (Zhang & Sun, 2014), cattle manure with 8.7 (Bah et al., 2014) and sewage sludge with 7.8 (Martín et al., 2015). pH presents a CV lower than 16% and follows a normal distribution, inducing that its mean real value ranges (considering the CI) between 4.9 and 5.3 with a real σ between 0.6 and 0.8. These mean value and range of variation of pH can be universally used for process design or environmental assessment.
The BMP of 460.0 NLCH4/kg VS (±87.6) is higher than most of organic waste BMPs: BMP from cattle manure and sludge sewage are 250 and 270 NLCH4/kg VS respectively (Lesteur et al., 2010). BMP values for FW follow a normal distribution with a CV of 19% that is close to being considered as low. Using the CI, the real mean ranges between 397.3 and 522.6 mLCH4/g VS and the real σ between 60.2 and 159.3. These values can be reasonably used, whatever the type of FW, for first simulation of biodegradable potential for example.
Neither of the other mean values for physicochemical, biochemical or elementary characteristics can be considered as representative of the whole FW types. Indeed they present either CV up to 33% or do not follow a normal distribution. As an example, VS presents a low CV (9.3%), but it is not considered as representative because it does not follow a normal distribution. Because of these variations, the comparison of these FW characteristics mean values with other waste makes no sense.
In general, FW prove to have varied proportions of nutriments and micronutrients and low presence of heavy metals. Nevertheless, once again these characteristics present very high CV (some of them exceed 100 and even 200%) and did not follow normal distribution.
The causes of variations and non-normality have been searched and are presented in the next section in order to provide a better understanding of FW characteristics variabillity and to identify more representative means depending on chosen categories.
Impact of FW categories on FW quality variation
The Kruskal–Wallis test was realized for all FW characteristics, except pH and BMP. Variations observed for nine characteristics demonstrated to be significantly influenced by the three studied categories (Table 7): geographical origin, collection source, and season.
Three characteristics showed to be impacted by the geographical origin of the FW: DM, TAN and Na content. In comparison to DM general mean (22.8%), FW from Asia and North America (NA) present lower values (18.5% and 18.1 respectively) than DM of FW from EU (23.7%). TAN value of FW from Asia shows to be lower than the TAN means of FW from EU and NA (249.5 against 989.0 and 1434.3 mg/L respectively). Meanwhile, Na values for FW from Asia shows to be the highest (5.3 against 0.5 and 0.9 %DM, respectively for EU and NA).
Recall of AD fundamentals and influencing factors
AD is a biological treatment which can lead to energy production, pollution control and fertilising substrate recovery from organic waste. Microorganisms that produce methane (CH4) utilise a limited number of substrates, mainly acetate, hydrogen (H2) and carbon dioxide (CO2). Consequently, a decomposition of raw substrate is necessary to provide these elements to the microorganisms. As well known, AD follows four key steps: hydrolysis, acidogenesis, acetogenesis and methanogenesis. These steps tend to work more or less simultaneously depending on availability of degraded molecules, environmental conditions and biomass development. Several studies already reviewed the fundamentals of AD, the environmental conditions and the inhibitors of the process (Deublein & Steinhauser, 2011). Thus this section only aims at summarizing the major phases of AD of OM and their main influencing factors in order to make then a link with the anaerobic treatability of FW.
In the hydrolysis step, the complex composite particles and particulate carbohydrates, proteins and lipids are broken down in water soluble monomers as glucose, amino acids, and long chain fatty acids (LCFA) (Batstone et al., 2002). This step follows variable kinetics, depending on the complexity of OM. This degradation is achieved mainly by bacteria secreting extracellular enzymes. The facultative anaerobic bacteria consume the remaining available oxygen after establishing anaerobic conditions, decreasing the reduction potential and adapting the conditions to the strictly anaerobic bacteria. In the acidogenesis step the monomers produced in the hydrolysis step are degraded in VFA, lactate, ethanol, H2 and CO2. This transformation is performed by hydrolytic fermentative bacteria (facultative anaerobic organism), which are also present in the hydrolysis step. However, this acidogenesis step has a faster kinetics than hydrolysis. In the acetogenesis step, the metabolites from hydrolysis and acidogenesis steps are transformed mainly in acetate (CH3COOH or C2H4O2), H2 and CO2. Two groups of acetogens may be identified: the obligate hydrogen producing acetogens, which degrade VFA, LCFA , alcohols and aromatics compounds in acetate and the homoacetogens, which produce acetate from H2 and CO2 (Anderson et al., 2003). Finally, the methanogenesis is the step responsible of CH4 production which remains distinct from the other steps because the methanogens micro-organisms are not bacteria but Archaea. This step can only work in strict anaerobic conditions.
