Golder Associates (UK) Limited Attenborough House Browns...

Golder Associates (UK) Limited Attenborough House Browns Lane Business Park Stanton-on-the-Wolds Nottinghamshire NG12 5BL England

Tel: [44] (0)115 9371111 Fax: [44] (0)115 9371100 E-mail: [email protected] http://www.golder.com/uk

OFFICES IN UK, IRELAND, FINLAND, GERMANY, HUNGARY, ITALY, FRANCE, SPAIN, SWEDEN, CANADA, USA, PERU, CHILE, BRAZIL, AUSTRALIA, SOUTH AFRICA,

FINAL REPORT ON

LANDFILL SETTLEMENT: ESTIMATING TIME TO COMPLETION

Submitted to:

DEFRA 17 Smith Square

London SW1P 3JR

DISTRIBUTION: 1 copy (PDF) - DEFRA 1 copy - Golder Associates (UK) Ltd July 2008 06529217.502/A.2

NEW ZEALAND, INDONESIA, HONG KONG, THAILAND Company Registered in England No 1125149. At Attenborough House, Browns Lane Business Park, Stanton-on-the-Wolds, Nottinghamshire, NG12 5BL.

July 2008 - i - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

EXECTUTIVE SUMMARY

Waste settlement analysis and prediction are crucial to understanding and managing the lifecycle of a landfill. Settlement determines the pre and post-settlement contours of the completed landfill and the planned filling volumes, it influences the progress of hydraulic and biodegradation processes, affects the performance of the landfill engineering and is part of the designation of when landfill sites have stabilised to the point that all management can be removed.

Landfill settlement predictions are typically carried out using time dependant methods, based on mathematical functions matched to previously recorded waste settlements. Despite the refinement of existing methods of settlement prediction, they have serious shortcomings in handling the organic fraction and the many factors that control its decomposition, and they are unable to account for changing landfill conditions, such as waste types and breaks in the filling phase. They are, therefore, difficult to use in a predictive manner and require recalibration for changing waste streams. The Hydro-Biological-mechanical (HBM) model is a settlement prediction tool, which considers the actual processes occurring within the waste that cause settlement. The HBM model is unique in that it provides a coupling between the, hydraulic, biodegradation and mechanical components of waste behaviour.

The principal objective of this research project has been the further development of the HBM model and preparation of an accompanying protocol to collect the necessary data and to analyse, interpret and predict the magnitude and time to completion of long-term landfill settlements of landfills in relation to site and waste factors, and site operations.

The project has focused on the acquisition of landfill settlement measurements and of site-specific and waste-specific data which influence future settlements of the waste, and then on the application of these data in the hydraulic-biodegradation-mechanical (HBM) model developed by Napier University.

In order to validate the performance of the HBM model, data on landfill geometry, filling history and waste characteristics have been collated from operators in the UK, USA, Australia and Hong Kong. The types of information requested from operators for use in the model comprised data normally gathered either for their own operational purposes or for compliance with permit conditions. Acquisition of adequately complete data sets proved to be challenging, and the quality and format of acquired data sets varied significantly.

Comparisons were made between results computed by the model and measured landfill settlement data, and it has been shown that the model can be used to reproduce measured settlement traces for a number of landfills with markedly different waste and site characteristics. A lack of complete data sets does mean that a fully rigorous validation was currently beyond the scope of the available data sets.

July 2008 - ii - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

The model is complex in its formulation and despite an effective graphical interface, it remains a relatively complex tool to implement. As with all finite element software, particularly bespoke codes such as this, it cannot be treated as a “black box”, and a sound understanding of the theory is required in order to implement the code. In its present form, it is considered to be an extremely useful research tool. However, users with the requisite experience and theoretical understanding can currently apply the model with relatively little effort to real cases and obtain settlement predictions with a sounder technical basis than the presently used time-dependant methods. Sensitivity and parameter analyses can be readily incorporated in the analyses.

Conclusions are drawn on the benefits to stakeholders of systematic monitoring and operational data collection; frequency of data measurements, especially of settlement; settlement prediction based on actual settlement processes; and on the present lack of guidance on settlement prediction methods which causes a wide variety of procedures to be submitted with planning applications.

With the aim of advancing landfill settlement data collection and settlement prediction, a landfill settlement protocol has been produced including a hierarchy of settlement prediction techniques.

July 2008 - iii - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

ACKNOWLEDGEMENTS

Golder Associates (UK) Limited, Napier University and the University of Loughborough gratefully acknowledge with thanks the contributions of the project steering group comprising Peter Braithwaite (Environment Agency and Defra), Alan Rosevear (now retired, formerly of the Environment Agency) and Nick Blakey (Defra).

The availability of the data records was essential to this project. The provision of landfill settlement and other data from a range of sites is gratefully acknowledged from the following companies:

• BRE, UK; • Sita UK; • Swire Sita, Hong Kong; and • Viridor Waste Management, UK. The attendance of the following at the Landfill Settlement Workshop, held on 29th April 2008, is gratefully recognised:

Peter Braithwaite (DEFRA/EA), Rob Marshall (EA), Nicola Ingrey (EA), Richard Moss (EA), Kevin Clarke (Biffa), Stuart Hayward-Higham (SITA), Chris Myers (WRG), Chris Ratcliffe (WRG), Richard Terry (Veolia), Andy Mackintosh (Hertfordshire CC), Roy Romans (Bedfordshire CC), Roy Leavitt (Essex CC), Tim Grindell (Cory Environmental), Nick Walker (Veolia), Neil Dixon (Loughborough University), Asif Siddiqui (Southampton University), Ken Watts (BRE), John McDougall (Napier University), and Adrian Needham, Gary Fowmes, Russell Jones (all of Golder).

July 2008 - i - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

TABLE OF CONTENTS

SECTION PAGE 1.0 INTRODUCTION......................................................................................... 1

1.1 Project Team ...........................................................................................3 1.2 Aims and Objectives................................................................................3

2.0 BACKGROUND .......................................................................................... 4 2.1 Settlement implications to Operators, Planning Authorities and Regulators ...........................................................................................................4 2.2 Existing Approaches to Long-Term Landfill Settlement...........................5

2.2.1 Simple Time-Dependent Methods ...............................................5 2.2.2 The Filling Phase .........................................................................8 2.2.3 Time Frame for Landfill Settlement Analysis .............................10 2.2.4 Interpretation of Collected Settlement Data as Simple Time-Dependent Process ...............................................................................11

3.0 DATA ACQUISITION ................................................................................ 17 3.1 Data Requirements for Assessing Landfill Settlement ..........................17

3.1.1 Input Data Requirements for HBM Model ..................................17 3.2 Availability and Acquisition of Data........................................................19 3.3 Extent of Search for Data ......................................................................21 3.4 Data Sets Obtained for the Project........................................................21 3.5 Challenges in Data Acquisition for the HBM Model...............................26

4.0 HBM MODEL OVERVIEW........................................................................ 29 4.1 Conceptual Framework .........................................................................29 4.2 Implementation ......................................................................................30 4.3 Applications of the Model ......................................................................31

4.3.1 Settlement..................................................................................31 4.3.2 Predicting Gas Production .........................................................31

5.0 DATA REQUIREMENTS FOR THE MODEL ........................................... 32 5.1 Hydraulic Data Requirements................................................................32

5.1.1 Van Genuchten ..........................................................................32 5.1.2 Residual Moisture Content ........................................................32 5.1.3 Specific Storage.........................................................................32 5.1.4 Hydraulic Conductivity ...............................................................32 5.1.5 Anisotropy with Respect to Hydraulic Conductivity, Ratio Horizontal / Vertical ...............................................................................35

5.2 Biodegradation Model Parameters ........................................................35 5.2.1 Maximum Hydrolysis Rate .........................................................35 5.2.2 Product Inhibition .......................................................................36 5.2.3 Digestibility.................................................................................36 5.2.4 Diffusion Coefficient ...................................................................38 5.2.5 Initial Solid Degradable Fraction ................................................38 5.2.6 Initial VFA Concentration ...........................................................39

July 2008 - ii - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

5.2.7 Initial Methanogenic Biomass ....................................................39 5.3 Mechanical Model Parameters ..............................................................40

5.3.1 Elastic Stiffness and Elasto-Plastic Stiffness .............................40 5.3.2 Critical State Friction Constant ..................................................40 5.3.3 Poisson’s Ratio ..........................................................................40 5.3.4 Initial Yield Stress ......................................................................40 5.3.5 Creep Viscosity ..........................................................................40 5.3.6 Decomposition-induced Void Change Parameter......................40 5.3.7 Decomposition Hardening .........................................................42 5.3.8 Dry Unit Weight (as placed) .......................................................43 5.3.9 Particle Weight...........................................................................43

6.0 DATA ASSEMBLY .................................................................................... 45 6.1 Data Required for HBM Validation ........................................................45

6.1.1 Input Parameters .......................................................................45 6.1.2 Output Parameters ....................................................................46

6.2 Data Acquisition.....................................................................................47 6.2.1 Solid Degradable Fraction .........................................................47 6.2.2 Waste Unit Weight .....................................................................49 6.2.3 Settlement Data .........................................................................50

6.3 Verification of Data ................................................................................50 7.0 USE OF THE HBM MODEL IN SETTLEMENT ANALYSIS..................... 51

7.1 Graphical User Interface (GUI)..............................................................51 7.1.1 User Modelling Experience of the GUI ......................................52 7.1.2 Discussion on Graphical User Interface ....................................55

7.2 Application of the Model ........................................................................55 7.2.1 Simple Predictive Modelling.......................................................55 7.2.2 Curve Matching Techniques ......................................................55

7.3 Modelling Discussion.............................................................................55 7.3.1 Phased Filling ............................................................................55 7.3.2 Numerical Difficulties .................................................................57

8.0 MODEL VALIDATION............................................................................... 59 8.1 Generic Modelling..................................................................................59

8.1.1 Simplified Modelling ...................................................................59 8.1.2 Parametric Study .......................................................................61 8.1.3 Generic Model including Filling Sequence.................................64

8.2 Site-Specific Predictions vs. HBM Predictions ......................................65 8.2.1 UK, Viridor #1 ............................................................................66 8.2.2 HONG KONG #1 (NENT Landfill) ..............................................79 8.2.3 UK, SITA #2 ...............................................................................81 8.2.4 UK, BRE #4 ...............................................................................82 8.2.5 UK, SITA #3 ...............................................................................86

8.3 Summary of Site-Specific Validation .....................................................88 8.3.1 Decomposition Induced Void Change Parameter......................89

July 2008 - iii - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

8.3.2 Parametric Discussion ...............................................................89 8.3.3 Influence of including Filling Sequence .....................................89

8.4 Issues Arising in Validating....................................................................90 9.0 SETTLEMENT IMPLICATIONS TO STAKEHOLDERS........................... 91

9.1 Operators...............................................................................................91 9.2 Planning Authorities...............................................................................92 9.3 Environmental Regulation Agencies......................................................93

10.0 PROPOSED LANDFILL SETTLEMENT PROTOCOL ............................. 95 10.1 Introduction............................................................................................95 10.2 Data Collection and Monitoring Framework ..........................................95 10.3 Settlement Prediction Protocol ..............................................................97 10.4 Simplified Settlement Prediction Protocol..............................................98 10.5 HBM Model Settlement Prediction Protocol ..........................................98

10.5.1 Role of HBM Input Parameters ..................................................98 11.0 CONCLUSIONS...................................................................................... 100

11.1 Landfill Settlement Monitoring .............................................................100 11.2 Collection and Assembly of Data other than Settlements ...................100 11.3 Data Collection and Assembly.............................................................101 11.4 Changing Waste Streams....................................................................102 11.5 Conclusions on Modelling....................................................................102

12.0 FUTURE DEVELOPMENTS................................................................... 104 12.1 Data Collection and Recording............................................................104 12.2 HBM Model Parameter Research........................................................104

12.2.1 Decomposition Induced Void Change Parameter (Λ)..............104 12.2.2 Digestibility and Solid Degradable Fraction .............................104

12.3 HBM Model Development Requirements ............................................105 12.3.1 Definition of Filling/Non Filling Events .....................................105 12.3.2 Infiltration Events .....................................................................105 12.3.3 Applied Loads ..........................................................................105 12.3.4 Settlement of Deactivated Zones.............................................106 12.3.5 Determination of End of Filling.................................................106 12.3.6 Moisture Content – Biodegradation Relationship ....................106 12.3.7 Boundary Conditions ...............................................................107

12.4 Settlement Prediction Guidance ..........................................................107 12.5 Gas Generation Prediction ..................................................................107 12.6 Probabilistic analyses ..........................................................................108

13.0 REFERENCES........................................................................................ 109

July 2008 - iv - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

LIST OF APPENDICES Appendix 1 Notes on Landfill Settlement Workshop, Birmingham 29 April 2008

Appendix 2 HBM Model Graphical User Interface Example

LIST OF TABLES Table 1: Summary of Monitoring of Completed Landfills.............................................................12

Table 2 Minimum Monitoring Procedures for Landfills, from Landfill (England and Wales) Regulations 2002............................................................................................................................20

Table 3: Hydro-Biological-Mechanical Model Interdependence ...................................................30

Table 4: Mineralisation Parameters and Values .............................................................................37

Table 5: Biodegradation Model Parameter Values: Default Values ..............................................38

Table 6: Decomposition Induced Void Change Parameter - Reference Values and Associated Phase Composition Changes Where dVs < 0..................................................................................41

Table 7: Derivation of Solid Degradable Content for MSW..........................................................48

Table 8: Summary of Solid Degradable Fraction by Landfill ........................................................49

Table 9: Generic Model Input Parameters......................................................................................61

Table 10: UK, Viridor #1 Parametric Variations ..........................................................................71

Table 11: Maximum Solid Organic Fraction at 9800 Days............................................................78

Table 12: Key Filling Dates for NENT ..........................................................................................79

Table 13: UK, SITA #3 Filling Sequence ......................................................................................87

Table 14: Required Site Specific Information for Waste Settlement Prediction............................96

Table 15: Beneficial Additional Monitoring ..................................................................................96

July 2008 - v - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

LIST OF FIGURES Figure 1: Form of Simple Time-Dependent Long-Term Settlement Functions Over Short time (1000 days)...............................................................................................................................6

Figure 2: Form of Simple Time-Dependent Long-Term Settlement Functions Over Longer Time (10,000 Days)..........................................................................................................................7

Figure 3: Settlement Time Relationships (Bjarngard & Edgers, 1990)...........................................8

Figure 4: Settlement Components .................................................................................................11

Figure 5: Assembled Settlement Strain Data Presented as Strain Curves where Time Zero Corresponds to the Start of Filling and the Time of the First Appearance of any Particular Trace Denotes the Delay in Settlement Monitoring. ......................................................................14

Figure 6: cα Data Obtained from Assembled Settlement Strain with Tref = 30 Days and Offset from Time Zero by the Period of Delay Between start of Filling and Start of Settlement Monitoring......................................................................................................................................16

Figure 7: HBM Model: Function and Determinability of Input Data ...........................................18

Figure 8: Schematic Representation of the HBM Conceptual Framework (McDougall, 2007) ...29

Figure 9: Finite Element Mesh Illustrating Filling (McDougall, 2007). .......................................31

Figure 10: Variation of Saturated Hydraulic Conductivity with Dry Unit Weight, from Data Published by Beaven (2000)...........................................................................................................33

Figure 11: Waste Saturated Hydraulic Conductivity Data (DM3 from Beaven, 2000) Presented as a Function of Void to Inert Phase Ratio ....................................................................35

Figure 12: Influence of Digestibility .............................................................................................36

Figure 13: Loss of Solid Degradable Fraction (Shown in Red) with Time...................................39

Figure 14: Changes in Position of Yield Surface with Decomposition (McDougall, 2007). ........43

Figure 15: Overview of Varying UK Waste Composition after Langer (2005)............................47

Figure 16: Mean UK Waste Composition with Average Deviation, after Langer (2005).............48

Figure 17: Data Flow and Management between HBM Model and GUI. McDougall (2005)......52

Figure 18: Modelling Landfill Settlement a) without Phased Filling b) with Phased Filling .......56

Figure 19: Modelling Results as a Comparison to the Predicted Behaviour .................................60

July 2008 - vi - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

Figure 20: Parametric Study using the Eneric Model (Long-Term View) ....................................62

Figure 21: Parametric Study (Short-Term View) ..........................................................................63

Figure 22: Including Filling Sequence (Long-Term, Post-Filling)) ..............................................64

Figure 23: Generic Model with Filling Sequence (Short Term, Post Filling) ...............................65

