Characterization and Quantification of Faecal Sludge from Pit Latrines

Characterization and quantification of faecal sludge from pit latrines

Thèse

Catherine Bourgault

Doctorat en génie des eaux Philosophiæ doctor (Ph. D.)

Québec, Canada

Caractérisation et quantification des boues fécales issues des latrines à fosses

Thèse

Catherine Bourgault Sous la direction de :

Paul Lessard, directeur de recherche

Caetano Chang Dorea, codirecteur de recherche

RÉSUMÉ

L’assainissement autonome (non raccordés à un système d’égout) est largement répandu dans les pays à faible revenu. Selon les estimations, 2.7 milliards de personnes seraient desservies principalement par des technologies d'assainissement individuelles, dont près de la moitié seraient situées en zones urbaines des pays en voie de développement (WWAP 2017). Or, l’ampleur de la couverture des installations sanitaires autonomes commence à poser de nombreux problèmes et défis techniques aux municipalités des régions concernées. En effet, l’augmentation croissante des fosses devant être vidangées, couplé aux manques d’organisation et de ressources, et à l’absence de connaissances précises sur les caractéristiques et les quantités de boues fécales générées, ne permettent pas actuellement d’assurer une vidange sécuritaire et de planifier la gestion efficace des boues fécales dans ces endroits. On estime ainsi que 74 % des fosses des installations sanitaires autonomes seront vidangées de façon non sécuritaire par année, c’est-à-dire que les boues fécales retirées seront déversées dans les cours d’eau ou terrains environnants sans traitement. Ainsi, dans la perspective d’organiser la gestion des boues fécales dans le contexte des pays en voie de développement, il conviendrait d’améliorer les connaissances sur les taux de remplissage des latrines à fosse, ainsi que sur les données de caractérisation des boues fécales.

L’objectif général de cette thèse est d’améliorer la gestion des boues fécales dans le contexte des pays en développement. Pour ce faire, une série d’expérimentations a été réalisée, s’attardant à des problématiques précises de caractérisation ou de quantification des boues fécales. De façon plus précise, la première étude (CHAPITRE 4) visait la modélisation du taux d’accumulation des boues fécales en utilisant une régression linéaire multiple. Les résultats obtenus revoient un modèle impliquant les valeurs du volume des fosses et de l’âge des latrines comme principale variables indépendantes. La valeur du coefficient de détermination (r2) qui a été obtenue est de r2 = 0.41. Ceci qui est légèrement supérieure à la valeur obtenue à l’aide du modèle de bilan de masse développé par Brouckaert et al. (2013), pour lequel seulement 50% des latrines étaient à moins de 40% d’erreurs.

Par la suite, des travaux visant la caractérisation de l’effet inhibitif de l’azote ammoniacal sur la digestion anaérobie des boues fécales au sein des fosses ont été réalisés (CHAPITRE 5). L’objectif était de valider l’hypothèse selon laquelle les fortes concentrations en azote au sein des fosses (causées par la présence d’urine) influenceraient les mécanismes de biodégradation, et conséquemment les taux d’accumulation. La méthodologie développée à d’abords été testée et validée avec des boues de digesteur anaérobie (provenant d’un réacteur complètement mélangé). Par ailleurs, le protocole développé n’a pas mené à des résultats concluants en utilisant des boues fécales. Effectivement, bien que plusieurs tests d’activité méthanogène spécifique (specific methanogenic activity - SMA) aient été réalisées (en changeant le ratio inoculum (boues)/substrat (acétate), et la durée), aucun des tests exécutés n’a démontré une production de méthane.

À la lumière des résultats précédents (CHAPITRE 5), des travaux supplémentaires ont été réalisés visant à caractériser l’activité microbienne des boues fécales et de la matière fécale fraîche en termes des populations spécifiques à la digestion anaérobie (e.g. bactéries hydrolytiques et fermentatives, et méthanogènes méthylotrophes, acétoclastiques ou hydrogénotrophes) (CHAPITRE 6). Pour ce faire, la méthodologie utilisée consistait à mesurer la production de gaz (comme indicateur de l’activité microbienne) des échantillons de boues mélangés avec différents substrats spécifiques (e.i. acétate, méthanol, formate, glucose). Les résultats issus de cette étude ont démontré que l’utilisation des substrats de formate et de glucose présente l’activité microbienne la plus élevée lorsque mélangés avec une solution de matière fécale fraîche (soit de 112.17 et de 76.41 ml gaz/ g SV, pour le formate et le glucose, respectivement). Alors que l’utilisation des substrats de glucose et de méthanol ont résulté à l’activité la plus élevée avec des solutions de boues fécales plus âgées (soit de 129.15 et de 85.42 ml gaz/ g SV, pour le méthanol et le glucose, respectivement). Par ailleurs, une absence de l’activité méthanogène a été observé

III

en utilisant de l’acétate comme substrat pour les deux types de boues. Ces résultats concordent avec la littérature, où des tests de séquençages d’ADN n’ont pas mené à l’identification des archaea méthanogènes acétoclasitques au sein d’échantillons de boues fécales issues de latrines à fosses (Byrne 2016, Torondel 2017). Les résultats issus de cette étude remettent donc en perspective la présence d’une digestion anaérobie acétoclastique au sein des fosses, hypothèse qui est largement présumée dans de nombreux guides de conception des latrines à fosses conventionnelles.

Finalement la dernière étude constituait une étude préliminaire sur la caractérisation des propriétés de séchage de la matière fécale fraîche (CHAPITRE 7). Plus particulièrement par la détermination et par la modélisation des isothermes de sorption. Cette étude s’inscrit dans le cadre actuel où les techniques de séchage pour le transport efficient des boues gagnent en popularité, alors qu’un manque de données sur les propriétés des boues persiste. Ainsi, les résultats obtenus démontrent que les isothermes de sorption tendent à suivre une forme sigmoïde, laquelle est caractérisée par deux points d’inflexion de surface et une augmentation rapide de la teneur en eau à des valeurs d'humidité relative supérieures à 75 % HR. D’une première part, ceci signifie que la teneur en eau de la matière fécale semblerait être plus facile à extraire jusqu'à une valeur d’environ 1 à 1.5g H2O/g d’échantillon - base humide. D’autre part, le modèle de Guggenheim-Anderson-de Boer (GAB) sembleraient être le plus approprié pour décrire les courbes d’isothermes, avec une estimation acceptable (Pr (>|t| > 0.05).

TABLE OF CONTENTS

RÉSUMÉ ...... III

LIST OF TABLES ...... X

LIST OF FIGURES ...... XII

LIST OF ACRONYMS ...... XIV

REMERCIEMENTS ...... XVI

AVANT-PROPOS ...... XVIII

1. INTRODUCTION ...... 1

2. LITERATURE REVIEW ...... 4

2.1 ON-SITE SANITATION: PIT LATRINE COVERAGE ...... 4

What is a pit latrine? ...... 4

Design criteria of pit latrines ...... 5

Pit latrine coverage in the world ...... 5

2.2 CONSEQUENCES ASSOCIATED WITH POOR PIT LATRINES MANAGEMENT ...... 6

2.3 FAECAL SLUDGE MANAGEMENT ...... 7

2.4 FS ACCUMULATION RATE IN PIT LATRINES ...... 9

Pit design criteria...... 9

Field studies on measurement of FS accumulation ...... 10

Emptying frequency ...... 13

Modelling of pit filling rate ...... 18

2.5 FACTORS AFFECTING ACCUMULATION RATES OF FS IN PITS ...... 20

Operational and design related parameters ...... 20

Biodegradation mechanisms related parameters ...... 25

2.6 FAECAL SLUDGE CHARACTERISTICS ...... 32

FS from different type of sanitation systems ...... 38

Effects of sludge age on faecal sludge properties ...... 39

Faecal sludge characterization from sampling point ...... 39

From rainy season to dry season ...... 40

2.7 CONCLUSION ...... 40

3. OBJECTIVES OF THE THESIS ...... 43

3.1 OUTLINE OF THE THESIS ...... 44

4. MULTIPLE LINEAR REGRESSIONS ANALYSYS FOR FS ACCUMULATION RATES

PREDICTIONS ...... 46

4.1 INTRODUCTION ...... 46

4.2 METHODOLOGY ...... 47

Data collection ...... 47

Multilinear regression model ...... 48

4.3 RESULTS AND DISCUSSION ...... 48

Modelling the accumulation rate ...... 48

Validation of the coefficients ...... 49

Analysis of model residuals ...... 51

Model validation ...... 52

4.4 CONCLUSION ...... 54

5. THE CHARACTERIZATION OF THE AMMONIA INHIBITION OF ANAEROBIC

DIGESTION OF FAECAL SLUDGE SAMPLES ...... 56

5.1 INTRODUCTION ...... 56

5.2 MATERIAL AND METHODS ...... 57

Fresh faeces, faecal sludge and anaerobic sludge sampling ...... 58

Preparation of the diluted faecal sludge for the SMA test ...... 60

Source of nitrogen ...... 60

Nutrient solution ...... 60

SMA-based anaerobic toxicity assay ...... 61

Validation of the SMA-based anaerobic toxicity assay protocol ...... 62

5.3 RESULTS AND DISCUSSION ...... 63

Characterization of the FF and FS ...... 63

Validation of the SMA-based anaerobic toxicity assay protocol ...... 65

Characterization of ammonia inhibition on fresh faeces and faecal sludge ...... 67

General Discussion ...... 69

5.4 CONCLUSION ...... 70

6. SPECIFIC METHANOGENIC ACTIVITY TESTS FOR THE CHARACTERIZATION OF

MICROBIAL ACTIVITY INTO PIT LATRINE ...... 71

6.1 INTRODUCTION ...... 71

Microbial ecology of anaerobic digestion ...... 71

Methanogens from the human gut ...... 74

Microbial ecology of pit latrines ...... 74

Microbial ecology of untreated wastewater – from conventional sewage ...... 76

6.2 MATERIAL AND METHODS ...... 79

Fresh faeces and faecal sludge sampling ...... 79

Choice of substrate for the SMA test ...... 80

SMA-based anaerobic assay ...... 80

Gas analysis ...... 81 VII

Biomethane potential (BMP) test ...... 81

6.3 RESULTS AND DISCUSSION ...... 82

BMP test of FS and FF ...... 82

Validation: SMA anaerobic sludge from a CSTR ...... 83

SMA from faecal sludge solution ...... 86

SMA from faeces solution ...... 88

General Discussion ...... 90

6.4 CONCLUSION ...... 92

7. MEASURING MOISTURE SORPTION ISOTHERM OF FRESH FAECES ...... 93

7.1 INTRODUCTION ...... 93

Moisture distribution characteristics measurement methods ...... 94

Moisture sorption isotherm ...... 95

Modelling of the moisture sorption isotherm ...... 97

7.2 MATERIALS AND METHODS ...... 97

Fresh faeces sampling and initial characterization ...... 98

Determination of sorption isotherms ...... 98

Mathematical models of the sorption isotherm ...... 101

7.3 RESULTS ...... 101

Fresh faeces characterization ...... 101

Moisture sorption isotherm from two different sample masses ...... 102

Modelling of isotherms ...... 103

7.4 CONCLUSION ...... 108

8. DISCUSSION GÉNÉRALE ...... 110

8.1 MODELISER LES TAUX D’ACCUMULATION DES BOUES FECALES AU SEIN DES LATRINES A FOSSES

A L’AIDE D’UNE REGRESSION LINEAIRE MULTIPLE ...... 110 VIII

Limites de l’étude et recommandations...... 110

8.2 CARACTERISER L’IMPACT DE L’AZOTE AMMONIACAL SUR LES PERFORMANCES DE DIGESTION

ANAEROBIE DES BOUES FECALES ET DE LA MATIERE FECALE ...... 111

Limites de l’étude et recommandations...... 112

8.3 CARACTERISER L’ACTIVITE MICROBIENNE DES BOUES FECALES ET DE LA MATIERE FECALE EN

TERMES DES POPULATIONS SPECIFIQUES A LA DIGESTION ANAEROBIE ...... 113

Limites de l’étude et recommandations...... 114

8.4 DETERMINER LES CARACTERISTIQUES DE SECHAGE DE LA MATIERE FECALE FRAICHE ...... 115

Limites de l’études et recommandations...... 116

9. CONCLUSION GÉNÉRALE ...... 117

10. REFERENCES ...... 120

ANNEX 1: DATA ...... 130

10.1 TODMAN ET AL. (2014) ...... 130

10.2 FOXON AND STILL (2012) ...... 131

10.3 SCHOEBITZ ET AL. (2016) ...... 131

ANNEX 2: NORMALITY OF THE DATA ...... 134

ANNEX 3: SUBMITTED PUBLICATIONS ...... 137

LIST OF TABLES

Table 2-1 Types of pit latrines (Nikiema et al. 2017) ...... 4

Table 2-2 Proposed accumulation rate (liters per capita per year, l/cap/yr) for pit latrine design purposes

(Wagner and Lanoix 1958) ...... 10

Table 2-3 Overview of studies that have focused on the measurement of accumulation rate...... 12

Table 2-4 Estimate of the emptying frequency of pit latrine ...... 13

Table 2-5 The faecal sludge accumulated in on-site sanitation facilities in 30 cities of Africa and Asia

(Chowdhry and Kone 2012) ...... 17

Table 2-6 Parameters of the Brouckaert et al. (2013) pit filling rate model ...... 18

Table 2-7 Summary of studies assessing FS accumulation rates (L/cap/day) or FS biodegradation and characteristics, with factors that influence them (adapted form (Nakagiri et al. 2015) ...... 21

Table 2-8 Summary of studies assessing FS accumulation rates (L/cap/day) or FS biodegradation and characteristics, with factors that influence them (adapted form (Nakagiri et al. 2015) (cont’d) ...... 21

Table 2-9 Operating conditions of the anaerobic digestion process ...... 29

Table 2-10 Claims made for pit latrine additives and the reality (Foxon and Still 2012) ...... 32

Table 2-11 Physical and chemical properties of faecal sludge ...... 34

Table 2-12 Faecal sludge characterization from different type of containments (Koottatep et al. 2013) ... 37

Table 2-13 Differences between dry and rainy season (Bassan et al. 2013) ...... 40

Table 2-14 Gaps in characterization and quantification of FS ...... 41

Table 4-1 Summary of regression coefficients for models for subsets of the data...... 49

Table 4-2 Analysis of variance table ...... 49

Table 5-1 Methods used for the physico-chemical characterization of FF and FS samples ...... 59

Table 5-2 Nutrient medium recipe (adapted from Angelidaki et al. 2006) ...... 61

Table 5-3 Characterization of the anaerobic sludge from the CSTR (n=3) ...... 63

Table 5-4 Physico-chemical parameters of faecal sludge and fresh faeces samples (n=4) ...... 64

Table 5-5 COD of fresh feaces from literature (adapted from (Rose et al. 2015) ...... 64

Table 5-6 Average methane production of anaerobic sludge in response to varied (NH4)2CO3 concentrations

(n=3) ...... 66

Table 5-7 Summary of SMA results for both FS and FF – ISR between 2.08 – 35.28 ...... 68

Table 6-1Stoichiometric equation and change of free energy of some acetogenic reactions ( at NTP – assuming neutral pH and (Henze 2008) ...... 73

Table 6-2 Proportion of the main phylogenetic groups among different digesters ...... 77

Table 6-3 Characteristics of inoculum and substrate ...... 83

Table 6-4 Characteristics of the anaerobic sludge from the CSTR ...... 84

Table 6-5 SMA results – anaerobic sludge CSTR ...... 85

Table 6-6 Characteristics of the faecal sludge solution (n=3) ...... 87

Table 6-7 SMA results – faecal sludge solution ...... 87

Table 6-8 Gas analyses for each substrate – faecal sludge solution...... 88

Table 6-9 Characteristics of the fresh faeces solution (n=3) ...... 89

Table 6-10 SMA results – fresh faeces solution ...... 89

Table 6-11 Gas analyses for each substrate – fresh faeces solution...... 90

Table 6-12 Compilation of SMA results ...... 90

Table 7-1 Relative humidity (%) of the saturated salt solution at 35°C (CRC, 1977) ...... 100

Table 7-2 Isotherm models used for fitting experimental data ...... 101

Table 7-3 Characteristics of fresh faeces (n=6) ...... 102

Table 7-4 Linear regression results for linearized BET model ...... 105

Table 7-5 Predicted value analysis for selected model for 1.5g and 5.0 g of faeces as initial mass ...... 107

LIST OF FIGURES

Figure 2-1 Population (%) served by different types of sanitation systems (WWAP 2017) ...... 6

Figure 2-2 Gaps in faecal sludge management (Strande et al. 2014) ...... 8

Figure 2-3 Comparison between the observed and simulated filling rates ...... 19

Figure 2-4 Four stages theoretical model of FS digestion into pits (Buckley et al. 2008) ...... 26

Figure 2-5 Evolution of total COD and organics solids (%) as a function of pit latrine depth (n = 16 at the same time), from fresh faeces (adapted from Nwaneri, 2009) ...... 28

Figure 4-1Comparison between the observed and simulated filling rates. Solid diagonal represents the perfect fit. Results of the highest accumulation rates observed are circled...... 50

Figure 4-2 Comparison between the observed and simulated filling rates...... 51

Figure 4-3 Analysis of model residuals ...... 52

Figure 4-4 Plot of predicted value against the observed values of log transferred accumulation rate (log

L/yr) for (red squares), (Still and Foxon 2012b) (green circles) and (Schoebitz et al. 2016)(blue diamonds)

...... 53

Figure 5-1 Faecal sludge sampling at Parc de la Jacques-Cartier. Sampling device and sampling interface are shown in the left picture, and FS consistency is shown in the right picture...... 59

Figure 5-2 SMA tests with anaerobic sludge and different concentrations of total ammonia nitrogen ...... 66

Figure 5-3 SMA tests with faecal sludge and different concentrations of total ammonia nitrogen (ISR =

2.18) ...... 68

Figure 6-1 Anaerobic digestion pathway and the microorganisms involved (Cavinato et al. 2017) ...... 72

Figure 6-2 Cumulative methane production from FS and FF sample as a substrate and CSTR sludge as the inoculum ...... 83

Figure 6-3 SMA (ml CH4/g VS) for anaerobic sludge ...... 84

Figure 6-4 SMA (ml gas/g VS) for faecal sludge...... 86

Figure 6-5 SMA (ml gas/g VS) for fresh faeces...... 89 XII

Figure 7-1 General shape of sorption isotherms for a hypothetical biological matrix showing the relationship between water content (Xeq) and water activity (aw) (adapted from (Vaxelaire 2001) ...... 96

Figure 7-2 Experimental set-up for the MSI determination ...... 99

Figure 7-3 Boxplot of equilibrium moisture content data for each given relative humidity at a temperature of 35˚C and an initial faecal sample with mass a) 1.5 g and b) 5g ...... 102

풂풘 Figure 7-4 Estimation of constants C and Mo for BET equation - linear regression on as a function (ퟏ−퐚퐰)푴 of aw ...... 105

Figure 7-5a -1.5g and 7-5b-5g faeces as initial mass. Experimental data (.)(Average) and predicted values;

BET (___), GAB (---), FH (…) model predicted adsorption at 34C...... 106

Figure 7-6 Desiccation rate for 5g sample at 35ºC and at different %RH ...... 108

XIII

LIST OF ACRONYMS

Acronyms Definition Aw Water Activity BOD5 [mg/l] Biological oxygen demand Cap Capita COD Chemical oxygen demand CSTR Continuous stirred-tank reactor DW Dry weight FS Faecal sludge FF Fresh faeces FSM Faecal sludge management GW Groundwater HH Household ISR Inoculum to substrate ratio NH3 Ammonia N/A Not available OSS On-site sanitation system PL Pit latrine ST Septic tanks SMA Specific methanogenic activity SVI [ml/g TSS] Sludge Volume Index TN Total nitrogen TAN Total Ammonia Nitrogen TSS [mg/l] Total suspended solid TS Total solids VSS [% SS] Volatile suspended solid VIP Ventilated improved pit VS Volatile solids Yr Year

XIV

¨Y en a qui ont tout’ pis tout’ les autres, y ont rien. Change-moi ça.¨

Richard Desjardins

REMERCIEMENTS

La liste de remerciements sera longue! Si un doctorat peut être considéré comme étant un cheminement individuel, il s’appuie, se construit, et serait impossible sans l’aide et les encouragements de plusieurs personnes et organismes que je tiens sincèrement à remercier.

Mes remerciements vont tout d’abord à mon directeur et mon co-directeur de recherche, le professeur Caetano Dorea et le professeur Paul Lessard, pour m’avoir soutenu dans toute cette aventure. Leurs nombreux conseils et remarques ont énormément contribué à l’avancement de mes travaux. Qu’ils en soient remerciés. Qu’ils soient aussi remerciés pour leur gentillesse, leur disponibilité et pour les nombreux encouragements qu’ils m’ont donnés tout au long de mes travaux de recherche. Et sans doute le plus important, je les remercie pour avoir cru en moi. Ainsi, ils m’ont donné le plus précieux des cadeaux : la confiance.

Je tiens également à remercier très sincèrement le prof. Chris Buckley, qui m’a soutenue lors de mon séjour en Afrique du Sud. Et tellement plus! Sa rencontre a été pour moi un point tournant dans mon cheminement, tant au niveau professionnel que personnel. Sa générosité et son attitude toujours enthousiaste à répondre à mes nombreuses questions ont certainement contribué à amener ma recherche à un autre niveau. Aussi, je lui suis extrêmement reconnaissante pour m’avoir présentée à de nombreux chercheurs du domaine. De chaleureux remerciements vont également à toute l’équipe du Pollution Research Group de l’Université du Kwazulu-Natal, en Afrique du Sud, pour laquelle j’éprouve désormais une grande affection.

Mes remerciements vont également au professeur Peter Vanrolleghem, au Dr. Yann Lebihan et Dr. Josiane Nikiema pour leurs disponibilités, ainsi que pour leurs conseils et avis donnés sur de nombreux sujets. Leurs expertises ont contribué de façon importante à l’orientation du projet et à mon cheminement professionnel. Je tiens également à remercier la professeure Céline Vaneeckaute pour avoir accepté de faire partie de mon jury de thèse.

De plus, ce travail n’aurait pas été possible sans le soutien du Conseil de recherches en sciences naturelles et en génie du Canada (CRSNG) qui m’a permis, grâce à la bourse de recherche Alexander-Graham-Bell, de me consacrer pleinement à l’élaboration de ma thèse. De plus, cette recherche a également été supportée par le programme de bourse à la découverte du CRSNG en lien avec les travaux du professeur Dorea. Merci également à Aerosan Toilets pour leur contribution à la partie séchage des boues de ma thèse. XVI

Ma profonde reconnaissance va à mes collègues et amis du département de génie civil et de génie des eaux de l’Université Laval. Je me considère tellement chanceuse d’avoir pu cheminer dans mes études étant entourée d’une telle équipe. Un merci particulier à Stephanie Guilherme, Antoine Thiboult, Thomas Maere, Elena Torfs, Cyril Garneau, Sylvie Leduc, Queralt Plana Puig, Sovanna Tik, Julia Ledergerber, Maxine Dandois-Fafard, Jean-David Therrien et Bernard Patry pour m’avoir toujours écoutée, et pour avoir partagé mes problèmes et victoires. Je suis profondément honorée de pouvoir désormais les compter parmi mes amies les plus chers. Go go go, on ne lâche pas !! ☺

Merci également à Michel Bisping pour son soutien dans l’élaboration de mon montage et pour ses nombreux conseils. Merci au secrétariat du département de génie civil et de génie des eaux de l’Université Laval pour son support avec mes nombreuses questions administratives. Je tiens également à remercier particulièrement Claire Remington, étudiante à la maîtrise de l’Université de Victoria, pour m’avoir aidé dans la vérification de l’anglais de mon document de thèse. Je tiens à souligner sa générosité et la remercie pour être une si belle personne.

Et finalement, il m’est impossible d’oublier ma merveilleuse famille, mes parents, ma sœur, mon beau- frère, leurs enfants, Clara et Philippe, ainsi que mon meilleur ami et amoureux, Tom. Pour eux, les mots me manquent. Je leur dois tellement que l’écriture ne me permet pas d’exprimer la hauteur de mes sentiments à leurs égards. J’ose espérer qu’ils le savent…

XVII

AVANT-PROPOS

Avant d’entreprendre la lecture de cette thèse certaines informations doivent être mentionnées.

Premièrement, sur le choix de la forme; la thèse se divise en 9 chapitres, dont 4 rapportent les travaux expérimentaux. Ces chapitres sont écrits sous forme d’article scientifique et rédigés en anglais afin de faciliter leurs publications. Un article et une courte communication ont d’ailleurs été soumis entre la date de la soutenance et le dépôt final de cette thèse, lesquelles sont présentés en ¨ANNEX 3¨. Plus précisément, voici la description des articles soumis:

Bourgault C, Lessard P, Remington C, and Chang Dorea C, (2018). Experimental determination of moisture sorption isotherm of faecal sludge, Manuscript submitted on December 4th for publication in the Water Journal. Accepted.

Bourgault C, Lessard P, and Chang Dorea C, (2018).Pit latrine design: Is it rocket science after all? Manuscrit submitted on September 12th for publication in the npj Clean Water journal. Under review.

Deuxièmement, il convient également de mentionner que cette étude fut la première sur le sujet des boues fécales à être réalisée à l’Université Laval. C’est-à-dire qu’aucune autre étude n’a été réalisée sur ce sujet avant celle-ci. Un temps considérable a donc été consacré à l’élaboration et l’adaptation des protocoles d’échantillonnage et de caractérisation des boues.

XVIII

1. INTRODUCTION

Dans les pays à faible revenu, les services d’assainissement sont souvent insuffisants et inefficaces, ceci notamment en raison d’un manque de ressources pour assurer une vidange sécuritaire des installations à fosse (e.g. latrine à fosse, fosse septique), de même qu’une insuffisance de stations de traitement fonctionnelles et adaptées aux boues fécales. Ce manque d’organisation sanitaire entraîne de nombreuses conséquences, dont notamment les rejets d’eaux usées non traitées dans l’environnement. En effet. si l’on estime qu’environ 70% des eaux résiduelles municipales et industrielles seraient traité dans les pays à revenu élevé, dans les pays à faible revenu, seul 8% de ces eaux usées subirait un traitement, quel qu’il soit

(WWAP 2017). La mauvaise gestion sanitaire entraîne également la prolifération des maladies fécales- orales tels que le choléra et la typhoïde. Selon l’Organisation Mondiale de la Santé, environ 2 millions de personnes meurent chaque année suite à de fortes diarrhées, les cas les plus répertoriés étant les enfants de moins de cinq ans (UNICEF/WHO 2011). Aussi, on estime que l’Afrique subsaharienne perd environ 5 % de son PIB à cause du manque d’assainissement, soit quelque 28,4 milliards de dollars chaque année

(WWAP 2017). En effet, en affectant directement la qualité de vie, la capacité de travail et l’éducation, un manque d’assainissement pèsera lourdement sur le dynamisme de l’économie. Une amélioration de l’offre des services sanitaires s’avère indispensable.

Or, la mise en place d’une gestion sécuritaire et efficace des boues fécales n’est pas une chose simple. Parmi les principaux obstacles, mentionnons la disponibilité des données de caractérisation et de quantification des boues fécales qui demeure un défi persistant. Comparativement à la caractérisation des eaux usées domestiques issues des systèmes de traitement centralisés (principalement retrouvés dans les pays à revenu

élevé), la caractérisation des boues fécales est moins bien documentée, voire carrément négligée dans plusieurs endroits. Ce manque de données peut entraîner des conséquences et des coûts importants, comme l’a démontré une étude portant sur la station de traitement des eaux usée de Ouagadougou, où une mauvaise

évaluation des charges organiques à traiter aurait conduit au surdimensionnement de la station; la capacité de la station serait deux fois plus grande que nécessaire selon l’étude (Bassan et al. 2013a). L’exemple de

Ouagadougou n’est certainement pas le seul. En effet, d’après une analyse de 2013, sur les 181 pays de l’étude, il a été démontré que seul 55 disposent d’information sur la génération, le traitement et l’utilisation des eaux usées, et les autres ne disposent que d’information partielle ou n’en ont pas du tout (Sato et al.

2013).

Ce projet de recherche vise à répondre à la problématique de caractérisation et de quantification des boues fécales, en se consacrant seulement sur les boues fécales issues des installations sanitaires autonomes de type latrines à fosse. Dans une première partie, une revue approfondie de l’état des connaissances sur l’utilisation des latrines, leurs caractéristiques, ainsi que sur les recherches antérieures portant sur la quantification des taux d’accumulation et sur la caractérisation des boues fécales sera présentée. Par la suite, selon une analyse des besoins ressortant de la littérature, une série d’expérimentations a été réalisée, s’attardant à des problématiques précises de caractérisation ou de quantification des boues fécales. De façon plus précise, la première étude (CHAPITRE 4) s’attarde à la modélisation des taux d’accumulation des boues fécales en utilisant une régression linéaire multiple. Par la suite les chapitres 5 et 6 constituent le cœur de la thèse en présentant les méthodes développées pour caractériser l’activité méthanogène au sein des fosses, le but étant d’étudier les principales conditions opératoires favorables (ou non) au développement des archaea méthanogènes au sein des fosses. Ceci est un sujet important visant à améliorer la compréhension des mécanismes de biodégradation des boues fécales, et par extension à la prédiction des caractéristiques des boues fécales et taux d’accumulation au sein des fosses. Finalement la dernière étude constitue une étude préliminaire sur la caractérisation des propriétés de séchage de la matière fécale fraîche, plus particulièrement sur la détermination et la modélisation des isothermes de sorption. Cette étude s’inscrit dans le cadre actuel où les techniques de séchage pour le transport efficient des boues gagnent en popularité, alors qu’un manque de données sur les propriétés des boues persiste.

Finalement, dès lors que l’on s’intéresse aux problématiques environnementales, on constate rapidement les nombreuses répercussions que celles-ci ont sur des milliers d’individus. La précarité de l’assainissement mondial en est un bon exemple. En effet, malgré de nombreux progrès visibles, le manque d’installations sanitaires demeure encore aujourd’hui une menace grave pour l’ensemble de la planète, laquelle pourrait inverser des décennies d’innovation si elle n’est pas traitée. Le message est donc clair; il faut s'attaquer à cette problématique de façon inclusive et coordonnée. Ce projet de recherche agit en ce sens.

2. LITERATURE REVIEW

2.1 On-site sanitation: pit latrine coverage

What is a pit latrine?

Pit latrines (PLs) are the simplest form of dry sanitation systems1, typically consisting of a pit dug in the ground covered with a slab for ease of use and a superstructure for user privacy. Although that is the basic design of a pit, it can vary in construction according to available water/space/construction materials, cultural norms, emptying practices, soil/hydrogeological conditions, climate, construction/operation/maintenance costs, etc. A summary of PL variants is described in Table 2-1

Table 2-1 Types of pit latrines (Nikiema et al. 2017) Type of latrine Description Traditional PLs - commonly used in developing countries, e.g. Africa, Latin America and Asia - consist of a simple pit, covered with logs - sometimes have walls - usually have no roofs - easy to build and require no specialized skills Slit-trench latrines - the simplest type of pit latrine - consist of a narrow (3 to 6 feet) trench, which is 1 to 2 meters deep - used by squatting, with the user’s feet positioned on either side of the trench - can have various structures for sitting on or leaning against (e.g. logs, planks, branches) placed horizontally over the vertical pit Conventional - similar to the traditional latrine, but built from solid materials, such as bricks improved PLs - usually have walls and roofs - can be improved by putting ashes in the latrine to reduce smells and flies Double PLs - consist of two shallow pits that are used alternately - built with a single concrete slab and superstructure that is mounted over one of the two shallow pits - considers recycling of materials Ventilated improved - consist of the normal pit, but with a pipe (vent pipe) fitted to the pit and a screen (fly pit (VIP) latrines screen) at the top outlet of the pipe. - an improvement to overcome fly, mosquito nuisance and odours Pour-flush latrines - are most common in southern Asia - have a water seal fitted to the drop hole, which prevents smells and flies from entering the shelter or superstructure through the pit - requires water to flush the toilet - could experience problems if water is scarce and/or with the water seal

1 Dry systems may include water. However, the amount of water used is low. In particular, faecal sludge is considered dry when the solids content is > 20% and moisture content < 80%. This system is not connected to sewers networks (Nikiema et al. 2017). 4

Table 2-1 Types of pit latrines (Nikiema et al. 2017) (cont’d) Type of latrine Description Twin pit water seal - main components of such a toilet are the water seal pan/ trap arrangement, squatting toilet platform, junction chamber, two pits and a superstructure - junction chamber has one inlet and two outlets (connected to the leach pits) Single offset pit - consists of water seal pan, a temporary platform, a junction chamber, a water seal toilet temporary/permanent superstructure and a single pit - pit is constructed away from the squatting platform and connected to the same by a pipe through a junction chamber Eco-san toilets - dry toilet, based on recycling principles: urine is collected separately from feces - consist of double vault compost latrines (two water-tight chambers)

Design criteria of pit latrines

According to best practices, pit latrines should be at least 30 meters from all water sources, and the bottom of the pits should be at least 1.5 meters above the hydrostatic level (Sphere Project 2011). As for the dimensioning of the pits, the model proposed by Mara (1984) is still widely used (Eq.2.1). Mara’s model calculates that the volume of the pits is equal to the rate of accumulation multiplied by the number of users and the life of the pits :

Volume of pits = r (accumulation rate) (l/cap/yr) x P (number of users) x n (lifetime) (year) (Eq.2.1)

In practice, such a model may not always be respected; the dense conditions encountered in urbanized areas may allow for the digging of pits at an ideal size and, as will be seen in the following section, it is difficult to accurately estimate the rate of sludge accumulation.

