EPA/600/2-87/059
August 1987
SIMULATION OF LEACHATE GENERATION
FROM MUNICIPAL SOLID WASTE
by
Neil D. Williams
Frederick G. Pohland
Kathleen C. McGowan
F. Michael Saunders
Georgia Institute of Technology
School of Civil Engineering
Atlanta, Georgia 30332
U.S. EPA Cooperative Agreement CR812580010
Georgia Tech Project No. E20-68A
Project Officer
Jonathan G. Herrmann
Solid and Hazardous Waste Research,Division
Hazardous Waste Engineering Research Laboratory
HAZARDOUS WASTE ENGINEERING RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
CONTENTS
Page
NOTICE ii
FOREWORD iii
ABSTRACT iv
LIST OF FIGURES vii
LIST OF TABLES xi
ACKNOWLEDGEMENTS xii
SECTION 1 - INTRODUCTION 1
SECTION 2 - CONCLUSIONS 6
SECTION 3 - RECOMMENDATIONS 8
SECTION 4 - REVIEW OF PREVIOUS MECHANISTIC MODELS OF LANDFILL
STABILIZATION 9
SECTION 5 - GTLEACH-I MODEL DEVELOPMENT 18
THE HYDROGEOLOGIC MODUEL 18
PROPERTIES OF A HOMOGENEOUS, ISOTROPIC, PARTIALLY
SATURATED, POROUS MEDIA 24
BIOLOGICAL MODULE 33
ORGANIZATION OF GTLEACH-I 41
SECTION 6 - EVALUATION OF GTLEACH-I 44
SECTION 7 - FITTING GTLEACH-I EXPERIMENTAL DATA .... 63
SIMULTANEOUS FITTING 63
SENSITIVITY ANALYSIS 69
SEPARATE FITTING 95
SECTION 8 - MODEL MODIFICATIONS 106
-
^^:^^ 19103
-
-------
CONTENTS (continued)
Page
FLOW MODEL
HYDROLYSIS RATE CONSTANT
SECONDARY RELEASE OF SUBSTRATE
DELAY IN ESTABLISHMENT OF ACIDOGENESIS
121
SECTION 9 - FITTING GTLEACH-I TO SECOND SET OF EXPERIMENTAL DATA. .
1 •)-]
SECTION 10 - SUMMARY AND DISCUSSION OF MODEL GTLEACH-I .....
139
REFERENCES ...................
APPENDIX A ................... 145
APPENDIX B ...................
VI
-------
LIST OF FIGURES
Page
1 Comparison of Unsaturated Flow Model and Four-Reactor
Simulations 12
2 Relationship Between Pressure Head and Volumetric
Water Content for Soils 27
3 Specific Moisture Content of Soils 30
4 Partially Saturated Hydraulic Conductivity of Soils . . 31
5 Microbial Ecology of the Anaerobic Digestion Process . . 35
6 Solid Substrate Concentration in Landfill Moisture,
Single-Pass Simulation
46
7 Acidogen Concentration in Landfill Moisture, Single-
Pass Simulation 47
8 Hydrolysis Product Concentration in Leachate, Single-
Pass Simulation 48
9 Methanogen Concentration in Landfill Moisture, Single-
Pass Simulation 49
10 Volatile Acid Concentration in Landfill Leachate,
Single Pass Simulation 50
11 COD Concentration in Landfill Leachate, Single-Pass
Simulation 51
12 Methane Production, Single-Pass Simulation 52
13 Solid Substrate Concentration in Landfill Moisture,
Recycle Simulation 54
14 Acidogen Concentration in Landfill Design, Recycle
Simulation 55
vii
-------
15 Hydrolysis Product Concentration in Landfill Leachate,
Recycle Simulation
16 Methanogen Concentration in Landfill Moisture, Recycle
Simulation
17 Volatile Acid Concentration in Landfill Design,
Recycle Simulation
18 COD Concentration in Landfill Design, Recycle
Simulation
19 Methane Production, Recycle Simulation 60
20 Comparison of Simulated Volatile Acid Concentration
with Experimental Data, Single-Pass
70
21 Comparison of Simulated Volatile Acid Concentration
with Experimental Data, Recycle ......... 71
22 Comparison of Simulated COD with Experimental Data,
Single Pass ............... 72
23 Comparison of Simulated COD with Experimental Data,
Recycle ................. ^3
24 Comparisons of Simulated Methane Production with
Experimental Data, Single-Pass ......... 74
25 Comparison of Simulated Methane Production with
Experimental Data, Recycle .......... 75
26 Single-Pass Sensitivity Analysis of NCSTR ..... 76
27 Recycle Sensitivity Analysis of NCSTR ....... 77
28 Single-Pass Sensitivity Analysis of XKH ...... 78
29 Recycle Sensitivity Analysis of XKH ....... 79
30 Single-Pass Sensitivity Analysis of QA ...... 81
31 Recycle Sensitivity Analysis of QA ....... 82
32 Single-Pass Sensitivity Analysis of XKA ...... 83
33 Recycle Sensitivity Analysis of XKA ........ 84
34 Single-Pass Sensitivity Analysis of YA ...... 85
35 Recycle Sensitivity Analysis of YA ........ 86
I
viii
-------
36 Single-Pass Sensitivity Analysis of XKDA 87
37 Recycle Sensitivity Analysis of XKDA 88
38 Single-Pass Sensitivity Analysis of QM 89
39 Recycle Sensitivity Analysis of QM 90
40 Single-Pass Sensitivity Analysis of XKM 91
41 Recycle Sensitivity Analysis of XKM 92
42 Single-Pass Sensitivity Analysis of YM 93
43 Recycle Sensitivity Analysis of YM 94
44 Single-Pass Sensitivity Analysis of XKDM 96
45 Recycle Sensitivity Analysis of XKDM 97
46 Comparison of Simulated Volatile Acid Concentrations,
Single-Pass 100
47 Comparison of Simulated Methane Production, Single-Pass . 101
48 Comparison of Simulated Volatile Acid Concentrations,
Recycle 102
49 Comparison of Simulated Methane Production, Recycle . . 103
50 Comparison of Constant Flow with Cyclical Flow,
Single-Pass 108
51 Comparison of Constant Flow with Cyclical Flow,
Recycle 109
52 Comparison of Effects of Varying Distributions of Flow
on Volatile Acid Concentrations, Recycle 110
53 Comparison of Varying Distributions of Flow on Methane
Production, Recycle Ill
54 Comparison of Effect of Value of Initial XMASSO and KH
on Volatile Acid Concentration, Single-Pass 113
55 Comparison of Effect of Value of Initial SMASSO and KH
on Methane Production, Single-Pass 114
56 Comparison of Effect of Value of Initial SMASSO and KH
on Volatile Acid Concentration, Recycle 115
ix
-------
57 Comparison of Effect and Value of Initial SMASSO and KH
on Methane Production, Recycle ...... 116
58 Effect of Secondary Release of Substrate on Volatile
Acid Concentration, Single-Pass ......... 117
59 Effect of Secondary Release of Substrate on Methane
Production, Single-Pass ........... 118
60 Effect of Lag in Initiation of Acidogenisis on Volatile
Acid Concentration, Single-Pass ......... 120
61 Comparison of Single-Pass and Recycle Volatile Acid
Concentrations, Chang ............ 122
62 Comparison of Single-Pass and Recycle Volatile Acid
Concentrations, Shaeffer and Yari ........ 123
63 Comparison of Simulated Volatile Acid Concentrations,
with Experimental Data by Chang ......... 124
64 Comparison of Simulated Methane Production with
Experimental Data by Chang ........... 125
65 Comparison of Simulated Volatile Acid Concentrations
with Experimental Data by Chang, QM = 2.5 ..... 127
66 Comparison of Simulated Volatile Acid Concentrations
with Experimental Data by Chang, QM = 2.5,
Lag = 10 days ............... 128
67 Comparison of Simulated Volatile Acid Concentrations
with Experimental Data by Chang, QM = 2.5, Y = 0.04,
Lag = 10 days ............... 129
68 Comparison of Volatile Acid Concentration Simulation
with Experimental Data by Chang, Recycle ...... 130
69 Comparison of Simulated Methane Production with
Experimental Data by Chang, Recycle .......
70 Comparison of Simulated Volatile Acid Concentration
with Experimental Data by Chang, Recycle, QM = 2.2/day 132
71 Comparison of Simulated Methane Production with
Experimental Data by Chang, Recycle, QM = 2.2/day . . ^33
72 Comparison of Simulated Volatile Acid Concentrations
with Experimental Data by Chang, Recycle, QM = 2.2/day,
SHO = 15,000 mg/L ........... .' 134
-------
73 Comparison of Simulated Methane Production with
Experimental Data by Chang, Recycle, QM = 2.2/day,
SHO = 15,000 mg/L 135
XI
-------
^
G°*
GO*
-------
ACKNOWLEDGEMENTS
This work was performed at the Georgia Institute of Technology under
Cooperative Agreement CR812580010 with the U.S. Environmental Protection
Agency. The authors wish to express their appreciation to Mr. James Brewer
and Mr. Deh-Jeng Jang for their assistance in the development of a database
management system which was used for model verification during the project.
Ms. Vicki Clopton and others also provided invaluable assistance in preparing
the various drafts of the report.
The EPA project monitor was Jonathan G. Herrmann, of the Land Pollution
Control Division, who provided guidance throughout the project and assistance
and guidance on the preparation of the final report.
xiii
-------
SECTION 1
INTRODUCTION
The generation of leachate from a landfill is a complex process
depending not only on the characteristics of the landfilled wastes, but also
on the interaction of the waste with water percolating through the landfill,
operational variables such as waste placement conditions, climatic condi-
tions, the physical design pf the landfill, and the interaction of the
landfilled waste with ground water, as discussed in Appendix A. Leachate
characterization is further complicated by the effects of microbial activity,
which mediates the fate of both hazardous and nonhazardous waste and their
potential for migration from the landfill site.
Leachate characteristics and the rate of leachate generation are
dependent on time and stage of landfill stabilization. Brunner (1979)
investigated the characteristics of leachate generated from municipal waste
disposal and found that the highest concentrations of a large number of
constituents were generated during the early stages of microbially mediated
stabilization. Therefore, if the purpose of any model is to predict changes
in chemical concentrations of leachate for assessment of migration potential
or liner/leachate compatibility, it is necessary to model the various phases
of landfill stabilization.
LANDFILL STABILIZATION
The fate of waste constituents disposed in a landfill can be envisioned
as a partitioning among the solid, vapor or aqueous phases. Various
microbial, chemical, and physical transformations alter the chemical and
physical nature of the waste and transfer it from one phase to another.
Likewise, waste constituents are transported out of the landfill system in
aqueous solution and suspension by washout of leachate and through the
evolution of gases.
Microbially-mediated reactions control the landfill environment for
some time after initial placement of waste and strongly impact the outcome of
other chemical and physical transformations leading to stabilization. Both
aerobic and anaerobic microbial processes occur in a landfill, however, free
oxygen availability is typically limited to early stages of the landfill
stabilization, and is often exhausted prior to the appearance of leachate.
For this reason, anaerobic activity usually establishes and controls leachate
quality during the active life of a landfill.
-------
Anaerobic processes are often described as multistep in nature. The
first step, hydrolysis of complex substrate, results in conversion of solid
organic matter to soluble compounds and reduction in the size of organic
molecules to facilitate transport across microbial cell membranes. The
second step, acidogenesis, is described as one in which acid-producing
bacteria utilize hydrolyzed organic molecules as carbon and energy sources,
producing an abundance of short-chain volatile acids as conversion products.
Carbon dioxide and hydrogen gas may also be evolved during this step. The
products of acidogenesis (volatile acids, CC>2 and H2) are converted to
methane gas by methanogenic bacteria in the final anaerobic process step.
In a steady-state anaerobic treatment process, the three steps occur
simultaneously at a rate controlled by the slowest step on the sequence so
that there is little accumulation of intermediate products over time.
However, in a non-homogeneous, batch-wise system such as a landfill, the
activity of acid-forming and methane-forming bacteria may not be in balance
at any particular location in the landfill at any one time. As a result,
landfills are often characterized by temporal stages caused by the predomi-
nance of different microbial populations at different times in the life of
the landfill. Pohland, et_ al_. , (1985) described five phases through which a
landfill proceeds as it becomes stabilized; initial adjustment, transition,
acid formation, methane fermentation and final maturation. These phases can
be outlined as follows:
Phase I: Initial Adjustment—
Initial waste placement and preliminary moisture accumulation.
Initial subsidence and closure of each landfill area.
Changes in environmental parameters are first detected to reflect
the onset of stabilization processes which are trending in a
logical fashion.
Phase II: Transition—
Field capacity is approached and leachate is formed.
A transition from initial aerobic to facultative anaerobic
microbial stabilization occurs.
The primary electron acceptor shifts from oxygen to nitrates and
sulfates with the displacement of oxygen by carbon dioxide in the
gas.
A trend toward reducing conditions is established.
Measurable intermediates such as the volatile acids first appear
in the leachate.
-------
Phase III: Acid Formation--
Intermediary volatile acids become predominant with the continu-
ing hydrolysis and fermentation of waste and leachate
constituents.
A precipitous decrease in pH occurs with a concomitant
mobilization and possible complexation of metal species.
Nutrients such as nitrogen and phosphorus are released and
utilized in support of the growth of biomass commensurate with
the prevailing substrate conversion rates.
Hydrogen gas may be detected and may affect the nature and type of
intermediary products being formed.
Phase IV: Methane Fermentation--
Intermediary products appearing during the acid formation phase
are converted to methane and excess carbon dioxide.
The pH is elevated from a buffer level controlled by the volatile
acids to one characteristic of the bicarbonate buffering system.
Oxidation-reduction potentials are at their lowest values;
sulfates and nitrates have been reduced to sulfides and ammonia.
Nutrients continue to be consumed.
Complexation and precipitation of metal species with sulfides
and organic ligands proceed.
Leachate organic strength is dramatically decreased in
correspondence with increases in gas production.
Phase V: Final Maturation--
Relative dormancy following active biological stabilization of the
readily available organic constituents in the waste and leachate.
Nutrients may become limiting.
Measurable gas production all but ceases.
Typical subsurface environmental conditions become reinstated.
Oxygen and oxidized species may slowly reappear with a
corresponding increase in oxidation-reduction potential.
More microbially resistant organic materials may continue to be
slowly degraded with the possible production of humic-like
substances capable of complexing and re-mobilizing heavy metals.
-------
Thus, microbially-mediated reactions accomplish first the transformation
and solubilization of waste material into the aqueous solution phase.
Further transformation into volatile acids followed by conversion to methane
result in transfer of conversion products to the vapor phase. The relative
rates of these transformations combined with the rate of moisture infiltra-
tion determine the concentration and mass flux of the various biochemical
intermediates and products in the leachate.
Superimposed upon the microbial transformations are solubility,
speciation, oxidation-reduction and vapor-liquid equilibria of both inorganic
and organic chemicals. These equilibria result in partitioning of chemical
components among solid, aqueous and vapor phases. Changes in these
equilibria as landfill stabilization progresses are evidenced by the decrease
in pH during the acid formation phase and subsequent increase in pH as
methane fermentation is established, mobilization of metals during the acid
formation phase and then precipitation principally with sulfides during
methane fermentation, and changes in oxidation-reduction potential from
positive to negative.
Sorption, the partitioning of organic and inorganic chemicals between
solution and solid surfaces, is a phenomenon not well understood in a
landfill setting. Many factors can affect the sorption of a substrate, such
as:
temperature;
pH and ionic strength for substances that ionize;
the presence of other sorptive substances which 'can compete for
the available sites; and,
particle size distribution and available surface area.
With regard to adsorption, current prediction is largely limited to single
solute systems and is based on experimentally determined isotherms. Most
information concerning adsorption in natural systems is limited to soil
surfaces, and little is known about the adsorptive capacity of municipal
solid waste.
One final transformation to be considered, particularly for toxic
organic compounds which may be microbially degraded at relatively slow rates
is chemical degradation. Although the mechanism for chemical degradation
depends on the specific organic chemical and environmental conditions, in an
aqueous environment away from light and oxygen, hydrolysis is typically the
dominant mechanism (Verschuren, 1983). Chemical-specific hydrolysis rate
constants are needed to predict the chemical degradation of organic chemicals
in the landfill environment.
One of the major difficulties in describing the fate of waste disposed
in a landfill stems from uncertainties in the relative impacts of the various
transformation and partitioning processes acting on the waste, leachate and
gas. Leachate and gas composition and quantity data reveal the net results
-------
of the various transformation and partitioning processes, but do not
necessarily indicate the path taken. Furthermore, data are seldom available
and sufficient to describe the composition of the solid phase. Thus, it is
difficult to assess the relative impacts of adsorption, complexation and
precipitation processes, since they all result in transfer of a component
from the aqueous to the solid phase or vice versa. Likewise, assessment of
the relative effects of microbial degradation, chemical conversion and
sorption on the fate of organic components disposed in a landfill is also
difficult.
-------
SECTION 2
CONCLUSIONS
A useful interpretation of leachate generation and leachate character-
istics is a highly complex and difficult undertaking, primarily because
landfill disposal of waste is site-specific. The local hydrogeology and
landfill design and operation all influence the potential for infiltration
and percolation and the rate of saturation of the waste materials. There-
fore, leachate characteristics and opportunity for migration will vary in
accordance with these factors.
Numerous evaluations and simulations of landfill performance have been
conducted to characterize leachate generation patterns. Some of the data
from these studies have been used to develop numerical predictive models
descriptive of leachate generation as a function of time. Based on a review
and further development of these models, the following conclusions are
presented:
1) Data descriptive of microbial mediation processes of landfill
stabilization are necessary components of an effective numerical
landfill simulation model and are influenced by waste type,
availability of nutrients, moisture content, and biological or
physical/chemical conversion.
2) Comprehensive analysis of waste characteristics and waste
constituent distribution must be available as input data for
predicting leachate constituent concentration changes as a function
of time, to indicate substrates susceptible to conversion, to help
assess any delays in the progress of landfill stabilization, and to
appropriately distinguish microbial mediation from simple washout.
3) There are no established techniques to yield data sufficient to
accurately predict hydraulic conductivity at landfills. Data
from simulation studies performed to date are presumptive of flow
conditions and generally ignore non-homogeneity of the leachine
matrix.
A) The GTLEACH-I model has been developed to simulate the microbiall
mediated processes of landfill stabilization in terms of hydroly-
sis of waste substrate, acid formation and methane fermentation
5) GTLEACH-I provides reasonable prediction of volatile acids and
generation during landfill stabilization and may be expanded to
-------
predict variations of pH, pE, and ORP as functions of time. These
latter relationships again require a comprehensive database which
must include acid-base and oxidation-reduction equilibria not
usually present in compilations of existing landfill analyses.
6) Calibration of the GTLEACH-I model has been limited by a general
lack of substrate specificity, the uncertainty of flow distribu-
tion and short-circuiting particularly during the early stages
of landfill stabilization, the possibility of retardation or
inhibition of microbial mediation, the influences of population
dynamics as substrate conversion proceeds and engulfs the waste
mass, and the potential for containment or release of the liquid
and gas transport phases.
-------
V'
* «i
tf
uD
^>*
? <®a , ^
'e Xoc
IP O"^ c
^6
-------
SECTION 4
REVIEW OF PREVIOUS MECHANISTIC MODELS OF LANDFILL STABILIZATION
A brief review of previous work on mechanistic models of leachate
quantity and quality during stabilization of municipal solid waste landfills
follows. The purpose of this review is to examine previous approaches and
assumptions in modeling landfill stabilization and to assess the degree of
success achieved in simulating experimental data.
The transfer of solid organic matter into a soluble form was assumed by
Straub and Lynch (1982) to follow a pseudo first-order kinetic reaction of
the form:
q
R = f- b (C - C) (1)
g S max
where;
RK = net rate of transfer of COD from solids to landfill moisture,
S = mass of solid COD/unit bulk volume of landfill,
SQ = initial or ultimate S,
m = dimensionless constant,
b = solubilization rate constant,
Cmax = saturation COD concentration in leachate, and
C = COD concentration in landfill moisture.
Thus, solubilization of solid substrates was treated as an equilibrium
process in which the rate of mass transfer decreased as the initial source
was depleted. Microbially-mediated aspects of solubilization were not
included in Equation 1 implicitly.
-------
Organic substrate (COD) utilization or removal by suspended micro
organisms was described as following the Monod kinetic model:
-k X S
Ru
m
where ;
RJJ = substrate utilization rate,
S = substrate concentration in leachate,
X = microbial cell concentration in leachate,
kQ = specific substrate utilization constant,
kjjj = half velocity constant, and
Y = microbial cell growth yield coefficient.
Microbial cell growth rate was then given by:
Rc = - Y RU - kd X (3)
where ;
k^ = endogenous death rate of microbial cells, and
RU and Y were defined previously.
Straub and Lynch (1982) applied the preceding mass transfer equations
to three models of moisture and contaminant transport: a single well-mixed
reactor model, a vertically cascaded reactors model, and a model of
unsaturated flow and contaminant transport through porous media. In the
first model, a landfill was considered to be a single well-mixed reactor
while the vertically-cascaded reactors model was composed of a number of'
these well mixed reactors connected in series. The vertically-cascaded
reactors model was thought to model vertical transport processes more
realistically than the single, well-mixed reactor. The moisture content f
the wastes in both models was assumed to be at field capacity. Flow int t-h
system, flow between reactors and effluent leachate flow were set equal
any instant in time for the vertically- cascaded model.
The model of unsaturated flow and contaminant transport through
media was based on an explicit finite-difference representation of the
vertical transport equations: ;
10
-------
_6_/ 69\_
6z\ 6zy
and,
|C = _q 6C + 1 6/E6C\ R + rC (5)
6t 6 6z 9 6z \ 6z/ 8
where;
9 = moisture content,
C = contaminant concentration in leachate,
t = time,
z = position (depth) in landfill,
k(9) = unsaturated hydraulic conductivity,
D(9) = unsaturated moisture diffusivity,
r = moisture sink,
q = moisture flux,
E = contaminant diffusion/dispersion coefficient, and
R = net contaminant generation rate.
Straub and Lynch (1982) applied the three transport models to a
simulation of a single-pass experimental landfill cell using data reported by
Pohland (1975). The ultimate source of leachable organics was estimated as
12% of the dry refuse mass. The value of m was assumed to be 2 as a
reflection of the decreasing availability of easily leachable organic
material as leaching proceeded, and Cmax was inferred from maximum COD level
typically observed experimentally (40,000 - 50,000 mg/L). The solubilization
rate constant and Monod parameter were estimated or adjusted to fit the test
data, as were the parameters for the vertical transport equations. The
vertically-cascaded reactors model and the model of unsaturated flow and
contaminant transport through porous media model exhibited very similar
behavior and fit the experimental data reasonably well as shown in Figure 1.
The single-reactor model showed somewhat poorer fit than the other two
models.
Edrogan (1982) presented two approaches for modeling leachate generation
and transport through a sanitary landfill, which were essentially the same as
those of Straub and Lynch (1982) • The first approach modeled the landfill
11
-------
20000 r
—— MEASURED
SIMULATED
200
400 600
TIME (DAYS)
800
1000
UNSATURATED TRANSPORT MODEL SIMULATION OF POHLAND DATA (1975)
(Straub and Lynch, 1982->
Q
O
O
20000
15000
10000
5000
MEASURED
SIMULATED
250 500 750
TIME (DAYS)
1000
FOUR-REACTOR MODEL SIMULATION OF POHLAND DATA (1975)
(Straub and Lynch, 1982)
FIGURE 1. Comparison of Unsaturated Flow Model and Four-Reactor
Simulations.
12
-------
as a number of continuously-stirred tank reactors (CSTR) in series, while the
second approach employed the concepts of flow through porous media. Both
approaches were applied for inorganic and organic pollutants and allowed for
solubilization of organics and inorganic matter by first-order kinetics;
removal of organic matter by microbial action of a single group of suspended
organisms; and accumulation of biomass by growth. The following assumptions
were made in developing the models:
medium is unsaturated;
moisture flow is unidirectional and moved vertically downward
through the landfill;
inorganic and organic contaminants are generated from the solid
phase by first-order reaction;
organic substances are utilized by microorganisms as a food and
energy source; and,
the roles of aerobic and anaerobic microorganisms were simulated
using Monod kinetic formulation.
The model was not applied to the prediction of experimental data so no
conclusion could be drawn regarding its usefulness.
Bernardes (1984) developed a model to describe fixation of heavy metals
in the codisposal of industrial sludges with domestic solid waste. Three
sub-models were presented: a biological model of the decomposition of
domestic solid waste over time; a hydrologic model for unsaturated flow as
applied to landfills; and, a chemical equilibrium model to predict metal
speciation with age of the landfill. Biodegradation of solid waste was
modeled in three steps. First, four major components consisting of cellu-
lose, protein, fat, and starch were hydrolyzed to soluble complex organic
matter, measured as COD. Hydrolysis was described as an enzyme-catalyzed
reaction which, under substrate to microorganism ratios, was a zero-order
kinetic expression:
= -K (V x 0) (6)
or,
M/M = 1 - K't (7)
o
where;
t = time (years),
13
-------
M = mass of solid present (kg COD),
MQ = initial mass of solid present (kg COD),
V = bulk volume of solid waste,
9 = moisture content, and
K1 = hydrolysis constant (year "*•).
Hydrolysis of each substrate was mediated by a specific group of microorgan-
isms, each with a specific hydrolysis constant. Hydrolysis was followed by
acidogenic and methanogenic phases. No further explanation of the biological
decay model was given.
Outputs from the biological model developed by Bernardes -(1984) were:
cellulose, protein and fat content in the solid phase as a function of time;
glucose, amino acid, propionic acid, butyric acid, hydrogen gas, soluble fat,
sulfide and sulfate concentrations in solution as a function of time; and,
concentration of five raicrobial populations (acid-formers metabolizing
glucose, acid-formers metabolizing araino acid, methanogens metabolizing
acetic acid, methanogens metabolizing propionic acid, methanogens metabo-
lizing hydrogen gas) as a function of time. Inputs to the biological model
included: coefficients for each of the enzyme-catalyzed reactions, density
of solid waste, initial moisture content, actual moisture content, desired
time step, porosity, molar volume of gas, and initial values of total C02
plus CH^ pressure.
Bernardes (1984) employed a computer model developed by McDuff and Morel
(1973), REDEQL 2, to evaluate chemical equilibria described by Sturam and
Morgan (1981). The user was required to identify all species present, the
total concentration of each element present, and either the pH or the total
equilibria, correct for ionic strength and temperature, and apply the proton
condition to avoid round-off errors at high ionic strengths. REDEQL 2 would
then output the fraction of the total concentration of a metal in solution in
each form and the fraction which precipitated.