Figure 24: Modelled vs. Measured Settlements for UK, Viridor #1 Landfill ...............................66

Figure 25: UK, Viridor #1 Modelling, Including Filling Sequence of 0 to 1500 Days.................67

Figure 26: UK, Viridor #1 Landfill: Solid Degradable Fraction (900 Filling Days) ....................68

Figure 27: UK, Viridor #1 Landfill: VFA (900 Filling Days).......................................................69

Figure 28: UK, Viridor #1 Landfill: Methanogenic Biomass (900 Filling Days) .........................69

Figure 29: UK, Viridor #1 Measured Settlement and Modelled Data, Including Filling of 900 to 1500 Days...................................................................................................................................70

Figure 30: UK, Viridor #1 Parametric Study, Hydraulic Conductivity.........................................72

Figure 31: UK, Viridor #1 Parametric Study, Solid Degradable Fraction ....................................73

Figure 32: UK, Viridor #1 Parametric Study, Dry Unit Weight ...................................................74

Figure 33: UK, Viridor #1 Parametric Study, Digestibility ..........................................................75

Figure 34: Influence of Infiltration on Settlement (Plot Includes 150 Days Filling) ....................76

Figure 35: VFA Plot for UK, Viridor #1_205: no Infiltration (First 1000 Days)..........................77

Figure 36: VFA Plot for UK, Viridor #1_206 30 mm Infiltration Events During Filling (First 1000 Days) .....................................................................................................................................77

Figure 37: UK, Viridor #1, with Infiltration, Post-Completion Settlement...................................78

Figure 38: Diagram Showing the Initial Unloaded FE Mesh (t=0), Filling Phases and Settlement Modelling Phases for NENT ........................................................................................80

Figure 39: Outline HBM Model Surface Elevation Predictions for NENT ..................................81

Figure 40: Comparison of HBM Model Output with NENT Data Presented as Surface Displacements Showing Influence of Λ (= -0.2 and -0.4). .............................................................81

Figure 41: Modelled and Measured Settlements at UK, SITA #2 Landfill...................................82

July 2008 - vii - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

Figure 42: UK, BRE #4 Landfill Modelling against HBM Model Predictions.............................83

Figure 43: Settlement Plots Including Filling................................................................................85

Figure 44: UK, SITA #3 Post Filling Settlement ..........................................................................86

Figure 45: Settlement Profiles for UK, SITA #3 with 3 Filling Events ........................................87

Figure 46: UK, SITA #3 Post-Filling Settlement Including Full Filling Sequence ......................88

Figure 47: Settlement Prediction Model Hierarchy.......................................................................97

Figure 48: HBM Input Protocol ....................................................................................................99

July 2008 - 1 - 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

1.0 INTRODUCTION

This report is the final report of the Defra Research Project “Landfill settlement: controlling the time to completion”. The project has the DEFRA reference number WR0301 (previously WRT381). The project has been undertaken by Golder Associates (UK) Limited (Golder) and Napier University, Edinburgh, and with review by Loughborough University.

Waste settlement analysis and prediction are crucial to understanding and managing the lifecycle of a landfill. Settlement determines the pre- and post-settlement contours of the completed landfill and the planned filling volumes, it influences the progress of hydraulic and biodegradation processes, affects the performance of the landfill engineering and is part of the designation of when landfill sites have stabilised to the point that all management can be removed. Despite the refinement of existing methods of settlement prediction, they have serious shortcomings in handling the organic fraction and the many factors that control its decomposition, and they are unable to account for changing landfill conditions, such as waste types and breaks in the filling phase. They are, therefore, difficult to use in a predictive manner and require recalibration for changing waste streams. An alternative, more fundamental approach to the estimation of landfill settlement has been developed based on combining individually proven models of hydraulic, biodegradation and mechanical (HBM) behaviour to give an integrated interpretation of landfill behaviour.

This project has focused on the acquisition of landfill settlement measurements and of site-specific and waste-specific data which influence future settlements of the waste, and then on the application of these data in the hydraulic-biodegradation-mechanical (HBM) model developed by Napier University. The HBM model and graphical user interface is freely available at http://www.sbe.napier.ac.uk/HBM/.

The principal objective of this research project is the further development of the HBM model and preparation of an accompanying protocol to collect the necessary data and to analyse, interpret and predict the magnitude and time to completion of long-term landfill settlements of landfills in relation to site and waste factors, and site operations. The intention within this project is to validate the HBM model as a research tool. Development of the HBM model to a commercially applicable settlement/time to completion predictive tool is beyond the scope of this project.

The study commenced with a search for field-scale settlement records and the landfill data that influence settlement from UK operators and other research-led field scale projects overseas. In the second phase, the settlement data were analysed using simple time dependent methods. This was done firstly to assess the nature and scale of the landfill settlement interpretation problem today and secondly to identify data sets suitable for further analysis using the HBM model. In the final phase, performance of the HBM model was assessed and evaluated. A Workshop was held on 29 April 2008 in Birmingham following issue of the draft final report, attended by representatives from landfill operators, planning authorities, the

http://www.sbe.napier.ac.uk/HBM/


Golder Associates

Environment Agency, Defra and the project team. Presentations on landfill settlement issues and on the project were each followed by discussion sessions. Notes on the Workshop are contained in Appendix 1.

The HBM model of waste degradation has been under development for several years to simulate the actual processes involved in the settlement of landfill wastes. This model has theoretical advantages over the currently-used simplified settlement prediction methods, in that the actual processes of waste settlement and decomposition are considered with coupled hydraulic, biodegradation and mechanical modelling. This model provides a more fundamental, and hence more comprehensive, prediction of short-term and long-term settlements and time to completion.

This report provides an assessment of the operational and quantitative performance of the HBM model against actual settlement data. In the context of this assessment, “operation” refers to the ease of use and stability of the model and graphical user interface (GUI), while “quantitative” performance refers to the resulting fit between model output and the actual settlement data.

The acquisition of actual landfill data under normal operating conditions, rather than data from laboratory or larger scale test facilities, has enabled full-scale testing and validation of the model to be undertaken for a variety of landfill cases. An early finding of the project was that acquisition of complete data sets of information from the records that landfill operators have normally collected was challenging. The process of collection, recording and storing settlement and other relevant data by operators is carried out in a variety of ways and for different purposes. This project has examined these procedures and provides recommendations for improvements to facilitate the use of these data in settlement calculations. The benefits to landfill operators of settlement prediction are important for operational and commercial reasons but are not necessarily widely accepted at present. These have been discussed in Section 9.0 with the aim of demonstrating the benefits of increasing the use of settlement analysis and prediction by landfill operators.

The time-dependent approach to landfill settlement analysis as currently used is compared with the more fundamental approach. It is not a simple comparison because this latter approach (HBM model) is part of a wider interpretation of landfill behaviour and offers more than just an analysis of post-closure settlement. Of particular relevance are the additional data and skills requirements of the more fundamental approach, although these enable the benefits of the fundamental approach to be realised. Time-dependent predictions will also benefit from the availability of the additional data.

A technical paper on the data collection and assembly was presented at the Sardinia waste management conference in 2007 (Needham et al., 2007) and an abstract for a second technical paper for presentation at Waste 2008 has been accepted (Needham et al., 2008).


Golder Associates

1.1 Project Team

The project team comprised:

Golder Associates (UK) Ltd Project management, data acquisition and assembly, HBM operation, interpretation and reporting.

School of the Built Environment, Napier University, UK

HBM model development and technical expertise on landfill settlement estimation methodology.

Department of Civil and Building Engineering, Loughborough University, UK.

Senior technical review and project guidance.

1.2 Aims and Objectives

The research project has the key objectives of:

• The evaluation and refinement of the HBM model using actual landfill data sets; and • The development of a landfill settlement protocol on the practical prediction of landfill

settlement. The project comprised three phases:

Phase 1 covered the assembly and evaluation of available data sets from landfill sites in the UK and overseas. Agreement was reached with five of the main UK landfill operators to provide data sets for the project. In addition, access was gained by the project team to a number of long term landfill data sets from several countries outside the UK. The data sets were assessed for completeness and suitable sets compiled for use in the HBM model.

Phase 2 focused on quantitative simulations of the assembled data sets. Model performance with differing waste composition, under different climatic conditions and other operational conditions have been assessed and the strengths and weaknesses of the HBM model evaluated.

Phase 3 concentrated on the distillation of the outcomes of Phases 1 and 2 into a landfill settlement protocol. The protocol provides guidance on the recording and compilation of landfill data relevant to settlement prediction, analysis of landfill settlement and establishes benchmark conditions based on the best available landfill data sets for use in the freely available HBM landfill settlement model.


Golder Associates

2.0 BACKGROUND

The Landfill Directive is changing the way in which waste is managed. Much more stringent criteria and controls are required for the active management and relinquishment of landfill permits. For current and future sites, the nature of the waste stream will change significantly. Municipal biodegradable waste diversion targets and waste pre-treatment will cause the types of waste that are landfilled to change and will lead to modifications in the way in which landfills are operated. The amount of degradable matter will reduce. Sustainable management of past, present and future landfills must ensure that at landfill completion, no significant contamination remains and that the site is mechanically stable.

The term completion here refers to a state in which a landfill site is no longer a threat to the environment and that all active management can cease. Mechanical stability, or settlement, of the waste pile is a key factor in the definition of completion. Whilst on the route to completion, waste settlement modifies hydraulic and gas permeabilities, and final load bearing capacity. Ideally, productive re-use of the site should be anticipated, in which case long-term mechanical stability should also be attained. The prediction of long-term mechanical stability in the form of settlement cessation is also required to determine closure (pre-settlement) and post-settlement surface profiles and hence available void space. (Waste Management Paper 26E, 1996).

The factors controlling biodegradation-related long-term landfill settlement are varied, many of which are not part of conventional geotechnical settlement models. Nevertheless, simple time-dependent methods based on soil mechanics have been used to analyse and determine long-term settlement. In this context, such methods are highly empirical.

2.1 Settlement implications to Operators, Planning Authorities and Regulators

The settlement of a landfill during filling and after completion of filling will have many implications to the main stakeholders involved, being the site operators and owners, planning authorities in administering planning policy and guidance, and the environmental regulators (principally the Environment Agency, Scottish Environment Protection Agency and the Environment and Heritage Service).

Planning permissions for landfill developments include the post-settlement levels or contours, and the agreed pre-settlement surface to be attained on completion of filling, with the intent that this will settle down to the desired post-settlement levels. Experience has frequently shown that the landfill surface settles below the planned post-settlement levels and forms irregular surfaces. The irregularity is caused by different waste types or filling rates in the different parts of the landfill. This has led to sometimes repeated re-filling to achieve the post-settlement levels, requiring new planning permissions, and causing disruption to restoration, capping and leachate and gas management systems.


Golder Associates

A sound understanding of settlement and reliable settlement predictions made during filling enable operators to maximise waste infill quantities, project future void space availability accurately, and provide necessary information for closure planning.

The substantial settlements that occur at landfills, and the differential settlements between wastes and the adjacent natural ground impose considerable challenges to the integrity of landfill engineering. Environmental regulators need to be satisfied that the containment systems, surface water drainage and landfill gas and leachate management systems operate as intended.

Arising from the findings of this project, the commercial, planning, technical and regulatory benefits to the main stakeholders of a fuller understanding of landfill settlement during and after filling are discussed in some detail in Section 9.0.

2.2 Existing Approaches to Long-Term Landfill Settlement

Historically, long-term landfill settlement has been regarded as a combination of mechanical creep, physico-chemical corrosion, biodegradation and ravelling (Sowers, 1973). For the purposes of analysis, these processes are lumped together and defined as a function of time. Several methods have been proposed for the interpretation of long-term landfill settlement.

2.2.1 Simple Time-Dependent Methods

2.2.1.1 Cα or Log-time

Early attempts to interpret long-term settlement employed the familiar log-time or ‘cα’ soil mechanics approach (Sowers, 1973):

tc Δ= logααε (2.1) where εα is long-term settlement strain, cα is the coefficient of secondary settlement, Δt is some time interval.

2.2.1.2 Power, Exponential, Hyperbolic, First Order kinetic

Other techniques include: power law (Edil et al., 1990); ‘Gibson and Lo’ exponential model (Edil et al., 1990); hyperbolic functions (Ling, 1998); and first order kinetic functions (FOK) (Park and Lee, 1997; Marques et al., 2003).

The main characteristic of these approaches is that they define settlement strain as a unique function of time. Over a relatively short period of time (2-3 years) their use may be expedient, with subtle differences in shape and form offering different degrees of fit to data.


Golder Associates

Figure 1 shows the characteristics of the different functional descriptions for input parameters that force the output forms to lie within a consistent and common range.

0.00

0.05

0.10

0.15

0.20

0.25

0 200 400 600 800 1000

Time [days]

Verti

cal s

train

Sowers

Power

Gibson & Lo

HyperbFOK

Figure 1: Form of Simple Time-Dependent Long-Term Settlement Functions Over Short Time (1000 days) However, over a longer time-scale, these techniques reveal other conceptual features. For example, from Figure 2 it can be seen that the first order kinetic and exponential forms (Gibson and Lo method) show an early termination of long-term settlement, whereas the power function increases monotonically in log-time. The former interpretation may be more appropriate but is more demanding of input parameters. In general, the simplicity of the Sowers or cα method has encouraged its continued use. Moreover some agreement on the range of cα values, albeit wide, can be found (Sharma and De, 2007).


Golder Associates

0.00

0.05

0.10

0.15

0.20

0.25

1 10 100 1000 10000

Time [days]Ve

rtica

l stra

in

SowersPower

Gibson & LoHyperb

FOK

Figure 2: Form of Simple Time-Dependent Long-Term Settlement Functions Over Longer Time (10,000 days) Despite its popularity, the cα method must be formulated to avoid the fact that it is not defined at time Δt = 0. This is usually achieved by an expression of the form:

⎟⎟⎠

⎞⎜⎜⎝

⎛ +Δ=

ref

ref

ttt

c logααε (2.2)

where tref is a reference time interval which serves the dual purpose of (a) ensuring the logarithmic term is defined at the start of monitoring, and (b) increasing the flexibility of the function to fit a wide range of data sets. In many cases tref is between 30-120 days – based on the observation that this is the time over which primary compression occurs (Sowers, 1973; Yen and Scanlon, 1975; Sharma and De, 2007).

2.2.1.3 Other Modifications

It has been observed (Bjarngard and Edgers, 1990) that long-term landfill settlement does not always follow a simple monotonic reduction in strain rate, see Figure 3. Jessberger et al. (1995) and others describe an acceleration in strain rate at about 365 days, which they attribute to the onset of biodegradation. Williams (1991) proposed an expression of the following form,

( )( )1−Β= − ktteβαε (2.3)

where B is a strain rate parameter, β refers to the microbiological activity and tk is the time at which the acceleration occurs. The approach seeks to accommodate this acceleration but


Golder Associates

essentially the approach calls for the a priori definition of the time of acceleration and appears to require back calculation of the other rate parameters from observed data.

Figure 3: Settlement Time Relationships (Bjarngard & Edgers, 1990) 2.2.1.4 Comment on Simple Time-Dependent Methods

Clearly the time-dependent methods are an exercise in curve fitting; they are highly empirical and can only be applied observationally or to near identical waste and landfill circumstances. Moreover, the implication of the reference time is that long-term settlement processes (such as decomposition) commence 30-120 days prior to the assumed start of long-term settlement. This is an awkward assumption for the interpretation of settlement in a real landfill site; it ignores the impact of the filling phase – a much longer period of time during which significant changes in the hydraulic, biochemical and mechanical properties of waste take place.

2.2.2 The Filling Phase

Decomposition will commence as soon as a layer of waste is placed; in fact, layers of waste near the base of a site may well turn methanogenic long before filling above has ceased. A waste pile of uniform hydro-bio-mechanical properties is a considerable simplification. There is, therefore, no one single reference time with which to analyse long-term post closure settlement. Indeed, it is this shortcoming that triggered the development of the ISP settlement model (Olivier et al. (2003a) and other workers to discretise the waste depth (Marques et al., 2003).


Golder Associates

2.2.2.1 ISP Model (Olivier and Gourc)

The ISP model (Olivier et al., 2003a) identifies nominal layers of waste that are disclosed in time, each of which has its own reference and current time. The controlling function is of the following form:

⎟⎟⎠

⎞⎜⎜⎝

⎛=Δ

i

cihihτ

τα log (2.4)

where Δhi is the long-term settlement of waste layer i, hi is the post-compression height of layer i, τi is the duration of filling of layer i and τ is the time since the beginning of filling. Surface settlement is the summation of individual layer settlements, both load and decomposition related.