Pit latrine coverage in the world

As we can see on Figure 2-1, approximately 1.3 billion people rely on some form of pit latrine systems as their primary mean of sanitation worldwide, most being used in developing countries (95%). This based on the population with faecal sludge management (FSM) demand and the percentage of the population served by dry or flush/poor flush pit latrines (yellow circles and light blue and blue section of (Figure 2-1)). Sewer

connections are mostly found in high-income countries, in middle-income countries of Latin America and in some urban areas in China (Kjellén et al. 2011). In general, the number of households connected to sewer systems correlates (to a greater or lesser extent) with the connections to a water supply, although always in much lower proportions (WWAP 2017).

Figure 2-1 Population (%) served by different types of sanitation systems (WWAP 2017)

2.2 Consequences associated with poor pit latrines management

In many urban regions of the world, the increasing number of pit latrines is causing numerous problems and technical challenges. Many municipalities have not yet put the necessary strategies, policies and budgets in place to maintain (e.g. desludging and treatment) these on-site sanitation systems, resulting in the contamination of the environment (Strande et al. 2014), and directly affecting the quality of life, the working capacity of the inhabitants, the education and the economy. It is estimated that only 26% of urban

and 34% of rural sanitation and wastewater services worldwide effectively prevent human contact with excreta along the entire sanitation chain and can therefore be considered safely managed (Hutton and

Varughese 2016). On average, high-income countries treat about 70% of the municipal and industrial wastewater they generate. That ratio drops to 38% and to 28% in upper middle-income and lower middle- income countries respectively, and in low-income countries, only 8% undergoes treatment of any kind

(WWAP 2017). As a result, water pollution is worsening in most rivers across Africa, Asia and Latin

America. In the seas and ocean, de-oxygenated dead zones caused by the discharge of untreated wastewater are growing rapidly, affecting an estimated 245,000 km2 of marine ecosystems, impacting on fisheries, livelihoods and food chains (WWAP 2017). Furthermore, poor wastewater management leads to the proliferation of faecal-oral diseases such as cholera and typhoid. According to the World Health

Organization (WHO), approximately 2 million people die each year due to severe diarrhoea; most of them are residents from poor peri-urban areas in developing countries (UNICEF/WHO 2011). From an economic perspective, scientific evidence has demonstrated that the economic cost associated with poor sanitation is substantial (Van Minh and Nguyen-Viet 2011). It was estimated that sub-Saharan Africa loses about 5% of its GDP due to lack of sanitation, or about $ 28.4 billion each year (WWAP 2017).

2.3 Faecal sludge management

Poor sanitation and its multiple consequences are a complex challenge. Many reasons could explain the difficulties that municipalities are facing in providing adequate sanitation services to their citizens due to numbers of interactions between various sectors (i.e. financial, environmental, political, etc.) Figure 2-2.

Figure 2-2 Gaps in faecal sludge management (Strande et al. 2014)

As can be seen in Figure 2-2, the implementation of safe faecal sludge management strategies implies several fields of interventions. These are (starting from the top): (1) the planning of sludge collection and transport, (2) establishing sanctions to counteract illegal dumping in the environment, (3) the availability of safe and suitable treatment options, (4) the acceptability of safe disposition options (5) adequate tariffs to ensure the sustainability of the sanitation structure, and (6) the implementation of a communication platform to ensure adequate interactions between the various stakeholders. The establishment of good practices for the safe management of faecal sludge (FS) represents major social, technologic and economic challenges for developing countries.

From a technical point of view, in order to respond to the ever-increasing quantities of FS generated in cities, the implementation of functional treatment plants is required. However, one major gap in developing appropriate and suitable FS treatment options is the ability to understand and monitor faecal sludge characteristics, faecal sludge quantification (accumulation rate) and the factors affecting the variabilities of 8

the sludge in both (characteristics and accumulation rates). The following sections present the state-of-the- art on sludge accumulation rates within pit latrine systems (section 2.4 and 2.5) and on characterization

(section 2.6).

2.4 FS accumulation rate in pit latrines

Given the fact that pits are not standardised environments (e.g. variability in volume, lined or not-lined, type of soil, etc.), knowing precisely the FS accumulation rate is often very challenging, as the collection of information can be too labour intensive (Strande et al. 2014). Methodologies for providing estimates of faecal sludge accumulation rates have thus been used by various authors and can be organized in 4 categories (detailed in 2.4.1 to 2.4.4), namely: (1) for design criteria purpose, (2) from direct measurement according to field studies, 3) based on emptying frequency from household surveys and (4) based on mass balance modeling.

Pit design criteria

One of the first accounts describing faecal sludge accumulation rates into pits were suggested by Wagner and Lanoix (1958) to provide basic recommendations for the sizing of pits. These estimates were based on theoretical assumptions of biodegradation (i.e. that wet pits could easily perform biodegradation processes) and field observations from data collected by the World Health Organization (WHO). The proposed accumulation rates are presented in Table 2-2.

Wagner and Lanoix (1958) suggest the FS accumulation rate is lower in wet pits where biodegradable materials are used for anal cleansing, this is based on the assumption that better consolidation, accelerated decomposition and removal of the finer material should be observed in wet conditions (Wagner and Lanoix

1958).

Table 2-2 Proposed accumulation rate (liters per capita per year, l/cap/yr) for pit latrine design purposes (Wagner and Lanoix 1958)

Description of pit latrine Pit accumulation rate (l/cap/yr) (wet or dry) With Biodegradable anal cleansing With Unbiodegradable anal cleansing material material Wet pits 40 60

Dry pits 60 90

The guidelines proposed by Wagner and Lanoix are still cited as references for PL design (Foxon 2010,

Franceys et al. 1992, Harvey et al. 2002, Heinss et al. 1998a, Heinss et al. 1998b, Murphy 2015, Still and

Foxon 2012a). However, they should be considered as a generalization since they are proposing that only two factors are affecting the filling rate of sludge: 1) the type of soil (sandy or clay) and 2) the type of material used for anal cleansing (i.e. biodegradable : water, paper, etc., or unbiodegradable: stones, corn cobs, coconut husk, etc.). As a matter of fact, several other parameters have proven to also impact the FS accumulation rates in pits (these parameters will be discussed in section 2.5.) Thus, guidelines in Table 2-

2 should only be used when no other local measurement of the actual filling rate is available (Franceys et al. 1992).

Field studies on measurement of FS accumulation

Table 2-3 presents a review of work that has focused on actively measuring the accumulation rate of FS in

PLs. Based on information obtained from household surveys, Baskaran (1962) reported a pit accumulation rate of 34 l/cap/yr in West Bengal (India), where ablution water was used (biodegradable anal cleansing material). In Brazil, an average pit filling rate of 47 l/cap/yr is reported in dry pit conditions; in other words, this means that a pit of 1 m3 capacity will serve a family of five for four years (Sanches and Wagner 1954).

Furthermore, in Zimbabwe, FS accumulation rates were reported to rarely exceed 20 l/cap/yr as the latrines observed in the study were regularly washed down (wet pits), and biodegradable material was used for anal

cleansing - all of which is supposed to contribute to lower the accumulation rate. Overall, the observations presented above are lower in comparison to guidelines proposed by Wagner and Lanoix (1958). This confirms the difficulty to extrapolate the accumulation rate estimations to other locations.

Furthermore, in more comprehensive studies commissioned in South Africa, the FS accumulation rate was measured over time (Bhagwan et al. 2008, Norris 2000, Still and Foxon 2012b). The sludge level was determined by measuring the vertical distance between a fixed datum (top of toilet seat) and the sludge, so the change in the vertical distance indicates the change in sludge volume (knowing the surface area of the pit) (Norris 2000). Information on the number of users came from a household survey (average number of users was used).

A great variability in results (3 to 264 l/cap/yr) has been observed, even for pits located in the same geographical area (i.e. a comparatively homogeneous environment) (Norris 2000). The most important factors affecting these rates were understood to be drainage from the pit and the extent to which pits are used for disposal of other household waste (Foxon 2010). The study do not indicate a strong relationship between household size and the rate at which the pit fills (Foxon 2010). This was also confirmed by Bakare

(2012), who showed that there was no correlation between available data for pit filling rates and reported number of users. Bakare (2012) concluded that this was due to uncertainty in the pit filling rate data and confusion surrounding the interpretation of average number of pit users. Indeed, the household size may have increased or decreased over the period that the pit has been in use, some members may have been away from the home during the day or week and have contributed far less to the contents of the pit than others, and children contribute less than adults to the pit (Foxon 2010).

Table 2-3 Overview of studies that have focused on the measurement of accumulation rate. Estimated Number of Frequency of Average Accumulation Average Location Reference age of latrine latrines accumulation volume of rates range accumulation (yr) monitored measurement pits (m3) (l/cap/yr) rate (l/cap/yr)

N/A 34 India (West (Baskaran 1962) Bengal) 25 India (West (Baskaran 1962, Wagner and Lanoix Bengal) 1958) 40 Philippines (Wagner and Lanoix 1958) 20 Zimbabwe 47 Brazil (Sanches and Wagner 1954) 3 11 14 within 28 1.96 13.1 to 34.0 24.1 months South-Africa 4 159 2 - 3 within 35 3.16 18.3 to 120.5 69.4 months 5 11 1 2.83 10 to 33.2 18.5 11 19 1 3.40 14 to 123 29 (median)

11 25 1 2 14 to 77 34 (median) (Bhagwan et al. 2008, Norris 2000, Still and Foxon 2012b) 9-12 100 N/A 5.40 1 to 109 39 (median) 3 4.2 11 to 146 48 (median) 10 40 N/A N/A N/A 21 (median) 10-14 35 35 2.25 3 to 264 19 N/A N/A N/A N/A 25 to 30 27.5 (implied) Tanzania 0.25-19 50 18 months 2.72 97 to 702 39 Tanzania

*Estimated with an average 6.3 users (Foxon and Still 2012)

Emptying frequency

For economic reasons, emptying services would most likely be claimed only when the pits are full.

Following this assumption, pit emptying frequency data may be used as a proxy to estimate the faecal sludge filling time (yr) into pits. Table 2-4 summarizes the average pit emptying frequencies reported in different studies.

Table 2-4 Estimate of the emptying frequency of pit latrine Emptying frequency Summary of methodology used Location Reference (yr) 0.25 – 1 (53%) Data used includes surveys, field observations and Kampala (Kulabako et al. 1 – 2 (40%) photography. Analysis was done on 30 randomly (Uganda) 2010) chosen households of Bwaise III Parish, a lower- income urban settlement. 5 Data is from a survey conducted with 1,500 lower- Kampala (Günther et al. income households. 45% of latrines are abandoned 2011) after 5 years because of filling up or breaking down. 4.2 (average) Data collected from Kumasi Waste Management, Kumasi (IRC 2006, ≥ 10 (high income includes department reports and the FS treatment (Ghana) Vodounhessi and areas) plant operator’s worksheets. von Münch 2006) ≥ 0.25 (low-income areas) 6 – 10 Data includes interviews and field observations; Ashanti (Appiah-Effah et 270 households were surveyed. region al. 2014) (Ghana) 2 (unlined) Data is from a cross-sectional survey in June 2008 Dar es (Jenkins et al., 6.5 (partially lined) of 662 residential properties across a sample of 35 Salam 2015) 8.5 (fully lined) unplanned low-income sub-wards. (Tanzania) 4.7 (drum/tire) 5.5 (other, mainly septic and sewer)

The assessment of the various studies presented in Table 2-4 highlights valuable information on the factors that can affect the filling time (yr) of pits. First, as observed with the FS accumulation rate (Table 2-3), the frequency of pit emptying varies significantly with observations ranging from pits that had never been emptied (> 10 yr) to some that were emptied monthly. As it was reported in Foxon (2010) for accumulation rates, Kulabako et al. (2010) also noted that the depth of the water table is likely to have the largest impact on the emptying frequency of pits. In the Kulabako study (n = 30), 53% of the respondents live in an area with a high-water table and stated to have emptied their latrine every 3 to 12 months, compared to the 40% of the respondents who live in relatively lower water table and emptied their latrine every 1 to 2 years.

In Kumasi, surveys with 20 households (5 in high income areas, 5 in medium income areas, 10 in low income areas) indicate an estimated average of 4.2 years (ranging from 10 years to only 3 months) for pit to fill up to its holding capacity (IRC 2006, Vodounhessi and von Münch 2006). Interestingly, values found in Kumasi are proportional to the household income, ranging from 10 years or more in high income areas, to as low as 3 months in low income areas (Vodounhessi and von Münch 2006). This might be caused by a higher solid waste content into pit from low-income areas where the pit is more likely to be used as a disposal site for garbage. Also, Günther (2011) conducted on-site surveys with 1,500 low income households in Uganda’s capital, and results show that 45% of the households surveyed stopped using their toilet within 5 years because it filled up (and it was difficult to be emptied), or because it broke down

(Günther et al. 2011).

Furthermore, Jenkins et al. (2015) surveyed latrine emptying services in unplanned communities in Dar es

Salaam, Tanzania. Collected data from a cross-sectional survey of 662 residential properties in 35 unplanned sub-wards show that lined pits, therefore watertight, were reported to fill up slower that in unlined pits. Indeed, lined pits required an average drainage frequency of 8.5 years, as compared to 2 years for unlined pits. This suggests that water infiltration (water from outside moving to inside the pit) would be

more important than liquid leaching (water from inside moving to outside). In the same study, Jenkins and colleagues also reported that the average emptying frequency (expressed in years) decreased with the increase in number of historical emptying; from 14.4 years for latrines emptied once, to 4.8 years for latrines emptied four or more times. This confirms local perceptions that it takes less time to refill a latrine once emptied (Jenkins et al. 2015) probably according to solids trash or sand accumulating at the bottom of the pits which is difficult to remove and structural damage to unlined or partially lined pits during emptying

(Jenkins et al., 2015).

In another study, Appiah-Effah et al. (2014) reported an average emptying rate of one every 6 – 10 years.

Data include interviews and field observations of 270 households in Ghana. However, one limitation is that the majority of surveyed households used shared facilities (69.6%) and may not have been aware of the emptying practices of the communal facilities (Appiah-Effah et al. 2014).

Lastly, the household survey results of a city-scale meta-analysis of on-site sanitation users’ perspectives and habits is presented in Table 2-5. A total of 13,143 household surveys and financial statements of 154 emptying service providers were compiled for 30 different municipalities in 10 countries (Chowdhry and

Kone 2012). The data collected includes the number of households per city with on-site sanitation systems

(PL and STs), and the average size of the pit/tank and the emptying frequency is presented. The accumulation rate assessed for each city varies from 21.4 to 941.5 l/cap/year. In general, the higher accumulation rates correspond to a higher use of STs (Chowdhry and Kone 2012). For example, in Burkina

Faso and Ethiopia, where pits dominate, the accumulation rate varied between 42.1 and 177.1 l/cap/yr as compared to Senegal where the STs were reported to fill up at a rate between 608 to 941.5 l/cap/year. Given the range of pit filling rate field measurements, this estimation from a city-scaled survey shows overall higher accumulation rates. The authors justified the variation on possible inaccuracy of some primary data.

Variations in the pit and tank design, size, intrusion of groundwater, amount of greywater disposal in the

latrines, along with other items like rags and garbage could also explain the large variability of the results

(Chowdhry and Kone 2012).

Table 2-5 The faecal sludge accumulated in on-site sanitation facilities in 30 cities of Africa and Asia (Chowdhry and Kone 2012) Estimated FS No. of people Average Average FS accumulated No. of HH emptying accumulated / using HH accumulation rate volume of pits Location (m3/yr) with OSS frequency HH (m3/yr) latrine (l/cap/year) (m3) (years) 1,247,193 181,066 7 14 492 10 1.5 Abuja 793,239 503,188 2 7 225 12 7.6 Addis Ababa 59,341 93,998 1 15 42 5 7.9 Bobo Dioulasso 2,079,107 167,874 12 14 885 3 0.2 Dakar 49,333 61,996 1 8 100 12 15.1 Dire 4,045 7,680 1 9 59 5 9.5 Fada N'Gourma 10,972 14,955 1 7 105 12 16.4 Hosaena Africa 1,829,663 275,248 7 18 369 12 1.8 Ibadan 691,903 124,735 6 8 693 3 0.5 Kisumu 4,604,702 502,358 9 12 764 3 0.3 Nairobi 439,122 275,208 2 9 177 5 3.1 Ouagadougou 201,514 30,111 7 11 608 3 0.4 Thies 696,960 56,941 12 13 942 3 0.2 Touba 218,022 29,003 8 14 537 10 1.3 Yenagoa 98,806 425,179 0.23 6 39 2 8.6 Delhi 564,689 333,747 1.69 5 338 3 1.8 Dhaka 90,149 24,835 3.63 5 726 3 0.8 Faridpur 166,466 212,231 0.78 4 196 1 1.3 Hai Phong 280,376 404,800 0.69 5 139 1 0.0 Hanoi 894,087 823,785 1.09 5 217 1 0.9 Ho Chi Minh City 126,004 101,714 1.24 7 177 2 1.6 Jaipur 1,013 6,971 0.15 6 24 2 13.8 Kampot Asia 892,051 380,327 2.35 5 469 3 1.3 Khulna 56,142 58,108 0.97 3 322 2 2.1 Kuala Lumpur 48,276 57,084 0.85 4 211 1 1.2 Kuala Terengganu 66,212 35,873 1.85 5 369 2 1.1 Madurai 44,443 46,833 0.95 3 316 1 1.1 Melaka 25,764 197,788 0.13 6 22 3 23.0 Phnom Penh 3,684 30,290 0.12 5 24 2 16.4 Siem Reap

Modelling of pit filling rate

The need to better understand the faecal sludge accumulation rates into pit latrines has motivated the development of a simple mass balance model (Eq. 2.2) The model characterizes the faecal sludge as a mixture of biodegradable material, unbiodegradable material and inorganic material (Brouckaert et al.

2013). Measurements made on 2 pits in eThekwini, South Africa, were used to determine parameters for the model. Then the model was validated using data from 15 other pits in the same area and filling rate data from previous South African studies (Brouckaert et al. 2013) (Eq. 2.2). Parameters of this model are defined in Table 2-6).

é v v e-rt - e-rT ù V(t,T) = Ru (1+ k b0 )(T - t)- ((1- k) b0 ) ê v v r ú ë u0 u0 û (Eq. 2.2)

Table 2-6 Parameters of the Brouckaert et al. (2013) pit filling rate model Parameter Unit Description of parameter

(m3/d) Rate of addition of unbiodegradable material (dry basis) Ru

vb0 (m3/m3) Ratio of biodegradable to unbiodegradable material fed vu0

K (m3/m3) Yield of unbiodegradable organic material from degradation of biodegradable material

r (d-1) Rate constant for biodegradation

In Brouckaert’s model, a PL is considered as an open system, unsealed, that can interact with the environment. The model includes: 1) the inflow of faecal matter into the PL, 2) the degradation process that takes place, 3) the new inert solid material formed, and 4) the outflow. In addition, unbiodegradable material

(e.g. garbage, rocks, glass, etc.) that is added to the PLs is also included in the inflow (Still and Foxon 2012,

Bakare 2014, Murphy 2015). However, the authors report that, given the uncertainties involved, it seems

18 unlikely that the design of a PL would be driven primarily by the factors described by the model, but rather by considerations of resources, cost and the subsequent treatment process (Brouckaert et al. 2013).

Nevertheless, the model may be useful to estimate some of the implications of any chosen system design

(Brouckaert et al. 2013).

Furthermore, Brouckaert’s model was assessed in Infakara, Tanzania where FS build-up in 28 pits was monitored. The performance of Brouckeart’s mass balance model is shown in Figure 2-3, which reported that the simulated filling rate was within 40% of the observed rate for only half of the latrines (12 latrines).

Figure 2-3 Comparison between the observed and simulated filling rates. The 1:1 line is shown and dashed lines correspond to 40% error Todman et al. (2014).

Todman et al. (2014) concluded that Brouckaert’s model could be used as a research tool to predict pits that fill unexpectedly fast or slow, but the model cannot provide an accurate prediction of the filling rate of an individual latrine given the uncertainty of the parameters due to limited data. Results from Todman et al.

(2014) also suggest that the presence of a layer of water at the top of the pit may increase the filling rate and that seasonal changes in groundwater flow into and out of the latrines has an important effect on the

19 filling rate (m/yr). This concludes the available information reported in the literature on the accumulation rates. The next section will discuss the factors affecting the accumulation rates of FS in pits.

2.5 Factors affecting accumulation rates of FS in pits

Many factors influence the filling rate of pits, some of them being operational and design related (i.e. number of users, age of the pits, geophysical and climatic factors, etc.) (section 2.5.1), and others being the result of the degradation processes occurring over time (i.e. aerobic and anaerobic biodegradation) (section

2.5.2). Table 2-7 summarized the studies assessing FS accumulation rates (L/cap/day) or FS biodegradation and characteristics, with factors that influence them (adapted form (Nakagiri et al. 2015).

Operational and design related parameters

+ Number of users

Studies relating sludge accumulation rates to number of users have reported contradictory results with the general perception that the filling rate increases with the number of users. Indeed, Norris (2000), Bakare

(2014) and Still and Foxon (2012b) concluded that the reported number of users does not represent how many people on average throughout the year use the toilet. Hence, the authors suggest that this parameter may not be reliable to describe the filling rate of household pit latrines (5 to 15 users). However, the case could be different in urban settings where pit latrine sharing leads to a higher number of users (Nakagiri et al. 2015).

Table 2-7 Summary of studies assessing FS accumulation rates (L/cap/day) or FS biodegradation and characteristics, with factors that influence them (adapted form (Nakagiri et al. 2015) Variable of Country Study/experimental design Remarks References interest

Per pit filling rate as a function of Number of users The filling rate is not a function of reported number of users, but number of users. South rather depends on a range of other factors not recorded. (Still and Africa Foxon 2012b) Throwing rubbish in a pit is susceptible to increase the filling rate Rubbish content Sorting and analysis of pit content by 50%

Analysis of amalgamated data No correlation (Pearson correlation coefficient of 0.203) between Number of users documented by (Still and Foxon sludge accumulation rate and number of users. 2012b)

South Commercial pit Laboratory experiments on pit The use of commercial pit latrine additives to treat pit latrine sludge (Bakare 2014) Africa latrine additives latrine samples content was unable to accelerate biodegradation rate and mass loss in the test units. Thus, no impact on the accumulation rate.

Laboratory batch experiments on Moisture No evidence that an increase in moisture content of samples from pit latrine samples VIP latrines reduced the sludge accumulation rate.

Flow and (Todman et al. Tanzania Field monitoring and The presence of a layer of water at the top of the pit may increase accumulation of 2014) measurements the filling rate. water

Table 2-8 Summary of studies assessing FS accumulation rates (L/cap/day) or FS biodegradation and characteristics, with factors that influence them (adapted form (Nakagiri et al. 2015) (cont’d)

Number of users Field monitoring and measurements (Norris 2000) No evidence of correlation can be found

21 South Africa No evidence was found to support the effects of Seasonal effects rain and moisture contents in the pit latrines monitored

Addition of The rate of anaerobic digestion of pit latrine moisture contents taken from the surface of the pit could South Laboratory experiments, batch anaerobic digestion on be accelerated by the addition of moisture (Couderc et al. Africa pit latrine samples 2008) Increasing No statistically significant increases in the rate alkalinity of gas production from the samples under anaerobic conditions.

South Anaerobic A serum bottle test was used to investigate anaerobic (Nwaneri Inconclusive results Africa digestion biodegradability of fresh faeces and pit latrine sludge 2009) from the different layers (depth) of a VIP pit South Adding non-degradable material to the pit (Brouckaert et Addition of waste Africa Developing and testing a simple mass balance model significantly influenced its filling al. 2013) South Commercial pit No statistically significant effect on rate of mass (Foxon et al. Laboratory experiments on pit latrine samples Africa latrine additives loss 2009) The longer the emptying period required by on- site sanitation unit, the lower the sludge (Koottatep et Thailand Age of the sludge Field investigations accumulation rate could be obtained. al. 2013)

Table 2-8 Summary of studies assessing FS accumulation rates (L/cap/day) or FS biodegradation and characteristics, with factors that influence them (adapted form (Nakagiri et al. 2015) (cont’d)

An experimental study employing two Operating a septic tank at 40°C had reduced the total

Effect of the laboratory-scale septic tanks fed with diluted volatile solids accumulation and would lengthen the Thailand (Pussayanavin et temperature septage and operating at temperatures of 40 and period of septage removal to twice that of the septic tank al. 2015) 30°C was conducted. operating at 30° C.

23 + Seasonal impact

Studies relating sludge accumulation rates to seasonal effects have reported contrasting results. Some studies reported no influence of the seasonal effect (rainy or dry seasons) on the variations of sludge accumulation rates in PLs in South Africa and Burkina Faso (Norris 2000, Bassan et al. 2013). However, this was not observed in Tanzania, where an important increase in pit content was monitored in the wet periods (rainy season) (Todman et al. 2014).

+ Sludge age

Koottatep (2013) reported that the FS accumulation rate decreases over time due to the stabilization or biodegradation process of organic contents in the sludge (in a septic tanks, ST). The author suggested that

FS accumulation rates for the first year of filling could therefore be highest, given that microorganisms require time to reach solids decomposition activity levels necessary to impact accumulation rates (Koottatep et al. 2014).

+ Addition of other wastes (rubbish)

A simple mass balance model of PL filling developed and tested by Brouckaert, et al. (2013) using data from VIPs in South Africa, predicted that adding non-degradable materials to the pit significantly influenced its filling.

+ Temperature

Impact of temperature was demonstrated by Pussayanavin et al. (2015) in an experimental study using two laboratory-scale STs fed with diluted septage and operating at temperatures of 40°C and 30°C. At steady- state conditions, there was more digestion of organic matter in the sludge layer of the ST operating at the temperature of 40°C, resulting in less total solids accumulation than in the unit operating at 30°C.

24 + Others

Finally, other operational and design related parameters could probably affect the accumulation rate of FS into pits. Parameters such as: the type of soils, diet of the users, the use of chemicals, etc. However, the lack of information in the literature limits their discussion. Further work would be needed aiming to standardize the FS accumulation monitoring into pits and to assess the key factors impacting on the accumulation rate. The next section presents the existing knowledge on the biodegradation mechanisms of faecal sludge into pits.

Biodegradation mechanisms related parameters

+ Anaerobic digestion

Knowledge of the degradation of faecal sludge within latrines is essential in predicting accumulation rates.

Because of the unfavorable redox conditions in pits (e.g. limited oxygen, high COD from the fresh sludge), the assumption of anaerobic biodegradation predominance is largely accepted. Based on faecal sludge characterization data, a theoretical model of biodegradation mechanisms has been developed describing in four stages the profile of FS digestion into pits (Buckley et al. 2008).

Figure 2-4 Four stages theoretical model of FS digestion into pits (Buckley et al. 2008)

Buckley et al.’s (2018) conceptual model (Figure 2-4) has been developed according to the following assumptions. 1) The first layer (i) which is composed of fresh faecal matter - and thus mainly easily biodegradable mass - is likely to decompose quickly under aerobic conditions. 2) The second layer (also aerobic) (ii) is characterized by a thin layer at the top of the pit where the hydrolysis of the substrate’s macromolecules (i.e. lipids, polysaccharides, proteins, etc.) into soluble fermentable compounds under aerobic and/or facultative anaerobic fermentative bacteria takes place. At this stage, the rate of decomposition is suggested to be limited by the hydrolysis of complex organic molecules into simpler compounds. 3) The third layer (iii), the main one, is suggested to be most probably governed by anaerobic biodegradation mechanisms because of the low oxygen concentrations at this depth. 4) Finally, the fourth layer (iv) would consist of inert sludge having reached a high degree of stabilization. Criticism of the

Buckley et al. (2008) theoretical model suggests that given the low numbers of aerobic bacteria in fresh feces it cannot be comfortably assumed that aerobic digestion is solely responsible for the degradation at

26 the surface layer (Torondel 2010); nor is it clear which are the microorganisms present that are accountable for the biodegradation.

Some laboratory batch experiments have been undertaken aimed to improve the understanding of biodegradability of faecal sludge and validate Buckley et al.’s conceptual model (Table 2-7). A study conducted by Nwaneri (2009) aimed to characterize the anaerobic biodegradability of FS samples at different depths (surface, 0.5 m, 1 m, 1.5 m) into pit latrines using an adaptation of the bio-methane potential

(BMP) test (Angelidaki et al. 2009). However, the results of gas production were highly variable; analysis of variance indicated that there was no significant difference (p > 0.05) in daily total gas produced from FS sludge, irrespective of the layer from which it was sampled from (Nwaneri 2009). Consequently, no conclusion could be drown from the BMP tests. Regarding the limitations of the bioassay, it should be noted that only four (4) samples were monitored (surface, 0.5 m, 1 m, 1.5 m), and those were collected in one single pit (which was not in use at the moment of sampling). Also, doubts have been presented about the monitoring devices used; the presence of leaks during the experimental period. Also, the absence of a mixing device into the serum bottle probably reduced the digestion performance, as compared to the suggested BMP protocol found in literature (Angelidaki et al. 2006, Rozzi and Remigi 2004).

Furthermore, Nwaneri (2009) measured the COD profile in a pit (n = 1), to get an idea of the organic carbon consumption. And as shown in Figure 2-5, a COD reduction up to 50% was observed between FF sample and the FS sample on top of the latrines. This indicates that the anaerobic digestion of FS into the pit is not the main biodegradation mechanism as supposed in the 3rd layer, as most of the COD seems to be degraded under aerobic conditions (approximately 50 %).

27 -

Figure 2-5 Evolution of total COD and organics solids (%) as a function of pit latrine depth (n = 16 at the same time), from fresh faeces (adapted from Nwaneri, 2009)

Factors that may explain the suggested inactivity of the anaerobic digestion include poor bioconversion conditions (including pH-alkalinity conditions, temperature and moisture content) and the presence of inhibitory substances (e.g. ammonia, sulfide, light metal ions, heavy metals, and organics) (Bakare 2014,

Foxon et al., 2009, Nwaneri 2009, Bhagwan et al. 2008, Couderc et al., 2008).

In another study, Couderc et al. (2008) measured the effect of additional moisture and/or alkalinity on the rate of anaerobic digestion of faecal sludge. Results reveal that increasing the liquid phase alkalinity through dosing with NaHCO3 had no discernible effect on the rate of anaerobic gas production. However, increasing the moisture content from 76% to 91% of anaerobically incubated pit latrine samples (where the test material was not already well stabilized) increased the rate of anaerobic gas production by between 0.006 and 0.02 mL gas/g total solids/day per 1% increase in moisture content (Couderc et al. 2008). It was thus concluded that increasing the moisture content has the potential to increase the rate of stabilization of buried organic material in the pit (Couderc et al. 2008). However, no evidence was found to link this effect of moisture on the actual solid reduction of the sludge into pits. It is rather proposed that compaction could

28 play a more important role on the volume reduction of sludge into pits than the moisture content (Bakare

2014).

In the light of this results, there seems to be a relative lack of evidence supporting the assumption of efficient anaerobic degradation process into pit latrines. Indeed, factors that account for the decomposition process

(assumed to be mostly anaerobic) aren’t yet conclusively determined and understood. Additionally, some systems are reported to fill up much faster than expected; indicating that the biological breakdown of faeces was not proceeding as would be expected in an anaerobic digester (Bhagwan et al. 2008). Furthermore, the poor conditions for the anaerobic process in pits (see Table 2-9) are suggesting incomplete anaerobic performances. While some authors have looked at the impact of pH-Alkalinity and temperature on the anaerobic degradability of the sludge (Table 2-7), further work would be needed regarding the high ammonia concentrations in pits.