Based on data gathered by Pohland (1980), Bernardes (1984) assumed that:
over the pH range encountered for leachates, the dominant carbonate
species are 1^03* and HC03";
the pH changes slowly enough to assume a series of metastable
equilibria;
an important fraction of the alkalinity is due to acetic acid'
and, '
initially the dissolved C02 and H2C03*, is in equilibrium with
HC03, but not with C02. Not until gas production is well
established is the H2C03*/C02 equilibrium attained.
14
-------
Mass balances for HCO^ and C02 were made taking into account biological
sources and gas transport. The pH was calculated from the mass balance and
equilibria expressions. The Davies expression (Stumm and Morgan, 198l) f°r
ion activity coefficients was used to correct the equilibrium constants for
ionic strength. The pE based on oxidation of glucose to acetic acid and CC>2
to methane was also calculated.
In developing the hydrologic model, Bernardes (1984) assumed that:
flow through landfills is characterized as unsaturated, therefore,
water which infiltrates the landfill moves downward due to
capillary action and gravity;
during periods of moisture inflow, the moisture profile moves
downward as a wetting front; and,
during periods with no moisture inflow, the porous media drains.
The wetting front velocity was defined as:
k(9l)-k(62)
dt 8X - 92
where;
dz.f/dt = wetting front velocity,
k(6) = hydraulic conductivity, and
8 = moisture content as a function of depth, z.
The moisture velocity for a drainage wave was:
dz dk //r Q
dt = d9 (f°r 9 = Constant)
An expression to relate unsaturated conductivity to moisture content was also
employed:
0-eo)3
k(0) = ks (E-0 ) (10)
o
where;
k(e) = unsaturated hydraulic conductivity,
15
-------
ks = saturated hydraulic conductivity,
90 = field capacity, and
E = porosity.
In cases with variable infiltration rates, where one front could overlap
another, the front with higher moisture velocity dominated.
Bernardes (1984) applied his model to simulate experimental results
obtained by Pohland (1979) for pH and pE of leachate; by Swartzbaugh (1979)
for metal speciation; and by Fungaroli (1979) and Swartzbaugh (1979) for
leachate flow. However, it is difficult to make definitive conclusions as to
the success of the model due to the limited number of data points that were
presented for comparison.
Review of the important efforts in development of mechanistic models of
landfill stabilization revealed two dramatically different approaches to
modeling microbial processes occurring in a landfill. On one hand, Straub
and Lynch (1982) employed a relatively simplistic model of landfill
stabilization involving release of COD through microbial metabolism. At the
other extreme, Bernardes (1984) developed a very complicated series of
microbial processes involving six groups of microorganisms following Monod
kinetics.
In selecting a model, one must consider the database available to
support and test the model. A highly sophisticated model is useless without
adequate data. The most commonly available leachate organic strength
measurements are COD and BODg. Less frequently gathered are total volatile
acid concentration. Finally, only studies such as those conducted by Pohland
and his co-workers (Chang, 1982; Schaffer, 1986 Yari, 1986) include analyses
of the individual volatile acids as well as many of the other indicator
parameters discussed previously.
The overwhelming requirements for Monod parameters and detailed analyses
of the solid waste implicit in the model of Bernardes (1984) are impediments
considering the current database available for landfill stabilization.
However, the limitation of the model of Straub and Lynch (1982 ) is that it
predicts only COD as a measure of leachate quality and fails to describe the
phases of landfill stabilization with important fluctuations in pH and ORP
and resulting implications for chemical equilibria. If a model of landfill
stabilization is to be eventually applied to the prediction of the rate of
hazardous as well as nonhazardous organic and inorganic constituents disposed
in a landfill, a model which includes elements such as fluctuations in pE and
ORP will be essential. One must also bear in mind, however, that landfills
are extremely complicated nonhomogeneous systems of microbial and chemical
reactions and equilibria for which an exact model would present a
considerable challenge and for which it would be even more difficult to
obtain the data necessary for verification.
Based on the preceding review, a model somewhere between the two
extremes and capable of utilizing available landfill or pilot-scale exr>e '
16
-------
mental data, yet sufficiently mechanistic that it is also capable of
simulating the observed trends in landfill stabilization, would be the most
desirable type of model to develop. Therefore, it was with this goal in mind
that the model described herein, GTLEACH-I was conceived.
17
-------
SECTION 5
GTLEACH-I MODEL DEVELOPMENT
The GTLEACH-I model is a component model composed of a hydrogeologic
module and biologic module. The modules comprising the GTLEACH-I model are
described below.
THE HYDROGEOLOGIC MODULE
The rate and quantity of water flowing through waste materials can be
correlated to the quantity and, to some extent, the characteristics of
leachate produced in landfills. This is not to imply that a simple washout
model adequately predicts leachate characteristics. Stated very simply, the
characteristics of the leachate depend primarily on the adsorption, solution,
movement and hydrolysis/conversion of waste materials described by gravity
flow models and the biologically-mediated stabilization processes described
by biological waste conversion models.
The latter biologically-mediated stabilization processes are dependent
on a large number of parameters, among which are the water content, porosity,
and distribution of water in the waste, the contact time between water
percolating through the landfill and the waste materials, and changes in
waste surface area, composition and porosity with time. These parameters
either affect or are dependent upon the quantity and rate of water flowing
into and through the waste materials.
The hydrogeologic model must account for all flow into and out of the
landfill and its constituents cells. Therefore, it is proposed that a quasi-
two dimensional deterministic model, similar to the HELP (Schroeder et al.,
1983) model, be incorporated into the hydrogeologic module of GTLEACH-I.The
HELP model performs a sequential time-based water budget for a landfill cell.
The water budget is based on soil and waste characteristics, climatological
data and landfill design parameters. The model can be described as a
component, semi-empirical numerical model which evaluates the effects and
interaction of runoff, evapotranspiration, percolation and lateral drainage.
The algorithms used in the HELP model are described by Schroeder et al.
(1983) and are not repeated herein.
In order to provide the data required to evaluated the time-dependent
progression of biologically-mediated stabilization, several modifications of
the HELP model are required. First, the flow model utilized in the HELP
model yields unrealistically high flow rates through the waste materials,
Therefore, it was anticipated that the vertical flow model would be replaced
with a different flow model which could provide more realistic flow rates and
18
-------
distribution of moisture content as a function of depth and time in the waste
material. In addition, surface run-on interaction with the site hydrology
and short-circuiting were considered necessary components. The flow models
and additional components are described in the following narrative.
Since the flow through the waste may be viewed as a series of reactors
where the water interacts with the waste materials over a finite period of
time, the mixing regimes within reactors are typically described as falling
between two idealized extremes; plug flow and completely-mixed flow. In an
idealized plug-flow reactor there is no axial mixing, only radial mixing, so
that differential cross-sectional fluid elements moving axially through the
reactor are homogeneous or completely mixed in the radial direction
(perpendicular to the direction of flow). There is no dispersion or mixing
between fluid elements in the axial direction. At the other extreme, an
idealized, continuously stirred tank reactor is instantaneously mixed with
the contents of the reactor and a fluid element leaving the reactor is of
identical composition to the contents of the reactor.
Mixing regimes for real or non-real reactors are typically modeled as
modified versions of one of the two idealized models (Levenspiel, 1972). The
dispersed plug-flow model and the tanks-in-series model, both single-
parameter models, are the two most commonly employed models for non-ideal
flow. The dispersed, plug-flow model or dispersion model possesses a plug-
flow mixing regime upon which is superimposed some degree of backmixing or
intermixing that is independent of position within the reactor. The model
assumes no stagnant pockets or gross short-circuiting of the fluid in the
vessel. As the mixing intensity, the adjustable parameter, is increased, the
mixing regime ranges from plug-flow to completely-mixed flow. Axial mixing
of fluid flow is described by an equation analogous to Pick's Law of
diffusion:
•2C (I1)
^ = D —2
6t 6x
where; D = longitudinal or axial dispersion coefficient,
C = constituent concentration,
t = elapsed time, and
x = distance in direction of flow.
According to Levenspiel (1972), the dispersion model satisfactorily
represents flow which does not deviate greatly from plug-flow. With
increasing axial dispersion, however, it becomes increasingly unlikely that
the assumptions of the dispersion model will be satisfied by the real system.
The other widely-applied model on non-ideal flow is the tanks-in-series
model. In this model the fluid is viewed as flowing through a series of
equal size, continuously-stirred tanks, the adjustable parameter being the
number of tanks in series. As the number of tanks is increased, the mixing
regime varies from a single, completely-mixed tank to plug-flow. Recycle of
effluent can be represented equally well by the dispersion model and the
19 :
-------
tanks-in-series model. "Which approach one takes is simply a matter of
taste, style and mood" (Levenspiel, 1972).
For reactors which cannot be satisfactorily represented by one of the
two single-parameter models, other more complicated multi-parameter models
may be considered. These models are typically formulated by interconnecting
several different regions of plug, mixed, dispersed, and stagnant flow in
various ways. For trickle beds, chromatographs and packed beds of porous
and/or adsorbing solids, a Guassian residence time distribution curve with an
extended tail is observed. This phenomena is explained as hold-up of fluid
by adsorption on the solid surface, trapping of fluid within pores, or hold-
up of fluid in the many stagnant regions formed at the contact points of the
packing within the reactor. In these cases, various multi-parameter models
have been applied when the single-parameter models were deemed inadequate.
One of the simplest and most effective of these models is a modification of
the tanks-in-series model in which each of the tanks is connected to a side
tank or deadwater tank. As the fluid moves through each vessel, part of the
fluid is held up in the side tank. The three adjustable parameters in this
model are the number of stages, the fraction of the total volume which is
stagnant, and the ratio of flow between the deadwater tank and the active
tank to the flow through the entire stage.
In selecting a flow model for a reactor, one would first select a
single-parameter model in order to simplify the procedure of fitting the data
to the models. If the fit of neither of the single-parameter models is
adequate, then a multi-parameter alternative which can accomplish the
satisfactory modeling of the mixing regime should be selected. One further
consideration in selection of a flow model for incorporation into a
predictive landfill model is the need for flexibility in the flow model to
accommodate landfills of varying configuration and composition. This is
particularly true for GTLEACH-I which was to be calibrated against pilot-
scale landfill data in which the depth of the reactor is typically greater
than the width, but was to be applied eventually to full-scale landfills
which are of much greater width than depth. In the former, one expects to
approach a plug-flow mixing regime whereas in the latter, substantial
deviations from plug-flow are expected.
Straub and Lynch (1982) have demonstrated that the dispersed-flow and
tanks-in-series models yield very similar results in simulating contaminant
transport through pilot-scale landfills. At this time, however, there is a
lack of data with which to evaluate the extent of deviation from plug-flow
encountered in full-scale landfills. Due to this uncertainty and the
resulting unreliability of the dispersed flow model for modeling these
systems, the tanks-in-series model was selected over the dispersion model for
modeling the mixing regime in GTLEACH-I. In addition, if ever the tanks-in-
series model was found to be inadequate when applied to full-scale landfill
systems, the tanks-in-series model could be modified to incorporate deadwater
tanks at each stage.
The tanks-in-series approach may be modeled using either a one-
dimensional saturated or unsaturated flow model. The saturated flow model is
based on the Darcy Equation: • ,
20
-------
q = kave i A (12)
where;
q = the flowrate through the waste,
kave = the average hydraulic conductivity of the waste,
i = the hydraulic gradient, dh/dl
h = the head difference across a layer,
1 = the height of the layer, and
A = the cross-sectional area of flow.
In a larger profile, the average hydraulic conductivity of the media may
be approximated using the following equation:
(13)
where;
di = the thickness of the ith waste layer,
kave = the hydraulic conductivity of the ith waste layer, and
N = the number of layers of waste.
A FORTRAN program, SAT, based on the above algorithm, was written and
used to evaluate the impact of flow rate on biologically mediated
stabilization. In addition to the saturated flow model, short-circuiting,
prior to reaching field capacity, was modeled assuming a low initial
effective porosity and a corresponding low initial cross-sectional area of
the flow path. The equation used to evaluate the flow rate through the waste
attributed to short-circuiting was evaluated using the following algorithm:
i A'
21
k
ave
N
Z di
1=1
N
i-i '
( di
k
\ ave
-------
where;
qs = flow rate of water which percolates through the
landfill prior to reaching field capacity
kwas = hydraulic conductivity of the waste,
i = hydraulic gradient which is assumed to be
equal to 1 for short-circuiting, and
A" = initial cross-sectional area of the flowpath.
The initial cross-sectional area of the flowpath may be evaluated using
the following equation:
A1 = ne A (15)
where;
A = the cross-sectional area of the landfill or
lysimeter, and
ne = the initial effective porosity of the waste which
is likely in the range of 0.01 to 0.1.
A listing of program SAT is presented in Appendix B.
As discussed subsequently, the saturated flow model did not adequately
model the flow through the waste materials. Results of the modeling using
the biological module of GTLEACH-I indicated that a partially saturated flow
model must be used to properly characterize flow through the waste materials.
As stated previously, the flow of water through partially saturated
soils may be described using the Darcy Equation. For flow in the vertical
direction, the velocity may be calculated as:
where;
v
z
velocity in the vertical direction,
= total potential or hydraulic head,
z = distance over which the head is dissipated, and
k = hydraulic conductivity of the porous media.
22
-------
In a saturated porous media, the hydraulic conductivity, k, is primarily
a function of the media structure, void ratio, state of stress, stress
history, the boundary condition affecting flow and the properties of the
permeant (water). In a partially saturated media, the hydraulic conductivity
is also a function of the volumetric water content. Therefore, in a
homogeneous, isotropic, partially saturated media, the Darcy Equation
becomes :
v = -k(6) (17)
z oz
The total potential in a partially saturated media consists of a
pressure head or matric potential, 4s and an elevation head, z, assuming the
velocity head, v^/2g, can be neglected. For unsaturated flow, the pressure
head, ty, is always less than zero, thereby indicating suction in the media
pores. If the interconnected media pores are modeled as capillary tubes, the
suction which develops in the pores may be described as surface tension
forces developing at the air /water interface of the capillary tube.
The general equation for one-dimensional flow through a partially
saturated, rigid porous medium is:
--<*> (18)
If it is assumed that
The media is homogeneous and isotropic;
Flow is one-dimensional with no lateral dispersion;
The fluid properties are constant with time;
The media porosity or void ratio is constant with time
(no sett lement ) ;
The fluid density is constant through the media profile;
The fluid in the media pores does not freeze;
The air phase remains at atmospheric pressure; and
There are no thennoosmotic or osmotic gradients affecting
flow,
then .the general equation for partially saturated flow becomes (Richards,
1931):
Since the volumetric water content, and hence the partially saturated
hydraulic conductivity are a function of the depth, the general equation may
be expanded to the Richard's Equation:
23
-------
_ 6k(9)
' T — " — - — -- -
6t
The specific moisture capacity, C(iji), is equal to the slope of the media
moisture/matric potential curve, or:
= — (21)
6iC
Therefore, the Richard's Equation may be modified as follows:
_, .. 6 ill 6 i,/n\ &ty\ 6k(6) fT)\
C(40 -r1- = -?— I k(9) -r2- I •= \t-t-)
The first term in the equation for partially saturated flow through a
porous media describes the distribution of water content as a function of
time or change in media storage. The second term describes the flux due to
hydraulic gradients and the third term describes the flux due to
gravitational potential.
Under saturated conditions, the total head is equal to the pressure head
and the hydraulic conductivity is a constant for the porous media and fluid.
The unsaturated flow conditions, Equation (22) reduces to:
69 6
^ = k ~z = Vz
which is the Darcy Equation for saturated flow.
Properties of a Homogeneous, Isotropic. Partially Saturated. Porous Media
The hydraulic conductivity and specific moisture capacity of
homogeneous, isotropic, porous media vary as a function of the volumetric
water content, 9, and the pressure head of matrix potential, i|i. Due to the
nonlinearity of these functions, a number of numerical approximations have
been developed based on curve fitting techniques. These relationships tend
then to be historetic, in that the relationship between volumetric water
content and pressure head is dependent on whether the porous media is
draining or wetting. The equation of volumetric water content for drainage
conditions developed by Brooks and Coray (1964) was:
24
-------
function:
6(40
where;
8r = the residual water content,
'l^b = the air entry pressure (saturation),
n = the soil porosity, and ,
X = the pore size distribution coefficient.
The specific moisture capacity, C(iji) = 66/6^ , is a discontinuous
• -i rtn •
(24)
C(40
(25)
-\
King (1965) developed a hyperbolic drainage model:
9(1(0 = (n6)
cosh
cosh
[fe)'«]
I*)'"]
- Y
+ Y
(26)
where: 5, ty, p, e, and Y are curve fitting parameters.
Rogowski (1971) used a logarithmic relationship for volumetric water
content as a function of the pressure head:
9010 =
> n
(27)
n+ot
where;
a =
(915-n)
In
(28)
25
-------
fy. r = 1. 5 x 10"1 cm,
J. - = mo±^z :-e content when ty=tl), r .
J-J 15
Clapp and Hornberger (1978) developed a mathematical model for the
wetting of soil using a power law and a portion pf a parabola. For a degree
of saturation < 92% the pressure head is given by:
4> = 4^ W~ (W< 0.92)
where;
Q ^
W = the soil wetness = Jr!_
e- 100
8S = saturated volumetric water content,
ips = saturated pressure head, and
b = an empirical constant.
When degree of saturation is between 92% and 100% the pressure head is
given by:
t|)=-m (W-n)(W-l) (0.92 W ^ I)
where;
^. ij'. b
m = ^ - I.T 11 _I.T '^ (31)
n - 2Wi- inr~lj and
i(i£ = the pressure head at W = 0.92.
Milly and Eagleson (1980) fit the nonlinear relationship between the
volumetric water content and the log of the pressure, pF = log (-i}i), with
three line segments as shown in Figure 2. The equation given for the
volumetric content is:
26
-------
IB.
ni
ni
u. ">.
0. -
CLAY
— Sflno
I I I
5 .18 .15 .28 .25 .30 .35 .40 .45 .59
MOISTURE CONTENT
b.
.a
3.a
FIGURE 2. Sandy and clayey soil moisture characteristics computed
from a continuous functional relationship, a. moisture
content and b. log of specific moisture capacity (After
Johnson et al., 1983)
27
-------
9(pF) -^ Inj exp[M[a1-s1(pF)]J + exp [ M[a2 - s2(pF)]] [
- -, Inj exp^M1 [a2-s2(PF)j] + [exp [M'9j]f + eu (33)
where; M and M1 are constants which control the curvature of the joining
segments.
For the linear portion of the capillary range between pFm^n and pFQ,
Equation 33 reduces to:
a2- s2 log (H|0, (34)
and the specific moisture capacity is given by
The specific moisture capacity for all values of 41 is continuous and is given
by:
-1
1 [-pqH,)qi~1-pq(-)q2~1l
J [ Plql ~P2q2 * J
r q3 v1!
-(P3) (-*) (36)
where;
Pi = eMal, (37)
P2 = «Ma2' (38)
P3 = eM'a2, (39)
PA = eM'u- i (40)
-------
q2 = -Ms2 log(e), and (42)
q3 = -M's2 log(e). (43)
These relationships for typical sand and clay soils are shown in Figure 3.
Partially Saturated Hydraulic Conductivity
The relationship between k(ijj) and 9 for a homogeneous, isotropic porous
media is nonlinear, as shown in Figure 4.
Several numerical models have been developed for partially saturated
hydraulic conductivity:
Logarithmic Model (Gardner, 1952):
k(40 = - - - — (44)
(b+(-<0 )
where; a,b and m are curve fitting constants.
Exponential Model (Gardner, 1958):
kW = k exp(-a40 (45)
s
where; ks = the saturated hydraulic conductivity.
Power Law Model (Brooks and Corey, 1964):
Power Law Model (Campbell, 1974):
k = w2b-3 (47)
where ;
29
-------
Saturation^
Range
PFmin
Capillary
Range
Adsorption
Range
Figure 3. Specific Moisture Content of Soils (After Johnson
1983)
30
-------
\
SAND
00
o
pF = log (-
Figure 4. Schematic of unsaturated hydraulic conductivity for
a sand and clay soil (After Johnson et. §.1. , 1983)
31
-------
k = , and (48)
is.
s
b = curve fitting parameter related to
the soil structure and gradation.
Power Law Model (Maulem, 1978):
k(l») = ks(Se)°-015w +3.0 (49)
where;
Se = the effective saturation = —^ (50)
n-9
y:
{
YwVdt a~G (51)
$ = + -
Yw = the unit weight of water.
The equation for w represents work required to drain one pore volume of
a homogeneous, saturated porous media. Maulem (1978) showed that this model
was a significant improvement over previous models for the prediction of
unsaturated hydraulic conductivity as a function of the volumetric water
content.
Applications to Waste Materials
Numerous numerical models have been developed to model one-
dimensional moisture migration in partially saturated soils. These models
typically utilize finite element or finite difference methods for solution.
Johnson, e£ al. (1983) performed a detailed review and comparison of the
following models through a layered soil profile:
UNSAT2 (Neuman, 1973; Neuman, et al., 1975)
Lappela (1981)
Reeves and Duguid (1975)
FEMWATER (Yen and Ward, 1980)
TRVST (Nardsimhan, 1975)
FLUMP (Neuman and Nardsimhan, 1977)
TRANSAT (Pickens, Gillham, and Camran, 1979)
32
-------
These simulations are useful for comparison of the various models. However,
there are several significant differences between models used for evaluation
of flow through partially saturated soils and models which will ultimately be
developed for flow through waste in landfills.
The application of the partially saturated flow equation and specific
moisture capacity and hydraulic conductivity relations to waste materials
typically involves a great deal of guess work. Until field capacity is
reached, the flow regime in a partially saturated waste cell is three-
dimensional. The assumption of one-dimensional flow is a gross over-
simplification of the problem. Moreover, experience indicates that leachate
is generated before field capacity is reached. This leachate is generated
primarily by short-circuiting through the landfill, although gravity flow of
liquid waste materials may also have an impact on initial leachate quantity
and characteristics. Therefore, the numerical model used to model the
generation of leachate should incorporate short-circuiting.
As a further complication, waste materials are utilized or converted
during the stabilization process. The surface area of the waste materials
tend to increase with time and fractions of the waste may be consumed. This
results in changes in the porosity, structure and even the composition of the
waste. The existing partially saturated flow models do not account for
changes in waste properties, whether incurred due to biological or physical
(compression) mediation. These changes may significantly affect the
partially saturated hydraulic conductivity of the waste. Unfortunately,
little data is available to define these changes in either temporal or
spacial dimensions.
BIOLOGICAL MODULE
The biological module of the GTLEACH-I model is a mechanistic three-
step model which extends the two-step model proposed by Straub and Lynch
(1982). It incorporates components from several other models and analyses
within the perspective of the sequences of events constituting the phases of
landfill stabilization. Consequently, a landfill may be considered to be a
fixed-film microbial treatment process operating in a batch-wise
configuration with continuous dilution and washout. The system is a batch
process since the substrate, the degradable fraction of refuse, is depleted
over time. Usually, both reactants and products of landfill stabilization
processes are continuously or semi-continuously diluted and removed from the
landfill in leachate and landfill gas. The microorganisms, as catalysts in
the reaction, occur largely in films on refuse surfaces and also as
aggregates in the interstices of the waste mass. The concentration of
suspended microorganisms in the leachate is expected to be minimal due to the
filtering action of the refuse, although experimental data confirming this
supposition are currently lacking. Based on landfill stabilization trends
observed by Pohland, gJt al- (1985), it is evident that the appearance and
eventual disappearance of volatile acids in leachate are primary indicators
of processes responsible for concurrent changes in COD, pH, and oxidation-
reduction potential within the landfill. Thus, a model which is capable of
simulating the changes in volatile acid concentrations in leachate over time
could ultimately form the foundation for simulation of other indicator
33
-------
parameters such as pH and ORP along with corresponding changes in the
mobility of hazardous constituents disposed in a landfill.
The three-step processes of hydrolysis/solubilization, acid formation
and methane fermentation described previously were logical choices on which
to base the development of GTLEACH-I. Since it has been applied successfully
in the design and operation of other anaerobic processes (Grady and Lim,
1980; Henze and Harremoes, 1983; Mosey, 1983; Lindgren, 1983), it was
reasonable to assume that the three-step process would also be successful in
modeling anaerobic stabilization in landfill systems. Futhermore, such a
mechanistic model could predict the concentration of volatile acid
intermediates as a function of time as well as the rate of conversion of
leachate COD to methane.
Hydrolysis/solubilization is an important process in the degradation of
organic matter, and it is considered by Eastman and Ferguson (1981) to be the
rate-limiting step in the acid phase of anaerobic digestion. The rate of
hydrolysis is affected by many factors including pH, temperature, microbial
biomass and associated substrate, the remaining concentration of particulate
substrate, and hydrolysis product concentration. It is evident that any one
or all of these factors may change with time in a batch-wise operation such
as a landfill.
Eastman and Ferguson (1981) postulated that, at constant temperature and
pH, hydrolysis is approximately first-order with respect to particulate or
solid organic substrate concentration. This approach was used to model
hydrolysis/solubilization in GTLEACH-I as expressed by the equation.
dM - -K M (52)
dt H
where;
M = solid substrate concentration expressed as if
it were suspended in landfill moisture, and
KH = hydrolysis/solubilization rate constant.
While the approximation of first-order kinetics for hydrolysis is reasonable
when considering the acid formation and methane fermentation phases of
landfill stabilization separately, difficulty is expected in modeling the
transition between the two phases when pH and other leachate characteristics
change dramatically. Additionally, as relatively easily hydrolyzed
substrates are exhausted and more refractory substrates begin to be utilized,
a corresponding change in the rate of hydrolysis should be manifest.
Mosey (1983) summarized the mechanisms involved in conversion of glucose
to carbon dioxide and methane as presented in Figure 5. Acid forming
bacteria, the fastest growing of the four depicted groups of bacteria,
ferment glucose to produce a mixture of reduced end products, e.g., acetic,
propionic and butyric acids. Other volatile acids are produced as well in'
-------
FIGURE 5
MICROBIAL ECOLOGY OF THE ANAEROBIC DIGESTION PROCESS
(Mosey. 1983)
ORGANIC MATTER
I AC ID-FORMING BACTERIA!
acetic
— pyruvic
•H, H,
butyric propionic
butyric acid propionic acid
_*
— acetic —-
acetic
[ACETOGENIC BACTERIA!
CHCOOH
[Acetoclastic METHANE BACTERIA)
[H,-utilising METHANE BACTERIAJ
C0
CHt * 2H20
35
-------
the fermentation of soluble organic matter other than glucose. Acetogenic
bacteria convert long-chain volatile fatty acids into acetic acid. The
existence of these acetogenic bacteria has not yet been proven except by
deduction from the inability of any known methanogenic bacteria to metabolize
propionate and butyrate. The growth rates of acetogenic bacteria are
relatively.low in comparison with acid-forming bacteria and can be easily
product-inhibited by the accumulation of dissolved hydrogen gas in the growth
media. An analogous though somewhat more complicated pathway exists for
mixtures of complex soluble substrate expected from the hydrolysis of the
degradable fraction of waste disposed in a landfill.
