British Geological Survey

TECHNICAL REPORT WC/95/7 Overseas Geology Series

A GROUNDWATER HAZARD ASSESSMENT SCHEME FOR SOLID WASTE DISPOSAL

B A KLMCK, M C CRAWFORD AND D J NOY

A report prepared for the Overseas Development Administration under the O D N B G S Technology Development and Research Programme, Project 9315

ODA clatrificadon : Subsector: Water and Sanitation Theme: W3 - Increase protection of water resources, water quality and aquatic ecosystems Project title: Hazard ranking system for solid waste disposal Project reference: R5564

Bibliographic refermce : Klinck B A andothers 1995. A groundwater hazard assessment scheme for solid waste disposal BGS Technical Report WC/95/7

k-eywords : Landfill, leachate quality, modelling, hazard ranking, DRASTIC, GOD, WASP

Fnmt wver illurmaria : Sampling leachate, Merida landfill, Mexico

8 NERC 1995

Keyworth, Nottingham, British Geological Survey, 1995

CONTENTS

EXECUTIVE SUMMARY

Acknowledgements

1 INTRODUCTION

1.1 Factors in assessing hazard

1.2 Approach

2 FACTOR GROUP CONTRIBUTION TO A CONCEPTUAL MODEL

2.1 Site Grow Factors

2.1.1 Site Size Factor

2.1.2 Site Climate Factor

2.1.3 Waste Composition Factor

2.1.4 Leachate Composition Factor

2.2 Geolopical and Hydrogeological Factor Group

2.2.1 The Unsaturated Zone

2.2.2 Aquifer Properties and Contaminant Transport

2.2.3 Hydraulic Gradient

2.2.4 Recharge

2.3 Fate Group

WCf95f7 i Issue 2

3 THE MODEL APPROACH

3.i The Conceptual Model

3.1.1 Background

3.1.2 The Model

3.2 Development of the Numerical Model

3.3 Parameter Variation

3.3.1 Site Group

3.3.1.1 Site Size

3.3.1.2 Site Climat, Factor

3.3.1.3 Waste and Leachate Composition Factors

3.3.2 Geological and Hydrogeological Factor Group

3.3.2.1 Unsaturated Zone

3.3.2.2 Aquifer Properties and Contaminant Transport

3.3.2.3 Hydraulic Gradient

3.3.2.4 Recharge

3.3.3 Fate Group

3.3.3.1 Proximity of Local Population

3.3.3.2 Distance to Nearest Abstraction Borehole and Volume of Groundwater Abstracted

3.3.3.3 Distance to the Nearest Surface Water

WCl9517 .. 11 Issue 2

4 CASE STUDIES

4.1 Indonesia

4.1.1 The Dago Landfill Site

4.1.1.1 Site Group

4.1.1.2 Geological and Hydrogeological Group

4.1.1.3 Fate Group

4.1.1.4 Hazard Ranking of the Dago Landfill Site

4.1.2 The Leuwigajah Landfill Site

4.1.2.1 Site Group

4.1.2.2 Geological and Hydrogeological Group

4.1.2.3 Fate Group

4.1.2.4 Hazard Ranking of the Leuwigajah Landfill Site

4.1.3 The Sukamiskin Landfill Site

4.1.3.1 Site Group

4.1.3.2 Geological and Hydrogeological Group

4.1.3.3 Fate Group

4.1.3.4 Hazard Ranking of the Sukamiskin Landfill Site

4.2 Mexico

4.2.1 The Bordo Landfill Site

WCl9517 iii Issue 2

4.2.1.1 Site Group

4.2.1.2 Geological and Hydrogeological Group

4.2.1.3 Fate Group

4.2.1.4 Hazard Ranking of the Bordo Landfill Site

4.2.2 The Merida Municipal Landfill, Yucatan

4.2.2.1 Site Group

4.2.2.2 Geology and Hydrogeology Group

4.2.2.3 Fate Group

4.2.2.4 Hazard Ranking of the Merida Landfill

4.2.3 Leon Guanajuato Municipal Landfill

4.2.3.1 Site Group

4.2.3.2 Geology and Hydrogeology Group

4.2.3.3 Fate Group

4.2.3.4 Hazard Ranking of the Leon Guanajuato Landfill

4.3 Case Studv Discussion

5 SUGGESTED METHODOLOGY FOR THE HAZARD RANKING OF

WASTE DISPOSAL SITES

6 REFERENCES

w c/9 517 iv Issue 2

Figures

Figure 1.1

Figure 1.2

Figure 1.3

Figure 2.1

Figure 2.2

The DRASTIC Methodology (based on Rosen, 1994)

Calculation scheme for GOD (based on Foster and Hirata, 1988)

The WASP index nomogram (from Parsons and Jolly, 1994b)

Conceptual model of possible flow paths associated with a waste disposal site

Comparison of waste composition for Bandung, Indonesia and Brussels,

Belgium

Figure 2.3

Figure 2.4

Figure 2.5

Figure 3.1

Composition of recycled waste for Leon Guanajuato, Mexico

Comparative plot of rainfall and chloride concentration for the Dago Landfill,

Bandung, Indonesia

Typical characteristic curves relating moisture content and relative hydraulic

conductivity to pressure head in the unsaturated zone. Above the water table, the

moisture content remains at near-saturated values until the pressure head reaches

the bubbling pressure (hb) at which it starts to drop. The curve becomes less

sharp for fine-grained or well-sorted media. A relative hydraulic conductivity of

1 .O corresponds to the saturated hydraulic conductivity. The conductivity

decreases as the pressure head becomes more negative. After van Genuchten

(1980).

Typical breakthrough curve for a solutekontaminant travelling through a porous

medium. For the 2-D saturated zone model presented here, the source term

concentration is assumed to be the same as that in the landfill i.e. C,/C, in the

infiltrating water reaching the water table = 1.0. However, it can be seen that

wc/95/7 V Issue 2

some contaminant reaches the water table long before CJC, = 1 .O and so the time

taken for C,/C, to reach 0.5 at the water table is taken as an average for the

infiltration time through the unsaturated zone.

Figure 3.2 Grid of elements generated by Prepwater around a landfill (stippled area) for the

2-D saturated zone finite-element model represented by the parameter values

given in Table 3.1. Note that the elements are smallest around the landfill where

concentration changes can be expected to be largest and so where the model

needs to be most sensitive.

Figure 3.3 Times for CJC, to reach 0.133, and C,/C, values after 100 days for a simulation

of 2-D saturated flow away from a landfill in an unconfined gravel aquifer with

a head gradient set to 1500 in a temperate climate. Parameter values for this

model are given in Tables 3.1 and 3.2.

Figure 3.4 Times for CJC, to reach 0.133 for a simulation of 2-D saturated flow away from

a landfill lying on fractured basement in a tropical climate. Parameter values for

the model are given in Table 3.3. A doubling in the area1 size of the landfill

decreases the times with the times closest to the landfill showing slightly larger

proportional changes due to the greater effect of the recharge mound. Travel

Time through the unsaturated zone is unaffected since this is modelled in one

dimension.

Figure 3.5 Data points depicting characteristic curves (cf Figure 2.6) for unsaturated silt

loam (taken from van Genuchten, 1980) fitted using the equations given in

Section 2.X and the parameter values given on the figures and in Table 3.2.

Parameter values were obtained using a non-linear least-squares technique.

Figure 3.6 Times for CJC, to reach 0.133 for a simulation of 2-D saturated flow away from

a landfill lying on an unconfined sand aquifer in a tropical climate. Parameter

values for the model are given in Table 3.3. Halving the transverse dispersion

wc/95/7 vi Issue 2

coefficient slightly decreases the travel times along the main flow direction and

significantly increases the times away from this orthogonal flow path. Time

through the unsaturated zone is unaffected since this is modelled in one

dimension.

Figure 3.7 Times for C,/C, to reach 0.133 for a simulation of 2-D saturated flow away from

a landfill lying on shale in an arid climate. Parameter values for the model are

given in Table 3.3. Decreasing the longitudinal dispersion coefficient increases

travel times with the greatest increases being away from the main flow path and

closest to the source. There is no simple correlation between the changes in times

and the change in the parameter. Time through the unsaturated zone is based on

a dispersion coefficient of lm.

Figure 3.8 Times for CJC, to reach 0.133 for a simulation of 2-D saturated flow away from

a landfill lying on an unconfined limestone aquifer in temperate and arid climates.

Times are given for four specified head gradients and other parameter values for

the model are given in Table 3.3. The change in infiltration rates between the two

climates is an order of magnitude. For the non-set gradient this is reflected by an

order-of-magnitude decrease in times. For the specified gradients, the climate

change makes little difference except for the smallest 1: 1000 gradient. Travel

times decrease in linear proportion to the increase in head gradient.

Figure 3.9 Times for CJC, to reach 0.133 for a simulation of 2-D saturated flow away from

a landfill lying on siltstone in temperate and arid climates. Parameter values for

the model are given in Table 3.3. The decrease in travel times with increasing

infiltration is not linear since the rates change by an order of magnitude between

the two figures whereas the times decrease by proportionately less nearer to the

landfill.

Figure 4.1 Location of Indonesian case study sites.

WC/95/1 vii Zssue 2

Figure 4.2 Location of Mexican case study sites.

Figure 4.3 Composition of waste received in 1993 by the Leon Gto. Landfill

Figure4.4 Comparison of scores from the DRASTIC, GOD, and WASP assessment

schemes for five of the six case studies described in Section 4. Sukamiskin is excluded because

the scores only consider the clay liner in this case. The scores have been normalised against each

assessment scheme's maximum so that they can all be plotted on a of 0 to lscale. The lower

figure shows the generic model's 1-D vertical travel time plotted against the normalised

DRASTIC score showing that the lower DRASTIC scores correspond well with the

lowestcalculated 1 -D travel times.

Figure 5.1

could play in the assessment of alternatives for a landfill waste disposal site.

Procedural flow diagram to illustrate the role that the generic model proposed here

WCl9517 viii Issue 2

Plates

Plate 2.1 Mass flow of waste at the Leuwigadja waste disposal site, Bandung, Indonesia.

Plate 2.2 Landslide into recently completed leachate oxidation ponds, Jelakong landfill

site, Bandung, Indonesia.

Plate 2.3 Recovered plastic bags waiting for collection at the Merida municipal waste

disposal site, Mexico.

Plate 2.4 Waste gleaning at the delivery point, Leuwigadja waste disposal site,

Bandung, Indonesia.

Plate 2.5 Discharge of untreated sewage at the Merida municipal waste disposal site,

Mexico.

Plate 2.6 Tannery waste being discharged at Leon Guanajuato waste disposal site,

Mexico, the fat from the processed skins is being recovered.

Plate 4.1 View into the head of the valley of the Leuwigadja landfill site.

Plate 4.2 Leachate draining from the Leuwigadja landfill into a stream used for rice paddy

irrigation.

Plate 4.3 Landfill gas collection system at Sukamiskin Landfill. The amount of

settlement is evident from the position of the base plate above the ground

surface.

Plate 4.4 A village on the edge of Sukamiskin Landfill. A leachate spring is evident in

the foreground.

WCl9517 ix Issue 2

Plate 4.5 An aerial view of the Merida landfill prior to the establishment of a covered

land raise. The lagoons at bottom right accept sewage and liquid industrial

wastes from tortilla manufacture.

Plate 4.6 Leachate seepage at the Merida landfill. The substrate is karstic limestone.

Plate 4.7 View towards the Leon Guanajuato landfill.

WCl9517 X Issue 2

Tables

Table 2.1 Percolation of rainfall into a landfill for different climatic settings

Table 2.2 Waste composition for selected developing countries

Table 2.3a Leachate analyses - Mexico

Table 2.3b Leachate analyses - Indonesia

Table 2.4 Organic compounds identified in Mexican and Indonesian Leachates

Table 2.5 Typical values of hydraulic conductivity ( d s e c )

Table 2.6 Porosity values for some common rock types

Table 2.7 Root constants and wilting points for common vegetation types

Table 3.1 Required parameters for Prepwater and Prepwaste to run the 2-D saturated zone

simulation. Values given are for the simulation presented in Figure 3.3.

Table 3.2 Required parameters for Prepwater and Prepwaste to run the 1-D unsaturated

zone simulation. Values given are for the times in Figure 3.3.

Table 3.3 Parameter values for the 2-D saturated zone simulations presented in the Figures.

Table 3.4 Parameter values for the 1-D unsaturated zone times given in the Figures.

Table 4.1 Comparison of the site scores using DRASTIC, GOD and WASP with the 1-D

numerical simulator.

WCf95f7 xi Issue 2

EXECUTIVE SUMMARY

This study reports on work carried out under the Overseas Development Administration (ODA)

and British Geological Survey (BGS) Technology, Development and Research (TDR)

Programme (Project 9315, R5565) as a contribution to the British Governement’s provision of

aid through technical assistance to developing countries. It focuses on the contamination threat

to groundwater quality posed by landfill leachate. The main objective is to provide an approach

for a robust means of ranking an existing or prospective waste disposal site within a hazard

framework.

Waste containment as a waste management strategy is expensive to implement and is beyond the

means of many third world countries. Dilute and disperse type landfills are the norm. In this

type of landfill the attenuation capacity of the unsaturated zone is exploited in order to reduce the

impact of any leachate on the groundwater system. With this in mind a method is needed to

estimate the impact of the hazard and outcomes of landfilling for the dilute and disperse case.

The controls on contaminant migration from landfills are examined and the parameters which

are important in establishing the hazard posed by a waste facility to groundwater quality

identified.

The report considers:

e the factors which contribute to landfill hazard

e the mathematical description of flow and contaminant transport in aquifers

numerical modelling and parameter estimation for landfill in a number of different

case studies from Indonesia and Mexico to demonstrate the proposed scheme and to

e

hydrogeological environments

e

compare the methodology with existing empirical assessment schemes.

The following factors, which need to be evaluated in assessing landfill or landraise impact on

groundwater quality, have been identified by Klinck, ( 1995):

wc19517 xii Issue 2

Site group: e site size

e waste composition

0 leachate composition

e site climate

Geological and hydrogeological group: e unsaturated zone character

0 aquifer properties

e hydraulic gradient

0 recharge

Fate group: e proximity of local population

0 volume of groundwater abstracted

0 distance to nearest abstraction boreholel spring

e distance to the nearest surface water

A number of empirical methods have been employed in evaluating groundwater vulnerability

which take account of the above listed risk factors. Three of them have been considered in this

study. The schemes used were DRASTIC, Aller et al., (1987); GOD, Foster, (1987) and Foster

and Hirata, (1991), and the Waste Aquifer Separation Principle, WASP, of Parsons and Jolly,

(1994a and 1994b). The latter scheme is specifically designed for landfill hazard ranking and

makes an assessment of the impact on the aquifer directly below the waste disposal site.

The hazard assessment scheme proposed here is based on the travel time of a conservative

contaminant, chloride, from a waste facility. It can be used to establish a hazard zone around

a waste site for different hydrogeological and climatic settings. For the purposes of the present

study a generic, deterministic model has been designed in order to examine a number of

hydrogeological scenarios. The use of mathematical models offers the advantage of being able

... wc19517 X l l l Issue 2

to test the sensitivity of the prediction to parameter variations where there is uncertainty in their

value. The model has both 1-D and 2-D numerical, contaminant transport simulators. The

impact of waste disposal is assessed not only directly below the waste site, but also in the

environs of the site and in this respect the approach is novel compared with existing schemes.

The main steps in constructing and

e develop a conceptual model

mplementing the numerical simulators were to:

design the model using a suitable computer code

e predict scenario response and determine factor response to parameter variation.

The conceptual model was developed with the following criteria in mind:

to include as many of the factors listed above as possible,

it had to be kept simple, but to be applicable to as many real situations as possible

using its numerical representation, to make extrapolation to real sites as quantified and non-

subjective as possible

To reduce subjectivity, it was decided to calculate time periods for pollutant migration to achieve

specific levels at various distances from a landfill. Although simple conceptual models are

bound to contain assumptions that do not apply in individual situations, it should be possible to

use the numerical representation of the model to make a semi-quantitative assessment of the

potential hazard at a certain distance from the landfill.

A comparison of the above mentioned ranking schemes and the numerical simulators was

conducted using data from the case studies. The following observations can be made:

DRASTIC is insensitive to high recharge, i.e. greater than 250mma-' and considers a

narrow range of hydraulic conductivity from 4.7 x 10-5 to 9.4 x ~ O - ~ ms-'.

GOD is the quickest scheme to use.

WCl95l7 xiv Issue 2

The computerised version of WASP was easy to use. Because the scheme is intended for

hazard assessment of prospective waste disposal sites, groundwater usage is considered.

The numerical simulator requires considerable expertise to run and obtain results. Using

the 1-D simulator, developed for this study, vertical travel times from the waste to the water

table are obtained. These times give a quantified groundwater pollution potential which

are directly comparable with the results of the less mathematically rigorous scoring

schemes.

All of the schemes tested can be used either to evaluate prospective sites or sites already

in existence.

The following procedure is suggested to evaluate the impact of a single site on the groundwater

system or to compare a number of sites for their suitability for waste disposal.

Carry out a preliminary desk study of the prospective areas. This will involve obtaining

topographical, geological and hydrogeological maps; climate data; site investigation

reports, and any other information in the public domain.

In many cases the basic information will not be available and a site visit will be essential.

During this visit a record of the geology, and geographic setting of the area should be made.

Apply GOD to rank the sites. GOD is the simplest scheme to implement and does not

require very detailed data to make an assessment. It gave comparative rankings to WASP

and DRASTIC.

The sites with the lowest GOD score should then be modelled using the WASP scheme or

the numerical, 1-D transport simulator. The sites with the lowest barrier factor scores in

WASP and the longest travel times using the simulator should then be considered for waste

disposal. The effect of the waste disposal operation on the fate group can be assessed using

the 2-D contaminant transport simulator.

Using the model and the guidelines provided, it should be possible to derive an approximate time

at which the landfill starts to pose a hazard at the distance of interest.

WCl95l7 xv Issue 2

Acknowledgements

This project would not have been possible in the first place without the support of the British

Government’s, Overseas Development Administration who funded the research. Field work in

support of the project was conducted in Indonesia and Mexico and many people gave us their

support. The following deserve special acknowledgment:

In Indonesia, we had the help of Gottfried Rolke of the German Advisory Team and his

colleagues of the Indonesian Directorate of Environmental Geology. Mr Roelke gave the project

valuable assistance in acquiring background data and project reports which are quoted

extensively in the case studies. Work in Bandung was greatly facilitated by Dr H Sudaril,

Director of PD Kebersihan (PDK), Bandung Municipality Cleansing Enterprise, and Ir

Somardjito PR, Technical Director of PDK, who organised site visits and supplied much of the

statistical information on Bandung quoted in this report.

In Mexico, many of our initial contacts were established by the British Council. In particular we

would like to thank the following: Ing Gabriela Rivera de la Torre, Mexico D.F. Servicio

Urbanos (Biogas y Reciclamiento), for her valuable logistic support in arranging our site visits

in Mexico City and in supplying unpublished data. Thanks also to Ing Felipe Polo Hernandez,

Director General of Sistema de Agua Potable y Alcantarillado del Municipio de Leon (SAPAL),

for making available to us the services of Ing Carlos Rodriguez Sanchez, Jefe del Departamento

de Proyectos Ecologicos. Ing Rodriguez Sanchez gave us valuable logistic support in arranging

our site visits in Leon Guanajuato, accompanying us in the field, and supplying unpublished

information which is used in this report. Ing Enrique Chaparro of the Leon Guanajuato

Municipality provided the data used to construct Figure 2.3 and Figure 4.6. In Merida, Ing

Miguel Villasuso Pino and his colleagues of the Universidad Autonoma de Yucatan, gave us

unstinting support, providing transport, arranging our site visits, informative discussion, and

making available copies of difficult to obtain published and unpublished data used extensively

in this report. The offices of the Comisi6n Nacional de Aguas (CNA) in Selaya and Merida

provided us with climate data; particular thanks to Ing Renan Mendez of the Merida office for

his help and for providing copies of open file reports. Ing Cesar Borges Zapata, formerly Sub-

Director de Aseo Urbano, Ayuntamiento de Merida, allowed me to copy his negatives used to

produce Plates 2.5 and 4.5.