Table 2-9 Operating conditions of the anaerobic digestion process Parameters Anaerobic digestion process Pit conditions

Temperature (°C) Psychrophilic: 0 - 20 15 – 301 Mesophilic: 20 - 45 Thermophilic: 50 - 65 pH 6.5 – 7.51 5.7 - 8.72 Alkalinity (mg/L) 2,000 – 4,000 - Inoculation 10% volume1 - Organic concentration (g COD/L) < 23 20 - 50 N Inhibition (mg TN/L) 1,700 – 14,0004 2,000 – 15,0003 VFA Inhibition (mg/L) > 3,000 2,0003

1Tonrondel et al. (2016), 2Zuma et al. (2015), 3 Strande et al. (2015), 4Chen et al (2008)

Nitrogen concentrations of faecal sludge can be very high (10-100 times the concentration in domestic wastewater) (Strande et al. 2014) (Table 2-8). Indeed, the ammonia concentration exceeds the threshold level that has been reported to inhibit anaerobic activity (Koster and Lettinga 1984): 50% reduction in activity has been observed at total ammonia concentrations over 1.7 g N /L. Strauss et al. (1999) suggest that the influent of an anaerobic lagoon (in tropical areas) should not exceed 400 to 500 mg - N / l to prevent

29 ammonia toxicity. Furthermore, the work carried out by Chaggu (2009) on excreta accumulation systems

(watertight pit) has also suggested that there is an inhibition of methanogenesis by ammonia (N-NH4) when the nitrogen concentrations exceed 2,800 mg N/L. The effect of pH on the distribution of the ammonia species (NH4 – NH3) has also to be considerate. Indeed, at pH levels about 7, the equilibrium is in favor of ammonia (NH3); at levels below pH 7, the ammonia ion is predominant (NH4). The effect of pH at various total ammonia levels (of both NH4 – NH3) was investigated during high-solids sludge digestion (Lay et al.

1998). The pH range was chosen to vary between 6.5 and 9.0, while ammonia concentration range varied from 0 to 6000 mg/l. At all pH levels, 5000 mg TN/l reduced the methanogenic activity by 50%. At pH of

9.0 and NH3 concentration of 900 mg/l, the digester could be operated satisfactorily and the authors concluded that the total ammonia level (NH4 – NH3) rather than the NH3, was a more significant factor in inhibiting methanogenic activity in high-solids digestion (Lay et al. 1998).

Considering this and the high nitrogen concentration of urine entering pit latrines (2,000-15,000 mg - N/L

(Torondel et al. 2016), it is possible that nitrogen concentrations are affecting anaerobic biodegradation mechanisms, resulting in an increased sludge accumulation rate. Further research on the inhibitory effects of ammonia nitrogen into pits is thus needed to quantify the impact of ammonia on the anaerobic biodegradation of faecal sludge.

+ Aerobic biodegradation

As for the study of the aerobic degradation of the FS into pits (first layer (i)), biodegradation batch experiments had been conducted by Bakare (2014). The method used consisted to aerate FS samples (50g diluted with tap water) with saturated air for 5 days (the samples were placed in open flasks). The measure of the biodegradability of the samples was the ratio of COD consumed during the aerated period with the initial COD of the faecal sludge solution. Thus, COD values were obtained before and after the aeration periods. No nutritional solution or inoculum was used in those experiments. Bakare’s results showed a decrease in the biodegradability (COD) depending on the depth of the pits; passing from 52.46% (surface

30 area), to 41.35% (0.5 meter), to 24.08% (1 meter), and 16.55% (1.5 meter) (Bakare 2014). This suggests that the organic charge is decreasing depending on the depth of the pits. The higher biodegradation rate observed between the surface and to 0.5 meter suggested that most of the biodegradation occurred in presence of oxygen. This could suggest that the main biodegradation pathway into pit is aerobic digestion.

Regarding the limitations of this study, the lack of nitrification control (inhibitor) agent must be considered, and this even though the testing period (5 days), which coupled with forced aeration, could have induced the development of nitrifying bacteria and influence the results. In addition, temperature fluctuations (the samples were at room temperature). Also, it would have been interesting to measure the biodegradation rate of fresh faeces, to compare with FS samples.

+ The use of some additives for enhancing the biological breakdown of wastes

The use of some chemical and biological additives for enhancing the biological breakdown of wastes has also been investigated. Chemical additives include strong acids and alkalis (such as sodium carbonate — soda crystals), organic solvents, and ammonia (considered as a Bio-Chemical treatment). They are usually meant to destroy pathogens when mixed with FS. As for biological additives (or bio-additives), they include microorganisms (bacteria) and extracellular enzymes (typically used in septic tanks) or lactic acid. They often aim at reducing the accumulation rates of sludge by enhancing the biological breakdown of wastes.

As shown in Table 2-10, despite the fact that chemical and biological additives are largely used, their performance is severely contested, though some positive impacts have also been reported, mostly during short term studies.

31 Table 2-10 Claims made for pit latrine additives and the reality (Foxon and Still 2012) Claim Reality Products contain microorganisms that can The microorganisms in the pit additive will be insignificant as biologically break down the material in the compared to the number of active microorganisms present in the pit to harmless compost and CO2. pit. Nutrients present in the additive ensure Pit sludge has no nutrient limitation; all nutrients required to sustain optimal growth conditions for microbial life (nitrogen (N), phosphorus (P), potassium (K) etc.) are microorganisms to break down pit present in excess of the growth requirement of microorganisms. contents. Microorganisms work as fast as they can in any given system. Chemicals or biochemical additives There is no chemical or biochemical product that will alter the stimulate the microorganisms in the pit to system, i.e. pit conditions. Consequently, the general pit conditions break down pit sludge faster. are more impactful. Addition of aerobic microorganisms create A system is aerobic or anaerobic depending on how much oxygen aerobic conditions in the pit that result in is present, not on how many oxygen-utilising microorganisms are rapid degradation. present. Accelerated breakdown of pit sludge There is no evidence of accelerated sludge breakdown. However, prevents fly larvae from growing in the pit even if there were, this would not prevent flies from laying eggs in sludge. the top layers where fresh material is constantly being added. Pathogenic bacteria and viruses usually do not survive outside of Addition of non-pathogenic bacteria in the their host (the human) for an extended period, especially under pit sludge out-compete and in fact eat disease- conditions. The major health hazard of pit sludge that has been in causing pathogenic microorganisms in the the ground for an extended period is helminth (worm) eggs. pit sludge, rendering it safe. Sometimes, they survive in PLs for periods exceeding 10 years and are impervious to pit additives. In all the research undertaken as part of the WRC projects, Odours reduce because of accelerated researchers did not notice any reduction in odour, even when sludge breakdown. householders claimed that odours were less.

Bakare (2014), Foxon et al. (2009) and Buckley et al. (2008) showed that many commercial PL additives had no statistically significant effect on the rate of mass loss on PL contents. It is concluded that the addition of additives does not impact the solids accumulation rates (Table 2-10).

This is concluding the section on the factors affecting accumulation rates of FS in pits. The next section will discuss the FS characteristics.

2.6 Faecal sludge characteristics

Faecal sludge is a mixture of human excrements and blackwater, with or without greywater, combined with sand, rocks, and other waste (i.e. anal cleansing materials, menstrual hygiene materials, nappies (diapers), plastics, paper and/or various chemical compounds). It is a raw or partially digested, slurry or semisolid,

32 that are disposed in the pits, tanks or vaults of on-site toilets and sanitation systems; instead of going into centralized wastewater treatment systems (Hines et al. 2017, Tilley et al. 2014).

In the past few years, efforts have focused to improve both quantity and quality of FS characterization data.

However, while the amount of available data increases, the lack of cohesion between studies limits conclusions that can be drawn. There are no standardized methods to determine FS properties and much of the existing knowledge is based on local observations (Strande et al. 2014). As was discussed for faecal sludge accumulation rate data, it is established that various factors can influence the properties of the sludge

(Bassan et al. 2013b, Buckley et al. 2008, Couderc et al. 2008, Schoebitz et al. 2014, Zuma et al. 2015).

These factors include: sampling location (in pits or in trucks), variances of sampling environment (e.g. in rainy or dry seasons), pit characteristics (lined, seal, size, etc.), management systems (e.g. mixed greywater or blackwater), average desludging interval, and other physical factors (e.g. soil, permeability, water table level, etc.). Consequently, comparing the FS characteristic values from each study should be considered as a generalization, since many unknown factors could explain the variability of the results.

In light of these limitations, the next section highlight the differences in characteristics between FS characterization data. More precisely, Table 2-11 presents physical-chemical data according to FS sources, number of samples, the country of the study and the sampling location. Table 2-12 presents data collected from different types of pits and sampling location.

As expected, data reported vary greatly. In the latter table, the standard deviations are almost as high as their average value. The coefficient of variation for all type of tanks and each parameter ranges from 38% to 98%. Bassan (2013b) measured similar standard deviation and average values. This confirms the non- homogeneous results when characterizing FS; data is widespread and does not focus around an average.

33 Table 2-11 Physical and chemical properties of faecal sludge Parameter FS source N samples Study location Sampling References location Pit Household ST Public toilet pH 4.7 – 8.6 10 Durban, South Africa Pit (Zuma et al. 2015) 7.0 – 8.7 5.7 – 8.9 PL (n = 23) Kampala, Uganda Truck (Schoebitz et al. 2016) ST (n = 28) 6.6 – 9.3 N/A Yaoundé, Cameroon Truck (Kengne et al. 2011) 1.5 – 12.6 205 United State ST (United States 5.2 – 9.0 N/A Europe/Canada ST Environmental Protection Agency 1994) Total Solids, 180,000 – 10 Durban, South Africa Pit (Zuma et al. 2015) TS (mg/L) 201,000* 3,515 – 122,581 493 – 66,078 PL (n = 23) Kampala, Uganda Truck (Schoebitz et al. 2016) (33,356 average) (14,912 average) ST (n = 28) 190,000 – 17 Durban, South Africa Pit (Foxon 2010) 710,000* 13,349 (average) 8,984 (average) PL (n = 48) Ouagadougou, Truck (Bassan et al. 2013b) 10,755 (std dev) 8,926 (std dev) ST (n = 18) Burkina Faso 5,020 – 71,007 10 Hanoi, Vietnam Truck (Schoebitz et al. 2014, Strande et al. 2014) 2,202 – 67,200 256 Bangkok, Thailand ST (Koottatep et al. 2005) 2,500 – 124,400 (TSS) 44 Yaoudé, Cameroun Truck (Kengne et al. 2011) 12,000 52,500 N/A Accra, Ghana ST (Kone and Strauss 2004) 4,500 – 14,000 N/A Dakar, Senegal N/A (Vonwiller 2007) (Walker 2008) 6,000 – 35,000 N/A Alcorta, Argentina ST (Kone and Strauss 2004) (SS) 1,132 – 130,475 205 USA ST (United States 200 – 123,860 Europe/Canada ST Environmental Protection Agency 1994) 6,000 – 90,000 N/A ST (Henze et al. 2002) ≤ 3 % ≥ 3.5 % (Heinss et al. 1998a)

* With density of 1,000 ml

34 Table 2-11 Physical and chemical properties of faecal sludge (cont’d) Parameter FS source N samples Study location Sampling References location Pit Household ST Public toilet Total Volatile 23.6 – 82.5 10 Durban, South Africa Pit (Zuma et al. 2015) Solids, TVS. 4,304 – 42,567 826 – 54,919 PL (n = 23) Kampala, Uganda Truck (Schoebitz et al. 2014, (mg/L) (mg/L) ST (n = 28) Schoebitz et al. 2016) 3,421 – 47,440 10 Hanoi, Vietnam Truck (Schoebitz et al. 2014) (mg/L) 59 (as% of TS) 68 (as % of TS) N/A Accra, Ghana ST (Kone and Strauss 2004) 50 (as% of TS) N/A Alcorta Argentina 66.8 – 68.1 (as% N/A Dakar, Senegal N/A (Walker 2008) (Vonwiller of TS) 2007) 73 (as % of TS) 256 Bangkok, Thailand (Koottatep et al. 2005) COD (mg/L) 16,729 – 224,373 10 Durban, South Africa Pit (Zuma et al. 2015) 6,740 – 100,017 742 – 91,850 PL (n = 23) Kampala, Uganda Truck (Schoebitz et al. 2016) ST (n = 28) 46,000 – 199,000 17 Durban, South Africa Pit (Foxon 2010) 12,437 (Average) 7,607 (Average) PL (n = 48) Ouagadougou, Truck (Bassan et al. 2013b) 12,045 (Std dev) 6,718 (Std dev) ST (n = 18) Burkina Faso 7,100 – 15,700 N/A Dakar, Senegal N/A (Vonwiller 2007, Walker 2008) 4,233 – 83,000 10 Hanoi, Vietnam Truck (Schoebitz et al. 2014) 7,480 – 72,500 42 Yaoundé, Cameroun Truck (Kengne et al. 2011) 7,800 49,000 N/A Accra, Ghana ST (Kone and Strauss 2004) 4,200 N/A Alcorta Argentina ≤ 10,000 20,000 – N/A Pits and (Heinss et al. 1998a) 50,000 Tanks 1,500 – 703,000 205 USA ST (United States 1,300 – 1,114,870 Europe/Canada Environmental Protection Agency 1994)

35 Table 2-11 Physical and chemical properties of faecal sludge (cont’d) Parameter FS source N samples Study location Sampling References location Pit Household ST Public toilet

BOD5 (mg/L) 840 – 2,600 7,600 ST (Kone and Strauss 2004) 40 – 78,600 205 USA (United States 700 – 25,000 Europe/Canada Environmental Protection Agency 1994) 5,208 3,860 3 each Accra, Ghana Truck (Jayathilake et al. 2017) 502 ± 36 640 ± 26 15 each Ouagadougou, Truck to be published Burkina Faso 5,198 ± 3199 50 trucks Benin Truck TKN (mg/L) 1,000 3,400 N/A (Katukiza et al. 2012) 9,300 – 74,000* Pits (Zuma et al. 2015) 869 ± 353 50 Benin Trucks (Jayathilake et al. 2017, to be published NH4-N (mg/L) 593 – 2,620 35 – 1,310 Truck (Schoebitz et al. 2016) 80 – 3,300 42 Yaoundé, Cameroun Truck (Kengne et al. 2011) 1,201 1,079 3 each Accra, Ghana Truck (Jayathilake et al. 2017, 29 ± 3 49 ± 4 15 each Ouagadougou, Truck to be published Burkina Faso 150 – 1,200 3,300 N/A (Kone and Strauss 2004) Total 6 – 474 65 – 2,040 (Schoebitz et al. 2016) phosphorus 0.86 1.92 3 each Accra, Ghana Truck (Jayathilake et al. 2017, (mg/L) 44 ± 4 68 ± 6 15 each Ouagadougou, Truck to be published Burkina Faso 490 ± 209 Benin Truck ≤ 1,000 2,000 – 5,000 Accra, Ghana Pits and ST (Heinss et al. 1998a)

36 Table 2-12 Faecal sludge characterization from different type of containments (Koottatep et al. 2013) Type Sample size Total Solids (mg/L) Volatile Solids COD (mg/L) (N) (mg/L)

Man-made concrete ST 10 39,812 ± 36,639 26,033 ± 25,466 32,186 ± 14,212

A single cesspool in clay and loamy soil areas 21 60,934 ± 52,330 35,925 ± 24,126 21,960 ± 10,012

A single cesspool in sandy soil areas 12 119,022 ± 51,645 64,220 ± 17,280 44,939 ± 35,442

bottom sludge bottom - Double cesspool in series 12 34,489 ± 23,880 22,760 ± 15,193 40,420 ± 19,739

Commercial plastic treatment package 10 217,189 ± 107,840 25,245 ± 9,711 38,994 ± 18,644 Thickened Man-made cement ST 10 17,425 ± 23,474 12,273 ± 18,051

A single cesspool in clay and loamy soil areas 21 10,054 ± 5,822 7,199 ± 4,419

Double cesspool in series 12 10,958 ± 8,500 7,206 ± 5,153

Fluidized sludge Fluidized Commercial plastic treatment package 10 189,974 ± 109,143 10,581 ± 10,805

FS from different type of sanitation systems

As reported in Table 2-12, Koottatep (2013) studied FS characteristics from samples collected in different types of containers (i.e. ST, single cesspool (partially lined pit), double cesspool (partially lined pits) in series and commercial STs systems) and from two different sites (i.e. bottom layer of the tank and fluidized sludge from an emptying truck) (Table 2-12). Higher average TS concentrations are reported with commercial systems for both the thickened-bottom sludge samples and the fluidized samples (217,200 mg/L and 189,974, respectively). Koottatep (2013) suggested that the higher TS concentrations in commercial STs are due to the use of toilet paper, which is not used in other systems. Additionally, volatile solids (VS) in the thickened-bottom sludge of commercial plastic units were only 11% of TS whereas those in other types of containment systems were in the range of 60 - 70% of TS (Koottatep et al. 2013). The authors also noted higher COD concentrations from samples collected from cesspools built in locations characterized by sandy soil. In such cases, the average COD was double that of cesspools found in clay and loamy soil areas (44,939 mg/L versus 21,960 mg/L respectively). The authors explained that by the result of the improved percolation of liquids in the sandy soils.

Overall higher concentrations are observed for waterless systems, PLs, as compared to STs. Indeed,

Schoebitz et al. (2016) reported average TS ranging between 493 – 66,078 with STs (N = 40), as compared to average TS ranging from 3,515 – 122,581 mg/L with PLs (N= 36). Higher values were also reported with COD measurement ranging from 6,740 – 100,017 mg/L and 593 – 2,620 mg/L with STs and PLs respectively (Schoebitz et al. 2016). FS from STs are, due to the long storage period, biochemically more stable than FS from installations, which are emptied more regularly on a weekly, monthly or yearly basis

(e.g. public toilet vaults).

38 Effects of sludge age on faecal sludge properties

According to Koottatep et al. (2013), the longer the period between two emptying is, the lower the FS accumulation rate. Higher FS accumulation rates were observed in systems with an emptying frequency of less than one year. Relatively lower FS accumulation rates were observed when the emptying period was between 2 and 6 years. A possible explanation for this is the anaerobic decomposition of organic contents in wastes. These results suggest an appropriate emptying period of at least 2 years to facilitate efficient biodegradation. The USEPA recommendation for emptying frequency is every 3-5 years for septic tanks

(United States Environmental Protection Agency 2002).

Faecal sludge characterization from sampling point

Because pits are not equipped with mixing devices, the location where the sample is collected will greatly influence its physical-chemical properties (Table 2-12). When looking at FS collected from PLs, considering only the sampling point (directly from pit or from collection truck), higher average TS, COD,

TKN and N-NH4 concentration (mg/L) are observed from samples collected directly from pits, as compared to samples from the collection truck. When sampling directly from PLs, Zuma et al. (2015) and Foxon

(2010) reported TS values ranging from 180,000 – 201,000 mg/L and 16,729 – 224,373 mg/L respectively as compared to 3,515 – 122,581 mg/L and 6,740 – 100,017 mg/L from PL sludge collected in an emptier truck (Bassan et al. 2013b, Schoebitz et al. 2016). This may be explained by the fact that some emptying trucks dilute FS prior to pit emptying. The impact of the sampling location seems to be less important with

FS collected from septage.

Zuma et al. (2015) also observed significantly higher values of the moisture, VS, COD, N-NH4, and TKN content in FS samples collected at the top (where sludge age was lower) versus the lower sections of the

PL. Koottatep et al. (2013) demonstrated that the characteristics of the fluidized sludge samples from a FS

39 collection truck (TS: 10,000 – 190,000 mg/L) have lower solids concentrations than those from thickened- bottom sludge (35,000 - 220,000 mg/L).

From rainy season to dry season

Bassan et al. (2013) compared physical and chemical characteristics of raw FS from collection and emptier trucks during rainy and dry season in Ouagadougou (Table 2-13). The results show that there is no statistically significant difference between the dry and the rainy seasons for the parameters measured. This suggests that the season has a low influence on the variability of faecal sludge.

Table 2-13 Differences between dry and rainy season (Bassan et al. 2013b) Parameters FS sampled during dry season FS sampled during rainy season Mean Standard deviation Mean Standard deviation SVI [ml/g TSS] 29 9 26 12 TS [mg/l] 10,658 8,264 12,919 10,989 TVS [% TS] 53 - 61 - TSS [mg/l] 6,826 5,032 11,084 10,406 VSS [% SS] 72 - 60 - COD [mg/l] 9,355 6,538 11,973 11,492 BOD5 [mg/l] 1,839 1,236 1,981 1,454 COD: BOD5 5.3 3.3 6.9 5.0 Source: Bassan et al. 2013

2.7 Conclusion

Overall, this literature review highlighted some of the major gaps in the characterization and quantification of faecal sludge from pit latrines. Table 2-14 summarizes the main issues emerging from the literature review, and the methodology proposed to address the gap.

40 Table 2-14 Gaps in characterization and quantification of FS Justification of the relevance and importance of Methodology to follow to address the Description of the gap addressing the gap gap

Quantification: Faecal 1a. Inability to • Increase the ability to predict pit filling rate per • Modeling the accumulation rate as a sludge (FS) adequately estimate the location characteristics. This will help to set-up function of the key impacting accumulation in pits and FS accumulation rate to guidelines and treatment plants for safe management of factors. factors which influence it extrapolate the quantity FS collected from these systems, e.g. enabling to

generated by a city design treatment systems with sufficient capacity, and to deliver expected quality outputs.

1b. Inability to predict • Factors that account for the decomposition process • Conducting specific methanogenic the performance of (presumed mostly anaerobic) in pit latrines aren’t yet activity tests and toxicity test, to biodegradation conclusively determined and understood. assess decomposition rates into pits. processes occurring in Additionally, some systems are reported to fill up . pits, or how much faster than expected; indicating that the accumulation can be biological breakdown of faeces is not proceeding as minimized. would be expected in an anaerobic digester. Factors that may explain the suggested inactivity of the anaerobic digestion include poor bioconversion conditions (including pH-alkalinity conditions, temperature and moisture content) and the presence of inhibitory substances (e.g. ammonia).

Characterization of 2a. No standard • Currently, standard methods for water, wastewater and • Conduct more extensive faecal sludge in pits and methods for sampling soil are being adapted but they are not necessarily the experiments on characterization of their variation over and analysis of faecal most suitable for faecal sludge, which differs in its FS. Development of new different treatment sludge characteristics and can be highly variable based on the characterization methods and processes local context and its typical heterogeneity. indicators adapted for the FS specific properties and suitable for a resource limited environment.

41 Table 2-14 Gaps in characterization and quantification of FS (cont’d) 2b. The inability to • The development of treatment adapted processes (i.e. anaerobic reactor, • Development of a characterization protocol and understand faecal wetland, drying systems, etc.) requires the knowledge of some validation for specific treatment process and sludge characteristics characteristics of the faecal sludge and the fresh faeces. designing of equipment and their correlation to specific treatment processes.

42 3. OBJECTIVES OF THE THESIS

The global objective of this thesis is to improve the knowledge on characterization and quantification of FS into pit latrines. Specific aims were to:

1) Model the faecal sludge accumulation rates within pit latrines using multiple linear regression (Table 2-13:

1a).

1a) Evaluate the applicability of the multiple regression method to quantify the influence of the pit

volume, number of users, and age of latrines on filling rates of faecal sludge within pits.

1b) Compare the performance of the multiple linear regression model with the existing mass balance

model of Brouckeart and al. (2013).

2) Characterize the impact of ammonia nitrogen on the anaerobic digestion (gas production) of faecal sludge

and faecal matter (Table 2-13: 1b, 2a).

2a) Evaluate the suitability of the specific methanogenic activity test for the characterization of the

influence of ammonia nitrogen on the anaerobic digestion of faecal sludge and faecal matter.

3) Characterize the microbial activity of faecal sludge and faecal matter in terms of microbial populations specific to anaerobic digestion (Table 2-13: 1b, 2a).

3a) Improve understanding of biodegradation mechanisms within pits.

4) Determine drying characteristics of fresh faeces (i.e. isothermal sorption, and evaporation rate) (Table 3-1:

2b).

43 4a) Produce experimental moisture sorption isotherms for fresh faeces using the standard gravimetric

method.

4b) Analyze the sorption isothems using mathematical models from food industries in order to predict

the sorption behavior of faeces.

3.1 Outline of the thesis

The first study (CHAPTER 4) evaluates the possibility of using a multiple linear regression model for the prediction of faecal sludge accumulation rates. This model is in order to address the difficulty of adequately estimating the FS accumulation rate to extrapolate the quantity generated by cities (1a. in Table 2-13). The approach is based on statistical analysis from the data available and discussed in the literature review

(CHAPTER 2). More precisely, the general idea is to verify whether faecal sludge accumulation rates could be adequately estimated using a multiple linear regression model by means of independent variables which are supposed to most impacting the filling rate, namely the volume of the latrine, the number of users, and the age of the latrine.

In CHAPTER 5 characterization of the ammonia inhibition of anaerobic digestion of faecal sludge and fresh faeces samples is presented. This is an important research topic that will address the problem of rapidly filling latrines (1b. and 2a. in Table 2-13). More precisely, it is hypothesized that the high concentrations of ammonia from urine may inhibit anaerobic biodegradation of the faecal sludge in pits, leading to faster sludge accumulation rates. The methodology developed is using specific methanogenic activity (SMA) and toxicity tests, to assess methane production rates from faecal sludge and fresh faeces under different ammonia concentrations.

The results of the SMA tests from CHAPTER 5 were further processed to look at the microbial activity of faecal sludge and fresh faeces (CHAPTER 6). More precisely, the study aims to monitor the microbial activity

44 of faecal sludge and fresh faeces in terms of relative physiological groups of methanogenic archaea in the sludge.

It is believed that a better understanding of the microorganisms in latrine pits can lead to a better understanding of the biological reactions, sludge characteristics, and solids reduction (1b. and 2a. in Table 2-13).

Finally, CHAPTER 7 is addressing the lack of information to understand faecal sludge characteristics and correlation to specific treatment processes (2b in Table 2-13) more specifically the lack of characterizing data on sludge dewatering. Dewatering can be a critical treatment process in faecal sludge management systems.

However, whereas a larger body of knowledge on dewatering characteristics of conventional wastewater sludge, relatively little is known about the solid-to-water bond strength in excreta and faecal sludge. The objective of this study is to conduct a preliminary assessment of the dewatering characteristics (i.e. moisture sorption isotherm, and evaporation rates) of fresh faeces to better inform the design and processes of faecal sludge management systems.

45 4. MULTIPLE LINEAR REGRESSIONS ANALYSYS FOR FS ACCUMULATION RATES PREDICTIONS

4.1 Introduction

Understanding faecal sludge (FS) accumulation rates into pits is essential for the proper sizing and planning of the infrastructure required for collection and transport networks, for discharge sites, treatment plants, and disposal options (Strande et al. 2014). However, it is a challenge to infer FS accumulation rates as there are many variables that need to be considered (see section 2.5). Some estimates have been proposed in the literature

(see section 2.4), however, these values are either independent observations which should be considered generalizations, or the method required to calculate the accumulation rate is labor intensive. Furthermore, a simple mass balance model has been developed using measurements of two pits in eThekwini, South Africa

(Brouckaert et al. 2013) (see section 2.4.4). However, given the uncertainties in the determination of the relevant parameters (i.e. biodegradation rate, addition of unbiodegradable materials, etc.), the authors concluded that an adequate prediction of the accumulation rate could not be driven primarily by the factors described by the model

(Brouckaert et al. 2013). Thereafter, the model was further used with actual filling rates measured in latrines in

Ifakara, Tanzania. The study concluded that the model provides an insufficient prediction of the filling rate of an individual latrine because, 1) there is a complexity of assumed occurring reactions (i.e. hydrolysis, fermentation, methanogenesis, etc.) which could influence the mass balance model, and 2) data availability for the model parameters is limited. Further work is needed to develop methodologies for the prediction of the faecal sludge accumulation rate into pits.

Using the observed FS accumulation rate and the data collected in Todman et al. (2014), multiple regression analysis for the prediction of the FS accumulation rate has been applied in this study. The objective was to assess whether faecal sludge accumulation rates could be estimated by a multiple linear regression model based on

46 independent variables more easily accessible than the values involved in the usual mass balance model, namely the volume of the latrine, the number of users, and the age of the latrine.

4.2 Methodology

Data collection

The multiple regression model was developed based on data reported in Todman et al. (2014), which followed the FS accumulation rate of 28 pit latrines located within rural and urban areas of Tanzania over a period of 18 months. Four of these observations were negative (indicating a decreasing accumulation rate) and were not included in the model; no temporal variations in user behaviour or environmental conditions parameters were included to the model that could explained the negative accumulation rates. The remaining 24 observations are shown in ANNEX 1. The entire database was log transformed because the distribution violated the assumptions of the residuals' normality and homogeneity. Results of the normality test are shown in ANNEX 2. The log transformation was applied to both the independent variables (volume of the pit, number of users and the age of latrine) and the dependent variable (the accumulation rate). This means that a small percentage change in one of the independent variables induces a proportional percentage change in the expected value of the dependent variable, assuming that the effects of the independent variables are multiplicative rather than additive in their original units. Also, log-normal distribution indicates that the averages values of FS accumulation rates for a number of latrines is not representative because of the asymmetric distribution of the data collected.

Furthermore, the predictive regression model was further tested using: 1) FS accumulation rates observed from pits located in different regions of South-Africa (Still and Foxon 2012b), and 2) from FS accumulation rates collected and analysed in Kampala, Uganda (Schoebitz et al. 2016). All data and a brief description of the studies are also given in ANNEX 1.

47 Multilinear regression model

The multiple linear regression model is described below. The regression coefficients (bo, b1, b2, and b3) were estimated by the lm function of RStudio based on the least squares method. RStudio is a free open-source integrated development environment (IDE) for R, a programming language for statistical computing and graphics.

3 Ln (accumulation rate - m /year) = b0 + b1* ln (X1) + b2* ln (X2) + b3*ln (X3)

(Eq. 4.1)

Where:

3 X1 = Volume of the pit (m ) X2 = Number of users (-) X3 = Age of the pit (yr)

4.3 Results and Discussion

Modelling the accumulation rate

Estimations of the model are presented in Table 4-1. Two variables: the volume of the pit and the age of the latrine, are significant on the regression model. However, the number of users is not significant because the

Pr(>|t|) failed the alpha limits of 0.1. This finding is coherent with the findings of the study by Still and Foxon

(2012), where no correlation between filling rates and the reported number of users per household was found.

Thus, the same conclusion may be drown, the authors suggested that the reported number of users is too vague and does not represent how many people on average throughout the year use the toilet, and therefore does not correlate well with the sludge accumulation rate (Still and Foxon 2012b). Consequently the number of users is not included in the model.

Table 4-1 Summary of regression coefficients for models for subsets of the data. Coefficients correspond to those in equation (4.1) Estimate Std. Error t value Pr(>|t|)*

(Intercept) 4.56 0.73 6.22 7.14e-06 *** Volume of the 0.87 0.29 2.94 0.001 ** pit Number of 0.15 0.36 0.40 0.690 users Age of the -0.14 0.08 -1.84 0.083. latrine *Signif. codes : 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Validation of the coefficients

According to the data in Table 4-1, the variable of the number of users has been removed from the equation.

The reduced model is then compared to the full model, using a partial F-test. The partial F-test is performed in

RStudio by comparing the sum of squared errors (SSE) from both the reduced model and the full model, and thereafter comparing the results using the Anova function. Results are shown in Table 4-2.

Reduced model: Yln ~ Vln + Tln (Eq.4.2)

Full model: Yln ~ Vln + Nln + Tln (Eq. 4.3)

Table 4-2 Analysis of variance table Models Res.DF RSS DF Sum of sq F Pr (>F)

Reduced 19 4.24

Full 18 4.20 1 0.038 0.16 0.69

The null hypothesis can’t be rejected since the Pr(>|t|) failed the alpha limits of 0.1 (Table 4-2), therefore there is no significant difference between both models Including the number of users does not improve the model.