Production of volatile acids was represented in GTLEACH-I by a single
Monod equation for a generalized group of acid-forming bacteria rather than a
more complicated set of equations for multiple populations of bacteria.
The overall growth rate for this group of bacteria was described by the
equation:
= - KDA XA (53)
dt KA + SH
where;
XA = acidogen concentration (mg/L),
Sjj = hydrolysis/solubilization product concentration (mg/L),
QA = maximum substrate utilization rate for acidogens (days'M
KA = half-velocity constant for acidogensis (mg/L),
YA = yield coefficient for acidogens (mass acidogens produced
per mass substrate removed), and
Kj-jA = decay rate for acidogens.
The major assumption underlying the selection of a single equation to model
behavior of a group of microorganism is that the relative activities of the
various populations of microorganism are constant with respect to each other
so that the net reaction follows Monod kinetics. While the validity of this
assumption may be questioned when applied to landfills, the single-equation
Monod model of acidogenesis was selected to allow application of the model to
the database currently available and to minimize the number of input
parameters to the model.
Methane is produced from acetic acid by acetoclastic bacteria and from
hydrogen and carbon dioxide by hydrogenophilic bacteria. Bacteria producing
methane from hydrogen and carbon dioxide are faster growing than those
metabolizing acetic acid, but because of their mutualism, the rate of methane
production is limited by the slower growing acetophilic bacteria (Henze and
36
-------
Harremoes, 1983). For this reason, a single Monod equation for methanogenic
microbial activity was employed in GTLEACH-I rather than two separate
equations, also serving to minimize the number of input parameters to the
model. The expression for the overall growth rate of the two groups of
methanogenic bacteria was employed in GTLEACH-I was:
dXM YM QM SH XM
— = - KDM*M (54)
dt KM + SA
where;
^A = v°latile acid concentration (mg/L),
Xj»j = methanogen concentration (mg/L),
Qft = maximum substrate utilization rate for acidogens (days'1),
KJ.J = half-velocity constant for methanegens (mg/L),
Yj^ = yield coefficient for methanogens (mass methanogens
produced per mass substrate removed), and
^DM = decay rate for acidogens.
One other aspect of methanogenesis to be considered is that methanogen
growth and concurrent methane production can be suppressed at pH levels below
the range of 6.8 to 7.5. This explains the virtual absence of methane
production observed during the acid phase of landfill stabilization when the
pH of the landfill moisture can be well below 6. That methanogenesis is
established at all is attributed to the heterogenous nature of landfills
which allows methanogens to begin to grow in small pockets of the landfill
not completely saturated with volatile acids. Methanogen growth gradually
spreads out from these pockets until a methanogen population is established
sufficient to reduce the concentrations of volatile acids in the landfill
moisture and increase the pH, thereby allowing unsuppressed methanogen growth
throughout.
As a first attempt at modeling landfill processes, it was decided to
provide for suppression of methanogenesis during the acid phase by simply
delaying initiation of methanogenesis for a specified period of time in
GTLEACH-I. Further work in the area of landfill modeling will require
development of the pH-inhibition term in Equation 54 for methanogenesis which
will allow for methanogen growth at a suppressed rate during periods of low
pH levels in leachate.
Diffusional resistance to reactions in a dense matrix such as a
microbial film or aggregate may have significant influence on reaction
kinetics by reducing the efficiency of the microbial mass (Henze and
Herremoes, 1983). Due to difficulties encountered in evaluating the total
microbial mass, the thickness of biofilms or aggregates, and due to
37
-------
heterogeneities in landfill environments, diffusional resistances were not
explicitly treated in GTLEACH-I. Instead, they were subsumed in the fitting
of the Monod rate constants to experimental data. Deviations of the fitted
kinetic parameters from literature values were expected to reflect not only
diffusional resistances, but also the effects of environmental heterogeneity
and differences in the type of solid substrate degraded.
On the basis of the preceding discussion, the following assumptions were
made in constructing the biological module of the GTLEACH-I model:
Solubilization/hydrolysis of solid substrate to soluble
intermediates were considered a first-order reaction with respect
to mass of solid substrate concentration or surface area.
Production of volatile acids from soluble intermediates by a group
of acidogens was considered a function of hydrolysis product
concentration according to Monod kinetics as well as a first-order
function of acidogen concentration.
Production of methane from volatile acids by methanegens was
considered a function of volatile acid concentration according to
Monod kinetics and also a first-order function with respect to
methanogen concentration.
The mixing regime of a single lift in a landfill was modeled as a
number of equal-sized, completely-mixed reactors in series.
By the time leachate generation begins, field capacity had been
attained and remained constant thereafter.
The flows of leachate into a landfill lift, through each reactor
within the lift, and out of the landfill lift were equal at any
instant in time so that there was no accumulation of moisture in
any portion of the landfill above field capacity. The flow values
provided in GTLEACH-I were constant, average values of daily flow.
(An assessment of the impacts of these flow assumptions is
contained herein. Since GTLEACH-I has not yet been combined with a
flow rate were necessary. When this work is integrated with a
companion flow model, flow and field capacity should be allowed to
vary from day-to-day in accordance with rates of moisture
applications and drainage.)
Solid substrate was also assumed to be fixed within the landfill.
Hydrolysis products and volatile acids existed only as part of the
landfill leachate, i.e., they did not sorb or precipitate, and,
therefore, could be washed completely out of the landfill given
sufficient flow.
Since inhibition or delay in the establishment of methanogenesis by
environmental conditions is a commonly observed phenomenon in
anaerobic processes, a simplistic approach for modeling the
38
-------
phenomenon was taken in GTLEACH-I by providing for a lag time of
specified length before establishment of methanogenesis.
Based on these assumptions, the GTLEACH-I model was constructed of five
differential equations solved simultaneously for each of a series of
continuously-stirred tank reactors. The total volume of moisture or field
capacity contained in the landfill lift or experimental landfill cell was
divided evenly among the tanks-in-series. The rate of change of solid
substrate mass, expressed as concentration in the volume of moisture
contained within a reactor, was a first-order decay as given by Equation 55:
dM
- = -% M (55)
dt
The rate x»f change of acidogen and methanogen masses, also expressed as
concentration in the volume of moisture contained within a reactor, followed
Monod kinetics as given by Equations 56 and 57:
QA SH XA
- K X (56)
dt KA + SH
<*% _ YM QM SH XM
dt KM + SA
- KDM XM (57)
The rate of change of hydrolysis product concentration in a continuously-
stirred tank reactor was a function of four processes: the rate of
production by hydrolysis, the rate of removal by acidogenesis, the rate of
flow of hydrolysis products dissolved in landfill moisture passing into the
reactor from the preceding reactor, and the rate of flow of hydrolysis
products carried out of the reactor and into the next. The differential
equation describing these simultaneous processes is given as:
dSH QA SH XA
= % M - + F(SHO - SH) (58)
dt KA + SH V
Likewise, the rate of change of volatile acid concentration in a
continuously-stirred tank reactor was also a function of four processes: the
rate of production of volatile acids by acidogenesis, the rate of removal by
methanogenesis, the rate of flow of volatile acids dissolved in moisture
passing into the reactor from the preceding reactor, and the rate of flow of
volatile acids carried out of the reactor and into the next. The
differential equation describing these simultaneous processes is given as:
39
-------
QA sH XA QM SA XM F(s - s ) (59)
= - + — SAO SA
dt KA + SH KM + SA v
In Equations 55, 56, 57, 58, and 59:
M = solid substrate concentration as if it were suspended
solids in landfill moisture (mg/L),
KJJ = hydrolysis/solubilization rate constant (days" ),
XA = acidogen concentration as if suspended in landfill moisture
(mg/L),
SH = hydrolysis/solubilization product concentration in landfill
moisture (mg/L),
Q^ = maximum substrate utilization rate for acidogens (days" ),
KA = half-velocity constant for acidogenesis (mg/L),
YA = yield coefficient for acidogens (mass acidogens produced
per mass substrate removed),
KDA = decay rate for acidogens,
SA = volatile acid concentration in landfill moisture (mg/L),
XM = methanogen concentration as if suspended in landfill
moisture (mg/L),
QM = maximum substrate utilization rate for methanogens
(days-1),
KM = half-velocity constant for methanogens (mg/L),
YM = yield coefficient for methanogens (days),
KDM = decay rate for methanogens (days),
F = moisture flow rate through reactor (L/day),
V = volume of reactor (L),
SHQ = influent hydrolysis/solubilization product concentration
(mg/L), and
SAO = influent volatile acid concentration (mg/L).
These five equations were solved simultaneously to give concentrations
of solid substrate, acidogens, methanogens, hydrolysis products, and volatile
acids as a function of time in the moisture contained in each of the reactors
in series. The composition of the moisture in the final reactor in the
40
-------
moisture entering the first reactor in series was equivalent to the leachate
composition exiting the final reactor in the series.
ORGANIZATION OF GTLEACH-I
GTLEACH-I is currently composed of one main program, PROGRAM GTLEACH-I
and five subroutines: INPUT, OUTPUT, FLOW, FCN, and FCNJ. A copy of the
FORTRAN source code for GTLEACH-I is provided in Appendix B.
GTLEACH-I is linked to a single precision IMSL Math/PC library (1984)
which contains DGEAR, a differential equation solving subroutine for initial
value problems composed of systems of differential equations which must be
solved simultaneously. Subroutine FCN is a subroutine called by DGEAR
containing the system of differential equations to be solved. When DGEAR is
called, DGEAR solves the differential equations given in FCN from initial
values taken from the current value of the matrix Y(N) and the independent
variable X. In this application:
X — t ime (days ),
Y(l) = solid substrate concentration (mg/L),
Y(2) = acidogen concentration (mg/L),
Y(3) = methanogen concentration (mg/L),
Y(4) = hydrolysis product concentration (mg/L), and
Y(5) = volatile acid concentration (mg/L).
DGEAR returns the solutions as new values in X and Y(N). Detailed
specifications for FCN and FCNJ (a dummy subroutine in this application) as
well as guidance for the use of DGEAR are given in the DGEAR section of the
IMSL Library Users Manual [IMSL, 1984].
The underlying structure of PROGRAM GTLEACH-I is composed of a pair of
nested Do-loops, (DO 175 1-1, NDAYS) and (DO 150 J=l, NCSTR). The outer
loop, the I loop, calls for the solution of the differential equations in
steps of days, and the inner loop, the J loop, for solution at each CSTR of
the landfill. DGEAR actually employs time-steps many orders of magnitude
smaller than one day in solving the equations, however, the composition of
the flow between CSTR's is only adjusted at the end of each day as controlled
by the I loop. This size time step is adequately small as long as the
solution exhibits the desired degree of smoothness.
GTLEACH-I allows for two modes of landfill operation, single-pass and
recycle, as determined by the state of the variable IRECYC in the set of
statements following the beginning of the I loop. If IRECYC is equal to
zero, then the concentration of hydrolysis products and volatile acids in the
influent to the first CSTR (SHO and SAO found in subroutine FCN) are set
i
41
-------
equal to zero, thereby resulting in a single-pass system. If IRECYC equals
1, then the concentration of SHO and SAO found are the same as in the
effluent from the last CSTR of the previous time steps (Y(4) and Y(5))-
Y(N) is a matrix containing the solutions to the N differential
equations for the most recently solved CSTR at the most recent time step.
The values in Y(N) are overwritten each time DGEAR is called. The matrix
YY(J,K) stores the values of the solutions Y(N) for an entire time step such
that J is the CSTR number and K is the number of the differential equation.
Thus, before DGEAR is called, the initial values of solution matrix Y(N) are
taken as the final values from the previous time-step for the same CSTR in
the do-loop (DO 50 K=1,N). The next statement sets the value of X, the
initial value of time for the calculation corresponding one day from XEND,
the end of the current time-step.
Once DGEAR is called and the solutions returned in Y(N), the influent
concentrations of SHO and SAO for the next CSTR are set to the effluent
values of the current CSTR from the previous time step as stored in YY(J,4)
and YY(J,5). The values of YY(J,K) are overwritten with the current
solutions of Y(N) in the do-loop (DO 100 K=1,N). The variable INDEX is reset
to 1 so that DGEAR may be provided new initial values at the next call.
Subroutine INPUT reads the input file PARAM.DAT which contains the input
parameters and run-specific information. The input format statements are
list-directed so that the input parameters may be entered in any format and
need only be separated by commas. Line 1 of PARAM.DAT should contain the
parameters: N, METH, MITER, INDEX, H, and TOL. These are parameters
required by subroutine DGEAR and are defined in the IMSL Library User's
Manual (IMSL, 1984). These parameters need only be altered if the GTLEACH-I
code is modified. The one parameter of interest in the list is N, the number
of differential equations to be solved by DGEAR.
The second line of PARAM.DAT contains the parameters: TVOL, XMASSO,
SHO, SAO, XAO, and XMO. They are defined as follows:
TVOL = total volume of moisture contained in the landfill at field
capacity,
XMASSO = total mass of food waste or degradable waste initially
disposed in the landfill,
SHO = concentration of soluble substrate in leachate at time
leachate first appears,
SAO = concentration of volatile acid in leachate at time leachate
first appears,
XAO = concentration of acidogens in landfill moisture at time
leachate first appears, and
XMO = concentration of methanogens in landfill moisture at time
leachate first appears.
The third line of PARAM.DAT contains the parameters IOUT and LAG. The
value of IOUT determines whether the output values of leachate strength are
given in concentration (mg/L) or in total mass contained in the leachate for
one day's flow. A value of 0 for IOUT selects concentratibn and a value of 1
42
-------
selects total mass. LAG allows for inhibition of methanogenesis by delaying
the time at which methanogen growth is initiated. LAG should be provided as
the number of days after leachate generation begins at which methanogen
growth begins.
Line A of PARAM.DAT contains the parameters XKA, XKDA, and YA which are the
Monod parameters for acidogens.
XKA = half-velocity constant for acidogens (mg/L),
XKDA = decay rate for acidogens (days'1), and
YA = yield coefficient for acidogens (mass biomass produced per mass
substrate removed).
Line 5 of PARAM.DAT contains the same parameters for methanogens:
XKA = half-velocity constant for -acidogens (mg/L),
XKDA = decay rate for acidogens (days'*), and
YA = yield coefficient for acidogens (mass biomass produced per mass
substrate removed).
Line 6 contains the values for XKH, QA and QM, defined as:
XKH = first-order rate constant for hydrolysis/solubilization
(days'1),
QA = maximum rate constant for acidogenesis (days'1), and
QM = maximum rate constant for methanogenesis (days'1).
Finally, Line 7 contains the run-specific values for NCSTR, NDAYS, and
IRECYC which are:
NCSTR = the number of CSTR's in series,
NDAYS = the number of days after leachate first appears for which
GTLEACH-I should be run, and
IRECYC = U tor single-pass mode, 1 for recycle mode.
Subroutine OUTPUT is called once every five time steps (days) to write the
solution to a file called RESULT.DAT. RESULT.DAT contains 7 columns of
effluent leachate data, namely: time (days) since leachate generation began,
the mass of degradable substrate remaining, the mass of acidogens, the mass
of methanogens, the mass of hydrolysis products, the mass of volatile acids
and the mass of methane produced.
43
-------
SECTION 6
EVALUATION OF GTLEACH-I
The GTLEACH-I model was applied in the simulation of a set of recent
landfill stabilization experiments conducted by Schaffer (1986) and Yari
(1986) which are the latest in a series of simulated landfill experiments by
Pohland and coworkers. The experiments were designed to capture the temporal
processes occurring within a single lift or cell of a landfill.
Two modes of leachate management were employed, single-pass and recycle.
In the single-pass experiment, moisture infiltrated the landfill, percolated
through, and exited from the bottom of the landfill much as in a typical
municipal solid waste landfill. Management of leachate in the recycle mode
involved collecting leachate from the bottom of the landfill
and reapplying it to the surface, allowing leachate to percolate through the
landfill repeatedly. Either mode of operation can be simulated by GTLEACH-I.
The results of fitting GTLEACH-I to the Schaffer (1986) and Yari
(1986) data are discussed subsequently. However, before proceeding to
fitting GTLEACH-I to experimental data, an examination of the various
parameters output by the model for a typical simulation serves to illustrate
the capabilities of GTLEACH-I. The example selected is the simulation of
the Schaffer (1986) and Yari (1986) data which, using only a single
group of parameters, provides the best possible fit to both the single-pass
and recycle data. The only difference between recycle and single-pass
simulation is that flow into the recycle cell has the same composition
as leachate exiting the cell whereas the composition of moisture flowing
into the single-pass cell was that of pure water. This simulation
employs the following input parameters:
Flow, F = 6 L/wk =0.86 L/day,
Moisture content, TVOL = 71 L,
Number of CSTRs in series, NCSTR = 3,
Initial mass of solid substrate, XMASSO = 11 kg,
Initial concentration of hydrolysis products, SHO = 40,000 mg/L,
Initial concentration of volatile acids, SAO = 5,000 mg/L,
Initial concentration of acidogens, XAO = 100 mg/L, and
Initial concentration of methanegens, XMO = 10 mg/L.
Kinetic parameters for the various processes are:
Hydrolysis Acidogenesis Methanogenes i s
KH = .0001/day QA = 3.2/day Q^ = 1.9/day
44
-------
KA = 200 mg/L KM = 500 mg/L
KDA = 0.5/day KDM = 0.02/day
YA = 1.0 YM = 0.02
Methanogen lag = 200 days
Basic criteria for selection of these parameters are discussed subsequently.
The simulations presented here provide an understanding of the general
characteristics of single-pass and recycle simulations, but as will be
demonstrated in the next section, details of the simulations can be modified
by varying the choice of input parameters.
Figures 6 through 12 illustrate graphically the variation in values of
the output parameters with time for the single-pass simulation. Over a
period of nearly 450 days the solid substrate, expressed as concentration in
landfill moisture, decreased from an initial value of 155,000 mg/L to a value
of 148,000 mg/L due to hydrolysis/solubilization. This represented a
decrease in available solid substrate of approximately five percent as shown
in Figure 6. It was difficult to assess whether or not this was a realistic
utilization of solid substrate because of the difficulty in measuring solid
substrate experimentally. In this work solid substrate was taken as the food
fraction of municipal solid waste, estimated to be 20% by weight
(Tchobanoglous et al., 1977). It is. unlikely that the entire food fraction
of municipal solid waste was degradable by anaerobic processes, however, a
better approximation was unavailable.
The mass of acidogens, expressed as concentration in landfill moisture,
reached a maximum value of more than 5,000 mg/L during the first week of
leachate generation (Figure 7). The concentration of acidogens quickly
decreased over the next two weeks, stabilizing at a value of approximately
300 mg/L. For the remainder of the simulation, the acidogen concentration
profile exhibited a slight but steady decrease. Comparison of Figure 8 with
Figure 7 indicated that trends in hydrolysis product concentration decreased
from an initial value of 40,000 mg/L to a value of 37 mg/L. This value was
maintained for the remainder of the simulation, indicating that the rate of
appearance of hydrolysis products by hydrolysis equalled the rate of removal
by acidogenesis and washout.
Methanogen growth was initiated on Day 200 in accordance with the
imposed lag. As illustrated in Figure 9, the initial concentration of 10
mg/L was sustained for the first 21 days then steadily decreased to
approximately 2 mg/L by Day 446.
Volatile acid concentration in the leachate (Figure 10) quickly
increased from the initial value of 5,000 mg/L to 17,500 mg/L within the
first week of leachate generation. The value subsequently decreased entirely
under the influence of washout until methanogenesis initiated on Day 200 and
further increased the rate of removal of volatile acids. The concentration
of volatile acids had decreased to 200 mg/L by Day 260, then gradually
trended upward to a value of 300 mg/L by Day 446. ,
45 :
-------
FIGURE 6
, .. .
(wg/L)
SOLID SUBSTRATE CONCENTRATION IN LANDFILL MOISTURE,
SINGLE-PASS SIMULATION
160000
140000
120000
100000
80000
D0000 "
A QQQQ
40000 -
oaaan
ZUaod
a .
— -1-
56
108
158
zee
25B
300
350
400
4E.0
TIME SINCE LEACHATE GEMEFtATION BIIGAN (DAYS)'
-------
FIGURE 7
XA
(ng/L)
6088
5088
4880
3880
iACIDOGEN CONCENTRATION IN LANDFILL MOISTURE,
SINGLE-PASS SIMULATIOIH
1168
156
20B
250
380
350
408
450
TIME SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 8
HYDROLYSIS PRODUCT CONCENTRATION IN LEACHATE,
SINGLE-PASS SIMULATION
4UUUH -T
irnnn I
JJOUo 1
qaaaa 1
JDDOD 1
9rnnfl U
£DUDU 1
cu
7aiaaa .
, _ « ^DuDa
(ng/L)
icaaa .
1DDUD
iaaaa .
IDDUD
SflBB -
R -
0 58 IBB 158 288 258 380 358
TINE SINCE LEACHATE GENERATION BEGAN (DAVS)
488
450
-------
FIGURE 9
14
12
18
XM 8
(ng/L)
6
4
2
8
B
METHAN'OGEN CONCENTRATION IN LANDFILL MOISTURE,
SINGLE-PASS SIMULATION
1100
280
256
380
358
4130
458
TIME SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 10
VOLATILE ACID CONCENTRATION IN LANDFILL LEiACHATE,
SINGLE-PASS SIMULATION
IbUOO
1ARRR -
J.DOOD
i/innn .
14000
mnnn J
SA imm
° ("9/L) ORRR
DODO
ARRR •
4RRR -
7RRR -
8
r\
i \
N
V
\
\
\
X
"^^•^1
fc-l
8
50
188
15B
280
258
3BH
35B
488
458
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 11
66066
46686
COD
(mg/L)
26888
18868
8
e
COB CONCENTRATION IN LANDFILL LEA'CHATE,
SINGLE-PASS SIMULATION
166 156 266 256 38H 356
TIME SINCE LEACHATE GENERATION BEG AIM (DAVS)
468
450
-------
FIGURE 12
METHANE
(L/day)
METHANE PRODUCTION,
SINGLE-PASS SIMULATION
21513
300
35H
400
456
TIME SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
The concentration profile of COD in leachate (Figure 11) was simply
constructed as the sum of hydrolysis product concentration and volatile acid
COD concentration. The shape of the COD concentration profile was similar to
that of volatile acid concentration with the addition of an initial peak
caused by the peak in hydrolysis product concentration.
Methane production (Figure 12) was initiated at a rate of 2 L/day,
corresponding with the initial 10 mg/L concentration of niethanogens. Methane
production decreased as the supply of volatile acids was depleted, falling to
a production level of half the initial value within 50 days.
Figures 1" through 19 illustrate the variation in values of the output
parameters with time for the recycle simulation. Solid substrate removal in
the recycle simulation exhibited the same five percent decrease over 450 days
as the single-pass simulation (Figure 13). As discussed earlier, it was
difficult to determine whether simulated levels of solid substrate were
representative of experimental values because of a lack of data. However,
because the recycle system was a conservative system (i.e., constituents were
not lost by washout) it was likely that the actual mass of solid substrate
removed in the simulation was a good approximation of the amount removed
experimentally during the acid phase if the amount of volatile acids produced
in the simulation closely approximated the amount of volatile acids produced
experimentally. In other words, if all hydrolysis products produced from
solid substrate were converted to volatile acids, and the simulation of
volatile acid concentration closely approximated the experimental volatile
acids by methanogenesis to complicate the mass balance), then it can be
deduced that the amount of solid substrate removed by hydrolysis/sol-
ubilization in the simulation was representative of experimental removal of
solid substrate. The terms "mass of volatile acid" and "volatile acid
concentration" can be used interchangeably only when flows are accurately
known so that the conversion from concentration to mass is accurate.
The acidogen concentration profile for the recycle simulation was
virtually identical to the single-pass simulation (Figure 14) as was the
hydrolysis product concentration profile (Figure 15). This similarity
indicated that the initial concentration of hydrolysis products was determi-
ned by the peak in acidogen concentration and that initial removal of
hydrolysis products was largely due to acidogenesis rather than to washout in
the single-pass simulation.
Differences between the recycle and single-pass simulations were largely
manifest through methanogen concentration, volatile acid concentration, and
methane production profiles. Methanogen concentration for the recycle
simulation (Figure 16) initiated at level of 10 mg/L on Day 200, increased
to a peak value of 670 mg/L on Day 366, then decreased thereafter. This was
in sharp contrast to the single-pass simulation which never attained a
methanogen concentration higher than the initial value. These differences
can be attributed to the fact that most of the volatile acids in the single-
pass simulation were removed by washout and were thus unavailable for
53
-------
-------
FIGURE 14
ACIDOGEN CONCENTRATION IN LANDFILL MOISTURE,
RECVCLE SIMULATION
oooo •
enact
buUO
4000
TUOO
XA
oaaa .
, yi « Jouu
(ng/L)
oaaa .
£OOD
innn •
IDDD
R
[1
I
8
5B
188
1!50
2160
250
300
350
400
450
TIME SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 15
HVDF10LYS1S PRODUCT CONCENTRATION IN LANDFILL LEACHATE,
RECYCLE SIMULATION
1OOOO T
qaoaa J
JuuDu
SH
S (rug/L)
icnaa .
iDtJDO
IRflRR
QQQQ
0
0
1
1
5
0 1(
)8 15
iO 2if
J0 2!
30 3(
JO 3!
30
468
450
TIME SINCE LEACHATE GENERATION BEGAN (DiAYS)
-------
FIGURE 16
XM
(mg/L)
208
188
168
148
128
188
88
68
48
28
B
METHANOGEN CONCENTRATION IN LANDFILL MOISTURE,
RECYCLE SIMULATION
1138 158 288 258 386 358
TINE SINCE LEACHAIE GENERATION BEGAN (DAVS)
488
45EI
-------
FIGURE 17
UOLATILE ACID CONCENTRATION IN LANDFILL LEACIHATE,
RECYCLE SIMULATION
lOOOU •
18888
16888
14888
12888
g (.a/L) 188B8
8888
6888
4888
2888
R -
»MB HB^MN Mt Ml
'
, — ^— -•
_iVMM«l
•-.
N.,^
\
\
\
\
\
V
8
SB
188
158
268
258
308
3'5B
488
450
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 18
COD
(mg/L)
68080
50080
40000 f
30000
20000 -
10000 -
6
0
COD CONCENTRATION IM LANDFILL LEACHATE,
RECVCLE SIMULATION
i;
56
100
15B
268
250
3HE)
350
400
456
TINE SINCE LEACHATE GENERATION BEGAN (DiAVS)
-------
FIGURE 19
METHANE PRODUCTION,
RECYCLE SIMULATION
METHANE
o (L/day)
100
90
80
70
60
40
30
20
10
0-1
0
Ei6 1100 150 200 2513 300 350
TIME SINCE LEACHATE GENERATION BEGAN (DAVS)
450
-------
utilization by methanogens, whereas all volatile acids removed in the recycle
system were removed by methanogenesis.