Thus the ISP model tackles a very significant factor controlling long-term landfill settlement. However, the approach demand detailed waste placement records, which on the basis of our experience currently appear not to be gathered at most UK landfills in a consistent form or to an adequate extent.

2.2.2.2 Thomas and Cooke Settlement Analyses

Thomas and Cooke (2007) adopted a settlement analysis technique considering waste input time and the nature of the waste. This model is based on consideration of the biological degradation processes occurring within the waste mass, and is based on the work by Young (1992). Thomas and Cooke (2007) identify that overburden pressure may not cause near-complete collapse of the micro voids formed by biodegradation, but offset this against the interaction on ravelling which is “likely to compensate to some degree for the full potential of settlement due to waste degradation not being achieved”, and suggest the calibration of models against measured data to quantify this response. The relationship between biodegradation, settlement, void change is addressed directly in the HBM model through the decomposition induced void change parameter, Λ , which is discussed in Section 5.3.6. Whilst Thomas and Cooke (2007) provide a discussion of the need for model requirements and Cooke et al. (2007) provide details on the comparison to measured data, insufficient information on the model principles and calculations are provided to allow a critical evaluation of model behaviour to be made.

2.2.2.3 Interpreting the Filling Phase - Establishing a Time Frame

The lack of a systematic record of waste placement, including surface levels, has turned out to be one of the major difficulties of this project. The Landfill Directive, and subsequent UK legislation, require data collection, and there are requirements under waste management licences and PPC permits for data collection; however, without specific data collection, compilation and presentation requirements or rationale for collecting such data, landfill


Golder Associates

operators are unlikely to make the necessary changes to their data collection and recording procedures to meet this objective, unless the benefits of more reliable settlement prediction become more widely accepted or are required by changes in regulatory requirements.

2.2.3 Time Frame for Landfill Settlement Analysis

In view of the general lack of systematic waste placement records, an early step in the project was therefore to establish a conceptual landfill settlement framework for the purposes of this study. Figure 4 identifies the main settlement processes and their incidence during the life of a landfill. For the purpose of representing landfill void volumes and as a benchmark for numerical analyses, there is an uncompressed, un-degraded waste volume, shown in the figure as a column. The waste in this state is considered with its as-placed (as-compacted) density. The components of waste settlement are described below.

Primary compression: during filling, the compressive effect of the weight of overburden soon exceeds that of compaction and an in situ waste density profile evolves. This is primary compression. Secondary settlement phenomena: conventionally treated as post closure phenomena, in fact these processes originate in the filling phase. Inappropriate interpretation of the origin of these processes seriously undermines simple landfill settlement analyses. Creep: regarded as significant during the filling phase but decreasing into the long-term. Biodegradation-related settlement: may also be significant during the filling phase but equally if not more significant into the long-term. In the operational phase, with filling and compaction occurring, self weight compression will dominate settlement. Some creep and biodegradation-related effects occur during this phase with the latter predominantly in the lower and older waste layers. The operational phase can vary significantly in length from one to two years to in excess of 10 years. On completion of filling, and disregarding capping, further settlement comprises physico-chemical and mechanical creep, and decomposition.

All settlement traces are simply drawn. During the filling phase, they are shown as contributions to settlement directly related to the amount of accumulating waste. During the post-closure phase, they are shown as monotonically decreasing (creep) and accelerating after closure to then decrease with time (biodegradation). In fact the pattern of displacement may be much more complicated than this with biodegradation-related processes being dependent on the evolution of a range of landfill conditions and not just time.


Golder Associates

Figure 4: Settlement Components Additional events such as placement of stockpiled materials on to the waste mass will result in further compression and secondary effects of such additional loading will include increased creep rate and potential reductions in hydraulic conductivity.

2.2.4 Interpretation of Collected Settlement Data as Simple Time-Dependent Process

The first task after collecting and assembling the settlement data, as described in Section 3.0, was to evaluate their quality and usefulness for the purpose of HBM modelling. The evaluation process involved the following:

1. Comparison of all settlement strain data. All settlement traces were plotted on a single diagram to ascertain the strain and time ranges and to identify completeness and gaps in individual traces. It was planned that the subsequent simulations would be based on the findings of this preliminary assessment; and

2. Comparison of all corresponding coefficients of secondary (biodegradation) settlement,

cα. This assessment exercise was performed in order to assess data quality when presented as cα values.

2.2.4.1 Settlement Strain Data

Settlement profiles were calculated using the surface levels and at given times. Strains were based on reported initial fill height or deduced from the first reported surface level and the base liner level. Ideally, some indication of the time that the waste had been in place was


Golder Associates

required but in most cases, the start and termination of filling of a cell was not explicitly recorded. This information may be obtained by personal communication but where sites have changed hands or ceased operation many years ago, this information may have been lost. There was rarely enough information to allow an infilling sequence to be well defined.

The start and end of filling enable the following to be derived:

1. An estimation of the age of the waste; and 2. A measure of the current time upon which the logarithmic function calculation is to be

made, i.e. tref and t. Table 1 shows the amount and type of information from which the time frame was established. As can be seen, only the start of filling and the start of monitoring are available from which a delay in monitoring is calculated. The information contained within Table 1 is used for positioning measured data within the appropriate time frame within time settlement plots, as shown in Figure 5. The filling times are representative of landfill phases and are used in calculating the time elapsed from the waste placement to the start on monitoring.

Table 1: Summary of Monitoring of Completed Landfills

Waste depth [m]

Start of

filling Start of

monitoring End of

monitoring

Delay in monitoring

[days]

Duration of

monitoring [days]

Settlement at start of

monitoring (-)

NENT :M1 35 Jul-95 Aug-02 Oct-06 2588 1522 NENT :M20 31 Mar-97 May-01 Nov-04 1522 1280 UK, SITA #1 27 Feb-03 Nov-04 Mar-07 639 850 0 UK, SITA #3 40 Mar-02 Jan-03 Sep-07 306 1704 0 UK, SITA #4 14 Jan-01 Mar-02 Jan-07 424 1767 0 UK, BRE #3 7 32 Jan-94 Aug-99 Jan-06 2038 2345 0.07100 UK, BRE #3 8 42 Aug-92 Aug-99 Jan-06 2556 2345 0.0830 UK, BRE #3 9 57 Aug-90 Aug-99 Jan-06 3287 2345 0.0900 UK, BRE #3 10 58 Aug-90 Aug-99 Jan-06 3287 2345 0.0600 UK, BRE #4 1 7 Jul-75 Jun-92 Aug-94 6180 791 0.1050 UK, BRE #4 4 24 Jul-75 Aug-89 Oct-91 5158 778 0.1700 UK, BRE #5 21 30 Jul-79 Aug-01 Aug-05 8067 1461 0.1170 UK, BRE #5 23 33 Jul-77 Aug-01 Apr-06 8797 1704 0.1306 UK, BRE #5 25 27 Jul-75 Aug-01 Apr-06 9528 1704 0.1145 UK, SITA #2 13 Jan-99 Mar-03 Oct-06 1520 1310 UK, SITA #1 NB1-5 30 Jan-68 Jul-87 Nov-94 7121 2680 0.1500 UK, SITA #1 NB1-4 30 Jan-68 Jul-87 Nov-94 7121 2680 0.1500 UK, SITA #1 PS1-x 10 Jan-68 Jan-88 Apr-94 7305 2282 0.1500 Sandtown AB-x 18 Oct-80 Oct-89 Aug-01 3287 4322 0.0630 UK, Viridor #1 J1 32 May-06 Jun-06 Jan-07 31 214 Lyndhurst 14 Dec-93 Feb-96 Jul-99 792 1246 NENT: M37 29 May-99 Jul-01 Oct-06 792 1918 In Figure 5, the settlement profiles are shown on a natural time scale offset from zero time by the delay in monitoring. This adjustment means the settlement traces are positioned according to the age of the waste (as depicted by the time at which filling began (= time


Golder Associates

zero)) and the start of the monitoring programme. The leftmost traces represent UK, Viridor #1, where settlement monitoring began only 31 days after filling. It is unusual for monitoring to begin so soon after completion of filling, however, closer inspection of these data shows that the end of filling for the site may be significantly later than the end of filling for a specific location. From the other data sets, it is shown that that settlement monitoring begins about two years or more after filling commenced.

It is interesting to observe the small strains that occurred in the very late (in terms of the landfill life cycle) data obtained from UK, BRE #5 – some 8000 - 9000 days after filling commenced. Note also that the Sandtown data are based on very few data points; the inclination of that trace around 7000 days may be flatter than is depicted here.


Golder Associates

0.00

00

0.05

00

0.10

00

0.15

00

0.20

00

0.25

00

0.30

00

020

0040

0060

0080

0010

000

Tim

e [d

ays]

Strain (-)

1200

0

UK

Viri

dor 1

.2

UK

BR

E 3

.4

UK

Sita

1.2

NE

NT

1.3

UK

Viri

dor 1

.1

Lynd

hurs

t 1.3

UK

BR

E 3

.3

UK

Sita

2

San

dtow

n 1.

2

Lynd

hurs

t 1.2

UK

BR

E 3

.2

UK

BR

E 5

.3

San

dtow

n 1.

1

Lynd

hurs

t 1.1

UK

BR

E 3

.1

UK

BR

E 5

.2

San

dtow

n 1.

5

UK

BR

E 5

.1

UK

Sita

1.5

UK

Sita

3

San

dtow

n 1.

4

UK

BR

E 4

.1

UK

Sita

1.4

NE

NT

1.2

San

dtow

n 1.

3

UK

BR

E 4

.2

UK

Sita

1.3

NE

NT

1.1

Figure 5: Assembled Settlement Strain Data Presented as Strain Curves where Time Zero Corresponds to the Start of Filling and the Time of the First Appearance of any Particular Trace Denotes the Delay in Settlement Monitoring.


Golder Associates

2.2.4.2 Cα data

Figure 6 shows values of cα calculated from the strain data shown in Figure 5 using:

⎟⎟⎠

⎞⎜⎜⎝

⎛⎟⎟⎠

⎞⎜⎜⎝

⎛ +Δ=

ref

ref

ttt

c

log

εα (2.5)

where Δt is time of settlement monitoring on completion of filling since the start of filling and tref = 30 days. Note that cα in this form is a secant value.

It is evident that cα for the recent settlement data increases from relatively low values to a long-term steady value. This is in the nature of the logarithmic function. At times around 1000-4000 days or more, the change in long-term strain is larger than the change in the logarithm of time so cα increases. At much longer times, the remaining settlement strain is small and the change in the logarithm of time is so small that cα tends to its long-term secant value. In this case, cα is tending towards long-term values of between 0.04 and 0.10. These data are consistent with values reported by Sharma and De (2007) and others for wastes not subjected to enhanced decomposition treatments including leachate recirculation.

Note that the ca calculation behind the data in Figure 6 will be sensitive to tref, in other words to a different interpretation of the fill period or median fill age. However, the rationale for such reinterpretation is not clear. In fact, the lack of a fundamental or factual basis upon which to implement such an adjustment adds to the argument for simulating these data using a more fundamental approach (e.g. the HBM model). Using the HBM model, the fill period is represented (in its simplest form) as a linear filling sequence. Start and finish times may be imprecisely known but a more representative ‘average’ waste age is automatically obtained and resulting biodegradation-related strains are more realistically distributed through the waste depth. Of course, the magnitude of biodegradation-related settlement, rather than its distribution, is not simply defined but a mechanism for interpreting and implementing its causes and effects is provided within the HBM model.

Comparisons between the measured data and the HBM model predictions are given in Section 8.2.


Golder Associates

0.00

0

0.02

0

0.04

0

0.06

0

0.08

0

0.10

0

020

0040

0060

0080

0010

0

Tim

e [d

ays]

Calpha

0.12

0

0012

000

UK

Viri

dor 1

.2

UK

BR

E 3

.4

UK

Sita

1.2

NE

NT

1.3

UK

Viri

dor 1

.1

Lynd

hurs

t 1.3

UK

BR

E 3

.3

UK

Sita

2

Lynd

hurs

t 1.2

San

dtow

n 1.

2

UK

BR

E 3

.2

UK

BR

E 5

.3

Lynd

hurs

t 1.1

San

dtow

n 1.

1

UK

BR

E 3

.1

UK

BR

E 5

.2

San

dtow

n 1.

5

UK

BR

E 5

.1

UK

Sita

1.5

UK

Sita

3

San

dtow

n 1.

4

UK

BR

E 4

.1

UK

Sita

1.4

NE

NT

1.2

San

dtow

n 1.

3

UK

BR

E 4

.2

UK

Sita

1.3

NE

NT

1.1

Figure 6: cα Data Obtained from Assembled Settlement Strain with Tref = 30 Days and Offset from Time Zero by the Period of Delay Between Start of Filling and Start of Settlement Monitoring.


Golder Associates

3.0 DATA ACQUISITION

3.1 Data Requirements for Assessing Landfill Settlement

The data acquisition requirements for assessing landfill settlement were to obtain sufficient, suitable information from a range of different landfill types and filling scenarios to apply to the HBM model so that it can be validated. The required information comprised details of landfill settlement and other specific influencing factors (e.g. rate of filling, depths of waste (further discussed in Section 3.4) and waste type (further discussed in Section 6.2.1)). These data were obtained from a range of different landfill types (e.g. quarry and land raise, deep and shallow, different waste types, different climatic environments, etc).

Acquisition of settlement and associated data was obtained from operating landfills within the UK and also overseas. The inclusion of data sets from non-UK sources was important as this allowed the HBM model to be validated for a range of waste types, operational conditions and climate not experienced in the UK. This has enabled a more robust validation to be carried out to allow future extrapolation for modified waste compositions likely to be experienced in the UK (e.g. in response to pre-treatment and legislation led practice).

3.1.1 Input Data Requirements for HBM Model

The key input data requirement for the HBM model is the assembly of waste deposition and landfill performance data in order that existing characteristics, trends, and other influencing factors might be identified and quantified, such that future predictions could be derived based on previous operational trends.

From the function and determinability of all the HBM input parameters, three distinct types of parameter can be identified: site-specific, waste-specific and generic.

The relationship between these types of parameter is presented in Figure 7.


Figure 7: HBM Model: Function and Determinability of Input Data A summary of these types of input parameter and their inter-relationship is presented below:

• Site-specific: These input data describe a site in terms of its geometry, filling schedule and operational circumstances, e.g. leachate recirculation and compaction. The importance of the filling schedule should not be underestimated; it enables the HBM model to capture a variation of waste properties with depth, even if the other parameters used are set to default values.

• Waste-specific: This term relates to the parameters that define the material being landfilled. For a region or waste type not previously modelled, some effort to characterise the waste may be required. This type of parameter is obtained through relatively simple laboratory testing, e.g. waste classification, loss on ignition, or through compositional analyses of typical waste streams.

Golder Associates


Golder Associates

• Generic: These comprise several parameters controlling fundamental system behaviours, especially the biochemical system. For example, the interaction of VFA (volatile fatty acids) and MB (methanogenic biomass) in the HBM model enables microbiologically mediated phenomena such as VFA souring, methanogenesis and (subsequently) hydrolysis-limited digestion, to be captured. A programme of testing for this type of parameter would be a significant undertaking. The default set contained within the HBM model is based on a literature review of anaerobic processes in general. Model testing has shown that this data set provides a credible interpretation of observed landfill behaviour. Modification of the default generic parameters should be made within the context of sensitivity analysis.

It should be noted that many of the input parameters share characteristics with more than one parameter type. Two such groups of combined parameters have been identified, as follows:

• Combined site-specific and waste-specific: This group contains parameters that are related to both site circumstances and waste type. Saturated hydraulic conductivity is a good example. Mechanically Biologically Treated (MBT) and raw Municipal Solid Waste (MSW) differ markedly in their saturated hydraulic conductivity. That said, the difference between them is no more than that occurring between the base and surface of a 30 m depth of MSW. A similar argument can be made for dry density.

• Combined waste-specific and generic: In this group are parameters that may be validated in order to strengthen the quantitative performance of the model. They are mainly constitutive parameters but include one process coefficient, b. The effort to validate these parameters would be focussed according to the principal aim of the modelling, e.g. hydraulic, biodegradation and/or mechanical.

3.2 Availability and Acquisition of Data

As part of the Landfill (England and Wales) Regulations 2002, landfill operators are required to record data to demonstrate that their operations are being carried out in compliance with current permitting requirements. Minimum monitoring requirements and frequencies from the regulations are summarised in Table 2.