Consequently, the accumulation rate model could be reduced (Eq.4.4) as a function of the volume of the pit and the age of the latrine:

ˆ y = 4.56 + (0.87*ln(x1)) -(0.14*ln(x3)) (Eq.4.4)

Interpretation of the model could be as follows: the volume of the pit contributed positively to increase the accumulation rate by a factor of 0.87, while the age of the latrine (time since the last emptying), according to the regression model, contributed negatively to the overall accumulation rate. This suggests that the sludge residence time into pits is inversely proportional to the accumulation rate. This is consistent with literature which suggest that higher age of pits reduces accumulation rates due to the stabilization or biodegradation processes of organic contents in the sludge (see section 2.5). Figure 4-1 depicts the predicted value using the reduced model versus the observations of the accumulation rate.

Figure 4-1Comparison between the observed and simulated filling rates. Solid diagonal represents the perfect fit. Results of the highest accumulation rates observed are circled.

The output of the reduced model shows an R2 of 0.41 and an adjusted R2 of 0.31. This is a slightly better prediction as compared to the performance of Brouckaert’s mass balanced model, which reported that the

50 simulated filling rate was within 40% of the observed rate (Figure 4-2) for only half of the latrines (12 latrines).

The circled dots in Figure 4-1 correspond to the three values of the highest accumulation rates observed by

Todman and al. (2014). Deeper investigation would be interesting to conduct on those pits, on additional parameters that could explain the rapid filling rates. For example, investigating the amount of rubbish present or the distance between the bottom of the pit and the water table.

Figure 4-2 Comparison between the observed and simulated filling rates. The 1:1 line is shown and dashed lines correspond to 40% error. Results for the wettest pits are circled and points a, b and c.

Analysis of model residuals

The analysis of the model residuals is presented below (Figure 4-3). The first graph (top left) shows residuals

(on the y axis) against fitted values (on the x axis). Residuals seem to be uniformly distributed; no tendency between residuals are observed which is necessary to validate the model. However data numbered 9, 13 and 18 are on the limits of the model. These data are associated with ether the highest or the lowest age of the latrine, namely 8, 0.5 and 0.25 age of the latrine respectively– data circled in red in Figure 4-1. The next plot (bottom left) shows the normal QQ-norm plot which shows a relatively good straight line, meaning that the errors are

51 normally distributed. The third plot (top right) is a repetition of the first plot, but on a different scale; it shows the square root of the standardized residuals against the fitted value. Again, no indication of any correlations is observed in that graph. The fourth plot (bottom right) shows standardized residuals as a function of leverage, along with Cook’s distance for each of the observed values that have the biggest effect on the estimated parameters. No residuals have a large Cook's distance which will justified their further examination.

Figure 4-3 Analysis of model residuals

Model validation

The predictive regression model was further tested using: 1) FS accumulation rates observed from pits located in different regions of South-Africa (Still and Foxon 2012b), and 2) from FS accumulation rates collected and analysed in Kampala, Uganda (Schoebitz et al. 2016). All data and a brief description of the studies are also given in ANNEX 1.

52 Figure 4-4 shows predicted accumulation rate (log (L)/year) against the observed values (log (L)/year) for

Todman and al (2014) (red squares), Still and Foxon (2012) (green circles) and Schoebitz et al. (2016) (blue diamonds). The idea was to apply the regression model developed with the data of Todman and colleagues to studies performed in different geographical regions.

Figure 4-4 Plot of predicted value against the observed values of log transferred accumulation rate (log L/yr) for (red squares), (Still and Foxon 2012b) (green circles) and (Schoebitz et al. 2016)(blue diamonds)

For the Still and Foxon (2012b) data (green circles), results seems to fit the regression model quite well. On the other hand, the model seems to systematically overestimate the accumulation rate when applied to the data collected in Schoebitz et al. (2016). One explanation may be that the accumulation rates observed in Schoebitz et al. (2016) came from different origins, including schools, universities, mosques, hotels, restaurants, etc. This might have led to an overestimation of the predicted values since the model was developed with observations made on household pit latrines (lower number of users).

53 Another limitation of the data from the Schoebitz et al. (2016) study is that solid waste (i.e. glass bottles, plastics bags, etc.) could be accumulating on the bottom of the pit, reducing the actual volume of the pit over time.

Indeed, the authors reported that solid waste was often removed from containment prior to the start of the emptying process. In some cases, the solids were buried in a hole on the property, but in other cases, it was placed back into the pit (Schoebitz et al. 2016). This certainly contributes to reducing the volume of pits which wasn’t accounted for. Furthermore, depending on the emptying episode, water inflow or outflow could have influenced the volume of the sludge and consequently the filling rate estimated. These two events can vary significantly based on the structure of the containment, the area (high and low groundwater level) and extreme weather (Schoebitz et al. 2016). Finally, although easier to measure than decomposition kinetics (which is necessary for the Brouckaert model), the volume of pits is not an easy value to be obtained. The information based on the households survey are qualitative values, certainly involves more error than a direct measurement.

Consequently, it is possible to suspect that the values provided are rather estimates instead of precise variables.

4.4 Conclusion

This study aimed at evaluating the prediction of faecal sludge accumulation rates using multiple linear regression. The results obtained refer to a valid model involving pit volume and latrine age values as independent variables. The variable of the number of users was not included in the model, having not respected the limit p- value of 0.1 - Pr (>|t| 0.1). With regard to validation, the value of the coefficient of determination (r2) obtained is r2 = 0.41. This value is slightly higher than the value obtained using the mass balance model developed by

Brouckaert et al. (2013), which reported that the simulated filling rate was within 40% of the observed rate

(Figure 4-2) for only half of the latrines (12 latrines). This reinforces the hypothesis that, given the poor quality of data obtainable from characterized pit latrines, complex modelling of the processes in pit latrines is not justified.

54 Finally, this preliminary study motivates further research that would refine the multiple linear regression model.

For instance, it would be interesting to enlarge the dataset on accumulation rates to be able to obtain a regression model with a better accuracy. Also, further work should include the impact on the accumulation rate of other parameters, such as average temperature, position relative to the water table, annual precipitation, etc.

55 5. THE CHARACTERIZATION OF THE AMMONIA INHIBITION OF ANAEROBIC DIGESTION OF FAECAL SLUDGE SAMPLES

5.1 Introduction

Due to their low-cost, simplicity of construction, and ease in operation and maintenance, pit latrines are the most common form of sanitation in the developing countries (Nakagiri et al. 2015). According to the Joint Monitoring

Program, approximately 1.3 billion people use a pit latrine as their primary mode of excreta disposal, 95 % of which reside in developing countries (Graham and Polizzotto 2013, WHO/UNICEF 2017). However, despite their widespread application and use, performance in terms of solids reduction are unpredictable and can be optimized.

Because of the unfavorable redox conditions into pits, standard practice manuals many times assume that anaerobic degradation occurs within pit latrines (Bhagwan et al. 2008, Buckley et al. 2008, Couderc et al. 2008,

Foxon et al. 2009). However, there is a relative lack of evidence supporting such an important assumption in the design of such systems. Indeed, pit latrines are usually regarded as mere collection systems; factors that account for the decomposition process (assumed to be mostly anaerobic) aren’t yet conclusively determined and understood. Additionally, some systems are reported to fill up much faster than expected; indicating that the biological breakdown of faeces was not proceeding as would be expected in an anaerobic digester (Bhagwan et al. 2008). Furthermore, the poor conditions for the anaerobic process in pits (see Table 2-9) are suggesting incomplete anaerobic performances. While some authors have looked at the impact of pH-Alkalinity and temperature on the anaerobic degradability of the sludge (Table 2-8), further work would be needed regarding the high ammonia concentrations in pits.

In this study, it is hypothesized that the high concentrations of ammonia from urine may inhibit anaerobic biodegradation of the faecal sludge in pits, leading to faster sludge accumulation rates. Previous studies (Sung and Liu 2003) have shown that such type of inhibition can be adequately characterized by adapted specific methanogenic activity (SMA) tests (i.e. batch inhibition assays). However, such an approach has only been applied to sludge from conventional anaerobic digesters and needs to be validated with sludge samples from pit latrines. The objective of this study is to characterize the inhibitory effect of ammonia on both fresh and aged faecal sludge samples, using adapted SMA tests.

5.2 Material and methods

Microbial activity under anaerobic conditions indicates the inherent ability of microbial populations to undertake the degradation of a designed substrate (Rozzi and Remigi 2004). Thus, microbial activity is an indicator to predict the biodegradation rate of an organic compound. Such quantification is normally assessed using the specific methanogen activity test (SMA), which evaluates the anaerobic sludge’s capability to convert an organic substrate into methane and is often expressed as the production of biogas on the specific rate of substrate consumption with reference to total volatile solids (g COD-CH4/g VS). (Dolfing and Bloeman 1985, Valcke and

Verstraete 1983).

Regarding inhibition tests, the main purpose is to assess the degree and the nature of inhibition of the given anaerobic sludge by a specific compound. In other words, indicates a detrimental effect that a test substance has on the activity of a microbial population (Rozzi and Remigi 2004). The experimental technique of assessing microbial inhibition does not differ in principle from the ones designed to measured microbial activity (Rozzi and Remigi 2004). The base activity is measured in a first step at a given concentration of the substrate-acetate

57 (g COD-CH4/g VS). Thereafter, the same concentration of substrate is applied to several vials that are each subjected to different concentrations of the inhibitory substance (ammonia in this case).

The next section describes the adapted method used to characterize the ammonia inhibition of anaerobic digestion of both faecal sludge and fresh faeces samples (simulating the top of a latrine). It includes the sludge sampling, the physical and the chemical analysis for the sludge characterization, the sample preparation and the experimental set-up. The methodology also includes a series of validation tests using active sludge from a conventional digester, to validate the experimental set-up.

Fresh faeces, faecal sludge and anaerobic sludge sampling

Fresh faeces were collected from the top of a decentralized sanitation system with urine separation. The age of the faeces was 24 hours or less. Once collected, the faeces samples were transported in a closed plastic container and characterized for TS and VS using methods 2540G (Total, Fixed, and Volatile Solids in Solid and Semisolid

Samples) from the Standard Methods (APHA 1998). Then faeces were diluted and homogenized with distilled water using a commercial blender (8100 Waring® blender). The quantity of faeces to be diluted was according the VS value of the sludge (aiming 1g VS per serum bottle). The sludge solution was further analysed for COD,

NH3 and TN according to HACH protocols (Table 5-1). The time between sampling and analysis was no more than 3 hours.

Faecal sludge was collected at provincial Parc de la Jacques-Cartier in a conventional pit latrine (no chemical was added to the pit). Approximately one liter of faecal sludge was collected from the back of the pit. The sludge samples were collected at 10 - 15 cm below the surface using a sampling device (Figure 5-1). The sampling period was carried out during the summer period, between May and September. Once collected, samples were transported to the laboratory and stored at 4C until beginning the laboratory analyses (Table 5-1). TS and VS analyses were conducted on the raw sample. Then, as described for the FF, the sludge was diluted with distilled water and

58 homogenized using a commercial blender (8100 Waring® blender). The sludge solution was further analysed for COD, NH3 and TN according to HACH protocols (Table 5-1). Finally, the time between sampling and SMA tests was not more than 2 weeks.

Figure 5-1 Faecal sludge sampling at Parc de la Jacques-Cartier. Sampling device and sampling interface are shown in the left picture, and FS consistency is shown in the right picture.

Table 5-1 Methods used for the physico-chemical characterization of FF and FS samples Methods Reference Parameters

COD Method HACH TNT 8000 (Hach 2013)

NH3 Method HACH TNT 832 (Hach 2013)

TN Method HACH TNT 827 (Hach 2013)

TS and VS Method 2540G (APHA 1998)

pH Electrometric glass probe (APHA 1998)

59 Finally, anaerobic sludge was collected from an anaerobic digester (CSTR) treating domestic wastewater. This was used as an active mature inoculum containing a wide spectrum of microorganisms. Approximately 10L was sampled from the bottom of the digestor. Once collected, the sludge was transported in a closed plastic bucket and characterized for COD according to HACH protocols and for TS, VS using methods 2540G (Total, Fixed, and Volatile Solids in Solid and Semisolid Samples) from the Standard Methods (APHA 1998).

Preparation of the diluted faecal sludge for the SMA test

Faecal sludge and fresh faeces solution samples were mixed with a sodium acetate solution U.S.P. grade from

Sigma-Aldrich. The dilution ratio was calculated from the sample’s volatile suspended solid contents (VS) to achieve an inoculum-substrate ratio (ISR) of 1.3 - 8 g VS/g COD, a recommended ISR for performing SMA in batch tests (Souto et al. 2010). The concentration of the sodium acetate solution was 2.5g COD/L to avoid toxicity; concentrations higher than 3g COD/l can be toxic to methanogenic bacteria (Souto et al. 2010).

Source of nitrogen

Ammonium carbonate, (NH4)2CO3, was used as a nitrogen source, as it has a similar composition to urea

(Procházka et al. 2012). The range of concentrations tested was from 1.5 to 10 g N-(NH4)2CO3/ L (1.5, 3, 5, 7.5 and 10) representative of the range of concentrations likely to be found in pit latrines (Chaggu 2009, Zuma et al. 2015).

Nutrient solution

A nutrient solution was used to maintain optimal pH conditions throughout the experiments. Anaerobic biodegradation produces an increased partial pressure of carbon dioxide, CO2, and a build-up of carbonic acid that could reduce pH of the sludge if there is insufficient buffer solution added. The nutrients medium is prepared from the following stock solutions (10 ml of A and 2 ml of B, Table 5-2), and mixed with 988 ml of distilled water.

Table 5-2 Nutrient medium recipe (adapted from Angelidaki et al. 2006) Component Stock Concentrations in g/L, solution in distilled water NH4Cl A 100 NaCl A 10 MgCl2 6H2O A 10 CaCl2 2H2O A 5 K2HPO4 3H2O B 200

SMA-based anaerobic toxicity assay

The SMA protocol consisted of mixing the sludge solution in 250ml serum bottles with a specific substrate

(acetate), nutrients solution, and different concentrations of the inhibitor (1.5g, 3g, 5g, 7.5g and 10g of N-

(NH4)2CO3)/L). Oxygen was removed from the bottles using a 60-s nitrogen (N2/CO2; 98%/2%; v/v) purge.

During the assays, samples were mixed and the temperature was controlled at 35°C ± 1°C using an incubator.

An Oxytop (monomeric) system was used to measure biogas volume produced, as a proxy for methanogenic activity. Five NaOH pellets were dosed inside a plastic holder with aeration holes just underneath the screw-cap of the Oxytop bottles, so the CO2 can be adsorbed and the pressure reading be assigned to methane. The ratio of the volume of the headspace to the volume of the liquid phase was adjusted in order to measure the maximum theoretical overpressure expected without need to equalize the pressure to the atmosphere (Wei et al. 2011).

Using the ideal gas law and considering the effect of the dissolution of methane in the bottle, the conversion of pressure to the volume (mL) of methane is done using equation 5.1:

æ V ö n = P g + K V CH4 CH4 ç H (CH4 ) l ÷ è RT ø (Eq. 5.1)

nch4 = number of moles of biogas produced at time t Pch4 = pressure measured on the OxiTop reader at time t (Pa) Vg = volume of the headspace in the bottle (m3) Vl = volume of sludge solution (m3)

61 R = perfect gas constant which is 8.314 472 Pa · m3 · K-1 · mol -1 T = temperature of the incubator (K) KH(CH4) = Henry’s constant for methane (Pa)

However, since methane is significantly less soluble than carbon dioxide and does not form other soluble

10 aqueous species (the Henry’s constant for methane (KH(CH4)) at 25°C is 5.96 x 10 Pa), the number of moles that was dissolved into the liquid is not considered in this study. So, the conversion of pressure to the volume (ml) of methane had been done, using the simplified equation:

P ´V CH4 g nCH = 4 R´ T (Eq.5.2)

And since one mole of gas occupies a constant volume at a given temperature and pressure, the volume of methane could be determined for each of the reactors. Finally, the microbial activity is expressed as the production of biogas per biomass of the anaerobic sludge relative to total volatile solids (g CH4 substrate/g VS inoculum).

Validation of the SMA-based anaerobic toxicity assay protocol The suitability of the SMA-based assay was first assessed using constant stirred anaerobic digester (CSTR) sludge (standard inoculum). The sludge was incubated at 35 C during 24h prior to the toxicity test. Table 5-3 shows the initial characterization of the anaerobic sludge. Except for the COD value, the average concentrations are in the same range of what has been reported in (Rodríguez-Méndez et al. 2015), when characterizing the anaerobic sludge from the same source (the same anaerobic digester (CSTR)). The COD value for this study was lower than the value reported in Rodríguez-Méndez et al. (2015). This might be explained by the fact that the characterization procedure for this study was done on sludge samples that were incubated for 24h at 35 C; micro-organisms could consume COD during that period.

62 Table 5-3 Characterization of the anaerobic sludge from the CSTR (n=3) Parameters This study Rodríguez- (n=3) Méndez et al. 2015 (n = 4) COD mg/L 18 400 ± 2310 25 470 ± 1040 TS (g/L) 24.4 ± 1.2 22.5 ± 1.1 VS (g/L) 18.7 ± 0.5 16.3 ± 0.1 pH 7.0 ± 0.1 6.8 ± 0.2 Total nitrogen (TN) (mg N/L) 1690 ± 117 1660± 80

5.3 Results and discussion

Characterization of the FF and FS Table 5-4 summarizes the average characterization values (n = 4) of both faecal sludge and fresh faeces samples.

The units of measurements have been chosen to allow comparison with literature. The average COD for fresh faeces (1395 mg COD/g dry sample) is in the same range of values found in the literature; values ranging from

567 to 1450 mg COD /g dry sample had been reported in different studies (Table 5-5).

COD values of FS from this study (1,140 mg/g dry sample) were lower than those measured by Zuma (2015), who reported a total COD median value of 1,700 g COD/g dry mass for FS samples in South Africa. As discussed in section 2.5, FS can undergo huge variations between different pits and localization.

63 Table 5-4 Physico-chemical parameters of faecal sludge and fresh faeces samples (n=4) Parameter Fresh faeces Faecal sludge

1395 ± 293 1140 ± 157 COD (mg COD/g dry sample) min :1158 min :966 max: 1789 max: 1355 0.20 ± 0.03 0.12 ± 0.03 TS (g/g wet) min :0.18 min: 0.12 max: 0.24 max: 0.15 0.17 ± 0.02 0.09 ± 0.02 VS (g/g wet) min :0.15 min :0.08 max: 019 max: 0.12 6.44 ± 0.4 7.24 ± 0.2 pH min :6.04 min :7.11 max: 6.83 max: 7.47 6.3 ± 0.4 8 .7 ± 4.3 Total nitrogen (TN) (mg N/g wet weight) min: 6 min: 2.9 max: 7 max: 16.1 2.7 ± 0.5 2.4 ± 1.2 Ammonia (N-NH ) (mg N/g wet weight) 3 min: 2.0 min: 0.4 max: 5.0 max: 4.4

Table 5-5 COD of fresh feaces from literature (adapted from (Rose et al. 2015) COD (mg COD/g dry References sample) 1450 Lopez Zavala et al. (2002) 1380 Almeida et al. (1999)

1130 Nwaneri et al. (2008)

567 Chaggu 2009

1448 Buckley 2008

TS, VS and pH values from the FF samples are slightly lower than those suggested as design criteria for on-site sanitation systems (Rose et al. 2015), which mentioned average values of TS, VS and pH of 0.25 g/wet weight,

0.22 g/g wet weight, and 6.6, respectively. However, considering the large number of parameters likely to influence the properties of faecal matter (see section 2.5), variabilities in characteristics were expected and are acceptable. Also, it seems that higher water content is observed in faecal sludge samples collected from the

64 study’s pit latrine as compared to values from Zuma (2015). An average TS concentration of 0.12 g/g wet sample was found for Quebec’s sample in comparison to 0.13– 0.31 g/g wet sample for South Africa samples (Zuma et al. 2015).

Overall, the comparison of the data collected in this study with those of the literature reflects the observed variability in the measured characteristics of faecal matter and faecal sludge. The factors that can influence the characteristics of faecal sludge have been discussed in section 2.5. Furthermore, the fact that there is no standardization between studies for the sampling, the conditioning (i.e. dilution, homogenization, etc.), and for the sludge analysis (i.e. HACH, titration, reflux, etc.) is also likely to have impacted the measured characteristics of the sludge.

Validation of the SMA-based anaerobic toxicity assay protocol

Figure 5-2 shows the methane production curve for: 1) the anaerobic sludge samples with ammonia concentration (ranging from 1.5 to 10 g (NH4)2CO3/L), 2) the anaerobic sludge without the addition of ammonium carbonate (NH4)2CO3, and 3) the control (sludge without acetate - the blank). From Figure 5-2, the inhibitory effect of nitrogen can be seen as lower methane productions correspond with higher nitrogen concentrations. Average SMA results are shown in Table 5-6.

Figure 5-2 SMA tests with anaerobic sludge and different concentrations of total ammonia nitrogen The ISR was 2.49 gVS/gCOD

Table 5-6 Average methane production of anaerobic sludge in response to varied (NH4)2CO3 concentrations (n=3)

Concentration of (g N - (NH4)2CO3)/L SMA ml CH4/g VS 0 63.44 ± 3.54 1.5 64.06 ±15.63 3 61.83 ± 12.72 5 53.13 ± 9.56 7.5 31.91 ± 7.36 10 26.20 ± 6.92

The results showed the highest methanogenic activity at lower nitrogen concentrations (<1.5 g N -

(NH4)2CO3)/L), and an obvious drop with nitrogen concentrations higher than 5 g N - (NH4)2CO3)/L. More precisely, total nitrogen concentrations of 5.0, 7.5 and 10.0 g N - (NH4)2CO3)/L caused methane production to drop by approximately 16%, 50% and 58% respectively. Those results were expected; similarly, Sung and Liu

66 (2003) observed improved methanogenic activity at lower nitrogen concentrations (<1.5 g N-NH4/l), as well as an inhibition of methanogenesis at higher nitrogen concentrations (> 4.0 g N-NH4/l) (Sung and Liu 2003). Also a 50% reduction in activity has been observed at total ammonia concentrations over 1.7 g N-NH4/L .

Characterization of ammonia inhibition on fresh faeces and faecal sludge

As compare to methane production curves using anaerobic sludge, when the SMA-based toxicity assay was applied using FS as an inoculum (Figure 5-3), the addition of a specific substrate (acetate) did not produce the typical SMA curve (as seen with the CSTR sludge). Blank solution (FS solution with no acetate) is also shown in Figure 5-3; cyan blue line. Same results were observed when the SMA-based toxicity assay was applied using

FF as an inoculum, as it is shown in Table 5-7 who summarizes all the tests that have been done, using FS or FF and acetate as a substrate. The duration of the test was based on the strength of the inoculum (when the gas production is reaching a plateau) because different ISR were tested. In most cases the higher the inoculum concentration, the quicker the test was completed. In some occasions, the duration of the test was prolonged to

16 and 29 days, to confirm that no activity will occur after the recommended 5 days of the test (Angelidaki et al. 2009). Overall, no tests produced the expected amount of methane ((ml CH4/g VS), suggesting that acetate is not an appropriate substrate when performing SMA with FF or FS, or that FF and FS have no anaerobic activity. Therefore, the hypothesis that ammonia inhibits the anaerobic digestion of FS in pits could not be confirmed using the SMA-based toxicity protocol.

Figure 5-3 SMA tests with faecal sludge and different concentrations of total ammonia nitrogen (ISR = 2.18)

Table 5-7 Summary of SMA results for both FS and FF – ISR between 2.08 – 35.28

g VS - inoculum g COD - substrate ISR Theoretical CH4 [COD] SMA (ml CH4/g VS) Days (ml CH4/g VS) 0.976 0.38 2.58 152 17.5 ± 1.52 11

1.1 0.20 8.09 80 29.35 ± 2.95 16 FS 1.6 0.30 5.40 120 25.23 ± 3.67 29 1.9 0.42 4.50 148 7.85 ± 0.36 16

0.5 0.13 2.08 51 22.02 ± 0.50 14 1.2 0.14 8.61 49 14.52 ± 1.35 7 FF 1.2 0.28 4.30 113 19.59 ± 2.93 9 3.75 0.63 5.95 252 33.28 ± 0.53 8 5.0 0.14 35.28 49 7.78 ± 0.30 6

68 General Discussion

The main results of experiments are summarized in Table 5-7. There was no change in methane production in response to varied concentrations of substrate or total ammonia concentration. Some hypotheses could explain this absence of microbial activities on the tested samples. First, the standard SMA test is based on the principle that the anaerobic sludge (inoculum) is well mineralized and has a low concentration of readily biodegradable organic matter. Hence, theoretically CH4 production can be estimated by adding known substrate concentrations

(i.e. 1g COD acetate). However, it seems that FS and FF can’t be considered as an exhausted anaerobic inoculum, as the concentration of easily biodegradable material (i.e. carbohydrates, protein, VFA) results in characteristics in-between those of an inoculum and a substrate (VS/TS > 75 %). The higher concentration of COD from the

FF and the FS samples may have result in a lower ISR than with the anaerobic sludge with less easily biodegradable material content.

Also, a recent study has shown that methanogen populations from the human gut are hydrogenotrophic, as compared to predominant acetoclastic microorganisms found in the majority of anaerobic reactors

(Vanderhaeghen et al. 2015). This could explain the results obtained in this study, suggesting that acetate doesn’t seem to be an appropriate substrate when performing SMA with FF or FS. Furthermore, relatively lower temperatures (15-25 C) and high ammonia concentrations (such as encountered in typical pit conditions), had been reported to sometimes lead to a shift from acetoclastic methanogenesis to hydrogenotrophic methanogenesis (Bocher et al. 2015). In other words, pit conditions may not be suitable for growing acetoclastic methanogen archaea, and thus explain the absence of anaerobic activity when performing SMA tests with acetate substrate on FS and FF samples.

Further tests are needed. At first it would be interesting to characterize the potential of methane production

(BMP test) of FF and FS samples. Indeed, the BMP test is used to determine the maximum biogas production of a given substrate (FS and FF), in optimal conditions of anaerobic digestion (e.g. pH, temperature, alkalinity, inhibiting substances, etc.). As compared to the SMA test, the BMP test uses the FS and the FF sample as the

69 substrate. This will therefore give an indication of the efficiency of FS and FF to be converted into biogas during optimal anaerobic digestion. Generally, digested sludge from a wastewater treatment plant is used as an inoculum.

Finally, to understand the degradation pathways of FS and FF in pit latrines, methanogenic pathways should be identified and their contribution to methane production assessed through similar SMA tests with appropriate pathway-specific substrates (i.e. glucose, methanol, and formate). By characterizing the microbial activity of microorganism populations in pit latrines, it should be possible to identify causes of inhibition. This will allow researchers to understand whether there are opportunities to minimize the inhibition of, or indeed optimize the performance of anaerobic degradation in pits.

5.4 Conclusion

The SMA protocol as described in this study failed to characterize the ammonia inhibition of anaerobic digestion of FS and FF samples. The experimental set-up used was appropriate, as the protocol produced expected results when using anaerobic sludge as an inoculum. Indeed, the results showed an improved methanogenic activity at lower nitrogen concentrations (<1.5 g N - (NH4)2CO3/L), and an obvious drop with nitrogen concentrations higher than 5 g N - (NH4)2CO3/L. However, results with FS and FF as inoculum were not as expected; no microbial activity had been recorded on any sample tested. Further tests would be needed to deepen the understanding of anaerobic degradation pathways of FS and FF in pits.

70 6. SPECIFIC METHANOGENIC ACTIVITY TESTS FOR THE CHARACTERIZATION OF MICROBIAL ACTIVITY INTO PIT LATRINE

6.1 Introduction

Results from the previous studies (CHAPITER 5) have raised some questions about the presence of acetoclastic archaea in latrine pits. Poor bioconversion and the fact that acetoclastic methanogens have never been isolated from the human gut (Vanderhaeghen et al. 2015) support this hypothesis. As the performance of an anaerobic digestion system is primarily linked to the structure of the microbial community present, the absence of acetoclastic methanogenesis in the pits would correspond to a poor bioconversion of the sludge, impacting the accumulation rate. The absence of acetoclastic methanogens results in the accumulation of acetate in the reactor and thus bringing the pH to a lower value than the threshold value for methanogenesis. A better understanding of the microorganism activity in latrine pits can lead to a better understanding of the biological reactions, sludge characteristics, and solids reduction.

Microbial ecology of anaerobic digestion

Anaerobic digestion is a complex biological processes. The stability of the reaction depends on a metabolic degradation cascade and requires a diverse population of bacteria/archaea (Schink 1997). Figure 6-1 shows the main anaerobic digestion pathways as well as the principal microorganisms believed to be involved in the anaerobic digestion process.

Figure 6-1 Anaerobic digestion pathway and the microorganisms involved (Cavinato et al. 2017)

Stable anaerobic digestion is accomplished by representatives of four major metabolic groups: 1) hydrolytic fermentative bacteria, 2) proton-reducing acetogenic bacteria, 3) hydrogenotrophic methanogens, and 4) acetoclastic methanogens. The first stage of anaerobic digestion involves the hydrolysis of the substrate’s macromolecules (i.e. lipids, polysaccharides, proteins, etc.) into soluble fermentable compounds by aerobic and/or facultative anaerobic fermentative bacteria. Because the hydrolysis of complex molecules is catalyzed by

72 several extra-cellular enzymes such as cellulases, proteases, and lipases, this step is known as the most activated reaction in anaerobic digestion (Amani et al. 2010).

In the second reaction, acidogenesis, monomers are degraded by acidogenic bacteria. This produces volatile fatty acids (VFA) (i.e. propionic, butyric, etc.), alcohols, organic acids, acetic acid (CH3COOH), hydrogen (H2) and/or carbon dioxide (CO2) depending on the biomass present (Figure 6-1). Then in the third step, intermediate compounds produced by acidogenic bacteria can be oxidized by acetogenic bacteria forming: 1) acetate (by homoacetogenic bacteria), or 2) hydrogen and carbon dioxide (by syntrophic acetogenic bacteria). The last are called syntrophic, because of their mutually beneficial association with methanogens. This syntrophic interaction of acetogens and methanogens is believed to be the most important relationship in anaerobic digestion (Amani et al. 2010). Since the degradation of alcohol and higher fatty acids (i.e. ethanol, butyrate, propionate and LCFAs palminate) under standard conditions are endergonic (Wang et al. 1999) (see Table 6-1), their degradation do not occur under standard conditions as the Gibb free energy is positive and the bacterial energy yield is negative

(Henze 2008). So, for this degradation to occur, there needs to be a low hydrogen concentration. This is achieved by the removal of hydrogen products by methanogenic hydrogenotrophic bacteria (Metcalf and Eddy (2003)).

This step is critical; the imbalance between the growth rates of fast-growing acidogenic bacteria (umax =1 h–1) and slow-growing acetogenic bacteria and acetoclastic methanogens (umax = 0.04 h–1) is known to be the primary cause of anaerobic reactor instability (Wang et al. 2009b).

Table 6-1Stoichiometric equation and change of free energy of some acetogenic reactions ( at NTP – assuming neutral pH and (Henze 2008) Compound Reaction ∆G°(kJ/mol)

73 Methanogenesis is the last (fourth) step. The methanogens are part of the Euryarchaeota kingdom, characterized by a slow metabolism and low exothermic associated reactions (Metcalf and Eddy, 2003). There are three principal groups of methanogens: hydrogenotrophic, acetoclastic, and methylotrophic methanogens. Hydrogen- consuming methane production results in greater energy gains for methanogens than acetate degradation, so they are most likely to happen naturally (von Stockar et al. 2006). However, due to the limited supply of hydrogen in most environments, the main anaerobic digestion pathway found in nature is the one performed by the acetoclastic archaea (70% of CH4) as compared to 30% by hydrogenation (Sylvestre, 2003).

Methanogens from the human gut

Hydrogenotrophic methanogens Methanobrevibacter smithii and Methanosphaera stadtmanae are the main archaeon found in human gut in terms of prevalence and quantity (Chaudhary et al. 2015). No acetocasics methanogens had ever been isolated from the human gut. M. smithii are using hydrogen (or formate) to reduce carbon dioxide and M. stadtmanae are using hydrogen to reduce methanol (Gaci et al. 2014). The hydrogen depletion optimizes the fermentation and modifies the metabolic pathways of fermentative bacteria (Nakamura et al. 2010).