WCI95l7 xvi Issue 2

Comments on a first draft of this report by John Bennett of the British Geological Survey are

appreciated. David Bailey of the Fluid Processes Group produced an extremely useful

spreadsheet application to calculate recharge which was used during the study. The report was

technically reviewed by Geoff Williams of the Fluid Processes Group. Geoff also gave us many

useful leads into the literature and provided us with his teaching notes which helped in writing

parts of Section 2.

WCf95f7 xvii Issue 2

1 INTRODUCTION

This study was undertaken as part of Project R5564, Hazard Ranking System for Solid Waste

Disposal, under the Overseas Development Administration (ODA) Technology Development and

Research (TDR) Programme. The programme contributes to the British Government’s provision

of aid through technical assistance to the developing countries. The project aim was to develop

a hazard ranking scheme for the waste disposal practices of developing countries using sites in

Mexico and Indonesia as case studies. The study focused on the contamination threat to

groundwater quality posed by landfill leachate. The main objective was to provide an approach

for a robust means of ranking an existing or proposed waste disposal site within a hazard

framework.

The following report examines the controls on contaminant migration from landfills and

identifies the parameters which contribute to the hazard to groundwater quality posed by a waste

disposal facilities. A hazard assessment scheme is proposed which is based on the travel time

of a conservative contaminant from a waste facility and which establishes a hazard zone around

the facility given different hydrogeological and climatic settings.

Although waste containment is the current trend in Europe and North America, the technology

is extremely expensive and is beyond the means of many third world countries. Mather, (1994)

notes that prior to 1972 dilute and disperse type landfills were the norm in Europe. In this type

of landfill the attenuation capacity of the unsaturated zone is exploited in order to reduce the

impact of any leachate on the groundwater system. Mather op.cit. argued that in developing

countries it is unnecessary to protect every minor aquifer from localised pollution and undesirable

to impose the expensive containment technologies of Europe and the USA. With this in mind

a method is needed to estimate the impact of the hazard and outcomes of landfilling for the dilute

and disperse case. Klinck, (1995) carried out a review of a number of hazard ranking schemes

which have been applied to waste and contaminated sites and proposed a list of factors which

need to be evaluated in assessing landfill or landraise impact on groundwater.

1.1 Factors in assessing hazard

we19517 1 Issue 2

Ideally at dilute and disperse sites degradation of the harmful organic compounds found in

leachate occurs. The leachate generated is allowed, in an uncontrolled way, to percolate through

the unsaturated zone down to the water table. The contaminant loading beneath a waste site is

a function of microbial degradation in the waste and the unsaturated zone attenuating capacity.

In establishing a scheme for dilute and disperse sites the unsaturated zone properties are a key

factor in the overall hazard impact. At the water table, where the hydraulic properties of the

saturated aquifer control the process, the remaining contaminant is diluted and carried away by

regional groundwater flow.

A look at the factors which were given consideration in the schemes reviewed by Klinck op.cit.

show a number of similarities. The following is a list of risk factors, divided into three groups,

which is considered to be important in evaluating a site for its impact on the groundwater system.

Risk Factor Groups

Site group: a site size

a waste composition

a site climate

a leachate composition

Geological and hydrogeological group: a unsaturated zone character

a aquifer properties

a hydraulic gradient

a recharge

Fate group: a proximity of local population

a distance to nearest abstraction borehole a volume of groundwater abstracted

a distance to the nearest surface water

1.2 Approach

A number of methods can be employed in evaluating the importance of the risk factors listed

we19517 2 Issue 2

above. Perhaps the simplest is to convene a group of experienced personnel and, for various

scenarios, assign a numerical rating to the risk factors based on the consensus of the group - this

is the Delphi method. The resulting schemes are generally questionnaire based and subjective.

Examples of such aquifer vulnerability assessment schemes are DRASTIC, Aller et al., (1987),

and GOD, Foster, (1987) and Foster and Hirata, (1991). The rankings derived can be plotted

onto maps and contoured. The Waste Aquifer Separation Principle, WASP, designed by Parsons

and Jolly, (1 994a, 1994b) is specifically designed for landfill hazard ranking. The scheme makes

an assessment of the impact on the aquifer directly below the waste disposal site.

The DRASTIC method is an implementation of the concept of vulnerability mapping and

attempts to generate a single vulnerability parameter using a fixed weighting scheme for seven

input parameters. These are: Depth to water, net Recharge, Aquifer media, Soil media,

- Topography, impact of vadose zone, and hydraulic Conductivity of the aquifer. The scheme for

weighting the different parameters was arrived at by a consensus procedure rather than being

based on a physical model. This approach can be criticized both because the effect of some of

the input parameters is uncertain and because the method of combining them is somewhat

arbitrary, (Van Stempvoort et al., 1992). Although originally developed for regional aquifer

protection planning the scheme has been successfully used to identify groundwater supplies that

are vulnerable to point source contamination from volatile organic compounds released by

industrial processes, Kalinski et al., (1994).

The DRASTIC index (or pollution potential) is calculated as follows:

Pollution Potential = D P , + rtR, + A,A, +S,S, + T,T, + I,&, + C,C,

where subscripts are r = rating for the site

w = importance weight of the parameter

A useful calculation algorithm was given by Rosen, (1994), and Figure 1.1 has been adapted

from this.

The GOD methodology is an organisational basis of risk assessment, which ranks Groundwater

occurrence, Overall aquifer class, and Depth to water table. Each component is ranked on a scale

of zero to one, the aquifer pollution vulnerability being defined as the product of the three

WCl9517 3 Issue 2

rankings, Figure 1.2. A drawback is that the scheme gives a vulnerability which is an

arithmetical artefact. For example if all three components are ranked 0.8 then the product is 0.52

which implies that the product of the vulnerabilities is less than the individual vulnerabilities.

For the purposes of the comparative exercise the geometric mean of the parameters is considered

to provide a better index. The implication being that each parameter has equal weight.

WASP is a scheme which has been developed for the South African, semi-arid climatic setting

and is the acronym for Waste Aquifer Separation Principle. Three factors are identified as being

important in the assessment of site suitability:

a threat factor

barrier factor

a resource factor

The threat factor is basically the product of the volume of leachate produced and the leachate

quality. It is obtained by consideration of the final design area of the waste facility and the waste

type. The barrier factor between a waste pile and an aquifer is defined by the character of the

unsaturated zone. This is where much of the leachate attenuation is expected to take place. The

factor is evaluated by calculation of the travel time from the base of the waste to the water table

using Darcy’s Law. The resource factor considers the strategic value of the groundwater to the

user. Assessment of this factor is questionnaire based and deals with both current and potential

usage. The rating consists of both a WASP index and a data reliability index obtained either by

a computer application or a questionnaire, the final analysis is done using a nomogram, Figure

1.3. Essentially the nomogram gives the arithmetic mean of the factor scores and assigns a

descriptor, e.g. suitable, unsuitable etc. There is no provision for varying recharge in the scheme,

however this is not seen as a major obstacle. The use of Darcy’s Law to calculate transport times

through the unsaturated zone using values of saturated hydraulic conductivity is a concern.

Travel time through the unsaturated zone will be underestimated with a resultant overestimate

in the barrier factor and possibly the exclusion of suitable sites.

As an alternative to subjective ranking schemes a more rigorous, stochastic or deterministic

approach can be taken using groundwater flow models to evaluate a number of scenarios.

Numerical rankings based on the outcome of the models can then be produced. Gutjahr, (1992)

lists the available modelling options as follows:

WCl9517 4 Issue 2

0 Deterministic porous-media models, where uncertainty enters primarily through

Stochastic porous-media models, where uncertainty enters through the treatment of the

Fracture media models, where both stochastic and deterministic features may appear.

parameter variations.

0

medium itself as well as through parameter variations. 0

Numerical models are only as good as the conceptualisation of the physical system that they are

supposed to represent.

For the purposes of the present study a generic deterministic model has been designed in order

to examine a number of hydrogeological scenarios. The impact of waste disposal is assessed not

only on groundwater quality directly below the waste site, but also in the environs of the site.

In this respect the approach is novel compared with existing schemes.

The use of models offers the advantage of being able to test factor sensitivity to parameter

variations and, if a sufficient number of realisations are carried out, a statistical analysis is

possible. The main problems with this approach are that it is impossible to examine every

permutation of variability in the above groups of factors and the approach is computationally

expensive in terms of CPU time.

This report sets out to examine:

0 the factors identified as contributing to landfill hazard

the mathematical description of flow and contaminant transport in aquifers

numerical modelling and parameter estimation for landfills in a number of different

landfills in Indonesia and Mexico to demonstrate the proposed scheme and compare it

a methodology for assessing the suitability of a site for waste disposal.

0

0

hydrogeological environments 0

with DRASTIC, GOD and WASP 0

2 FACTOR GROUP CONTRIBUTION TO A CONCEPTUAL MODEL

A number of scenarios can be envisaged to illustrate possible contaminant migration from a

WCl95f7 5 Issue 2

waste disposal site. The usual method of depicting these scenarios is by means of conceptual

models which are then formulated as a numerical simulator. The conceptual model shows all the

possible interactions between a waste facility and the groundwater system. The potential of a

landfill to pollute derives from its ability to generate leachate which can migrate downwards to

the underlying groundwater table or alternatively run off and contaminate surface water. The

conceptual model attempts to embody all of the factors listed above. The factors subsequently

become the variables of the numerical simulator. Figure 2.1 is the conceptual model for a

landfill and landraise adopted for the present study. The black arrow represents infiltration; the

red arrows represent contaminant pathways, and the blue arrows show the groundwater

movement. The hazard that the landfill poses to the groundwater system depends on the red

pathways and how they interact with the blue pathways. To further develop the model the groups

of factors listed above have to be considered in terms of the parameters which contribute to the

risk.

2.1 Site Group Factors

In both Indonesia and Mexico waste disposal to land raise is preferred to infilling valleys or

quarries. Although not a primary consideration addressed by the proposed hazard ranking

scheme the site topography can, in itself, be a hazard. In Plate 2.1 waste mass flow at the

Leuwigadja disposal site near Bandung, Indonesia, has occurred. The mass movement was

caused by a combination of deposition on a steep slope, and water saturation of the waste

reducing its internal coefficient of friction. In this case the waste travelled down the valley

demolishing houses and covering rice paddies to a depth of more than one metre. Plate 2.2

illustrates that even at carefully engineered sites such as the new Jelakong site at Bandung,

Indonesia , slope instability can cause problems. Here there has been a landslide into the recently

completed leachate oxidation lagoons.

Turning again to Figure 2.1 the site group factors contribute to the volume of leachate that is

generated and its quality. The factors can be combined to arrive at an estimate of potential

leachate generation and quality which are a function of site size, infiltration and waste type. This

is examined further below.

WCl95l7 6 Issue 2

2.1.1 Site Size Factor

The site size determines the surface area of waste that can accept infiltration from precipitation.

Larger sites generate more leachate than smaller sites in the same climatic setting and for a

similar waste type.

2.1.2 Site Climate Factor

This factor comes into play through two distinct, but related processes of precipitation and

evapotranspiration. The quantity of leachate generated is a function of the water budget of the

landfill. This will depend upon the original moisture content of the waste and upon how much

water incident on the waste ultimately becomes leachate. Effective rainfall percolating through

waste is one of the main controls on the volume of leachate generated by a waste site and in the

recharge to the regional aquifer.

Holmes, (1980) discussed moisture content and water retention of domestic refuse and showed

that the field capacity of domestic waste varies over a range from 29 to 42% by volume.

Cointreau, (1982) quotes a range of between 29 and SO%, the higher percentage corresponding

to Bandung, Indonesia. Studies by Campbell, (1982) in the UK indicate that, on an annual basis,

leachate production can surpass 50% of incident rainfall once the absorptive capacity of the waste

is exceeded, while Blakey, (1982) has demonstrated that for a bare soil, landfill cover, in the UK,

annual infiltration is 55% of rainfall, declining to 36% for a vegetated surface.

An estimate of the amount of leachate generated can be obtained by applying a water balance

calculation. The equation for landfills and landraises situated above the water table is given by

the Department of the Environment, DOE, (1978) and Holmes, (1980), it may be stated as:

percolation = P - Ep f R + (Ld - Ed) + AS

where P = precipitation

Ep = actual evapotranspiration

R = run-off

Ld = volume of liquid disposed

WCl95l7 7 Issue 2

Ed = actual evaporation from liquid disposal

AS = change of moisture storage within the landfill and its cover

In an assessment scheme for a site with very limited data availability it would be impracticable

to carry out the above calculation. An attempt to arrive at a figure by considering typical rainfall

figures is presented in Table 2.1 and is based on the assumption that 50% of precipitation will

infiltrate.

Table 2.1 Percolation of rainfall into waste for different climatic settings

(Rainfall data from Critchfield, 1983).

Tropical Rain Forest I 1500 - 2200

Tropical Monsoon I 1500 - 3700

Wet and Dry Tropics 1000 - 1500

Tropical Aridsemi-arid 1 - 500

Dry subtropics 500

Humid subtrop. 750 - 1500

Marine 500 - 2500

750 - 1100

750 - 1850

500 - 1500

up to 130

200

350 - 750

250 - 1200

Sub-tropical and tropical arid regions are to be found in many developing countries and the

presence of large soil moisture deficits means that the potential for waste disposal sites to

generate leachate is very low. The coupling of low levels of leachate production with an

unsaturated zone means that the pollution threat to groundwater is also very low. Blight et al.,

(1992) conclude that to be able to predict whether or not a waste facility sited in an arid area will

produce leachate requires a knowledge of the detailed distribution of precipitation and

evaporation throughout the year. Studies of a site in Cape Town indicate the potential to produce

leachate is about 200 mm per year for a soil moisture deficit of 600 mm. On the Witwatersrand,

with a soil moisture deficit of 800 mm, a figure of 130 mm was calculated.

2.1.3 Waste Composition Factor

WCl9517 8 Issue 2

The actual waste composition influences the composition of the leachate which is ultimately

generated. Unfortunately a source book of information on leachate quality as a function of waste

composition for different countries is not available. Table 2.2 is a modified version of data

presented by Cointreau, (1 982), it compares waste composition with economic development.

Cities on the left of the table are classed as middle income while those on the right are classed

as low income. The most notable feature of the tabulated data is the very high content of

putrescible material and the low content of recyclable material such as glass, paper and metals

in the wastes. Figure 2.2 highlights this contrast by comparing Brussels, an industrialised

developed western European city, and Bandung, a low to middle income city in Indonesia.

Holmes, (1993) notes that for a typical Asian city 75% of the waste is putrescible and has a

density of 570 kgm-3. For a Middle Eastern city the figure is around 50% with a density of 21 1

kgm". Figures for Bandung, Indonesia, Bandung, (1990) are similar where organic waste

contributes 74% of the total, the moisture content is about 65%, and the density is about 225

kgm-3. Recent work in Mexico by Klinck, (1993) suggests that a typical putrescible content may

be as high as 98% with a density of 350 kgm".

The widespread practice of informal recycling activity may explain, to some extent, the very high

organic matter contents present in landfilled wastes. The most common materials recovered from

the waste are paper, cardboard, glass, metal and plastic bags, (Plate 2.3). This recycling process

often begins even before the waste leaves its point of origin. Indeed in Bandung, Indonesia, it

is estimated that the waste is gleaned at least three times prior to being landfilled, (Plate 2.4).

Figure 2.3 illustrates the composition of the recycled component for the city of Leon Guanajuato,

Mexico, where approximately one percent of the total disposed waste is recovered at the landfill.

A major component of the recycled waste is tortilla, the hard corn pancake eaten throughout

Mexico, which is re-manufactured into animal feed.

A hazardous component of waste is often sewage sludge which may consist entirely of raw

sewage from septic tanks. Plate 2.5 shows raw sewage being discharged from a collection

vehicle to the Merida landfill site in Mexico. Peniche A. et al., (1993) have demonstrated the

presence of faecal coliforms, faecal Streptococcus, Clostridium and Salmonella in such sludges.

Previously, before the construction of oxidation ponds, the sewage was retained in waste-bunded

lagoons built directly onto the karstic limestone which underlies the site. An indication of the

possible level of impact below the waste site is provided by Cruikshank et al., (1980) who report

wc/95/7 9 Issue 2

coliform concentrations of between 1500 and 8100 total per 100 ml for the aquifer beneath

Merida, and attributed it to leakage from unsewered septic tanks.

2.1.4 Leachate Composition Factor

Leachate is the liquid produced as water percolates through the waste. It is derived from the

original moisture content of the waste and infiltrating water. Leachates contain a variety of

pollutants which can contaminate groundwater including bacteria and viruses. Christensen et al.,

(1994) state that landfill leachate can be characterised as a water based solution of four groups

of pollutants:

dissolved organic matter expressed as chemical oxygen demand (COD) or total organic

carbon (TOC), including methane and volatile fatty acids.

anthropogenic organic compounds associated with household and industrial use and

generally present in very low concentrations. These compounds include, among others,

aromatic hydrocarbons, chlorinated solvents and phenols.

inorganic macro components: calcium, magnesium, sodium, potassium, ammonium, iron,

manganese, chloride, sulphate and bicarbonate.

Heavy metals: cadmium (Cd), chromium (Cr), copper (Cu), lead (Pb), nickel (Ni) and

zinc (Zn).

Leachate composition varies amongst landfills; however, because of the high organic content of

the wastes generated by developing cities in general, leachates derived from them are strongly

organic in character. The development of the leachate chemistry is controlled by oxidation -

reduction reactions, acid - base reactions, complexation of heavy metals, precipitation1

dissolution, and adsorption1 desorption processes. Leachate quality varies throughout the

operational life of the landfill and long after its closure. During the early stages of waste

degradation and leachate generation the composition is acidic and high in volatile aliphatic acids.

This phase is often described as the acetogenic phase. As degradation of the waste progresses

WCl9517 10 Issue 2

conditions in the landfill become more anaerobic and the strongly reducing methanogenic phase

is initiated.

Harmsen, (1983) has shown that during the acidic phase the content of free volatile fatty acids

can exceed 95% TOC. In the methanogenic phase volatile fatty acids are present in very small

quantities while the majority of the organic compounds are high molecular weight, humic and

fulvic acids. Heavy metal contents also show differences between the two phases. In the low

pH, acetogenic phase, metals are more soluble and form complexes with free volatile fatty acids.

The methanogenic phase is characterised by a rise in pH and generally an overall decrease in

heavy metal concentrations due to precipitation; although organic complexing can reverse this

trend.

The main chemical changes involved in methanogenesis are defined by the following reactions:

Aerobic respiration

CH,O + 0,

Denitrifkation

CH,O + 4/s NO, + 4/j H'

Mn(1V) reduction

CH,O + 2 MnO, + H,O

Nitrate reduction

CH,O + I/, NO, + H'

Fe(II1) reduction

CH20 + 8 H++ 4Fe(OH),

Sulphate reduction

CH,O + '/, SO:- + '1, H+

Methanogenesis

CH,O + '/,CO2

-> CO,+H,O

-> CO, + 2/j N, + 7/j H20

-> 2Mn2+ +3H,O+CO,

-> CO, + '/,NH4++ '/,H,O

-> 4 Fe2+ + 1 1H,O + CO,

-> '/,HS- + H,O + CO,

-> I/, CH4 + CO,

The CO, produced in the above stages would tend to form HqO

reaction:

by the following

H2O + CO, -> H+ + HC0,-

wc/95/7 11 Issue 2

The reaction accounts for the very high bicarbonate values found in mathanogenic

leachates.

Given the similarity in the waste composition in developing-countries, discussed in

section 2.1.3, is it possible to adopt a typical leachate composition for input into a generic

flow model? A study of Tables 2.3a and 2.3b demonstrates the difficulty of doing this.

There is a wide variation in the concentration of determinands which seems to be a

reflection of the state of the landfill, e.g., either acetogenic or methanogenic. The samples

from Merida, Mexico, are anomalous in that they indicate methanogenic and acetogenic

stages in the same waste cell, samples MlO and M11. The sulphate concentrations

indicate that in one part of the fill sulphate reduction is occurring whereas in another part

of the fill sulphate remains high. Bicarbonate concentration, which is also a reflection

of the bacterial activity in the fill, is higher in the reduced leachates, e.g. M1 1, Table 2.3a

and 1-6, Table 2.3b.