Typical reporting requirements are based on the original provisions of the Landfill (England and Wales) Regulations 2002, which have now been incorporate into the Environmental Permitting (England and Wales) Regulations 2007. However, landfill permits can include for site-specific variations, which are incorporated as part of bespoke permits, providing data evaluation is not adversely affected. Bespoke permits can also require other site-specific information to be reported, e.g. rain gauge data for surface water monitoring assessments or an increased frequency of monitoring.


Golder Associates

Table 2: Minimum Monitoring Procedures for Landfills, from Landfill (England and Wales) Regulations 2002 Operational phase After-care phase Leachate volume Monthly Every six months Leachate composition Quarterly Every six months Volume and composition of surface water

Quarterly Every six months

Potential gas emissions and atmospheric pressure

Monthly Every six months

Groundwater level Every six months Every six months Groundwater composition May be monthly, quarterly or annually

depending on site-specific requirements) (depending on site-specific requirements)

Structure and composition of landfill body

Annually N/A

Settling behaviour of the level of the landfill body

Annually Annually

Data for the structure and composition of the landfill should include details of the surface occupied by waste, the volume and composition of waste, method of depositing, time and duration of depositing and calculation of the remaining capacity available at the landfill.

The following information was requested as HBM model input parameters on the basis that this information should be readily available to landfill operators as part of their operational reporting requirements:

• Settlement records; • Cell/landfill dimensions (areas, waste depths); • Filling sequence; • Waste composition (types and quantities); • Rate of filling; • Lift thicknesses and compaction equipment; • Time of capping; • Landfill gas generation data; • Leachate management (e.g. was recirculation carried out); and • Climatic records. Settlement monitoring records were requested for the following:

• During filling; and • Post-filling. To assess the adequacy and reliability of each specific data indicator type, details of the actual period, frequency and method of data collection were also requested. This information was required to facilitate the evaluation and ranking of the data sets.


Golder Associates

3.3 Extent of Search for Data

The acquisition of data for the project was positively supported by major landfill operators, both UK and internationally.

Data sets were sought to be representative of a wide range of landfills and waste compositions, including old mineral workings, land raise disposal sites, municipal waste landfills and sites taking mainly commercial and industrial waste. Some operators who expressed support for the project were unable to provide data as they were considered either commercially confidential or not available in an accessible format.

Each of the data sets received was reviewed and scrutinised for evidence of relationships between indicators and controlling conditions as part of a pre-selection process before assembly and use in the HBM model. Details of all the data sets obtained are summarised below.

3.4 Data Sets Obtained for the Project

The following section summarises details of the information obtained from the various UK and overseas landfill operations, which has been gratefully received for this project. The source of each data set has been identified; however, the UK sites have been identified anonymously. The Hong Kong Government has approved the acknowledgement of the Hong Kong site as the NENT Landfill.

UK, SITA #1

This site comprises an approximately 150 ha land raise type landfill, on former agricultural land and aggregate extraction works. Waste disposal commenced in 1967 and remains operational, with a maximum pre-settlement waste thickness of 65 m. The landfill was initially a co-disposal site up to 2004, and is now a non-hazardous waste site. The filling history of the site is complex, due to it size, with filling sequences distributed around the site at various times to meet void requirements. Most cells were filled as single events, although some cells have been filled as part of multiple events (typically in two phases, but some cells filled in three of more periods, i.e. mainly within the middle of the site). The northern areas of the site were capped and restored in 1996.

Settlement monitoring data were provided from the late 1980’s up to approx 1994 for the older areas of the landfill at the north of the site. This part of the site was capped in 1996. Quarterly topographical survey data were also provided for the period from 2001 to 2006, which also provided details of the site, selected cell geometries and operation sequences including filling and capping schedules. Quarterly data on tonnages and waste volumes for the period 1994 to 2005 and waste inventory details (including details of waste composition and waste density) for the period 2004 to 2006 were also provided. Landfilling for the period


Golder Associates

2001 to 2006 typically comprised 75% biodegradable waste and 25% essentially inert commercial/industrial wastes. Tonnages per quarter typically varied between approximately 200,000 up to 300,000 tonnes, depending on demand, with initial waste densities typically determined in the range 0.95 to 1.10 t/m3 for this period. Landfill gas production details for the whole site were provided from 1980 onwards, together with leachate composition data for the period from 2003 to 2007, with some details of limited leachate recirculation.

UK, SITA #2

This landfill is approximately 25 ha, primarily of the disposal of commercial and industrial waste, with waste intakes in the order of 220,000 to 300,000 tonnes per annum. Owing to the nature of the waste, there is currently little gas generation and the dry waste input and the use of temporary clay caps has resulted in minimal leachate production. Effectively no leachate was generated until 2006.

Quarterly topographical survey data were provided for the period from 2000 to 2007, which also provided details of the site, selected cell geometries and operation sequences including filling and capping schedules. Quarterly data on tonnages and waste volumes (including details of waste composition and waste density) for the period 1998 to 2004 were also provided. Landfill gas production details for the whole site were provided from 1998 onwards.

UK, SITA #3

This is a steep sided quarry landfill, filled from 1993 to present, with a pre-settlement waste depth of 32 m. The landfill was initially a co-disposal site and is now a non-hazardous waste site for commercial and industrial wastes with small quantities of municipal solid waste, of low to moderate biodegradable content. Capping commenced in 2005.

The data set was acquired for a steep wall lined cell within the landfill. Quarterly topographical survey data were provided for the period from 2001 to 2007, which also provided details of the site, selected cell geometries and operation sequences including filling and capping schedules. In particular, one cell (Cell F) comprises a steep wall lining system with slip inducing membrane, which does not "piggy back" any adjacent cells. Quarterly data on tonnages and waste volumes (including details of waste composition and waste density) for 2002, (i.e. appropriate for Cell F) were also provided. Landfilling for 2002 (Cell F) typically comprised 85% domestic and civic amenities waste and 15% essentially inert commercial/industrial wastes. Waste input for 2002 was approximately 280,000 tonnes, of which 40,000 tonnes were deposited in Quarter 1, with the remaining 240,000 tonnes deposited over Quarters 2, 3 and 4. Landfill gas production details for the whole site were provided from 1993 onwards.


Golder Associates

UK, SITA #4

This landfill was formed from a worked out sand quarry, with a pre-settlement waste depth of 26 m. The landfill was initially a co-disposal site and is now a non-hazardous waste site for municipal solid waste and commercial/industrial waste.

Settlement monitoring data for Cell 4 were provided for a seven month period between September 2002 and April 2003. Quarterly topographical survey data were also provided for the period from 2000 to 2007, which also provided details of the site, selected cell geometries and operation sequences including filling and capping schedules, noting that the site remained uncapped several years. Quarterly data on tonnages and waste volumes for the period 2004 to 2006 and waste inventory details (including details of waste composition and waste density) for the period 1998 to 2006 were also provided. Leachate composition data were provided for 2007 together with landfill gas production details for the whole site from 1998 onwards.

UK, SITA #5

This is a steep sided landfill in a former rock quarry which commenced filling in 2006. The waste thickness varied from 18 m to 45 m, with waste composition comprising 80% biodegradable fraction.

Settlement monitoring data were provided for a seven month period between January to August 2002. Topographical survey data were provided for the period from 2001 to 2006, which included details of the site, selected cell geometries and operation sequences including filling and capping schedules.

UK, SITA #6

This is a steep sided quarry landfill, with a waste thickness of up to 25 m, with a low biodegradable content.

Topographical survey data were provided for the period from 2001 to 2006, which provided details of the site, selected cell geometries and operation sequences including filling and capping schedules. Annual data on tonnages of waste for the period 2004 to 2006 were also provided.

UK, SITA #7

Topographical survey data were provided for the period from 2002 to 2004, which provided details of the site, selected cell geometries and operation sequences including filling and capping schedules.


Golder Associates

UK, Viridor #1

This landfill was filled from July 2002 to April 2005, with a waste depth ranging from 27 m to 35 m. The waste comprised approximately 75% municipal soil waste and 25% commercial/industrial waste.

Settlement monitoring data for a recently landfilled cell were provided for a seven month period between June 2006 and January 2007. Topographical survey data were provided for the period from 2005 to 2007, which also included details of the site, selected cell geometries and operational sequences including filling and capping schedules. Quarterly data on tonnages of waste (including waste composition) for the period 2002 to 2005 were also provided, as were monthly rainfall data for the period October 2001 to January 2007, together with details of leachate volume, composition, extraction for the period 2003 to 2007. Leachate had been removed from the cell to comply with the permitted leachate depth, although no significant re-circulation was carried out. Average landfill gas composition was also provided for the period from January 2006 to July 2007.

UK, BRE #1

Settlement monitoring data were provided between the period December 1999 to February 2003, together with details of the commencement and completion of filling and original waste thickness.

UK, BRE #2

Settlement monitoring data were provided between the period May 2000 to July 2002, together with details of the commencement and completion of filling and original waste thickness.

UK, BRE #3

Settlement monitoring data were provided, in the form of percentage compression of original waste thickness for a period of up to approximately 6.5 years from completion of filling.

UK, BRE #4

The current landfill site comprises two adjacent former clay pits separated by a natural bund. The total current landfill area covers approximately 74 ha. There is a 10 to 20 m difference in elevation between the bases of the two pits with an average depth of each pit of 20 m. Side slopes vary, with some steep (> 1:2) and the older parts of the natural clay site was unlined. Filling is to achieve a single domed profile with a maximum pre-settlement waste depth of around 60 m. Waste disposal began around 1980 and continues, principally comprising household refuse plus commercial and inert waste.


Golder Associates

Settlement monitoring data were provided, in the form of percentage compression of original waste thickness. The data ranged from a period of approximately 5 months up to 19 years from completion of filling. Summary details of the sequence of waste filling and capping for ‘Pit 4’ and ‘Pit 5’ (i.e. Cells 1, 2, 3, 4 and 5) were also provided.

UK, BRE #5

Settlement monitoring data were provided between the period August 2001 to April 2006, together with details of commencement and completion of filling and original waste thickness.

NENT Landfill, Hong Kong (also identified in this report as Hong Kong, #1)

This landfill covers approximately 60 ha, with a maximum waste thickness of 140 m, equivalent to a maximum waste capacity of 35 million cubic metres. Waste disposal commenced in 1995, comprising municipal, commercial and special waste types.

The NENT data set describes the filling of a landfill cell in two distinct stages: Phase 1 between July 1995 and September 1998 and Phase 2 between January 2001 and August 2002.

Settlement monitoring data on a fortnightly basis of 54 surface markers were provided for a period between September 1998 (i.e. end of Phase 1) and October 2006 (approximately 3.5 years). Annual topographical survey data were also provided for the period from 1999 to 2006.

Weekly and monthly data on waste mass composition were provided for a 9 year period between November 1997 and November 2006. Daily site-wide data on leachate quantity and landfill gas production and composition were provided, together with monthly and quarterly leachate quality data, for the period from 1995 to 2006.

Australia, Lyndhurst Landfill, Victoria

Details of an experimental test cell at Lyndhurst landfill, Victoria, Australia were obtained from Yuen and McDougall (2003).

The experimental test cell is approximately 1.4 ha in area, comprising partly landfill and land raise, on a compacted clay base and side-wall lining system, with a waste thickness of 10 to 15 m. Waste disposal comprised municipal soil waste with inert material provided as daily cover. Filling commenced in December 1993 and was completed in December 2005.

Data from detailed instrumented monitoring of the Lyndhurst experimental test cell included site-specific data such as cell geometry, filling schedule, compacted density, daily cover, leachate re-circulation and waste-specific data such as waste composition and temperature,


Golder Associates

moisture content leachate and landfill gas composition and production. Full details of all the monitoring data obtained from the Lyndhurst experimental test cell can be found in Yuen (1999).

USA, Central Solid Waste Management Centre (CSWMC), Sandtown (Delaware Solid Waste Authority)

Details of CSWMC landfill facilities in Sandtown were obtained from the paper by Morris et al. (2003).

The CSWMC began operation in October 1980 with the opening of the 3.6 ha Area A disposal cell. After 2 years of landfilling in this cell, the facility was expanded in October 1982 by constructing a 7.3 ha Area B disposal cell adjacent to Area A. Areas A and B (i.e. 10.9 ha in total) were constructed with a geomembrane liner system and a leachate collection system. At different operational and post-closure periods, leachate was recirculated in Area A/B using: (1) leachate injection wells (operational period); (2) surface spray irrigation (operational and post closure periods); and/or (3) top surface application via leachate recirculation fields (post-closure period). Area A/B was closed in October 1988.

Measurement of landfill settlement in Area A/B was based on landfill volume estimates taken from aerial surveys taken on four separate occasions: October 1989; February 1992; June 1995; and August 2001. Settlement calculations were based on the difference form the as built landfill base plan and the respective cover topography at each date. Settlement of the base was not considered which may have lead to an over estimate in volume calculations.

3.5 Challenges in Data Acquisition for the HBM Model

The main challenge in the acquisition of data for the HBM model has been obtaining reasonably complete site-specific and waste-specific data in a format which was readily assembled as HBM model input parameters. In England and Wales, site-specific data are recorded as part of the PPC permit (now the Environmental Permitting (England and Wales) Regulations 2007) or earlier Waste Management Licence requirements. In addition, for some operators, data are recorded and compiled for the operator’s own operational and commercial reasons.

The quality and completeness of the data sets improved in the more recent records, as the appreciation of the benefits of systematically recording data and of compiling them into an integrated data set increased. The ideal data set would be the complete list of data as set out in Section 3.2 and would enable the influence of all the input parameters to be assessed, from the initial filling of a landfill site through to final mechanical equilibrium, and in terms of the environmental conditions for each of the hydraulic, biodegradation and mechanical systems.


Golder Associates

UK and international operators and researchers have been very supportive of the data requests for this project. However, a complete set of the requested site and waste-specific data was not able to be obtained for any one single landfill. Some operators advised that their data were kept in formats and locations suitable for their own operational use and were not in a condition which was readily retrievable or suitable for the project. As such, those operators were unable to contribute further.

Some mandatory reporting intervals were not ideally suitable for predicting settlement. Under the Landfill (England and Wales) Regulations 2002, settlement monitoring of the waste mass post-filling is only required on an annual basis. Some operators undertake settlement monitoring during filling at a much greater frequency per annum, which provides actual settlement trends based on operational conditions which can be used for establishing future settlement predictions. Settlements measured by fixed survey markers placed in a cap, or temporary cover on the waste, provided the most accurate and usable data. Any subsequent filling phases can also be readily identified, providing the fixed survey markers are re-established in approximately the same position in plan in the new surface.

Settlements estimated from periodic topographical surveys of the cap or restored surface of the landfill or landfill cells and the data used to create digital ground models provide only a general indication of landfill settlement. Topographical surveys do not necessary provide a comparison of the same position for each monitoring interval and do not readily identify any filling sequences of unusual features within the monitored period. As such, settlement profiles derived from topographical surveys are generally less accurate, more time consuming to assemble into a usable form, and required a greater number of assumptions for use in settlement composition.

One operator who supplied data uses waste densities, determined on a quarterly basis from waste tonnages and topographical surveys throughout waste filling, as the principal basis for analysing waste settlements to predict void space for operational planning and commercial benefit. These data are used to enable estimates of future settlements to be made which, in turn, allows better predictions to be computed of remaining void space, future waste intakes by weight and the lifespan of the landfill. Also, these data are used to support discussions with regulators regarding environmental controls and agreeing pre-settlement waste levels. The records are collected on a quarterly basis, as more widely spaced collection is seen as being too infrequent for their purposes.

Individual cells can be filled quickly to their pre-settlement levels within a site and capped but elsewhere within the same landfill, filling may be much slower because of the site geometry and planned filling sequence. Alternatively, cells may be capped temporarily and then subjected to a later period of filling. Over time, the composition of the waste stream can also alter which is currently happening in the UK caused by changes in waste streams and as the restrictions on the landfilling of biodegradable wastes in the Landfill Directive take effect. In


Golder Associates

these cases, each completed cell has to be considered independently as the settlement characteristics of the wastes in the different cells can be substantially different.

Analysis of values and trends in waste densities also indicates changes in the nature of the waste streams over time, i.e. changes to the proportions of commercial and municipal solid waste. Recent work by Dixon and Langer (2005) and Zekkos et al. (2005) has provided an available framework within which to associate waste composition with mechanical properties. Information on the biodegradability of municipal soil waste is obtainable from National Household Waste Analysis Programme (NHWAP) and the Landfill Allowance Trading Scheme (LATS) testing.

Landfill gas production and composition (reported on a monthly basis for operational landfills in the UK) is usually only available for a landfill site as a whole, rather than any discrete cell or cells, which makes it difficult to compare gas generation predictions made by the HBM model with site gas monitoring records.