Microbial ecology of pit latrines

Torondel et al. (2016) investigated the diversity and composition of bacterial communities found in pit latrines from two different geographical regions (Tanzania and Vietnam) using in-depth DNA sequence-based characterization. Torondel’s results revealed that the most abundant phylum was the Firmicutes, followed by the Bacteroidetes, Proteobacteria and Actinobacteria in order of decreasing abundance. Suppositions on the metabolic functions of these microorganisms in pits are as follows: the Firmicutes groups are syntrophic bacteria, which can degrade volatile fatty acids such as butyrate. This degradation produces H2, which can then be degraded by hydrogenotrophic methanogens (Riviere et al. 2009). Then, Bacteroidetes are known to intervene in the degradation of proteins and are able to ferment amino acids to acetate (Riviere et al. 2009). Representatives of the Proteobacteria are believed to be microorganisms involved in the first steps of the degradation (Riviere et

74 al. 2009). They also include a wide variety of pathogens, such as Escherichia, Salmonella, Vibrio, Helicobacter,

Yersinia, and many other notable genera (Schröder 1981). Finally, Actinobacteria are starch hydrolyzing bacteria.

The results from Torondel (2016) seem representative of bacteria in human faeces. Previous pyrosequencing studies have revealed that members of Bacteroidetes, Firmicutes (mainly class Clostridia) and Actinobacteria accounted for the majority of bacteria in human faeces. However members of Proteobacteria were rarely detected

(Hashimoto et al. 2014). The phylum Synergistetes was also found to be present in pits - at significantly higher proportional abundances in the Tanzanian pit latrines compared with Vietnam pit latrines (Torondel et al. 2016).

Synergistetes are gram-negative anaerobic bacteria and are known for inhabiting many anaerobic environments, e.g. animal gastrointestinal tracts, human mouth, soil, anaerobic digesters, oil-field waters, and goat rumen

(Jumas-Bilak et al., 2007; Vartoukian et al., 2007). Most of them are mesophiles with thermophiles pertaining only to some genera such as Thermovirga, Thermanaerovibrio, or Anaerobaculum (Militon et al. 2015). They can break amino acids and provide short-chain fatty acids and sulphate for terminal degraders such as the methanogens and sulphate-reducing bacteria (Vartoukian et al., 2007). Consequently, the presence of

Synergistetes in the pit latrine may be associated to amino acid degradation, without any clear demonstration.

Also, certain species from this phylum have been identified to play an important role in the degradation of sludge and food waste for production of biogas in anaerobic digesters (Riviere et al. 2009, Wang et al. 2018).

As for the statistical analysis of Tonrondel’s results, there are clear differences between the proportional abundances of phyla between the two countries (Tanzania and Vietnam) (P < 0.001). It also had been observed that each latrine has a distinct bacterial community composition. However, differences in composition between latrines were far greater than differences within latrines (Torondel and al. 2016). Finally, the microbial communities were from aerobic and anaerobic environments suggesting both degradation processes occurring within the pit latrine (Torondel and al. 2016). However, one major limitation of this study is the absence of

Euryarchaeota characterization. This is important since the Euryarchaeota include methanogens which are, as

75 described in section 6.1.1, the final step of the anaerobic digestion process. However, no information is available on the methanogen species in pits through Torondel’s study and the study does not specify whether analyses were aiming to characterize methanogens in pit latrines, or if no methanogen species was isolated from the pits.

The methanogens characterization in pits would have been important, allowing to compare pits to anaerobic digesters. Indeed, as described in the literature, an anaerobic digester typically contains more than 1012 cells/μl with an average of 108 cells/μl methanogens (Amani et al. 2010). It would have been interesting to see if methanogens in pits were in the same range of magnitude - 108 methanogens.

In another study, (Byrne et al. 2017) studied the microbiological analysis of four (4) pour-flush active and standing leach pits in South Africa over an 11 month period. Molecular microbial analysis of the pit contents

(using Illumina sequencing of the 16S rRNA gene) showed that the main populations in PL pits are:

Bacteroidales, followed by the Firmicutes, Proteobacteria and Synergistetes in order of decreasing abundance

(Byrne and al., 2017). Actinobacteria were not isolated from pour flush pit latrine studied by Byrne (2017) as compared to Torondel’s study (2016). One hypothesis could be the sanitation system user diet differences;

Actinobacteria are starch hydrolyzing bacteria. Regarding the presence of Eurcharchea, evidenced by the archaeal dominance of Methanocorpusculum, a member of the Methanomicrobiales, supporting the hypothesis of heterotrophic bacteria predominance in pits (Byrne et al., 2017). Furthermore, it seems that whether the sample is collected near the front (closest to the input pipe) or the back of the pit does not appear to make an appreciable difference (Byrne et al. 2017). Also, like Torondel et al. (2016) reported, results showed that both aerobic and anaerobic degradation occurred in active pits.

Microbial ecology of untreated wastewater – from conventional sewage

Shanks et al. (2013) study the microbial diversity found in untreated sewage. The sewage profile includes a core human faecal signature made up of several abundant taxonomic groups within the Firmicutes, Bacteroidetes,

Actinobacteria and Proteobacteria phyla (Shanks et al. 2013). Furthermore, Rivière et al. (2009) studied the prokaryotic community of seven anaerobic sludge digesters by constructing and analysing a total of 9890 16S

76 rRNA gene clones. Proportion of the main phylogenetic groups among the different digesters are shown in Table

6-2.

Table 6-2 Proportion of the main phylogenetic groups among different digesters (Riviere et al. 2009)

As shown in Table 6-2 the major bacteria phylum level groups are Chloroflexi, Betaproteobacteria,

Bacteroidetes and Synergistetes. This composition of bacteria is similar to what had been found in pit latrine and in human faeces, except for Chloroflexi. As this last group is abundant in many environments, several studies have attempted to investigate their metabolic function and showed their potential role in the degradation of carbohydrates (Sekiguchi et al., 2001; Kindaichi et al., 2004; Ariesyady et al., 2007). Finally, as for the archaea domain, Rivière’s study demonstrates the predominance of methanogens of the type Methanosarcinales

(acetoclastes) in the reactors.

77 Overall, the studies presented above shed light on the great diversity of microbes present in the different types of sludge (sewage sludge, faecal sludge and faecal matter) and their cross similarity, with the presence of common dominating bacteria groups (i.e. Firmicutes, Bacteroidetes, Actinobacteria, Synergistetes and

Proteobacteria). One major difference is the type of methanogens present in faecal matter and faecal sludges as compared with those found in anaerobic digesters. Indeed, the latter are colonized mostly by acetoclastic methanogens, whereas none of these were found in latrines nor in the faecal matter. Indeed, hydrogenotrophic methanogens are the main archaeon found in the human gut in terms of prevalence and quantity (Chaudhary et al. 2015). This reinforces the hypothesis of the absence of acetoclastic methanogens in the pits.

Furthermore, previous studies have used molecular techniques by identifying the DNA sequence of microorganisms. These techniques require a high level of skill, advanced equipment, and thus restrict its application in the field (Dolfing and Bloeman 1985). Also, one major limitation of 16S rRNA gene sequencing is that it is unable to distinguish between live and dead/inactive microbes, potentially limiting the ability to detect the level of growth and activity of the microbial communities present (Torondel et al. 2016). However, characterizing the amount of active methanogenic population would be essential to understand the anaerobic degradation of faecal sludge in pits. Culture and isolation is one technique that make it possible to determine the microbial activity associated with each digestion step. However, this method has important disadvantages, in particular the difficulty of cultivating microorganisms in anaerobic conditions.

As describe in Chapter 4, microbial activity under anaerobic conditions may also be assessed using the specific methanogen activity test (SMA), which evaluates the anaerobic sludge’s capability to convert an organic substrate into methane. Methods of this type have been successfully used to verify which specific microbial groups (i.e. acidogenic, acetogenic, methanogenic) prevail under a particular condition (Souto et al. 2010).

This chapter presents the application of a methanogenic activity test procedure to monitor the microbial activity of faecal sludge and fresh faeces in terms of relative population levels of methanogenic species by using different

78 test substrates. This will lead to differentiate the various physiological groups of methanogenic archaea in the biomass and the non-methanogenic organisms as far as they act as producers of methanogenic substrates.

6.2 Material and methods

The next section describes the methodology used to characterize the microbial activity of faecal sludge and fresh faeces samples in terms of relevant populations implicated in the anaerobic digestion process. This includes sludge sampling, the physical and the chemical analysis for the sludge characterization, the sample preparation and the experimental set-up. The biomethane potential (BMP) test was also conducted on FS and FF samples to determine their expected maximum biogas production. Biomethane potential tests are measuring the maximum amount of biogas or bio-methane produced per gram of volatile solids (VS) contained in the sludges used as substrates in the anaerobic digestion process. The BMP tests should not to be confused with the SMA tests which are assessing the activity of the microorganism in sludge (as an inoculum).

Fresh faeces and faecal sludge sampling

Samples) from the Standard Methods (APHA 1998). Then faeces were diluted and homogenized with distilled water. The quantity of faeces to be diluted was according previous results of VS (aiming for 1g VS per serum bottles). Homogenization of the sludge solution was made in a commercial blender (8100 Waring® blender), the sludge solution was mixed for 30 sec. The sludge solution was further analysed for COD, NH3 and TN according to HACH protocols (Table 5-1). The time between sampling and analysis was no more than 3 hours.

79 As for the faecal sludge, it was collected at the provincial Parc de la Jacques-Cartier in a conventional pit latrine

(no chemical was added to the pit). Approximately one liter of faecal sludge was collected from the back of the pit.

The sampling period was carried out during the summer period, between May and September. The sludge samples were collected at 10 - 15 cm below the surface using a sampling device (sticks and a bucket). Once collected, samples were transported to the laboratory and stored at 4°C temperature until the start of the laboratory analyses (See Table 5-

1). TS and VS analysis was conducted on raw samples. Then as described for the FF, the sludge was diluted with distilled water. The quantity of faecal sludge to be diluted was according previous results of VS (aiming 1g VS per serum bottles). The sludge solution was further analysed for COD, NH3 and TN according to HACH protocols (Table 5-1). The time between sampling and SMA tests was not be more than 2 weeks.

Choice of substrate for the SMA test

In order to determine the activities of different trophic groups, a variety of substrates was used. Formate

(HCOONa), methanol (CH3OH), and acetic acid (CH3COOHNa) (0.25g DCO/bottles) were used to test for hydrogenotrophic, methylotrophic and acetoclastic methanogenic activity, respectively. For determination of acidogenic activity, 1 g/l glucose (C6H12O6) was used.

SMA-based anaerobic assay

The FF or FS samples were placed in 250 ml ± 0.5 mL serum bottles together with a specific substrate (acetate, formate, glucose, and methanol). The volume of the headspace/volume of the liquid phase ratio was adjusted to

200 ml/50 ml to reach the maximum theoretical overpressure expected without the need to equalize the pressure to atmospheric pressure. For statistical robustness, the assay was performed in triplicates for each substrate added system. The biogas production from the tested substrates was determined by subtracting the biogas production generated by the blanks- sludge solution without substrate. Nutrient and basic anaerobic media were added to each bottle (20ml). The anaerobic media preparation protocol can be found in Angelidaki and Sanders

(2006) (Table 5-2). During the assays, samples were mixed and the temperature was controlled at 35°C ± 1°C.

80 An Oxytop© system was used to measure biogas volumes, as a proxy for methanogenic activity. The test period of the assay was limited to a maximum of 12 days and only the initial linear rate of the biogas accumulation curve was considered for the measure of the production rate (SMA in ml gas / g VS). This was done avoid the interference with potential bacterial growth and adaptation changes during the assay.

Gas analysis

The gas production measurement was conducted by the INRS laboratory (Quebec) using a TRACE 1310 gas chromatograph with flame ionization detector (FID). The volume of gas produced in each Oxytop® batch reactor was obtained by multiplying the headspace volume by the gas percentage (mL/mL) in the headspace as determined by GC analysis. In addition, the calculated value of gas production was converted to STP conditions

(0°C, 1 atm). Finally, the gas production assay was referenced to the sample mass (mL gas/g VS).

Biomethane potential (BMP) test

The BMP tests were conducted based on the most recent recommendations resulting from an international workshop with over 40 participants from 30 laboratories around the world (Holliger et al. 2016). The following procedures were used:

• All tests were carried out in triplicate;

• Blank assays (background methane production from the inoculum) were done;

• The duration of the BMP tests was based on the daily methane production during three consecutive days when <1% of the accumulated volume of methane (i.e. BMP 1%) was observed;

• The BMP results were expressed as the volume of dry methane gas under standard conditions (273.15 K and 101.33 kPa) per mass of volatile solids (VS) added;

• NaOH pellets were used to capture the CO2, so results are expressed in methane production.

81 • The BMP is determined by subtracting the methane production of the blanks from the gross methane production of the substrate.

The BMP test was used with an inoculum taken from a well-functioning digester, a continuously stirred anaerobic reactor treating municipal wastewater. Characteristics of the inoculum are found in Table 5-3. The results of the BMP tests are calculated using the following formula:

2 2 2 BMPsubstrate = BMPaverage,substrate ± (SDblank ) + (SDsubstrate ) control control control (Eq.6.1).

BMP: gas production (ml CH4/g VS)

SD: standard deviation

6.3 Results and discussion

BMP test of FS and FF

Figure 6-2 shows results from BMP tests using anaerobic sludge from an anaerobic reactor (CSTR). The methane produced was 47.31 ± 1.81 ml CH4 /g VS for the FS sample, and 53.10 ± 2.12 ml CH4/g VS for the FF sample. As can be seen on Figure 6-2, it took approximately 450 hours (21 days) before the gas production increments were below 1% per day; indicating that the reaction was near completion. Table 6-2 describes the inoculum and substrate used in the BMP experiments.

Figure 6-2 Cumulative methane production from FS and FF sample as a substrate and CSTR sludge as the inoculum

Table 6-3 Characteristics of inoculum and substrate Component COD VS (g/L) pH Volume (L) ISR (mg/L) Anaerobic sludge 24 750 ± 590 25.60 ± 1.30 7.47 0.025 FS - solution 275g wet sample/L 34 375 ± 950 24.75 ± 0.02 7.11 0.010 1.86 FF- solution 125 g wet sample/L 38 537 ± 985 21.21± 0.30 6.83 0.010 2.33

The BMP values measured in this study are of the same order of magnitude as compared with literature. Methane productions of 50 and 45 ml CH4/g VS had been recorded for VIP toilets and public ablution block toilets respectively (Chris et al. 2014).

Validation: SMA anaerobic sludge from a CSTR

Table 6-3 shows characteristics of the anaerobic sludge used and the expected CH4 production from substrate calculated using the theoretical conversion factor of COD to ml of CH4 (40 mlCH4/gCOD) at 35°C ± 1°C temperature and normal pressure, and the mass (g) of COD of the substrate solution in each bottle. Figure 6-3

83 show the results of the SMA test using anaerobic sludge from a CSTR as an inoculum mixed with three different substrates (acetate, formate and methanol). As expected, the dynamics of the reaction is different for each substrate, providing information on the different microorganisms involved and the kinetics of bacterial growth.

Table 6-4 Characteristics of the anaerobic sludge from the CSTR

Component DCO VS (g/L) pH CH4 expected (mg/L) from substrate (ml CH4/g VS) Anaerobic sludge 26 110 ± 910 24.10 ± 0.11 7.51 49

Figure 6-3 SMA (ml CH4/g VS) for anaerobic sludge

As expected, all three main pathways for methanogens (i.e. hydrogenotrophic, methylotrophic and acetoclastic) are found in the sludge. In typical municipal anaerobic digesters, about 70% of the CH4 is produced from acetate, and the rest from hydrogen (H2) and carbone monoxide (CO) and a minimal amount of CH4 is produced via methylotrophic methanogenesis. Concerning the kinetics, can be seen from Figure 6-3, the anaerobic reaction using acetate (purple line) is characterized by a constant gas production rate of approx. 0.44 ml CH4/g VS per hour (10.56 ml CH4/g VS per day), with a maximum potential production gas of 57.2 ± 5.1 ml CH4/g VS, reached

84 after 110 h (4.5 days). This maximum is slightly higher than the expected theoretical methane production, which was calculated at 49 ml CH4/g VS. However, considering the variability in characteristics of faecal sludge

(section 2.6), this difference is acceptable given the COD variability in each sample. Furthermore, as compared to potential methanogenic activities of sewage digester, the SMA results for acetoclastic methanogenic activity are the same of what has been found in literature, where a rate of 0.44 CH4/gVS per hour had been found on anaerobic sludge digestion with acetate (Valcke and Verstraete 1983). As for the methane production with the methanol substrate (green line), a lag phase of 75 hours (3 days) is observed, followed by an exponential gas production rate with a maximum rate at 0.82 ml CH4/g VS per hour. Then the maximum gas production stagnated at 28.9 ± 3.3 ml CH4 /g VS. Finally, the methane production from formate forms a bell shape; methane production increases at an average rate of 0.25 ml CH4/g VS per hour, reaching a maximum at 27.67 ml CH4/g

VS and then decreases at the same average rate of 0.25 ml CH4/g VS per hour. Table 6-5 shows the cumulative

SMA for each substrate.

Table 6-5 SMA results – anaerobic sludge CSTR

Substrate SMA rate (ml CH4/g VS * hour) SMA ml CH4 /g VS measure

Acetate 0.44 57.2 ± 5.1

Formate 0.25 35.6 ± 2.7

Methanol 0.82 28.9 ± 3.3

As can be deducted from Table 6-5, the maximum gas production with formate and methanol is lower than expected; the theoretical maximum production of methane is 49 ml CH4/g VS (Table 6-5). One explanation is that some of the COD might have been used as source of reducing equivalents for microbial growth. Also, the pH at the end of the experiment with formate was high (8.01 ± 0.18). At this pH, CO2 can dissolve in the sludge solution and could explain the loss in pressure monitored in the Oxytop® bottles. Also, this increase in pH could have led to a decline in activity of the microorganisms. The ideal pH for methanogenic activity is between pH 6 and 8. More buffer solution should be added in the serum bottle for next experiments.

SMA from faecal sludge solution

Figure 6-4 presents the SMA results using a FS solution (275g FS/L) and Table 6-6 reflects characteristics of the sludge used. The use of acetate as a substrate does not result in the production of biogas. This suggests the absence of acetoclastic methanogens in the faecal sludge sampled. Table 6-7 shows the SMA rate and the maximum volume of gas produced with each substrate. Values shown are the average of triplicate repetitions for each substrate. The values for each substrate are calculated by subtracting the maximum potential result from the blank (which is why the SMA results from acetate are negative). This can be explained by the variance of the results. Alternatively, one could deduce a certain inhibition of the anaerobic digestion of the samples by the addition of acetate.

gas/g VS) faecalsludge

SMA(ml

Figure 6-4 SMA (ml gas/g VS) for faecal sludge

86 Table 6-6 Characteristics of the faecal sludge solution (n=3) Component COD VS (mg/l) pH (mg/l) FS solution (275 g FS/L) 157.2 ± 17.6 10.8 ± 1.1 7.05 ± 0.08

Table 6-7 SMA results – faecal sludge solution Substrate Final pH SMA rate (ml gas/g VS * hour) MAX SMA (ml gas /g VS)

Acetate 7.24 -0.001 ± 0.37 -8.88 ± 6.15 Formate 7.91 1.11 ± 0.80 28.47 ± 3.95* Methanol 7.17 0.18 ± 0.12 129.15± 1.29 Glucose 6.90 3.98 ± 2.27 85.42 ± 11.62 *gas sampling for formate was done at t = 25h

Positive microbial activity is observed with glucose and methanol. For the glucose, the microbial activity starts after the substrate is introduced, and the maximum gas production rate reaches 3.98 ml gas/g VS per hour.

However, from the chromatograph analysis, carbon dioxide is the main gas produced from glucose (Table 6-8).

This indicates a glucose fermentation described by the stoichiometry equation:

C6H12O6 + 2 H2O = 4 H2 + 2 C2H4O2 (acetic acid) + 2 CO2 (Eq.6.4)

As for the methanol pathway, gas production is observed after a lag phase of approximately 100 hours (4.16 days), and the average gas production rate is 0.31 ml gas/g VS per hour. Table 6-8 shows the chromatograph analysis. The specific gas produced column corresponds to the total gas measured, minus the reported value from the blank. This facilitates a comparison between the pressure build up measured in the Oxytop® bottle and the amount of gas measured from the chromatograph. As can be seen, the total amount of gas measured from the chromatograph doesn’t match the total amount of gas reported on the Oxytop® bottles. It seems that the

SMA measured from the Oxytop® overestimates the value of the methane and carbon dioxide quantified by the chromatograph. This might be explained by the fact that other gasses (i.e. hydrogen, nitrogen, water vapor,

87 hydrogen sulphide) could have contributed to the pressure in the bottle, but are not analysed by the chromatograph.

Table 6-8 Gas analyses for each substrate – faecal sludge solution

Substrate CH4 CO2 Total gas Methane (%) Specific gas Biogas from the (ml/gVS) (ml/ gVS) (ml/gVS) produce (ml gas/ substrate (ml gas /g gVS) * VS) * Acetate 10.4 35.2 45.6 22.8 0.9 - 8.88 ± 6.15 Formate 42.3 12.1 59.5 71.1 14,8 28.47 ± 3.95 Methanol 88.0 44.9 132.9 66.2 88.2 129.15 ± 1.29 Glucose 18.4 99.5 117.9 15.6 73.2 85.42 ± 11.62 Blank 18.1 26.6 44.7 *Gas from substrate minus gas from the blank

SMA from faeces solution

Figure 6-5 presents the SMA results using FF solution (125 g FF /L). Again, acetate does not lead to the production of biogas. Positive microbial activity is observed only from glucose and formate respectively. As was observed with the FS solution, for the glucose, the microbial activity started immediately at an average rate of 1.82 ml CH4/g VS per hour. The maximum cumulative gas production for glucose reached 76.41 ml gas/g

VS. Furthermore, as was observed with the anaerobic sludge (Figure 6-5), the SMA curve with formate is showing a bell shape; increasing at an average rate of 5.65 ml CH4/g VS per hour, reaching a maximum at

112.77 ± 0.80 ml CH4/g VS and then decreasing at the same average rate of -5.72 ml CH4/g VS per hour. The cumulative SMA results are shown in Table 6-10 (values shown are the average of the triplicate for each substrate)

gas/g VS) faecalsludge

ml SMA SMA (

Figure 6-5 SMA (ml gas/g VS) for fresh faeces

Table 6-9 Characteristics of the fresh faeces solution (n=3) Component COD V.S. (mg/l) pH (mg/l) FF 125 g/L 231.3 ± 12 14.7 ± 1.1 7.00 ± 0.1

Table 6-10 SMA results – fresh faeces solution Substrate Final pH SMA rate (ml biogas/g VS hour) MAX SMA ml gas /g VS measure

Acetate 7.49 0.64 ± 0.36 -5.06 ± 1.30 Formate 8.24 5.65 ± 1.23 112.77 ± 0.80 Methanol 7.26 0.84 ± 0.13 31.10 ± 0.60 Glucose 6.85 1.82 ± 0.71 76.41 ± 11.70

The results from the gas analysis are shown in Table 6-11. Again the specific gas produced column corresponds to the total gas measured, minus the reported value from the blank. This facilitates a comparison between the pressure build up measured into the Oxytop® bottle and the amount of gas measured from the chromatograph.

As observed with the SMA test using the FS solution, the volume of gas measured by the chromatograph does not equal the total volume of gas measured with the Oxytop® bottles. This might have been caused by the fact that hydrogen gas might have contributed to the pressure elevation in the bottle but not measured by the gas analysis.

89 Table 6-11 Gas analyses for each substrate – fresh faeces solution

Substrate CH4 CO2 Total Methane (%) Specific amount of Biogas from the (ml) (ml) biogas gas produce (ml substrate (ml gas/ (ml gas/ gas/ gVS) * gVS) * gVs) Acetate 0.58 3,46 4.04 14.3 -31.08 -5.06 ± 1.30 Formate 34.88 68.21 103.09 33.8 67.97 112.77 ± 0.80 Methanol 0.79 39.6 40.39 1.96 5.27 31.10 ± 0.60 Glucose 0.06 80.48 80.54 0.07 45.42 76.41 ± 11.70 *Gas from substrate-gas from the blank

General Discussion

Table 6-12 summarized all the experiments that have been conducted with FF and FS samples (Note that Table

6-12 does not include chromatograph gas analysis). First, when comparing SMA results from FF and FS samples

(Figure 6-4, 6-5 and Table 6-12), similar kinetics of reaction are found. Also, there is a clear difference in the

behavior based on the nature of the substrate/inoculum combination. Formate and glucose substrate solutions

produce biogas when inoculated with FF, as compared to glucose and methanol when inoculated with FS.

Table 6-12 Compilation of SMA results

g [VS]- g [COD] - CH4 expected SMA rate (ml gas/g VS No ISR MAX SMA inoculum substrate from [COD] min) -0.008 ± 0.020 Acetate -3.04 ± 0.97 1 1.04 0.25 4.16 87 0.056 ± 0.165 MeOH 31.49 ±1.94 1.011 ± 0.168 Formate 23.10 ±2.07 0.064 ±0.12 Acetate -6.99 ± 1.38 0.090 ± 0.38 MeOH 29.44 ±1.87 FS 2 0.97 0.20 4.85 70 3.34 ± 3.47 Glucose 51.59 ±1.39 0.003±0.002 Formate 0.12 ± 0.02 -0.01 ± 0.026 Acetate 4.65 ± 2.74 3 0.78 0.2 3.90 70 0.07 ± 0.01 MeOH 34.97 ± 2.51 0.21 ± 0.17 Formate 27.93 ± 1.19 0.49 ± 0.27Acetate -25.58 ±1.67 0.60 ± 0.04 MeOH 10.72 ± 0.59 4 1.01 0.28 3.6 98 12.7 ± 2.6 Formate 101.84 ± 4.84 FF 9.1 ± 4.3 Glucose 56.52 ± 1.08 0.4 ± 0.3 Acetate 8.82 ± 0.28 5 0.80 0.16 4.4 63 0.30 ± 0.20 MeOH -1.67 ±1.60 2.47 ± 0.26 Formate 40.11±1.79

90 For both FS and FF samples, the addition of glucose led to a quick microbial response, produced by acidogenic bacteria according to gas analysis; the proportion of CO2 from the biogas was measured at 84.4 % and 99.9 % for FS and FF samples respectively. This suggested that acidogenesis can be achieved from the sludge samples tested with glucose. As for the production of methane from formate in FF solution, this is what is expected since the methanogenic ecology from the human gut is characterized by only heterotrophic archaea (Vanderhaeghen et al. 2015).

In both cases for FF and FS, acetate does not seem to be an appropriate substrate for gas production, indicating the absence of acetoclastic archaea in the sludge. This supports previous results from chapter 5 and the study of

Byrne (2016), where there is a dominance of heterotrophic methanogens in pits. This also demonstrated that a pit latrine cannot be compared to anaerobic digesters (on which some design guidelines make assumptions).

According to the results presented, the presence of anaerobic digestion processes inside pit is not guaranteed.

This outcome has important ramifications. For examples, poor bioconversion of FS is associated with malodorous smells - mainly due to butyric acid and paracresol, which elicit disgusting or repulsive response to pit latrine users (Njalam’mano and Chirwa 2018). In negative reaction to these unpleasant odorous, people are deterred from using the pit latrines thereby resorting to open defecation, if private and secluded space availability is not a constraint (Njalam’mano and Chirwa 2018). This leads to public health burdens and polluting the environment.

More experiments would be necessary to deepen the knowledge of the presence (or not) of acetolactic archaea into pits. Future work should include using sludge from different latrines and localizations. The monitoring of the volatile fatty acids of the sludge samples would also be necessary (investigate the acetate build-up into sludge).

More experiments would also be necessary aiming to characterize the methylotrophic activity in pits and the inconsistency of hydrogenotrophic methanogen activity in FS samples. Indeed, SMA tests No. 1 and 3 (from

91 Table 6-12) measured the activity of hydrogenotrophic methanogens (from formate), but test No. 2 registered no activity of hydrogenotrophs. The justification for the variability in hydrogenotrophic methanogen activity in

FS samples is still unknown. To evaluate the hydrogenotrophic die off in faecal matter, it would be interesting to follow the specific microbial activity of fresh faeces over time and under different experimental conditions

(i.e. temperature, pH, nutrients).

6.4 Conclusion

The microbial activity of different bacteria/archaea involved in the anaerobic digestion of faecal sludge and fresh faeces has been characterized in this study. To do so, an adaptation of the SMA with Oxytop devices was used.

The results with FS samples demonstrated that the biogas production was absent with acetate, second-lowest with formate, and highest with methanol. As for FF samples, gas production was absent or very low with acetate and methanol, and formate exhibited higher gas production. For both FS and FF samples, the addition of glucose led to a quick microbial activity response, suggested that acidogenesis can be achieved.

This study demonstrated pit latrines cannot be compared to anaerobic digesters (on which some design guidelines make assumptions) in terms of microbial activity. The presence of anaerobic digestion processes inside pit is not as supposed. This might be caused by the absence of acetoclastic methanogens in the faecal matter, combined with poor environmental conditions that limit the growth of acetoclastic archaea into pits. This instability of the anaerobic digestion certainly accounts for the variability in characteristics of faecal sludge, as well as their rapid accumulation rates in pits. Consequently, using the adapted SMA methodology to improve the knowledge on biodegradations mechanisms from FS in pits, this study provides the basis for better design of latrines. This could lead to bolster the role of latrines from simple containment and partial treatment to engineered systems with predictable efficiencies and improved biodegradation.

92 7. MEASURING MOISTURE SORPTION ISOTHERM OF FRESH FAECES

7.1 Introduction

In low-income countries, the most common forms for excreta disposal are onsite, non-sewered sanitation systems (NSS) (Nakagiri et al. 2015). If safely managed, these sanitation facilities can provide a hygienic and affordable method for excreta disposal, meaning that the faecal sludge is safely contained, collected, treated and disposed. However, in many regions of the world, current faecal sludge management practices and approaches are insufficient and often unsafe. Many municipalities have not yet put the necessary strategies, policies, and budgets in place to maintain (e.g. desludging and treatment) these NSS systems, resulting in the contamination of the environment (Strande et al. 2014) and spread of diarrheal diseases. It is estimated that only 26% of urban and 34% of rural sanitation services worldwide effectively prevent human contact with excreta along the entire sanitation chain and can therefore be considered safely managed (Hutton and Varughese 2016). In addition, inadequately managed sanitation facilities have degraded water quality in most rivers across Africa, Asia and Latin America, directly affecting the quality of life, the working capacity of the inhabitants, the education and the economy (WWAP 2017). Better methods for the safe management of FS from NSS are urgently needed.

As FS consists of over 85% water (Strande et al. 2014), dewatering processes (i.e. drying beds, thermal drying, centrifuges, etc.) for the safe and cost-effective transportation of faecal sludge are advantageous.

These techniques can reduce water content resulting in benefits such as associated transport costs reductions, as well as enhanced pathogen inactivation (Kone et al. 2010), which reduces the health risks of handling the faecal matter. However, studies on dewatering of FS are relatively limited given its importance and they report highly variable results (Gold et al. 2017, Heinss et al. 1998b). The physical water mobility

93 of FS is not fully understood, possibly limiting the progress in the development and optimization of FS dewatering technologies.

Moisture distribution characteristics measurement methods

Whereas there is a large body of knowledge concerning dewatering performances and moisture distribution characteristics of conventional activated sludge (Chen et al. 1997, Colin and Gazbar 1995, Katsiris and

Kouzeli-Katsiri 1987, Lee and Lee 1995, Lee and Hsu 1995, Vaxelaire 2001, Vaxelaire and Cézac 2004), direct transfer of moisture distribution characteristics from wastewater sludge to FS is not possible due to considerable differences in characteristics (e.g. water content, organic content, etc.) (Strande et al. 2014).

Nevertheless, wastewater literature can provide insights into methods that could be applied to the measurement of faecal sludge moisture distribution characteristics. Methods include those based on 1) isothermal drying curves, 2) freezing properties (i.e. the dilatrometric analysis, the differential scanning calorimetry, or the differential thermal analysis (DTA)), 3) mechanical strain tests (i.e. the centrifugal settling test, the filtration test, and the expression test), and 4) moisture sorption isotherms (Vaxelaire and

Cézac 2004).

Vaxelaire and Cézac (2004) present a review of these methods within activated sludge. From their study, the authors show that no technique may be automatically preferred to another (Vaxelaire and Cézac 2004), but each method has specific limitations and should be adequately chosen according the material to be analyzed (Vaxelaire and Cézac 2004). For example, the isothermal drying curves approach does not measure the water distribution as an intrinsic parameter of the sludge. As such, the experimental results depend on the operating conditions (Vaxelaire and Cézac 2004). In comparison, the techniques that are based on the freezing properties of water tend to give a broad spectrum of results which can limit a detailed analysis of the sludge conditioning or dewatering (Vaxelaire and Cézac 2004). According to Chen et al.