At the Merida site the variation in chemistry appears to be a function of the moisture

content in different parts of the waste. The samples were collected during the wet season

and it is suggested that where rapid infiltration has occurred the redox state of the waste

has changed from methanogenic to acetogenic due to the presence of dissolved oxygen

in the infiltrating water causing methanogenic bacteria die off. This interpretation

suggests that the waste cell can, very rapidly, cycle from the methanogenic to acetogenic

phase in response to wet - dry season cycles.

A component that is not affected by biodegradation is chloride which behaves

conservatively. For the Indonesian sites presented in Table 2.3b there is a range in

chloride from 189 to 2330 mgl-'. These figures at first sight do not seem to offer much

hope for a source term; the Indonesian leachates are, on average, lower in chloride than

the Mexican ones. Perhaps this difference reflects seasonal variation of leachate

chemistry and volume being generated. For instance during the rainy season or shortly

thereafter one would expect leachate concentrations to drop due to dilution by infiltrating

rain water. Conversely during the dry season one would expect more concentrated

leachates to develop due to evapotranspiration from the waste. Indeed, early work by

Robinson and Maris, (1979) showed that leachate flow and strength in U.K. landfills is

seasonal and irregular, depending upon rainfall and evaporation.

wc/95/7 12 Issue 2

Rosadi and Sukrisno, (1993) document a limited time series of data for the Dago Landfill

in Bandung, Indonesia. The data are examined in relation to rainfall in Figure 2.4. In

Bandung the peak of the rainfall event occurs between October and March with the

lowest rainfall occurring in June and July. Although the chemical data are incomplete

tentative conclusions can be drawn from the plot. There appears to have been a period

of leachate concentration during the 1991 dry season due to evaporative losses, followed

by dilution caused by rainy season infiltration flushing through the waste in the early part

of 1992. Clearly there is some lag between the rainfall events and the drop in chloride

concentration from a maximum in excess of 4000 mgl-'. Similar plots can be constructed

for bicarbonate and TOC.

Evidence of leachate dilution during precipitation events also comes from the Merida site

in Mexico. Sauri Riancho and Castillo Borges, (1992) and Castillo Borges and Sauri

Riancho, ( 1993) record partial analyses of leachate collected during composting

experiments on Merida solid waste. They concluded that the leachate quality was not

only a function of the period of composting, but also precipitation event frequency. They

noted that after long, dry periods, leachates were more concentrated with conductivity

values > 15000 pmho cm-' dropping as low as 880 pmho cm-' during the spring.

The chloride data for the Mexican sites ranges from 863 to 6880 mgl-'. Referring back

to Table 2.3a, sample M9 was collected from the leachate deriving from old waste

whereas samples MIO and M11 were collected from wastes deposited in the last four

years at the Merida site. Sample M11 represents diluted leachate whereas M 12 is a

concentrated leachate collected directly at the waste face. An average value for a source

term is difficult to come up with and perhaps it is more meaningful to select a value

which would represent the peak impact. Once again we are faced with the problem of not

knowing how representative these samples are of leachates in general.

Table 2.4 summarises the main organic components which were identified in the

leachates from the Mexican and Indonesian sites. The GC-MS traces obtained show many

recurring peaks, all the samples are complex, nearly all components being present in

concentrations greater than 1Opgl-' and with maximum concentrations reaching 40mgl-'.

A concern is the appearance of the chlorinated solvent, tetrachloroethane, in both

Mexican and Indonesian leachates at concentration levels of over 0.5 mgl-' . Chlorinated

WCl95l7 13 Issue 2

solvents are persistent, carcinogenic and mutanogenic; other solvents identified were

xylene, toluene, and N,N-dimethyl formamide. Schultz and Kjeldsen, ( 1986) have

demonstrated that the occurrence of these compounds in a leachate is typical of landfills

used for the disposal of domestic and industrial wastes. Industrial organic compounds

associated with rubber manufacture, e.g. sulphur, thiophene derivatives, thiazole

derivatives and n-butyl benzene sulphonamide were noted in samples I-6,7.

Sample M4 contained a large concentration of nonanoic acid which has many industrial

applications, and a range of terpenes and derivatives which are used as flavourings and

fragrance compounds. Sample M6 was the most complex with many industrial

intermediates identified. Many responses gave poorly characterisable spectra belonging

to alcohols, ketones and esters and non-identifiable spectra representing oxidation,

polymerisation, or degradation products of the leachate.

Concentrations of the heavy metals are variable in the leachates analysed. In the case of

the Leon Guanajuato landfill site the high input of tannery waste is clearly reflected in the

high levels of chromium, both in the leachates and the run-off from the waste. (samples

M6 to M8, Table 2.2a; Plate 2.6). However, there is no systematic picture of heavy metal

precipitation as a function of age of waste and the idea of acetogenic - methanogenic

cycling in the waste is supported.

WCl9517 14 Issue 2

Table 2.4 Organic compounds identified in Mexican and Indonesian Leachates

M2

M4

1-3

r-4

1-6

1-7

Compound

toluene, xylene, tetrachloroethane, diethyltoluamide, alkanes, dibutyl

phthalate, bis(2-ethylhexyl) phthalate, caprolactum, derivatives of thiazole

and thiophene

siloxane, xylene, tetrachloroethane, methyl palmitate, dibutyl phthalate,

bis(2-ethylhexyl) phthalate

toluene, siloxane, xylene, isopropyl benzene, siloxane, nonanoic acid

isomer, ketone, alkylbenzene, dibutyl phthalate, bis(2-ethylhexyl) phthalate,

palm i t ic acid, a1 kane

toluene, siloxane, xylene, tetrachloroethane, ketone, alkane,

diphenylmethane, myristic acid, di isobutyl phthalate, palmitic acid,

hexanedioic acid, bis(2-ethylhexyl) phthalate, oleic and stearic acids

siloxane, tetrachloroethane, alkene, thiophene derivative, dibutyl phthalate,

oleic and stearic acids, bis(2-ethylhexyl) phthalate, alkane, palmitic acid

toluene, siloxane, xylene, 1 -ethyl cyclo hexanol, alkane, 2-methyl propanoic

acid, cyclohexane carboxylic acid, benzothiazole, lauric acid, diethyl

toluamide, myristic acid, pentadecanoic acid, dibutyl phthalate, alkane,

bis(2-ethylhexyl) phthalate, decanoic acids, palmitic acid

chloroiodomethane, toluene, siloxane, xylene, tetramethyl pyrazine,

nonanoic acid isomer, bis(2-ethylhexyl) phthalate, stearic acid, palmitic

acid, dibutyl phthalate, N-butyl benezene suphonamide.

2.2 Geological and Hvdrogeological Factor Group

The position of the water table in relation to the base of the waste is extremely important.

Provided that no permeable pathways exist then the thicker the unsaturated zone the more

capacity is potentially available to attenuate any leachate released by the waste. If the

water table is within the landfill then there is the possibility of nearby abstraction wells

drawing in leachate from the landfill. The proximity of an abstraction to a site combined

with its abstraction rate will influence movement of contamination by increasing the

1s Issue 2 WCl9517

hydraulic gradient in the direction of the abstraction well.

Once contamination reaches the water table saturated flow processes will take over and

soluble contaminants will move down the hydraulic gradient to points of outflow such

as rivers, springs and abstracting boreholes. At some point along the flow path the

concentration of the contaminants will reach a safe level due to dilution processes and

bacteria will die off and no longer pose a risk to water supply. A primary objective is to

define this safe distance for various geological - hydrogeological scenarios.

2.2.1 The Unsaturated Zone

The unsaturated zone is that part of the aquifer which lies between the surface and the

water table; it is also known as the vadose zone. This zone is the first defence between

the aquifer and the leachate generating waste site. Void spaces in the rock are only

partially filled with fluid and the rest of the volume is taken up by air. There will be a

tendency for contaminants to be held in the profile by the negative pressure head, a

characteristic of the vadose zone. The residence time of this phenomenon will depend

to a large extent on the moisture content which in turn controls the unsaturated hydraulic

conductivity and transport rate. The retention of contaminants in the unsaturated zone

will therefore increase the opportunity for degradation and leachate rock interactions

leading to attenuation.

The attenuation capacity depends on the presence of a chemically active rock type such

as clay, which usually has a high cation exchange capacity, or limestone, which

encourages biodegradation and buffering, or sediment with a high organic matter content.

Mather, (1989) reports that the results of UK landfill research demonstrate that significant

anaerobic biodegradation is possible in the unsaturated zone. This is especially

noticeable for landfills situated on the calcareous Chalk where the presence of CO,

reduces the pH of the unsaturated zone. On the other hand sites on the carbonate free

Permo-Triassic Sandstones demonstrate little attenuation. Mather op.cit. concluded that

the most important factor controlling the biodegradation in the unsaturated zone is the

significantly higher buffering capacity of the calcareous rocks. An earlier study by Rees,

( I 98 1 ) identified ion exchange and dilution as major factors in leachate attenuation in the

we19517 16 Issue 2

Lower Chalk, but concluded that biodegradation was a significant process in the

overlying gravels.

The key parameters which determine the flow rates or conversely the retention times of

contaminants in an unsaturated profile are thickness, moisture content, and the

unsaturated hydraulic conductivity. The last two parameters are related by the

unsaturated form of Darcy’s Law, which mathematically is given by Richard’s Equation:

ae at

where - change in volumetric water content with time

K($) = the unsaturated hydraulic conductivity at a given matric potential, $

V@ = the gradient q j the soil water potential @

For one dimensional flow the above equation can be reduced to:

The equation is difficult to handle, it is not only highly non-linear, requiring numerical

methods for its solution, but also requires a knowledge of the relationship between soil

moisture content and suction ($), (see for example Lappala et al., 1987). Again, very

little information is available in the literature for these relationships for various rock

types.

The matric potential, q, is related to the volumetric moisture content, 8, and is dependant

on the type of the media or soil. At the water table, the soil is saturated and the water

content, e,, is equivalent to the porosity. Moving up from the water table, the soil

remains saturated until the pressure head has become sufficiently negative that water can

begin to drain. This critical head is known as the bubbling pressure. The water content

then continues to decline until it reaches a minimum, e,, below which a reduction in

wcr9517 17 Issue 2

pressure head has no effect. This produces a soil-moisture characteristic curve as shown

in Figure 2.5, (Van Genuchten, 1980). These curves are different for different media and

typically, the transition from 8, to Q becomes sharper as the pore-size distribution

narrows or the hydraulic conductivity increases.

The curves are generally fitted using empirical solutions, and the relationship of water

content to pressure head used here is:

where h, and a are constants for a particular soil. This is similar to that proposed by Van

Genuchten, ( I 980) if 8, is taken to be zero. The unsaturated hydraulic conductivity can

then be derived from 0 using another empirical relationship:

where K,= saturated conductivity and p is another constant. (8/8Jp is termed the relative

hydraulic conductivity and its relationship to pressure head is illustrated in Figure 2.5.

2.2.2 Aquifer Properties and Contaminant Transport

Groundwater flow under saturated conditions is governed by Darcy's Law which may be

expressed in the following way for the one dimensional case as:

where v = average linear velocity (ms-')

k = hydraulic conductivity

i = hydraulic gradient

wc/95/7 18 Issue 2

q = effective porosity

Note that the hydraulic conductivity is no longer dependant upon moisture content.

However, both the hydraulic conductivity and porosity are dependant on the lithology.

Values of hydraulic conductivity are given by Freeze and Cherry, (1979) and are

tabulated in Table 2.5. These are typical values for the porous medium, however it must

be realised that where significant fracturing is present then by-pass flow can operate

which effectively short circuits the system and significantly reduces travel times.

The effective porosity may be described as that porosity which actively contributes to the

flow of groundwater. It takes a range of values depending on the type of material, e.g.,

for open structured gravels it can be as high as 40% while for a shale the range of values

is from zero to 10%. Fractured rocks too can have high porosities, as high as 50%.

Table 2.5 is modified from Freeze and Cherry, (1979) and lists the range of values of

porosity for some common rock types.

Steady state groundwater flow in three dimensions is modelled by the Laplace equation:

= o d2h d2h + Ty- + ' 7 2

d2h Tx- ax2 a y 2

for the 1-D case with recharge this may he written as :

d 2h dx

Tx- - - -4

where: T, is the transmissivity in the x-direction,

it is the product qf hydraulic conductivity and saturated thickness

q is the recharge

w c / 9 5 /7 19 Issue 2

An analytical solution of the above equation for the boundary value problem of a

water table aquifer with a landfill on one boundary with a discharge rate Q, an aquifer

recharge rate q, and a fixed head condition is:

4 2 h = --(L - x’) + g(L - x) + h, -‘ 2T T

where: hx is the head at distance x from the landfill

L is the distance ,from the landfill to the fixed head boundary

h, is the head at the,ftxed head boundary

The above equation is readily handled by a spread sheet and groundwater flow

velocity can be calculated. Hence an idea of travel time to a specific distance from

the landfill can be obtained. However, because of the quadratic term in the above

equation, it is necessary to define the head gradient over small intervals of the flow

path, the use of an average gradient introduces significant error. It is suited to

modelling situations where a low hydraulic gradient is present, e.g. river floodplains.

wc/95/7 20 Issue 2

Table 2.5 Typical values of hydraulic conductivity (ms-I) after Freeze and Cherry,

( 1979)

LITNOLOGY HYDRAULIC CONDUCTIVITY

Karst Limestone

Fractured Igneous and metamorphic

Limestone and dolomite

,Sandstone

10-6 - 10-*

Massive igneous and metamorphic

Clay

Shale

Silty sand

Silt

Sand (clean)

1 0 . ~ - I O - ~

5x10" - .5x10-'o

> IO-"'

1 0 - 8 - 10-1'

> 1 o - ~

io-? - 10-7

10-5 - 10-9

1 0-2 - 5x 10.'

LITHOLOGY

11 Gravel I 1 - 10-?

-Qmlr

POROSITY (%)

Table 2.6 Porosity values for some common rock types

Kars t Limes tone

Fractured Igneous and

metamorphic

Limestone and dolomite

Sandstone

Massive igneous and metamorphic

Clay

Shale

5 - 50

0 - 10 ( 5 - SO for basalt)

0 - 20

5 - 30

0 - 5

25 - 40

0 - 10

Gravel 25 - 40

11 Sand I 2.5 - 50

wc19517 21 Issue 2

When a conservative contaminant travels through a porous aquifer medium its movement

is governed by the advection dispersion equation (ADE) which can be expressed as

follows:

where: D,, is longitudinal hydrodynamic dispersion coefficient

v, is the average linear velocity in the x-direction

C is the mass per unit volume of solute

The longitudinal hydrodynamic dispersion can be thought of as a parameter that accounts

for the mixing of solute due to mechanical effects in the direction of flow and diffusion

around the grains in the aquifer matrix. The hydrodynamic dispersion is defined

mathematically as:

D, = v,a + q D'

where: a is the dispersivity

D* is the diffusion coefficient

For most situations the effects of diffusion can be ignored retaining the term involving

average linear velocity and dispersivity. Dispersivity has dimensions of length and is

strongly scale dependant, laboratory experiments indicating centimetre scale values

while field experiments indicate values of ten's of metres.

The ADE assumes that the porous medium is homogeneous, isotropic and saturated with

fluid. An analytical solution, suitable for the situation of a landfill releasing leachate into

an aquifer, with the following initial and boundary conditions:

C(x,O) = o x 2 0

C(0, t) = C" t 2 0

WCf95f7 22 Issue 2

C(-,t) = o t 2 0

is given by Fetter, (1993) as:

where: CO is the initial concentration,

L is the distance from the point of injection to the point of measurement.

The ADE is also used to estimate the transport and attenuation of bacteria and

contaminants which are reversibly adsorbed and result in a retardation in transport rate.

Isotherms relate the amount of a species sorbed to the concentration in the aqueous

phase. If the sorbed concentration of a contaminant A is S then:

The measured relationship between A and AS at a constant temperature is known as the

sorption isotherm. This is often expressed by the Langmuir Isotherm which is:

WT 1 + h,CA

where: W, = number of moles of solute per gram of rock

W,= number of moles of adsorption sites per gram of rock

B, is a constant

c, is the concentration of the contaminant A in solution.

the Freundlich Isotherm is more useful and is given by:

w cl9 517 23 Issue 2

where b, is a constant

l/n is a measure of the non linearity with respect to cA

When l/n = 1 then the isotherm is linear and

where K, the distribution coefficient, is an equilibrium thermodynamic constant for a

particular reaction.

K,,'s are relatively easy to determine empirically in the laboratory by batch sorption

experiments. However, equilibrium is rarely achieved and the retardation distribution

ratio, R,, which does not necessarily imply equilibrium or reversibility, is more often

used.

The retardation, R,, of a species relative to the flow of water can be determined using the

following equation:-

where p is the density of the aquifer material.

Such retardations are readily incorporated into analytical solutions of the ADE (see for

example Van Genuchten, 198 1) and are used for lack of better data, even though, strictly

speaking, the isotherms may not be linear.

wc/95/7 24 Issue 2

Examples of the application of the ADE in modelling the long term impacts of leachate

plumes from waste disposal sites are given by Valochi and Herzog, (1988). Dasgupta

et al., ( I 984) employed a finite difference approximation to show that the ADE is

insensitive to chemical parameters, but sensitive to the dispersion coefficient and the

groundwater flow velocity. Serrano, (1 992) has demonstrated the effect of recharge on

the velocity field of contaminant transport and the evolving nature of dispersion

coefficients.

Kent et ul., (1985) use a nomogram solution to predict plume migration and determine

the plume centre line concentration. The nomogram can also be used to predict a solute

concentration in space and time down stream from the source. A similar approach by

Loxham, (1988) is used to determine the breakthrough times of contaminant arrival at

specified distances from a site. Both approaches provide a deterministic estimate of the

hazard posed by a contaminant source by comparing the derived plume concentrations

with approved drinking water standards. A limitation is that transport through the

unsaturated zone is not predicted and recharge is not considered.

2.2.3 Hydraulic Gradient

The third variable in the Darcy equation is the hydraulic gradient which is influenced by

the physical setting of the site. The steeper the gradient the faster the travel time. The

problem in the field is how to determine a value of the gradient without drilling

boreholes. If records of groundwater levels are available for at least three boreholes in

an area it is possible to calculate the gradient of the water table in any particular direction.

If this information is not available then a rough and ready estimate of the hydraulic

gradient can be obtained from the ground slope, the assumption is that the water table is

coplanar with the surface topography. Furthermore, if one assumes that the aquifer is

isotropic, i.e. that the hydraulic conductivity is the same in all directions, then

groundwater flow will be perpendicular to the groundwater contours.

wc19.517 25 Issue 2

2.2.4 Recharge

Recharge to aquifers can occur from a number of sources. Ward and Robinson, (1 990)

lists the following:

e infiltration as part of the total precipitation on the ground surface

seepage through the bed and banks of surface water bodies

leakage from associated aquifers and aquitards

artificial recharge due to irrigation; leakage from supply mains, and

e

e

e

groundwater augmentation schemes

In most situations direct infiltration from precipitation is the major component of the

recharge. Water balance calculations are made to assess the amount of recharge. The

methodology is very similar to that already presented for calculating leachate production

in a landfill and is based on the concept of soil moisture deficit, SMD. When a soil’s

capillary forces are in equilibrium with gravity it is said to be at field capacity. Any

precipitation which falls in excess of the field capacity can become groundwater recharge

or runoff. After a rainfall event ceases water is lost by evaporation and plant uptake and

soil moisture is gradually depleted to produce a soil moisture deficit, i.e. the amount of

water required to return to field capacity. Initially the potential evaporation rate is met

by water from the soil moisture storage until a critical value, C, known as the root

constant. Beyond this evapotranspiration is assumed to occur at 10% of the potential

rate. With continuing evaporation the soil will reach a point where plants can no longer

extract water and this is termed the wilting point, D. The Penman - Grindley model is

the simplest and most widely used soil moisture budgetting method which takes into

account the concept. The following scheme of calculation is modified from Lerner et al.,

( 1 990):

psmd, = (ro+ smd, + ae,) - p,

r, = - psmd,

smd,,, = psmd, - r,

when psmd, < 0

when psmd, > 0

Actual evaporation is derived from the potential evaporation as follows:

WCl9517 26 Issue 2

ae, = pe,

ae, = 0.1 pe,

ae, = p,

when smd, < C or p, >pe

when C I smd, < D and p, < pe

when smd, = D and p, < pe

where: smd, = soil moisture deficit at the start of time period i

ae, = actual evaporation during period i

pe, = potential evaporation during period i

p, = precipitation during period i

r, = recharge during period i

psmd, = intermediate variable

ro =run off

C = root constant

D = wilting point

It is generally found that using time periods greater than a day produces overestimates of

actual evaporation and underestimates of recharge. The above algorithm can be

programmed into a spreadsheet and the recharge calculated from daily data of

precipitation and evaporation. The choice of root constant and wilting point depends on

the particular crop cover. Table 2.7 is based on Shaw, (1988) and Lerner, et al., (1990)

and lists some common root constants and wilting points.