Based on the availability of settlement data, and accompanying information on waste composition and filling sequencing, the following sites, or discrete cells within landfill sites, were selected for comparisons between the HBM model and the site-derived settlement data:

• UK, Viridor #1; • Hong Kong, #1 (NENT); • UK, SITA #2 • UK, SITA #3; and • UK, BRE #4. These sites also provided a range of waste types, ages and compositions. UK, Viridor #1 provided the most complete information about filling and the shortest time between filling and the onset of settlement monitoring. UK, BRE #4 provided a broad range of settlement data from the different ages and settlement magnitudes. SITA #2 and SITA #3 settlement data were derived from settlement profiles and also contained variable and irregular filling sequences, useful for assessing these influences within the HBM modelling.

Settlement data are available in the form of post-filling settlement markers and survey data. Settlement during the filling stages can be estimated from survey data; however, filling events skew the analysis of settlement magnitude.


Golder Associates

4.0 HBM MODEL OVERVIEW

4.1 Conceptual Framework

The HBM model provides a framework for the integrated analysis of the hydraulic, biodegradation and mechanical behaviour of landfilled waste or other degradable soils (McDougall, 2007). Building on individually-proven models of hydraulic, biodegradation and mechanical behaviour, the HBM model gives a synergistic interpretation of landfill behaviour with relatively light input parameter requirements.

The HBM model comprises three main system models and link routines, through which the algorithm passes, as shown in Figure 8. It is in the link routines that the most recent system variable values are used to update the conditions within each system model.

HYDRAULICHYDRAULICUnsaturated flow model

(hydraulic pressure head ↔ moisture content)

BIODEGRADATIONBIODEGRADATIONTwo stage anaerobic digester

(VFA ↔ methanogenic biomass ↔ solid degradable)modified enzymatic hydrolysis of solid degradable

“HBM”Conceptualframework

MH LinkMH Linkupdates:

volumes/ratioshydraulic conductivity

HB LinkHB Linkupdates:

volumes/ratios

MECHANICALMECHANICAL“bio-visco-elasto-plasticity”

(load, time ↔ displacement)

creep settlement

creep-hardening

BM LinkBM Linkupdates:

volumes/ratiosdVV = ΛdVS

bio- softeningmesh displacement

Figure 8: Schematic Representation of the Hbm Conceptual Framework (Mcdougall, 2007) As isolated system models, the hydraulic and mechanical models are close to practice, and are the basis of well established design tools. However, in the HBM framework each system model can modify parameters in other systems. These system interdependencies are the innovative aspect of the HBM model. They are less well understood and have necessitated a fundamental review of the performance of the combined framework.


Golder Associates

Table 3 adds some substance to the state of model development in isolation and the interdependencies within the HBM framework. There are significant challenges in the interdependencies; for example, the mechanical consequences of decomposition have, until now, received little attention in either the landfill or geotechnical research communities. But the potential rewards are significant. If the interdependencies can be sensibly understood, then hitherto disparate behaviours can be analysed in a much more meaningful and coherent context.

Table 3: Hydro-Biological-Mechanical Model Interdependence

Interpretation/Interdependence Systems Hydraulic Biodegradation Mechanical

Hydraulic Long history of unsaturated flow modelling. Commercial software available.

Hydrolysis of solid organic matter is moisture-limited

Impact of moisture/suction on compressibility or strength unquantified

Biodegradation Moisture is consumed during hydrolysis.

Long history of two-stage anaerobic digestion modelling. HBM uses moisture dependent enzymatic celluylolytic hydrolysis

Few studies of impact of decomposition on mechanical behaviour. HBM introduces decomposition-induced void change.

Mechanical Variation of density with depth and dual porosity control retention and permeability.

Porosity constrains mobility of microbes.

Well-established legacy of mechanical modelling in unsaturated soils. Commercial software available.

4.2 Implementation

The HBM model is implemented using the finite element method with each system model sharing a common mesh. Using this method it is possible to account for material and operational features such as complex section geometry, waste heterogeneity, anisotropic hydraulic conductivity and simulation of the filling phase.

Operation of the HBM model, i.e. preparation of simulation input data and interrogation of output data, is through the graphical user interface – HBM GUI.


Golder Associates

drainage layerleachate discharge

measured here

elements of placed fillexposed to rainfall/evaporation,

biodegrading, density modified by overburden

elements of fill yet to be disclosedexposure to rainfall/evaporation &

biodegradation begins on disclosure

rainfall/evaporation

a finite element given initial mass, organic fraction, moisture content,

absorptive capacity

drainage layerleachate discharge

measured here

elements of placed fillexposed to rainfall/evaporation,

biodegrading, density modified by overburden

elements of fill yet to be disclosedexposure to rainfall/evaporation &

biodegradation begins on disclosure

rainfall/evaporation

a finite element given initial mass, organic fraction, moisture content,

absorptive capacity

Figure 9: Finite Element Mesh Illustrating Filling (McDougall, 2007). 4.3 Applications of the Model

4.3.1 Settlement

The time at which settlement is completed can be estimated using the model. The effect of recirculation of leachate and water ingress can be included to simulate those processes adopted to accelerate biodegradation and reduce the time to completion. The method can also be used to predict the settlement that has occurred at any time in the future.

4.3.2 Predicting Gas Production

The model can be used to estimate the time and relative location of maximum gas production. If the phased filling option is adopted in the model, the gas production at relative positions within the landfill can be estimated. This will also indicate when a decline in gas production will occur and indicate an optimum depth for retro-drilled gas wells providing a detailed compositional input dataset is available.


Golder Associates

5.0 DATA REQUIREMENTS FOR THE MODEL

This section details the input parameters required for the HBM model, the meaning of the individual parameters and an indication of how they can be derived or from what literature sources the default values have been sourced. The input parameters can have an influence on all aspects of the HBM model behaviour through the associated link routines. The link routines are controlled directly by the code and are a function of the input individual hydraulic, biodegradational and mechanical inputs, therefore require no controlling input factors.

The purpose of this section of the report is to provide an overview of the input parameters, not to provide a full theoretical justification for the use of the parameters; however, in places it is considered appropriate to discuss the parameter in greater depth to explain the associated influence on the model.

5.1 Hydraulic Data Requirements

5.1.1 Van Genuchten

Van Genuchten α and n values refer to the moisture retention properties of the waste mass. It is envisaged that generic default values will be adopted for these parameters α = 1.4 and n=1.6 are derived from laboratory tests by Kazimoglu et al. (2005).

5.1.2 Residual Moisture Content

The residual moisture content is expressed by weight ratio. Default values will be adopted for this parameter of 0.25 based on neutron probe data obtained by Yuen (1999).

5.1.3 Specific Storage

Whilst this parameter is still contained within the input console, it is a relict parameter from a previous model version and an input parameter is not required.

5.1.4 Hydraulic Conductivity

The hydraulic conductivity, k, has a wide range of values as waste has highly variable components, types and amounts of cover soil differ between sites, the percentage of inert and industrial wastes varies, and placement procedures play an important role.

5.1.4.1 Uniform Hydraulic Conductivity

Literature values range from 10-3 to 10-9 ms-1. A value of 5 x 10-5 ms-1 has been selected as a plausible median figure for waste. Beaven (2000) shows data from the Pitsea compression


Golder Associates

cell and identifies a density (hence depth) dependant hydraulic conductivity function, which should be used, where possible, in lieu of a single value. Qian et al. (2002) summarised the literature values for hydraulic conductivity of municipal solid waste, and concluded that the average (mean) hydraulic conductivity for MSW was approximately 1 x 10 -5 ms-1, with the mean values ranging from 1.5 x 10 -6 to 4 x 10 -4 ms-1. Assumed literature values may be used where site data are not available.

5.1.4.2 Depth Dependant Hydraulic Conductivity

It is known that the saturated hydraulic conductivities of a sample of waste can vary by three or more orders of magnitude when compressed over the range of stresses exerted within 20 m depth of waste (Powrie et al., 1998). In response, the HBM model allows for the saturated hydraulic conductivity to be controlled by the void phase volume of each element. A relationship between saturated hydraulic conductivity and the void phase volume as a ratio of the solid inert phase volume is used. The influence of decomposition on hydraulic conductivity, through its impact on void volume, can be realised as follows.

From data obtained by Beaven (2000) on the saturated hydraulic conductivity of household wastes in a large (2 m diameter x 2 m high) compression cell shown in Figure 10, it is evident that the relationship between dry unit weight and saturated hydraulic conductivity can be approximated by a function of the form:

).exp( dsat cBk γ= (5.1)

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

0.0 2.0 4.0 6.0 8.0 10.0

Dry unit weight [kN/m3]

Sat

urat

ed h

ydra

ulic

con

duct

ivity

[m/s

]

DM1

DM3

PV1

AG1

y = 0.15exp(-2x)

Figure 10: Variation of Saturated Hydraulic Conductivity with Dry Unit Weight, from Data Published by Beaven (2000) where γd is the dry unit weight, B and c are fitting parameters. Approximate values of B = 0.15 and c = –2.0 can be seen for DM3, a MSW, shown in Figure 10. However, in a degrading soil comprising inert and degradable solid phases which have different phase


Golder Associates

densities (or specific gravities), the overall dry unit weight does not necessarily correspond to a unique volumetric state.

A more useful controlling variable would be the void to inert phase ratio, ei, given by,

SI

Vi V

Ve = (5.2)

where VV is the void volume and VSI is the solid inert phase volume (McDougall and Pyrah, 2004). By assuming an inert/degradable waste composition and corresponding unit weights, equation (5.2) can be expanded to define ei in terms of overall dry density, i.e.

( ) ( ) 11....

1.

−−

−−

=ωγ

ωγωγ

γ

wSD

wSI

d

wSIi G

GGe (5.3)

where GSI is the specific gravity of the inert phase component, GSD is the specific gravity of the degradable phase component, γw is the unit weight of water and ω is the mass fraction of solid degradable matter as a proportion of total solid mass.

Data from Figure 11 below can then be re-interpreted as a function of ei. Figure 11 shows the DM3 conductivity data calculated in this way, assuming the solid degradable dry weight fraction of sample DM3 is 0.54, the specific phase weights of the inert phase GSI.γw = 17 kNm-3 and degradable phase GSD.γw = 7.3 kNm-3. The fitted function is logarithmic in form:

(5.4) 9.48 )(0.1 isat exek −=

Equation (5.4) is currently hard-coded into the HBM model and can be selected in preference to a fixed input saturated hydraulic conductivity. One should be aware, however, that preferential flow pathways are common in waste, so bulk saturated hydraulic conductivity might misrepresent field conditions.


Golder Associates

1.0E-08

1.0E-07

1.0E-06

1.0E-05

1.0E-04

1.0E-03

0.0 2.0 4.0 6.0 8.0 10.0

Void to inert phase ratio

Sat

urat

ed h

ydra

ulic

con

duct

ivity

[m/s

]

DM3 (Beaven, 2000)

Function

Figure 11: Waste Saturated Hydraulic Conductivity Data (Dm3 from Beaven, 2000) Presented as a Function of Void to Inert Phase Ratio 5.1.5 Anisotropy with Respect to Hydraulic Conductivity, Ratio

Horizontal / Vertical

An anisotropy value of 1 is assumed as a default value. However, waste heterogeneity and the inclusion of daily cover soils along with compression induced stratification are likely to result in anisotropy in many cases, requiring modification of this parameter from the default.

5.2 Biodegradation Model Parameters

5.2.1 Maximum Hydrolysis Rate

Maximum hydrolysis rate represents a limiting rate on the hydrolysis reactions within the biodegradation model. An assumed value derived from volatile fatty acid growth vs. time plots is considered to be appropriate for this parameter. Assumed values have been taken from Barlaz et al. (1989) (1800 g.m-3

(aq).day-1) and Jones and Grainger (1983) (3000 g.m-3

(aq).day-1). Solid loss may also be considered. Cecchi et al. (1988) and Wang and Banks (2000) suggest values of 4000 – 5000 g.m-3

(aq).day-1. A default value in the model of 2500 g.m-3

(aq).day-1 is considered appropriate for MSW. With other processed waste stream wastes (e.g. MBT), the material is some way along its decomposition 'path' so conceptually it may be better simply to manipulate digestibility parameters to manage (factor down) maximum hydrolysis.


Golder Associates

5.2.2 Product Inhibition

This parameter is based on volatile fatty acid concentration and allows for acid accumulation. A default value of 2 x 10-4 m3.g-1 is assumed which results in peak Volatile Fatty Acid (VFA) values of about 16,000 g/m3.

5.2.3 Digestibility

The digestibility, n, represents how degradable the organic matter is. Wald et al. (1984) gave a value of 0.7 for straw rice, a lignified cellulose which can be likened to MSW (Bookter and Ham, 1982).

A high value of n represents a high percentage of rapidly degrading material; hence the degradation rate remains high (see Figure 12).

Deg

rada

tion

rate

Percentage material degraded 0 100 %

n > 1

n = 1

n < 1

Figure 12: Influence of Digestibility The digestibility can be used to alter the decomposition rate and can be used to represent the quantity of low, medium and high degradability content material within the waste. A digestibility, n, value of 1 gives a linear relationship between material degraded and relative digestibility. A value below 1 represents lower quantities of rapidly degradable matter, while a value greater than 1 represents higher quantities of such materials.


Golder Associates

It is not expected that quantitative values for this parameter will be derived from MSW samples. Whilst the parameter can be qualitatively manipulated to represent differing waste streams, it is a recommended area of further research that this parameter is linked to a waste classification system that can be applied by waste operators to give consistent and reliable link between the composition of the waste and the digestibility and solid degradable fraction percentages.

5.2.3.1 Methanogenesis

The depletion of the methanogenic substrate and methanogen growth are described by Monod kinetics, hence for Methanogenic Biomass (MB) accumulation (rj ):

( ) mck

ckr

MCj +

= 0 (5.5)

where k0 is the maximum specific growth rate, kMC is the half saturation constant and m is the MB concentration. What is c? The rate of VFA depletion, rh, is directly related to MB accumulation through a cell/substrate yield coefficient, Y:

Yr

r jh = (5.6)

The MB decay rk is given by: (5.7) mkrk 2= where k2 is the methanogen death rate. Estimates for the methanogenic parameters were originally sought from a literature review but there were little data relating to MSW and those which were available covered a range of values (see Table 4). These data provided the starting point for methanogenesis parameter selection and were subsequently refined following a parametric sensitivity study (McDougall and Philp, 2001).

Table 4: Mineralisation Parameters and Values

Reference K0 KMC Y k2 Straub and Lynch (1982) Model waste

0.03 day-1 5000 mg/L 0.04 0.01 day-1

Lee and Donaldson (1985) Cellulose

0.5 day-1 4200 mg/L 0.75 0.02 day-1

Viturtia et al. (1995) Pig manure

0.57 day-1 3280 mg/L 0.19

El-Fadel et al. (1996) Various

0.25 day-1 500 mg/L 0.06 0.03 day-1


Golder Associates

As a result of that study, the HBM GUI derived input files were populated with a set of default parameters, as given in Table 5.

Table 5: Biodegradation Model Parameter Values: Default Values

b k0 k2 kMM kVFA n Y Description Enzymatic

hydrolysis methanogen

growth methanogen

decay Half rate

Product inhibition

digestibility yield coefficient

Units g.m-3(aq).day-1 day-1 day-1 g.m-3 m3/g(aq) # g.g-1

Default 2500 0.02 0.002 4000 2e-4 0.7 0.08

5.2.4 Diffusion Coefficient

This is a fundamental numerical function of the model, however, the input value is insignificant compared to the Growth and Decay functions. The value of 0.05 m2.day-1 is acceptable to maintain numerical stability. This is not to be treated as an input parameter for user alteration.

5.2.5 Initial Solid Degradable Fraction

Initial solid degradable fraction represents the percentage degradable material. This is typically derivable from landfill operator-supplied data and represents the solid phase which can be lost during the decomposition process. Figure 13: represents the phase relationships during degradation, and shows the loss of solid degradable fraction in a single element.


Golder Associates

Figure 13: Loss of Solid Degradable Fraction (shown in red) with Time

Whilst the parameter can be estimated from site-derived data for differing waste streams, it is a recommended area of further research that this parameter is linked to a waste classification system that can be applied by waste operators to give consistent and reliable solid degradable fraction percentages and digestibility parameters for use in the model.

Classification of waste to the Consolidated European Waste Catalogue is an existing requirement on landfill operator, however this document contains broad generic categories, and material specific classification would be beneficial.