The volatile acid concentration profile for the recycle simulation
(Figure 17) was also markedly different from the single-pass simulation
(Figure 10). The volatile acid concentration increased from the initial
value of 5,000 mg/L to 17,5000 mg/L within the first week and continued to
increase gradually until Day 200. On Day 200, corresponding with initiation
of methanogenesis, the volatile acid concentration began to decrease rapidly,
falling below 1.0 mg/L by Day 390. Volatile acid concentration then gradually
trended upward for the remainder of the recycle simulation. The COD
concentration profile (Figure 18) was again very similar in shape to the
volatile acid concentration profile with the exception of the initial peak.
Methane production in the recycle simulation (Figure 19) followed the
trend of methanogen growth. Methane production increased rapidly, peaking at
670 L/day on Day 366, then decreased more rapidly than the decrease in
methanogen concentration, falling to a value of 2 L/day by Day 368 as the
concentration of volatile acids was depleted.
Figures 6 through 19 have been presented as a means to convey the
mechanics of GTLEACH-I. However, most of the parameters output by GTLEACH-I
are seldom measured in landfill experiments. For this reason fitting of the
model to experimental data must be limited to those parameters which are
measured. Of those parameters output by GTLEACH-I, available experimental
data are typically limited to COD, volatile acid concentration and methane
production.
Although COD is the most frequently measured leachate parameter in
landfill experiments, it can be the most unreliable of the three due to
losses of volatile acids by volatilization during the COD digestion procedure.
Not only is COD likely to be inaccurate in terms of other commonly available
leachate characteristics measured during landfill experiments, it can also be
misleading in its interpretation since COD analysis measures not only COD
contributed by hydrolysis products and volatile acids, but also other
constituents of leachate such as extracellular enzymes, microbial material
and refractory components of the solid waste. Thus it was difficult to
compare experimentally measured COD with the sum of COD contributions of
simulated volatile acids and hydrolysis products.
Volatile acid concentration in leachate can be accurately measured by
gas chromatographic techniques. Since volatile acids are produced in
landfill leachate almost solely as a result of microbial activity, volatile
acid concentration could be directly related to the processes simulated by
GTLEACH-I. In fact, measurement of volatile acid concentration in leachate
allows monitoring of the net effects of three of the processes occurring in
the landfill: production of volatile acids by acidogens, removal of volatile
acids by methanogens, and removal of volatile acids by washout. Thus
volatile acid concentration was a key parameter in fitting GTLEACH-I to
experimental data.
61
-------
As long as accurate gas volume measurements are made and there are no
gas leaks in landfill experiments, the methane fraction of landfill gases can
be accurately measured by gas partitioning and converted to volume.
Measurements of methane production from landfills allows monitoring of the
activity of methanogens.
The greatest difficulty in verifying the three-step microbial model is
the frequent lack of any direct experimental measurement of hydrolysis
product concentration. Without direct measurement, accurate modeling of
hydrolysis product concentration can only be inferred from correlation of
simulated and experimental volatile acid concentration and methane production
62
-------
SECTION 7
FITTING GTLEACH-I TO EXPERIMENTAL DATA
The GTLEACH-I model has been applied to the simulation of a landfill
experiment begun by Schaffer (1986) and completed by Yari (1986). In the
experiment, two pilot-scale (208-L) landfill cells were filled with shredded
municipal solid waste (MSW). One of the cells was operated in a leachate
recycle mode while the other was operated in single-pass mode.
The cells were each packed with 82 kg of shredded MSW having a moisture
content of 33%. The MSW absorbed an additional 44 liters of applied water
before attaining field capacity. Because the waste was shredded, actual
determination of the food fraction was not possible, so it was assumed that
the refuse contained 20% food waste or degradable waste on a mass weight
basis (Tchobanoglous et al., 1977). Thus the GTLEACH-I input parameters
XMASSO, the total mass of degradable solid substrate, and TVOL, the field
capacity of the cell, were set at 11 kg and 71 liters, respectively.
The rate of moisture application for the single-pass cell was 6 L/wk
once field capacity was attained (equivalent to 127 cm/yr of rainfall with
100% infiltration). The average daily flow, F, was then 0.857 liters. For
the recycle cell, approximately 25 L/wk of leachate were recycled through the
cell, giving an average daily flow of 3.6 liters. These high rates of
moisture application were designed to provide accelerated stabilization of
the municipal solid waste, but also to delay the onset of the methane
fermentation phase intentionally imposed on both the recycle and single-pass
cells. These delays in natural methanogenesis activity prompted eventual
seeding of the cells with digested sludge and development of methanogenesis
activity after 200 days of leachate production. Therefore, a time lag of 200
days was selected before initiation of the methanogen population in this
simulation.
SIMULTANEOUS FITTING
Fitting of the GTLEACH-I to the Schaffer (1986) and Yari (1986) data was
initiated by first selecting values for the kinetic constants. Values of
kinetic growth constants for acidogens and methanogenic bacteria compiled by
Henze and Harremoes (1983) are included in Tables 1 and 2. The tables
present values for maximum specific growth rates, umax, rather than maximum
substrate utilization rates, Q. Maximum substrate utilization rates can be
obtained by dividing maximum growth rate by a corresponding yield
coefficient. Median values of kinetic growth constants shown in the tables
were selected as initial values from which to begin the fitting procedures.
63
-------
Few literature values are available for hydrolysis rate constants.
Eastman and Ferguson (1981) presented a value of 0.3/day at 35°C for a first-
order hydrolysis rate constant measured in a separate acid-producing reactor
fed with raw domestic primary sludge. The hydrolysis rate constant measured
in Eastman and Ferguson's (1981) experiment with primary sludge would not
necessarily be expected to apply to the hydrolysis of shredded municipal
solid waste. Large particles with low surface-to-volume ratios such as
shredded MSW would be hydrolyzed more slowly than smaller particles such as
those resulting in primary sludge fed to completely mixed digestors.
Starches, proteins and cellulose are degraded at different rates. In
addition, the overall first-order hydrolysis rate constant represents the sum
of individual processes occurring in the landfill system and would reflect
the relative composition of the substrate. Therefore, the initial value
selected for KJJ was based on an order-of-magnitude approximation of the
Eastman and Ferguson (1981) value. Parameters selected for initial
simulation were:
Flow, F = 6 L/wk =0.86 L/day
Moisture content, TVOL = 71 L,
Number of CSTRs in series, NCSTR = 3
Initial mass of solid substrate, XMASSO = 11 kg,
Initial concentration of hydrolysis products, SHO = 40,000 mg/L,
Initial concentration of volatile acids, SAO = 5,000 mg/L,
Initial concentration of acidogens, XAO = 100 mg/L, and
Initial concentration of methanogens, XMO = 10 mg/L.
Kinetic parameters for the various processes were:
Hydrolysis Acidogenesis Methanogenesis
% = O.I/day QA = 3.5/day Q^ = 1.5/day
KA = 200 mg/L KM = 500 mg/L
KDA = 0.5/day KDM = 0.02/day
YA = 0.5 YM = 0.02
Methanogen lag = 200 days
A preliminary simulation was conducted using these values. Simulated
results of volatile acid concentration as a function of time were compared
with experimental results. The parameters were then adjusted to fit the
experimental results of volatile acid concentrations with time in subsequent
simulations.
Several objectives were applied as a first attempt at fitting the
parameters. An effort was made during parameter adjustment to fit both
recycle and single-pass volatile acid data simultaneously with more emphasis
placed on the recycle cell as it provided a clearer indication of the impact
of microbial processes without washout effects. Although, microbial popu-
lations in the two types of reactors were not expected to have exactly
64
-------
the same growth rate parameters, it was decided that an initial simultaneous
fitting would provide a set of parameters from which to explore the
differences in kinetic constants imposed by the two leachate management
options. A second initial objective was to avoid underestimating the peak
volatile acid concentration observed in the data, as this represented the
most aggressive leachate produced with the potentially greatest
environmental impact. Gross fitting of the model was accomplished by
adjusting the maximum rate constants; K^, QA, and 0^, for these parameters
were expected to show the largest deviation from literature values. Fine-
tuning was then conducted with small adjustments of the half-velocity
constants, decay rates and yield coefficients. In accordance with the above
objectives, the following group of kinetic parameters were finally selected
as producing the best fit for both recycle and single-pass data
s imultaneous1v:
Hydrolysis Acidogenesis Methanogenesis
KH = 0.0001/day QA = 3.2/day QM = 1.9/day
KA = 200 mg/L KM = 500 mg/L
KDA = 0.5/day KDM = 0.02/day
YA = 1.0 YM = 0.02
Methanogen lay = 200 days
The only major deviation in the fitted parameters from literature values
was the hydrolysis rate constant which was three orders of magnitude lower
than the literature value reported by Eastman and Ferguson (1981). That the
simulated value of hydrolysis rate constant differed from the Eastman and
Ferguson (1981) value was not surprising considering the previous discussion
of factors influencing hydrolysis rate constants. However, the degree of
deviation from the Eastman and Ferguson (1981) value is cause for question
and is examined subsequently.
The fitting value of maximum substrate utilization rate for acidogens of
3.2/day was at the low end of the range provided in Table 1. The fitted
half-velocity constant for acidogens was consistent with experimental values
reported in Table 1 as was the decay rate for acidogens. The value of the
yield coefficient of 1.0 for acidogens was approximately twice the typical
values reported in Table 1.
The maximum substrate utilization rate for methanogens of 1.9/day was
somewhat higher than the rates measured in acetate enrichment experiments
reported by Ghosh and Klass (1978) shown in Table 2. For many of the
experiments reported in Table 2, however, the maximum substrate utilization
rate for methanogens could not be determined due to insufficient information
(either no maximum specific growth rate or no yield coefficient). The half-
velocity constant for methanogens resulting from fitting of GTLEACH-I was
well within the range of values reported in Table 2 as were the decay rate
and yield coefficient.
65
-------
TABLE 1
GROWTH CONSTANTS OF ACID PRODUCING ANAEROBIC BACTERIA
Maxl>uii Yield Half Substrate Decay
specific coeffl- velocity removal rate
YA KA % KDA
d"1 kg VSS/ mq COD/L kg/COD/ d~
kg COD B3 (kg VSS'd)
JO 0.17 23 6.1
0.34 0.43
250-3200 0.88
150 1
2000 2.J
0.12 0.08
3.8 0.40 37000 0.79
>1.33 -0.54 0.87
0.41 0.10
0.26
0.15 19J 7.5
2.7
7.2 0.14 370
3.8 0.28 18300
1.7 0.13 30900
Teape- Culture/
rature substrate
•c
37 Mixed anaero*
bid (dextrose)
35 Acid producing
sludge
Mixed anaero-
bic (model)
Mixed anaerobic
Mixed anaerobic
37 Acid producing
sludge (sludge
feed
38 Mixed
Theoretical
(glucose)
35 Theoretical/
Model (glucose)
33 Mixed (design)
35 Glucose enrlch».
35 Sevage sludge
35 Cellulose
enricha
Literature
Ghosh and Pohland (1974)
Eastaan and Ferguson (1981)
Llndgren (1982)
Mueller and Manclnl (1975)
Young and McCarty (1967)
Ghosh et al. (1975)
Andrews and Pearson (1965)
Speece and McCarty (1964)
Sykes (1975)
McCarty (1971)
Cujer and Zehnder (1982)
Ghosh and Klass (1978)
Ghosh and Klass (1978)
Ghosh and Klass (1978)
66
-------
TABLE 2
GROWTH CONSTANTS OF METHANE PRODUCING ANAEROBIC BACTERIA
Maxlaui Yield Half Substrate Decay
specific coeffi- velocity (HAc or rate
rate cient constant HAc Equlv.)
removal rate
".ax
d kg VSS/ mq HAc/L kg/HAc/ d"1
leg COD (kg VSS'd)
2.0 4
0.08- 0.02 2 4-4 8
0.09
0.02 2.4-4.8
0.05 2.4
0.05 0.04
0.34 0.04 160. 8.2 0.02
19
0.045- 200 4.5-7.5 0.01-
0.055 0.02
0.26 10
3.4 600
8
80
Teape- Culture/ Literature
rature remark
•c
Mixed anaerobic Mueller and Manclnl (1975)
•ent
30 • " • " "
20 " " " " "
Mixed anaerobic Young and McCarty (1976)
35 Acetate enrich- Lawrence and McCarty (1969)
•ent
33 Anaerobic sludge Kaspar (1977)
30 Model Hartitann (1981)
30 Acetate enrich- Cappenberg (1975)
nent
37 Anaerobic sludge Ghosh and Pohland (1974)
Mixed anaerobic/ Llndgren (1982)
model
0.03
8
80
160
950 0.4
1000 0.8
10-12
60 7.5
0.02 35
25
35
35
35
(Model) Mosey (1983)
(Fixed bed/model) Svitzenbaum (1982)
(Exp.bed/«ooel)
Acetate enrichment Speece et al. (1982)
(Theoretical/ McCarty (1971)
model)
67
-------
TABLE 2 (continued)
GROWTH CONSTANTS OF METHANE PRODUCING ANAEROBIC BACTERIA
Maxima
specific
rate
UMX
d'1
0.4
0.24
1.4
0.5-0.7
>1.33
0.3
0.49
Yield Half Substrate Decay
coef fl- velocity (HAc or rate
dent constant HAc Equlv.)
reaoval rate
YM
kg VSS/ mg HAc/L kg/HAc/ d"1
kg COD • (kg VSS'd)
(0.039)
0.06 2
0.04- 154 8.1 0.01-
0.05 0.04
333 4.8
869 4.7
0.04 70 0.02
0.07- (0.6-
0.09 0.9)
0.03- (0.7-
0.04 0.9)
300
0.03- 300
0.04
2.1-2.2
3.7
0.14 0.02
200
0.28 3900
Tempe- Culture/ Literature
rature remarks
•c
(Theoretical) Bauchop and Elsden (1960)
(Hodel/Hlxed Andrews and Graef (1971)
culture
35 Reference Unavailable
30
25 " "
(Model) Reference Unavailable
30 Mixed (soured Lettlnga et al. (1980a)
sugar-beet waste)
30 "
SO Pure culture Zlnder and Hah (1979)
36 Pure culture Smith and Hah (1978)
30 Acetate enrich*. fol et al. (1982)
38 Acetate enrlchn. " " " "
38 Mixed Andrews and Pearson (1965)
33 Hlxed (design) Guger and Zehnder (1982)
35 Acetate enrlchm. Ghosh and Klass (1978)
68
-------
A graphical display of the comparison between the experimental total
acids data and the simulated results are shown in Figures 20 and 21. The
corresponding results for COD are also shown in Figures 22 and 23 and for
methane production'in Figures 24 and 25. It should be pointed out that this
simulation was the same as that presented previously. As a first trial, the
shapes and magnitudes of the simulated curves showed reasonable
correspondence with the data. There were, however, several important
deviations from the experimental data.
For this single-pass simulation, the entire simulated profile of
volatile acid concentration was shifted approximately 20 days earlier than
the experimental data. Furthermore, the simulated volatile acid curve
underestimated the concentration at which the experimental data leveled out.
Both these deviations were manifest to a lesser degree in the COD
simulation. Examination of methane production indicated that the initial
concentration of methanogens (10 mg/L) was too high and needed to be lowered
to allow for an increase in methane production to the peak value (Figure 9).
For the recycle simulation, the criteria which required that peak
volatile acid concentrations not be underestimated resulted in poorer
fitting of the volatile acid data between Days 20 and 250 than might
otherwise have been obtained (also shown in the COD curve). The methane
production simulation for the recycle mode showed good correspondence with
the data, although the initial value of methanogen concentration was too
high in this case as was evidenced by the abrupt initiation of methane
production.
Before further discussion of reasons for the preceding deviations from
experimental data and exploration of remedies which could be made, a
sensitivity analysis of the volatile acid profiles performed on the fitted
parameters was considered instructive.
SENSITIVITY ANALYSIS
By selection of a range of values for the various model parameters, the
impact of each value on model output was examined. As demonstrated in
Figures 26 and 27, varying the number of CSTR's in series from 1 to 5 had
only a small effect on the single-pass simulation and virtually none on the
recycle simulation. The effect of increasing the number of CSTR's was to
shift the washout curve of the volatile acids from one resembling a tracer
study for a completely mixed reactor to one resembling a plug flow reactor.
Since there was no washout in the recycle system, the sole removal mechanism
for volatile acids was by microbial degradation which was not significantly
affected by the mixing regime.
The results shown in Figures 28 and 29 demonstrate the effects of
adjustments to the value of the first-order rate constant for
hydrolysis/solubilization, KH. Increasing the value of KH from 0.00005/day
to 0.0005/day shifted both the recycle and single-pass simulated volatile
acid curves upward, with a greater change seen in the recycle simulation.
The effect served to slightly improve the fit on the single-pass data, but
produced a negative effect on the fit of the recycle data. This increase in
69
-------
FIGURE 20
COMPARISON OF SIMULATED VOLATILE ACID CONCENTRATION UITH
EXPERIMENTAL DATA CSCHAFFER 1986, YAH I 1986), SINGLE-PASS
/ /T \
(ng/L)
18000 -
160019 -
1 ^IQQlCJ
17RHIR -
itMWJ
D060 "
Aocum
a .
"'X
\.
V. " *'
•
•
•
• •
\
\
\
\-
\
•m
\
\Vio
- '
"^— -
• _ •
•—-x™^,^^.
^ *
1 •
. • • "
„
Ei8
109
150
200
250
300
350
460
458
TIME SINCE LEACH ATE GENERATION BEGiftIN (DAYS)
-------
FIGURE 21
COMPARISON OF SIMULATED VOLATILE ACID CONCENTRATION UITH
EXPERIMENTAL DATA (SCHAFFER 1986, YARI 1986), RECYCLE
20000 -r-
18800 -
•IVA
(ng/L)
8000
6000 -
0
IBB
iir>e
200
250
300
358
400
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 22
COMPARISON OF SIMULATED COD UITH EXPERIMENTAL DATA
(SCHAFFER 1986, YARI 1986), SINGLE-PASS
uuuuu -
caaaa .
OUDDU
COD
(ng/L)
jnaan .
£UOOO
a .
.
k
- \
.-
\
\
X
t--.
vx
•^-0;
^" • '
. . • . • •
• • • • .
. •
IBB
158
288
250
3BB
358
4BB
45B
TIME SINCE LEACH ATE GENERATION BEGAN (DiftYS)
-------
FIGURE 23
COMPARISON OF SIMULATED COD WITH EXPERIMENTAL DATA
(SCHAFFER 1-986, YiARI 1986), RECYCLE
68088 -i T—
COD
(mg/L)
48888
38888
B
188
158
208
258
300
350
488
450
TIME SINCE LEACIHiATE GENERATION BEGAN (DiAVS)
-------
FIGURE 24
2.5
1.5
METHANE
(L/day)
8.5
B
COMPARISON OF SIMULATED METHANE PRODUCTION UITH EXPERIMENTAL I'ffilA
(SCHAFFER 1986, VAR) 1986), SINGLE-PASS
\
58
JIBB
156
290
2513
308
358
488
458
TIME SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 25
78 T
METHANE
(L/day)
COMPARISON OF SIMULATED METHAINE PRODUCTION UITH EXPERIMENTAL DiATift
(SCHiAFFER 1986, YARI 1'986), RECY'CLE
)!)l)
JStl 40IB
450
TIME SI flu
lull UKGAN (DAYS)
-------
FIGURE 26
SINGLE-PASS SENSITIVITY ANALYSIS OF NCSTR
TUA
(ng/1)
NCSTR=5
— NCSTR=3
NCSTR=1
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 27
HECYCLE SEMSITIUITV ANALYSIS OF NCSTH
28000
16000
14000
TUA
(ng/1)
10000
8000
6000
4000
0
S
5
MCSTR=5
B
50 100 150 200 250 300 350 488 458
TIME SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 28
SINGLE-PASS SENSITIUITV ANALYSIS OF XKH
1 DOUR •
loDDu
1(.nna .
Ibutju
i Aooa
ITCmO '
1 7OC4O
l£ODu
'" i nnnn
, y , , 1UDUD '
(ng/1)
oaaa .
OUDO
caaa
bOtJo
Aaaa
TODO
7RRO
i.ODO -
-1
(
"\
\
« J-
* •
•
•
3 Ei
*
• •
V '
1^
^
^
e K
%
,-..._
! \S •
^
318 1!
•
^^^*Hk
*^JO||
i8 21
,te^;
38 ?.!
•••
*-«*•.««.
B0 31
38 3!
.
.-•"
'm™w^ymi
30 41
w»««w«
1 ""
38 4'
58
— XKH--:. 0085
— XKH::. 8081
/DAV
•• - • VlfU- nMRCic;
Alln— .DiJowo
/DAV
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 29
RECVCLE SENSITIVITY ANALYSIS OF XKH
£uemo
t nnaa
lOUOU
™ i anna
/ y , x IDtlOtl
(ng/1)
f CtCiCk
^4 QQQ
7RRR
a
.1
""•"""'
*
•
•
"
BS8»ffl5fl
•
•i
B *
*\"
•:v-
S3^5-S
•
•
• • •
•
5^5
•
•
\l
\
• *^!
'-•i
-•••
«^
\
1
i
\
•\
\
I \
\\
\1
XKH=.B0B5
/DrW
— XWi=.0BBl
B
EiB
1010 15B 2BB
3BB 35B 4BB 456
TINE S;INCE: LEACHATE GENERATION BEGAN (DAYS:)
-------
volatile acid concentration with increase in hydrolysis rate constant
supports the hypothesis of Eastman and Ferguson (1981) that hydrolysis is
the rate-limiting step in the acid phase of anaerobic processes. Thus,
increasing the rate of production of hydrolysis products also produced a
corresponding increase in volatile acid production.
The effects of changing the maximum rate constant for acidogenesis, Q^,
are examined in Figures 30 and 31. Decreasing the value of Q^ from 3.5/day
to 3.0/day increased the magnitude of both single-pass and recycle volatile
acid curves. This result can also be explained by a hydrolysis rate-
limitation in the acid phase of anaerobic digestion. Increasing the maximum
substrate utilization rate constant of acidogenesis alone should have
increased the rate of production of volatile acids. However, if the
increase in the maximum rate constant for acidogenesis produced a decrease
in the concentration of hydrolysis products greater than the increase in the
maximum rate constant for acidogenesis, the overall rate of production of
volatile acids would decrease according to Equation (59).
Moderate changes in the half-velocity constant for acidogenesis, K^,
from 100 mg/L to 500 mg/L had very little effect on the single-pass and
recycle simulations as shown in Figures 32 and 33. Results of adjustments to
the yield coefficient for acidogenesis are shown in Figures 34 and 35.
Moderate changes in the yield coefficient for acidogens, Y^, from 0.75 to
1.25 also had virtually no effect on the simulations for either flow regime.
Finally, Figures 36 and 37 reveal that adjustments to the value of the decay
rate for acidogens had no impact on either the single-pass or recycle
simulations. Thus, the major parameter of consequence for acidogenesis was
the maximum rate constant.
Adjusting the value of the maximum rate constant for methanogenesis, QJ.J,
from 1.5/day to 2.5/day, as shown in Figures 38 and 39, had virtually no
effect on the single-pass curve largely because it was overshadowed by the
effects of washout. A significant impact on the recycle curve was observed
adjusting the rate constant for methanogenesis. Decreasing the value of Q«
from 1.9/day to 1.5/day increased the time it took for the volatile acid
concentration curve to descend toward zero. Increasing QM to 2.5/day
shortened that time.
Adjustment of the value of the half-velocity constant for raethanogenesis
affected the simulations to a limited extend as shown in Figures 40 and 41.
Increasing the value of K^ from 100 mg/L to 1000 mg/L increased the time at
which the concentration of volatile acids descended toward zero in the
recycle mode. The same increase in KM caused a slight increase in the level
at which the volatile acid concentration leveled off in the single-pass
mode. Both of these results reflect a decrease in the overall rate of
removal of volatile acids by methanogenesis.
As illustrated in Figures 42 and 43, the yield coefficient for methano-
gens, YJ.J, had a very small impact on the single-pass simulation, yet a very
dramatic impact on the recycle simulation. A decrease in the yield
coefficient from 0.05 to 0.005 prevented the establishment of the population
80
-------
FIGURE 30
SIHGLE-PASS SENSITIUITV AMALVSIS OF QA
*
QA=3.5/DAV
— QA=3.2/DAV
QA=3.8/DAY
58
188 158 288 258 388 358 488 458
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
oc
ho
TUA
(ng/1)
FIGURE 31
1R1ECVCLE SENSITIVITY ANALYSIS OF QA
&UUUO
1 PARR .
lOUUD
1 f,RRR •
ibUuo
i4RQa .
IHDtJU
1 7R(in
l^ODQ
1 nnctn
lOOOD
RRRR
6000
Aaaa
luoo
7QQQ
^DDU
fl
^^•iMHM^
PftHKKKMKt^1
M
.
X
^•^^^™
•
"-•
QA=3.B/DAY
QA=3.2/DAY
•- QA=3.5/DAY
0 50 1610 150 200 258 300 350 480 458
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 32
SINGLE-PASS SENSITIVITY ANALYSIS OF XKA
£OPOD
1 OQQQ
lODOD
\caaa
iDUOO
14H cm
llOtlt)
4OQQQ
l&tHJt)
THA
IQoaa
(mg/1) 10000
oaaa
ODDu
AROfi •
DUDD
4nn(i
nooo
2888
• ">'-.
•
£ •
.
• •
v
\
%.
\ ,s
'•t
:&
\
%
s%
'""-.-.,:«
»
"••
W-
• * « •
•"
•"""""'~'"'"
— XKA=5BB
wg/1
— XKA=28B
mg/1
~..» VI/A— Ida
AKH-lUt)
mg/1
8
EiB
1881 158 288 258 388 358 480 458
TINE SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 33
RECYCLE SENSITIVITY ANALYSIS OF XKA
£ooao •
ioaaa .
lOOUO
i Anna .
IbODu
1 4RRR
14DDD
1 RRRR
/ / • v IDuuo
(rug/1)
RRRR
ARRR -
DODO :
OQQQ
R
WIOOOMtwl
"
•
•
[tWXKMK(W°*0
•
0 i.
*•
• * *
*
Hxxftepo°4>Rftfi
•
Ka*nL_, „_
•
•
s
\
* '
\
\
•
,
\
\
\...
- XKA =588
wg/1
— XKA=2B8
mg/1
oooi VV6-1RQ
AKH-ltJa
ng/1
B 50 186) 158 288 25B 388 35B 4B8 458
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 34
SINGLE-PASS SENSITIUITV ANALYSIS OF VA
00
Ln
zuuuu
j nana
lottUo •
4 fL QQQ
IbHUB •
i OQCICI
L/.OUO
Til A
foaaa
(mg/1) 10008
aana
DODO
(.ana
buuu
ddflfl
nuuu
20BB
:%4
. '-''^
•5
;; "
t; '
1
;:
•
• •
\
\ ••
"—
%
._
^!