Runoff is linked to the topography and in general the steeper the slope the higher the

runoff. Blakey, (1992) quotes the following formula to estimate runoff:

R = C.P

where R = runoff

C = runoff coefficient

P = uniform rate of rainfall intensity

Typical values of C for a bare soil are given as a function of slope. For a flat slope of

~ 2 % a value of 0.6 is quoted, increasing to 0.7 for a 2 - 10% slope and up to 0.82 for

slopes > 10%. A common empirical approach is to assume that 10% of precipitation goes

to runoff.

WCl9517 27 Issue 2

Table 2.7 Root constants and wilting points for common vegetation

Vegetation Root Constant

Type (mm,

Wilting Point

(mm)

Grassland

Root Crops

Bare Fallow

Cereals

Woodland

2.3 Fate G r o w

75 125

100 150

25 25

140 203

200 250

The fate group takes account of the distance from the waste disposal site to the nearest

vulnerable population. Discounting the usual nuisances of litter, birds, smells and vermin

associated with waste disposal sites, the risk is dependant on whether human activity

interacts with the hydrogeological system. This interaction can take the form of

abstracting water from boreholes down gradient of the site, or modifying the hydraulic

gradient by pumping to intercept contamination; taking water from streams which may

be in hydraulic contact with a contaminated aquifer, or from surface water runoff from

the landfill/ landraise.

The scale of the interaction is dependant on both the distance of the abstraction from the

landfill and the volume abstracted. Pumping has the effect of causing drawdown of the

water table and hence an increase in the hydraulic gradient in the vicinity of the well.

Referring back to Darcy’s Law this will have the effect of increasing the average linear

velocity and hence reducing the travel time of a contaminant to the well, dispersion will

also be reduced.

WC/95/7 28 Issue 2

3 THE MODEL APPROACH

The main steps in constructing and implementing the numerical simulator were to:

develop the conceptual model

predict scenario response and determine factor response to parameter

design the model using a suitable computer code

variation. The latter is a form of sensitivity analysis.

- 3.1 The Conceptual Model

3.1.1 Background

The conceptual model is limited by the fact that it has to be representable using equations

similar to those given in Section 2. With this in mind, the criteria in producing the

conceptual model were, in order of priority:

to keep it simple to use and simple to apply to real situations

using its numerical representation, to make extrapolation to real sites as quantified

and non-subjective as possible

to include as many of the factors described in Section 2 as possible

To reduce subjectivity, the approach used was to calculate time periods for migrating

contamination to achieve specific concentrations at various distances from the landfill.

Although a simple conceptual model is bound to contain assumptions that do not apply

in every individual case, it was anticipated that the numerical model could yield a semi-

quantitative assessment of the potential hazard at a certain distance from the landfill.

From Section 2.2, it is clear that two sets of equations and parameters are required to

solve for water and contaminant movement in the unsaturated and saturated zones.

Therefore, the numerical representation of the conceptual model must treat these

separately. Initially, it was intended to use simple analytical solutions for contaminant

migration in a 2-D vertical plane. However, it was found that in attempting to introduce

WCI 951 7 29 Issue 2

more factors in accordance with the third objective listed above, the equations became

too complex and the results were very dependent on solution assumptions. Comparison

with results from a finite-element code indicated that the latter were more reliable and,

although this increased complexity for the user, it was decided that this presented the

better option if a clear user interface and instructions are provided.

3.1.2 The Model

In the conceptual model adopted here, the waste disposal site sits on an unsaturated

geological media which overlies an unconfined aquifer (Figure 2.1). For simplicity, the

conceptual model assumes the saturated and unsaturated zones are composed of the same

homogeneous media. It does not allow for changes in hydraulic or geological properties

within each section of the transport path. For the numerical solving of the model, the

path is treated in two sections, the results of which must be summed in order to obtain

the total travel time. These are:

(i) The time taken for the pollutant to travel vertically downwards in a line through the

unsaturated zone from the landfill to the water table. The time is recorded when the

solution entering the saturated zone or groundwater has reached the concentration

ratios (i.e. C,/C,, where = concentration at the point of observatiog, C =

concentration in the source or landfill) of 0.001, 0.5. This is because the pollutant

will reach the water table in trace quantities long before it achieves the same

concentration as in the landfill (Figure 3.1). These trace quantities will be

significant for particularly hazardous pollutants or bacteria hence the time taken to

reach a C,/C, of 0.001 should be used. For other pollutants, the time taken to reach

a C,/C, of 0.5 can probably be used as an average.

(ii) Treating the infiltration from the unsaturated zone below the landfill as a source,

the travel time in the saturated zone is calculated across a two-dimensional,

horizontal plane in which the positions of the landfill boundaries are specified. The

two boundaries of the plane that are roughly parallel to the dominant flow direction

are treated as no-flow boundaries. The head at the outflow boundary is fixed by

considering this to be represented by a river or surface water. The other boundary

wc/95/7 30 Issue 2

can either be fixed-head to create a topographic hydraulic gradient or set to no-flow.

A borehole abstraction point can be specified in terms of location and volume of

water abstracted. The time is recorded when a predetermined C,/C, ratio is reached

at specified observation points.

The following assumptions are important:

0

0

0

0

0

0

3.2 _I

Because the unsaturated zone is treated as a I-D vertical column, no account is

taken of lateral dispersion and the resulting wider area of pollutant reaching the

water table outside of the boundaries of the landfill.

It is assumed that the landfill is unlined. However, the unsaturated vertical section

could be used to model infiltration through a clay liner and the result added to the

other two section results if desired.

Seasonal variations in recharge are ignored.

The source term from the landfill for the 2-D model is assumed to be constant

throughout the simulation. For some of the longer timescales presented here this

would not be the case.

Surface runoff is assumed to be absent. If this is significant then it may provide a

much quicker travel time to the point of interest, e.g. the high Cr values in surface

streams due to leaching of tannery waste in the Leon landfill.

Changes in hydraulic conductivity and other aquifer properties through each section

of the travel path are ignored.

The aquifer is flat and the thickness is constant, although a hydraulic gradient can

be imposed.

Development of the Numerical Model

Two finite-element codes, based on FEMWATER, Yeh and Ward, ( 1980), to calculate

the water fluxes and velocities and FEMWASTE, Yeh and Ward, (1 98 l ) , to calculate the

pollutant travel times, are used. Four front-end programs for these codes have been

written which allow the setting up of the above sections with a variety of different

parameters; two deal with the saturated zone and two with the unsaturated zone. Tables

3.1 and 3.2 detail the parameters requested by these programs. A grid of rectangular

WCf95f7 31 Issue 2

elements based on the size of the area or length of the line to be modelled is automatically

generated by the front-end programs. For the 2-D horizontal modelling, the elements

generated are smallest around the landfill where numerical changes are expected to be

greatest (Figure 3.2). Observation points can be set on the 2-D plane in relation to the

position of the landfill.

A program is also supplied which takes the output from the 2-D modelling and gives the

user travel times for entered C,/C, ratios. The entire system is written in FORTRAN77

and compiled for an Il3M PC or compatible. To increase portability, no graphical

packages are included, but a front-end menu-driven interface is provided. (The program

can be obtained from the authors on request.)

_. 3.3 Parameter Variation

Initially, it was intended to run a series of simulations for different types of geological

media with average hydraulic parameters (Table 3.3; cf Tables 2.5 to 2.7) under different

climatic conditions. Two figures were to be produced for each simulation showing travel

times for chloride to reach significant C,/C, ratios, and C,/C, ratios after 100 days (e.g.

Figure 3.3). The latter is based on the maximum time pathogenic bacteria may survive

(Canter et ul., 1987), and therefore the maximum distance from the landfill which may

be affected. The user could then select the simulation which most closely resembled the

situation of interest and extrapolate the timeslratios to fit the characteristics of the site

being examined. However, the dependence of results on some parameters which are very

difficult to “average” in the model (e.g. hydraulic gradient, depth to water table, leachate

composition) and the resulting high degree of uncertainty that is introduced during

extrapolation meant that this was unsatisfactory with regard to the second criterium for

the model given in Section 3.1.1. Therefore, it is preferable that the user is given the

model and enters the data as best known for the site, thereby eliminating a lot of the

extrapolation. Where the user is particularly uncertain of a parameter, the model can be

run for a range of values to determine model sensitivity.

This section considers the parameters that the user must enter, discusses what values

might be applicable and outlines the effect of uncertainty in them. The layout follows the

WCl95l7 32 Issue 2

factor groupings given in Section 2.

3.3.1 Site Group

3.3.1.1 Site Size

For the unsaturated zone, modelling is one dimensional and so the areal extent of the

landfill is not important. For the 2-D models, the four corners of the landfill are specified

giving a rectangular area within the modelling plane through which the specified

infiltration behaviour of the landfill is taken to apply. For most of the simulations

presented here, a 10000m2 landfill is placed between 100 and 200m from the left (upper)

boundary giving a distance of lOOOm to the right or outflow boundary (Figure 3.2).

Changes in the areal size of the landfill are most significant for high infiltrationlrecharge

rates where either the hydraulic conductivity is low (e.g. Figure 3.4) or, if the

conductivity is high, where the regional head gradient is low.

Figure 3.4 considers chloride migration in a fractured basement lithology which has a low

conductivity (1 0-7 ms-') under tropical climate conditions. Doubling the landfill area

reduces the travel time to the river boundary lOOOm away to 65 years from 76.4 years.

However, the proportional decrease is not constant along the main travel path so no

simple correction can be applied. This is because changing the landfill size affects

mainly the recharge mound local to the landfill and so the head gradient is not altered in

a linear manner. This also means that if the infiltration rate over the landfill is similar to

the general infiltration rate, increasing the areal size will not affect the head gradients.

If a topographic head gradient is imposed for a media with a high hydraulic conductivity

such as a gravel, this will dominate over a gradient due to recharge. Hence increasing the

landfill area will not produce a major effect on the main flow path times. For the same

scenario as Figure 3.3, but with a 20000m2 landfill, the transit time for chloride at the

river boundary comes down from 123 days to 115 days. However, if the topographic

gradient of 1 :500 is removed from this scenario to simulate flat terrain, the times become

7 1.3 and 57.7 years. This difference is of a similar order to the effect on the fractured

basement lithology in Figure 3.4.

we19517 33 Issue 2

Irrespective of changes along the main flow path due to changes in the head gradient, it

should be borne in mind that increasing the size of the landfill increases the source term

for the volume of leachate thereby increasing dispersion path lengths. Additionally,

changes in the shape of the landfill from the square used here become important in this

respect.

3.3.1.2 Site Climate Factor

For the 2-D modelling, the effect of climate is dealt with by two parameters; infiltration

rate over the area, and infiltration rate over the waste site. Section 2.2.2 described how

infiltration over a landfill can be higher than the surrounding area once the waste

becomes saturated. A value of 50% of rainfall has been used here. As mentioned above,

the modelling results obviously become more sensitive to this parameter if the area1 size

of the landfill is large with respect to the modelling area and if the difference between

general and landfill infiltration rates is high. As with site size, this parameter is only

significant in scenarios with either low topographic head gradients in high conductivity

media or with low conductivity media. Sensitivity of the 2-D modelling to infiltration

rate in general is discussed under Section 3.3.2.4 below.

For the 1-D unsaturated zone modelling, the landfill infiltration rate is very important in

determining travel time.

3.3.1.3 Waste and Leachate Composition Factors

The difficulties of estimating average leachate composition based on average waste

composition have been described in Section 2.1.4. Although some pollutants such as

heavy metals and organic compounds are much more hazardous than others, without

knowing the concentration in the source term it is not possible to assess the significance

of different calculated values of C,/C,. The model allows the user to specify the C,/C,

ratio of interest based on site-specific information about leachate composition and the

contaminant of interest. If these are unknown or if the model is being used in a general

manner, a useful yardstick may be to look at chloride and bacteria (Figure 3.3). Chloride

wc/95/7 34 Issue 2

can be assumed to be conservative (K, = 0) whilst bacteria are only conservative in media

with fractures or large pore spaces where they are not filtered out or adsorbed.

The recommended upper limit for chloride in drinking water is 400mgl ' (Water Supply

(Water Quality) Regulations, 1989), whereas landfill sites often have leachate with

concentrations in excess of 4500mgl-' (Christensen et al., 1994). Using a measured value

from this study of 3000mgl-', a C,/C, ratio of 0.133 would therefore be hazardous to

drinking water supplies. The times given in the 2-D scenarios presented here are based

on this. Additionally, bacteria may only survive for up to 100 days under favourable

conditions, Canter, et al. (1987), but any registered C,/C, value can be considered

significant. Therefore, a contour should be drawn around all registered values within a

100-day travel-time from the landfill.

Uncertainty in significant C,/C, values is difficult to manage since the change of the ratio

at a point over time is not linear, so extrapolation from one value to another is not

possible. It is probably better to assume a conservative, significant ratio as a worst case.

3.3.2 Geological and Hydrogeological Factor Group

3.3.2.1 Unsaturated Zone

Each medium has a characteristic soil moisture and relative hydraulic conductivity curve

similar to those shown in Figure 2.6, but with different curvatures. Of the parameters

required to model the unsaturated zone given in Table 3.2, the saturated hydraulic

conductivity, depth to the water table, and infiltration rate do not affect the shape of the

curves. They do, however, affect the actual pressure heads achieved and, in combination

with distribution coefficient and density, affect the travel time. Different hydraulic

conductivities, densities, landfill infiltration rates and distribution coefficients can be

entered for the unsaturated and saturated zones, but generally are not. Variations in these

parameters are therefore discussed under aquifer properties below. The depth to the

water table is an important parameter and errors in calculated time due to uncertainties

in depth are essentially linearly proportional.

wc/95/7 35 Issue 2

The porosity determines the upper asymptotic value of the characteristic curve without

changing the pressure head. Therefore, changing the porosity changes the shape of the

curve. Hence the fitting parameters, for the two curves in Figure 2.6, using the numerical

solutions given in Section 2.2.1, are porosity (=e,), H,, a , and 13. The approach taken

here is to fit the characteristic curves for sand, silt and clay given by Van Genuchten,

( I 980), using a non-linear parameter estimation technique to obtain values for H,, a , and

13 given a saturated porosity (e.g. Figure 3.5). For similar media with similar porosities,

these can then be used to estimate travel times for different infiltration rates, saturated

conductivities, and depths to water table. If the site medium has a different curve or

porosity, then the user will have to interpolate between the values given. Table 3.4

illustrates the difference in travel times resulting from changing the fitting parameters

between the curves used.

If this approach introduces too much uncertainty, then a more conservative approach can

be taken by assuming saturated vertical flow as was done by Parsons and Jolly, (1 994b)

in the WASP scheme. This may, in fact, be realistic if a hourglass-shaped zone of

saturation exists between the landfill and water table. The travel time for such a section

can be calculated by assuming the average velocity, v, is equal to the saturated hydraulic

conductivity divided by the effective porosity, the hydraulic gradient is one in this case.

The analytical solution below (Fetter, 1993) can then be used to calculate when C,/C, =

0.5 assuming a K, of zero:

where: L = depth to water table in metres

a L = dispersion coefficient (usually 0.1L)

t = time in seconds.

The times from this equation are compared to those calculated using the unsaturated

approach in Table 3.4 and it can be seen that there can be an order of magnitude

WCf95f7 36 Issue 2

difference.

3.3.2.2 Aquifer Properties and Contaminant Transport

Hydraulic conductivity and porosity are important aquifer properties with respect to water

and therefore contaminant transport. All rock-types give a range of these parameters

according to features such as cementation, fracturing, and sorting. Therefore, in the

scenarios presented here, values have been extracted from ranges such as those given in

Tables 2.5 and 2.6 to give a good spread of modelling values (Table 3.3). The user

should either have a good idea of the values applicable to the area or be able to extract

suitable values from the tables. To model the conservative case, the higher conductivity

and lower porosity limits should be used in conjunction as these will give the fastest

travel times

In Table 3.3 it can be seen that the main change in the 2-D section for models considering

fractured basement and shale is a change in hydraulic conductivity from 10-7 to w9 ms-I.

Under arid conditions, this results in an increase in travel time at the river boundary (see

Figure 3.3 for scale) from 406 years to 3390 years. Obviously these times are of little

importance in terms of landfill hazard since the source term will not be present for this

period of time. However, they are presented here for illustration of the effect of

parameter uncertainty. Changing from sandstone to weathered basement under tropical

conditions causes a much smaller relative change, but still results in a twofold increase

from 59 to 107 years. Hydraulic conductivities below 10-9 ms-' typical of non fractured

clay and crystalline rocks, are usually lower than the infiltration rate (see 3.3.2.4), and

tend to give fluxes which are too low to model successfully; the numerical model either

Fails to converge or models water tables above ground level. Therefore, if a hydraulic

conductivity is lower than the infiltration rate, the infiltration rate should be used to

calculate travel time instead of hydraulic conductivity. This will give a worst case

scenario. For the 2-D modelling, it is usually assumed that the medium is homogeneous

so that hydraulic conductivity in orthogonal directions is equal. However, different

values can be entered if desired and this will affect travel times away from the main flow

path.

WCl95l7 37 h u e 2

There is also a porosity difference between sandstone and weathered basement and this

will contribute to the time differences mentioned above. The change in porosity between

shale and fractured clay (0.05 to 0.35) leads to an increase in travel time of 3390 to 23671

years.

For 2-D modelling it is always assumed that the aquifer is unconfined, i.e., the top surface

is always set to be above the water table. This makes the actual height of the top of the

aquifer unimportant since the position of the water table will always be the same within

the aquifer. However, the elevation of the base of the aquifer in relation to the fixed head

at the outflow boundary is important as this will change the saturated thickness or

transmissivity of the aquifer. Again, this is not an easy change to extrapolate since the

relationship between travel time and aquifer thickness is not linear. Doubling the fixed

head boundary to 50m for the scenario of siltstone in a tropical climate (Table 3.3) leads

to an increase in travel time from 238 to 294 years. Doubling it again to lOOm gives a

time of 448 years. The elevation and therefore head of the fixed head boundary should

be straight forward to estimate from topographic maps. However, estimating the relative

position of the base of the aquifer to this may be more problematic.

For contaminant transport, the effect of dispersion may also be one of the main causes

of uncertainty in the modelling and is also one of the most difficult to quantify. Both

travel path sections require a longitudinal dispersion coefficient (a,), with the 2-D

section requiring an estimate of a transverse (aT) coefficient as well. The value of these

coefficients is not well established and cannot be related to rock-type. It has been

suggested that a linear dependency between a L and distance can be assumed, Pickens

and Grisak, (1981), and a L is estimated here at 10% of the mean travel distance as

recommended. The transverse coefficient is also taken here by convention as one fifth

of the longitudinal coefficient (Walton, 1985). Walton (1985) also suggests that a

transverse coefficient equal to one tenth of the longitudinal coefficient can be used. The

effect of halving a, is that the travel times away from the landfill source parallel to the

longitudinal dispersion direction decrease whereas those away from this line increase,

Figure 3.6. Dispersion is a very important process in low conductivity media where large

dispersion coefficients are probably not realistic. If the longitudinal coefficient is set at

a maximum assumed value of 3m, Walton, (1985), and the transverse coefficient is

reduced accordingly, then in the shale example of Figure 3.7 all the travel times increase

we19517 38 Issue 2

with the largest relative changes being in the calculated times closest to the landfill and

up gradient from the landfill. These time changes cannot be related to the changes in the

coefficients by using a simple equation, and so guidance in the alteration of the models

for site-specific dispersion data cannot be given.