5.2.6 Initial VFA Concentration

This is an arbitrary value and has little influence over model performance. Hence an initial value of 300 g.m-3 is suggested for use in the model.

5.2.7 Initial Methanogenic Biomass

Unlike initial VFA concentrations, the initial biomass values has an influence on the hydrolysis reactions within the model. An initial value of 250 g.m-3 is suggested following sensitivity analysis by McDougall and Philp (2001), and is taken as the default value within the model.


Golder Associates

5.3 Mechanical Model Parameters

The mechanical model in the HBM model is based on the modified Cam Clay model and the initial model parameters represent the critical state parameters from the Cam Clay model.

5.3.1 Elastic Stiffness and Elasto-Plastic Stiffness

The elastic stiffness, κ, represents the elastic stress - void ratio relationship (unloading, reloading line). A value of 0.072 is taken as the model default taken from analysis by McDougall and Hay (2005) from data from Olivier et al. (2003).

Elasto-plastic stiffness, λ, is the elasto-plastic stress – void ratio relationship (loading line). A value of 0.23 is taken as the model default from analysis by McDougall and Hay (2005) from data from Olivier et al. (2003). Values derived by Zhang (2007) from Beaven (1999) suggest a λ value of 0.279.

5.3.2 Critical State Friction Constant

The Cam Clay Critical state constant, M, is taken as 1.2; this parameter is related to shear failures in the waste. In the analyses carried out in this report the waste is considered without a temporary or final waste slope which may induce instability, therefore, this parameter is not a controlling factor.

5.3.3 Poisson’s Ratio

The Poisson’s ratio, μ, represents the compressibility of the material and the relationship between vertical and horizontal strains. A value of 0.35 is used, derived from back analysis by McDougall and Hay (2005) from data from Olivier et al., (2003).

5.3.4 Initial Yield Stress

A value of 30 kPa is taken as the initial yield stress representing previous compaction forces.

5.3.5 Creep Viscosity

A value of 0.0015 is used from back analysis by McDougall and Hay (2005) from data from Olivier et al. (2003).

5.3.6 Decomposition-induced Void Change Parameter

The decomposition-induced void change, Λ, is a very significant property to the overall settlement magnitude and represents the void ratio change as a function of solid loss due to degradation. Incorporation of this parameter is a fundamental development of the HBM


Golder Associates

model, however, there is a dearth of knowledge that surrounds the physical effect of decomposition of the waste structure. This is a significant area requiring further research.

Biodegradation-induced settlement is difficult to analyse using conventional soil mechanics volume descriptors because of the changing solid phase mass and volume. For example, the degradation of solid mass may be accompanied by an equivalent (to the solid volume lost) increase in void volume. There is then no change in overall volume although the void ratio has increased and the waste has become more skeletal. Alternatively, decomposition may be accompanied by contemporaneous particle rearrangement resulting in overall settlement. In this case the associated change in void ratio is a function of changes in both void and solid phase volumes; it may even remain unchanged despite overall settlement. McDougall and Pyrah (2004) proposed a constitutive relationship between decomposition of solid degradable fraction, i.e. a change in solid phase volume VS, and the induced change in void volume VV at constant stress, of the form.

SV dVdV Λ= (5.8)

where Λ is the decomposition (or degradation)-induced void change parameter.

Table 6 summarises changes in volumetric state variables and likely mechanical consequences associated with key values of Λ.

Table 6: Decomposition Induced Void Change Parameter - Reference Values and Associated Phase Composition Changes Where dVs < 0

Λ Void ratio Overall volume Phase composition and its expected strength -1 Maximum increase No change Much looser and possibly weaker 0 Increase Reduction Looser and possibly weaker e No change Large reduction No change >e Decrease Maximum reduction More compact and possibly stronger e = void ratio

There are two ways in which Λ can be used to interpret biodegradation-induced settlement. Firstly, Λ can be used to quantify the overall impact of decomposition on waste settlement, i.e. to predict the long-term volumetric state for a given combination of material composition and environmental control parameters. On the other hand Λ can be used incrementally to distinguish periods of zero overall volume change from periods of contemporaneous settlement, even accelerated settlement. In either case, the form of Λ can be calculated from compression tests in which the progress of decomposition is known. There are, however, little such data available. Quantification of Λ was obtained from a large-scale long-term laboratory test on waste refuse (Olivier and Gourc, 2007). A value of Λ ≈ –0.65 was obtained from Olivier and Gourc (2007). Other tests on sand-gypsum and sand-halide mixes, from which the soluble particles were gradually dissolved, indicate that the incremental value of Λ is not constant but changes with the progress of decomposition; more interestingly, the form of the change is not erratic but follows a steady path (unpublished data). If more data on the


Golder Associates

overall and incremental values of Λ can be obtained, then the HBM model will operate in a more reliable predictive mode.

Void phase changes are realised as one-dimensional vertical-only deformations. Quantitatively meaningful simulations of the full HBM formulation can therefore only made on vertical columns.

5.3.7 Decomposition Hardening

Whether used as an indicator of overall or incremental behaviour, the use of Λ offers two important advantages over simpler methods. Firstly, biodegradation settlement is not treated as a time-dependent process. Time is communicated through solid organic matter depletion, which is controlled by the biodegradation model.

In this way there is a maximum rate of depletion but within that rate, under the influence of moisture deficit/addition or acid accumulation or changing crystallinity, for example, decomposition may slow down, accelerate, or stop completely.

Secondly, this approach allows for the definition of hardening or (more likely) softening with decomposition. Note, however, that the hardening being considered here occurs in response to a change in phase composition, which in turn is induced by degradation of the solid phase and hence controlled by the biodegradation model.

5.3.7.1 Form and Implementation of the Biodegradation-Hardening Rule

It is evident from Table 7 that by combining the void ratio e, with Λ, a simple means of controlling both biodegradation-induced hardening and softening is obtained. When Λ < e, decomposition leads to an increase in void ratio, whereas when Λ > e, void ratio decreases. If it is assumed that changes in void ratio affect waste in the same way as in conventional soils, i.e. by triggering changes in the yield condition, then a biodegradation-hardening rule can be defined thus:

( )S

Sd

VdV

edh Λ−Ω= (4.9)

where dhd is the increment in yield surface tip stress due to biodegradation, Ω [kPa] is a decomposition hardening multiplier that relates the magnitude of tip stress increments to increments of strain. The qualitative behaviour of the yield condition and its response to the biodegradation-hardening rule is illustrated in Figure 14. When Λ < e, decomposition is associated with reductions in yield stress, the material softens. In contrast, when Λ > e, solid volume loss produces an increase in yield stress and the material hardens.


Golder Associates

Note also that Ω is a parameter about which very little is known, however, it is possible to perform sensitivity analyses and explore likely value ranges and a value of Ω = 2 has been established in this way. Whilst the parameter is user defined it is suggested a value of 2 is adopted.

0

10

20

30

40

0 10 20 30 40 50 60 70Mean stress

She

ar s

tress Hardening

Softening

σy

e<Λ

e>Λ

0

10

20

30

40

0 10 20 30 40 50 60 70Mean stress

She

ar s

tress Hardening

Softening

σy

e<Λ

e>Λ

Figure 14: Changes in Position of Yield Surface with Decomposition (McDougall, 2007). 5.3.8 Dry Unit Weight (as placed)

This can be estimated from site-supplied data or estimated from input volumes and tonnages. It should be noted that this can be a crude assessment, as settlement and compaction due to overburden pressure can result in uncertain results. Waste components have a controlling influence on the average unit weight of the waste mass. Individual waste components have a wide range of particle unit weights and these can change with time owing to degradation of components with organic content which will result in a loss of mass, changes in size and alteration of the mechanical properties (e.g. compressibility and shear strength). These effects will also change the unit weight of the component. A default value of 5 kN m-3 is taken as a default parameter.

It should be noted that this parameter refers to dry waste and may relate to significantly higher values at field capacity.

Depth dependency occurs as a result of the HBM model (transient stage) solution.

5.3.9 Particle Weight

The particle weight represents the assumed mean particle densities for inert and degradable waste components of the waste. Values of 17 and 7.3 kN m-3 respectively are chosen as defaults. Site-derived estimates from particle types should be used in lieu of these values where possible.


Golder Associates

The inert unit weight of rock has been assumed as 27 kN. m-3, hence it is suggested that this higher value be used where the unit inert fill is only rock, building rubble et cetera.


Golder Associates

6.0 DATA ASSEMBLY

6.1 Data Required for HBM Validation

This section identifies the parameters that should preferably be validated and those that it is critical that they are validated. Two notations are used in this section of the report to identify the two types of parameter – one is shown in bold font and the other in italic font.

Parameters in bold:

• These values should be material-specific; and • The values are key controlling factors and it is required that these are validated against

site-derived data to validate the performance of the model. Parameters in italics:

• Material-specific values are desirable. Where measured values cannot be obtained, a representative value should be selected, with justification, where possible; otherwise the default values should be selected; and

• Validation of these parameters against site-specific data would increase confidence in the model predictions.

A brief description is given with each parameter to identify which can be acquired from site-derived data.

6.1.1 Input Parameters

• Saturated hydraulic conductivity

This parameter is not available from the site-derived data. A default value of 5 x 10-5 m.s-1 has been adopted and parametric studies on these values taken. Hydraulic conductivity of waste is highly complex due to the homogeneous nature of the waste and the likelihood of preferential flow path formation, so simplification is required for modelling; however, it is reasonable to adopt a depth-dependant profile for the waste.

• Initial solid degradable fraction

This can be estimated from data provided by landfill site operators. The site data must be interpreted to select the percentage that is actually degradable.

Derivation of this parameter is considered in Section 6.2.1


Golder Associates

• Dry unit weight (as placed)

This parameter was provided for several of the sites; however, concerns still surround high values that have been derived from survey/input tonnages as additional settlement, primarily due to compaction under self weight, will give high input-tonnage:volume-filled ratios.

• Digestibility

Absolute values of digestibility are not obtainable from site-derived data. This value can be used to control the generic consistency in terms of rapidly or slowly degrading material in a qualitative manner, using values in the range of 0.07 and 1.5 for slowly degrading and rapidly degrading waste respectively.

• Elastic stiffness, κ and Elasto-plastic stiffness, λ

Acquisition of waste stiffness is not possible from site-derived data, and it is recommended that for this aspect of the model, the default factors are adopted.

• Decomposition-induced void change (Λ)

This parameter is estimated. In practice, this is a very challenging parameter to estimate, as it has no obvious manifestation in terms of the waste composition/type or physio-chemical properties. This parameter requires significant further research to allow estimates of this parameter to be made with confidence in response to the material properties. Current parameters have been selected based on default parameters derived from limited research, and parametric studies have been carried out to estimate appropriate Λ values.

6.1.2 Output Parameters

• Settlement

Settlement profiles have been provided by the site operators. Settlement data acquired from settlement markers are preferable to survey data, due to the much better accuracy compared to topographic surveys, and as additional waste inputs are easily distinguishable. Topographical survey data have been used to derive waste filling and subsequent settlement, although exact confirmation of timings and filling has proven to be challenging.

• Gas production

Acquisition of gas production data has proved difficult. On some sites these data have not been provided and, where it is available, it is not recorded on an individual cell basis, therefore is difficult to confirm that gas production is from the area of study.


Golder Associates

• Leachate quantity and leachate quality (COD/VFA)

Some data have been provided by landfill site operators, however, specifying the area of interest from the general site information has been difficult; hence it has not been possible reliably to compare the data derived from the supplied records to the modelled predictions.

• Mass degraded

This information has not been tested nor is it currently available from landfill operators. It could be obtained from physical sampling and testing of the decomposed waste mass but this would require a large number of samples to tackle the waste heterogeneity. Decomposed material recovered during the drilling of gas or leachate wells could be sampled and analysed for this parameter, and is an area for further research.

6.2 Data Acquisition

6.2.1 Solid Degradable Fraction

To derive the solid degradable fraction, a generic degradable fraction for MSW was derived from average UK waste percentage composition percentage by Langer (2005), see Figure 15.

Figure 15: Overview of Varying UK Waste Composition after Langer (2005) The mean UK municipal solid waste composition and variability is shown in Figure 16.


Golder Associates

Figure 16: Mean UK Waste Composition with Average Deviation, after Langer (2005) The degradability by component presented by Fricke et al. (1999) considered the material loss during biodegradation of individual waste components. The biodegradation material loss has been applied to the categories of waste presented by Langer (2005) as shown in Table 7.

Table 7: Derivation of Solid Degradable Content for MSW

Component Percentage in UK MSW after Langer (2005)

Percentage material with biodegradable content

Percentage biodegradability, after Fricke et al. (1999).

Actual degradable % of total

Paper / card 31.2 31.2 76 23.7 Plastics 8.8 1.0* 11 1.0 Organics 31.7 31.7 76 24.1 Glass 8.0 0 0 0 Metals 7.2 0 0 0 Inorganic 9.6 0 0 0 Hazardous wastes

0.2 0 0 0

Other 3.2 0 0 0 % of total 64% 49% *Assumed Value

Waste composition analyses have been carried out by others (National Assembly for Wales 2002; Environment Agency Wales, 2007 ) and the biodegradable content of MSW of ~61% have been derived. The inputs into the HBM model should take into account the amount of material expected to degrade hence this may be lower than the total material with biodegradable content. A total actual biodegradable percentage of 49% was developed for MSW for this project as an estimated biodegradable mass loss. This is not considered a


Golder Associates

conclusive figure and estimation of biodegradation percentages of MSW and other wastes is beyond the scope of this project. For each site, site-specific data should be employed.

Changing waste types as a result of legislation and other factors will affect the relevance of the historical data presented above. It should be noted that current biodegradable waste reduction targets only apply to MSW while the non-MSW stream still has a large biodegradable content.

In the case of modelling the UK, Viridor #1 Landfill, where there is a 73.5:26.5 ratio of MSW: Inert waste, the above 49% factor for MSW has been applied to derive an overall solid degradable fraction of 36%.

For the other sites, modelled percentage degradable fraction has been derived from the waste classifications provided. These classifications have been carried out in accordance with the Consolidated European Waste Catalogue. A degradability of 49%, derived as shown in Table 7 has been applied to MSW. Other components have been assigned percentages in accordance with their relative organic contents.

The solid degradable fractions derived for other sites are listed in Table 8.

Table 8: Summary of Solid Degradable Fraction by Landfill

Site Solid Degradable Fraction of Waste UK, Viridor #1 0.36 UK, SITA #3 0.46 UK, SITA #4 0.49 UK, SITA #2 0.45

NENT 0.55 A waste classification system could be adopted to include a digestibility and solid degradable fraction for use in such analyses allowing consistency in the derivation of the model input parameters.

6.2.2 Waste Unit Weight

In several cases, the unit weight of the waste as placed was provided by the landfill operators. There was a significant range in the values reported. The unit weight should be explicitly stated in the model as the dry unit weight as placed. Bulk unit weights must be corrected for moisture content to give a dry unit weight. This was carried out using assumed literature values for the moisture content as the site specific moisture content of the waste was not available.

Current efforts to increase recycling of inert wastes has led to reductions in the unit weight of waste deposited at landfills, limiting the applicability of published waste disposal figures.


Golder Associates

6.2.3 Settlement Data

Settlement has been recorded using two methods:

• Settlement monitoring points (usually placed after landfill capping); and • Interpolation from topographical surveys. Data from settlement monitoring points are more accurate than topographical survey data. The data must be accompanied by the times of filling and capping, and installation of settlement monitoring points in order to make a comparison between the modelled and measured settlements. Where information on the timing of settlement monitoring installation relative to filling and capping is not available, the relative portion of the predicted curve cannot be compared to the measured settlement curves.

The data acquired from topographical surveys are less accurate than where settlement monitoring points are used, firstly due to the infrequent time frame at which surveys are available (usually every 3 or 6 months), differences in the surveying process (e.g. the density of measurement positions) and in the digital terrain model derivation, and also due to assumptions made when inferring level heights from contoured plots. The surveys do tend to yield information on the timing of filling, although subsequent filling events can be difficult to identify, and the magnitude of filling events are challenging to interpret as settlement of underlying material also occurs.

6.3 Verification of Data

The assembled data sets were analysed by an independent assessor to comment of the data trends, anomalies and how the data should be used in the model validation. It was suggested that a single site should initially be considered in greater detail to assess the model performance. UK, Viridor #1 Landfill was selected for this purpose, as the short but detailed settlement records would allow a short-term estimation of model performance to be made prior to attempting long-term predictions. The results of this validation are discussed in Section 8.0.