Comparison between FW quality and optimal conditions for AD treatment
Considering the whole range of values for physicochemical characteristics (especially VS, DCO and BMP) shown in Table 8, FW show a good potential for AD treatment. However, within the scientific literature dealing with AD of food waste, several authors highlighted the occurrence of instabilities in the treatment that cause the falls of pH, a lower gas yield and the increase of CO2 content in biogas (Mata-Alvarez et al., 1992). Traditionally, remedial measures are applied as alkali addition, feed interruption and mixing with a nitrogen-rich supplement (Jiang et al., 2012). The results of the present study allow proposing appropriate AD configuration to avoid instabilities linked to the FW characteristics values as a function of different categories, and to improve FW adequacy with AD treatment.
The statistical analysis (section 3.1) stated that the initial pH of FW is 5.1 (±0.7) whatever the FW categories. Even if this range of pH is favourable for fermentative bacteria that could easily develop in the first days of the AD process, a higher pH is necessary in the digester in order to favour the development of methanogen microorganisms. Moreover, this low pH combined with the potentially high carbohydrates and proteins contents of FW (respectively 36.4 %VS (±20.8) and 21.0 %VS (±13.0)), may explain the rapid acidification of reactors, caused mainly by the high production of VFA and the high concentration of ammonium ion (NH4+) in the digesters (Deublein & Steinhauser, 2011; Mata-Álvarez, 2003; Scano et al., 2014). Consequently, inhibitions of acetogenesis and methanogenesis steps could be evidenced. On the other hand, the structural carbohydrates contents (CEL with and HEM) may be higher than 8.7 and 9.4 % of VS respectively and may induce an incomplete degradation of the OM (only 50% of CEL is decomposed during AD at 37°C for 30 days) (Haug, 1993). Concerning micro-nutriments, Ca (1.6 %DM (±1.3)), K (1.2 %DM (±0.7)), Mg (0.2 %DM (±0.2)), Na (2.2 %DM (±2.9)), P (0.5 %DM (±0.3)) and Fe (482.5 ppm (±815) are present in FW and, according to Walker (Walker et al., 2010), their concentrations are close to those recommended to AD. Only the nitrogen/phosphorus ratio might sometimes be lower than the value recommended by Walker to improve the AD process (N/P=7).
Looking more accurately at the characteristic values found per FW categories, some additional attention points may be highlighted according to the FW geographical origin on one hand, and the FW producer and collection sources on the other hand.
Concerning the geographical origin, the statistical study showed that:
– TAN concentration for Asia (249.5 mg/L) is considered close to the optimal concentration (Liu & Sung, 2002) and TAN concentration from NA (1434.3 mg/L) and EU (989.0 mg/L) are higher but not considered as inhibitory as discussed previously.
– The high Na content in Asian FW (5.3 %DM) can be considered as moderately inhibitory of AD (Moletta, 2011), thus a special attention is advised in this continent.
Concerning the producer and collection sources, the waste characteristics highlighted that:
– The mean value of DM in FWGW (39.4%, Table 8) is higher than in the other subcategories. The treatment of such a FW in wet AD may be more difficult because of higher needs for additional water in the digester and because of mixing difficulties. Thus for such a collection type, a dry process might be preferable.
Lacks in terms of characterization of FW for a better understanding of ad stability
In this study we have shown high variations in FW characteristics. However, the chosen categories explain the variations of only 9 of the 37 studied characteristics. Therefore, other factors should be studied to identify other causes of variations of the FW characteristics. As an example, the typology of the FW compounds might be interesting since FW are composed by a variable content of fruits, vegetables, cereals, bakery, meat, dairy, among others (Heaven et al., 2013), as shown in Table 9. Furthermore, details about the different FW sorting or collection techniques or about the initial degradation degree caused by the FW storage before AD may be interesting to understand the variations of FW characteristics
Table of contents :