The other main aquifer parameters for pollutant transport are aquifer density and the

distribution coefficient, K,, which is also a contaminant property. The value of I$, is

different for each element or compound for a particular aquifer material and for a multi-

component leachate presents an infinite number of modelling possibilities. The times on

the scenarios presented are for the transport of chloride which is assumed to be

conservative, i.e. it is not retarded and K, = 0. To examine sensitivity to changes in K,,

porosity or aquifer density, the model can be re-run or, for speed, using the following

simple algorithm:

time = time for chloride from model run x ( 1 + (K, x densitylporosity))

3.3.2.3 Hydraulic Gradient

A hydraulic gradient may be imposed due to topography as well as due to volumes of

recharge and discharge. For aquifers with low hydraulic conductivities, the recharge

usually generates a large gradient. In such cases, unless it is extremely high, the

topographic head gradient can be ignored and the leftlupper boundary of the model can

be set to no-flow. However, for lithologies with high hydraulic conductivities, the

imposed topographic gradient can be the main driving force of contaminant flux and can

introduce large uncertainties. As a first approximation, the left boundary can be set to

reflect the topographic gradient minus the depth to the water table. The decrease in travel

times due to increasing gradient is essentially linear for imposed gradients significantly

higher than that due to the recharge (Figure 3.8; compare 1:500 and 1: lOO times).

However, if the imposed gradient is small, then the recharge mound over the landfill can

start to cause unusual modelling results such as decreasing travel times with decreasing

recharge (Figure 3.8; compare 1 : 1000 times between arid and temperate climates).

we19517 39 Issue 2

3.3.2.4 Recharge

Three suggested infiltration rates are given in Table 3.3 based on precipitation rates in

three climatic zones: arid, temperate, and tropical, Table 2.2 and Critchfield, (1983). The

derivation of infiltration rate from precipitation rate is hampered by lack of knowledge

and the variability of phenomena such as evapotranspiration and surface runoff. It can

be set arbitrarily at a percentage (e.g.5%) of total precipitation. For media with

conductivities of 10-9 ms ' , only the more arid infiltration rates are lower than the

conductivity and this is necessary for the numerical code to converge. Therefore, under

higher recharge conditions or higher infiltration rates, surface runoff from the landfill

may be a more rapid transport mechanism which is not allowed for here.

Any variation in recharge/infiltration rate will be reflected in the calculated heads. For

media with high hydraulic conductivities, variations will make little difference to the

calculated travel times if a high assumed topographic hydraulic gradient is specified

(Figure 3.8; compare 1 : 100 times for arid and temperate climates). However, as was

stated above, care must be taken if imposing a low topographic gradient. For media with

lower conductivities, the difference between travel times caused by changes in

rechargehnfiltration can be large. Figure 3.9 shows the difference in chloride travel times

in a siltstone between arid and temperate climates. This is analogous to changing the

infiltration rate from 5% to 50% under arid conditions.

3.3.3 Fate Group

3.3.3.1 Proximity of Local Population

This is not included implicitly in the model. By using the observation points, the user can

calculate the time taken for the C,/C, ratio to reach significant levels in groundwater at

the distance from the landfill of the population centre. They must then use a separate

method to determine the expected ingestion levels and dose-response relationships which

are not the subject of this report.

w c / 9 517 40 Issue 2

3.3.3.2 Distance to Nearest Abstraction Borehole and Volume of

Groundwater Abstracted.

The model accepts the locality of a borehole and the pumping rate. These are specific to

individual cases and are therefore not included on any of the generic figures given here.

The volume of water abstracted from surface water is not catered for as it is assumed that

this does not affect the fixed head boundary.

3.3.3.3 Distance to the Nearest Surface Water

This is implicit in the setting of the fixed head boundary. The size of the entire 2-D

modelling grid should equate to the distance of the landfill from the nearest river or

surface water body which can be assumed to represent a fixed head. A grid of 1200 by

1 OOOm has been used for the 2-D modelling scenarios presented here. As well as setting

the four corners of the grid, the user is asked to input the average element size of the grids

used to model both the I-D and 2-D sections and the modelling results do show some

sensitivity to this. Increasing the average size decreases the number of elements in the

model thereby speeding up the computation, but decreasing the sensitivity. Conversely,

decreasing the element size increases the sensitivity and the computation time, and

smaller time steps have to be used in the calculation of transit times. The models will

only allow a certain number of elements to be generated and so this provides a smallest

size limit. Given the other uncertainties in the modelling, high sensitivity is not essential

and an element size should be selected to give reasonable computation times.

Once the leachate-contaminated water has reached the boundary, the dilution effect of the

surface water is not considered. This could, knowing surface water flow rate, be

calculated separately since the volume of water with a known C,/C, ratio reaching the

boundary is given in the output of the model.

wc/95/7 41 Issue 2

4 CASE STUDIES

The following section of the report describes Indonesian and Mexican sites which were

visited during the project. Each site is evaluated using the numerical simulator developed

during the project. These evaluations are compared with evaluations made using existing

methodologies. For the purpose of the comparison exercise GOD, Foster and Hirata,

( I 991) and Foster, (1987); DRASTIC, Aller, et al., (1987), and WASP, Parsons and Jolly,

(1994a and 1994b) are used.

4.1 Indonesia

The selected case studies are all in the vicinity of Bandung. Bandung is situated on the

island of Java in an intermontane basin which is surrounded by active volcanic peaks.

The population density is 25000 inhabitants per square kilometre and the total population

is about 1.6 million, (Siebenhuner et al., 1993); for the Greater Bandung area it is said

to be three million, (Suhari and Siebenhuner, 1993).

The climate is tropical monsoon with a short season of low precipitation from June to

September and a monthly mean temperature of 28°C. For the purposes of this study site

climates are assumed to be the same as that recorded at the Bandung Meteorological

Station.

The city is industrialised with an economy based on textiles, armaments, machine

production, and leather processing. Figure 2.2 gives an indication of the waste

composition for Bandung and according to Bandung, (1 990) 20% of this is of industrial

origin.

Suhari and Siebenhuner, (1993) indicate that there are more than a dozen active and

abandoned landfills in the Bandung Basin whose site selection was based more on

proximity to the waste generating centre than geological considerations. Four of these

sites were visited in and around Bandung, (Klinck, 1993), to assess disposal practice and

three of these, Dago, Leuwigadja and Sukamiskin have been selected as case studies,

Figure 4.1.

wc/95/7 42 Issue 2

4.1.1 The Dago Landfill

The Dago Landfill, which operated from 1971 until 1989, is situated on the northern hill

slopes of Bandung, about six kilometres from the city centre, on the watershed between

the Cikapundung River to the northwest, and an unnamed river to the southeast.

The following account is based on the work of Aust and Wiriosudarmo, (1990) and

Rosadi and Sukrisno, (1993).

4.1.1.1 Site group

The waste site forms an almost symmetrical flat topped hill with runoff in all directions.

The site occupies an area of six hectares and has an estimated volume of 750,000 m' of

domestic waste. The waste thickness is between 25 and 30 metres.

The leachate has a chloride concentration of up to 4635 mgl-' combined with high

sodium, potassium, calcium, bicarbonate and boron; trace metals are also elevated.

Recharge into the landfill is estimated to be about 1100 mm a-'.

4.1.1.2 Geological and Hydrogeological Group

The landfill is on the site of a pre-existing quarry into volcaniclastic sand and gravel

which is locally called trass.

the groundwater table is between eight and 21 metres below the surface.

the trass has a hydraulic conductivity ranging from 1 ~ 1 0 . ~ to 3x10.' ms-'.

the hydraulic gradient is 2% in an easterly direction.

the aquifer is about 28 metres thick and rests on seven metres of tufaceous clay

the regional recharge is about 750 mma-'

an estimated porosity for the trass is 10%

WCl9517 43 Issue 2

4.1.1.3 Fate Group

Groundwater pollution has been proved in dug wells 120 metres from the site. The

nearest houses are less than 20 metres from the site and the impact of sewage on the

shallow wells is also evident.

4.1.1.4 Hazard Ranking of the Dago Landfill

A weighted DRASTIC score has been calculated as follows:

Depth to groundwater score (1 5 - 23 metres) = 15

Net Recharge (25+ cm) = 45

Aquifer Media (sand and gravel) = 18

Soil Media factor (absent) = 20

Topography factor (0 - 2) = 10

Impact of vadose zone media factor (sand and gravel) = 40

Evaluation of Hydraulic Conductivity factor = 5

Overall Score = 148

Using GOD, the groundwater occurrence is scored 1.0, the overall lithology is scored

0.75, and depth to groundwater is scored 0.8. Taking the geometric mean of these scores

gives an aquifer pollution vulnerability of 0.84 which is extremely vulneruhle.

Using the WASP scheme it is found that the barrier score is independent of the value of

hydraulic conductivity used since a barrier factor of 10 is calculated using a K value of

either 1 ~ 1 0 . ~ ms-' or a K value of 3x1& rn-d . Some uncertainty in the score is also

introduced by not knowing the percentage of groundwater used for supply. The site gives

a WASP index including uncertainty of between 7.0 and 7.3 which is unsuituhle.

Using the numerical simulator the landfill was set up as a square 245 m by 245 m (i.e. six

hectares in area) with a 2% hydraulic gradient from the western model boundary to the

east. Recharge (1 100 mma-' and 750 mmd ) and porosity (0.1) were taken as given

WCl95l7 44 Issue 2

above. The density of the trass was assumed to be 2000 kgm-3. The base of the aquifer

was set at 0 m as1 and the eastern fixed head boundary was set to 28 m to give the correct

aquifer thickness. Observation points were set at distances of 120 m along the flow path,

and at 45" and 90" to it. No borehole abstraction was considered.

The two main variables for which values are uncertain are the dispersion and the

hydraulic conductivity. The model was run for conductivities of 10-4, 10-s and 3 x 10.'

ms-'. At high conductivities, which will give the worst case travel times, the observation

points at 45" and 90" were only affected if the dispersion coefficient was set at a high

value (100 m). The travel times for C,,C, to reach 0.1 (health risk levels if initial

concentration = 4000 mgl-I) at 120 m using these values were:

Conductivity (ms-') 1 x 1O4

Along flow path

45" to flow path

1.2 x 106 s

(14.5 days)

7.4 x 10' s

90" to flow path 2.3 x 1 0 7 s

I 10-5 3 x 10.'

3.2 x 107 s 1.2 x 108 s

1.7 x 10's 5.2 x 10's

3.5 x 10* s 2.0 x 1 0 9 s

(1 1 years)

Decreasing the dispersion coefficient by an order of magnitude at a conductivity of 10"

ms-' increased the travel times by around a factor of four.

Because of the flow rates at the higher hydraulic conductivities, high C,/C, values are not

attained and so if low values are not a hazard, then the site will be safe. However, along

the flow path, 120 m is within the 100-day limit for pathogenic bacteria to survive and

so this a major risk. Away from the flow path, depending on the conductivity and the

dispersion coefficients, water can be taken from quite close to the landfill without a risk

of contamination. The effects of surface runoff on water quality are ignored.

When the time for infiltration through the unsaturated zone is taken into account, all the

observation points lie outside the 100-day limit. For an unsaturated thickness of 10 m

with sandy characteristics, the travel time for C,/C, to reach 0.5 is 8.5 x 10' s (98 days).

wc/95/7 45 Issue 2

For a thickness of 20 m, this increases to 1.05 x 107 s (121 days).

4.1.2 The Leuwigajah Landfill

Leuwigajah is located in a hilly environment some ten kilometres to the west of Bandung.

The site has been operating since 1986 as a valley infill.

4.1.2.1 Site Group

The waste site has a surface area of eight hectares and accepts both domestic and

industrial wastes, especially textile residues. Leachate analyses are given in Table 2.3b.

Recharge into the waste is of the order of 1000 mma-I, leachate flows from the waste as

a small stream into the valley below.

4.1.2.2 Geology and Hydrogeology Group

Aust and Wiriosudarmo, (1 990) describe the site as being an old andesite quarry on a

steep valley slope, Plate 4.1 The main lithology at the site is fractured andesite and

volcanic brecci a

0 the groundwater table is about 24 metres below ground level

e the hydraulic conductivity has been determined to lie between 10-7 and 10-Xms-'

there is uncertainty generally in the porosity of fractured andesites and a range of

0.1 to I % is estimated

0 the hydraulic gradient varies between 2% and 10% towards the west

although, as for the Dago site, the Bandung climate data has been used, the 0

aquifer recharge in the vicinity of the Leuwigajah site is probably quite low due

to runoff on the valley sides. It may not exceed 100 mmd' and could be as low

as 50 mma-'. However, considerable runoff into the waste is to be expected and

the estimate for leachate generation is correspondingly low.

WCl95l7 46 Issue 2

4.1.2.3 Fate Group

The distance to the nearest settlements is about 400 metres. Two serious problems at this

site are waste mass flow, Plates 2.1 and 4.1 and leachate runoff contaminating

downstream surface water, Plate 4.2. The first mass flow occurred in 1989 followed by

a major slide in December 1989 when about 20000 m3 of waste slid onto paddy fields at

the foot of the valley. For geotechnical reasons alone the site is unsuitable, and remedial

works have been proposed and carried out on a number of occasions, (Oeltzschner et al.,

1992 and van der Wall et al., 1992), but were destroyed by the most recent, major slide

which occurred in 1994.

4.1.2.4 Hazard Ranking of Leuwigajah Landfill

A weighted DRASTIC score has been calculated as follows:

Depth to groundwater score (23 - 30m) = 10

Net Recharge (25+ cm) = 45

Aquifer Media (weathered fractured basalt) = 27

Soil Media factor (absent) = 20

Topography factor ( 1 8+) = 1

Impact of vadose zone media factor (fractured basalt) = 45

Evaluation of Hydraulic Conductivity factor = 3

Overall Score = 151

Using GOD, the groundwater occurrence score is 1 .O, the overall lithology rating is 0.6F,

the F suffix indicating fracturing, and the depth to groundwater score is 0.6. The overall

score using the geometric mean is 0.71F which is on the limit between highly and

extremely vulnerable. By scoring the overall lithology as recent volcanic lavas at 0.8FF

gives an extreme rating.

Using WASP a fracture porosity has to be assumed and a suitable range of values is 0.1

to 1 .O%. The calculated index is between 6.5, with a marginal rating, and 7.7 with an

WCl9517 47 Issue 2

unsuituble rating, depending on the combination of hydraulic conductivity and porosity

used. This arises because the barrier factor score varies from a maximum of 10 using the

low value of porosity and higher hydraulic conductivity to 6.5 for the lower hydraulic

conductivity and higher value of porosity.

This site was numerically simulated using a 282 m by 282 m landfill. The recharge

values were taken as 1000 mma-' and 100 mma-' which gave a hydraulic gradient greater

than the 10% specified. Observation points were set up at distances of 400 m along the

flow path and, again, at 45" and 90" to it. Porosity was taken as 0.1 and 0.001 to simulate

fracture flow. For lower porosities the model failed to converge. The travel times for

C,/C, to reach 0.1 were:

Conductivity (ms-') 1 x 10-7 1 10-7

(porosity = 0.1) (porosity = 0.001)

6.0 x 10' s 9.2 x 106 s Along flow path

(1 9 years) (106 days)

45" to flow path 2.43 x 109 s 2.7 107

90" to flow path 6.65 x 109 s 4.44 1 0 7

(77 years)

1 x 10-*

1.9 1 0 9

7.67 x 1 0 9 s

2.09 x 10" s

(660 years)

The fastest travel time exceeds the pathogenic bacteria 100-day limit and so the site does

not give such a poor rating as the Dago site. However, this ignores the surface run-off

and waste mass flow which are the major impacts at this site.

The time for infiltration through an unsaturated zone of 24 m assuming a saturated

conductivity of 10-7 ms-' and characteristics similar to a silt, is calculated as 5.27 x 107

s (1.67 years) for a porosity of 10% and 5.2 x 106 s (60 days) for a porosity of 1 %.

4.1.3 The Sukamiskin Landfill Site

This site is located on a hill top interfluve, at Pasir Impun, seven kilometres to the

wc/95/7 48 Issue 2

northeast of Bandung. The site was in operation from 1987 to 1991 as an engineered

containment site. The base of the site was lined with a clay mineral liner which,

according to van der Wall et al., (1992), was emplaced to a thickness of 0.3 metres with

a hydraulic conductivity of 10-9 ms-'.

Subsequent settlement of the waste has required that a further raise be emplaced on the

flat top to reduce infiltration through the surface. Visual evidence of the waste settlement

is provided by the base plates of the gas vents now being above ground level, Plate 4.3,

and gas bubbling up through surface ponded water in collapse depressions.

Leachate from perched levels in the waste and site drainage is channelled to two

oxidation ponds. A concrete-lined peripheral drain is in place to stop leachate entering

nearby agricultural land and housing.

4.1.3.1 Site Group

The waste site has a surface area of 14 hectares and accepts domestic waste. Leachate

analyses are presented in Table 2.3b. The presence of springs from the waste indicates

that the waste is saturated and an infiltration rate of about 1000 mma-' is estimated.

4.1.3.2 Geology and Hydrogeology Group

The surface lithology is a silty clay derived from volcanic tuffs which is estimated to be

seven metres thick and overlies a volcanic breccia at least 27 metres thick. The strata

show a gentle dip to the south of about seven degrees. The breccia aquifer is locally

semi-confined by the clayey silt.

For the semi-confined breccia we have:

e the piezometric surface is 8 metres above ground level

x 10" ms-'

the hydraulic conductivity has been determined to lie between 9 . 5 ~ 1 0 . ~ and 1.5

WCf95f7 49 Issue 2

0 values of the hydraulic conductivity of the silt layer of 10" to 10-6 ms-' have been

determined by infiltration tests.

For the unconfined breccia we have:

0 a water table at 3.8 to 10 metres below ground level

the aquifer recharge in the vicinity of the site is probably quite low due to runoff

the hydraulic gradient is 2% with flow to the southeast and southwest.

0

on the valley sides and the presence of the confining layer and may not exceed

100 mma-l

a porosity of between 0.1 and I .O% for the breccia can be assumed 0

4.1.3.3 Fate Group

The nearest habitations and agricultural activity are on the slopes of the landfill, Plate 4.4.

Surface water is already contaminated from runoff from the site. Surface water around

the site is contaminated from runoff from the landfill and also by sewage.

4.1.3.4 Hazard Ranking of the Sukamiskin Landfill

A weighted DRASTIC score has been calculated as follows and is based on the presence

of a clay liner:

Depth to groundwater score ( 5 - 10 m) = 35

Net Recharge (1 00 mm a-' through liner) = 15

Aquifer Media (weathered igneous) = 9

Soil Media factor (clay) = 2

Topography factor (flat top) = 10

Impact of vadose zone media factor (weathered igneous) = 20

Evaluation of Hydraulic Conductivity factor = 3

Overall Score = 94

wci95i7 50 Issue 2

Using GOD it is only appropriate to consider the lining layer in the calculation.

Therefore groundwater occurrence scores 0.2 for the confined situation. The overall

lithology is scored 0.4 the lowest score possible and the thickness score is 1.0. The

geometric mean score is 0.43 which implies moderate aquifer pollution vulnerability.

If the upper, silty clay layer is considered in combination with the liner then the score is

reduced to 0.4 again with a moderate rating.

Using WASP, the liner layer in combination with the thickness of the unsaturated zone

in the breccia is used in the calculation. A value of 40% was used for the porosity of the

liner material, and is based on its residual moisture content. The WASP index is 6.4 and

the interpretation is marginal. The WASP index is insensitive to inclusion of the silty

layer in the barrier factor score for the confined situation because of the high value of

infiltration for this unit. If the liner is ignored in the calculation then the WASP score is

8.1 with an unsuitable ranking.