X
. "
":*^
M • •
58*^.^....,,
".•
1 •^•"^•"•••^•••" -^
-^:-"":~''"
— VA=1.Z5
~ V<^-1 R
in— x • o
^ VA=8.75
8 EIB 1B0I 158 ZBB Z5B 308 358 488 450
TINE SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 35
RECYCLE SENSITIVITY ANALYSIS OF YA
™A
£OUOU '
toaaa
Loooo
\Aaaa
lIuoU
17(1(1(1
II.OOO
RRRR
(.ana .
DtJUU
Aaaa
TOtJO
2000
a
WtOOOOMtMi
"
•
.
•
tteoociot
"•-
. • *
*
• • -
flM
"""s
•
*
V
V
"X
\
\
,
\
•y
\
\.
- YA=1.25
— VA~1 R
""" YA=B.75
e E;B
iB0i 150 zee zse see 350 -\m 459
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 36
SINGLE-PASS SENSITIVITY ANALYSIS OF XKDA
TUA
00
—J
£UUUU
4 OAQQ
i dftna •
llooo
loaaa
l&OOO
1(1(1(1(1
lauuD
RRRR
ARRR -
DUDD
4dOP) •
tODD
2000
R -
\.
••H.
1 •• ''
: * •
| •
•i •
•
• .
'V.'
1 '•
\
\
.^
5!U
' X.
•v-
\;
— k*
**'M
'••:-. i, . .
;_,.,; , M- , ,.. -
> "*
_-."
/DAY
» XKM=8.5
/DAY
/DAY
u
Eie
150
250 300 :&B 400
TIME SINCE LEACHATE GENERATION BEGAN CDAVS)
-------
FIGURE 37
RECYCLE SENSITIVITY ANALYSIS OF XKDA
oo
oo
£OOOO •
1 QOOO
laotJtt
4 £.GIGICt
IbodtJ
1 dftnn
14Hua
1 7OOCI
l£uuu
1 RflRR
, y , . lUUDD
(ng/1)
RRRR
Anntt
bOuo
dp) on
louo
2008
a
MK^MIMxIMMHl
•
. •
•
•
(;9f t««KWH» »
•
\
•
»•
•
^TOTtCMMK.*
. • "
•
* .
•
""-s
•
••
*>*,.
^%
•X
^t
\
"\
\
N
,
\
\
X.
—T-» VW^fWWWWh.
XKDA =1.0
/DAY
-- XKDA=0.5
/DAY
OCMH vurkA a oc
AKUfl-o.^b
/DAY
0 58 1010 158 200 ?.58 300 2158 488 458
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 38
SINGUI-PASS SENSITIVITY ANALYSIS OF QM
ZHUHU
4 OQQQ
looUU
i £.ana
IbHoo '
4 Anna
14DUD
TUA
laaaa
/ yi v lUUUD
(ng/1)
oaaa
oauu
(.aaa .
DOt)U
2BR0 •
fl
*\
\
•
:
•
• .
i " '
; •
V
\ "•
\
,'
'. .
\
*•?.
X
* • •
"^^
•^*H
Ki>z-£+-\?^SJHSiR
.«_
C1M=2.5/DAY
— C1M=1.9/I)AY
CJM=1.5/DAY
8
56 100i 150 2B6 250 300 350 400 450
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 39
RECYCLE SENSITIUITY ANALYSIS OF QI1
8
QM=2.5/DAY
— qn=l.
QN^l.S/DAY
8 56 1881 150 280 250 300 350 488 458
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 40
SINGLE-PASS SENSITIVITY ANALYSIS OF XKM
ZHUUU •
4 nnnn
4C.QQQ
Ibuao
4 Anna
llooo
1 RRCin
x y I x lOOOD
(ng/1)
annn
ODDU
A (ton
Duuo
28BB
a
p-^. ^
V
^
* •
•
X
r*
!
s
T
.
•
i
\ .
V
._
*•"•
s
^\
•
"••
.
•"*
— XKM=1BHO
ng/1
•- XKH=58«
ng/1
,,..v vi/M-inct
Ai\ri-itJii
wg/1
8
58
158 288 258 388
48B 458
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 41
RECYCLE SENSITIVITY ANALYSIS OF XKH
IVA
(ng/1)
VO
K)
2080
0
0
n.,/1
XKM:=1B8
ng/1
IB
258 300 358 488 450
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
28888
TUA
(ng/1)
FIGURE 42
SINGLE-PASS SEMSITIVITV ANALVSIS OF VM
VM=.B5
— VM=.82
VM=.8B5
8
EIB 188 158 288 250 388 358 4BB 458
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE A3
RECYCLE SENSITIVITY ANALYSIS OF YM
TUA
(ng/1)
YM=.85
— YN=.B2
— Yh=.BB5
B
Efl IBB 158 288 250
388
488 458
TIME SIMICE LEACHATE GENERATION BEGAN (DAYS)
-------
of methanogens, thereby shortening the time before removal of volatile acids
was essentially complete.
The effect of modifications in the decay rate for methanogens on the fit
of the simulations to the data is demonstrated in Figures 44 and 45. As
observed for changes in the yield coefficient, changes in the decay rate for
methanogens from 0.02/day to 0.05/day produced an effect similar to decreas-
ing the yield coefficient for methanogens, resulting in a flat volatile acid
curve. Correspondingly, a decrease in the decay rate, KD^» shortened the
time for essentially complete removal of the volatile acids.
A summary of the results of the sensitivity analysis is presented in
Table 3. One striking observation derived from the results of the sensiti-
vity analysis was that the establishment of methanogenesis had very little
impact on the volatile acid profile for the single-pass system, while it had
a very important impact on the volatile acid profile for the recycle system.
Upon closer examination it seems that, in general, the volatile acid profile
for the single-pass system was less sensitive to changes in all of the
parameters when compared to the response of the recycle system. The most
obvious explanation for this behavior was that the washout effects of the
single-pass system masked the response of the system to changes in the
parameters. Thus it would be expected that a decrease in the rate of
moisture application to the single-pass system should cause the single-pass
system to behave in a manner more similar to the recycle system as the
effects of washout were diminished. In fact, examination of an additional
set of experiments conducted by Chang (1982) reveals this to be the case as
is demonstrated subsequently.
SEPARATE FITTING
Based on further understanding of the model gained from the simultaneous
fitting exercise and sensitivity analysis, an improved fit of the model to
the data from the Schaffer (1986) and Yari (1986) experiment was attempted.
In this case, the two leachate management options were fit separately, i.e.,
a different set of kinetic constants was fit for each of the two leachate
management options. Additionally, the fit of the methane production
simulation to experimental data was examined to provide an additional test
of GTLEACH-I.
Kinetic parameters selected in the final separate fit for the single-
pass simulation are:
Hydrolysis Acidogenesis Methanogenesis
KH = 0.0008/day QA = 3.2/day QM = 1.5/day
KA = 200 mg/L % = 500 mg/L
KDA = 0.5/day KDM = 0.02/day
YA = 1.0 YM = 0.028
Methanogeri lay = 200 days
95 :
-------
FIGURE 44
SINGLE-PASS SENSinUITV ANALVSIS OF XKDh
\Aaaa
iVawj
iaaaa .
f y , v J.UODD
(mg/1)
flflRfl
OODD
ARRR -
DODO
40RR •
TOIIU
?f)(i(t •
^DOO
a
l'\
1 X
.„
•
•
f
•
' •
'
5' "•
\
s.
i
•
,'
V
X
1*
"%<^™
5-^_.
•.-
.
."•
,_,
/DAV
/DAV
/IIAV
Ei0
IBB 150 206 258 380 3S0 408 45H
TIhE SINCE LEACHATE GEHERATIOM BEGAN (BAYS)
-------
FIGURE 45
RECYCLE 8EN8ITIUITY ANALVSIS1 OF XKDM
TUA
(ng/1)
XKDN=B.B5
/DAV
/DAV
XKDN=0.BB5
/WAV
B
16i8 15B 2BB 258 388 358 488
458
TIhE SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
TABLE 3
RESULTS OF SENSITIVITY ANALYSIS
ON VOLATILE ACID CONCENTRATION PROFILE
Single-pass Mode
Recycle Mode
Increase- f CSTR'e
Increase KH
Decrease Q*
Changes In K,
Changes in Y,
Changes in K»*
Decrease QH
Increase K»
Decrease Y.
Decrease Km
approach plug flow
increase magnitude of
concentration profile
increase magnitude of
concentration profile
little or no effect
little or no effect
little or no effect
little or no effect
increase final concentration
attained
little or no effect
little or no effect
little or no effect
increase magnitude of
concentration profile
increase magnitude of
concentration profile
little or no effect
little or no effect
little or no effect
increase tine to removal
increase time to removal
by methanogens
increase time to removal
by methanogens
increase time to removal
by methanogens
98
-------
The changes which occurred in the kinetic parameters selected in the
separate fitting from those in the simultaneous fitting of the single-pass
data were: the hydrolysis rate constant increased from 0.0001 to
0.0008/day, the maximum substrate utilization rate for methanogenesis
decreased from 1.9 to 1.5/day, and the yield coefficient for methanogenesis
increased from 0.02 to 0.028. In addition to changes in kinetic parameters,
the initial concentration of methanogens was decreased from 10 mg/L to 0.1
mg/L. Figures 46 and 47 compare the results obtained from the separately
fit parameters with those from the simultaneous fit for the single-pass
data.
A markedly improved fit of the volatile acid data was obtained in the
separately fit single-pass simulation when compared to the simultaneous fit.
The separately fit simulated volatile acid concentration curve closely
followed the experimental data as it decreased from the initial peak. The
experimental concentration of volatile acids reached a low point near Day
225 then increased slightly for a period of 75 days until leveling out. The
separately fit simulation did not follow these latter trends exactly, but
rather smoothed them by slightly overestimating the low and underestimating
the eventual leveling out of the curve. This was still a substantial
improvement over the simultaneously fit simulation.
Unfortunately, improvement in the simulation of methane production could
not be obtained without sacrificing the fit of the volatile acid data. In
order to elevate the levels of methane production in accordance with the
data, the maximum substrate utilization rate or the yield coefficient for
methanogens would have been increased (effectively increasing the growth
rate of methanogens). The resulting increase in methanogenesis would have
decreased the concentration of volatile acids in the leachate resulting in a
poorer fit of the volatile acid data.
Figures 48 and 49 compare two alternative improved fits to the recycle
data of Schaffer (1986) and Yari (1986). The two sets of kinetic parameters
employed in these fits were:
SEPARATE FIT 1
Hydrolysis Acidogenesis Methanogenesis
KH = 0.0005/day QA = 3.5/day QM = 1.9/day
KA = 200 mg/L KM = 500 mg/L
KDA = 1.0/day KDM = 0.02/day
YA = 0.5 YM = 0.03
99
-------
FIGURE 46
o
o
TVA
COMPARISON OF SIMULATED VOLATILE ACID CONCENTRATION
Ull'H EXPERIMENTAL DATA (SCHAFFER 19EI6, VARI 1986), SINGLE-PASS!
50 IBiB 158 288 258 380 358 480 45B
TIME SIINCE LEACHATE GEMERATIOM BEGAM (DAVS)
SIMUILTANEOUS
FIT
S:E:PARATE
FIT
-------
FIGURE 47
COMPARISON OF SIMULATED METHANE PRODUCTION
UITH EXPERIMENTAL DATA (SCIHAFFER 1986, VARI 1986), SINGLE-PASS
2
1.5
METHANE
(L/day)
1
1
8.5
0
T ""• " "•'•
1 * A"" '
k
\
\
\
\
\
\
!
V
.a*******
•
•
•
m M
^
i*"**""'
*
.
^
. t. •'.•& "' ' ^*^*%
•- .
~^~-
"'
— SIMUILTANEOUS
FIT
- SEPARATE
FIT
0 SB ieiei 150 zee ZSB aee 350 400 <&o
TIME SINCE LEfiCHfiTE GENERATION BEGAN (DAVS)
-------
FIGURE 48
o
ro
COMPARISON OF SIMULATED UOLATILE ACID CONCENTRATION
UITH EXPERIMENTAL DATA (SCHftFFER 1986, YAH I 1986), RECYCLE
18888
El 58 1813 156 288 253 360 3!5B 480 458
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
SEPARATE
FIT 1
SEPARATE
FIT 2
-------
FIGURE 49
COMPARISON OF SIMULATED METHANE PRODUCTION
UITH EXPERIMENTAL DATA (SCHAFFEFI 1986, VARI 1986), RECYCLE
14er-
iee
o
U)
METHANE
(L/day)
48
e
80
00
-
e sia IBB 150 zee 258 see 350 400 4se
TUNE SINCE LEACHATE GENERATION BEGAN (DAYS)
— SEPARATE
FIT 1
SEPARATE
FIT 2
-------
SEPARATE FIT 2
Hydrolysis Acidogenesis Methanogenesis
KH = 0.0005/day QA = 3.5/day QM = 1.9/day
KA = 200 mg/L % = 500 mg/L
KDA = 1.0/day KDM = 0.02/day
YA = 0.5 YM = 0.028
The only difference between these two sets of parameters was in the yield
coefficient for methanogens which was slightly higher for SEPARATE FIT 1
than for SEPARATE FIT 2. Changes in parameters when compared to simultane-
ously fit parameters were:
hydrolysis rate constant was increased from 0.0001 to 0.0005/day;
- maximum substrate utilization rate for acidogens was increased from
3.2 to 3.5/day;
- the decay rate for acidogens was increased from 0.5 to 1.0/day;
the yield coefficient for acidogens was decreased from 1.0 to 0.5;
the yield coefficient for methanogens was increased from 0.02 to
0.028 and 0.03;
the initial concentration of hydrolysis products was decreased from
40,000 mg/L to 20,000 mg/L; and
- the initial concentration of methanogens was decreased from 10 mg/L
to 1.0 mg/L for both separate fits.
Substantial improvements were effected by the separate recycle fits
(Figure 48) when compared to the simultaneous fit (Figure 21). First,
decreasing the initial concentration of hydrolysis products decreased the
initial peak in volatile acid concentration which, while underestimating the
initial peak in the experimental data, allowed the simulation to follow the
experimental data closely as the concentration of volatile acids gradually
increased from Day 50 to Day 250. Examination of Figures 48 and 49
indicated that while the higher yield coefficient gave better fit to the
volatile acid data as the concentration decreased from Day 275 to Day 400,
there was sacrifice in the fit of the methane production data when compared
with SEPARATE FIT 2. Decreasing the initial concentration of methanogens in
the separate fit served to eliminate the abrupt initiation of methanogenesis
observed in the simultaneously fit data (Figure 19). It should be noted,
however, that both separate simulations overestimated the peak methane
production data.
The key differences between the input parameters of the separately fit
single-pass and recycle simulations were:
104
-------
initial concentration of hydrolysis products for the single-pass
simulation was 40,000 mg/L, and was 20,000 mg/L for the recycle
simulation;
the hydrolysis rate constant was nearly twice as high for the
single-pass simulation than for the recycle simulation as observed
through the product of the yield coefficient and the maximum
substrate utilization coefficient;
the acidogen growth rate was nearly two times higher for the single-
pass simulation than for the recycle simulation as observed through
the product of the yield coefficient and the maximum substrate
utilization coefficents;
the decay rate for acidogens was two times higher in the recycle
simulation than in the single-pass simulation; and,
the maximum substrate utilization rate for methanogens was lower ia
the single-pass simulation than in the recycle simulation.
The higher initial concentration of hydrolysis products in the single-
pass simulation when compared to the recycle simulation can be attributed to
the greater volume of moisture circulating through the recycle system when
compared to the single-pass system. This would act initially to dilute the
hydrolysis product concentration to a greater degree in the recycle system
than in the single-pass system.
The lower hydrolysis rate constant in the recycle system may be the
result of product inhibition resulting from the accumulation of hydrolysis
products in the recycle system. The concentration of hydrolysis products in
the recycle simulation was 266 mg/L while the concentration of hydrolysis
products in the single-pass simulation was 37 mg/L.
The lower acidogen growth rate in the recycle system may also be the
result of product inhibition cuased by the accumulation of volatile acids in
the recycle system. Whether this would also explain the increase in the
decay rate is unknown.
In order for hydrolysis products to be product inhibited in the recycle
system, it must be presumed that hydrolysis would not be the rate-limiting
step under such conditions but that the rate of hydrolysis was controlled to
some degree by the rate of removal of hydrolysis products by acidogenesis.
Thus, the hypothesis of Eastman and Ferguson (1981) was contradicted in the
case of the recycle system.
Although a first comparison of Figures 47 and 49 might seem to indicate
that greater difficulty was encountered in fitting the methane production
data from the single-pass system as opposed to the recycle system, in fact
the converse was true. When the magnitudes of the methane production values
were compared, it was evident that the separately fit recycle simulations of
methane production showed a difference of as much as 50 L/day from the
experimental methane production data, while the single-pass separate fit
simulation showed a difference of less tha 2 L/day. '
105
-------
SECTION 8
MODEL MODIFICATIONS
Thus far the results of two alternative approaches to fitting GTLEACH-I
to experimental data collected by Schaffer (1986) and Yari (1986) have been
considered. In the first case, single-pass and recycle experimental data
were fit using a single set of kinetic parameters, while in the second case
the two experiments were fit with two separate sets of input parameters. It
was concluded that differences among the two landfill environments created by
the two leachate management options resulted in the need to fit kinetic
constants separately.
It was suggested that the recycle environment resulted in product
inhibition of hydrolysis and acidogenic processes, neither of which were
included in the GTLEACH-I microbial model. Two approaches to this limitation
could be taken. Either development of a separate group of rate constants for
recycle systems could be used to account for product inhibition, as was done
in the analyses herein, or more complex equations could be developed which
explicitly account for product inhibition of hydrolysis and acidogenesls.
Several other difficulties remain to be addresed in applying the
GTLEACH-I model to simulation of the Schaffer (1986) and Yari (1986) data.
FLOW MODEL
One of the most pressing issues, not only for application to the Yari
(1986) and Schaffer (1986) experiment, but for future modeling efforts as
well, is the necessity of a flow model. Because of the regular, weekly rate
of moisture application in the Schaffer (1986) and Yari (1986) experiment, a
reasonable fit of the data was obtained with GTLEACH-I by averaging the
moisture application rate to assume an average daily flow. Such an assump-
tion would not be possible if moisture were applied in a more seasonal manner
with accompanying irregular drying and wetting cycles.
Although daily leachate flow measurements were not available to quantify
the flow regimes within the single-pass experiment, a tracer study performed
by Schaffer (1986) during the acid formation phase and another by Yari (1986)
during the methane formation phase, were available. The tracer study
performed during the acid formation phase indicated a moisture residence time
of nearly five weeks, while the study performed during the methane-formation
phase indicated a moisture retention time of 14-17 weeks. This information
was not easily converted to leachate flow rate, however, it did indicate
that leachate was probably released more quickly after moisture application
toward the beginning of the experiment than it was near th end.
106
-------
In an attempt to capture the range of possible flow regimes encountered
during the Schaffer (1986) and Yari (1986) experiments, a series of simula-
tions were conducted which varied the rate of leachate flow among the days of
the weekly cycles. The results of causing the entire six liters of water
applied to pass though the cell on the day of application, with no flow of
leachate on subsequent days until moisture application seven days later, is
shown for the single-pass simulation in Figures 50 and 51. This simulation
effected no change in the shape of the volatile acid concentration profile,
however, it did produce small oscillations in the curve not unlike those
observed in the experimental data. The effect on the methane production
curve was to slightly increase the rate of methane production—a slight
improvement in the fit of the simulation.
Two variable-flow simulations were conducted for the recycle mode: in
the first case, all 25 liters of recycled leachate passed through the cell on
the day of application; in the second case, flow decreased throughout the
week beginning with 10 liters on the day of application, then 5, 4, 3, 2, 1,
0 on the next six days, respectively. Results of these simulations shown in
Figures 52 and 53 indicated a more dramatic impact than occurred for the
single-pass simulation. Increasing the variability of the flow through the
recycle reactor decreased the magnitude of the volatile acid concentration
and methane production profiles for the recycle simulation. More interesting
was the marked similarity to the data of the large oscillation produced in
the volatile acid simulation during the first 50 days.
The preceding simulations provided evidence that a variable flow model
could serve to improve the prediction of the GTLEACH-I model. Thus the
importance of an accurate flow model should not be underestimated, even for
systems where moisture is applied on a regular, cyclical basis. The
necessity for an accurate flow model would most certainly increase for
systems with irregular rates of moisture application such as would occur in a
full-scale landfill.
As indicated previously, most landfills can be characterized as
consisting of an unsaturated moisture zone underlain by a saturated zone
where leachate collects. For this reason the obvious choice of a flow model
for landfill simulation would be an unsaturated flow model which could
account for changes in moisture content of the unsaturated zone with rates of
moisture infiltration and evapotranspiration. However, there is difficulty
in applying unsaturated flow models to landfills due to frequent lack of
necessary input parameters such as: unsaturated conductivity with moisture
content and time. Further complicating the problem is the fact that all of
these parameters change with the age of the landfill as the landfill becomes
more consolidated and stabilized.
HYDROLYSIS RATE CONSTANT
A second problem encountered in the previous simulations of the Schaffer
(1986) and Yari (1986) data was that the hydrolysis rate constants for all
the simulations were several orders of magnitude lower than the value
reported by Eastman and Ferguson (1981). Although applicability of such a
value, measured in a conventional stirred reactor, to a landfill system can
I
107 ;
-------
FIGURE 50
COMPARISON OF CONSTANT FLOU
Ull'H CYCLICAL VARIABLE FLOU, SINGLE-PASS
16088
TVA
g (ng/L)
e
FH
•
• .
•
"*?~
p
r~
V
-\
%
— CO'NSTANT
FLOU
— VARIABLE
FLOU
0 58 1010 150 200 250 300 350 488 458
TINE SIIHCE LEACIHATE GENERATION BEGAN (DAYS)
-------
FIGURE 51
COMPARISON OF CONSTANT FLOW
UITH 'CYCLICAL VARIABLE FLOU, SIN'GLE-PiASS
£ •
In
.O •
1C
.b
1A .
.1
1 7
1. L
METHANE
(L/day) *
00
.0
0t
.O
0A
. 4
07
0.
1
a s
0 11
^iei i!
50 21
^—^»r
38 2!
••
..r^— •**
M 31
•
•
H
-y^iM^^
30 3!
•
•
•
£$r
30 4(
• • .
' •^:-
^
JB '«
>8
— CO'NSTANT
FLOW
~~ UAiRIABLE
F'LO'U
TIME SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 52
COMPARISON OF EFFECTS OF UARVING DISTRIBUTIONS OF FLOU
ON VOLATILE ACID CONCENTRATION, RECVCLE
TVA
1OOOO •
1 c^nnn
IDDtJU
1 Anna
lIHtjU
1 ?nnn
/•ana
bOOO
TUdtJ
n
• *
If"
\r
H
.
.
T i".-*1-^1*
<-*""
ii " *
,
•^ •
m
" * u
^
.
•
.-.-
.
, *
•^x
^A—
v\
^L 5i
\
\
i
V
\
— FLOU=3.6 L/day
- FLOU=18,E;,4,3,2,1
L/day
— FLOU=25,EI,8,B,8,8
Q
L/day
El 58 IBB 158 2HQ 258 38B 35B
458
TIME SINCE LEACHAI'E GENERATION BEGAN (DAVSI)
-------
FIGURE 53
COMPARISON OF EFFECTS OF VARYING DISTRIBUTIONS OF FLOU
ON METHANE PRODUCTION, RECYCLE
^B-
on
oo
7O
I U
(.a .
DO
CO
METHANE bd
-------
surely be questioned, it is nevertheless valid to question the fitted values
as well. Examination of Equation (55) for hydrolysis reveals two components
of the hydrolysis reaction rate: the hydrolysis rate constant and the
concentration of solid substrate. One possible explanation for the
excessively low hydrolysis rate constants fitted to the data may be that
there was an overestimation of biodegradable substrate fraction of the solid
mass. This fraction is difficult to measure and is frequently omitted from
compositional analysis of refuse.
The effects of decreasing the initial degradable mass of solid substrate
by order-of-magnitude steps and correspondingly increasing the hydrolysis
rate constants are demonstrated in Figure 54, 55, 56 and 57. These changes
produced increasingly poorer effects on the fits of both volatile acid and
methane production data for the single-pass simulation. In the case of the
recycle simulation, decreasing the degradable substrate from 11 kg to 1.1 kg
and increasing the hydrolysis rate constant from 0.0005/day to 0.005/day
decreased the magnitude of the volatile acid concentration curve, but did not
unduly affect the fit of the volatile acid data. This same decrease on the
overall production of methane, improving the fit of the methane production
data. Thus, no specific conclusion could be drawn as to the validity of the
approximation of 20% degradable substrate in the municipal solid waste used
in the GTLEACH-I model simulations.
SECONDARY RELEASE OF SUBSTRATE
An upward trend in volatile.acid concentration was observed in the
single-pass experiment data after approximately 250 days, but not reflected
in the simulation. One explanation for such a phenomenon could be that it
was the result of a secondary release of substrate. In such a scenario,
easily degradable solid substrate began to be exhausted by Day 200 so that
volatile acid production decreased. Meanwhile, volatile acids continued to
be washed out of the reactor. The resulting depletion of volatile acids in
the landfill moisture caused solubilization of the more refractory solid
substrate. This process would then result in the production of additional
volatile acids and concomitant methane production.
An attempt to simulate this effect was made by increasing the available
mass of solid substrate on Day 250. Results of this simulation are shown for
the single-pass mode in Figures 58 and 59. A fairly accurate representation
of the observed volatile acid and methane gas behavior was produced by this
simulation.
DELAY IN ESTABLISHMENT OF ACIDOGENESIS
One final problem encountered in simulation of the Schaffer (1986) and
Yari (1986) data was that the peak volatile acid concentration, particularly
for the single-pass experiment, appeared later than predicted by the
simulation. One possible explanation for this could be that short-circuiting
of the moisture flow'through the reactor during the first 30 to 75 days of
leachate production made less solid substrate available than would be the
situation otherwise. To test this hypothesis, the available mass of solid
substrate was incrementally increased over the first 50 days of leachate
112
-------
FIGURE 54
COMPARISON OF EFFECT OF VALUE OF INITIAL XNASSO AND KH
ON VOLATILE ACID CONCENTRATION, SINGLE-PASS
1UUOU -
itaao
ioooo
4AOOa
llODO
17000
l£Xluu
1OOOO
TUA 1BB0B
(n>9/L) flROO
Dotm
ABOO -
ODDu
4000
7RRA
a
r\
I %
! W
I 'I
' . '»
1 '-
! •
• .
'
•: •
>
!j
|
i;
•
.
V '
% '
'4y
\ ^
"
.._
^..