It was noted in the data verification process that the level of detail in the recorded data was variable. A data collection and recording framework would be beneficial to standardise the data recording. This has been discussed in Section10.2.


Golder Associates

7.0 USE OF THE HBM MODEL IN SETTLEMENT ANALYSIS

7.1 Graphical User Interface (GUI)

The Graphical User Interface (GUI) is a Windows© based environment in which to prepare, run, and interrogate simulations of landfill behaviour made using the Hydro-Bio-Mechanical (HBM) model or engine. The GUI comprises two main parts: the Pre-processor and the Postprocessor: An example of the GUI including screenshots with commentary is given in Appendix 2.

• Pre-processor: The pre-processor enables all the input data files to be prepared and written; and

• Postprocessor: The post-processor provides for the graphical interpretation of output

data files. The main engine is written in Fortran 95. It is the passing of data from the pre-processor to the main engine and then to the post-processor that dictates the nature of data file management. A schematic view of the data flows between the GUI and main engine is shown in Figure 17.

The first objective in the development of the GUI has been to facilitate input data file preparation and output data interrogation. Neither data formatting nor verification checking is implemented in the GUI.


Golder Associates

Figure 17: Data Flow and Management between HBM Model and GUI. McDougall (2005) 7.1.1 User Modelling Experience of the GUI

This section details the user experience in the application of the graphical user interface of the HBM model.

The graphical user interface presents a clear framework for parameter input. The logical sequence of parameter input is consistent with requirements of the model, allowing the user to step through the tabs one by one during the filling sequence.

7.1.1.1 General Input Console

The model must be set up and run on two separate occasions for each modelling simulation. The first run defines the initial hydraulic conditions of the model then secondly, the model can be run for the “transient” analysis, where the full HBM settlement simulation is run over the prescribed number of days. A drop down menu allows the user to select initial or transient analyses. This two staged modelling process requires an understanding of the modelling process and is not initially intuitive, however, it is a simple procedure to implement.


Golder Associates

The timestep and number of days is also inputted on the general input tab of the pre-processor. It is suggested by the authors that greater guidance on the critical timestep is given in the user manual, as there is a temptation to increase the timestep in order to reduce solution time.

7.1.1.2 Grid Definition and Boundary Conditions

The pre-processor allows an inexperienced user easy access to the code’s functions.

The grid is defined by specifying the grid dimensions as the corner points of the model geometry. Any change to the co-ordinates results in all boundary conditions requiring to be re-input, so it is suggested that care be taken when specifying the initial geometry.

The mechanical boundary conditions are assigned to nodes at the corner of the modelling elements by flags, whereby 0 = no fixity, 1 = fixity in the “x” direction and 2 = fixity in the “y” direction.

The hydraulic boundary conditions present the most complex input screen as the hydraulic boundary, position, and magnitude must be specified. It is recommended that a good understanding of the physical meaning of the boundary conditions is gained before the code is operated. The user manual gives a concise description on the use of Permanent Dirichlet and Neumann boundary conditions. It would be beneficial to add discussion on how to apply these codes to initial, transient and infiltration stages of modelling in the user manual.

7.1.1.3 Material Properties

The material properties input screen initially only allows access to the solid degradable fraction and the unit weight. Advanced access must be selected to allow all other input parameters to be altered. This discourages the user from making changes to the input parameters without consulting the guidance literature, which is considered to be beneficial for the correct application of the model. Care should be taken to ensure consistency between data sets when filling is specified, as sudden changes in material parameters can cause numerical difficulties.

7.1.1.4 Filling Sequences

The filling and infiltration boundary condition can be used in conjunction with the specified material types to identify a filling sequence. It should be noted that a large value should be specified for the co-ordinates of the last line on the filling sequence, and a high day number should be specified on the filling and infiltration sequence to avoid any “end of file” numerical problem. Whilst this did not pose any difficulties, it is suggested that this could be automatically generated by the code.


Golder Associates

The reported filling sequences frequently had stages of filling, separated by periods of settlement. It was not initially clear how to define the periods of settlement using the GUI. Definition of filling in areas outside the modelled grid allowed these periods of settlement to occur. It is strongly recommended that the GUI be revised to include a simple function of periods of “non-filling” during the filling sequence. Whilst the code could still be programmed to operate by filling areas outside the modelling mesh, the user interface change would allow the user more accessible ways to represent a staged filling sequence.

7.1.1.5 Infiltration

Defining an infiltration sequence is relatively simple in the GUI, however, difficulties have been experienced when assigning large numbers (>30) of infiltration events. Although there are 100 lines for infiltration events available in the GUI, only the first 30 of these are written to the data file, thus some infiltration events maybe missed and “end of file” errors may occur, as the final value (usually prescribed at 10,000 days) is omitted. It is suggested that, if possible, the process of writing the information to the data file be amended to allow 100 lines to be included. However, if this is not numerically possible, the number of available lines in the GUI should be reduced to 30.

7.1.1.6 Graphical Outputs

The graphical output section of the pre-processor defines what information will be available in the postprocessor. It is recommended that a range of timesteps be defined throughout the model cycle, so the evolution of the model can be assessed to ensure that the observed patterns are sensible. When defining the output points, care should be taken not to confuse node and element numbers, which can be checked in the geometry panel. This area of the input was often overlooked when running the model and changes in the number of timesteps and the geometry limit the use of the postprocessor if the graphical output pane is not also updated.

7.1.1.7 Numerical Controls

Numerical solutions in the HBM model are achieved using a Gauss-Seidel iterative solver. The numerical controls tab contains the numerical time stepping factors. These have not been altered as part of this investigation.

7.1.1.8 Postprocessor

The postprocessor allows graphical representation of all model outputs, enabling detailed interrogation of the model to be carried out. Whilst this function was useful for initial assessment of the model behaviour, extracting raw data from the output data files was found to be preferable for interrogation of model behaviour, although this will be at the preference of the user.


Golder Associates

Where users are new to the HBM model, the graphical postprocessor allows the user to appreciate the behaviour of different parts of the model, in particular the settlement, volatile fatty acid and methanogenic biomass concentrations, and the evolution of these with time, and allows checks to be carried out on model behaviour.

7.1.2 Discussion on Graphical User Interface

As is typically the case with numerical modelling codes, the presence of a graphical user interface allows code operators greater access to complex functions of the model which can lead to overconfidence in the model behaviour. It does, however, allow the user to spend time understanding the theory of the code rather than having to learn how to compose input data files. It is likely that experienced code operators will modify the code using both the graphical interface and direct data file access.

Experience in this project has shown that the graphical user interface makes the HBM modelling code accessible for the inexperienced user, and basic modelling techniques can be acquired within a single day of operation.

7.2 Application of the Model

7.2.1 Simple Predictive Modelling

Simple predictive modelling can be carried out using known and best estimate input parameters. This should be supplemented by sensitivity analyses concentrating on the most critical parameters and those where the greatest uncertainty exists.

7.2.2 Curve Matching Techniques

The HBM model may be used as a simple predictive tool based solely on the estimated input parameters. For additional accuracy the model can also matched to initial settlement profiles and extrapolated, based on as many known parameter values as possible. The values of the unknown parameters can be varied to adjust the computed settlement to the known settlement profile and other measured outputs (such as gas/leachate production) to fit the measured profiles. Future settlement and other outputs can be predicted from the end of the curve matching with greater confidence than for a simple predictive model.

7.3 Modelling Discussion

7.3.1 Phased Filling

The phased filling option can be selected by specifying material numbers 4 to 8 to the elements within the modelling mesh. An additional material property must be specified (with the material property number 9). The parameters of material number 9 contain the same properties as the degradable model, except for the following parameters:


Golder Associates

• A maximum hydrolysis rate = 0; and • A hydraulic conductivity = 1 x 10-15 m.s-1. Where model filling is specified by the presence of user defined materials identifying numbers 4 to 8, material property number 9 is automatically specified and gravity is set to zero for all of the elements at the start of the modelling process. As the specified filling sequence occurs, the elements are systematically revealed in accordance with the specified filling sequence. As the elements are revealed, gravitational forces are applied to the element and the material property number changes from 9 to the specified number (between 4 and 8 inclusive).

When the filling stage is included, initial compression settlement due to material self weight, which is otherwise omitted (see Figure 18). It is apparent that the compression under self weight is only considered when the filling stage of the model is selected as part of the analysis. When the material is “wished into place” in a single instantaneous lift, the compression under self weight is omitted from the model, and the resultant material densities are lower than where filling is specified.

The primary purpose of the phased filling sequence is to define the model conditions in terms of stress conditions and biodegradation state, at the start of the “settlement phase” (see Figure 18)

time

Settlement

time

Settlement b) With filling

Filling phase

Post filling has no

primary compression,

even where filling is

not specified Settlement phase

a) Without filling

Figure 18: Modelling Landfill Settlement a) without Phased Filling b) with Phased Filling


Golder Associates

The filling sequence will occur as a series of steps as the finite element modelling grids are disclosed during the phased construction. This can be envisaged as an imaginary line rising up the prescribed part of the model, at a prescribed rate.

Filling in the present version HBM model is a continuous process. Whilst the rate of filling can be altered between different filling phases, breaks in the filling sequence could not initially be defined to allow settlement to occur prior to another lift being placed. Following greater experience at using the model, techniques to address this have been identified (see Section 12.3).

During the phased filling process, the initial modelling mesh is defined at the start of modelling. As the lower layers in the model are activated, settlement occurs in these lower layers. This settlement also influences the recorded settlement in upper layers, even though they are yet to become active.

Although elements above the current filling line have not yet been disclosed, they still exist within the model and hence settlement in underlying layers affects the position of overlying elements. This is identified by parallel settlement traces for the upper layers of the model, recording identical settlement traces. This is a function of the finite element formulation, and whilst it is theoretically possible to fill to a specified height, this would require re-formation of the mesh and associated parameter and boundary acquisition at each stage to account for the change in the material properties.

This does represent ‘real world’ behaviour as the position of the next lift of waste is altered by settlement of the previous lift, however, it does make filling to a known height challenging as the height of the initial modelling mesh must be overestimated to account for the settlements which occur during the filling process. This can be a time consuming iterative process as every time the geometry is altered in the pre-processor GUI, the boundary conditions must be re-specified.

In many situations, particularly when constructing the lining system in lifts of known height (typical practice for steep sided landfill lining systems, Fowmes et al., 2007), the waste lifts are overfilled at each lift to account for the settlement of previous layers. An example of this is if 0.5 m of settlement has occurred in the first 3 m waste lift, 3.5 m will be placed in the second lift to account for this. In this example, the HBM model would specify another lift of 3 m, and require an additional lift at the end of the modelling to make up for the height loss.

7.3.2 Numerical Difficulties

Difficulties were experienced in modelling the phased construction sequence where a model solution could not be reached. The model stopped running during the model cycling process, sometimes closing itself down, other times freezing prior to the full number of runs being achieved. This was found to be due to an inconsistency in the Van Genuchten values between the model 4 (active elements) and model 9 (not yet activated elements) parameters. This caused a numerical incompatibility as the element became active (hence model cycling


Golder Associates

ceased). To avoid numerical problems, parameters should be consistent between the inactive cells and the reactivated cells (model parameter sets 4 and 9 respectively). Numerical difficulties were encountered whereby the model could not find a solution. This manifested itself through the cycling window disappearing, ceasing, or an error message being presented.


Golder Associates

8.0 MODEL VALIDATION

This section contains comparisons between the site-derived settlement data and the output data derived from the HBM model. The structure of this section begins with generic modelling and parametric studies, then validation against site-specific settlement data, with site-specific input parameters and waste geometry.

8.1 Generic Modelling

8.1.1 Simplified Modelling

The generic models represent a 30 m height landfill and was represented using a 2 dimensional modelling mesh measuring 50 m wide. The model has vertical sides which represent roller boundary conditions. This therefore represents a section of the landfill at the centre of the site, unaffected by geometric edge effects.

For the simplified modelling, the filling sequence was omitted, allowing a direct comparison between the influence of modelling parameters. To allow comparison with the all of the site-derived data, the model was run to a total of 9800 days (27 years). This allows the long and short-term influence of modelling variations to be considered.

Figure 19 shows a comparison between the modelled data using default input parameters (run: Generic_001) and the landfill settlement from multiple landfill sites. The settlement has been normalised to unit strain (dimensionless parameter) to allow direct comparisons.

The generic input parameters show a reasonable comparison between the default parameters and the overall settlement profile, falling within the bounds identified by the site-derived data set. It is evident from the dataset that there is significant variability associated with the reported settlement data.

The UK, SITA #4 dataset represents an outlier with very large early settlement. These data are taken from surveys and may be erroneous.


Golder Associates

0.00

00

0.05

00

0.10

00

0.15

00

0.20

00

0.25

00

020

0040

0060

0080

0010

000

Tim

e [d

ays]

Strain (-)

0.30

00

1200

0

USA

DSW

1.1

USA

DSW

1.5

UK B

RE

3.3

UK B

RE

5.3

USA

DSW

1.4

Figure 19: Modelling Results as a Comparison to the Predicted Behaviour

Hong

Kon

g 1.

1Ho

ng K

ong

1.2

UK S

ita 3

UK B

RE

3.1

UK B

RE

3.2

UK B

RE

3.4

UK B

RE

4.2

UK B

RE

4.1

UK B

RE

5.1

UK B

RE

5.2

UK S

ita 2

UK S

ita 1

.2UK

Sita

1.3

UK S

ita 1

.4UK

Sita

1.5

Mod

el D

efau

ltHo

ng K

ong

1.3

USA

DSW

1.3

UK V

irido

r 1.2

Aust

ralia

1.3

UK V

irido

r 1.1

Aust

ralia

1.1

USA

DSW

1.2

Aust

ralia

1.1


Golder Associates

8.1.2 Parametric Study

A parametric study has been carried out to assess the influence of input parameters on the long-term settlements. The parameters used in the parametric investigation are shown in Table 9 with Figure 20 showing the influence of the parametric changes described in Table 9. Model Generic_001 uses model default parameters. For ease of comparison, in all runs 001 to 006 the material filling stage has been omitted, hence degradation of all layers begins at the same time. In runs 008 and 009 a filling sequence of 3000 and 300 days respectively is included.

Table 9: Generic Model Input Parameters

Run number

Decomposition-induced Void

Change

Solid Degradable

Fraction Digestibility

Maximum Hydrolysis

Rate

Filling Days

Generic_001 (model default)

-0.65 0.4 0.7 2500 0

Generic_002 -0.3 0.4 0.7 2500 0 Generic_003 -0.65 0.6 0.7 2500 0 Generic_004 -0.65 0.4 0.1 2500 0 Generic_005 -0.65 0.4 1.5 2500 0 Generic_006 -0.65 0.6 0.7 5000 0 Generic_008 -0.65 0.4 0.7 2500 3000 Generic_009 -0.65 0.4 0.7 2500 300 A change in the decomposition induced void change parameter (Λ) is shown to have a large effect on the settlement generated by the model. This void change parameter is directly proportional to the settlement which occurs as a result of biodegradation which represents most of the post-filling settlement in a landfill. This parameter also has the most uncertainty, as there is little site-derived research available on the values that should be selected for this parameter, and this is considered to be an important area of further research.


Golder Associates

Figure 20: Parametric Study using the Generic Model (Long-Term View)

0.00

00

0.05

00

0.10

00

0.15

00

0.20

00

020

0040

0060

0080

00

Tim

e [d

ays]

Strain (-)

0.25

00

1000

012

000

Gen

eric

006

Gen

eric

005

Gen

eric

004

Gen

eric

003

Gen

eric

002

mod

el d

efau

lt


Golder Associates

Run Generic_003 shows an increase in settlement with increased solid degradable fraction. However, when looking at the short-term view of the same data (shown in Figure 21), the solid degradable fraction only has an effect after 400 days. When the maximum hydrolysis rate is doubled (run Generic_006), the solid degradable fraction increase is shown to have an effect on settlement from the start of the settlement profile, indicating that this is the controlling factor over the initial part of the settlement profile.

Runs Generic_004 and Generic_005 show the influence of digestibility. A reduction in digestibility to 0.07 shows a significant decrease in the long-term recorded settlement, however, it is clear from the settlement trace that settlement is still active at 9800 days. An increase in the digestibility to 1.5 results in a more rapid initial settlement (shown in Figure 21), however, the ultimate settlement achieved is the same as in the default case (Generic_001), and this parameter hence controls the time to completion but not the total settlement magnitude.