(1997), combining freezing methods of Thermal Gravimetry Analysis and Differential Thermal Analysis

94 (TGA/DTA) can adequately characterize the moisture content of activated sludge (Chen et al. 1997).

However, these methods require sophisticated apparatus that may not be available in a limited context.

Regarding techniques based on the mechanical strain tests, the measurement performed under the centrifugal field is not well adapted to extremely compressible and viscoelastic materials such as activated sludge (and is also not suitable for faecal sludge) (Vaxelaire and Cézac 2004). Finally, the moisture sorption isotherm method (MSI), which is largely used in food industries (Al-Muhtaseb et al. 2002), has been successfully used to describe sorption behavior of activated sludge (Vaxelaire 2001). The main limitation of this method is the duration of the measurement (a few days/weeks), which is potentially incompatible with biological materials as their properties change with time (Arlabosse et al. 2003). However, the use of natural anti-bacterial such as Thymol can be used to avoid potential bacterial growth in the samples (Labuza

1984). Also, additionally, the MSI method is a very low-cost and simple technique, based on the relationship between the equilibrium moisture content (EMC) of a product placed at different relative humidity environments (RH). The next section is presenting the MSI principle.

Moisture sorption isotherm

Moisture sorption isotherms are a graphical representation describing the sorption process of water molecules into a specific material (Figure 7-1). It shows where water molecules are progressively and reversibly released from hygroscopic forces in biological material as a result of mainly capillary effects and direct bonding (Caballero-Cerón et al. 2015). More precisely, the capillary effect refers to the ability of the water to flow in the capillary spaces of a material without any external forces applied, even if oppositions to gravity. Figure 7-1 is showing a general sorption isotherms for a hypothetical biological matrix.

Figure 7-1 General shape of sorption isotherms for a hypothetical biological matrix showing the relationship between water content (Xeq) and water activity (aw) (adapted from (Vaxelaire 2001)

The Zone A-B in Figure 7-1 is characterized by water movement to a molecular monolayer on the surface of the material. In this zone, the water molecules are progressively adsorbed to constitute a monolayer covering the entire external surface of the product. It is the result of Van der Waals forces between hydrophilic groups and water molecules (Jannot 2008 ). These forces represents strongly bound water, and consequently the enthalpy of vaporization will be considerably higher than the one of pure water (Jannot

2008 ). The transition to the next zone occurs when the entire surface is saturated. So, zone B-C is described as a multilayer characterized by the adsorption of water molecules on the saturated monolayer resulting in the creation of more layers (Jannot 2008 ). The isotherm in this zone can be represented graphically as a linear function (Labuza 1984). Zone C-D, capillary water, represents higher water activities where it is possible to find water in the liquid form in the material’s capillaries (Jannot 2008). The properties of water in this zone are similar to those of the free water that is held in large capillaries or crevices. Finally, for aw

= 1, the curve tends to an asymptote which corresponds to free water (Vaxelaire 2001). MSI can be modeled to predict the water mobility behavior in an organic matrix.

96 Modelling of the moisture sorption isotherm

Different equations have been used to model the moisture sorption of water in various organic materials

(Al-Muhtaseb et al. 2002, Caballero-Cerón et al. 2015, He et al. 2013, Janjai et al. 2010, Kumar et al. 2014,

Labuza 1984, Orikasa et al. 2010, Purohit and Rao 2017, Trujillo et al. 2003, Zhang et al. 2017). Some models are based on the theory of sorption, and others are either semi-theoretical or purely empirical (Janjai et al. 2010). Theoretical models are based on the multilayer sorption theory of small molecules onto solid surfaces and combine two modes of sorption. The first sorption mode is the formation of a monolayer of sorbate molecules on the surface of the sorbent (Guillard et al. 2013) – which corresponds to zone A-B in

Figure 7-1. This is characterized by the Langmuir model. The second sorption mode is the multilayer condensation of the sorbate onto the sorbate monolayer, whereby the properties approach those of the pure liquid (Guillard et al. 2013) – corresponding to zone B to C in Figure 7-1. The best known example of such multilayer sorption isotherm models is the Brunauer-Emmett-Teller (BET) (Brunauer et al. 1940, Thiele

1953) equation in which the multilayer condensation is combined with the Langmuir sorption model for the formation of the monolayer (Guillard et al. 2013).

In this study, the moisture sorption isotherm of fresh faeces was investigated. Then, from the experimental results, the data was fit to mathematical models in order to predict the sorption behavior of faeces. Finally, the evaporation rate was determined from the equilibrium data. This preliminary work will be a very useful tool for the description of the dewatering processes and to determine the energy requirements of potential further processes.

7.2 Materials and methods

A static gravimetric method was adapted for Labuza (1984) to determine the sorption characteristics of the

FS samples. Two initial masses of FS were tested (1.5 g and 5g) to study the effect of sample size on the

97 water sorption isotherm. Three commonly used mathematical models were then evaluated for best fit with experimental results.

Fresh faeces sampling and initial characterization

Surficial FS was collected from a decentralized sanitation system with urine separation in Quebec City

(QC). The age of the faeces was estimated to be of 24 hours or less. Once collected, the samples were transported in a closed plastic container and characterized for its total solids, volatile solids and water content using methods 2540G (Total, Fixed, and Volatile Solids in Solid and Semisolid Samples) from the

Standard Methods (APHA 1998). The time between sampling and analysis was no more than 3 hours

Determination of sorption isotherms

The standard static gravimetric method was implemented to determine the sorption characteristics of the fresh faeces samples (Labuza 1984). The principle is as follows: a sample of the product is placed in an enclosure maintained at temperature TºC (35ºC±1 ºC) and a relative humidity (HR). The sample is weighed at regular intervals until its mass no longer varies; this is known as the equilibrium moisture content. Then, knowing its wet mass, the equilibrium water content (Xe) can be deduced, and the couple (HR, Xe) provide one point of the isotherm of sorption or desorption.

Figure 7-2 Experimental set-up for the MSI determination

Figure 7-2 illustrates the set-up used for each sample. Two initial masses of fresh faeces were used (1.5 g and 5g) to study the effect of sample size on the water sorption isotherm. Fresh faeces samples were manually homogenized using a stainless-steel spatula. Then the samples were weighed and placed on small aluminum crucibles. The samples were weighed before and after placing in the crucibles. Attention was paid to have samples with approximately the same surface area, forming samples with similar spherical shape of 0.5 mm radius for the sample size of 1.5g and 1cm for the 5g samples. One drop of Thymol, a natural anti-microbial agent, is brushed onto the bottom of each aluminum crucible to prevent the mold growth observed in early trials. Each sample was then placed on an inert PVC plastic tube fixed to a Mason jar which was used as desiccator (Figure 7-2). These jars with the samples were incubated at 35 ± 1 °C until equilibrium of the mass was reached.

Five saturated salt solutions were used to reproduce a range of relative humidity values (RH) ranging from

0.06 to 0.96 (

Table 7-1). Each ACS grade salt was mixed with distilled water (approximately 100 ml) until a solution with excess crystals was formed. The saturated salt solutions were poured into Mason jars to a 1 cm depth

(Figure 7-2).

Table 7-1 Relative humidity (%) of the saturated salt solution at 35°C (CRC, 1977) Salt Relative humidity (%) Potassium Hydroxyde (KOH) 6 Calcium Chloride (CaCl2) 24 Sodium Chloride (NaCl) 75 Ammonium Chloride (NH4Cl) 82

Potassium Sulphate (K2SO4) 97

The weight of the samples was measured each 24 hour until the sample reached its mass equilibration. For the purpose of this experiment, equilibrium in moisture content between a sample and its surrounding environment was defined as when there was less than 2% weight difference over a period of 24 hours.

Calculation of the percentage of mass change over a period of 24 hours was done using Equation 7.1.

DM 24h ´100 Mw- Md (Eq.7.1)

Where:

DM24h = Mass of the sample at time x (g) – mass of the sample at time (x-24h) (g)

Mw = Initial weight of the “wet” sample (g)

Md= Dried mass (i.e. total solids) of the sample (g)

When the mass equilibrium of the sample was reached (i.e. a difference in weight of less than 2% after

24h), equation 7.2 was used to measure the final moisture content of the sample (Labuza 1984).

é%H O ù (W -W ) + 2 ×W f i ê 100 i ú m= ë û é100 - (%H O)ù W × 2 i ê ú ë 100 û (Eq. 7.2)

100 M = final moisture content g H2O/g dry Wf = final weight of a sample at equilibrium Wi = initial weight of the same sample

% H20 = wet basis % moisture content of the wet sample

Mathematical models of the sorption isotherm

Table 7-2 lists the mathematical isotherm models used in this study for comparison to the empirical data.

Models were selected based on their effectiveness for describing isotherms of high moisture content of biological materials (e.g. municipal sludge, food, plants). Nonlinear data analysis and curve fitting were executed in Rstudio.Version (2017)(RCoreTeam 2017). The quality of each model was assessed with regard to the standard error of estimate, percent root mean squared errors (% RMSE), and the maximum adjusted

R-square. Also, Student test was applied on regression coefficient for significance (Pr (>|t| > 0.05).

Table 7-2 Isotherm models used for fitting experimental data Isotherms Equation Parameters

Brunauer-Emmett-Teller (BET) M Ca M = Moisture content (MC) M = 0 w (1- a )(1+ Ca - a ) C = constant w w w Mo = Monolayer MC M CK 'a M = Moisture content (MC) M = 0 w Guggenheim-Anderson-de Boer (GAB) (1- K 'a )(1- K 'a + Ca K ') C, K’ = constants w w w Mo = Monolayer MC Flory-Huggins M = Aexp(Baw) A, B = constants

7.3 Results

Fresh faeces characterization

Table 7-3 shows the average value of faeces characteristics measured in this study. These characterization values are lower than those suggested as design criteria for on-site sanitation systems (Rose et al. 2015), which reported average values of TS,VS and pH of 32 and 22 mg/g wet weight, and 6.6, respectively.

However, considering the large number of parameters likely to influence the properties of the faecal matter

101 (e.g. the dietary intake of food, metabolism, gut microflora, etc.), variabilities in characteristics were expected and deemed acceptable.

Table 7-3 Characteristics of fresh faeces (n=6) Parameters Fresh Faeces TS (mg/g wet weight) 21.2 ± 0.17 VS (mg/g wet weight) 17.0 ± 0.02 VS/TS (%) 81.1 ± 0.03 pH 6.4 ± 0.40

Moisture sorption isotherm from two different sample masses

Figure 7-3 a) 1.5g sample and b) 5g sample show sorption isotherm results for 5 g samples and for 1.5 g samples - both at 35°C±1°C. The boxplots are depicting the data in quartiles that allow quick graphical comparisons between the different %RH results. The bottom and top of the box are the first and third quartile, and the band inside the box is always the second quartile (the median). Ends of the whiskers are the min and max of all the data.

Figure 7-3 Boxplot of equilibrium moisture content data for each given relative humidity at a temperature of 35˚C and an initial faecal sample with mass a) 1.5 g and b) 5g

As can be seen in both boxplots, overall trends indicate achievement of sigma-shaped moisture sorption isotherms characterized by two characteristic bends (Figure 7-1). The first bend occurs at an aw of 0.05 -

102 0.2 and the second one at around 0.75 – 0.8. According to Labuza (1984) this is caused by the additive effects of Raoult’s law, capillary effects, and surface H2O interaction. More precisely, as referred to the standard isotherm shape (Figure 7-1), between 0 and 6% RH is suggested to result from the Van der Waals forces on water molecules and constitutes a molecular monolayer on the surface of the product (Jannot

2008) (Zone A-B). Then, between 6% and 75%, which corresponds to the multilayer zone, water molecules are adsorbed on the saturated monolayer resulting in the creation of more layers (zone B-C) (Jannot 2008).

In this region, water is held in the solid matrix by capillary condensation. This water is thus available as a solvent for low-molecular weight solutes and for some biochemical reactions (Al-Muhtaseb et al. 2002).

Finally, between 75 and 97%, water is suggested to be bond due to macro-capillary forces or as part of the fluid phase in high moisture materials (Zone C-D). This shows early all the properties of bulk water, and thus can act as a solvent (Al-Muhtaseb et al. 2002). Microbial growth becomes a major deteriorative reaction at this zone (Al-Muhtaseb et al. 2002). Finally, for aw = 1, the curve tends to an asymptote which corresponds to free water (Vaxelaire 2001).

Furthermore, this sigma–shape that all fresh faeces isotherms seem to correspond to, have been reported for most food product with high moisture content (aw over 0.9) (Labuza 1984). This is as expected since the composition of faeces will be similar to most food in terms of molecules present, namely carbohydrates, lipids, protein. Finally, the effect of particle sizes and water activity on moisture content conforms to what was expected; lower moisture content with decreasing sample size, as it can be seen when comparing Figure

7-3a and 7-3b.

Modelling of isotherms

7.3.3.1 Initial conditions

The nonlinear isotherm equation presented in Table 7-2 was used to fit the experimental data presented in

Figure 7-3a and Figure 7-3b. For the GAB isotherm model, which includes three variables, determination

103 of initial conditions was necessary. This was possible by linearization of the BET equation where the monolayer (Mo) mass can be estimated. More precisely, from BET equation (Eq.7.3):

M Ca M = 0 w (1- a )(1+ Ca - a ) w w w (Eq.7.3)

The result of the linearization of equation (7.3) is:

a 1 C -1 w = + a a M M c M C w (1- w ) o o (Eq.7.4)

(Eq.7.5)

Equation 7.5 shows a linear equation of the form Y = I + Saw, where I represents the intercept, and S

풂풘 represents the slope. Using linear regression on as a function of aw, both constants C and Mo can (ퟏ−퐚퐰)푴 be estimated from the BET equation.

104

풂풘 Figure 7-4 Estimation of constants C and Mo from BET equation - linear regression on as a (ퟏ−퐚퐰)푴 function of aw Residual standard error: 0.8927 on 20 degrees of freedom, Multiple R-squared: 0.7051, Adjusted R-squared: 0.6904, F-statistic: 47.83 on 1 and 20 DF, p-value: 1.023e-06

Table 7-4 Linear regression results for linearized BET model

Coefficient Estimated Std. Error T value Sig

Intercept -0.1873 0.3246 -0.577 0.57.

Slope 3.4260 0.4954 6.916 1.02e-06 ***

Figure 7-4 and Table 7-4 represent the linearization of equation (Eq. 7-3) where Mo and the C constant of

BET model can be determined using Equation 7.5

C -1 1 S= I = M C M C o o

C -1 1 M o = = SC IC

105 C -1 1 M o = = 3.4260C -0.1873C

C = -17.29 1 M o = = 0.3088 -0,1878 + 3.4260

7.3.3.2 Fitting experimental data

Figure 7-4a and 7-4b depict experimental data fit with and BET, GAB, and FH models at 35°C±1°C and for both 1.5 g and 5.0 g of initial sample mass. Table 7-5 summarises the least squares analysis to determine the best data fit for the model. For both sample sizes, the FH model was rejected since model parameter A

(i.e. constant) failed the Student test for significance (Pr (>|t| > 0.05). The GAB and the BET models are both valid, since the equation parameters (C, Mo, K’) were satisfactory (Pr (>|t|) = 0 to 0.05). For BET, Mo is associated to the moisture content (dry basis) corresponding to an adsorbed monolayer (BET) and C and

K’ are constants related to the temperature effect. Further testing should be performed at across the range of humidity values to improve the calibration of the models and better assessment of the most suitable one for FS.

Figure 7-5a -1.5g and 7-5b-5g faeces as initial mass. Experimental data (.)(Average) and predicted values; BET (___), GAB (---), FH (…) model predicted adsorption at 34C.

106 Table 7-5 Predicted value analysis for selected model for 1.5g and 5.0 g of faeces as initial mass

Sample Standard Model Parameters Value t-value Pr (>|t|) Mass error M0 0.309 - - - 1.5 g C −24.800 23 690 −1.047 0.308 BET M0 0.309 - - - 5.0 g C 19.904 5 711 −3.485 0.002 ** M0 0.309 - - - C −4.997 1.658 −3.013 0.00746 ** 1.5 g < 2 × 10−16 K’ 0.973 0.007 142.140 *** GAB M0 0.309 - - - C −4.519 1.497 −3.017 0.00747 ** 5.0 g < 2 × 10−16 K’ 0.985 0.006 157.107 *** A 0.144 0.091 1.582 0.131 1.5 g 4.17 × 10−5 B 3.739 0.696 5.373 FH *** A 0.403 0.234 1.725 0.0986 5.0 g B 2.942 0.639 4.602 000139 ***

Finally, Figure 7- 6shows the evolution of the desiccation as a function of time. The mas of removed water was calculated using equation 7-6

W -W = t-1 t V (Eq. 7-6)

Where

Wt-1 is the mass of sample at time t-1 W is the mass of sample at time t (24 hours). V is the volume of sample (cm3).

107

) )

O/cm

2 Mass of of Mass removed(gH water

Time (day)

Figure 7-6 Desiccation rate for 5g sample at 35ºC and at different %RH (Black (6%RH), green (29%RH), blue (75%RH), purple (89%RH), yellow (96%RH)

As can be seen in Figure 7-6, the highest amount of water could be removed from the sample with the lowest relative humidity. Equilibrium moisture content (EMC) was obtained after approximately 6, 12 and

14 days for sample at 6% 24%-75% and 89-97% of RH.

7.4 Conclusion

The preliminary assessment of the drying characteristics (i.e. moisture sorption isotherm and desiccation rates) of fresh faeces was investigated at different relative humidity (6% - 97% RH) at 35°C±1°C using a static gravimetric method.

Overall moisture sorption isotherms trends indicated achievement of a sigma-shape characterized by two characteristic bends (caused by the additive effects of Raoult’s law, capillary effects, and surface H2O interaction) and an increase of moisture adsorption at higher values of relative humidity. This means that, in this study, water in fresh faeces samples seems to be more easily extracted until a written activity of

108 approximately 75%. Experimental data seemed to follow the three-parameter Guggenheim–Anderson–de

Boer (GAB) model with a reasonable fit (Pr (>|t| < 0.05)). However, further tests should be realized at lower relative humidity values (i.e. 6%, 10% 15%, 20%) to improve the calibration of the model.

As expected, the moisture isotherm at 35ºC±1 ºC resulted in a lower final water content, as higher temperatures corresponded to higher desiccation rates. At low moisture contents the heat of sorption values was higher, meaning that more energy was needed to dry fresh faeces. With regards to desiccation rates, results confirmed that at lower relative humidity and higher temperatures, the highest amount of water could be removed per unit of time.

These results add to the very limited data on the characterization of fresh faeces and represents one of the few datasets on desiccation properties for this type of matrix. Future work will include the effect of ventilation rates. This could serve as a basis for the design of on-site sanitation systems for in situ desiccation of faeces intended for solids reduction and pathogen inactivation.

109 8. DISCUSSION GÉNÉRALE

La prochaine section fait une synthèse des principaux points de discussion qui ressortent des diffèrents chapitres présentés dans cette thèse.

8.1 Modéliser les taux d’accumulation des boues fécales au sein des latrines à fosses à l’aide d’une régression linéaire multiple

Les résultats obtenus ont permis de tirer des conclusions intéressantes. Premièrement, il a été démontré que la répartition des données (à savoir : le volume des fosses, le nombre d’usagers, le temps depuis la dernière vidange et les taux d’accumulation) correspondait à une distribution log-normale; ceci indiquant une répartition des données asymétrique, caractérisé par une grande proportion de valeurs qui soit relativement faible (moyenne des données faible), et une petite répartition de valeurs plus élevées (indiquant une variance

élevée). En d’autres mots, cela voudrait dire que la valeur moyenne des données est peu représentative à cause de la distribution asymétrique (distribution log-normale). Par la suite, mentionnons que le modèle de régression semblerait plus efficace que le modèle par bilan de masse développé par Brouckaert et al. (2013).

En effet, la valeur du coefficient de détermination (r2) était de r2 = 0.41 (r2 ajusté de 0.31) pour le modèle de régression linéaire, comparativement à la moitié des latrines étudiées (12 latrines) par le modèle de

Brouckaert et al. (2013) estimées à plus de 40% d’erreur. De plus, un avantage du modèle de régression est l’emploi de variables qui soient plus facilement quantifiables (i.e. volumes des fosses, temps depuis la dernière vidange) que les paramètres impliqués dans le modèle de Brouckaert et al. (2013) (i.e. taux d'addition de matière non-biodégradable, rapport entre la matière biodégradable et la matière non- biodégradable, rendement de dégradation de la matière biodégradable, constante de biodégradation).

Limites de l’étude et recommandations

Plusieurs suppositions sont à considérer sur les données qui ont été utilisées dans l’élaboration du modèle de régression multiple. Par exemple, il est souvent observé qu’une proportion de déchets solides (e.g.

110 bouteilles de verre, sacs plastique etc.) n’est pas évacuée lors de la vidange des fosses, pouvant ainsi s’accumuler au fond des latrines au cours du temps. Parfois les déchets seront brulés sur place, parfois ils seront remis dans les fosses une fois les boues fécales vidangées (Schoebitz et al. 2016). Il serait donc nécessaire de quantifier cette proportion de déchets solides non évacuée, pour ainsi éviter les risques d’une surestimation des valeurs de taux d’accumulation avec l’usage des latrines. Une autre limitation importante est la quantité des données utilisées pour élaborer le modèle. En effet, comme discuté dans la revue de littérature, le suivi des taux d’accumulation n’a pas fait l’objet de plusieurs études, et le manque de standardisation entre les études limite les comparaisons, ainsi que la taille du jeu de données disponible pour analyses. Finalement, la qualité des données doit être considérée. Par exemple l’information relative

à l’âge des latrines est recueillie par sondage auprès des usagers, et par conséquent doit être considérée comme une estimation plutôt qu’une valeur mesurée.

Néanmoins, malgré ces limitations, l’analyse des valeurs obtenues par régression linéaire multiple motive la poursuite des travaux en ce sens. Il serait ainsi intéressant de poursuivre l’analyse en augmentant le nombre de données pour l’élaboration du modèle, en contactant les chercheurs du domaine et en collectant les valeurs non publiées. De plus, il conviendrait d’élargir l’étendue des paramètres du modèle en y incluant, par exemple : le type de sol dans laquelle la fosse est construite, la distance avec la nappe phréatique, etc.

8.2 Caractériser l’impact de l’azote ammoniacal sur les performances de digestion anaérobie des boues fécales et de la matière fécale

La méthodologie développée a permis de valider le protocole expérimental en utilisant des boues issues d’un digesteur anaérobie en opération, et en les mélangeant avec différentes concentrations d’azote (1.5,

3.0, 5.0, 7.5, 10.0 g(NH4)2CO3/L). Conformément à ce qui était attendu, les résultats ont démontré une légère augmentation de la production de méthane (1%) avec des faibles concentrations en azote (<1,5 g N

111 - (NH4)2CO3) / L), et une baisse évidente de la production de méthane avec des concentrations en azote supérieures à 5 g N - (NH4)2CO3 (< 5 g N - (NH4) 2CO3) / L). De façon plus précise, des concentrations en azote de 5.0, 7.5 et 10.0 g N - (NH4)2CO3) / L, ont entraîné une baisse de production d'environ 16%, 50% et 58%, respectivement.

Par contre, le protocole SMA n’a pas permis la caractérisation de l’impact des concentrations d’azote ammoniacal sur la digestion anaérobie des boues fécales au sein des latrines à fosses. Effectivement, bien que plusieurs tests d’activité méthanogène spécifique (SMA) aient été réalisés (en changeant le ratio inoculum /substrat (acétate), et la durée des tests), aucun des tests exécutés n’a démontré une activité méthanogène. À la lumière des résultats obtenus, de travaux supplémentaires ont été menés visant la caractérisation de l’activité microbienne au sein des fosses, lesquels ont été présentés au CHAPITRE 6.

Limites de l’étude et recommandations

Parmi les limites de l’étude, rappelons d’emblée la difficulté de travailler avec des boues issues des latrines

à fosse, compte tenu de la grande hétérogénéité des caractéristiques de celles-ci. Aussi, mentionnons que les boues fécales utilisées provenaient d’une seule latrine, dans laquelle l’accès aux boues n’était possible que par le trou de l’interface utilisateur. Ainsi, il était difficile d’échantillonner les boues aux mêmes endroits, ce qui s’ajoute aux paramètres pouvant affecter la variabilité en caractéristiques de celles-ci (e.g. solide totaux, âge des boues, etc.). De plus, mentionnons que les boues ont été prélevées au Québec durant la période d’activité des parcs, laquelle s’étend d’avril à octobre. Les boues ont donc été soumises aux aléas de la température, pouvant varier de plusieurs degrés durant cette période. Ceci pourrait avoir affecté l’activité microbiologique des boues.

Finalement, il convient de surcroît de mentionner que cette étude fut la première en son genre à être réalisée

à l’Université Laval. C’est-à-dire qu’aucune autre étude n’a été réalisée sur les boues fécales issues des

112 latrines à fosses avant celle-ci. Un temps considérable a donc été consacré à l’élaboration des protocoles d’échantillonnage et de caractérisation des boues. Des travaux supplémentaires seraient nécessaires visant l’amélioration des techniques. Par exemple, il serait intéressant d’étudier l’effet de la congélation des boues sur les propriétés physico-chimiques et biologiques de celles-ci, dans l’optique de pouvoir prolonger la période de conservation des boues.

8.3 Caractériser l’activité microbienne des boues fécales et de la matière fécale en termes des populations spécifiques à la digestion anaérobie

Cette étude a permis de caractériser l’activité microbienne d’échantillons de boues fécales et de la matière fécale fraiche en termes des populations spécifiques à la digestion anaérobie (e.g. bactéries hydrolytiques et fermentatives, et méthanogènes méthylotrophes, acétoclastiques ou hétérogénotrophes). De façon plus précise, les résultats ont démontré que l’utilisation des substrats de formate et de glucose ont enregistré l’activité microbienne la plus élevée lorsque mélangés avec une solution de matière fécale fraiche (soit de

112.17 et de 76.41 ml gaz/g SV, pour le formate et le glucose, respectivement) ; ceci indiquant une population active de méthanogènes hétérogenotrophes et de bactéries fermentatives. Alors que l’utilisation des substrats de glucose et de méthanol ont résulté à l’activité la plus élevée avec des solutions de boues fécales (soit de 129.15 et de 85.42 ml gaz/ g SV, pour le méthanol et le glucose, respectivement); indiquant, cette fois-ci, une population active de méthanogènes méthylotrophes et de bactéries fermentatives. Par ailleurs, dans les deux cas (pour les échantillons de boues fécales et de matière fécale fraîche), l'acétate ne semble pas être un substrat approprié pour la production de gaz, ce qui pose l’hypothèse d’une absence des méthanogènes acétoclastiques dans les échantillons de ces deux types de boues. Cette contribution est sans doute la plus importante de cette thèse, et vient appuyer les résultats précédents (CHAPITRE 5) et les études de Byrne et al. (2016) et Torondel et al. (2016), où des tests de séquençages d’ADN n’ont pas mené à identifier les archaea méthanogènes acétoclastiques au sein d’échantillons de boues fécales issues de latrines à fosse. Aussi, dans les deux cas, l'ajout de glucose a conduit à une réponse d'activité microbienne

113 rapide, laquelle serait produite par des bactéries acidogènes selon l'analyse des gaz ; la proportion de CO2 provenant du biogaz a été mesurée à 84,4% et 99,9% pour les échantillons de boues fécales et de matière fécale, respectivement. Or, la réaction d’acitogénèse en absence de méthanogènèse pourrait mener à une accumulation des sous-produits de décomposition anaérobique, tel que les acides gras volatiles (i.e. acide acétique, propionique, butyrique, etc.), lesquelles sont associés à des odeurs non souhaitables. Une étude récente a d’ailleurs démontré une accumulation d’acide butyrique au sein des fosses étant l’une des principales causes responsables des mauvaises odeurs se dégageant des fosses (Njalam’mano and Chirwa

2018). Or, les mauvaises odeurs sont l’une des principales raisons pouvant mener à l’abandon des latrines par ses usagers (pour effectuer la défécation en plein air) (Njalam’mano and Chirwa 2018). Et outre les problèmes d’odeurs, la faible activité des méthanogènes et l’activité des bactéries acidogènes pourrait expliquer le faible taux de dégradation des solides dans les latrines. En effet, l’acidogenèse n’entraîne pas la réduction des solides. Ainsi, il serait intéressant de poursuivre les analyses de caractérisation des boues fécales au sein des fosses, notamment en effectuant la méthodologie développée sur des échantillons de boues fécales provenant de différentes latrines.

Limites de l’étude et recommandations

Parmi les principales limitations, mentionnons l’absence de suivi des acides gras volatils au sein des fosses, ce qui aurait donné une meilleure idée sur l’état des réactions anaérobiques au sein des fosses. Aussi, mentionnons encore une fois que les tests d’activités ont été réalisés en utilisant les boues issues d’une seule latrine. De plus, bien que la méthodologie développée prévoit une période d’activation des boues de 24h

(incubation à 35C±1C), le délai pouvant parfois atteindre jusqu’à deux semaines avant la réalisation des tests SMA aurait possiblement pu influencer l’activité des bactéries au sein des boues. Finalement, d'autres expériences seraient nécessaires, notamment pour approfondir la connaissance sur la présence et le rôle de bactéries méthylotrophes dans les boues fécales, et sur la variabilité de l'activité méthanogène hydrogénotrophique. En effet, certains tests SMA (n°1 et 3 du tableau 6-13) ont mesuré l'activité des

114 méthanogènes hydrogénotrophes (à partir du formate), mais l'essai n° 2, quant à lui, n'a enregistré aucune activité. De plus, il serait intéressant de suivre l'activité microbiologique de la matière fécale au cours du temps et dans différentes conditions expérimentales (e.g. température, pH, etc.). Enfin, rappelons l’importance de caractériser l’activité microbienne d’échantillons de boues fécales provenant de différentes latrines (e.g. localisation, type de sol, âge des boues etc.). Il serait également intéressant d’effectuer les tests

SMA en combinaison avec les tests de séquençage d’ADN pour avoir une vue complète de l’activité microbiologique au sein des fosses. Ceci confirmant l’hypothèse selon laquelle il y aurait une absence des méthanogènes acétoclastes au sein des latrines.

8.4 Déterminer les caractéristiques de séchage de la matière fécale fraîche

Cette étude a permis de démontrer que les isothermes de sorption qui caractérisent la matière fécale tendent

à suivre une forme sigmoïde (comme d’ailleurs la plupart des produits alimentaires ayant une teneur élevée en eau), laquelle est caractérisée par deux points d’inflexion de surface (aux environs de 6 % et 75% en humidité relative) et une augmentation rapide de la teneur en eau à des valeurs d'humidité relative de 75% et plus. D’une première part, ceci signifie que l’eau contenue dans la matière fécale semblerait être plus facile à extraire jusqu'à environ 75% en activité de l’eau; en dessous de cette valeur, il faudra donc s’attendre

à devoir fournir de plus grandes quantités d’énergie pour sécher les échantillons. D’autre part, les valeurs expérimentales semblaient suivre le modèle à trois paramètres de Guggenheim-Anderson-de Boer (GAB) avec une estimation acceptable (Pr (> | t | <0.05). Rappelons également que la modélisation des isothermes de sorption constitue une étape indispensable dans l’élaboration des technologies de séchage des boues. De plus, il convient de mentionner que la méthode qui a été utilisée (à savoir la détermination des isothermes de sorption par voie gravitaire) est simple et facilement réalisable dans un contexte en ressources limitées

(pays en voie développement).

115 Par ailleurs, il serait intéressant de poursuivre la recherche, notamment en mesurant les isothermes de sorption à différentes températures. Ceci permettrait de calculer la chaleur de sorption, soit l’énergie requise pour casser les forces intermoléculaires entre les molécules d’eau et la surface du solide.

Limites de l’étude et recommandations.

La principale limite de cette étude repose sur le fait que les expérimentations ont été réalisées à seulement une température d’incubation, soit à 35C±1C. Des travaux supplémentaires seraient donc nécessaires pour caractériser la mobilité de l’eau à différentes températures. À cet effet, des expérimentations sont en cours

(dans le cadre de travaux de maitrise), lesquels visent justement la détermination des isothermes à différentes températures. L’idée est également d’ajouter des valeurs d’humidité relative (e.g. 10%, 15%

HR), en espérant ainsi une meilleure modélisation des courbes d’isothermes.