V^V
V
•
bfc
[i. "^
^»a«
•^
;^
* _ i >_
kH«»*»«i»»,
••»—•.-,
KMbOO (Ht rfKMXI
B
»^,
MBOIMCK^t^K
— XHASSO=11 kg
KH=0.0088/day
— XNASSO^l.l 1kg
KH=0 .«08/da«j
'- XNASSO^.ll kg
KH=0.l38/day
0
58
1BI0 158 288 256 388 350 400 450
TINE SINCE LEiACHATE GENERATION BEGAN
-------
FIGURE 55 •li;;i< A4
COMPARISON OF [EFFECT OF VALUE OF INITIAL XNASSO AND KH
ON NETHANE PRODUCTION, SINGLE-PASS
t.-
10 .
.0
1 ft-
1A .
.4
1 7 -
L .£.
METHANE ,
(L/DAV) l
00
, o
0fc
04 .
.4
07 .
• £
R
-.
^--^^
•
•
„
2±C£2^kiaiM
-"
•
•
^,'''
AlUCMlIUM^Mia
• ^
•
/
— XMASSO-11 ka
KH=B.BB08/DAV
— XMASSO=1.1 kg
KH=B.BB8/DAV
— XMASSO=B.ll kg
KH-R HiR/DAV
!9B 180 158 i»8B 258 308 358 408 45H
TIME SINCE LEACHATE GENERATION BEGAN
-------
FIGURE 56
COMPARISON OF EFFECT OF UALUE OF INITIAL XMASS'O AND KH
ON UOLftTILE ACID CONCENTRATION! RECVCLE
18868
16888
14888
12888
TUA 1000B
(ng/L) 8888 Jf^~
6888
4888
2888
8
£1
»1ASSO::11 kg
KH=:EI.B€IB5/day
XMAS;SO=:1.1 kg
XMASISO=:.ll kg
100 158 ZOO 2:50 30B 350 488 458
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 57
COMPARISON OF B1FFECT OF UALUE OF INITIAL XMASSO AND KH
ON METHANE PRODUCTION, RECVCLE
128 T-
METHANE
(L/day)
XMASSO=: 11 kg
KH=0.0EI85/day
XMASSO-1.1 kg
KH=0.0EI5/daiy
XMASSO=:.ll kg
KH=.85/day
188 158 288 258 388 358 4138 45B
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 58
EFFECT OF SECONDARY RELEASE OF SUBSTRATE
ON UOLATILE ACID CONCENTRAT ION, SINGLE-PASS
J.OUUU -
IbtjUIJ
1 QQCkCI
(ng/L) RRRR
ouUu
XCIQC|
OQQQ
0.
I
r\
A
• .
•
"r...
3 £
•
k_ ...
\.
\
0 1
\. .
XJ
90 1!
*-L
i0 2^
" w^
W 2\
--"T"rr~
>0 3(
J0 3E
. • " "
>0 4f
10 4E
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 59
EFFECT OF SECONDARY RELEASE OF SUBSTRATE
ON METHANE PRODUCTION, SINGLE-PASS
f. •
In
.O
If. .
.O
1A
• 4
1 9 •
1 .L
METHANE
(L/day) * 1
0p .
• O
0A J
• D 1
04 .
07 .
• £
Q
•
—— - "
•
•
•
»
^.
^^ >*
•t
•
•
Jl
/
/r
/
/ •
/
• •
•
0
100 156 200 258 300 350
TIME SINCE LEfiCHATE GEMFJIATION IIKGAN (DAVS)
400
4E.B
-------
production to simualte decreasing short-circuiting with time. There was no
significant change in the volatile acid concentration curve resulting from
this modification for either the single-pass or recycle operational modes.
Thus, it seems that the early behavior of the volatile acid concentration
profile was not influenced greatly by initial low availability of solid
substrate caused by the simulation selected. However, short-circuiting may
be involved in a different manner. It is possible that peaks in volatile
acid concentrations result when pools of volatile acids isolated within the
landfill material are contacted by the flowing moisture during an infiltra-
tion event thereby causing a sudden release of volatile acids. Such
heterogeneities within the landfill would alter the degree of saturation wtih
time and are difficult to predict and model.
Another explanation for the delayed peaking of the voaltile acid
concentration data could be related to the nature of the processes occurring
within the population of acidogens. It shoud be expected that even after
leachate production begins, there is still much transition occurring within
the mLcrobial populations of the landfill cell and that acidogenesis is not
established as quickly as modeled in . a single Monod equation may
be too simplistic to model the changing nature of the relative populations of
the various species of acidogenic bacteria as acidogenesis is established.
On solution to this problem is to avoid modeling this transitionary
period, but instead to focus modeling efforts on the time of the peak
volatile acid concentration and the subsequent volatile acid concentration
profile. Shifting the volatile acid concentration peak forward in time was
accomplished by suppressing acidogen activity for a specified period of time,
e.g., 20 days, as shown in Figure 60. This was also the technique employed
in accouting for pH inhibition of methanogens in GTLEACH-I. This technique
produced a satisfactory simulation of experimental data but could not be
effective without a priori knowledge of the length of the lag inhibition
period.
The discussion in this section has presented modifications to '""TLEACH-I
which would allow its use in a more diagnostic and/or predictive manner.
Development of an unsaturated flow model for GTLEACH-I was concluded to be
critical to future modeling efforts, particularly when extended to full-
scale landfills with intermittent infiltration of moisture. An investigation
of the degradable fraction of municipal solid waste, possibly even
subcategorization of this fraction (e.g., carbohydrate, protein, lipid and
cellulose fractions) would make possible a more accurate model of
hydrolysis/solubilization processes. Finally modifications to the microbial
equation for acidogenesis and methanogenesis which would accout for product
inhibition of acidogenesis and p i hibition of methanogenesis are suggested.
119
-------
FIGURE 60
ro
o
18888 T
TUA
(ng/L)
16888 —
14888—
12888
18888 --
•
8888 -.-
6888 —
4888
2888
a .
8
EFFECT OF LAG IN INITIATION OF AC IDOGEHESIS
ON UOLATILE ACID CONCENTRATION, SINGLE-PASS
f\
.\
1 •
•
»
•
V
Y •
\
V
\
.
\-.-
H. "
X,
H^__
•*•"•*—*
50
100 150 288 258 386 358
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
488
4561
-------
SECTION 9
FITTING GTLEACH-I TO SECOND SET OF EXPERIMENTAL DATA
As a final part of this investigation, the GTLEACH-I model was applied
to a second set of experiments conducted several years earlier in the same
containers used by Schaffer (1986) and Yari (1986). In experiments by Chang
(1982), 66 kg of shredded municipal refuse on a dry basis were packed into
170 liters, in comparison with the 55 kg into 170 liters employed by Schaffer
(1986). Initial field capacities established in the Chang (1982) and Schaffer
(1986) experiments were similar. The rate of moisture application in the
single-pass experiment conducted by Chang (1982) was one liter per week as
compared to Schaffer's six liters per week. Chang's recycle experiment
included the net addition of 0.5 L per week of deionized water as well as
complete recycle of leachate. Therefore, the average weekly flow through the
recycle cell increased by 0.5 L/wk throughout the experiment. Another major
difference between the Chang (1982) experiment and the Schaffer (1986) and
Yari (1986) experiments was that although methanogenesis was inhibited during
the first 20 to 50 days after leachate generation began, it was established
without seeding in the Chang (1982) experiment.
The single-pass and recycle volatile acid profiles for the two Chang
(1982) experiments are shown in Figure 61. A comparison of this figure with
Figure 62 showed that there was much greater similarity between the volatile
acid concentration profiles for the two modes in the Chang experiment than
the Schaffer (1986) and Yari (1986) experiments. This was due to diminished
effects of washout resulting from lower moisture application rates in the
Chang (1982) single-pass experiment which caused the single-pass system to
behave more like the recycle system than was the-case in the Schaffer (1986)
and Yari (1986) experiments.
The results of simulation of the volatile acid concentration profile for
the Chang (1982) single-pass experiment employing the same kinetic parameters
as applied in the separate simulation of Schaffer (1986) and Yari (1986)
single-pass data are presented in Figures 63 and 64. There was an
overestimation in the simulation of approximately 200 days for the time to
virtually complete removal of the volatile acids from the system. Likewise,
there was a gap of nearly 100 days between the peak in methane production
shown in the experimental data and that of the simulation. The magnitude of
the methane production curve was also grossly overpredicted. (Problems with
overprediction of methane production may have been due in part to experimen-
tal difficulties, possibly a leak in the landfill cell.)
Adjustments to the input kinetic parameters were made in order to
improve the fit of GTLEACH-I simulations to the Chang (1982) single-pass
121
-------
FIGURE 61
COMPARISON OF SINGLE-PASS AND RECYCLE VOLATILE ACID CONCENTRATION
DMA COLLECTED BV CHANG (1982)
2500fl r-
NJ
NJ
TUft
(ng/L)
SINGLE-PASS
RECYCLE
6 26 40 60 80 10)0 126 146 160 1H0 200
TINE SINCE LEACHATE GENERATION BEGAN (DAVS)
-------
FIGURE 62
COMPARISON O'F SINGLE-PASS AhID' RECYCLE VOLATILE ACID' CONCENTRATION
DATA COLLECTED BY SCHIAFFER (1986) AND YARI (1986)
TUA
§ ("9/L) 8000
RECYCLE
— SINGLE-PASS
8
250 3130 350 400 4158
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 63
COMPARISON OF SIMULATED UOLATILE ACID CONCENTRATION
(USING PARAMETERS FROM SCHAFFER (1986) AIND YARI (1986))
UITH EXPERIMENTAL DATA BY CHANG (1982), SINGLE-PASS
£DOQO -
7RRRA -
icncta .
latJiJO
TVA
(ma/L)
iaaaa .
loDOO
caaa .
DUOO
A -
•
r~ -
/••'
•
^~ — • — -.^-
•
• • .
"""""^ "*'"'-»-».
• • • • •
\
^X "
^s,
50 100 158 ?m 250
TIME SINCE LEACHATE GENERATION BEGiAIN (DAYS)
3HB
-------
FIGURE 64
COMPARISON OF SIMULATED METHANE PRODUCTION
(USING PARAMETERS FROM SCHAFFER (1986) AND VARI (1986))
UITIH EXPERIMENTAL DATA BY CHAN'G d'982), SINGLE-PASS
NETHANE
(L/day)
10)0
150
28iB
25B
3B8
TIME SINCE LEACHATE GENERATION BEGAM (DiftYS)
-------
data. The results of sequential adjustments in the parameters are demonstra-
ted in Figures 65 through 67. In Figure 65, the maximum substrate utiliza-
tion coefficient for methanogens, Q^, was increased from 1.9 to 2.5/day
resulting in a shortening of the time to removal of volatile acids. In Figure
66, the lag time before initiation of methanogenesis was decreased from 50 to
10 days, further shortening the time to removal of volatile acids. Finally,
in Figure 67 the yield coefficient for methanogens was increased from 0.02 to
0.04, sharply increasing the rate at which volatile acids were removed.
The results of simulation of the Chang (1982) recycle experiment
employing the kinetic parameters from the separate fit of the Schaffer (1986)
and Yari (1986) data are presented in Figures 68 and 69. These results show
a somewhat improved fit than was the case for the single-pass simulation.
There was a 50-day overestimation in the time to removal of volatile acids.
The peak in methane production was also 50 days later than in the
experimental data. In addition, the peak in methane production still
continued to be grossly overestimated.
Adjustments of parameters were made to improve the fit of GTLEACH-I
simulation of the Chang (1982) recycle data. The effects of increasing the
maximum substrate utilization rate for methanogens, Q^, from 1.9/day to
2.2/day are shown in Figures 70 and 71. The result is to shorten the time to
removal of volatile acids and correspondingly shorten the time to peak
methane production. A subsequent change in initial hydrolysis product
concentration, SHO, from 20,000 mg/L to 15,000 mg/L produced the results
shown in Figures 72 and 73. The effects were to slightly decrease the
magnitudes of both volatile acid concentration and methane production
profiles. The final result was a reasonably good fit of the volatile acid
concentration data, however, the magnitude of the methane production curve
continued to be overestimated, possibly explained by leaks in the experimen-
tal cells which may have resulted in failure to capture the total gas volume
generated.
The difference between the single-pass kinetic parameters fit to the
Schaffer (1886) and Yari (1986) data and those fit to the Chang (1982) data
were:
the maximum substrate utilization rate for methanogens was 1.9/day
for the Schaffer (1986) and Yari (1986) simulation, while for the
Chang (1982) simulation it was 2.5/day; and
the yield coefficient for raethanogens in the Schaffer (1986) and
Yari (1986) simulation was 0.02 and for the Chang (1982) simulation
it was 0.04.
This resulted in an overall increase by 2.5 times in the maximum growth rate
constant for methanogens of the Chang (1982) simulation over the Schaffer
91986) and Yari (1986) simulation. This may reflect differences in the
inhibitory natures of the two environments. Due to higher rates of moisture
application in the Schaffer (1986) and Yari (1986) experiments, the landfill
to be in contact with the low pH leachate longer with more likelihood of
stronger inhibition.
126
-------
FIGURE 65
15880
(mg/L)
10000
B
COMPARISON OF SIMULATED VOLATILE ACID CONCENTRATION
UITH EXPERIMENTAL DATA BY CHANG (1982), S IINGLE-PASS
QM=2.5/DAV
5H 108 158 2H0 258
TUNE SINCE LEACH ATE GENERATION BEGAN (DAYS)
3B0
-------
FIGURE 66
COMPARISON OF SIMULATED VOLATILE ACID CONCENTRATION
UITH EXPERIMENTAL DATA BY CHANG (1982), SINGLE-PASS
QM=2.5/DAY, LAG=18 DAYS
M TUA
<» (ng/L)
^sooo -
•?afiaa
ZoUUD
ic ana
IDtJOU •
iftaaa
ItJorJu ~
c,f\f\f\ .
H
f
•
•
r- — ^
F-' "
\ ;;
**
X
..N
8 1(
V. . . .
30 1'
ie 2(
10 2!
;e ae
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 67
S TUA
(ng/L)
COMPARISON OF SIMULATED VOLATILE ACID CONCENTRATION
UITH EXPERIMENTAL DATA BY CHANG (1982), SINGLE-PASS
QM=2.5/DAY, Y=0.04, LAG=10 DAYS
\
160 150 ?m 250
TINE SINCE LEACIHATE GENERATION BEGAN (DAYS)
300
-------
FIGURE 68
COMPARISON OF VOLATILE ACID CONCENTRATION SIMULATION
(USING PARAMETERS FROM SCHiAFFER (19:86) AND YiARI (1986))
UITH EXPERIMENTAL DATA BY CHANG (1982), RECYCLE
TUA
Ug/L)
180
\
158
zee
258
368
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 69
COMPARISON OF SIMULATED METHANE PRODUCTION
(USI ING PARAMETERS FROM SCHAFFER (1986) AMD MR I (1986))
IUITH EXPERIMENTAL DATA BY CHANG (1982), RECYCLE
METHANE
(L/day)
90
88
78
68
50
^g
38
28
18
8
8
58
108
158
288
250
308
TINE SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 70
C'OHPARISON OF SIMULATED VOLATILE ACID CONCENTRATION
UITH EXPERIMENTAL DATA BV CHANG (1982), RECYCLE
QM=Z.2/daiy
TUA
(mg/L)
iee
158
2BB
258
388
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
-------
FIGURE 71
COMPARISON OF SIMULATED METHAINE PRODUCTION
UITH EXPERIMENTAL DATA BV CHANG (19821), RECYCLE
1=2.2/day
120-r
180
OJ
U)
METHANE
(L/day)
B
308
TIME SINCE LEACHfiTE GENERATION BEGAN (HAVS)
-------
FIGURE 72
TUA
(wg/L)
COMPARISON OF SIMULATED VOLATILE ACID CONCENTRATION
UITH EXPERIMENTAL DATA BV CHANG (1982), RECYCLE
QM=2.2/day, SH8=15,8e0 mg/L
so iee 150 zee 258
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
308
-------
FIGURE 73
8
COM PARA I SON 'OF SIMULATED NETHAINE PRODUCTION
WITH EXPERIMENTAL DATA BY CHANG (1982), RECYCLE
QM=2.2/day, SIHO=15,888 ng/L
TUA
(mg/L)
LUU
98
on
oo
78
68
58
48
38
28
18
a
58 IBB 158 2810 258
TIME SINCE LEACHATE GENERATION BEGAN (DAYS)
308
-------
The only difference between the kinetic parameters fit to the Schaffer
(1986) and Yari (1986) recycle experiments and those fit to the Chang (1982)
recycle experiment was that the maximum substrate utilization rate for
methanogens was 1.9/day for the Schaffer (1986) and Yari (1986) simulations
and 2.2/day for the Chang (1982) simulation. This again may be related to
the increase in moisture application rates of the Schaffer (1986) and Yari
(1986) experiments which increased the opportunity for contact of the
methanogens with the low pH leachate thereby increasing inhibition. If
indeed this was the explanation for the difference in simulated maximum
growth rate for methanogens in the two experiments, it must be inferred that
simply lagging the initiation of methanogenesis was not sufficient for
modeling pH inhibition. The lag was an oversimplification of the pH
inhibition process. A more accurate representation of the process would
require a relationship between moisture movement per unit volume of the
landfill and the degree of inhibition of the methanogenic population, i.e.,
an unsaturated flow model coupled with a biological model containing an
inhibition function.
Thus the only major difference between the microbial kinetics observed
in the Schaffer (1986) and Yari (1986) experiments and the Chang (1982)
experiments, in both recycle and single-pass cases, was in the methanogen
growth rate constant. The hydrolysis/solubilization and acidogenesis
constants were consistent between the two experiments. This lends strong
support to the three-step model and provides impetus for improvement of the
inhibition function at least in the third step.
136
-------
SECTION 10
SUMMARY AND DISCUSSION OF MODEL GTLEACH-I
The model presented in this report is a three-step simulation of the
general anaerobic microbial processes occurring within landfill systems. A
reasonable model of volatile acid production and removal has been presented,
laying the foundation for future work expanding into the inorganic aspects of
landfill stabilization and eventually into a model of pH and pE which would
greatly enhance prediction of the fate of hazardous organic constituents co-
disposed with municipal waste in landfills. A major accomplishment of the
model was the application of kinetic constants for hydrolysis/solubilization
and acidogenesis in the simulation of two sets of experimental data having
substantially different moisture application rates. The model was also shown
to be a useful diagnostic tool in examining differences in microbial popula-
tions resulting from different leachate management options.
Two major areas of improvement are recommended for GTLEACH-I. The first
improvement would be in the microbial model of methanogeneis, since the
simplistic lag model put forth in this work was insufficient to account for
pH inhibition of methanogenesis. Development of a pH inhibition model would
help to correlate degree of inhibition with pH and the rate of moisture flow
through the landfill.
Beyond development of an improved model of pH inhibition of methanogens,
the other major area of improvement to GTLEACH-I lies in modeling of moisture
flow within landfills. Comparison of constant flow versus intermittent flow
dramatized the -importance of accurate flow simulation in modeling landfill
stabilization. Eventual coupling of the microbial model developed in
GTLEACH-I with a companion unsaturated flow model would do much to enhance
the ability of the model to simulate behavior of landfills during
stabilization.
The inability of the model to simulate fluctuations in volatile acid
concentrations during the first 50 or so days of landfill life following
leachate generation was attributed to effects of short-circuiting and the
transitional nature of the microbial populations during this period. Due to
the difficult, yet short-lived nature of this problem, it is recommended that
immediate modeling efforts focus on the subsequent volatile acid profile.
Further research into the composition of municipal solid waste in terms
of degradable fraction, as well as carbohydrate, protein, lipid and cellulose
subtractions is also recommended. Acquisition of such data would allow for
improvement of the hydrolysis/solubilization step in the r,TLEACH-I.
137
-------
Furthermore, such data is necessary in order to project the length of time
required to complete landfill stabilization and the associated magnitude of
leachate strength and migration potential.
138
-------
REFERENCES
Andrews, J.F. and E.A. Pearson, 1965. "Kinetics and Characteristics of
Volatile Acid Production in Anaerobic Fermentation Processes". Inter-
national Journal of Air and Water Pollution, v. 9, pp. 439-461.
Andrews, J.F. and S.P- Graef, 1971. "Dynamic Modeling and Simulation of the
Anaerobic Digestion Process". Anaerobic Biological Treatment Processes.
Advances in Chemistry Series 105. American Chemical Society, Washington,
DC, pp. 126-162.
Barber, C., 1981. Leaching in Hazardous Wastes in Landfills. Water
Research Centre, Environmental Protection Division, Steverage Laboratory,
London, United Kingdom.
Bauchop, T., and S.R. Elsden, 1960. "The Growth of Micro-Organisms in
Relation to Their Energy Supply". Journal of General Microbiology, v. 23,
pp. 457-469.
Bernardes, Gyro, Jr., 1984. Modeling Aspects of Domestic Solid Waste
Landfilling Related to Heavy Metal Fixation. Masters Thesis, University of
Texas at Austin.
Brooks, R.H. , and A.T. Corey, 1964. Hydraulic Properties of Porous Media,
Hydrology Paper No. 3, Colorado State University, Fort Collins.
Brunner, D.R., 1979, "Forecasting Production of Landfill Leachate",
Municipal Solid Waste: Land Disposal, Fifth Annual Research Symposium,
U.S. EPA Municipal Environmental Research Laboratory, Cincinnati, Ohio,
EPA-600/9-79-023a.
Campbell, G.S., 1974. "A Simple Method for Determining Unsaturated Con-
ductivity from Moisture Retention Data", Soil Science, v. 117, pp. 311-314.
Cappenberg, T.E., 1975. "A Study of Mixed Continuous Cultures of Sulfate-
Reducing and Methane Producing Bacteria". Microbial Ecology, v. 2, 60-72.
Chang, K. , 1982. Go-Disposal of Low-Level Radioactive Waste Within Sanitary
Landfills' A Thesis in partial requirement for the degree of Doctor of
Philosophy, School of Civil Engineering, Georgia Institute of Technology.
Clapp R.B., and G.M. Hornberger, 1978. "Empirical Equations for Some Soil
Hydraulic Properties." Water Resources Research, v. 14, pp. 601-604.
139
-------
Eastman, J.A. and J.F. Ferguson, 1981. "Solubilization of Particulate
Organic Carbon During the Acid Phase of Anaerobic Digestion". Journal
of the Water Pollution Control Federation, v. 53, pp. 352-366.
Erdogan, H., 1982. "Prediction Models for Sanitary Tandfills". Environ-
mental Engineering.
Fungaroli, A.A. and R.L. Steiner, 1979. Investigation of Sanitary Land-
fill Behavior, Vol. I. U. S. EPA Municipal Environmental "esearch Tabor-
atory, Cincinnati, Ohio. KPA 600/2-79-053a.
Ghosh, S. and others, 1975. "Anaerobic Acidogenesis of Wastewater Sludge".
Journal of the Water Pollution Control Federation, v. 47, pp. 30-45.
Ghosh, S. and D.L. Klass, 1978. "Two-Phase Anaerobic Digestion". Process
Biochemistry, v. 13, p. 15.
Grady, C.P.L. , Jr. and H.C. Lim, 1980. Biological Wastewater Treatment;
Theory and Applications. Marcel Dekker, Inc., New vOrk, NY.
Gujer, W. and A.J.B. Zehnde, 1982. "Conversion Processes in Anaerobic
Digestion". Presented at lAWPR-Seminar on Anaerobic Treatment, Copenhagen,
Denmark.
Hartmann, L.W., 1981. Anaerob Spildevandsbehandling. (Anaerobic Treat-
ment of Waste Water) (in Danish). Masters Thesis, Department of Sanitary
Engineering, Technical University of Denmark.
Henze, M. and P. Harremoes, 1983. "Anaerobic Treatment of Wastewater in
Fixed Film Reactors — A Literature Review", Water Science and Technology
v. 15, pp. 1-101.
Houle, M.J., D.E. Long, R.E. Bell, J.E. Soyland, and R.R.Grabbe, 1977.
"Effect of Municipal Landfill Leachate on the Release of Toxic Metals from
Industrial Wastes", Management of Gas and Leachate in Landfills: Proceed-
ings of the Third Annual Municipal Solid Waste Research Symposium, U.S. EPA
Municipal Environmental Research Laboratory, Cincinnati, Ohio
EPA 600/9-77-026.
IMSL, 1984. User's Manual, Math/PC-Library; Fortran Subroutines for
Mathematical Applications on a Personal Computer. IMST , fnc. 198''.
Johnson, T.M., T.H. Larson, B.L. Herzog, K. Cartwright, C.J. Stohr,
and S.J. Klein, 1983. A Study of Trench Covers to Minimize Infiltration
of Waste Disposal Sites, U.S. Nuclear Regulatory Commission, Washington,
DC, NUREG/CR-2478.
Kaspar, H., 1977. PhD Thesis, Swiss Federal Institute of Technology,
Nr. 5984 (in Zehnder, 1978).
Khaleel, R. and T.C. Yeh, 1985. "A Galukin Finite-Element Program for
Simulating Unsaturated Flow in Porous Media". Groundwater, v. 23,
pp. 90-96.
140
-------
King, L.G., 1965, Description of soil characteristics for partially
saturated flow. Soil Science Society of America Journal, v. 29,
pp. 359-362.
Lappella, E.G., 1980. "Modelling of Water and'Solute Transport Under
Variably Saturated Conditions." Proceedings of the Interagency Workshop
on Radioactive Waste Modelling. Denver, Colorado.
Lawrence, A.L. and P.L. McCarty, 1969. "Kinetics of Methane Fermentation
in Anaerobic Treatment". Journal of the Water Pollution Control Federation
v. 41, pp. Rl - R7.
Lettinga,
-------
Narasimhan, T.N. , 1975, A Unified Numerical Model for Saturated-Unsaturated
Groundwater Flows, Doctor of Philosophy dissertation, University of
California, Berkeley, California.
Neuman, C.P-, 1973, ''Saturated-Unsaturated Seepage by Finite Elements".
Journal of the Hydraulics Division of the American Society of Civil Engi-
neering v. 99, pp. 2233-2250.
Neuman, S.P. , R.A. Feddes, and E. Bresler, 1975, "Finite Element Analysis
of Two-dimensional Flow in Soils Considering Water Uptake by Roots:
I Theory". Soil Science Society of America Proceedings, v. 39,
pp. 224-230.
Neuman, S.P. , and T.N. Narasimhan, 1977, "Mixed Explicit-Implicit Iterative
Finite "lement Scheme for Diffusion-Type Problems: I. Theory". Inter-
national Journal for Numerical Methods in Engineering, v. 11, pp. 309-323.
Pickens, J.F., R.W. Gillham, and D.R. Cameron, 1979, "Finite-Element
Analysis of the Transport of Water and Solutes in Tile-Drained Soils".
Journal of Hydrology, v. 40, pp. 243-264.