Where the aquifer is semi-confined and the piezometric surface lies within the confining

layer, the conceptual model on which the numerical simulator is based does not apply

because the confining layer has different hydraulic properties to the aquifer. The

simulator can be used for the unconfined part of the aquifer. The best approximation is

obtained by using the 1 -D vertical, unsaturated flow model to simulate penetration firstly

through the clay liner with one simulation and then through a maximum of ten metres of

unsaturated breccia using a second simulation. Flow away from the site in the aquifer can

then be approximated with the 2-D model by fixing the hydraulic gradient at 2% and

treating it as unconfined.

For the clay layer, the infiltration rate has to be set lower than the saturated conductivity

(10-9 ms-') and so it was arbitrarily set at 5 x 10" msl . Using the characteristics for a

clay given in Table 3.4, a porosity of 0.45, and a density of 2000 kgm-', the travel time

for C,/C, to reach 0.5 through a thickness of 0.3 m was calculated as 1.95 x 10* s (6.2

years). For an infiltration rate of 1000 mma-' through 3.8 m of unsaturated breccia, the

travel time is calculated as 7.2 x 105 s (8.3 days) assuming a saturated worst-case

conductivity of 1.5 x 10-6 ms-', a porosity of 0.01 (1%) and a density of 2400 kgm-'. This

increases to 9.7 x 106 s (1 12 days) for a 10 m thickness.

w Cf 9 51 7 51 Issue 2

Once in the volcanic breccia aquifer, the 2-D simulation using a landfill 374 m by 374

m and the same breccia properties as above gives a travel time for C,/C, to reach 0.1 at

1000 m along the flow path as 7.3 x 107 s (2.3 years). Recharge outside the landfill was

taken as 100 mma-'. The C,/C, value is considered realistic since the analyses given in

Table 2.3b are diluted.

4.2 Mexico

The object of selecting Mexico for case study sites was to expand the range of geological

settings to include limestones and examine the behaviour of the hazard ranking scheme

in a semi-arid environment. A total of four sites were visited, (Klinck, 1993), and

samples of leachate collected, Figure 4.2 and Table 2.3a. Three of these sites are

described below.

4.2.1 The Bordo Landfill, Mexico D.F.

The Bordo Landfill site is on the outskirts of Mexico City and close to the international

airport. It is the principal waste disposal site for Mexico City, accepting about 50% of

the waste generated, and has been in operation for about eight years. Waste placement

is on three raises each with a cover layer of silt. The total thickness of deposited waste

is eight metres. Gas control is by passive venting, and leachate is removed to high

density, polyethylene (HDPE) lined, oxidation ponds where it is allowed to evaporate.

4.2.1.1 Site Group

No details of the site size are available and 14 hectares is a rough estimate based on the

site visit. Leachate analyses are presented in Table 2.3a. Domestic and medical wastes

are accepted the latter being emplaced in an engineered, HDPE, lined cell. The site

climate is semi-arid and potential evaporation exceeds precipitation; however a figure of

130 mm a-' for infiltration from the waste is a likely estimate, Table 2.1. The presence

of solvents and industrial organic intermediates in the leachate, Table 2.3b, indicate that

WCl9517 52 Issue 2

industrial waste is accepted.

4.2.1.2 Geology and Hydrogeology Group

The Bordo landfill is situated in a swampy area which was formerly a lake bed.

Lacustrine silts and clays with intercalated sands underlie the site to a depth of 35 metres.

The following values of the hydraulic parameters are based on Rudolph et al., (1991).

e the hydraulic conductivity of these sediments is generally between 2 . 3 ~ 10-9 ms-'

the porosity is very high at 80 to 90%.

the hydraulic gradient is downward and may be taken as 1.0, the regional

the silts contain saline groundwater with a water table at between 0.3 and one

and 5.6x10-' ms-l ranging to 2 . 5 ~ 1 0 - ~ ms-', where it is believed that fractures exist.

0

e

hydraulic gradient is low.

e

metre below the surface.

The principal Mexico City aquifer consisting of alluvial and volcaniclastic sediments

underlies the lacustrine sequence and is between 100 to 400 metres thick.

4.2.1.3 Fate Group

The surface water and lacustrine groundwaters are unsuitable for domestic use and are

not believed to be exploited in the vicinity of the site.

4.2.1.4 Hazard Ranking of the Bordo Landfill

A weighted DRASTIC score has been calculated as follows and is based on the hazard

to the principal aquifer:

Depth to groundwater score = 5

Net Recharge (25+ cm) = 30

wc/95/7 53 Issue 2

Aquifer Media (thin bedded shale /sand) = 15

Soil Media factor (shrinking clay) = 14

Topography factor (0-2) = 10

Impact of vadose zone media factor (clayhilt) = 5

Evaluation of Hydraulic Conductivity factor = 3

Overall Score = 82

The above score assumes that the contamination of the upper lacustrine confining

sequence is not a problem, however it might be that local users abstract from this upper

sequence. In this case the DRASTIC, Depth to groundwater score is S O and the overall

score recalculated on this basis is now 127.

Using GOD the assessment is based on the upper lacustrine sequence only. For

groundwater occurrence the sequence may be considered to be confined and score 0.2.

The overall lithology scores OSF, and if the depth to groundwater is taken as the total

thickness of the lacustrine aquitard then the score is 0.39F and moderute aquifer

pollution vulnerability. The F subscript indicates fracturing in the lacustrine clays.

It is at the Mexican sites, because of their semi-arid climatic setting, that the WASP

scheme is best suited. If only the upper lacustrine layer is used in the calculation, the

WASP index is calculated at 5.3 and the site is classified as suitable. The threat factor

is 7.5 and the barrier factor is 4.7 based on the presence of fractured silts. However, the

resource factor, assuming no groundwater potential, is low 2.0 and effectively reduces

the overall rating. Because of uncertainty in usage of lacustrine sequence groundwater,

the resource factor could be higher and the site rating would become marginal.

The conceptual model of Figure 2.1 cannot be used to represent this site since the main

process is vertical infiltration through saturated media and the aquifer is at least partially

confined. The 2-D model could be used to represent the aquifer if more detail concerning

gradients and abstraction points were available.

The vertical saturated flow velocity is equivalent to the conductivity divided by the

porosity (0.8) giving a maximum of 7.0 x 10.' ms-' or 3.1 x 10-7 ms-] in fractured areas.

wc/95/7 54 Issue 2

For a 35 m thickness of sediments, the time for infiltration will therefore be 5 x 10' s

(15.8 years) or 1.13 x 10' s (3.6 years) ignoring any dispersion and retardation. Despite

the low infiltration quantities (130 mma-'), given the area1 size of the site this may be

significant depending on the flow rate in the aquifer.

4.2.2 The Merida Municipal Landfill, Yucatan

The Merida Municipal Landfill is situated on the northern outskirts of the city of Merida

and is operated by the municipality. The site has a long history of uncontrolled dumping

and waste burning with commercial, industrial and domestic waste being tipped, Plate

4.5. More recently an attempt has been made to place the waste into a compacted raise

with a soil cover. This methodology has reduced the risk of the waste catching fire and

also the nuisance from air borne pollution and smells.

4.2.2.1 Site Group

The waste disposal site is estimated to cover an area of about 175 hectares to an average

depth of 3 metres. Leachate analyses are given in Table 2.3a and because of the very high

evaporation rate from the waste are strongly mineralised. Data averaged over 43 years

show that during most months evaporation exceeds precipitation. Most of the infiltration

into the waste occurs during the rainy season. Over the year recharge is of the order of

200 to 300 mmd' with strong leachates being produced, Plate 4.6. Rapid recharge to the

underlying aquifer via fissures is assumed.

4.2.2.2 Geology and Hydrogeology Group

The Merida Landfill site is emplaced directly onto a karstic, Tertiary, marine limestone

of Miocene - Pliocene age, (DGGTN, 1984), with a thin (< lm, DGGTN, 1985) to non

existent soil cover. The following hydraulic data are used to assess the hazard ranking:

WCl9517 55 Issue 2

0 the hydraulic conductivity is very high due to the fissured nature of the rock.

Mendez Ramos, (1993) indicates values of the order of 10-3 to 104 ms-' , while

Sanchez y Pinto, (1 989) quotes values ranging from 3 . 7 ~ 10-5 for calcarenites to

3.38~10.' ms-' for recrystallised limestone.

0 both matrix and fissure porosity are high, the former ranging between 40 and

the water table is about 1.3 metres below ground level.

the hydraulic gradient is low at 7.5x10-' msl with flow in northwesterly and

most of the recharge occurs in August and September. Rapid bypass flow can be

50%, (Sanchez y Pinto, 1989). 0

0

westerly direction towards to the coast.

0

assumed during rainfall events and a high percentage of rainfall is estimated to

go to recharge, i.e. 150 mma-I.

4.2.2.3 Fate Group

A contaminant plume has been detected moving away towards the west. The nearest

population centre under threat is Dzitya about five kilometres from the site. Groundwater

is the sole source of water for agriculture, industry and domestic consumption.

4.2.2.4 Hazard Ranking of the Merida Landfill

A weighted DRASTIC score has been calculated as follows:

Depth to groundwater score = 50

Net Recharge (1 50 mm a-') = 30

Aquifer Media (karst) = 30

Soil Media factor (thin) = 20

Topography factor (flat) = 10

Impact of vadose zone media factor (karst) = 50

Evaluation of Hydraulic Conductivity factor = 30

Overall Score = 220

WCf95J7 56 Issue 2

Using GOD a rating of 1 .O is given for groundwater occurrence, 1 .OFF for the overall

lithology and 1.0 for depth to groundwater. The overall score is 1.0 with an extreme

aquifer pollution vu1 nerabil i ty.

The WASP index is calculated to be 9.7 with a highly unsuitable rating. Both the threat

factor and barrier factor scores are at a maximum of 10, the former because of the

hazardous nature of the waste. The resource factor is 9.3.

This is a large site represented by a 418 m by 418 m square in the simulation. The

hydraulic gradient was allowed to be determined by the recharge and was not fixed at 7.5

x 10.' since this caused a large recharge mound around the landfill. Recharge was set at

150 mma-' regionally and 200 mma' over the landfill Porosity was taken as 0.45 and

density was assumed to be 2500 kgm". The base of the aquifer was set at zero m as1 and

the eastern fixed head boundary was set to 1 m. Observation points were set at a distance

of 5 km along the flow path, and at 45" to it, and at a distance of 500 m at 90" to it. No

borehole abstraction was considered. Times for C,/C, to reach 0.1 were:

Conductivity (ms-') 1 x 10-3 1 10-4

Along flow path 1.39 1 0 9 ~ 4.4 1 0 9 s

(44 years)

For a typical fracture porosity of 1 %, the fastest travel time along the flow path becomes

3.1 x 107 s (1 year) indicating the importance of this parameter.

Generally, depending on the dispersion coefficients, the observation points away from

the main flow path did not display significant contamination. Although contamination

is obviously a problem close to the landfill, as is illustrated by the extreme ratings given

by the other schemes, the distance to the fate group means that this site gives a large

travel time. However, the limestones are known to be highly fissured and these fissures

will provide rapid flow paths through the aquifer.

Infiltration through the 1.3 m of limestone making up the unsaturated zone was calculated

to take 7.65 x 107 s (2.4 years) assuming a worst-case conductivity of 10.' ms-', a porosity

WCl9517 57 Issue 2

of 0.45, infiltration of 200 mm-h and characteristics similar to a sandstone. This

decreases to 1.7 x 106 s (20 days) if a porosity of 1 % is used. Again, it is believed that

heavy rainfall events on the waste will cause rapid leachate discharge via fractures to the

groundwater table.

4.2.3 Leon Guanajuato Municipal Landfill

The Leon Guanajuato Municipal Landfill has been in operation since 1986. The site is

located seven kilometres to the southwest of the city towards San Francisco del Rincon

and within a production well field known as the 'Bateria Poniente'. It was originally

envisaged that the site would have a working life of 14 years, however the available space

has been utilised within eight years and a new site is actively sought, (Ingenieria

ecologica S.A., 1994). The reasons for the shortened life of the landfill are due to a

doubling of the waste volume generated since 1990, and the hazard of leachate escaping

from the site across the main highway during periods of heavy precipitation, (Klinck,

1993).

4.2.3.1 Site Group

The landfill is sited in a small southeast-trending tributary valley of the Rio Turbio, Plate

4.7. The head of the valley has been widened to facilitate operation and to provide cover

material.

Three layers of waste have been deposited in benches to a thickness of about eight metres

and cover an area of about five hectares. Each layer is covered with silty sand and

passive gas venting allows the escape of landfill gas and prevents spontaneous

combustion. The landfill accepts domestic, medical, commercial and industrial waste,

including the waste from the major tannery industry in the city, Plate 2.6 and Figure 4.3.

Leachate seeps from the waste and is discharged, without treatment, off site into the

valley of the Rio Turbio. Leachate analyses are presented in Table 2.3a and attention is

drawn to the high levels of chromium present. The climate is semi-arid and a water

WCf95f7 58 Issue 2

balance estimate of leachate generation is 260 mma-' of which half comes from

precipitation, the remaining directly from the waste.

4.2.3.2 Geology and Hydrogeology Group

According to the Guanajuato geological map, (DGGTN, 1983), the area of the landfill

is underlain by Tertiary rhyolites and tuffs with a cover of volcaniclastic conglomerates

and sandy siltstones. Good exposures of the silty tuffaceous material are to be seen in the

sides of the valley at the landfill.

0 BGS et al., (1994) suggest that transmissivity of the volcaniclastic aquifer ranges

from 50 to 1700 m'd-'. A value of hydraulic conductivity for these types of rock

is probably in the range of 10.' to 10-7 m s-' 0 the porosity is estimated at 30%

the water table is between 150 and 160 metres below ground level and the aquifer 0

is unconfined although locally may be semi-confined because of discontinuous

clay layers in the sequence 0 according to the Guanajuato hydrogeological map, (DGGTN, 198 1) the

groundwater flow is regionally to the south-southwest. However, due to heavy

pumping south of Leon, a cone of depression has developed which causes flow

to be towards the southeast. An estimate of the hydraulic gradient, based on the

DERNA, (1988) study, is 0.015 at the edge of the cone of depression.

0 regional recharge is estimated at 130 mma ' using the climatic data for Leon

Guanajuato for 1992.

4.2.3.3 Fate Group

Although the landfill is well removed from the city a public supply well is located on the

down gradient perimeter of the site. In the past it has become bacteriologically

contaminated due to surface runoff from the landfill entering the headworks. A small

farm is located about 200 metres from the site and leachate occasionally overflows into

the gardens. During periods of heavy rain sheet flooding of the site occurs and silt and

diluted leachate (Table 2.3a) flow out of the site onto the main highway.

WCl9517 59 Issue 2

4.2.3.4 Hazard Ranking of the Leon Guanajuato Landfill

A weighted DRASTIC score has been calculated as follows:

Depth to groundwater score (>30 m) = 5

Net Recharge (260 mm a-') = 45

Aquifer Media (sand and gravel) = 18

Soil Media factor (silty loam) = 8

Topography factor (flat) = 10

Impact of vadose zone media factor (sand and gravel) = 30

Evaluation of Hydraulic Conductivity factor (using k= l x 10-4 ms-')= 6

Overall Score = 122

The GOD score has been calculated as follows: groundwater occurrence unconfined,

0.75; overall aquifer class, mixed sequence of recent lavas tuffs and gravels, 0.75, and

depth to groundwater 0.4. These scores have a geometric mean of 0.6 1 which indicates

a high aquifer pollution vulnerability.

The WASP scheme has been implemented for the range of hydraulic conductivity given

above, i.e. from 10-3 to ms-'. Using the higher value the barrier factor score is 10.0

and the WASP index is 9.2 which is a highly unsuitable rating. The reason for the high

WASP score is due to all the factor scores being high. Using the lower value of

hydraulic conductivity the barrier factor reduces to 3.5 and the WASP index is 7.0, with

an unsuitable rating.

The main problems associated with this site may well be the direct discharge of runoff

into the Rio Turbio and the proximity of a public supply well. Neither of these can be

adequately represented by the model, although both are clearly at risk. Infiltration and

contamination of the aquifer can, however, be considered. The site is represented by a

223 m by 223 m square and, because there is no clear fate group, observation points are

fixed at 1000 m from the landfill along the flow path and at 4.5" to it. Recharge was set

at 260 mmd' over the landfill and 130 mmd' elsewhere. This recharge generates a larger

WCl9517 60 Issue 2

hydraulic gradient than that specified and so the upper boundary is not fixed. Porosity

was set at 0.3 and density at 2000 kgm-3. Times for C,/C, to reach 0.1 were:

Conductivity (ms-') 1 x 10.' 1 x 1 0 . ~

Along flow path 1.02 1 0 9 s 8.8 x 109 s

(32 years)

2.85 x 109 s

(90 years)

45" to flow path 2.75 x 10" s

However, the Leon Landfill has a particularly high concentration of Cr (5.43 mgl-I).

Given a safe limit of 50 ,@I, this means that a C,/C, ratio of 0.01 is significant. This

reduces the shortest travel time to 6.5 x 108 s (20 years) assuming no retardation. This

highlights the importance of considering waste/leachate composition in assessing hazard.

Additionally, the depth to the water table and the large thickness of unsaturated material

to attenuate the leachate decrease the potential hazard. It proved difficult to model this

thickness using the 1 -D simulation assuming silt-like characteristics. The p factor had

to be reduced considerably from its fitted value and, since this relates saturated and

unsaturated conductivity, the calculated times may be too short. However, in this sense

they are conservative and calculated times were 7.38 x 107 s (2.3 years) for a conductivity

of 10-3 ms-' and 3.48 x 10' s ( I I years) for a conductivity of 10.' ms-I.

- 4.3 Case Study Discussion

From application of the ranking schemes to the case studies, the following observations

can be made:

DRASTIC is insensitive to high recharge, i.e. greater than 250mma-' and

considers a narrow range of hydraulic conductivity from 4.7 x 10.' to 9.4 x ~ O - ~

ms-'. This range does not cover the lithologies listed which are used to calculate

the impact of the vadose zone and aquifer media types indices. Interpretation of

the lithology rating requires expert judgement, e.g for basalt the rating ranges

from 2 to 10 for the impact of the vadose zone factor and this range produces a

wc/95/7 61 Issue 2

difference in the total score of 40 points. Notwithstanding the above the scheme

is thorough and more sensitive to parameter variation than either GOD or WASP.

GOD is the quickest scheme to use. The use of the geometric mean of the three

contributing factor scores removes the arithmetical artefact of using their product.

However, unlike DRASTIC, no weighting is given to the factors in arriving at an

overall score.

The computerised version of WASP was easy to use. Because the scheme is

intended for hazard assessment of prospective waste disposal sites, groundwater

usage is considered. However, we found it difficult to answer some of the

questions posed because of lack of site-specific information. To avoid

introducing too much uncertainty at this stage a consistent approach to assessing

groundwater usage is required by the user.

The numerical simulator requires considerable expertise to run and obtain results.

Values for the hydraulic parameters required by the model can normally only be

obtained by detailed site investigation studies. The alternative is to use ranges of

values as given in Section 2 and evaluate the site sensitivity to parameter changes

to cover the possible site variability. This approach would give a range of travel

times to the fate group. Using the 1 -D simulator vertical travel times from the

waste to the water table are obtained. These times give a quantified groundwater

pollution potential which can be compared with the results of the less

mathematically rigorous scoring schemes.

All of the schemes tested can be used either to evaluate prospective sites or sites

already in existence.

Figure 4.4 shows DRASTIC scores normalised to a maximum score of 235, the GOD

rating and the WASP index normalised to a maximum of 10. Table 4.1 summarises the

results of applying the ranking schemes and numerical simulations for each of the case

study sites.

WCl95l7 62 Issue 2

Table 4.1 Comparison of the site scores using DRASTIC, GOD and

WASP with the 1-D numerical simulator.

SITE I DRASTIC I GOD WASP I SIMULATOR

Dago

Leuwigajah

Sukamiskin

Bordo

Merida

Leon

The distribution of scores for DRASTIC, WASP and GOD is similar although for

Both WASP and GOD generally give a slightly higher score than DRASTIC.