Chapitre 1 : Contexte et problématique de la thèse
1. Les déchets alimentaires
1.1. Production des déchets alimentaires et impact sur l’environnement
1.2. Exigences réglementaires spécifiques aux déchets alimentaires
1.3. Caractéristiques physicochimiques et typologie des déchets de cuisine .
2. La digestion anaérobie des déchets de cuisine
2.1. Etapes principales
2.2. Conditions physico-chimiques du processus
2.3. Potentiels et difficultés de la digestion anaérobie de déchets de cuisine
2.4. Pertinence du développement de systèmes de digestion anaérobie décentralisés .
2.5. La méthanisation par voie sèche des déchets de cuisine.
3. Prétraitement de déchets organiques
3.1. Prétraitements sur diffèrent substrats organiques
3.2. Prétraitement aérobie des déchets de cuisine
4. Objectifs de la thèse et méthodologie générale
4.1. Etape 1 : Caractérisation des déchets de cuisine
4.2. Etape 2 : Etude de l’impact du prétraitement aérobie sur la traitabilité anaérobie des biodéchets de cuisine
4.3. Etape 3 : Mise en oeuvre du prétraitement en amont d’un digesteur de laboratoire
Etape 1 : Caractérisation des déchets de cuisine
Chapitre 2 : Etat des connaissances sur la caractérisation des déchets de cuisine
2.1. Collection of literature data
2.2. Studied FW characteristics
2.3. Information characterizing data origins
2.4. Statistical analysis
3. Results of the statistical review of FW characteristics
3.1. Mean characteristics and representativeness
3.2. Impact of FW categories on FW quality variation
4. Discussion: Pros and cons of FW characteristics for valorisation through AD
4.1. Recall of AD fundamentals and influencing factors
4.2. Comparison between FW quality and optimal conditions for AD treatment
4.3. Lacks in terms of characterization of FW for a better understanding of ad stability
Chapitre 3. Influence de la composition et des caractéristiques physico-chimiques, biochimiques et microbiologiques sur la biodégradabilité de déchets de cuisine
2. Materials and methods
2.1. Food waste collection and sampling
2.2. Samples preparation
2.3. Statistical analysis and FW characteristics relationships
3. Results and Discussion
3.1. Characterization of the FW
3.2. Relationships between FW characteristics
Etape 2 : Etude de l’impact du prétraitement aérobie sur la traitabilité anaérobie des biodéchets de cuisine
Chapitre 4. Etude de l’impact d’une pré-aération sur les caractéristiques biochimiques et biologiques du déchet de cuisine et sa biodégradabilité anaérobie
2. Materials and methods
2.1. Composition and sampling of food waste feedstock
2.2. Experimental pre-treatment set-up
2.3. Preparation of the samples
2.4. Physical-chemical and biochemical analyses
2.5. Anaerobic biodegradability tests
2.6. Determination of enzymes
2.7. Determination of the microbial community
3. Results and discussion
3.1. Impact of O2 concentration on FW biodegradation kinetics
3.2. Anaerobic biodegradability of pre-treated FW
3.3. Changes in the biochemical characteristics of FW after aerobic pre-treatment
3.4. Effect of aeration on the biological activity in the FW
Etape 3 : Couplage d’un prétraitement aérobie en amont d’une digestion anaérobie par voie sèche de déchets de cuisine
Chapitre 5 : Etude de la performance du pilote de digestion anaérobie par voie sèche sur des déchets de cuisine
2. Matériels et méthodes
2.1. Substrat utilisé
2.2. Le pilote de méthanisation par voie sèche
2.3. Conditions à tester
2.4. Planification des expériences
2.6. Suivi analytique
2.7. Bilan matière par suivi de la DCO et de l’azote total (NTK).
2.8. Modélisation de cinétiques de production de méthane
2.9. Traitement statistique des résultats du plan d’expérience
3. Résultats et discussion
3.1. Potentiel biodégradable des substrats traités.
3.2. Validation de la reproductibilité de la mise en oeuvre de la méthanisation.
3.3. Performances de la digestion anaérobie en fonction des paramètres de mise en oeuvre et du substrat
3.4. Effet statistique de la mise en oeuvre du procédé et du substrat sur la performance de la digestion anaérobie
Chapitre 6 : Effet du prétraitement aérobie sur la performance de la digestion anaérobie en voie sèche des déchets de cuisine
2. Matériels et méthodes
2.1. Plan d’expérience
2.2. Description des essais expérimentaux
2.3. Analyse des paramètres de suivi de la méthanisation
2.4. Modélisation des cinétiques de production de méthane
3.1. Evaluation globale de la mise en oeuvre de la méthanisation
3.2. Performances de la digestion anaérobie en fonction des prétraitements appliqués
3.3. Modélisation des cinétiques de production
3.4. Analyse statistique des résultats
Chapitre 7 : Conclusions et perspectives
1. Avancées des connaissances et perspectives sur les caractéristiques des déchets dits de cuisine
2. Avancées et perspectives sur la compréhension de l’effet d’un prétraitement aérobie et des processus inhérents
3. Avancées et perspectives sur le couplage de procédés : prétraitement aérobie + digestion par voie sèche de type LBR
4. Questions d’ingénierie et de recherche au-delà des axes de la thèse
Synthèse des références