116 9. CONCLUSION GÉNÉRALE

Cette recherche a permis d’améliorer les connaissances en termes de caractérisation et de quantification des boues fécales au sein des latrines à fosses, en répondant aux différents objectifs (CHAPITRE 3). En résumé, les principales conclusions tirées (par objectif) sont :

Objectif 1 : Modéliser les taux d’accumulation des boues fécales au sein des latrines à fosses à l’aide d’une régression linéaire multiple

1) Un modèle de régression linéaire multiple (r2 = 0.41) reliant le volume des fosses et l’âge des boues

a été déterminé pour la prédiction des taux d’accumulation des boues fécales au sein des latrines à

fosses.

2) La répartition des données (à savoir : le volume des fosses, le nombre d’usagers, le temps depuis

la dernière vidange et les taux d’accumulation) correspondait à une distribution log-normale.

3) L’efficacité du modèle de régression s’est montrée légèrement plus efficace que le modèle par bilan

de masse développé par Brouckaert et al. (2013), en utilisant les mêmes données de la littérature.

Objectif 2 : Caractériser l’impact de l’azote ammoniacal sur les performances de digestion anaérobie des boues fécales et de la matière fécale

4) La méthodologie développée a permis de mesurer l’effet de l’azote ammoniacal sur la digestion

anaérobie de boues provenant d’un digesteur anaérobie, mais pas pour la digestion anaérobie

d’échantillons de boues fécales et de matière fécale.

5) Aucun des tests exécutés utilisant de l’acétate comme substrat n’a démontré une activité

méthanogène (production de biogaz).

117 Objectif 3 : Caractériser l’activité microbienne des boues fécales et de la matière fécale en termes des populations spécifiques à la digestion anaérobie

6) L’utilisation du substrat acétate n’a pas démontré une activité méthanogène, supportant l’hypothèse

d’une absence des archaea méthanogènes acétoclastiques dans les échantillons de ces deux types

de boues.

7) L’ajout des substrats de formate et de glucose a enregistré la production de biogaz la plus élevée

lorsque mélangés avec une solution de matière fécale fraiche (soit de 112.17 et de 76.41 ml gaz/ g

SV, pour le formate et le glucose, respectivement).

8) L’ajout des substrats de glucose et de méthanol a résulté en la production de biogaz la plus élevée

avec des solutions de boues fécales (soit de 129.15 et de 85.42 ml gaz/g SV, pour le méthanol et le

glucose, respectivement).

Objectif 4 : Déterminer les caractéristiques de séchage de la matière fécale fraîche.

9) Les isothermes de sorption des échantillons de matière fécale présentent une forme sigmoïde, se

caractérisant par deux points d’inflexion de surface (aux environs de 6 % et 75% en humidité

relative) et une augmentation rapide de la teneur en eau à des valeurs d'humidité relative de 75 %

et plus.

10) Le modèle à trois paramètres de Guggenheim-Anderson-de Boer (GAB) est le modèle décrivant le

mieux les isothermes obtenus expérimentalement avec une estimation acceptable - Pr (> | t | < 0.05).

Somme toute, il ne fait aucun doute que ces différents résultats obtenus dans le cadre de cette recherche doctorale formeront les bases, servant de références pour des travaux futurs. Par exemple, la poursuite des travaux visant une amélioration du modèle de régression linéaire multiples est désormais justifiée (i.e. augmentation du jeu de données, étendre le nombre de paramètres, etc.). Il s’agit donc d’un pas de plus vers

118 une prédiction des taux d’accumulation des boues fécales au sein des latrines à fosse, ce qui est essentiel pour l’organisation d’une gestion des boues fécales efficace dans les municipalités des pays en voie de développement.

De plus, cette recherche a permis l’élaboration et la mise à l’essai d’une méthodologie de caractérisation de l’activité microbienne spécifique aux réactions anaérobiques de différents types de boues (i.e. matière fécale et boues fécales). Les résultats obtenus sont prometteurs ayant soulevé l’hypothèse d’une absence de digestion anaérobique acetoclastic au sein des fosses. Il s’agit d’une contribution importante puisque, rappelons-le, on estime que 1.3 milliard de personnes utilisent les latrines à fosses comme principal et seul système d’assainissement. Ainsi, la poursuite des travaux en ce sens ouvre la voie vers une optimisation de la conception et de l’opération des latrines visant une réduction des taux d’accumulation (amélioration de la biodégradation) et un meilleur contrôle des caractéristiques des boues.

Finalement, la connaissance des isothermes de sorption renseigne sur la mobilité de l’eau au sein de la matière fécale; information importante pour le développement des technologies de séchage. Des travaux supplémentaires sont actuellement en cours (travaux de maîtrise) à l’Université de Victoria, et en collaboration avec l’Université Kwazulu-Natal, en Afrique du Sud, utilisant la méthodologie développée dans le cadre de cette thèse et en testant différentes températures et types de boues (i.e. latrines à fosses, boues issues de toilette à séparation d’urine). Un article a été soumis (ANNEX 3) et un second est prévu à l’issue de ces travaux, s’ajoutant ainsi aux connaissances limitées concernant le séchage des boues.

119 10. REFERENCES

Al-Muhtaseb, A.H., McMinn, W.A.M. and Magee, T.R.A. (2002) Moisture sorption isotherm characteristics of food products: A Review. Food and Bioproducts Processing 80(2), 118-128.

Amani, T., Nosrati, M. and Sreekrishnan, T. (2010) Anaerobic digestion from the viewpoint of microbiological, chemical, and operational aspects - a review. Evironmental review 18(1), 255-278.

Angelidaki, I., Alves, M., Bolzonella, D., Borzacconi, L., Campos, J.L., Guwy, A.J., Kalyuzhnyi, S., Jenicek, P. and van Lier, J.B. (2009) Defining the biomethane potential (BMP) of solid organic wastes and energy crops: a proposed protocol for batch assays. Water Science Technology 59(5), 927-934.

Angelidaki, I., Alves, M., Bolzonella, D., Borzacconi, L., Campos, L., Guwy, A., Jenicek, P., Kalyuzhnui, S. and Van Lier, J. (2006) Anaerobic Biodegradation, Activity and Inhibition (ABAI) Task Group Meeting 9th to 10th October 2006, in Prague.Institute of Environment and Ressource Technical University of Denmark.

APHA (1998) Standard Methods for the Examination of Water and Wastewater, Washington DC.

Appiah-Effah, E., Nyarko, K.B., Gyasi, S.F. and Awuah, E. (2014) Faecal sludge management in low income areas: a case study of three districts in the Ashanti region of Ghana. Journal of Water Sanitation and Hygiene for Development 4(2), 189-199.

Arlabosse, P., Rodier, E., Ferrasse, J.H., Chavez, S. and Lecomte, D. (2003) Comparison between static and dynamic methods for sorption isotherm measurements. Drying Technology 21(3), 479-497.

Bakare, B.F. (2014) Scientific and management support for ventilated improved pit Latrines (VIP) sludge content. Master Thesis University Kwala-Zulu Natal, Durban.

Baskaran, T.R. (1962) A decade of research in environmental sanitation. . Indian Council on Medical Research. Special Report Series (No. 40).

Bassan, M., Mbéguéré, M., Tchonda, T., Zabsonre, F. and Strande, L. (2013a) Integrated faecal sludge management scheme for the cities of Burkina Faso. Journal of Water Sanitation and Hygiene for Development 3(2), 216-221.

Bassan, M., Tchonda, T., Yiougo, L., Zoellig, H., Mahamane, I., Mbéguéré, M. and Strande, L. (2013b) Characterization of faecal sludge during dry and rainy seasons in Ouagadougou, Burkina Faso. 36th WEDC International Conference, Nakuru, Kenya.

Bhagwan, J.N., Still, D., Buckley, C. and Foxon, K. (2008) Challenges with up-scaling dry sanitation technologies. Water Science and Technology 58(1), 21-27.

Bocher, B.T.W., Cherukuri, K., Maki, J.S., Johnson, M. and Zitomer, D.H. (2015) Relating methanogen community structure and anaerobic digester function. Water Research 70(Supplement C), 425-435. 120 Brouckaert, C.J., Foxon, K.M. and Wood, K. (2013) Modelling the filling rate of pit latrines. WATER SA 39(4), 555-562.

Brunauer, S., Deming, L.S., Deming, W.E. and Teller, E. (1940) On a theory of the van der Waals adsorption of gases. Journal of the American Chemical Society 62(7), 1723-1732.

Buckley, C.A., Foxon, K.M., Brouckaert, C.J., Rodda, N., Nwaneri, C., Balboni, E., Couderc, A. and Magagna, D. (2008) Scientific Support for the Design and Operation of Ventilated Improved Pit Latrines (VIPs) and the Efficacy of Pit Latrine Additives. . Commission, W.R. (ed), Pretoria, South Africa

Byrne, A., Sindall, R., Wang, L., de los Reyes III, F.L. and Buckley, C. (2017) What happens inside a pour-flush pit insights from comprehensive characterization, 40th WEDC International Conference, Loughborough, UK, 2017.

Caballero-Cerón, C., Guerrero-Beltrán, J.A., Mújica-Paz, H., Torres, J.A. and Welti-Chanes, J. (2015) Water Stress in Biological, Chemical, Pharmaceutical and Food Systems. Gutiérrez-López, G.F., Alamilla-Beltrán, L., del Pilar Buera, M., Welti-Chanes, J., Parada-Arias, E. and Barbosa-Cánovas, G.V. (eds), pp. 187-214, Springer New York, New York, NY.

Cavinato, C., Ugurlu, A., de Godos, I., Kendir, E. and Gonzalez-Fernandez, C. (2017) Microalgae-Based Biofuels and Bioproducts. 7- Biogas production from microalgae, Woodhead Publishing.

Chaggu, E.J. (2009) Sustanable Environmental Protection Using Modified Pit-Latrine. Ph.D Thesis, Wageningen University, The Netherlands.

Chaudhary, P.P., Gaci, N., Borrel, G., O’Toole, P.W. and Brugère, J.-F. (2015) Molecular methods for studying methanogens of the human gastrointestinal tract: current status and future directions. Applied Microbiology and Biotechnology 99(14), 5801-5815.

Chen, G., Hung, W., Chang, I., Lee, S. and Lee, D.J. (1997) Continuous classification of moisture content in waste activated sludges. Journal of Environmental Engineering - ASCE 123(3), 253-258.

Chowdhry, S. and Kone, D. (2012) Business analysis of fecal sludge management: Emptying and Transportation Services in Africa and Asia. Report. The Bill & Melinda Gates Foundation, Seattle, USA.

Chris, R., Alison, P. and Elise, C. (2014) The biochemical methane potential of faecal sludge, Cranfield University.UK.

Colin, F. and Gazbar, S. (1995) Distribution of water in sludges in relation to their mechanical dewatering. Water Research 29(8), 2000-2005.

Couderc, A.A., Foxon, K., Buckley, C.A., Nwaneri, C.F., Bakare, B.F., Gounden, T. and Battimelli, A. (2008) The effect of moisture content and alkalinity on the anaerobic biodegradation of pit latrine sludge. Water Science Technololgy 58(7), 1461-1466.

121 Dolfing, J. and Bloeman, W.G.B.M. (1985) Acitivity measurements as a tool to characterize the microbial composition of methanogenic environments. Journal of Microbiological Methods 4(1), 1-12.

Foxon, K. (2010) Scientific support for the design and operation of ventilated improved pit latrines. Report. Pollution Research Group School of Chemical Engineering University of KwaZulu-Natal, Durban, South Africa

Foxon, K. and Still, D. (2012) Tackling the challenges of full pit latrines.Water Research Commission Petroria, South Africa

Foxon, K.M., Mkhize, S., Reddy, M. and Buckley, C.A. (2009) Laboratory protocols for testing the efficacy of commercial pit latrine additives. WATER SA 35(2), 228-235.

Franceys, R., Pickford, J. and Reed, R. (1992) A guide to the development of on-site sanitation. Report. World Health Organization.Geneva, Switzerland

Gaci, N., Borrel, G., Tottey, W., O'Toole, P.W. and Brugere, J.F. (2014) Archaea and the human gut: new beginning of an old story. World Journal Gastroenterology 20(43), 16062-16078.

Gold, M., Harada, H., Therrien, J.-D., Nishida, T., Cunningham, M., Semiyaga, S., Fujii, S., Dorea, C., Nguyen, V.-A. and Strande, L. (2017) Cross-country analysis of faecal sludge dewatering. Environmental Technology, 1-11.

Graham, J.P. and Polizzotto, M.L. (2013) Pit latrines and their impacts on groundwater quality: a systematic review. Environ Health Perspect 121(5), 521-530.

Guillard, V., Bourlieu, C. and Gontard, N. (2013) Food Structure and Moisture Transfer a Modeling Approach, Springer, Dordrecht, the Netherlands

Günther, I., Horst, A., Lüthi, C., Mosler, H., Niwagaba, C.B. and Tumwebaze, I.K. (2011) Where do Kampala's poor "go"? - urban sanitation conditions in Kampala's low-income areas. Report. ETH, Zurich, .

Hach (2013) Water analysis guide GMBH, H.L. (ed), HACH, Düsseldorf, Switzerland. .

Harvey, P.A., Baghri, S. and Reed, R.A. (2002) Emergency Sanitation: Assessment and programme design. WEDC, Loughborough University, UK.

Hashimoto, K., Matsuda, M., Inoue, D. and Ike, M. (2014) Bacterial community dynamics in a full-scale municipal wastewater treatment plant employing conventional activated sludge process. Journal of Bioscience and Bioengineering 118(1), 64-71.

He, X., Lau, A.K., Sokhansanj, S., Lim, C.J., Bi, X.T., Melin, S. and Keddy, T. (2013) Moisture sorption isotherms and drying characteristics of aspen (Populus tremuloides). Biomass and Bioenergy 57(Supplement C), 161- 167.

122 Heinss, U., Larmie, S. and Strauss, M. (1998a) Solids Separation and Pond Systems for the Treatment of Faecal Sludges in the Tropics. Swiss Federal Institute for Environmental Science and Technology. Department for Water and Sanitation in Developing Countries.

Heinss, U., Larmie, S.A. and Strauss, M. (1998b) Solids Separation and Pond Systems For the Treatment of Faecal Sludges In the Tropics.Report. SANDEC, E. (ed).

Henze, M. (2008) Biological Wastewater Treatment : Principles, Modelling and Design, IWA Pub, London,UK.

Henze, M., Harremoës, P., la Cour Jansen, J. and Arvin, E. (2002) Wastewater Treatment: Biological and Chemical Processes, Springer-Verlag, Berlin. Germany.

Herwijn, A. J. M. (1996). Fundamental aspects of sludge characterization Eindhoven: Technische Universiteit Eindhoven the Neterlands. DOI: 10.6100/IR4584.

Hines, W.W., Montgomery, D.C., Goldsman, D.M., Borror, C.M., Reny-Nolin, E. and Adjengue, L. (2017) Probabilités et statistique pour ingénieurs, Chenelière éducation, Montréal (Québec), Canada.

Holliger, C., Alves, M., Andrade, D., Angelidaki, I., Astals, S., Baier, U., Bougrier, C., Buffière, P., Carballa, M., de Wilde, V., Ebertseder, F., Fernández, B., Ficara, E., Fotidis, I., Frigon, J.-C., de Laclos, H.F., Ghasimi, D.S.M., Hack, G., Hartel, M., Heerenklage, J., Horvath, I.S., Jenicek, P., Koch, K., Krautwald, J., Lizasoain, J., Liu, J., Mosberger, L., Nistor, M., Oechsner, H., Oliveira, J.V., Paterson, M., Pauss, A., Pommier, S., Porqueddu, I., Raposo, F., Ribeiro, T., Rüsch Pfund, F., Strömberg, S., Torrijos, M., van Eekert, M., van Lier, J., Wedwitschka, H. and Wierinck, I. (2016) Towards a standardization of biomethane potential tests. Water Science and Technology 74(11), 2515-2522.

Hutton, G. and Varughese, M. (2016) The Cost of Meeting the 2030 Sustainable Development Goal Targets on Drinking Water, Sanitation, and Hygiene. Technical paper, World Bank/Water and Sanitation Programme (WSP), Washington, DC.

IRC (2006) The value of environmental sanitation - Case studies. Report. McIntyre, P. (ed), Delft, the Netherlands.

Janjai, S., Lamlert, N., Tohsing, K., Mahayothee, B., Bala, B.K. and Müller, J. (2010) Measurement and modeling of moisture sorption isotherm of litchi (Litchi Chinensis Sonn.). International Journal of Food Properties 13(2), 251-260.

Jannot, Y. (2008 ) Isotherme de sorption : modélisation et charactéristiques. Pdf.

Jayathilake, N., Fernando, S., Keraita, B., Paul, J. and Drechsel, P. (2017) Review of Guidelines and Regulations for Fecal Sludge Management.International Water Management Institute (IWMI).CGIAR Research Program on Water, Land and Ecosystems (WLE). In preparation, Colombo, Sri Lanka.

Jenkins, M.W., Cumming, O. and Cairncross, S. (2015) Pit latrine emptying behavior and demand for sanitation services in Dar Es Salaam, Tanzania. International Journal of Environmental Research and Public Health 12(3), 2588-2611. 123 Katsiris, N. and Kouzeli-Katsiri, A. (1987) Bound water content of biological sludges in relation to filtration and dewatering. Water Research 21(11), 1319-1327.

Katukiza, A.Y., Ronteltap, M., Niwagaba, C.B., Foppen, J.W.A., Kansiime, F. and Lens, P.N.L. (2012) Sustainable sanitation technology options for urban slums. Biotechnology Advances 30(5), 964-978.

Kengne, I.M., Kengne, E.S., Akoa, A., Bemmo, N., Dodane, P.H. and Koné, D. (2011) Vertical-flow constructed wetlands as an emerging solution for faecal sludge dewatering in developing countries. Journal of Water Sanitation and Hygiene for Development 1(1), 13.

Kjellén, M., Pensulo, C., Nordqvist, P. and M., F. (2011) Global Review of Sanitation System Trends and Interactions with Menstrual Management Practices. Report.Stockholm Environment Institute., Sweden.

Kone, D., Cofie, O. and Nelson, K. (2010) Low-cost options for pathogen reduction and nutrient recovery from faecal sludge. IDEAS Working Paper Series from RePEc.

Kone, D. and Strauss, M. (2004) Low-cost Options for Treating Faecal Sludges (FS) in Developing Countries – Challenges and Performance.Proceeding for the 9th International IWA Specialist group conference on wetlands systems for water pollution control, Avignon, France.

Koottatep, T., Eamrat, R., Pussayanavin, T. and Polprasert, C. (2014) Hydraulic evaluation and performance of on-site Sanitation systems in central Thailand. Environmental Engineering Research 19(3), 269-274.

Koottatep, T., Surinkul, N. and Panuvatvanich, A. (2013) Accumulation rates of thickened-bottom sludge and its characteristics from water-based onsite sanitation systems in Thailand. Proceeding for the second international conference on Faecal Sludge Management (FSM2) Durban, South Africa

Koottatep, T., Surinkul, N., Polprasert, C., Kamal, A.S.M., Koné, D., Montangero, A., Heinss, U. and Strauss, M. (2005) Treatment of septage in constructed wetlands in tropical climate: lessons learnt from seven years of operation. Water Science and Technology 51(9), 119.

Koster, I.W. and Lettinga, G. (1984) The influence of ammonium-nitrogen on the specific activity of pelletized methanogenic sludge. Agricultural Wastes 9(3), 205-216.

Kulabako, R.N., Nalubega, M., Wozei, E. and Thunvik, R. (2010) Environmental health practices, constraints and possible interventions in peri-urban settlements in developing countries – a review of Kampala, Uganda. International Journal of Environmental Health Research 20(4), 231-257.

Kumar, A., Ketel, S., Vance, K., Oey, T., Neithalath, N. and Sant, G. (2014) Water vapor sorption in cementitious materials—measurement, modeling and interpretation. Transport in Porous Media 103(1), 69-98.

Labuza, P.T. (1984) Moisture Sorption: Practical Aspects of Isotherm Measurement and Use., Americn Association of Cereal Chemists, St. Paul, Minnesota, USA.

Lay, J.-J., Li, Y.-Y. and Noike, T. (1998) The Influence of pH and Ammonia Concentration on the Methane Production in High-Solids Digestion Processes. Water Environment Research 70(5), 1075-1082. 124 Lee, D.-J. and Lee, S.F. (1995) Measurement of bound water content in sludge: The use of differential scanning calorimetry (DSC). Journal of Chemical Technology & Biotechnology 62(4), 359-365.

Lee, D.J. and Hsu, Y.H. (1995) Measurement of bound water in sludges: A comparative study. Water Environment Research 67(3), 310-317.

Limpert, E., Stahel, W. and Abbt, M. (2001) Log-normal distributions across the sciences: Keys and clues. BIOSCIENCE 51(5), 341-352.

Mara, D. 1984. The design of ventilated improved pit latrines (English). Technical Advisory Group (TAG) technical note; no. 13. Washington, DC : The World Bank. http://documents.worldbank.org/curated/en/618101468749362028/The-design-of-ventilated-improved-pit- latrines.

Metcalf and Eddy ((2003)) Water Engineering. Treatment and Reuse, Boston, USA.

Militon, C., Hamdi, O., Michotey, V., Fardeau, M.-L., Ollivier, B., Bouallagui, H., Hamdi, M. and Bonin, P. (2015) Ecological significance of Synergistetes in the biological treatment of tuna cooking wastewater by an anaerobic sequencing batch reactor. Environmental Science and Pollution Research 22(22), 18230-18238.

Murphy, C. (2015) Solids Accumulation Rates of Latrines at Rural Schools in Nimba County, Liberia. Master Thesis, University of South Florida, Fl, USA. .

Nakagiri, A., Niwagaba, C.B., Nyenje, P.M., Kulabako, R.N., Tumuhairwe, J.B. and Kansiime, F. (2015) Are pit latrines in urban areas of Sub-Saharan Africa performing? A review of usage, filling, insects and odour nuisances. BMC Public Health 16, 120.

Nakamura, N., Lin, H.C., McSweeney, C.S., Mackie, R.I. and Gaskins, H.R. (2010) Mechanisms of microbial hydrogen disposal in the human colon and implications for health and disease. Annual Review of Food Science and Technology 1(1), 363-395.

Nikiema, J., Bourgault, C., Amoah, P. and Nartey, E. (2017) Environmental & Health Risk Assessment of On-site systems (Pits and Septic tanks) – Literature Review. World Bank Report - In process India.

Njalam’mano, J.B.J. and Chirwa, E.M.N. (2018) Isolation, Identification and characterisation of butyric acid degrading bacterium from pit latrine faecal sludge. Chemical Engineering Transaction 64.

Norris, J.A. (2000) Sludge Build-Up in Septic Tanks, Biological Digesters and Pit Latrines in South Africa. Report. Water Research Commission. , Petroria, South Africa

Nwaneri, C.F. (2009) Physico-chemical characteristics and biodegradability of contents of ventilated improved pit latrines (VIPs) in eThekwini Municipality. Master Thesis., Kwazulu-Natal, Durban.

Orikasa, T., Wu, L., Ando, Y., Muramatsu, Y., Roy, P., Yano, T., Shiina, T. and Tagawa, A. (2010) Hot air drying characteristics of sweet potato using moisture sorption isotherm analysis and its quality changes during drying. International Journal of Food Engineering 6(2). 125 Park, S., Cui, F., Mo, K. and Kim, M. (2016) Mathematical models and bacterial communities for ammonia toxicity in mesophilic anaerobes not acclimated to high concentrations of ammonia. Water Science and Technology 74(4), 935-942.

Procházka, J., Dolejš, P., Máca, J. and Dohányos, M. (2012) Stability and inhibition of anaerobic processes caused by insufficiency or excess of ammonia nitrogen. Applied Microbiology and Biotechnology 93(1), 439-447.

Purohit, S.R. and Rao, P.S. (2017) Modelling and analysis of moisture sorption Isotherm of raw and pregelatinized rice flour and Its crystalline status prediction. Food Analytical Methods 10(6), 1914-1921.

Pussayanavin, T., Koottatep, T., Eamrat, R. and Polprasert, C. (2015) Enhanced sludge reduction in septic tanks by increasing temperature. Journal of Environmental Science and Health, Part A 50(1), 81-89.

RCoreTeam (2017) R: A language and environment for statistical computing.R Foundation for Statistical Computing.

Vienna, Austria.

Riviere, D., Desvignes, V., Pelletier, E., Chaussonnerie, S., Guermazi, S., Weissenbach, J., Li, T., Camacho, P. and Sghir, A. (2009) Towards the definition of a core of microorganisms involved in anaerobic digestion of sludge. ISME Journal 3(6), 700-714.

Rodríguez-Méndez, R., Lessard, P. and Yann, L. (2015) Digestion anaérobie des résidus d'abattoirs de veaux de lait : caractérisation, traitement et modélisation. Thèse., Université Laval Qc, Canada

Rose, C., Parker, A., Jefferson, B. and Cartmell, E. (2015) The Characterization of feces and urine: A review of the literature to inform advanced treatment technology. Critical Reviews in Environmental Science and Technology 45(17), 1827-1879.

Rozzi, A. and Remigi, E. (2004) Methods of assessing microbial activity and inhibition under anaerobic condition: a literature review. Environmental Science and Bio Technology 3, 95-115.

Sanches, W.R. and Wagner, E.G. (1954) Experience with excreta-disposal programmes in rural areas of Brazil. Bulletin of the World Health Organization 10(2), 229-249.

Sato, T., Qadir, M., Yamamoto, S., Endo, T. and Zahoor, A. (2013) Global, regional, and country level need for data on wastewater generation, treatment, and use. Agricultural Water Management 130, 1-13.

Schoebitz, L., Bassan, M., Ferré, A., Vu, T.H.A., Nguyen, V.A. and Strande, L. (2014) FAQ: faecal sludge quantification and characterization–field trial of methodology in Hanoi, Vietnam. proceeding of the 37th WEDC International Conference, Hanoi, Vietnam, 2014.

Schoebitz, L., Bischoff, F., Ddiba, D., Okello, F., Nakazibwe, R., Niwagaba, C.B., Lohri, C.R. and Strande, L. (2016) Results of faecal sludge analyses in Kampala, Uganda: Pictures, characteristics and qualitative observations for 76 samples. Eawag: Swiss Federal Institute of Aquatic Science and Technology. Report. , Dübendorf, Switzerland. 126 Schröder, H. (1981) S. M. Th. D. Brock, Biology of Microorganisms (3rd Edition). Zeitschrift für allgemeine Mikrobiologie 21(6), 475-476.

Shanks, O.C., Newton, R.J., Kelty, C.A., Huse, S.M., Sogin, M.L. and McLellan, S.L. (2013) Comparison of the microbial community structures of untreated wastewaters from different geographic locales. Applied and Environmental Microbiology 79(9), 2906-2913.

Shapiro, S.S. and Wilk, M.B. (1965) An analysis of variance test for normality (complete samples). Biometrika 52(3/4), 591-611.

Souto, T., Aquino, S., Silva, S. and Chernicharo, C. (2010) Influence of incubation conditions on the specific methanogenic activity test. Biodegradation 21(3), 411-424.

Sphere Project (2011) Sphere Handbook: Humanitarian charter and minimum Sstandards in disaster response, available at: http://www.refworld.org/docid/4ed8ae592.html [accessed 23 May 2018].

Still, D. and Foxon, K. (2012a) Tackling the challenges of full pit latrines; Volume 1: Understanding sludge accumulation in VIPs and strategies for emptying full pits.Water Research Commission.

Still, D. and Foxon, K. (2012b) Tackling the challenges of full pit latrines. Volume 2: How fast do pit toilets fill up? A scientific understanding of sludge build up and accumulation in pit latrines.Report. Water Research Commission.Petroria, South Africa

Strande, L., Ronteltap, M. and Brdjanovic, D. (2014) Faecal Sludge Management – Systems Approach for Implementation and Operation. IWA Publishing.London, UK. .

Sung, S. and Liu, T. (2003) Ammonia inhibition on thermophilic anaerobic digestion. Chemosphere 53(1), 43-52.

Thiele, H. (1953) The Dynamical Character of Adsorption, von H. J. de Boer. Oxford University Press, 1953. 1. Aufl. X V, 239 S., 45 Abb. gebd. s. 30.—. Angewandte Chemie 65(16), 431-431.

Tilley, E., Ulrich, L., Lüthi, C., Reymond, P. and Zurbrügg, C. (2014) Compendium of Sanitation Systems and Technologies. 2nd Revised Edition., Dübendorf, Switzerland.

Todman, L.C., van Eekert, M.H.A., Templeton, M.R., Hardy, M., Gibson, W.T., Torondel, B., Abdelahi, F. and Ensink, J.H.J. (2014) Modelling the fill rate of pit latrines in Ifakara, Tanzania. Journal of Water, Sanitation and Hygiene for Development 5(1), 100-106.

Torondel, B. (2010) Sanitation Venture Literature Review: on-site sanitation waste characteristics.Report. , London School of Hygiene and Tropical Medecine, London.

Torondel, B., Ensink, J.H.J., Gundogdu, O., Ijaz, U.Z., Parkhill, J., Abdelahi, F., Nguyen, V.-A., Sudgen, S., Gibson, W., Walker, A.W. and Quince, C. (2016) Assessment of the influence of intrinsic environmental and geographical factors on the bacterial ecology of pit latrines. Microbial Biotechnology 9(2), 209-223.

127 Trujillo, F.J., Yeow, P.C. and Pham, Q.T. (2003) Moisture sorption isotherm of fresh lean beef and external beef fat. Journal of Food Engineering 60(4), 357-366.

UNICEF/WHO (2011) Progress on sanitation and drinking water: 2010 update., France

United States Environmental Protection Agency, U. (1994) Guide to Septage Treatment and Disposal. (Document EP/625/R-94/002).

United States Environmental Protection Agency, U. (2002) Onsite Wastewater Treatment System Manual. United States Environmental Protection Agency, U. (ed).

Valcke, D. and Verstraete, W. (1983) A practical method to estimate the acetoclastic methanogenic biomass in anaerobic sludges. Water Pollution Control Federation 55(9), 1191-1195.

Van Minh, H. and Nguyen-Viet, H. (2011) Economic aspects of sanitation in developing countries. Environ Health Insights 5, 63-70.

Vanderhaeghen, S., Lacroix, C. and Schwab, C. (2015) Methanogen communities in stools of humans of different age and health status and co-occurrence with bacteria. FEMS Microbiology Letters 362(13), 92.

Vaxelaire, J. (2001) Moisture sorption characteristics of waste activated sludge. Journal of Chemical Technology and Biotechnology 76(4), 377-382.

Vaxelaire, J. and Cézac, P. (2004) Moisture distribution in activated sludges: a review. Water Research 38(9), 2215-2230.

Vodounhessi, A. and von Münch, E. (2006) Financial challenges to making faecal sludge management an integrated part of the ecosan approach: case study of kumasi, Ghana. Water Practice and Technology 1(2).

Vonwiller, L. (2007) Monitoring of the faecal sludge treatment plant Camberène in Dakar, Eawag, Dübendorf, Switzerland.

Wagner, E.G. and Lanoix, J.N. (1958) Excreta Disposal For Rural Areas And Small Communities.World Health Organization Switzerland

Walker, M. (2008) Performance of the FSTP Rufisque and its impact on the WSP in Dakar. Report. Eawag. Dübendorf, Switzerland

Wang, P., Wang, H., Qiu, Y., Ren, L. and Jiang, B. (2018) Microbial characteristics in anaerobic digestion process of food waste for methane production–A review. Bioresource Technology 248, 29-36.

Wei, Y.X., Ye, Z.F., Wang, Y.L., Ma, M.G. and Li, Y.F. (2011) Enhanced ammonia nitrogen removal using consistent ammonium exchange of modified zeolite and biological regeneration in a sequencing batch reactor process. Environmental Technology 32(12), 1337-1343.

WHO/UNICEF (2017) Progress on drinking water, sanitation and hygiene: 2017 update and SDG baselines.World Health Organization (WHO) and the United Nations Children’s Fund (UNICEF), Geneva, Switzerland

128 WWAP (2017) The United Nations World Water Development Report 2017: Wastewater, The Untapped Resource., UNESCO, Paris, France

Zhang, L., Sun, D.-W. and Zhang, Z. (2017) Methods for measuring water activity (aw) of foods and its applications to moisture sorption isotherm studies. Critical Reviews in Food Science and Nutrition 57(5), 1052-1058.

Zuma, L., Velkushanova, K. and Buckley, C. (2015) Chemical and thermal properties of dry VIP latrine sludge. Water SA. 41(4), 534-540.