The relative difference between the scores of WASP and GOD and WASP and

DRASTIC is more variable than the relative difference between GOD and

DRASTIC. This is probably due to the consideration of groundwater potential

and usage in WASP.

Sukamiskin only the liner was taken into account for the GOD score.

The same distribution as calculated from the schemes can be approximated by the

1 -D unsaturated, vertical travel time. However, the numerical simulator actually

quantifies the distribution showing a small difference in score can represent a

large difference in travel time. The times are much more sensitive to porosity and

hydraulic conductivity than the scoring schemes.

Use of the 2-D numerical simulator allows a consideration of the fate groups. For

instance a site with the highest hazard potential with respect to groundwater may

pose the lowest threat to the fate group, e.g. Merida, Table 4.1. In this sense, the

2-D part the simulator package is a first line risk assessment tool.

0.63 0.84 0.7 - 0.73 6.92

0.64 0.7 1 0.65 - 0.77 7.72

0.4 0.43 0.64 8.4

0.35 0.39 0.53 8.04

0.94 1 .o 0.97 6.23

0.52 0.6 1 0.7 1 7.87

WCl9517 63 Issue 2

5 SUGGESTED METHODOLOGY FOR THE HAZARD RANKING OF

WASTE DISPOSAL SITES.

The primary objective of developing a generic, numerical simulator, to assess waste

disposal sites, must be to improve present assessment procedures, whilst keeping costs

and evaluation time as low as possible. A detailed hydrogeological model cannot be built

for each site because of costly parameterisation requirements. However, a numerical

simulator allows quantification of the magnitude and timing of impacts and aids the

design of mitigation strategies and monitoring programmes. It also promotes

consideration of siting and abstraction with respect to the fate group; aids assessment of

uncertainty, and highlights gaps in data. This study has assessed whether it is possible

to use a generic simulator for any site utilising the same amount of information that might

be available through a desk-study for other, more empirical assessment schemes. Whilst

the results from the simulator cannot be considered to be accurate scientific prediction,

because of uncertainties in key parameters such as: dispersion coefficients, unsaturated

zone characteristics, and infiltration rate, the model was successfully applied in each case

study. Therefore, it does provide a useful and cheap quantification tool to be used

alongside other methods.

Based on the comparative evaluations carried out for the case study sites the following

procedure is suggested for hazard ranking a waste disposal site. This procedure can be

adopted both to evaluate the impact of a single site on the groundwater system or to

compare a number of sites for their suitability for waste disposal.

e Carry out a preliminary desk study of the prospective areas. This will involve

obtaining topographical, geological and hydrogeological maps; climate data; site

investigation reports, and any other information in the public domain.

In many cases the basic information will not be available in which case a site visit

is essential. During this visit a record of the geology, and geographic setting of

the area should be made.

e Apply GOD, using this information to rank the sites. GOD is the simplest

assessment scheme to implement and does not require very detailed data to make

WCl9517 64 Issue 2

an assessment. It gave comparative rankings to WASP and DRASTIC.

a The sites with the lowest GOD score should then be modelled using the WASP

scheme or the 1-D transport numerical simulator. The sites with the lowest

barrier factor scores in WASP and the longest travel times using the simulator

should then be considered for waste disposal. The effect of the waste disposal

operation on the fate group can be assessed using the 2-D contaminant transport

si mu1 ator.

a After preferred site selection, a more detailed site characterisation programme can

be performed to allow a site-specific hydrogeological model to be parameterised.

From the above case studies it is evident that sites located on silty material pose less of

a risk to a potential fate group because of the long contaminant transport times involved.

The Sukamiskin site in Indonesia, which had a relatively thin natural clay liner emplaced,

demonstrates the advantage of selecting sites on low permeability strata if total leachate

containment is sought. The trade off is that some form of leachate treatment is then

required, e.g. leachate oxidation lagoons. This type of site should be considered if

industrial wastes are to be disposed. If the site is only to accept municipal waste then a

site with a thick silty unsaturated zone is to be preferred. Such sites would have the

lowest barrier factors in the WASP scheme and the longest travel times using the 1-D

simulator. The above procedure is summarised as a flow chart in Figure 5.1.

Using the model and its guidelines, it should be possible to derive an approximate time

at which the landfill starts to pose a hazard at the distance of interest. Results and the

assessment of uncertainty from the numerical simulator in combination with a dose-

response analysis could form the basis for a risk assessment.

WCl95l7 65 Issue 2

6 REFERENCES

Aller, L., Bennett, T., Lehr, J., Petty, R. J. and Hackett, G. (1987). DRASTIC: A

standardized system for evaluating ground water pollution potential using hydrogeologic

settings. U. S. Environmental Protection Agency, Ada, Okla. NWWAEPA Series. Report

NO. EPA-600-12-87-035.

Aust, H. and Wiriosudarmo, S. (1990). Geological Reconnaissance Survey of Potential

Areas for Waste Disposal Sites in the Bandung Plain. Directorate of Environmental

Geology, Bandung, Indonesia. Project CTA 108, Environmental Geology for Landuse

and Regional Planning. Report No. 6.

Bandung, Municipality of. (1990). Urban solid waste in Bandung: towards the

development of an integrated resource recovery. Proceedings, The World Congress of

Local Authorites,jor Sustainable Future (New York, September 5 - 8, 1990),

BGS, CNA and UAC (1994). Effects of Wastewater Reuse on Urban Groundwater

Resources, Leon, Mexico. BGS Technical Report WD/94/25.

Blakey, N. C. (1982). Infiltration and absorption of water by domestic wastes in landfills

- research carried out by the Water Research Centre. Proceedings, Landfill Leachate

Symposium (Cockcroft Hall, Harwell),

Blakey, N. C. (1992). Model Prediction of Landfill Leachate Production. Landfilling of

Waste: Leachate, Ed. T. H. Christensen, R. Cossu and R. Stegmann. Elsevier Apllied

Science, London, 17-34

Blight, G. E., Hojem, D. J. and Ball, J. M. (1992). Production of Landfill Leachate in

Water-deficient areas. Landfilling of Waste: Leachate, Ed. T. H. Christensen, R. Cossu

and R. Stegmann. Elsevier Apllied Science, London,

Campbell, D. J. V. (1982).

Proceedings, Landfill Leachate Symposium (Cockcroft Hall, Harwell),

Absorptive capacity of refuse - Harwell Research.

wc/95/7 66 Issue 2

Canter, L. W., Knox, R. C. and Fairchild, D. M. (1987).

Protection. Lewis Publishers Inc., Chelsea, Michigan.

Ground Water Quality

Castillo Borges, E. R. and Sauri Riancho, M. R. (1993). Concentracion organica del 10s

lixiviados generados durante e1 composteo de desechos solidos municipales con

diferentes metodos de aeracion. Proceedings, IX Congreso Nacional de la Sociedad

Mexicana de Ingenieria Sanitaria y Ambiental: Calidad Ambiental para e1 Desarrollo

Sustentable (Mexico City), VI 5 - V20.

Christensen, T. H., Kjeldsen, P., Albrechtsen, H.-J., Heron, G., Nielson, P. H., Bjerg, P.

L. and Holm, P. E. (1994). Attenuation of Landfill Leachate Pollutants in Aquifers.

Critical Reviews in Environmental Science and Technologv, 24, 1 19-202.

Cointreau, S. (1982). Environmental Management of Urban Solid Wastes in Developing

Countries. The World Bank. Report No. 5.

Critchfield, H. J. ( 1983). General Climatology. Prentice-Hall Inc., Englewood Cliffs,

NEW Jersey.

Cruikshank, G., Aguirre M., D., Kraemer M., D. and Craviota G., E. (1980). Bacterial

contamination of the limerstone aquifer beneath Merida, Mexico. Aquifer Contamination

and Protection. Project 8.3 International Hydrogeological Programme., Ed. R. E. Jackson.

UNESCO, 34 1-345

Dasgupta, D., Sengupta, S. , Wong, K. V. and Nemerov, N. (1984). Two-dimensional

time-dependent simulation of contaminant transport from a landfill. Applied

Mathematical Modelling, 8, 203-2 10.

DERNA ( I 988). Estudio Geohidrologico en las inmediaciones de la cuidad de Leon Gto.

Report prepared for Sistema de Agua Potable y Alcantarilado del Municipio de Leon Gto.

(SAPAL). Report No.

DGGTN (1 98 1). Guanajuato F14-7, Carta Hidrologica de Aguas Subterraneas 1 :250 000.

1st ed. Direccion General de Geografia del Territorio Nacional.

WCl9517 67 Issue 2

DGGTN (1983). Guanajuato Hoja F14-7, Carta Geologica 1 :250 000. 1 st ed. Direccion

General de Geografia.

DGGTN (1984). Merida F16-10, Carta Geologica 1:250 000. 1st ed. Direccion General

de Geografia.

DGGTN (1985). Merida F16-10, Carta Hidrologica de Aguas Subterraneas 1 :250 000.

1 st ed. Direccion General de Geografia del Territorio Nacional.

DOE (1 978). Co-operative Programme of Research on Behaviour of Hazardous Wastes

in Landfill Sites. HMSO, London.

Fetter, C. W. (1 993). Contaminant Hydrogeology. Macmillan Publishing Company,

New York.

Foster, S. and Hirata, R. (1991). Groundwater Pollution Risk Assessment: a

methodology using available data. Pan American Centre for Sanitary Engineering and

Environmental Sciences, Lima, Peru.

Foster, S. S. D. (1987). Fundamental Concepts in Aquifer Vulnerability, Pollution Risk

and Protection Strategy. Proceedings, Vulnerability of Soil and Groundwater to

Pollutants, International Conference (Noordwijk aan %e, The Netherlands), Proceedings

and information No. 38, 69-86.

Foster, S. S. D. (1993). Characterisation of groundwater quality problems in Asia -

Pacific region. Report on visit to Thailand and the Philipines on Nov - Dec 1993. BGS

Hydrology Series Report. WDlOSl93156R.

Freeze, R. A. and Cherry, J. A. (1979). Groundwater. Prentice-Hall Inc., Englewood

Cliffs. N.J.

Gutjahr, A. L. (1992). Hydrology. Techniques for Determining Probabilities of

Geologic Events and Processes, Ed. R. L. Hunter and C. J. Mann. International

Association of Mathematical Geology, 75-98

WCl95l7 68 Issue 2

Harmsen, J. (1983). Identification of organic compounds in leachate from a waste tip.

Water Research, 17, 699-705.

Holmes, J. (1 993). Waste managemnt

Management, 8- 14.

practices in developing countries. Waste

Holmes, R. ( 1980). The water balance method of estimating leachate production from

landfill sites. Solid Wastes, 70, 20-33.

INGINIERIAECOLOGICA S.A. (1994). Estudio de Localizacion del Sitio para Relleno

Sanitario de la Ciudad de Leon, Gto. Bufete de Ingenieriaecologica S.A., Leon Gto.,

Mexico.

Kalinski, R. J., Kelly, W. E., Bogardi, I., Ehrman, R. L. and Yamamoto, P. D. (1994).

Correlation between DRASTIC vulnerabilities and incidents of VOC contamination.

Ground Water, 32, 3 1-34.

Kent, D. C., Pettyjohn, W. A. and Prickett, T. A. (1985). Analytical Methods for the

Prediction of Leachate Plume Migration. Ground Water Monitoring Review, 46-59.

Klinck, B. A. ( 1993). Hazard Ranking Scheme for Landfills: the search for Mexican case

study sites. Visit report Mexico 5th - 19th December 1993. British Geological Survey,

Technical Report WE/93/35.

Klinck, B. A. (1995). A Review of Hazard Ranking Schemes for Solid Waste Disposal.

British Geological Survey, Technical Report WC/94/76.

Krost, K. J., Pellizzari, E. D., Walbum, S. G. and A, H. S. ( 1 982). Collection and analysis

of hazardous organic emissions. Analytical Chemistry, 54, 8 10-8 17.

Lappala, E. G., Healy, R. W. and Weeks, E. P. (1987). Documentation of Computer

Program VS2D to Solve the Equations of Fluid Flow in Variably Saturated Porous

Media. U.S. Geological Survey. Water-Resources Investigations Report. 83-4099.

wc/95/7 69 Issue 2

Lerner, D. N., Issar, A. S. and Simmers, I. (1990). Groundwater Recharge: A Guide to

Understanding and Estimating Natural Recharge. Heisse, Hannover.

Loxham, M. (1988). Transport Modelling and Risk Analysis as the Basis for Assessing

Hazardous Waste Sites. Land Disposal of Hazardous Waste. Engineering and

Environmental Issues, Ed. J. R. Gronow, A. N. Schofield and R. K. Jain. Ellis Horwood

Ltd., Chichester, 1 1 3-124

Mather, J. (1 994). Waste Disposal to Landfill. Proceedings, Geoscience and the Urban

Environment in Developing Countries (Geological Society, London)

Mather, J. D. (1 989). The Attenuation of the Organic Component of Landfill Leachate

in the Unstaturated Zone: A Review. Quarterly Journal of Engineering - Geology, 22,24 1 -

246.

Mendez Ramos, R. (1 993). Generalidades sobre la propuesta de reglamentacion del

aquifer0 de Yucatan. Comision Nacional Del Agua, Gerencia Estatal en Yucatan,

Subgerencia de Administracion del Agua, Open File Report.

Oeltzschner, H., Wiriosudarmo, S. and Zainal Abidan, D. (1992). Comparitive Site

Rating of Potential Areas for Waste Disposal (Sanitary Landfilling) in the Bandung Area.

Directorate of Environmental Geology, Bandung, Indonesia. Project CTA 108,

Environmental Geology for Landuse and Regional Planning. Report No. 21.

Parsons, R. and Jolly, J. ( 1 994a). WASP Manual - Waste Aquifer separation Principle.

Water Research Council, Pretoria, South Africa. WRC Report No. TT67/94.

Parsons, R. and Jolly, J. ( 1994b). The development of a systematic method for evaluating

site suitability for waste disposal based on geohydrological criteria. Water Research

Commission, Pretoria, South Africa.WRC Report No 485/1/94.

Peniche A., I., Sauri R., M. and Koh H., C. J. (1993). Microbiologia del composteo de

desechos solidos municipales con diferentes metodos de aeracion. Proceedings, Calidad

Ambiental para e1 Desarrollo Sustentable (Mexico, D.F., Mexico), v21 - v26.

WCl9.517 70 Issue 2

Pickens, J. F. and Grisak, G. E. (1 98 1). Scale-dependant dispersion in a stratified granular

aquifer. Water Resources Research, 17, 1 19 1 - 121 I .

Rees, J . F, (1981). Landfill Leachate Attenuation in the Lower Chalk: the role of

Microbial Processes. Harwell Laboratory, Environmental & Medical Sciences Division,

Atomic Energy Research Establishment. Report No. AERE-R 1027 1.

Robinson, H. D. and Maris, P. J. (1979). Leachate from Domestic Waste: Generation,

Composition, and Treatment. A Review. Pollution B Division, Water Research Centre.

Report No. TR 108 HMS 16859.

Rosadi, D. and Sukrisno (1993). Groundwater Quality in the Surrounding of he Dago

Landfill Site. Directorate of Environmental Geology, Indonesian Geologica Survey.

Report of Environmental Geology for Landuse and Regional Planning Project. Report

No. 31.

Rosen, L. (1994). A Study of the DRASTIC Methodology with emphasis on Swedish

Conditions. Ground Water, 32, 278-285.

Rudolph, D. L., Cherry, J. A. and Farvolden, R. N. (1991). Groundwater Flow and Solute

Transport in Fractured Lacustrine Clay Near Mexico City. Water Resources Research,

27 ,2 187-220 1.

Sanchez y Pinto, I. A. (1 989). Estudio del comportamiento de la contaminacion del agua

subterranea generada por la disposicion de desechos solidos a cielo abierto. Universidad

Autonoma de Yucatan, Facultad de Ingenieria. Open File.

Sauri Riancho, M. R. and Castillo Borges, E. R. (1992). Caracterisicas de 10s lixiviados

producidos durante e1 composteo de 10s desechos solidos de Merida. Proceedings,

Sociedad Mexicana de Ingenieria Sanitaria y Amhiental: VIII Congreso Nacional -

Acciones para un ambiente limpio. (Cocoyoc, Mexico.),

Schultz, B. and Kjeldsen, P. (1986). Screening of Organic Matter in Leachates from

Sanitary Landfills Using Gas Chromatography Combined with Mass Spectrometry. Water

WCl9 51 7 71 Issue 2

Research, 20, 965-970.

Serrano, S. E. (1992). The Form of the Dispersion Equation Under Recharge and

Variavle Velocity, and Its Analytical Solution. Water Resources Research, 28, 1801 -

1808.

Shaw, E. S. (1988). Hydrology in Practice. Van Nostrand Reinhold (International),

London.

Siebenhuner, M., Silitonga, P. H., Sudradjat, A. and Toloczyki, M. (1993).

Environmental Geology for Landuse and Regional Planning - Greater Bandung Area,

Indonesia. Federal Institute for Geosciences and mineral Resources, Hannover.

Suhari, S. and Siebenhuner, M. (1993). Environmental geology for land use and regional

planning in the bandung Basin, West Java, Indonesia. Journal of Southeast Asian Earth

Sciences, 8, 557-566.

Valochi, A. J. and Herzog, B. L. (1988). Mathematical modelling of the Transport of

Pollutants from Hazardous Waste Landfills. Land Disposal of Hazardous Waste.

Engineering and Environmental Issues, Ed. J. R. Gronow, A. N. Schofield and R. K. Jain.

Ellis Horwood Ltd., Chichester, 153-1 64

van der Wall, R., Wiriosufarmo, S. and Zainal Abidin, D. (1992). Site Selection for

Domestic Waste Disposal Sites in the Hilly Surroundings of the Batujajar and Bandung

Plains. Directorate of Environmental Geology, Bandung, Indonesia. Environmental

Geology for Landuse and Regional Planning, Project CTA 108. Report No. 24.

Van Genuchten, M. T. (1 980). A closed form equation for prediciting the hydraulic

conductivity of unsaturated soils. Soil Science of America Journal, 44, 892-898.

Van Genuchten, M. T. (1981). Analytical solutions for chemical transport with

simultaneous adsorption, zero-order production, and first-order decay. Journal of

Hydrology, 49, 21 3-233.

WCl9517 72 Issue 2

Van Stempvoort, D., Ewert, L. and Wassenaar, L. (1992). Aquifer Vulnerability Index:

A GIS compatible method for groundwater vulnerability mapping. National Hydrology

Research Institute.

Walton, W. C. (1985). Practical Aspects of Groundwater Modeling. National Water

Well Association, Worthington, Ohio, USA.

Ward, R. C. and Robinson, M. (1990). Principles of Hydrology. McGraw-Hill Book

Company, London.

Yeh, G. T. and Ward, D. S. (1980). FEMWATER: A finite element model of water flow

through saturated - unsaturated media. Oak Ridge National Laboratory, USA. Report No.

ORNL- 5567.

Yeh, G. T. and Ward, D. S. (1981). FEMWASTE: A finite element model of

contaminant transport through saturated - unsaturated media. Oak Ridge National

Laboratory, USA. Report No. ORNL - 5601.