Pohland, F.G. 1975. Sanitary Landfill Stabilization with Leachate Recycle
and Residual Treatment, U.S. EPA, Municipal Environmental Research Labor-
atory. Cincinnati, Ohio. EPA 600/2-75-043..
Pohland, F.G., 1980. "T.eachate Recycle as a Management Option," Proceed-
ings of the Leachate Management Seminar, University of Toronto, Toronto,
Ontario, Canada.
Pohland, F.G., J.T. Dertien, and S. B. Ghosh, 1983. "Leachate and Gas
Quality Changes Touring Landfill Stabalization of Municipal Refuse,"
Proceedings of the Third International Symposium on Anaerobic Digestion,
Boston, Massachusetts.
Pohland, F.G., J.P. Gould, R.E. Ramsey, B.J. Spiller, and W.R. Esteves,
1981. "Containment of Heavy Metals in Tandfills with Leachate Recycle,"
Land Disposal; Municipal Solid Waste, Proceedings of the Seventh Annual
Research Symposium, U.S. EPA Municipal Environmental Research Laboratory,
Cincinnati, ^hio, EPA 600/9-81-002a.
Pohland, F.G., S.R. Harper, K.C. Chang, J.T. Dertien, and E.S.K. Chian,
1985. "Leachate Generation and Control at Landfill Disposal Sites", Water
Pollution Research Journal of Canada, v. 20, no. 3.
Pohland, F.G., D.E. Shank, R.D. Benson, and H.H. Timaerman, 1979. "Pilot
Scale Investigations of Accelerated T.andfill Stabilization with Leachate
Recycle in Municipal Solid Waste Land Disposal", Municipal Solid Waste.
Land Disposal, Fifth Annual Research Symposium. U.S. Municipal Environ-
mental esearch Laboratory, Cincinnati, Ohio. EPA 600/9-79-023a.
142
-------
Pohland F.G. and S.R. Harper, 1985. Critical Review and Summary of
Leachate and «*« Production From Landfills. U.S. EPA Hazardous Waste
Engineering Research Laboratory, Cincinnati, Ohio EPA 600/2-86-073.
Pol L.W. Hulshoff and others, 1982. "Granulation in UASB-Reactors".
Presented at lAWPR-Seminar on Anaerobic Treatment, Copenhagen, Denmark.
Reeves, M. , and J.O. "uguid, 1975, Water Movement Through Saturated-
Unsaturated Porous Media; A Finite-Element Galerkin Model, Oak Ridge
National Laboratory, Oak Ridge, Tennessee, ORNL-4927.
Richards, L.A. , 1931. "Capillary Conduction of Liquids Through Porous
Mediums". Physics, v. 1, pp. 318-333.
obinson, H.D., B. Tech, C. Barber and P.J. Maris, 1982. "Generation
and Treatment of Leachate from Domestic Wastes in Landfills". Water Poll-
ution Control, 1982, p. 465.
Rogowski, A.S., 1971, "Watershed Physics: Model of the Soil Moisture
Characteristic", Water Resources Research, v. 7, pp. 1575-1582.
Schaffer, T.R., 1986. Fate of Selected Organic Priority Pollutants During
the Acid Formation Phase of Landfill Stabilization. A Special Problem
in partial fulfillment of the Requirements for the Degree of Master of
Science in the School of Civil Engineering, Georgia Institute of
Technology.
Schroeder, A.C., A.C. Gibson, and M.D. Smolen, 1983, The Hydrologic
Evaluation of Landfill Performance (HELP) Model, Volumes I and II, U.S.
EPA Municipal Environmental Research Laboratory, Cincinnati, Ohio.
EPA 530-SW-009 and EPA 530-SW-010.
Smith, M.R. and R.A. Mah, 1978. "Growth and Methanogenesis by Methano-
sarcina Strain 227 on Acetate and Methanol". Applied Environmental
Microbiology, v. 36, pp. 870-879.
Speece, R.E. and P.L. McCarty, 1964. "Nutrient Requirements and Biological
Solids Accumulation in Anaerobic Digestion". Advances in Water Pollu-
tion Research; Proceedings of the International Conference, Vol. 2,
''xford, nited "ingdom.
Speece, .E., G.F. Parkin, and D. Gallagher, 1982. "Nickel Stimulation of
Anaerobic'nigestion". Environmental Studies, Drexel University,
Philadelphia.
Straub W.A. and D.R. Lynch, 1982. "Models of Landfill Leaching: Moisture
Flow and inorganic Strength". Journal of the Environmental Engineering
Division of the American Society of Civil Engineers, v. 108, pp. 251-268.
Stumm W. and J.-J. Morgan, 1970. Aquatic Chemistry — An Introduction
Finrf,al<»infl Chemical Equilibria in Natural Waters. Wiley-Interscience,
London, United ingdoi.
143
-------
Swartzbaugh, J.T., R.L. Hentrich, Jr., and G. Sabel, 1979. Evaluation of
Landfilled Municipal Industrial Solid Wastes. Report to EPA Contract
No. 68-03-2120, Cincinnati, Ohio.
Sykes.R.M., 1975. "Theoretical Heterotrophic Yields". Journal of the
Water Pollution Control Federation, v. 47, pp. 591-600.
Tchobanoglous, G., H. Theisen, and R. Eliassen, 1977. Solid Wastes;
Engineering Principals and Management Issues. McGraw-Hill, New York, NY.
Van den Berg, L., 1977. "Effect of Temperature on Growth and Activity of
a Methanogenic Culture Utilizing Acetate". Canadian Journal of
Microbiology, v. 23, pp. 898-902.
Walsh, J.J., W. G. Vogt, and W.M. Held, 1983. "Demonstration of Landfill
Gas Enhancement Techniques in Landfill Lysimeters," Proceedings of the
GRCDA Sixth International Landfill Gas Symposium. Industrial Hills,
California.
Yari, S., 1986. Fate of Selected Organic Porosity Pollutants During the
Methane Fermentation Phase of Landfill Stabilization. A Special Problem
in partial fulfillment of the requirements for the Degree of Master of
Science in the School of Civil Engineering, Georgia Institute of
Technology.
Yeh, G.T., and D.S. Ward, 1980, FEMWATER; A Finite-Element Model of Water
Flow Through Saturated-Unsaturated Porous Media: Oak Ridge National
Laboratory, Oak Ridge, Tennessee. ORNL-5567.
Young, J.C. and P.L. McCarty, 1967. "The Anerobic Filter for Waste Treat-
ment". Proceedings of the 22nd Industrial Waste Conference, Purdue
University, Lafayette, Indiana.
Zinder, S.H. and R.A. Mah, 1979. "Isolation and Characterization of a
Termophilic Strain of Methanosarcina Unable to Use H2~C02 for
Methanogenesis". Applied Environmental Microbiology, v. 38, pp. 996-1018.
144
-------
APPENDIX A
DATA ASSESSMENT
INTRODUCTION
The Georgia Institute of Technology research team conducted a review of
pertinent available reports of studies on leachate generation and leachate
characteristics. The review focused on municipal solid waste (MSW) landfills
and pilot-scale studies of leachates generated from MSW and codisposal
wastes. Data assessment criteria were investigated to determine the
parameters required to model lechate generation in a MSW landfill. It is
anticipated that these data assessment criteria may be used to evaluate data
from previous landfill investigations or landfill simulation studies. The
assessment criteria may also be used for guidance in the preparation of
future sampling plans for proposed landfill investigations or landfill
simulations.
DATA ASSESSMENT CRITERIA
The generation of leachate from a waste containment structure is a
complex process dependent on a large number of variables. Due to the large
number of variables and the difficulty in isolating the variables to evaluate
potential impact on leachate generation, it is very difficult to quantify and
interpret leachate characteristics. However, the ability to predict the life
of synthetic or natural liner materials, the potential for leakage, and the
impact on the environment requires some quantification of leachate volume and
chemical characteristics based on setting, design, waste, and operational
variables.
Notwithstanding the difficulty in prediction of leachate characteris-
tics the available data indicate that the basic processes of MSW landfill
stabilization are generally independent of setting, waste, and operational
variables. Moreover, the basic processes of biologically-mediated stabiliza-
tion are similar in lysimeters, pilot-scale landfills and full-scale
landfills. However, the extent and rates at which the processes occur and
the resultant effect on leachate constituent concentrations is highly
dependent on these variables.
The data assessment criteria were developed to provide a mechanism to
evaluate the completeness of the data. The data were screened to determine
if sufficient data were present to characterize and model the landfill or
landfill simulation (i.e., pilot-scale landfill or lysimeter) through the
phases of microbial stabilization. The data assessment criteria used in the
145
-------
screening and evaluation of available data are presented and discussed below.
These data assessment criteria are divided into five principal areas:
1) Hydrologic Variables;
2) Waste Containment Structure Design Variables;
3) Waste Characteristics;
4) Operational "rocedures Affecting Leachate Generation; and
5) Monitoring Data.
Hydrologic Variables
The leaching of liquids through waste materials and the subsequent
development of leachate in a landfill is affected by the geologic and
hydrologic settings, climate and hydrology as shown in Figure Al. Land use
data are also important in the evaluation of potential impacts on receptors.
Site Geologic Model—
The geologic setting is typically described using conceptual regional
and site geologic models. The regional geologic model provides a basic
understanding of the depositional environment of the soil and rock units in
the vicinity of the site. The site geologic model provides a detailed,
idealized summary of data obtained during the site investigation. These data
typically include boring logs, visual descriptions (micro-logs) of oil
specimens and rock cores, and laboratory analyses.
The boring logs may be linked using fence or block diagrams to evaluate
the continuity of individual soil or rock units. Of particular interest are
the continuity of water bearing units, which may contribute seepage through
the liner into the leachate collection system or provide pathways for the
release of contaminants, and confining layers, which act as vertical barriers
for contaminant migration. The leachate constituents may also interact with
the soils, resulting in attentuation, retardation, or fractionation.
The primary properties of interest in the model are those which relate
to the transmission of fluid through the layer. These are the soil density,
water content, gradation, percent passing the Number 200 sieve, and the
hydraulic conductivity or coefficient of permeability. For rock units, the
evaluation of properties affecting transmission of fluid is more difficult.
An indication of the transmissivity of the rock units may be obtained from
core recovery data, in situ evaluation of hydraulic conductivity and regional
information on well yield and gradients.
The effectiveness of confining layers and soil barrier layers is
typically based on an evaluation of the soil's hydraulic conductivity.
However, when soil materials come in contact with high strength leachates,
the cation exchange capacity, initial pore water chemistry, state of stress,
structure and fabric of the soil may also be important.
The geologic model is required when the landfill is constructed below the
original ground surface in such a way that infiltration of ground water may
contribute to the quantity of leachate produced.
146
-------
PRECIPITATION
EVAPOTRANSPIRATION
GROUND-WATER
INTRUSION
LEACHATE
W1THDRAML
A,. WAIER
-------
Site Hydrogeologic Model--
The regional and site geologic models provide a conceptual description
of the occurrence and use of ground water in the vicinity of the site. The
regional hydrogeologic model is a companion of the regional geologic model
which identifies the major aquifers in the region. The model provides
descriptions of the direction of ground water flow, velocity, confinement,
and basic water quality.
The site hydrogeologic model describes the occurrence of ground water in
the soil and rock units at the site. The water bearing units are typically
classified as perched water zones, ground water table zones, or unconfined,
leaky or confined aquifers. Data from slug tests or pump tests conducted at
the site are used to evaluate the in situ hydraulic conductivities,
transmissivities, and storativities of the individual units or layers. Water
level data are typically obtained from monitoring wells at the site to
evaluate the potentiometric surface for each aquifer as a function of time.
Ground water specimens are also obtained for baseline chemical analyses.
These analyses typically include all the parameters to be monitored in the
monitoring system. The site hydrogeologic model typically focuses on those
water bearing units which are in the same horizon as the landfill and the
water bearing unit immediately below the landfill.
Ground water use data may also be important in the evaluation of
potential impacts on receptors. These data typically include the locations,
depths and withdrawal rates of wells in the vicinity of the site, and the
locations and types of other potential receptors such as environmentally
sensitive areas, basins, lakes and rivers.
Since ground water may contribute to the available water driving the
development of the leachate at a landfill, it is necessary to evaluate the
regional and site hydrology and to assess their impact on a specified site.
The site hydrogeologic model is required when the waste containment facility
is constructed below the original ground surface in such a way that
infiltration of ground water may contribute to the quantity of leachate
produced.
Climatic Data--
The infiltration and percolation of rainfall into landfills provides the
major driving force behind the generation of leachate. Therefore, climatic
data affecting the quantity of water percolating into a landfill are
essential in the evaluation of leachate quantities and characteristics.
These climatic data include daily rainfall, solar index, plant uptake, daily
temperature, relative humidity and wind speed over time. The climatic data
are required whenever the waste containment structure is constructed in such
a way that it is open to the elements.
Site Hydrology--
The management of surface water at the site has a major impact on the
quantity of water which may enter a landfill. A site plan which details the
surface water management strategy should be reviewed to evaluate the
potential for surface water run-on. In addition, it is important to obtain
data relating to the run-off from the final soil cover. These data would
148
-------
include the type of vegetative cover, the type of soil and the nature of the
cover.
iption of surface features such as streams, rivers, lakes and
basins should also be provided. This information is typically used in the
site hvdrogeologic model. The hydrologic data are required whenever there is
a possibility of surface run-on affecting the quantity of leachate produced
in the landfill-
Waste Co"f-ainment Facility Design Variables
Type of Landfill"
The type of landfill and its purpose have a major impact on the type,
quantity, and rate of generation of constituent concentrations of the
leachate. The evaluation focuses on three types of landfills: hazardous
waste landfills permitted under RCRA; codisposal landfills which contain both
municipal refuse and quantities of hazardous or industrial waste; and, MSW
landfills which contain primarily municipal refuse. The design variables and
model evaluations described in this report are primarily related to the
evaluation of leachate characteristics at MSW landfills. However, the
approach and the data requirements are very similar for codisposal and
hazardous waste landfills as well as pilot-scale or simulated landfills and
lysimeters.
Cell Geometry- -
Most landfills are traditionally separated into a series of cells which
reduce the initial volumes of water to be handled in a leachate collection
system. The cells may be excavated and constructed below the existing ground
surface or embankments may be constructed and the cell installed above the
ground surface. The configuration of the cell is primarily a function of the
local site geology, hydrogeology (depth of the water table), and waste
management factors (e.g., site access, size of the facility).
The primary cell geometry variables affecting the leachate are the
length, width and depth of the cells, as these variables are related to the
waste quantity, and the contact time of the water percolating through the
landfill Side slope data, embankment design data, cut and fill areas and
cell locations should also be *™1"' °£
locations^ "e teacnX. Suction points can be evaluated or identified.
• eters, the basic cell geometry, method of construction,
*or lysi t treatnient of boundaries is required to evaluate the
quantity of ™fe> rcoiating through the landfill. It is also important
contact time ot J^J for liquid or gas leakage from the lysimeter and to
to assess the p nteraction between the waste and the structure sidewalls.
evaluate potential ±"*-
Liner D68*8"."; he soil liner is important for the evaluation of
n16 hate compatibility and the analysis of potential migration from the
liner/ leacna mary variables affecting the performance of the liner are
landfill- classification, the layer thickness, the density and water
the soil type
149
-------
content, the soil gradation (and hydrometer analysis), the liquid and plastic
limits, and the hydraulic conductivity. The field quality control (QC) data
are also important in the evaluation of the continuity of the clay liner.
Where synthetic liners are used in conjunction with a soil layer to form a
composite liner, it is also important to assess the effects of interaction
between the synthetic liner and soil. Soil liner design information is not
typically required for lysimeters.
Liner Design (Synthetic)--
Synthetic liner design data are required whenever synthetic liner
materials are used in the construction of the waste containment structure.
The effectiveness of a synthetic liner is primarily a function of the polymer
type and density, a method of seaming, the liner thickness, stress/strain
behavior, and the installation methodology and quality control. In landfills
where a leak detection layer is installed, the performance of the liner may
be evaluated directly.
The chemical compatibility of the synthetic liners is typically
evaluated using the EPA 9090 evaluation. In the absence of data from the
9090 test, liner compatibility may be evaluated for specific waste streams
based on general polymer compatibility data provided by the manufacturer.
Leachate Collection, Leak Detection and Underdrain Layers--
The leachate collection system design information is required for all
waste containment structures. The design of the leachate collection, leak
detection and underdrain layers affects the quantity of water which is
collected in the leachate collection system, the contact time between the
liquid and the waste, and the time required to reach field, capacity in the
landfill. The primary variables affecting the performance of the drainage
layers are the flow capacity and the thickness of the drain. For aggregate
drains, the type of aggregate, grain size distribution, gradient, and
hydraulic conductivity control the flow capacity of the drain. For synthetic
drainage materials, the type of drain, the polymer type and density,
gradient, and the transmissivity under load control the flow capacity of the
drain. Flow capacity reduction due to microbial growth in the drainage
layers should be assessed.
When geotextiles are used for filtration or separation, the flow through
the geotextile into the drainage layer becomes important. The primary
variables affecting flow normal to the geotextile are the type, thickness,
porosity and hydraulic conductivity ratio of the geotextile. Flow reduction
due to microbial growth in the drainage layers should be addressed in the
evaluation of leachate management for the landfill.
Final Cover--
The design of the final cover system has a major impact on the
generation of leachate. Since the driving force behind the generation of
leachate is infiltration and percolation, the design of the cover system,
which limits the amount of water available for percolation, is critical to
the evaluation of leachate generation as a function of time. The primary
variables required to evaluate infiltration through the cover are the final
cover profile, the soil types, the average densities and water contents of
150
-------
each soil layer at placement, the liquid and plastic limits, grain size
distribution and vertical hydraulic conductivity of the soil layer, and the
type of vegetative cover. In addition to these parameters, the overall
surface water management program should be investigated. This would include
the design and layout of the surface water management facilities on the
cover.
Since the quantity of water which percolates through the landfill is
much greater prior to closure, it is important to determine when the final
cover is placed on each cell of the landfill. The cover design information
is required for all waste containment structures which are constructed and
operated out of doors.
Daily Cover--
The placement of daily or intermediate cover affects the migration of
rain water and leachate through the landfill. The primary variables required
to assess the impact on migration are the type of soil, thickness and
locations of the soil layers, the compactive effort and the hydraulic
conductivity. In many landfills, the daily cover is removed in some areas,
particularly adjacent to the cover on a slope, to promote vertical
percolation. It is important to record the locations of all areas where the
daily cover is removed.
Daily cover information is required whenever intermediate soil layers
are placed between waste layers in the landfill.
Gas Collection System--
Where gas collection systems have been installed, preferential paths for
the migration of leachate may have been created. Therefore, the design of
the gas collection system, its layout and the probability of leachate
migration through or around the gas collection system should be assessed.
On-Site Surface Structures or Appurtences
Surface impoundments and tanks are often used to temporarily store
leachate or storm water prior to collection or treatment. These impoundments
and tanks may leak, creating a ground-water mound near portions of the
landfill. The impact of these facilities on the potentiometric surface and
groundwater quality should be assessed. The primary variables required to
evaluate the potential for leakage from impoundments or ponds are the size of
the pond, the depth of fluid in the pond, the pond location, and the liner
type, thickness, and hydraulic conductivity.
Shallow groundwater monitoring wells in the vicinity of the impoundment,
tanks or ponds should be evaluated to determine if leakage has occurred.
Water level readings from the wells should be recorded on a regular basis to
assess the impact of potential leakage on the potentiometric surface.
These data are only required when the waste containment facility is
constructed below the original ground surface and there is a reasonable
possibility that flow from the surface impoundments or tanks may contribute
to the generation of leachate.
151
-------
Waste Characteristics
Waste characteristics have a major impact on leachate composition and
constituent concentrations, stabilization processes, time required for each
phase of stabilization to occur, and the rate of migration of leachate
through the waste. Due to the heterogeneity of typical municipal wastes,
sufficient analyses are required to characterize the chemical composition of
the waste. Such analyses would likely be expensive, and would be applicable
only to the wastes at the landfill where the analyses are performed. In the
absence of such chemical composition data, it has been traditional to
classify wastes according to their origin.
Composition of Solid Wastes--
Pohland and Harper (1985) classified solid wastes into five major source
categories including residential, agricultural, commercial, municipal, and
industrial. Waste originating from each source category are typically
dominated by a limited number of constituents. For example, the major
constituents of residential waste are rubbish, food and garden wastes, paper,
plastics, glass and ash. Household hazardous wastes are also included in the
residential waste category, but are generally relatively minor compared to
the total volume of the major constituents. Agricultural wastes consist
primarily of crop and animal wastes, food wastes, rubbish, and chemicals such
as pesticides, herbicides and insecticides. The major constituents of
commercial waste are rubbish, food waste,construction/demolition debris, ash
and chemical waste residues. Municipal wastes often consist of combinations
of the other categories with an abundance from the residential and commercial
sectors. Industrial waste streams are a function of the industry from which
they originate. These wastes typically consist of biological and chemical
sludges, chemicals, rubbish, ash, construction/demolition debris and chemical
waste residues.
Food wastes, rubbish, garden wastes, crop and animal residues and
sludges would contribute organic components to solid wastes. These organic
compounds often serve as growth substrates and nutrients for the microbial
populations present in a landfill as stabilization proceeds. These wastes
often also contribute moisture for leachate formation and a biological
inoculum for biological activity leading to conversion products of volatile
organic acids and gas. Chemical sludges, chemical waste residues, chemicals
and industrial and municipal sludges typically contribute inorganic as well
as more complex organic constituents which provide sources of hazardous or
toxic heavy metals and organic chemicals. The volume of hazardous or toxic
heavy metals and organic chemicals is generally small in an MSW landfill, but
could constitute a potentially large source of hazardous constituents in
codisposal or hazardous waste landfills.
The character of leachate is largely a function of the characteristics
of the waste in the landfill. Where significant quantities of heavy metals
or hazardous organic constituents are included in the waste, these
constituents should be expected to appear in the leachate depending on the
phase of stabilization. Where waste materials are disposed which lead to
very high (>12) or very low (<2) pH, or when highly polar organic chemicals
are present, the interaction of the leachate with the liner should be
152
-------
assessed to evaluate potential changes in liner properties which may lead to
increased rates of leakage or seepage from the landfill.
Relatively high or low pH conditions or the presence of toxic inorganic
and/or organic chemicals may inhibit the microbially mediated processes of
landfill stabilization. However, research indicates that disposal of
appropriate proportions of industrial and municipal wastes may be
accomplished with no adverse effects on the stabilization processes (Houle, eX
1977; Barber, 1981; Pohland, et al., 1981; and Walsh, et al., 1983). In some
cases, the moisture from industrial sludges actually accelerates the
stabilization process (Walsh, et al., 1983)iwhen inhibition of
the microbial processes occurs, the stabilization processes typically
progress at a reduced rate (if at all) as the leachate tends to become more
concentrated. This may increase the potential for migration of certain
inorganic and organic constituents which become more mobile under acid or low
pH conditions.
Operational Procedures Affecting Leachate Generation
Two landfills, in similar regions of the country, with similar geologic
and hydrogeologic settings and identical waste streams can have vastly
different leachates resulting from differences in the operational procedure
for the landfills. The primary operational variables affecting leachate
generation are the waste placement conditions and methods, the leachate
collection procedures and system design, and possible leachate recirculation
procedures.
Waste Placement--
The method of waste placement affects both the density and hydraulic
conductivity of the waste after compaction. The vertical hydraulic
conductivity of the waste is primarily a function of the structure, density,
water or liquid content, and particle size distribution of the waste. The
relative compaction may be estimated based on the type of waste, the initial
moisture content, the method of placement (end dump, spreading, etc.), lift
thickness prior to compaction, and type of compactor used for compaction.
Very little is known about the vertical hydraulic conductivity in saturated
or partially saturated waste materials. The vertical saturated hydraulic
conductivity may vary from greater than IxlO"1 cm/sec for poorly compacted
construction debris to less than IxlO"6 cm/sec for some sludge materials. In
addition, density changes, changes in moisture content and differential
settlement in the waste occur as a function of time during the stabilization
process. Therefore, it is likely that the vertical hydraulic conductivity of
the waste also changes with time.
The cell dimensions and dates of waste and final cover placement are
also important for the evaluation of vertical percolation. Information
regarding the placement and compaction of the waste materials is required for
all waste containment structures.
Leachate Management--
The management of leachate in the landfill is extremely important for
th evaluation of leachate generation and chemical characteristics. The
153
-------
primary leachate management variables affecting leachate composition are the
detention times, locations of leachate detention (e.g., landfill, sump,
surface impoundment) and locations of leachate removal.
In addition to the leachate management information, data concerning the
removal of leachate from the landfill are important. These data include the
dates and volumes of leachate removed, the locations from which leachate was
removed, the depths of leachate in each cell, and the first date leachate was
detected in the leachate collection system.
Field monitoring data, such as the chemical analysis of the leachate for
the indicator parameters discussed under "Leachate Monitoring," and visual
descriptions such as the color, turbidity and odor may also be useful.
The leachate management information is essential to evaluate the rate of
generation and quantity of leachate and to evaluate the distribution of water
as a function of time and space in the waste. Since the characteristics of
the leachate are controlled by these variables, leachate management
information is required for all landfills.
Leachate Recirculation--
Landfill stabilization processes tend to be accelerated when the
moisture content remains at or near the field capacity of the waste.
Leachate recirculation increases the moisture content of the waste, more
evenly distributes nutrients and enhances biomass/substrate contact, and
maintains the waste at or near the field capacity, thus accelerating the
stabilization processes.
The primary variables which must be considered to evaluate the impact of
leachate recirculation on the stabilization processes are the volumes, and
rates and extent of application, recirculation schedules, and the chemical
character of the recycled leachate. Gas generation data are also important
to the overall evaluation.
Gas Management--
While the gas management system does not have an immediate impact on
leachate characteristics, monitoring of the gas collection or gas venting
system can reveal valuable information regarding the condition of stabiliza-
tion in the landfill. Gas production occurs mainly during the later phases
of landfill stabilization and typically indicates that the stabilization
process is in an advanced state of progression. The primary variables which
should be evaluated are the date of first measurable gas generation, the
dates of gas measurements and the gas generation rates and quality. In order
to model the landfill or lysimeter through the various stages of microbial
stabilization the gas generation data are required for all of the waste
containment structures.
Monitoring Data
Monitoring data are an essential source of information required to
evaluate the maturity of a waste containment facility. Analyses are
typically performed on the leachate, ground water, surface water, gas, and
154
-------
wastes. The results of these analyses, combined with information regarding
the frequency of monitoring and the sample locations may be used to predict
short- and long-term leachate constituent concentrations.
Leachate Monitoring Data—
Monitoring data may prove to be a very valuable source of information in
the evaluation of leachate characteristics. As discussed by Pohland and
Harper (1985), there are several indicator parameters which may be used to
describe the presence, intensity and longevity of the phases of landfill
stabilization. When applied to leachate, these indicator parameters provide
insight into the mobility of certain organic and inorganic constituents.