129 ANNEX 1: DATA

10.1 Todman et al. (2014)

Table A.1.1 presents the observed FS accumulation rate of 24 pit latrines located within rural and urban areas of

Tanzania. The accumulation rates were measured every two months over a period of 18 months using a laser sensor. Along with the monitoring of the accumulation rate, the physical characteristics of the pits (e.g. geometry, lining, roof, etc.) were recorded and the households were surveyed to provide information about pit latrine use and practices (i.e. volume of the pit, number of users). The emptying frequency was used as a proxy for the age of the latrine.

Table A.1.1: Plan area, average user numbers, estimated age of latrine and observed filling rate of the latrines (Todman et al. 2014) Volume of the pit (m3)* Average user number Age of latrine (yr) Observed filling rate (L/yr) 1.50 11 19 97.3 3.60 5.2 0.25 251.1 2.38 7 11 232.1 1.44 6 1 206.5 2.24 9.8 3 347.1 2.20 5.2 2.5 277.5 1.54 4 3 143.3 1.48 3.8 1 330.5 2.96 14 0.25 106.1 3.66 8 16 480.3 4.52 9.4 7 130.0 1.36 6.8 2 232.9 2.62 9.8 4 111.7 4.00 2.8 8 276.6 2.00 6.8 0.5 633.6 2.02 8.8 0.25 199.9 3.08 9.8 3 265.4 2.42 5.5 3 702.4 2.62 7 0.25 506.8 2.24 7.8 0.25 180.7 5.06 7 1.5 225.2 1.66 9.2 0.75 614.3 1.82 6 1 97.3 2.98 5.6 1 251.1 *Based on an average depth of approximately 2 m

130 10.2 Foxon and Still (2012)

Data reported in Still and Foxon (Still and Foxon 2012b) are shown in Table A.1.2. Since the data was expressed in l/cap/year, results were multiplied by the average number of users (6.3) to fit the model unit (l/year). Also, only the median of the accumulation rate per region was reported in the study, so the median value was used to validate the model.

Table A.1.2: Volume of pit, average user numbers, emptying frequency and observed filling rate of the latrine adapted from Still and Foxon (2012b). Average Average Average age Median volume of number of of latrine accumulation pit (m3) users (yr) rate (L/yr) 1.96 6.3 0.33 151.8 3.16 6.3 0.25 437.2 2.83 6.3 0.20 116.5 3.40 6.3 0.09 182.7 5.40 6.3 0.09 245.7 4.20 6.3 0.09 302.4 2.25 6.3 0.07 119.7 2.00 6.3 0.10 214.2

10.3 Schoebitz et al. (2016)

As part of a collaborative research project conducted by Eawag/Sandec and Makerere University, faecal sludge samples from both pit latrines and septic tank were collected and analysed in Kampala, Uganda, from November

2013 to April 2014 (Schoebitz et al. 2016). Only data from pit latrines were used in the study (Table A.1.3). The information collected in the study was also linked to household surveys (i.e. time since the last desludging, number of users, etc.) and operational characteristics (i.e. volume of containment, the % of the sludge emptied, type of on- site systems – pit latrine or septic tank). The accumulation rate was not directly measured but deducted using equation (A.1):

Vc´ %emptied Time (A.1) 131 Where Vc represent the volume of the containment. Using equation A.1, the calculation of the accumulation rate is therefore based on the hypothesis that the observed percentage of the pit emptied (% of the volume of the pit) is considered as the sludge that had been accumulated since the last emptying (the accumulation rate). It was also assumed that the pit was fully emptied at each desludging. Consequently, based on this assumption, only the data associated with small containments (< 15 m3) were used from that study. According to the authors, it was uncommon that a truck (trucks have a range of volumes from around 2.7 m3 up to 15 m3) would not be full after emptying the pit, the desludging services are paid by truck load not by volume (Schoebitz et al. 2016).

Data collected by Schoebitz (2016) and the accumulation rate estimation are presented in Table A3.

Table A.1.3 Observation of pit latrine emptying, data collected from Schoebitz et al. (2016) Sample % Volume of the pit (m3) Average number of users Age of the pits (yr) Accumulation Emptied rate (L/yr) 14 100 10 250 0.13 10000 19 50 8 18 0.67 4000 20 40 5 25 0.25 2000 22 100 2 100 0.21 2000 24 57 7 5 1.50 3990 28 33 12 45 1.00 3960 30 100 5 45 0.25 5000 34 100 4 15 0.67 4000 35 100 8 80 0.42 8000 36 93 4 27 1.33 3720 38 54 12 20 0.83 6480 40 65 10 7 1.50 6500 41 100 3 10 1.00 3000 42 25 12 30 0.50 3000 43 100 2 20 0.13 2000 44 57 7 70 0.67 3990 46 100 5 30 0.33 5000 50 100 3 5 5.00 3000 53 100 2 7 0.67 2000 55 100 1 12 1.50 1000 56 67 9 20 0.83 6030 60 100 4 7 2.00 4000 67 56 8 600 0.08 4480 71 50 13 20 0.25 5000 73 33 12 280 1.00 3960

132

133 ANNEX 2: NORMALITY OF THE DATA

The quantile-quantile plot and Shapiro-Wilk normality test were used to verify whether the data were normally distributed. From Figure (A.1.1) shown below the S-shape of the Q-Q graph suggested the non-normality of the data. This hypothesis is also supported by the analysis of the shape of the histogram. Indeed, the shape of the histogram does not expose a bell shape, usually attributed to normal distribution data, but rather seems to demonstrate a skewed distribution, usually associated with lognormal distribution. Moreover, skewed distributions are particularly common with data showing low mean values and large variances, and when values cannot be negative (Limpert et al. 2001) - as is the case with accumulation rate, volume of pits, number of users and the age of the latrine

134

Figure (A.1.1) show histogram and quantile-quantile plot conducted on Todman et al. (2014) data. 135

The Shapiro-Wilk normality test was applied to validate the non-normality of the data. This statistical test evaluated the null-hypothesis that the population is normally distributed.

Table A.2.1: Result of the Shapiro-Wilk test on data Parameter p-value Volume of the pit 0.041 Number of users 0.234 Age of the latrines 9.092e-06 Accumulation rate 0.007

As we can see in Table A.2.1, the p-value is less than the alpha level (0.05) for the parameter of the volume of the pit, age of the latrine, and the accumulation rate. Only the population represented by the number of users follows a normal distribution. Thus, the null hypothesis is rejected and there is evidence that the data tested are not from a normally distributed population for that parameters (Shapiro and Wilk 1965).

136 ANNEX 3: SUBMITTED PUBLICATIONS

Communication Experimental Determination of Moisture Sorption Isotherm of Fecal Sludge

Experimental Determination of Moisture Sorption Isotherm of Fecal Sludge

Catherine Bourgault 1,*, Paul Lessard 1, Claire Remington 2 and Caetano C. Dorea 1,2

1 Département de génie civil et de génie des eaux, Université Laval, Québec, QC G1V 0A6, Canada; [email protected] (P.L.); [email protected] (C.C.D.) 2 Department of Civil Engineering, University of Victoria, Victoria, BC V8P 5C2, Canada; [email protected] * Correspondence: [email protected]; Tel.: +01-581-985-5958

Received: 4 December 2018; Accepted: 8 February 2019; Published: date

Abstract: Dewatering and drying of fecal sludge (FS) is a key treatment objective in fecal sludge management as it reduces volume (thereby reducing emptying frequency and associated transportation costs), inactivates pathogens, and is beneficial and/or necessary to resource recovery activities such as composting and combustion as fuel. However, studies on dewatering performances of FS are limited. The physical water distribution of such matrices is not fully understood, limiting the progress in the development and optimization of FS dewatering technologies. The objective of this study is to present a gravimetric method intended to assess the dewatering characteristics and associated modelling of FS through moisture sorption isotherms. Samples were placed in airtight jars containing different saturated salt (NaOH, CaCl2, NaCl, KCl, K2SO4) solutions to reproduce a range of relative humidity values (6 to 97%). Results confirmed the achievement of characteristic sigma-shaped moisture sorption isotherms with increasing moisture adsorption at higher values of relative humidity. Furthermore, experimental data best fit the three-parameter Guggenheim–Anderson–de Boer (GAB) model. This method can be replicated to contribute critical data about the characterization of fecal sludge, a seriously under- researched matrix.

Keywords: desiccation; drying; fecal sludge management; moisture sorption isotherms; sanitation

1. Introduction In low-income countries, the most common forms for excreta disposal are on-site, non-sewered sanitation systems (NSS) [1]. If safely managed, these sanitation facilities can provide a hygienic and affordable method for excreta disposal, meaning that the fecal sludge (FS) is safely contained, collected, treated, and disposed of. Such systems would contribute to the Sustainability Development Goals technology spectrum for safely managed sanitation services [2]. However, in many regions of the world, current FS management practices and approaches are insufficient and often unsafe. Many municipalities have not yet put the necessary strategies, policies, and budgets in place to maintain these NSS systems (e.g., desludging and treatment), resulting in the contamination of the environment [3] and spread of diarrheal diseases [4]. It is estimated that only 26% of urban and 34% of rural sanitation services worldwide effectively prevent human contact with excreta along the entire sanitation chain and can therefore be considered safely managed [5]. Additionally, inadequately managed sanitation facilities have degraded water quality in most rivers across Africa, Asia, and Latin America, directly affecting quality of life, 137 working capacity of the inhabitants, education, and economies [6]. Better methods for the safe management of FS from NSS are urgently needed. The moisture content of fresh faeces ranges from 63 to 86% and the fecal sludge collected by vacuum trucks in urban areas can contain more than 95% water [7,8]. As FS has a high water content, dewatering processes (i.e., drying beds, thermal drying, centrifuges, etc.) for the safe and cost-effective transportation of FS are advantageous. These techniques can reduce water content, resulting in benefits such as reductions in associated transport costs, as well as potentially enhanced pathogen inactivation [9,10], which can reduce the health risks of handling the fecal matter. However, studies on the dewatering of FS are relatively limited given its importance. The physical water mobility of FS is not fully understood, possibly limiting the progress in the development and optimization of FS dewatering technologies. Future research is needed to develop predictive models of sludge characteristics on dewatering, based on a fundamental understanding of FS dewatering mechanisms [8]. Moisture sorption isotherms (MSIs) are a graphical representation describing the sorption process of water molecules into a specific material at a specific temperature. They illustrate where water molecules are progressively and reversibly released from hygroscopic forces in biological material as a result of mainly capillary effects and direct bonding [11]. This method is largely used in the food industry [12], and has been successfully used to describe the sorption behavior of activated sludge [13,14]. However, direct transfer of moisture distribution characteristics from wastewater sludge to FS is not possible due to considerable differences in composition (e.g., organics, total solids, etc.) [3]. Given the lack of desiccation-related data on FS and the potential for MSIs, our aim was to test the potential of an experimental method to produce moisture sorption isotherm data for FS and to then evaluate various mathematical models to determine which best describes the datasets.

2. Materials and Methods A static gravimetric method was adapted {Jannot, 2008 #670} to determine the sorption characteristics of the FS samples. Two initial masses of FS (1.5 g and 5.0 g) were tested to study the effect of the mass of the samples on the water sorption isotherm. Three commonly used mathematical models were then evaluated for best fit with the experimental results.

2.1. Fecal Sludge Sampling and Initial Characterization Surficial (i.e., top 5 to 10 cm exposed to air) FS was collected from a decentralized sanitation system in Quebec City (QC) that has urine separation and separate toilet paper disposal. The age of the FS was estimated to be 24 h or less. Once collected, the samples were transported in a closed plastic container and characterized for their total solids, volatile solids, and water content using methods 2540G from the Standard Methods [15]. The time between sampling and analysis was no more than 3 h.

2.2. Preparation of Saturated Salt Solution Five saturated salt solutions were used to reproduce a range of relative humidity (RH) values between 0.06 and 0.97 (Table 1) [16]. Each ACS grade salt was mixed with distilled water (approximately 100 mL) until a solution with excess crystals was formed. The saturated salt solutions were poured into Mason jars to a depth of 1 cm (Figure 1).

Table 1. Relative humidity of saturated salt solutions at 35°C [16].

Salt Relative Humidity (%) Sodium hydroxide (NaOH) 6 Calcium chloride (CaCl2) 24 Sodium chloride (NaCl) 75 Potassium chloride (KCl) 82 Potassium sulphate (K2SO4) 97 138 2.3. Determination of Sorption Isotherms The FS samples were manually homogenized using a stainless-steel spatula. One drop of thymol was brushed onto the bottom of each aluminum crucible to prevent mold growth. Then, the samples were placed and weighed in the aluminum crucibles. Samples were spread to have approximately the same circular shape with a radius of 0.5 or 1 cm for the 1.5 g and 5.0 g samples, respectively. Each crucible was then placed on an a PVC tube fixed to an airtight Mason jar with a saturated salt solution to control the relative humidity (Figure 1). These Mason jar desiccators were incubated at 35 ± 1°C. Samples were weighed at regular intervals (24 h) until their mass varied by less than 2% W/W, the adopted operational threshold for equilibrium moisture content determination (approximately two weeks). It was ensured to limit the opening time of the jars to avoid humidity loss or gain. Finally, the equilibrium water content was calculated from the dry mass providing a point of the isotherm of sorption when coupled with RH values.

Figure 1. Experimental set-up for the moisture sorption isotherm (MSI) determination of fecal sludge (FS).

2.4. Mathematical Models of Sorption Isotherms Table 2 lists the mathematical isotherm models used in this study for comparison to the empirical data. Models were selected based on their effectiveness for describing isotherms of high moisture content of biological materials (e.g., municipal sludge, food, plants). Nonlinear data analysis and curve fitting were executed in the R development core system (2013) [17]. The quality of each model was assessed with regard to the standard error of estimate and the residual standard error. Additionally, the Student’s test was applied on regression coefficient for significance (p-Value < 0.05).

Table 2. Isotherm models used for fitting experimental data.

Isotherms Equation Parameters Brunauer–Emmett–Teller (BET) M Ca M = Moisture content (MC) M = 0 w (1- a )(1+ Ca - a ) C = constant w w w M0 = Monolayer MC Guggenheim–Anderson–de Boer M CK 'a M = Moisture content (MC) M = 0 w (GAB) (1- K 'a )(1- K 'a + Ca K ') C, K’ = constants w w w M0 = Monolayer MC Flory–Huggins (FH) A, B = constants M = Aexp(Baw)

3. Results and Discussion

3.1. Moisture Sorption Isotherm from Two Different Sample Masses 139 Figure 2 shows representative sorption isotherm results for 1.5 g and 5.0 g samples. The boxplots present data in quartiles, allowing comparisons between the different RH results. Overall trends for both sample masses indicate achievement of sigma-shaped moisture sorption isotherms characterized by two characteristic bends. The first bend occurs at an RH of 0 to 6% and the second one at around 75 to 82% for both sample masses. According to Labuza (1984) [18], this is caused by the additive effects of Raoult’s law, capillary effects, and surface water interaction. More precisely, RH values between 0 and 6% are suggested to result from the Van der Waals forces on water molecules and constitutes a molecular monolayer on the surface of the product [19]. Then, between 6% and 75% RH, water molecules are adsorbed on the saturated monolayer, resulting in the creation of more layers [19]. In this region, water is held in the solid matrix by capillary condensation. This water is thus available as a solvent for low-molecular-weight solutes and for some biochemical reactions [12]. Finally, between 75 and 97% RH, water is suggested to be bonded due to macro-capillary forces or as part of the fluid phase in high-moisture materials. This shows early all the properties of bulk water, and thus can act as a solvent [12]. Microbial growth becomes a major deteriorative reaction at this zone as water is available for bacteria (pathogens) [12]. Finally, for RH = 100%, the curve tends to an asymptote which corresponds to free water [14]. These sigma-shape isotherms that FS sample seem to correspond to have been reported for most food products with high moisture content [18]. This is as expected since the composition of FS is like most food in terms of molecules present, namely carbohydrates, lipids, and proteins [7].

Figure 2. Box and whisker plots (min/max, lower/upper quartiles, and median) of equilibrium moisture content data for each given relative humidity at a temperature of 35°C and an initial FS sample with masses of (a) 1.5 g and (b) 5.0 g.

A qualitative comparison of the experimental data at 35°C, shown in Figure 2, suggests that the equilibrium moisture content reached by the 5.0 g sample mass is higher than that reached by the 1.5 g sample mass. Quantitatively, the sample mass has a significant impact on equilibrium moisture content at a given relative humidity (α < 0.001).

3.2. Modelling of Isotherm

Fitting Experimental Data Figure 3 depicts experimental data (average) with and Brunauer–Emmett–Teller (BET), Guggenheim– Anderson–de Boer (GAB), and Flory–Huggins (FH) models at 35°C and for both 1.5 g and 5.0 g initial sample masses. Table 3 summarizes the parameter estimates, standard error, and the residual standard error (RSE) to determine the best data fit for the model. The FH model was rejected for both sample masses, and the BET model was rejected for the 1.5 g sample mass, since model parameters (i.e., constant) failed the Student’s test for

140 significance (p-value < 0.05). The GAB and BET models are both valid for the 5.0 g sample mass since the equation parameters (C, M0, K’) were satisfactory (Pr (>|t|) = 0 to 0.05). For BET, M0 is associated with the moisture content (dry basis) corresponding to an adsorbed monolayer (BET) and C and K’ are constants related to the temperature effect. Further testing should be performed across the range of humidity values to improve the calibration of the models and provide better assessment of the most suitable one for FS.

Figure 3. Experimental data ((average) and predicted values; Brunauer–Emmett–Teller (BET) (___), Guggenheim– Anderson–de Boer (GAB) (---), and Flory–Huggins (FH) (…) model predicted adsorption at 35°C for (a) 1.5 g and (b) 5.0 g FS as initial mass.

Table 3. Predicted value analysis for selected models for 1.5 g and 5.0 g of fecal sludge (FS) as initial mass.

Sample Standard Model Parameters Value t-value Pr (>|t|) RSE Mass error M0 0.309 - - - 1.5 g 2,533 on 18 DoF C −24.800 23 690 −1.047 0.308 BET M0 0.309 - - - 5.0 g 2.837 on 22 DoF C 19.904 5 711 −3.485 0.002 ** M0 0.309 - - - C −4.997 1.658 −3.013 0.00746 ** 1.5 g 1.302 on 18 DoF < 2 × 10−16 K’ 0.973 0.007 142.140 *** GAB M0 0.309 - - - C −4.519 1.497 −3.017 0.00747 ** 5.0 g 2.361 on 22 DoF < 2 × 10−16 K’ 0.985 0.006 157.107 *** A 0.144 0.091 1.582 0.131 0.8884 on 18 1.5 g 4.17 × 10−5 B 3.739 0.696 5.373 DoF FH *** A 0.403 0.234 1.725 0.0986 5.0 g 1.46 on 22 DoF B 2.942 0.639 4.602 000139 *** Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’; DoF: degrees of freedom.

4. Conclusions The preliminary assessment of the moisture sorption isotherm of fecal sludge was investigated at 35°C using a static gravimetric method. Overall moisture sorption isotherm trends indicated achievement of a sigma shape characterized by two characteristic bends and an increase of moisture adsorption at higher values of relative humidity. This means that, in this study, water in fresh faeces samples seems to be more easily extracted until a water activity (aw) of approximately 75%; more energy will be required to lower the water activity below this

141 point. This sigma shape that fecal sludge isotherms seem to correspond to has also been reported for most food products with high moisture content (aw over 0.9) [18]. This is as expected, since the composition of faeces will be like most food in terms of molecules present, namely carbohydrates, lipids, and protein. Furthermore, the effect of the mass of the sample and water activity on moisture content conforms to what was expected, namely lower moisture content with decreasing mass of the sample used, as can be seen when comparing Figure 2a and 2b. Experimental data seemed to follow the three-parameter Guggenheim–Anderson–de Boer model with a reasonable fit (p-Value < 0.05). However, further tests should be realized at lower relative humidity values to improve the calibration of the model. This study validated the potential suitability of the gravimetric method for the characterization and modelling of the moisture sorption isotherms of fecal sludge. The results obtained represent one of the few datasets on dewatering properties for this type of matrix, and it is hoped that this method can be replicated to improve our understanding of fecal sludge desiccation properties. Future work using this approach includes determining the heat of sorption (or the solid–liquid bond strength); applying the approach to more samples from varied sources; analyzing the potential impact of chemical oxygen demand, pH, and total volatile solids on the equilibrium moisture content of fecal sludge to determine whether there are any universalities in the physical water distribution; comparing the physical water distribution of fresh faeces to that of fecal sludge; and investigating the impact of ventilation rates on drying rates. This could serve as a basis for the design of on-site sanitation systems for the in situ desiccation of fecal sludge intended for solids reduction and pathogen inactivation.

Author Contributions: Conceptualization, C.B. and C.C.D.; Methodology, C.B. and C.C.D.; Validation, C.B. and C.C.D.; Formal Analysis, C.B.; Investigation, C.B.; Data Curation, C.B.; Writing-Original Draft Preparation, C.B. and C.R.; Writing-Review and Editing, P.L. and C.C.D.; Supervision, P.L. and C.C.D.; Project Administration, C.C.D.; Funding Acquisition, C.C.D.

Funding: This research was partially funded by Aerosan Toilets (grant number SIRUL/116523) and the Humanitarian Innovation Fund (grant number 27319). C.B. was supported by Alexander Graham Bell Canada Graduate Scholarships from the Natural Sciences and Engineering Research Council (NSERC).

Acknowledgments: Michel Bisping is acknowledged for the technical support.

Conflicts of Interest: The authors declare no competing interests. Funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.099

References

1. Nakagiri, A.; Niwagaba, C.B.; Nyenje, P.M.; Kulabako, R.N.; Tumuhairwe, J.B.; Kansiime, F. Are pit latrines in urban areas of Sub-Saharan Africa performing? A review of usage, filling, insects and odour nuisances. BMC Public Health 2015, 16, 120. 2. World Health Organization (WHO)/ the United Nations Children’s Fund (UNICEF). Progress on drinking water, sanitation and hygiene: 2017 update and SDG baselines; WHO and UNICEF: Geneva, Switzerland, 2017. 3. Strande, L.; Ronteltap, M.; Brdjanovic, D. fecal Sludge Management—Systems Approach for Implementation and Operation; IWA Publishing: London, UK, 2014. 4. Mara, D.D.; Feachem, R. Water- and excreta-related diseases: Unitary environmental classification. J. Environ. Eng. -ASCE 1999, 125, 334–339. 5. Hutton, G.; Varughese, M. The Cost of Meeting the 2030 Sustainable Development Goal Targets on Drinking Water, Sanitation, and Hygiene; Technical paper; World Bank/Water and Sanitation Programme (WSP): Washington, DC, USA, 2016. 6. World Water Assessment Programme (WWAP). The United Nations World Water Development Report 2017, Wastewater: The Untapped Resource; United Nations Educational, Scientific and Cultural Organization (UNESCO): Paris, France, 2017. 7. Rose, C.; Parker, A.; Jefferson, B.; Cartmell, E. The Characterization of feces and urine: A review of the literature to inform advanced treatment technology. Crit. Rev. Environ. Sci. Technol. 2015, 45, 1827–1879. 8. Gold, M.; Harada, H.; Therrien, J.-D.; Nishida, T.; Cunningham, M.; Semiyaga, S.; Fujii, S.; Dorea, C.; Nguyen, V.-A.; Strande, L. Cross-country analysis of fecal sludge dewatering. Environ. Technol. 2017, 39, 3077–3087. 9. Kone, D.; Cofie, O.; Nelson, K. Low-cost options for pathogen reduction and nutrient recovery from fecal sludge; Reports, No H042609; International Water Management Institute: Colombo, Sri Lanka, 2010. 142 10. Hui, Y.H., Food drying science and technology: Microbiology, chemistry, applications; DEStech Publications: Lancaster, England, 2008; p. 792. 11. Caballero-Cerón, C.; Guerrero-Beltrán, J.A.; Mújica-Paz, H.; Torres, J.A.; Welti-Chanes, J. Moisture Sorption Isotherms of Foods: Experimental Methodology, Mathematical Analysis, and Practical Applications. In Water Stress in Biological, Chemical, Pharmaceutical and Food Systems; Gutiérrez-López, G.F., Alamilla-Beltrán, L., del Pilar Buera, M., Welti-Chanes, J., Parada-Arias, E., Barbosa-Cánovas, G. V., Eds.; Springer New York: New York, NY, USA, 2015; pp. 187–214. 12. Al-Muhtaseb, A.H.; McMinn, W.A.M.; Magee, T.R.A. Moisture sorption isotherm characteristics of food products: A Review. Food Bioprod. Process. 2002, 80, 118–128. 13. Herwijn, A.J.M. Fundamental Aspects of Sludge Characterization. Ph.D. Thesis, Technische Universiteit Eindhoven, Eindhoven, the Neterlands, 1996. DOI:10.6100/IR4584. 14. Vaxelaire, J. Moisture sorption characteristics of waste activated sludge. J. Chem. Technol. Biotechnol. 2001, 76, 377–382. 15. American Public Health Association (APHA). Standard Methods for the Examination of Water and Wastewater; APHA: Washington, DC, USA, 1998. 16. CRC Handbook of chemistry and physics. CRC Press: Boca Raton, FL, USA, 1977. 17. RCoreTeam. R: A language and environment for statistical computing. R Foundation for Statistical Computing; RCoreTeam: Vienna, Austria, 2017. URL: http://www.R-project.org/. 18. Labuza, P.T. Moisture Sorption: Practical Aspects of Isotherm Measurement and Use. Americn Association of Cereal Chemists: Saint Paul, MN, USA, 1984; Vol. Library of Concress Catalog Card Number: 83-73312. 19. Jannot, Y. Isotherme de sorption: Modélisation et charactéristiques; University of Bordeau I, laboratoire de TREFLE, France, 2008.

143

https://www.nature.com/npjcleanwater/

Brief communication Pit latrine design: Is it rocket science after all? Catherine Bourgault1, Paul Lessard1 & Caetano C. Dorea1,2,* 1Département de génie civil et de génie des eaux, Université Laval, Québec, QC, Canada 2Department of Civil Engineering, University of Victoria, Victoria, BC, Canada

Abstract

Pit latrines are one of the simplest forms of onsite sanitation. Their design guidelines presume anaerobic stabilisation of pit contents, hindering accumulation of solids within these systems. It has been assumed that such stabilisation occurs through acetoclastic methanogenesis. However, this has not been verified until this study, which was aimed at assessing methanogenic activity of faecal sludge. It has been shown that the acetoclastic pathway was not significant in the faecal sludge and fresh faeces samples studied. Whereas implications for pit latrine design cannot be inferred from this study, further research towards the characterisation of other methanogenic pathways may be informative with to this end.

Brief Communication

Recent Sustainable Development Goal estimates are that 24 % of the world’s population practice open defecation or rely on an unimproved sanitation1. Pit latrines are one of the most basic and common options on the sanitation ladder with an estimated 2 billion users globally. In the past decade, attention has been focused on the study of faecal sludge accumulation rates within onsite sanitation variants. Latrine design is relatively simple and considers the stabilisation of pit contents, which should in principle reduce the frequency of pit emptying. These units, when filled, require emptying, an expensive option for resident and a hazardous activity for many informal workers of this often unregulated sector. Moreover, when pit contents are disposed of in an unsanitary fashion, public and environmental health are compromised through exposure to faecal contamination and increased risk of diarrhoeal disease transmission.

144 The design principle of pit latrines has remained virtually unchanged in over half a century since first guidelines were published. The basic postulation is that biodegradation of pit contents occurs naturally and will hinder the accumulation of faecal sludge within vaults. This has been modelled in an analogous way to an anaerobic digestor2, which assumes a predominant mechanism of acetoclastic methanogenesis. Yet, this fundamental assumption has never been tested and specific methanogenic activity (SMA) tests could confirm or refute this as a plausible degradation mechanism. The objective of this study was to gain a first insight into the methanogenic activity of relevance to pit latrines.

SMA was assessed through manometric biogas yields with an OxiTop® Control AN 12 system (WTW, Germany).

Both fresh faeces (FF) with less than 24 hours and aged faecal sludge (FS) samples were collected from vaults of onsite sanitation systems, with and without urine diversion, respectively. Sludge from an anaerobic digestor (AD) treating local wastewater biosolids was also sampled as an acetoclastic methanogenesis seed sludge reference.

Triplicate samples were incubated at 35 °C in parallel to sample blanks with different substrates (i.e. acetate, formate, methanol, and glucose) with fixed substrate to inoculum rations using reagent grade chemicals (Sigma-

Aldrich). This was done to identify substrate-specific activity (i.e. acetoclastic, hydrogenotrophic, methylotrophic, and acidogenic) following a similar rationale used for such assays3. Biogas was analysed independently using a

TRACE™ 1310 Gaschromatograph (ThermoFisher Scientific, Germany). SMA assays for AD samples, revealed biogas (mostly methane) conversion kinetics from acetate-fed samples as expected from literature, confirming the prevalence of the acetoclastic methane-forming pathway for this type of sludge. SMA assays of FS samples indicated relatively low activities acetate (Figure 1). Although the biogas yield from acetate was lower than the blank no statistically significant difference (p<0.01) was observed between the blank (no substrate) and the acetate fed samples (i.e. acetoclastic methanogenic activity was not importance for the aged FS). Formate resulted in a relatively modest biogas conversion activity from FS. Highest SMA FS biogas yields resulted from methanol conversion. The intermediary acidogenesis step was confirmed with similar trends for both FS and FF samples

(Figures 1 and 2) through the production of CO2 when glucose was used as a substrate. This may corroborate indications of incomplete anaerobic breakdown of pit contents as suggested by chemical analysis of volatile compound emissions from pit latrines4. Neither methylotrophic (contrary to FS sample) nor acetoclastic activities 145 were statistically significant (p<0.01) in relation to sample blank when methanol and acetate were used, respectively, as a substrate in FF tests. SMA tests with FF samples showed presence of initial formate-induced

(possibly hydrogenotrophic) activity. However, the rapid production and consumption of biogas (other than CO2 and methane possibly not measured by gas chromatography) highlights the need for further experimentation to better characterise methanogenic activities in sludge samples.

Hydrogenotrophic methanogensis is one of the most prevalent methanogenic pathways in the human gut and stools7 (i.e. pit latrine input) and thus merits further investigation. Presence of hydrogenotrophic methanogens in latrines has been confirmed by other studies5,6, but their identification by microbial DNA sequencing does not necessarily indicate their viability/activity, which this study addressed. Importantly, no significant acetoclastic activity was observed and these preliminary SMA results point towards a possible shift in methanogenic activity as the fresh faeces ages. Interestingly, initial studies (from around the time pit latrine designs were first published) on hydrogenotrophic methanogensis were motivated by the “space race”8. The significance and potential role of these microbes in pit latrine design for passive treatment of human waste is still unknown, but further research is warranted. So, it is possible that pit latrines may not be that simple after all and their design may involve some rocket science!

Acknowledgements

Authors would like to thank the financial support from the Natural Sciences and Engineering Research Council

(NSERC) of Canada through an by an Alexander Graham Bell Canada Graduate Scholarship and a Discovery

Grant.

References

1. World Health Organization & UNICEF. Progress on Drinking Water, Sanitation and Hygiene - 2017

Update and SDG Baselines. (2017).

2. Brouckaert, C., Foxon, K. & Wood, K. Modelling the filling rate of pit latrines. Water SA 39, 555–562

(2013).

146 3. McKeown, R. M., Scully, C., Mahony, T., Collins, G. & O’Flaherty, V. Long-term (1243 days), low-

temperature (4–15°C), anaerobic biotreatment of acidified wastewaters: Bioprocess performance and

physiological characteristics. Water Research 43, 1611–1620 (2009).

4. Lin, J. et al. Qualitative and Quantitative Analysis of Volatile Constituents from Latrines. Environ. Sci.

Technol. 47, 7876–7882 (2013).

5. Torondel, B. et al. Assessment of the influence of intrinsic environmental and geographical factors on the

bacterial ecology of pit latrines: Bacterial ecology of pit latrines. Microbial Biotechnology 9, 209–223

(2016).

6. Bryne, A., Sindall, R., Wang, L., de los Reyes III, F. L. & Buckley, C. What happens inside a pour-flush

pit? Insights from comprehensive characterization. in 40th WEDC International Conference (2017).

7. Vanderhaeghen, S., Lacroix, C. & Schwab, C. Methanogen communities in stools of humans of different

age and health status and co-occurrence with bacteria. FEMS Microbiology Letters 362, fnv092–fnv092

(2015).

8. Grimble, G. Fibre, fermentation, flora, and flatus. Gut 30, 6–13 (1989).

147 Figure Legends

Figure 1. Typical biogas conversion trends from different substrates tested on faecal sludge (FS) samples.

Figure 2. Typical biogas conversion trends from different substrates tested on fresh faeces (FF) samples.

148