WCl9517 73 Issue 2

EXISTING DATA JUDGEMENT

PARA METER C LASSlFICATlO N

PARAM ETER WElG HTlNG

t

DRASTIC INDEX

Figure 1.1 The DRASTIC Methodology (based on Rosen, 1994)

GROUNDWATER OCCURRENCE

(INPUT)

OVERALL LITHOLOGY OF AQUIPERM OR AQUITARD

( i ) CONSOLIDATION ( i i ) CHARACTER

F degree of fissuring

A relative altenuatlon CaPaCl:Y (clay conrent)

I 1 1 I I I -I , w r d E r i E G L L c:'r MO 1:t i i ~ r E t l lL l i

(INPUT)

DEPTH TO GROUNDWATER TABLE (unconfined) OR

STRIKE (confined) X 0 4 0 5 O G 07 0 8 0 9 1 0

I 1 1

EX TR t hl E POLLUTION

I? I I 1 (OUTPUT)

Figure I .2 Calculation scheme for GOD (from Foster and Hirata, 1988)

10 8 6 4

W A S P I N D E X

2 0 2 4 6 8 1 0

RESOURCE FACTOR SCORE

h i g h l y unsuitable UnSulldblw marg ina l s u i l a b l o h l g h l y s u i l a b l e

I N T E R P R ETAT I 0 N

Figure I .3 The WASP index noinogram (from Parsons and Joliy. 1994b)

wci95i7

Figure 2.5: ’l’ypical characteristic c u r ~ e s relating moisture content and relative h j draulic coiiductivit! to pressure head 111 the unsaturated /one. rlho\ e the water table, the moisture coiileiil remains at near-saturated values uiitil the pressure head reache\ the huhhliiig pressure (h b) at M hich it starts to drop. The cun e becomes le$\ sharp tor tine-grained o r ~ell-\orted nicdia. ‘1 relatire h)draulic conductivitg of 1 0 correspondc to the satiiratcd liqdraulic conducti\ ity The conducti\ ity decreases as the prcwire head becomes more iiegatii e .\fter I im (ienuchteii (1980).

1 .o

0

Q- 0.5 U

0.0

Time of first appearance time (s)

Figure 3.1 : Typical breakthrough c u n e for a solutelcontaminant t r a elling through a porous medium. For the 3-D saturated zone model presented here, the source term concentration is assumed to bc the same as that in the landfill i.e. C,/C, in the infiltrating water reaching the mater table has reached 1.0. Howeker, i t can be seen that some contaminant ~ 1 1 1 reach the Rater table long before Cl/C,, = 1.0 and so the time taken for Cl/C,, t o reach 0.5 at the mater table is taken as an average for the inf litration time through the unsa turated Lone.

Figure 3.2: Grid of elements around a landfill (stippled area) generated bjr the numerical front-end program for the 2-0 saturated Lone fin1 te-element model represented by the parameter I alues giyen in Table 3.1. Note that the elements are smallest around the landfill n here concentration changes can be expected to be largest and so where the model needs to be most scnsiti\k

Gra\rel - temperate climate (Head Gradient set t o 1500)

Tinit! for C$Co to reuch 0.133 for chloride

No-flow boundarj

90') metres 0 @ > 128 h y s

@ > 128 clays

> 128 c€uys

> 128 days @ 700

> 128 days @ 25.6 davs /

3.6 duys - 3(X) duvs

- I -L hlain flow direction I

No-flow boundarj

Time through unsaturated zone of 1Om = 2 years

C//C, aper 100 h y s n-uvel tinle with K = 0

No-flow boundarj

Figure 3.3: Times for Cl/Co to reach 0.133: and C,/C, lalues after 100 days for a simulation of 2-D saturated flow away from a landfill in an unconfined gravel aquifer with a head gradient set to 1:500 in a temperate climate. Parameter values for this model are gil'en in Tables 3.1 and 3.3.

Fractured Basement Tropical Climate

Fixedhead orno-flow boundary

- 900 metres 0 @ > 160years > 160years -

- 700 8 55 years 0 76 years 20Years @ 16.9years

: J 0 @e @18.8years@ 47years 76.4yeard

- t r 411 days4.75 years - 300 -

- Main flow direction - - 100

200 400 600 800 1000 metres 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 .

Landfill Area = 20000m2

Fixedhead orno-flow boundary

- 900 metres 0 149 years 8 I42 years

-

- 700

- 0 4 years @ 31.1 years@ 53 years

. ........... E] {<{<if{ 7 Q 15. I years@ 38 years 65 years t

. if{<{<< -

4 years - 300 - - r Main flow direction - - 100

200 400 600 800 1000 metres ? ' 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 .

No-flow boundary Time through unsaturated zone of 10m = 166 days

f

River boundary - fixedhead

f

River boundary - fixedhead

Figure 3.4: Times for Cl/Co to reach 0.133 for a simulation of 2-D saturated flow away from a landfill lying on fractured basement in a tropical climate. Parameter values for the model are given in Table 3.3. A doubling in the area1 size of the landfill decreases the times with the times closest to the landfill showing slightly larger proportional changes due to the greater effect of the recharge mound. Travel time through the unsaturated zone is unaffected since this is modelled in one dimension.

0.4

0.3

0.3

0.1

.O1 .1 1 10 100

Pressure Head (-m)

Line sholvs fit to data for Silt Loam G.E.3 (1 an Genuchten, 1980) using parameter \ dues H, = 7.3, a = 1.28

1 0

1

. I

. 0 1

2 ,001

.000 1

.00CX) 1

.0 1 . I 1 10 100

Pressure Head (-m)

Lines shomSs fit t o data for Silt Loani G.E.3 (\-an Genuchten, 1980) using parameter value 0 = 8 (upper line) and 16.4 (lo\ver line)

Figure 3.5: Data points depicting charactenstic curles (cf Figure 2.5) for unsaturated silt loam (taken froin van Genuchten, 1980) fitted using the equations in Section 2.2.1 and the parameter lalues gi\ en in Table 3.2 and on the figures. Fitting \T. as perfanied using a non-linear least-squares technique.

Sand - tropical climate (Head Gradient set to 1:500)

- 900 metres 0 0 > 44.4 years > 44.4 years -

- 700 0 37. I years

Time for CiCo to reach 0.133 for chloride

Fixedhead orno-flow boundary

Transverse dispersion coefficient = 20m

145 days 1.44 years - 300 - - -L Main flow direction - - 100

600 800 1000 metres 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . I . 1 . 1 L

200 400

0 9 years

Transverse dispersion coefficient = 1 Om

No-flow boundary

900 metres

209.1 years 700 0

016 .3 years e 5 9 . 6 years

[ @ 05.8years 017years 37.1 years 2 1.42 years

1 3 0 0

L Main flow direction 1 rl .2:0, I , 4:O. ,6:0, I . 8:O. I , 1000 I . metres I

Fixed head or no-flow boundary

No-flow boundary

Time through unsaturated zone of 10m = 432 days

f

River boundary - fixedhead

River boundary - fixedhead

Figure 3.6: Times for Cl/Co to reach 0.133 for a simulation of 2-D saturated flow away from a landfill lying on an unconfined sand aquifer in a tropical climate. Parameter values for the model are given in Table 3.3. Halving the transverse dispersion coefficient slightly decreases the travel times along the main flow direction and significantly increases the times away from this orthogonal flow path. Time through the unsaturated zone is unaffected since this is modelled in one dimension.

Shale - arid climate

- 900 metres 0 0 > 9500 years > 9500 years -

Time for C$, to reach 0.133 for chloride

Longitudinal dispersion coefficient = IOOm

3390 0 0 years

2430 years

8.87 Years @ 751 years

@ 80 $837 years 0 2088 years 3390 years

3oo 53.9 years 2i1 years

-L Main flow direction Fixed head or no-flow boundary

200 400 600 800

No-flow boundary

Longitudinal dispersion coeflicient = 3m

No-flow boundary

900 metres 0 0 > 16000years > 16000 years

7795 years 700 0 0 801 7 years 3581 years @ 2573 years I

@ 2063 @ 3676years 5070 years years

332 862years 300 years

Fixed head or no-flow boundary

I Main flow direction I - 100

200 400 600 800 1000 metres 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 . 1 .

No-flow boundary

Time through unsaturated zone of 10m = 92 years

3

River boundary - fixedhead

3

River boundary - fixedhead

Figure 3.7: Times for CllCo to reach 0.133 for a simulation of 2-D saturated flow away from a landfill lying on shale in an arid climate. Parameter values for the model are given in Table 3.3. Decreasing the longitudinal dispersion coefficient increases travel times with the greatest increases being away from the main flow path and closest to the source. There is no simple correlation between the changes in times and the change in the parameter. Time through the unsaturated zone is based on a dispersion coefficient of lm.

Limestone

- 100 2(K) XK) mo 800 1000 metres

I . I . I . I . I . I . I . I . ~ . ~ . ~ .

?‘iine,for C i C , to reach 0.133for chloride

Temperate c1iniate;time.v = no set head gradient gradient = 1:1~x10 gradient = I:5Cx) gradient = 1: 100

No-flow boundaq

900 metres 25.4 years 37.4 yeurs 18.Y.veur.y 22.9 years 10.3years 11.7years

02 .31 y e a r s 0 2.44 years

7.7 years 5.48 year5 2.8

7cK) 0.6 yeurs 0

8.78 years 23.1 yecm 43.7 wars 3.96 yearc 0 1 I .6 vears 24.6 vearr ....=.. @ @1.76 vears 5.54 years 12.6 years 0.36 years 1 .I1 vears 2.53 years

2.15 years 0.95 vearr 156cla?’s I Xlanflon direchon I 31 days

1;ixed head or no-flon boundaq

- 900 metres 73.5 years 5.4 years 2.7 years

-

- 700 2 I8 rluys - 0

250 years 368 years 21.4 years 23.4 years 10.6 years 11.8 years

0 2 . 3 3 years 0 2.44 years

86.2 yeurs 227year.y 431 years 3.33 years@ 11.2 years 25.7 years 1.56 yeurs 5.3 years 12.9 years 130 clays 1 .I years 2.54 years

t 22.2 veurs 2S9 days 14‘ duvs 31.3 duvs

300

1 d hlan f l o w duectmn I Iqixed liead or no-flon bourn*

3

River boundary - fixedhead

>

Kivcr boundary

, ~ fisedhead

So-flow boundary Time through unsaturated zone of lOm = 5.9 years

Figure 3.8: ’ h i e s for C~!’CO to reach 0.133 for a siinulatiorl of Z-L) saturated flow away from a landfill 1jing on an unconfined limestone aquifer in temperate aid arid climates. ‘Times are given for four specified head gradients and other parameter \-dues for the m d e l are given in’rable 3.3. The change in infitration rates between the two climates is a1 order of ina,ghide. For the non-set gradient this is reflected by an order-of-magnitude decrease in times. For the specified gradients, the cliniate change makes little difference escept for the smallest 1: loo0 gradient. Travel times decrease in linear proportion to the increase in head ,oratlient.

Si1 tstone

3

Time for C1/Co to reach 0.133for chloride

Temperate Climate

No-flo537 bomdq

Fixedhead or ncsflow b<>m*

- cX)O nictres 0 @ 1068 vears 1370 years

246 years - 700 0 @ 345 years

-

88.7 y ears @ 75.7 years

@ @ 0 @ 84.6 yews 0 2 14 y a r s 364 vears rf : .*.* t 2

5.4 years 2 1.2 years - 300 - - L Main flow direction -

- 100 200 4CK) hoo 800 IC)O mctrcs

1 . 1 . 1 . 1 . 1 . 1 . ~ . ~ . ~ . ~ ' ~ '

Time through unsaturated zone of 1Om = 5.6 years

Fixedhead or no-llow bouldlrj

Arid Clima.te

- 900 mctrcs 0 @ 5830vears 7290 vears

131 5 ycars - 700 0 @ 1#8 vears

-

466 years 0 402 years r J - T Z @ @@ 04 .56 years @ 1182 years 2183

years

32.3 115 years - 390 years - - Main flow direction . - 100

200 403 600 800 1ooC) rnctres 1 . 1 . 1 , 1 . I . I . l . i . l . ~ . ~ L

River

- fixedhead b0UlndXY

River boundary - fixed head

Figure 3.9: Times for ClKo to reach 0.133 for a simulation of 2-D saturated flow away from a landfill lying on siltstone in temperate and arid climates. Parameter values for the model are given in Table 3.3. The decrease in travel times with increasing infiltration is not linear since the rates change by an order of magnitude bctwecn the two f iprcs whcrcas the tinies decrcase by proportionatcl) lcss ncarer to the landfill.

J elakong +

Figure 4.1 Location of Indonesian case study sites

P a c i f i c

+Study Sites I Mexico City 2 L e o n Guanajuato 3 Merida

Figure 4.2 Location of Mexican case study sites

E 0

CO .I c I

.I i m

9 E e e 2

1

cc

h E .I

.m : c I

P F P

1 .o

0.8

0.6

0.4

0.2

0.0 -I -I

Dago Leuwi Bordo Merida Leon

Sites

0 I,eon

4

3 -

2 -

1 -

hl enda * Dago Leuwigadja

0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

DRASTIC GOD WASP

DRASTIC score

Figure 4.4: Comparison of scores from the DRASTIC, GOD, and WASP assessment schemes for five of the six case studies descnbed in Section 4. Sukamiskin is excluded because the scores only consider the clay liner in this case. The scores have becn normalised against each assessment scheme's maximum so that the] can all be plotted on a 0 to 1 scale. The louer figure shonrs the generic model's 1-D lcrtical t ra \d time plotted against the normalised DRASTIC score s h m ing that the loww DRASTIC scores correspond well \frith the Ion cst calculated 1 -D tra\.el times.

I

M

I

I

4

V

I

e4

I

9

n

'I

F "I

y N ~ ~ ~ ~ ~ 0 0 0 0 l n 0 > 0 0 0 0 0 0 0 0 0 0 0 0

v v

0 0 0 0 0 0 0 0 0 0 0 0 r ? ~ ~ " y ? F F l - F m F

~ E v v v v v v v v v v v v 0 0 0 0 0 0 0 0 0 0 0 0 1 m m m m m m m m m m m m

~ F l - F N O l - 0 0 0 0 F 0 0 0 0 0 0 0 0 0 0 0 0 0

Q E 0 0 0 0 0 0 0 0 0 0 0 0 d v v v v v v v v v v v v

c u m b

I c u r n c u F

0 0 0 0 v v v v m m m m q o o o o 0 0 0 0

3 E O O O O v v v v c u c u c u c u r q q q q

E v v v v 8 0 0 0 0 1

F - 7 - F

r q q q q 5 0 0 0 0

E v v v v

# I # # T # * # #

# # # # f i E # # # # I # * # #

c u m c u c u '0000

T O O 0 0

'0000

Z E O O 0 O

Q E O O O O

~ E o o o o

V v v c u c u c u c u

v v v v c u c u c u c u i v v v v

0 0 0 v v v w o r n ocuo

0 0 0 v v v c u t o c u

0 0 0 v v v

6 , m c u

0 0 0

0 0 0

0 0 0

0cu0

v ) m a c u ( 0 c u

0 0 0

* a 0 - c o b

0 0 0

bco0

0 0 0

c u a c u

0 0 0 v v v b 7 - 7

0 0 0

6 , m c o m c u c u

0 9 0

? ? 0

? ? - :

0 0 0

Cu.97

0 t c 5 0

' 4 " "

' 4 b h CO-??

9""

c o c o b

- c u b

b m m

0 0

m b 6 ,

N O 0

F b c o

- 0 0 a9"T

F F

0 0 v v m m 0 0

0 0 v v c u c u 0 0 v v

0 0

0 0

F ?

0 9 0 0 v v

# O # Y

0 #

# # # # # # # # # # N N

0 0 v v

c u N

0 0 v v c u m

o a v v

$ ?

0 9

0 9

0 c F F

? C o c V L

m v :

o c V L

? C

F Y

? C o c v \ m y F C 0 C

? C F T

o c I

E 2 c v)

r, '5

m

Parameter entry for modelling the saturated zone

Pararn eter

Pre pwa ter/F E M WATE R

Left Model Boundary Right Model Boundary Bottom Model Boundary Top Model Boundary Average Element Size

Landfill Left Boundary Landfill Right Boundary Landfill Bottom Boundary Landfill Top Boundary

Borehole x-coordinate Borehole y-coordinate

Aquifer x-conductivity Aquifer y-conductivity Aquifer porosity

Aquifer Base Elevation Aquifer Top Elevation

General Recharge Rate Landfill Recharge Rate Borehole Extraction Rate

Left Boundary Head Right Boundary Head

Prepwaste/FEM WASTE

Distribution Coefficient Bulk Density of Rock Longitudinal Dispersivity Transverse Dispersivity

Units Value used in Figure 3.2

m m m m m

m m m m

m m

m l s m l s

fraction

m m

m l s m l s

m 31s

m m

0 1200

0 1000 2 5

100 200 450 550

-1 -1

1.00E-02 1.00E-02 0.45

0 30

1.60E-09 1.60E-08

0

27.4 25

m3/kg 0 kglm3 1700

m 100 m 2 0

Table 3.1 : Parameters required for Prepwater and Prepwaste to run the 2-D saturated zone simulation. Values given are for the simulation in Figure 3.3

Parameter entry for modelling the unsaturated zone

Parameter

PrepwaterIFEM WATER

Depth to Water Table Srid Size

Satu rated Conductivity Porosity

Infiltration Rate (Same as Landfill Recharge Rate)

Soil Moisture Parameter HZERO Soil Moisture Parameter ALPHA Soil Moisture Parameter BETA

Prepwaste/FEM WASTE

Distribution Coefficient Bulk Density of Rock Longitudinal Dispersivity

Units

m m

m/s fraction

m/s

m3/kg kg/m3

m

Value used in Figure 3.2

1 0 1

1.00E-02 0.33

1.60E-08

1 1 1

0 1700

1

Table 3.2: Parameters required for Prepwater and Prepwaste to run the 1-D unsaturated zone simulation. Values given are for the times in Figure 3.3

Rock-type Dry density Porosity Conductivity Maximum head Aquifer thickness kglm3 % mls m m

Gravel 1700 3 3 1.00E-02 27.4 3 0 Sand 1600 4 0 1.00E-04 27.4 3 0 Silt 1400 45 1.00E-06 80 85 Limestone 2500 10 8.00E-05 27.4 3 0 Sandstone 2300 2 5 3.00E-05 28.5 3 0 Weathered basement 2400 4 0 1.00E-05 35 40 Fractured basement 2400 5 1.00E-07 250 275 Shale 2500 5 1.00E-09 600 750 Fractured Clay 2000 35 1.00E-09 600 750

Non-fractured clay and intact basement are assumed to have a conductivity too low (1.00E-11) for modelling Homogeneous media - conductivity in x and y directions are equal Model area always 1200 1000m with an average element size of 25m Landfill always 100m by 100m in size with lower left corner at (x=lOOm, y=450m) Longitudinal dispersivity = 0.1 * mean travel distance (i.e. 100m) Transverse dispersivity = longitudinal dispersivity * 0.2 (i.e. 20m) River boundaly head always set to 25m; head gradient set to 1 in 500 where less than this Aquifer thickness set to keep water table below surface but close to base of landfill for recharge conditions Borehole abstraction not considered

These sets are run for three recharge types

Recharge type Precipitation % as recharge Recharge rate mmlyear % mls

Arid 100 5 1.60E-10 Temperate 1000 5 1.60E-09 Tropical 2000 5 3.20E-09

Landfill recharge rate is set at 50% i.e. one order of magnitude higher

Contaminant Kd Concentration Health

mgl l mgl l

Chloride 0 3000 400 Bacteria 0

m3lkg in landfill guideline Potentially

hazardous CIICo

0.1 33 any 100 day limit

100 days taken as maximum for bacterial survival irrespective of conditions such as temperature, nutrient supply etc (Canter et al., 1987)

Table 3.3: Parameter values for the 2-D saturated zone simulations presented in the figures

Parameter values for modelling different unsaturated zone media

Sand Silt

Porosity Ho a

Ip Infiltration (m/s) Saturated Conductivity (m/s)

Time (s) Saturated time (s)

0.25 0.4 2.75 f0.06 7.30 f 0.01 2.40 f 0.12 1.28 f 0.02 6.91 f 2.31 16.4 f 1.2

3.20E-08 3.20E-08 3.00E-05 1.00E-06

(B=3) (B=8) 2.40E+07 8.00E+07 8.30E+04 3.95E+06

Clay

0.45 188.7 f 10 0.62 f 0.02 37.4 f 7.6

1.60E-10 1.00E-09 (B=37.4) 2.60E+10 4.50E+09

IDepth to water table = 10m; dispersivity = l m

Table 3.4: Parameter values for the 1-D unsaturated zone modelling times given in the figures.

Site Visits

Figure 5.1 : Procedural flow diagram to illustrate the role that the generic model proposed here could play in the assessment of alternatives for a landfill waste disposal site.

Assemble data

Desk Study 1 . _ _ . - -

Select new sites

I I no

F:mpirical Assessment e.g. M>D

I I