Table Al provides a summary of parameters which provide useful information
regarding the stabilization processes in a landfill.
As suggested in Table Al, there are physical, chemical or biological
indicator parameters which provide information required to evaluate the
interrelationships of the landfill stabilization processes. As an example,
pH and ORP are physical-chemical paramaters used to indicate acid-base and
oxidation-reduction conditions critical to the evaluation of the acidogenesis
and methanogenesis processes. The chemical and biological parameters of COD
and 8005, respectively, are used to evaluate relative biodegradability. The
chemical parameters, nitrogen and phosphorus, are used to evaluate nutrient
sufficiency and condition (aerobic/anaerobic) of a particular phase.
In addition to the parameters discussed above, the following additional
parameters may be used to assess stabilization in terms of their significance
as indicated below:
Parameter Process
Total Alkalinity Buffer Capacity
Heavy Metals Potential Inhibition
Specific Conductance Ionic Strength/Activity
Chlorides Migration Potential
Bacteria & Viruses Health Hazards
Nitrates & Sulfates Oxidizing Potential or Condition
As illustrated in Figure A2, the concentrations of selected indicator
parameters vary during each phase of landfill stabilization. Review of
available data indicate that while some contradictions occur, indicator
parameter concentrations for leachates from both landfills and lysimeters,
whether with or without microbial inhibition or leachate recycle, will be
similar but a varying times, and perhaps for different reasons.
For instance the magnitude of leachate concentrations during each phase
of stabilization is strongly affected by dilution. However, dilution does
of affect the total mass of leached constituents in space and time.
Unfortunately the effects of dilution and the distribution of indicator
oarameter concentrations with time are often poorly recorded, making it
difficult to assess which phase of stabilization a landfill is currently
undereoing In addition, there may be discrete compartment of pockets of
activity where the stabilization processes are occuring at different stages
155
-------
TABLE Al. LANDFILL LEACHATE AND GAS CONSTITUENT CONCENTRATION RANGES
ENCOUNTERED IN THE LITERATURE*AND THEIR RELATIVE SIGNIFICANCE
TO THE DEGREE OF LANDFILL STABILIZATION.
Phase or Biological Stabilization
Leaohate or Ca«
Const Ituant
Biochemical Oiygen
Devnd
(BOD,)
Cheejlcal Oiygen
Deejand
(COD)
Total Organic
Cartxxi (TOC).
•g'l
Total Volatile
Adda (TVA).
••/I a* Acetic
Acid
900,/COD
Ratio
COD/TOC
Ratio
Total KJeldahl
Kltrogen (TKX)
•g/1
nitrate nitrogen
< *>,--«>.
•g/I
Aaannia Nitrogen
(KMj-H)
M/l
NHj/TCT
Ratio
Total Pnoapnata
(PO(--P).
•t'l
Tranaltlon Phaae
100-10.900
Influence at dilu-
tion and aerobic
solubll nation of
«asta or(anlca
"90-18.000
Trending In a slail-
lar faanlon to BOD;
100-3. OOO
Beginning to app«ar
•a • result or
aerobic soluolllKa-
tlon
100-3.000
Or Juat beginning to
appear aa a raault
or aoluftlllxatlon
0.23-0.87
Inenaatnf blod«-
iradaDllltr or
aoluDlllxatloa
«.3-«.8
Loo oildatlon atata
or organlca
180-860
0.1-5.1
Inortaalnf du« to
oildatlon or
i»»nnla
120-125
O.I-O.J
0.6-1.7
told Formation Phaaa
I.OOO-5T.700
Aecuaulatlon or bio-
dagradabla orjanlo
aolda dua to a«tnano-
ganie la(
1.500-71.100
Trending In a alallar
faanlon to BOD;
500-27.700
Incraaalng rapldlri
accuanilatlon du« to
Mthanoganlc lag
3.000-l8.aoo
Soluollliatlon or orga-
nic polyaara to •onoattrai
txta-oildatlon to «ola-
tlla aalda
0.»-0.8
Hlgn blod«gradabllltr
2.1-3."
Low to BOdcrata oilda-
tlon atata or organloa
1 '-1.970
Hay t>« low dua to •!-
eroftlal aaalallatlon
or nltroganoua ooar
pounda
0.05-19
D«craaalng du« to ra-
duetlon to laannla. or
*2 «••
2-1.030
Inoraaalng dua to
nltrata reduction and
protain Draakdown
0-0.98
Protain Braakdovn: Bio-
logical ualatlitlon
0.2-120
Biological aaalaila-
tlon and Batal
coaplaxatlon
Hatnana Ftraantatlon
Phaaa
600-3. «00
Convaralon or bloda-
gradabla organlca to
gasaoua end products
(CMt and CO2)
5*0-9.760
Trandlng In a alallar
faanlon to BOD;
300-2,230
Conversion or
-------
TABLE Al LANDFILL LEACHATE AND GAS CONSTITUENT CONCENTRATION RANGES
ENCOUNTERED IN THE LITERATURE AND THEIR RELATIVE SIGNIFICANCE
TO THE DEGREE OF LANDFILL STABILIZATION (continued).
M/l aa CaCOj
Sollda (TS). 2. '50-2.050
M/l
pH 6.7
Oildatlon-Aaductlon -«0 Co -80
Potential (OBP).
•v
Coppar. 0.085-0.39
M/l
Iron. 64-312
•I' I
Laad. 0.001-0.004
M/l
Magnaalua. 66-96
M/l
'Manfanaaa 0.6
•1/1
Ntckal. • 0.02-1.5!
M/l
Potaaaluai. 35-2.300
M/l
SodliM. 20.7.600
M/l
Zinc. 0.06-21
M/l
Tot«l CollCor*. 10° to 10*
CFU/tOO al
r«cil ColKorai. 10° to 10*
cru/ioo ai
Fecal Str.ptococol. 10° to 10*
cru/ioo «i
Inertaalnf du« to
volatllt acid rorna-
tlon and blcaroonala
dlaaolutton
•.120-55.300
Incraaalnf du« to 9olu-
bllttatlon or orianlca
and aodlllutlon of
a* tali
•.7-T.7
Loo du« to tolatlla
acid acouaMlatlon
•90 to -2«0
Dacraaalnf dua to tha
daplatlon of oiyian
0.005-2.2
90-2.200
0.01-1 .M
J-l.HO
0.6-11
O.OJ-79
35-2.300
0.65-JZO
10* to 10*
10° to 10*
10° to I06
760-5.050 200-3.520 liO-9.650
Dccraaalng dua to »ola-
tlla acid reaoval
2,090-6,110 I,I60-«,6>0 i,<60-55,300
6.3-S.l 7.1-S.J «.7-8.«
Inervaalnf dua to rola-
ttla aold raooval and
blcarbonata dlaaolutlon
-70 to -2«0 >97 to -163 -2«0 to -16
O.OJ-O.U 0.02-0.56 0.005-2.2
Dacraaalnf (cooplaiatlon)
115-336 »-20 «-2.200
Oaoraaalni (ooaplaiatlon)
O.OI-O.t 0.01-0.1 0.001-1.4*
Oacraaalnf (coaplaiatlon)
91-505 91-190 3-1. KO
Oaeraaatni (eoaplaiatlon)
0.6 0.6 u.6-«i
Oacraaalnc (coaplaiatlon)
0.01-1.0 0.07 0.02-79
D««raaainc (conplaiatlon)
35-2.300 35.2.300 35-2.300
20-7.600
0.1-6.0 O.I 0.06-220
Cjaantlaily abaant Abaafit 0-10*
Eaaantlallr abaant Abaant 0-10*
Caaantlally abaant Abaatit 0-10^
Reproduced trom ^'•"%
best available copy. %,ii^
157
-------
TABLE Al. LANDFILL LEACHATE AND GAS CONSTITUENT CONCENTRATION RANGES
ENCOUNTERED IN THE LITERATURE AND THEIR RELATIVE SIGNIFICANCE
TO THE DEGREE OF LANDFILL STABILIZATION (continued).
Viruses.
PFU/100 »1
Conductivity,
u«noa/c*)
Chloride
(Cl'l.
M/l
Sulfate
(S0ta).
M/l
Sulf Lde
(S-).
M/l
Cadalu*.
M/l
Chroalue).
M/l
M«ihan«,
J
Carbon Dloilde.
1
Nitrogen Gas,
Oiyitn,
I
Hydrogen.
I
«
*. '50-3. 3'Q
30-5.000
Biologically stable:
good Indicator or
wasnout
lO-«50
Essentially absent
190-«90
0.023-0.26
0-10
Product of aerobic
decomposition of
organic*
70-80
• ir
20
Influence of trapped
air
Essentially aoavnt
In tn« pr«s«ne« of
oi)fg»n
CaMntlally aoa«nt
1. 600-17. 100
Incr»*lng du« to «oel-
11 tat Ion of Mt«U
30-5.000
St»0l«i good hydraulla
tracer
10-3.2*0
llxatlon tn«n d*cr«aa-
Ing aa anaerobic condi-
tion* art •*t«oill*n«d
0-8 8
B«flnn ng to appear and
Inere* ing due to
sulfat reduction under
anaero lo condition*
70-3.900
0.06-10
Very lo- «H)i
bio *«t*ooll**i
10-30
Increaalng due to
M*ate decomposition
M-80
tlon with CO;
0-5
Oecre**lng due to
ivroeto utll nation;
ahlft toward* anaero-
QlO MtaDOllM
0-2
Beginning to appear aa
oiygen 1* depleted!
accueiulate* until
•etnanogene*!* occur*
E**entl*llr aDaent
2.900-7.700
Oeorea*ln« due to Betal*
covpltiatlon «ltn
wHflde*
30-5.000
Stable; good hydraulic
tracer
Abaent
0.9
Low due to heavy «etal
precipitation
7*-*W
Decreaalng due to coa-
pletatton and precipi-
tation
0.05
Decreasing due to co«-
plexatlon. precipita-
tion with aulflde*
30-60
recovery
30-60
Decreasing to 20 <20-80
introduction of »lr
>5 0-20
Increasing due to
Introduction of air
0-2
essentially absent
"Hange* of constituent concentration* w«re oolleoted from the reference* and data presented In the App«ndlce*.
158
-------
1.0
gO.8
u
a
•o
o
M
-------
and rates, often making it difficult to distinguish between phases.
Therefore, in order to properly assess the impact of dilution, indicator
parameters should be frequently evaluated on a mass basis. Such an
evaluation requires careful monitoring of leachate volumes and evaluation of
infiltration, percolation and recycle quantities and rates. Where flow data
are not recorded, it may not be possible to distinguish between the intensi-
ties of various phases of landfill stabilization or those consequenced by
microbial mediation or dilution or both.
Ground-water and Surface Water Monitoring Data--
Ground-water and surface water monitoring are required to provide an
early warning system in the event of leakage from a waste containment
facility. Ground-water and surface water monitoring data are also useful for
the evaluation of seasonal variations in the potentiometric surface, which
may affect the direction and velocity of ground-water flow.
The ground-water and surface water elevations should be obtained on a
regular basis and the potentiometric surface plotted as a function of time.
Samples should be obtained and analyzed for the same constituents and
indicator parameters monitored in the leachate. These data are required for
all waste containment structures where ground water may enter the leachate
collection system.
Gas Monitoring Data--
Gas monitoring data are required for the evaluation of landfill
maturation. Since methane production is pronounced during the methane
fermentation phase of landfill stabilization, careful monitoring of gas
generation rates at the landfill is required to indicate when the landfill
has completed the active phases of stabilization and has reached the relative
dormancy of the final maturation phase. These data are required for all
waste containment facilities.
Waste Monitoring Data--
Many landfills are segregated for various wastes, leading to variations
in leachate characteristics from cell to cell or even point to point within a
cell. Data regarding the segregation of different types of wastes should be
scrutinized to assess potential impacts on leachate characteristics.
In addition, the physical characteristics of the waste such as density,
porosity, specific surface area and hydraulic conductivity change as a
function of time and depth. In order to model flow through a landfill, it
may be necessary to measure the variation in the parameters as a function of
time.
General Perspective
The evaluation of leachate volumes and chemical characteristics is
extremely complex and dependent on a large number of variables. The impact
of many of these variables on leachate generation has often been overlooked,
resulting in wide variation in interpretations and prediction of actual
performance.
160
-------
An important factor in the generation of leachate is the quantity of
water entering and the rate at which water percolates though the waste
materials. The hydrologic and waste containment facility design variables
etermine the rate of leachate generation progress of microbially mediated
waste stabilization and the leachate and gas characteristics with time.
igure 1 shows typical contributions and withdrawals of water at a landfill
cell. The hydrologic and landfill design variables may be used to perform a
water balance for each landfill cell. The water balance may be used to
estimate the quantity of water percolating into the waste and the total
volume and rate of leachate produced.
Since the microbially-mediated processes of landfill stabilization are
controlled primarily by the waste type, the availability of nutrients and the
moisture content of the waste, waste characterization and waste placement
data are vital. These data are used to estimate the initial substrate
availability and vertical hydraulic conductivity of the waste. The hydraulic
conductivity, combined with information regarding the waste porosity may then
be used to evaluate the rate at which field capacity would be reached, the
potential magnitude of short-circuiting and the rate of leachate production.
The waste characteristics and waste constituent concentrations may also be
used to predict leachate constituent concentrations as a function of time and
to assess the magnitude and impact of possible retardation or inhibition of
stabilization.
161
-------
APPENDIX B
PROGRAM GTLEACH-I
C
C
C THIS PROGRAM IS DESIGNED TO SIMULATE THE FOLLOWING CHARACTERISTICS
C OF LEACHATE EMANATING FROM SANITARY LANDFILLS DURING STABILIZATION
C —VOLUMETRIC FLOW RATE, CHEMICL OXYGEN DEMAND, VOLATILE ACID
C CONCENTRATION, AND PH.
C
C
EXTERNAL FCN.FCNJ
COMMON YY(20,6),NDAYS,NCSTR,VOL,TVOL,ZZ(20,6)
COMMON XKH,YA,QA,XKA,XKDA,YM,QM,XKM,XKDM,F,LAG
COMMON SAO,SHO,XMASSO,XAO,XMO,IRECYC,I,J,K,IOUT,GASMET
COMMON N.METH,MITER, INDEX,IWK(5),IER
COMMON Y(5),WK(86),X,TOL,XEND,H
C
C
OPEN (8,FILE='PARAM.DAT',STATUS='OLD')
OPEN(7,FILE-'RESULT.DAT')
C
C
C
CALL INPUT
C
C INITIALIZES VALUES FOR ALL CSTRS BASED ON INPUT DATA
C
DO 25 JJ=1,NCSTR
YY(JJ,1)=XMASSO
YY(JJ,2)=XAO
YY(JJ,3)=0.0
YY(JJ,4)-SHO
YY(JJ,5)=SAO
25 CONTINUE
C
WRITE(7,33)
33 FORMAT(2X,
-------
ELSEIF (IOUT.EQ.1) THEN
C
WRITE(7,55)
55 FORMAT(lX,l(days)',3X,'(kg/m3)',4X,'(mg/l)',6X,I(mg/l)',6X,
1'(grams)'.4X,'(grams)',4X,'(grams)',5X'(g/day)f)
C
C
C
ELSE
END IF
C
DO 175 I-l.NDAYS
Cj
XEND-FLOAT(I)
C*
c
C AT SPECIFIED TIME=LAG, INITIALIZES CONCENTRATION OF METHANOGENS
C
IF (I.EQ.LAG) THEN
DO 65 JJ = 1, NCSTR
YY(JJ,3) = XMD
65 CONTINUE
ELSE
ENDIF
C
C FOR FIRST 50 DAYS INCREMENTS XMASSO
C
IF (I.LT.51) THEN
DO 66 JJ=1,NCSTR
YY(JJ,1)=YY(JJ,1) + 2.2E+05/TVOL
66 CONTINUE
ELSE
ENDIF
C
C
C AT TIME=200 DAYS INCREASES XMASSO TO SIMULATE NEW SUBSTRATE
C
IF (I.EQ.200) THEN
DO 77 JJ-L,NCSTR
YY(JJ,1)= (1.0E+08)/TVOL
77 CONTINUE
ELSE
ENDIF
C
C
C
C
C AT EACH NEW TIME STEP THE INFLUENT VALUES SAO AND SHO ARE
C SET FOR THE FIRST CSTR DEPENDING ON RECYCLE STATUS.
C
IF (IRECYC.EQ.O) THEN
163
-------
SHOO.O
SAD-0.0
ELSEIF(IRECYC.EQ.1)THEN
SHO-Y(4)
SAO-Y(5)
ELSE
ENDIF
C
DO 150 J=1,NCSTR
C
C SETS INITIAL VALUES OF Y'S FOR CSTR EQUAL TO THOSE FOR THE SAME
C CSTR AT THE PREVIOUS TIME STEP AS STORED IN YY.
C
C
DO 50 K=1,N
Y(K)-YY(J,K)
50 CONTINUE
C
C INITIALIZES VALUE OF TIME FOR EACH CSTR CALCULATION
C
X=XEND-1.0
C
C CALLS DGEAR TO SOLVE DIFFERENTIAL EQUATIONS FOR CSTR
C
CALL DGEAR (N.FCN.FCNJ,X,H,Y,XEND.TOL.METH,MITER,
1INDEX,IWK,WK,IER)
IF(IER.GT.128) GO TO 999
C SETS INFLUENT VALUES FOR NEXT CSTR STEP
C
SHO=YY(J.4)
SAO=YY(J,5)
C STORES SOLUTIONS IN MATRIX YY FOR USE AS INITIAL VALUES IN
C NEXT TIME STEP
C
DO 100 K = 1,N
YY(J,K) = Y(K)
100 CONTINUE
C
INDEX-1
C
150 CONTINUE
C
C CALLS OUTPUT TO WRITE VALUES OF LAST CSTR TO RESULT.DAT
C
IF(I.EQ.II)THEN
CALL OUTPUT
11=11+5
ELSE
ENDIF
C
C SEE DGEAR DOCUMENTATION FOR EXPLANATION OF INDEX
C
164
-------
175 CONTINUE
STOP
999 CONTINUE
STOP
END
C
C
SUBROUTINE FCN(N,X,Y.YPRIME)
C
C CONTAINS THE SYSTEM OF DIFFERENTIAL EQUATIONS FOR DGEAR
C
C
REAL Y(N),YPRIME(N),X
COMMON YY(20,6),NDAYS,NCSTR,VOL,TVOL,ZZ(20,6)
COMMON XKH,YA,QA,XKA,XKDA,YM,QM,XKM,XKDM,F,LAG
COMMON SAO,SHO,XMASSO,XAO,XMO,IRECYC,I,J,K,IOUT,GASMET
C
C
C Y(1)=XMASS=SOLID SUBSTRATE CONCENTRATION
C Y(2)=XA=ACIDOGEN CONCENTRATION
C Y(3)=XM=METHANOGEN CONCENTRATION
C Y(4)=SH=HYDROLYSIS PRODUCT CONCENTRATION
C Y(5)=SA=VOLATILE ACID CONCENTRATION
C
C
YPRIME(1)=-XKH*Y(1)
YPRIME(2)=YA*QA*Y(4)*Y(2)/(XKA+Y(4)) - XKDA*Y(2)
YPRIME(3)=YM*QM*Y(5)*Y(3)/(XKM+Y(5)) - XKDM*Y(3)
YPRIME(4)=XKH*Y(1) - QA*Y(4)*Y(2)/(XKA+Y(4)) + F*(SHO-Y(4)
1)/VOL
YPRIME(5)=Y(4)*Y(2)/(XKA+Y(4)) - QM*Y(5)*Y(3)/(XKM_Y(5) + F*
1(SAO-Y(5))/VOL
C
RETURN
END
C
C
C
SUBROUTINE FCNJ(N,X,Y,PD)
C
C'DUMMY FUNCTION FOR MITER=0
C
INTEGER N
REAL Y(N),PD(N,N),X
RETURN
END
C
C
C
SUBROUTINE INPUT
° COM10N YY(20,6),NDAYS,NCSTR,VOL,TVOL,ZZ(20,6)
165
-------
COMMON XKH,YA,QA,XKA,XKDA,YM,QM,XKM,XKDM,F,IAG
COMMON SAO,SHO.XMASSO,XAO,XMO,IRECYC,I,J,K,IOUT,GASMET
COMMON N,METH,MITER,INDEX,IWK(5),IER
COMMON Y(5),WK(86),X,TOL,XEND,Ii
C
READ (UNIT=8,FMT=*) N.METH,MITER,INDEX,H.TOL
READ (UNIT-S.FMT-*) F,TVOL,XMASSO,SHO,SAO,XAO,XM3
READ (UNIT=8,FMT=*) IOUT,LAG
READ (UNIT=8,FMT-*) XKA.XKDA.YA
READ (UNIT=8,FMT=*) XKM.XKDM.YM
READ (UNIT=8 ,FMT=**) XKH.QA.QM
READ (UNIT-8,FMT=*) NCSTR.NDAYS,IRECYC
C
VOL=TVOL/NCSTR
XMASSO = XMASSO*!.OE+06/TVOL
C
RETURN
C
END
C
SUBROUTINE OUTPUT
C
COMMON YY(20,6),NDAYS,NCSTR,VOL,TVOL,ZZ(20,6)
COMMON XKH,YA,QA,XKA,XKDA,YM,QM,XKM,XKDM,F,LAG
COMMON SAO,SHO,XMASSO,XAO.XH),IRECYC,I,J,K,IOUT,GASMET
C
C CALCULATES MASS OF METHANE PRODUCED
C
GASMET=0.0
DO 195 M=1,NCSTR
GASMET = GASMET + VOL*0.267*QM*YY(M,5)*YY(M,3)/((XKM+
1YY(M,5))*100.)
195 CONTINUE
C
IF (IOUT.EQ.O) THEN
C
YY(NCSTR,6)-YY(NCSTR,4) + YY(NCSTR,5)*2 .5
WRITE (7,203)I,(YY(NCSTR,M),M=1,6),GASMET
203 FORMAT (IX,14,2X,E8.3,2X,E8.3,2X,E8.3,2X,E8.3,2X,E8.3,2X,E8.3,
12X.E8.3)
C
C CONVERTS TO MASS IF IOUT=1
C
ELSEIF (IOUT.EQ.1) THEN
C
ZZ(NCSTR.l) = YY(NCSTR.l)
ZZ(NCSTR,2) =• YY(NCSTR,2)
ZZ(NCSTR,3) = YY(NCSTR,3)
ZZ(NCSTR,4) - YY(NCSTR,4)*F/100.
ZZ(NCSTR,5) = YY(NCSTR,5)*F/100.
ZZ(NCSTR,6) = ZZ(NCSTR,4) + ZZ(NCSTR,5)*2.5
166
-------
WRITE (7,205)I,(ZZ(NCSTR,M),M=1,6),GASMET
FORMAT (1X,I4,2X,E8.3,2X,E8.3,2X,E8.3,2X,E8.3,2X,E8.3,2X,E8.3,
12X.E8.3)
ELSE
END IF
RETURN
END
c cccccccccccccccccccccccccccccccccccccccccccccccc
c C c
^ C THE PROGRAM SOLVE FLUID VELOCITY C
C C OF C
I C GROUNDWATER (SOIL SATURATED) C
^ CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
C PROGRAM SAT
DIMENSION XK(10),XL(IO),HEAD(100),TIME(100),TR(100),QDUMP(100)
CHARACTER FIN*8,FOUT*8
WRITE(*,6)
6 FORMATC INPUT NAME OF INPUT FILE (EX-.XYZ .DAT) -' \)
READ(*,8)FIN
8 FORMAT(A8)
WRITE(*,7)
7 FORMATC INPUT NAME OF OUTPUT FILE (EX-.XYZ .OUT) -' ,\)
READ(*,8)FOUT
OPEN(5,FILE=FIN,STATUS='OLDf)
OPEN(6,FILE=FOUT,STATUE='NEW )
C
C
C INPUT DATA:
C
C NLAYER : NO. OF LAYERS
C NCYCLE : NO. OF CYCLES
C XK
C XKSS :
C ' EN
C TFSS
C XL
C QDUMP
C AREA
C TIME
C
C
READ(5,*)NLAYER,AREA,XKSS ,EN
DO 10 1=1,NLAYER
READ(5,*)XK(I),XL(I)
XL(I)=XL(I)*12*2.54
HYDRAULIC CONDUCTIVITY OF SOIL LAYER (CM/SEC)
HYDRAULIC CONDUCTIVITY OF SHORT CIRCUITING (CM/SEC)
EFFECTIVE POROSITY (-) (0.01 to 0.1)
ELAPSED TIME OF SHORT CIRCUITING (10 TO 100 DAYS)
THICKNESS OF SOIL LAYER (FT.)
QUANTITY OF WATER DUMPED IN LANDFILL (CU-FT.)
AREA OF LANDFILL (SQ-FT.)
ELAPSED TIME OF EACH DUMPED WATER (DAYS)
167
-------
10 CONTINUE
READ(5,*)NCYCLE,TFSS
TFSS=TFSS*3600*24
DO 20 I-l.NCYCLE
READ(5,*)QDUMP(I),TIME(I)
HEAD(I)=QDUMP(I)/AREA
HEAD(I)=HEAD(I)*12*2.54
TIME(I)=3600*24*TIME(I)
TR(I)=TIME(I)
TR(I)=TR(I)/7
20 CONTINUE
XKAVG-0.
XLTOTAL=0.
DO 100 I=1,NLAYER
XKAVG=XKAVG+1/XK(I)
XLTOTAL=XTOTAL=XL(I)
100 CONTINUE
XKAVG=1/XKAVG
H=0.
TTR=0 .
TESTTIME=0.
DO 200 I=1,NCYCLE
DT=TR(I)
H=H+HEAD(I)
IF(TTR.LE.TFSS) THEN
QSS =• XKSS*EN
Q=QSS+XKAVG*H/XLTOTAL
ELSE
Q=XKAVG*H/XL TOTAL
END IF
QFLOW=«Q*AREA/(12*2.54)
WRITE(6,*)TESTTIME,QFLOW,H/(12*2.54)
DO 210 J-1,7
TTR=TTR+DT
DH-ABS(XKAVG*H*DT/XL TOTAL)
H=H-DH
IF (TTR.LT.TFSS) THEN
QSS-XKSS*EN
Q=QSS+XKAVG*H/XLTOTAL
ELSE
Q^XKAVE * H/XLTO TAL
END IF
TESTTIME=TTR/(3600* 24)
QFLOW=-Q*AREA/(12*2.54)
WRITE(6,*)TESTTWE,QFLOW,H/(12*2.54)
210 CONTINUE
200 CONTINUE
TTOTAL=TTR/(3600* 24)
WRITE(6,5)QFLOW,TTOTAL
5 FORMATC//,' THE GROUNDWATER FLOW RATE = ',E10.2,'CU-FT./SEC'
1 ,' IN ELAPSED TIME = ',F8.3,' DAYS')
STOP
END
168
-------
|