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

0.00 200.00 400.00 600.00 800.00 1000.00 1200.00Time [days]

Stra

in (-

)

model default

Generic 002

Generic 003

Generic 004

Generic 005

Generic 006

Figure 21: Parametric Study (Short-Term View)


Golder Associates

8.1.3 Generic Model including Filling Sequence

The default model has been run for a 300 day and 3000 day filling sequences, representing filling rates of 0.01 m.day-1 (run Generic_008) and 0.1 m.day-1 (run Generic_009) respectively. The results of this analysis are shown in Figure 22.

0.00

00

0.05

00

0.10

00

0.15

00

0.20

00

0.25

00

0.30

00

020

0040

0060

0080

0010

000

Tim

e [d

ays]

Strain (-)

1200

0

3000

day

s fill

ing

San

dtow

n 1.

1S

andt

own

1.5

UK

BR

E 3

.3U

K B

RE

5.3

Sand

tow

n 1.

4

Figure 22: Including Filling Sequence (Long-Term, Post-Filling))

NE

NT 1

.1N

ENT

1.2

UK S

ita 3

UK

BR

E 3

.1UK

BR

E 3

.2U

K B

RE

3.4

UK

BR

E 4

.2UK

BR

E 4

.1U

K B

RE

5.1

UK B

RE

5.2

UK

Sita

2U

K S

ita 1

.2UK

Sita

1.3

UK

Sita

1.4

UK S

ita 1

.5

Mod

el D

efau

ltS

andt

own

1.3

NEN

T 1.

3UK

Viri

dor 1

.2Ly

ndhu

rst 1

.3U

K Vi

ridor

1.1

Lynd

hurs

t 1.2

300

days

fillin

g

San

dtow

n 1.

2Ly

ndhu

rst 1

.1


Golder Associates

Figure 22 shows the post-filling settlements over 9800 days, whilst Figure 23 shows the same analytical post filling settlement data presented for the first 1200 days. It can be seen that with 3000 days filling, the post-filling settlement is significantly reduced as degradation, particularly in the lower layers of waste in the model, has slowed due to limited availability of solid degradable material. Figure 23 shows that in the short-term, the model with 300 days filling has greater initial settlement, as degradation processes are more active at the start of monitoring. In the long-term there is no discernable difference in the predicted settlement where zero filling days and 300 filling days are selected.

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045

0.05

0.00 200.00 400.00 600.00 800.00 1000.00 1200.00

Time (post filling, days)

Stra

in (-

)

model default (zero days Filling)

With filling (3000 days)

With filling (300 days)

Figure 23 Generic model with filling sequence (short term, post filling) 8.2 Site-Specific Predictions vs. HBM Predictions

In this section, data from specific sites have been compared to HBM model predictions. Input geometry and parameters have, where possible, been derived from the data provided by site operators.

The following sites have been selected for comparisons between the HBM model and the site-derived settlement data:

• UK, Viridor #1; • NENT ; • UK, SITA #3; • UK, SITA #2; and • UK, BRE #4 . These sites have been selected based on the availability of settlement data and accompanying information on waste composition and filling sequencing.


Golder Associates

These sites also provided a range of waste ages and compositions. UK, Viridor #1 Landfill provided the most complete information about filling and the shortest time between filling and the onset of settlement monitoring. The waste stream was assigned a lower solid degradable fraction due to the increased inert content. UK, BRE #4 provided a broad range of settlement data from the different ages and settlement magnitudes. UK, SITA #3 and UK, SITA #2 settlement data were derived from settlement profiles and contain irregular filling sequences.

8.2.1 UK, Viridor #1

Modelling of UK, Viridor #1 Landfill has been carried out assuming a 30 m high waste mass, with a 73.5:26.5 MSW:commercial waste ratio, the latter waste type being assigned a solid degradable fraction of 36%. A series of settlement monitoring points were monitored over a period of 217 days. Filling of the site had been carried out between 2002 and 2005 and capping had been carried out between 2003 and 2005. Point J7 is located towards the edge of the site, and this may account for the low settlement recorded. Also point J7 was in an area capped in June 2003, and there may have been a significant time lapse from the capping of this area to the start of monitoring. Points J8 and J9 were capped in April 2004; J6, J5 and J4 in September 2004; and J1, J2 and J3 in May 2005, although it is not clear when these areas were filled, or the time from completion of filling to capping, or the settlement that had already occurred prior to monitoring. Whilst the settlement plots shown in Figure 24 are normalised to zero days and zero settlement, some time will have elapsed from the end of filling and some settlement will have occurred, and it is likely the actual time zero curves will have greater initial settlement than the curves shown.

0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0 50 100 150 200 250 300

Time [days]

Stra

in (-

)

J1

J2

J3

J4

J5

J6

J7

J8

J9

modelled - zero filling days

Figure 24: Modelled vs. Measured Settlements for UK, Viridor #1 Landfill


Golder Associates

8.2.1.1 Including Filling Phases

Figure 25 shows the influence of inclusion of the filling sequence on post-filling settlement. The zero days filling and 150 days filling show a delay as the biodegradation process begins. Settlement rate is at a maximum after approximately 1 year, with the rate and magnitude decreasing as additional filling days are included. It is evident from Figure 24 and Figure 25 that the short-term settlements from the landfill are better represented if the initial filling sequence is included. This section carried out a parametric study on the duration of the filling phase. The boundary of the filling phase has been taken from 150 days to approximately 4 years, which is considered to be an appropriate range to consider based on-site-derived data.

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

0 50 100 150 200 250 300

Time [days]

Stra

in (-

)

Zero filling days

150 filling days

300 filling days

600 filling days

900 filling days

1200 filling days

1500 filling days

Figure 25: UK, Viridor #1 Modelling, Including Filling Sequence of 0 to 1500 days


Golder Associates

Figure 26 shows the effect of solid degradable fraction at a variety of levels throughout the landfill. It can be seen that when filling reaches the upper level (signified by degradation in the upper layer) the solid degradable fraction in the lowest layer has degraded to near zero. This results in a slowing overall settlement rate post-filling as the solid degradable material availability becomes the limiting factor on the degradability in the lower layers. Figure 27 and Figure 28 show the VFA and methanogenic biomass for the 900 filling days modelling predictions. The VFA shows a delay for each zone as it is disclosed and a small drop (or decrease in the rate of accumulation in the increase in VFA) where a subsequent lift is disclosed.

Figure 26: UK, Viridor #1 Landfill: Solid Degradable Fraction (900 Filling Days)


Golder Associates

Figure 27: UK, Viridor #1 Landfill: VFA (900 Filling Days) Figure 28: UK, Viridor #1 Landfill: Methanogenic Biomass (900 filling days) The solid degradable fraction for 900, 1200 and 1500 days filling is compared to the measured data in Figure 29. These show a good correlation with the measured profiles

300


Golder Associates

compared to where the filling sequence is omitted, however, as discussed previously, the steep initial plots may be appropriate if the settlement between capping and installation of settlement markers had been included.

0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014

0 50 100 150 200 250 300

Time [days]

Stra

in (-

)

J1

J2

J3

J4

J5

J6

J7

J8

J9

900 filling days

1200 filling days

1500 filling days

Figure 29: UK, Viridor #1 Measured Settlement and Modelled Data, Including Filling Of 900 To 1500 Days


Golder Associates

8.2.1.2 Parametric Analysis

To consider the influence of the parametric analysis, the zero filling day’s case has been taken as the base condition. This allows easier comparison between the influence of the varied parameters and the measured settlement.

Table 10: UK, Viridor #1 Parametric Variations

Hydraulic Conductivity

Solid Degradable Fraction

Dry unit weight Digestibility

(m.s-1) (-) (kN.m3) (-) UK, Viridor #1_001 5 x 10-5 0.36 5 0.7 UK, Viridor #1_002 5 x 10-4 0.36 5 0.7 UK, Viridor #1_003 5 x 10-6 0.36 5 0.7 UK, Viridor #1_004 5 x 10-5 0.2 5 0.7 UK, Viridor #1_005 5 x 10-5 0.5 5 0.7 UK, Viridor #1_006 5 x 10-5 0.36 4 0.7 UK, Viridor #1_007 5 x 10-5 0.36 6 0.7 UK, Viridor #1_008 5 x 10-5 0.36 5 0.07 UK, Viridor #1_008 5 x 10-5 0.36 5 1.5 Hydraulic Conductivity

The hydraulic conductivity was varied as shown in Table 10. Figure 30 shows that neither increasing nor decreasing the hydraulic conductivity at the site had a discernable effect on the observed settlement. This is believed to be due to the hydrostatic initial moisture conditions and lack of infiltration limiting fluid migration in the model, and hence this was not a controlling parameter.


Golder Associates

0

0.005

0.01

0.015

0.02

0.025

0.03

0 50 100 150 200 250 300 350 400 450

Time [days]St

rain

(-) 001

002003

NB 3 lines are overlaid

Figure 30: UK, Viridor #1 Parametric Study, Hydraulic Conductivity Solid Degradable Fraction

The solid degradable fraction was varied as given in Table 10 and the results are shown in Figure 31. The solid degradable fraction decreased from 0.36 (default – UK, Viridor #1_001) to 0.2 (UK, Viridor #1_004) reducing the settlement magnitude at 400 days from 0.754 m to 0.656 m.

This is due to the solid degradable material beginning to run out and thus become a limiting factor on the degradation rate. The increase from 0.36 to 0.5 solid degradable fractions causes a small increase in the settlement at 400 days, however, the curves indicate that this is not a controlling factor below 300 days. Figure 21 confirms the fact that an increased solid degradable fraction only has significant effect on the settlement after approximately 1 year. Prior to this the limiting factor on settlement magnitude is the onset of biodegradation reactions, and not the availability of material to degrade. When a filling sequence of greater than 1 year is included, this “lag” time in the onset of degradation is not observed and the influence of higher solid degradable fractions occurs earlier in the settlement trace.


Golder Associates

0

0.005

0.01

0.015

0.02

0.025

0.03

0 50 100 150 200 250 300 350 400 450

Time [days]

Stra

in (-

) 001004005

Figure 31: UK, Viridor #1 Parametric Study, Solid Degradable Fraction Dry Unit Weight

The dry unit weight was varied as shown in Table 10. Whilst the filling sequence, and hence compression under self weight, is omitted from this model, the dry unit weight still influences the creep compression component of settlement. The results of the parametric investigation are shown in Figure 32. A decrease in settlement magnitude at 400 days of 7% occurs with a drop in unit weight from 5 kN.m-3 to 4 kN.m-3. An increase in unit weight from 5 kN.m-3 to 6 kN.m-3results in a 4% relative increase in settlement at 400 days.


Golder Associates

0

0.005

0.01

0.015

0.02

0.025

0.03

0 50 100 150 200 250 300 350 400 450

Time [days]St

rain

(-) 001

006007

Figure 32: UK, Viridor #1 Parametric Study, Dry Unit Weight Digestibility

The digestibility was varied as shown in Table 10. The default 0.7 is compared in Figure 33 to values of 1.5 and 0.07. The lower digestibility value of 0.07 represents material which takes longer to degrade. It does not represent a lower degradable fraction, hence, the smaller settlement values at a given time will be accompanied by a longer time to completion.


Golder Associates

0

0.005

0.01

0.015

0.02

0.025

0.03

0 50 100 150 200 250 300 350 400 450

Time [days]

001

Figure 33: UK, Viridor #1 Parametric Study, Digestibility 8.2.1.3 Infiltration

Infiltration during filling has been assessed by running the 150 day filling model with infiltration events of 15 or 30 mm, every 10 days during filling. Figure 34 shows the influence of infiltration during the filling sequence. Infiltration is not continued beyond filling as the model is considered to be capped and infiltration becomes negligible. Figure 34 includes the settlement of the upper surface of the model during the filling sequence. It should be noted that this is not an absolute settlement as the material has not yet been filled to the full height; however it provides a useful comparison of the behaviour during this stage of the modelling process.

Stra

in (-

)

008009


Golder Associates

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

0 100 200 300 400 500 600

time (days)st

rain

(-)

150 filling days

150 filling days, with 15 mm every10 days rain

150 filling days, with 30 mm every10 days rain

Figure 34: Influence of Infiltration on Settlement (Plot Includes 150 Days Filling) Interestingly the settlement at 550 days is decreased by the presence of the infiltration events. This is believed to occur by the resulting increase in VFA concentrations caused by the infiltration events. Figure 35 shows the first 1000 days of the model with no infiltration, where as Figure 36 shows the VFA traces where 15 mm of infiltration is added each day during the filling sequence.


Golder Associates

Figure 35: VFA Plot for UK, Viridor #1_205: no Infiltration (First 1000 Days) Figure 36: VFA Plot for UK, Viridor #1_206 30 mm Infiltration Events During Filling (First 1000 Days)


Golder Associates

When considering the long-term post-completion settlement plots, initial biodegradation-induced settlement is inhibited by the high VFA, however, the settlement exceeds the no infiltration model after approximately 1500 days (see Figure 37).

0

0.02

0.04

0.06

0.08

0.1

0.12

0 1000 2000 3000 4000 5000 6000 7000 8000 9000time [days]

stra

in (-

)

205 - 150days filling204 - 150d filling + infilltration (15 mm / 10days)206 - 150d filling + infilltration (30 mm / 10days)

Figure 37: UK, Viridor #1, with Infiltration, Post-Completion Settlement Table 11 shows the remaining solid degradable fraction (SDF) in the uppermost element (representing the highest remaining SDF) after 9800 days. The solid organic fraction at 9800 days is lower where infiltration occurred during filling, indicating that the infiltration does reduce the time to completion.

Table 11: Maximum Solid Organic Fraction at 9800 Days

Run Number Details Solid Degradable Fraction (Kg/m3) 205 No infiltration 3.93 204 15 mm every 10 days during filling 2.58 206 30 mm every 10 days during filling 1.58 This demonstrates the complexity of the degradation process, and that time assumptions can not be made with regards the infiltration and the effect on settlement. The rainfall input over approximately a six month period is shown to increase VFA concentration, delays the degradation, but ultimately results in more rapid degradation of the solid degradable fraction. However, the high VFA concentrations also cause a “lag” in onset of the biodegradation induced settlement. The site-derived data are not considered to be comprehensive enough to calibrate such parameters against, however, this section does highlight the importance of acquiring good rainfall data but, more importantly, the keeping of detailed records of the time


Golder Associates

of filling and capping. The model is also capable of representing leachate inputs which have non-zero VFA and methanogenic biomass values, however, assessment of these aspects of the model is considered to be beyond the scope of the current investigation.

8.2.2 HONG KONG #1 (NENT Landfill)

The NENT data set describes the filling of a landfill cell in two distinct stages. A one-dimensional vertical column comprising 58 elements was used with the default parameter set. The column was dimensioned (by trial and error) so that the final self-weight compressed height coincided with the reported elevation of the settlement stations at the start of the two monitoring phases. Within each filling phase, elements of waste were placed at a constant fill rate; surface displacements were output during the intervening and final long-term settlement phases.

The NENT data are unusual in that settlement readings were made every two weeks. Also provided were waste composition data. Table 12 and Figure 38 show the filling sequence at site NENT.

Table 12: Key Filling Dates for NENT

Event Date Elapsed Time [days]

Start of filling phase 1 July 1995 0 Start of settlement monitoring Sept 1998 1185 End of monitoring/ Start of filling phase 2

Jan 2001 2029

End of filling phase 2/ Start of settlement monitoring

Aug 2002 2592

End of monitoring Feb 2006 3877


Golder Associates

Filling 1

Settle

Settle

Filling 2

t = 0 1185 2029 2592 3877 days

Figure 38: Diagram Showing the Initial Unloaded FE mesh (t=0), Filling Phases and Settlement Modelling Phases for NENT Occasional model output has been extracted and is presented as surface elevations to show the entire filling and settlement sequence in Figure 39. There is no data for the filling phases so a more telling comparison is obtained through comparison of model predictions and field settlement data Figure 40). In this data both settlement phases with the absence of filling data shown as a horizontal line connecting the two displacement curves. There is of course a problem with the definition of the settlement strain when the underlying fill depth is increased in this way.

The first thing to notice from Figure 40 as has already been highlighted, is the generally satisfactory performance of the default data set. Beyond this, the influence of decomposition effects and their control using Λ may be observed. At Λ = -0.2, the model over-predicts settlement at all stages. A reduction of Λ to –0.4 improves the overall fit but does not indicate that any further improvements can be achieved by a simple change in Λ alone because the rate of settlement in the first phase is over predicted, whereas in the later phase, the rate is under-predicted. This type of insight is very valuable for the understanding and interpretation of decomposition mechanics generally and the theoretical development of the HBM model in particular. For example, this output points to the non-uniformity of Λ during the decomposition process, which has already been observed in the case of particle dissolution (McDougall, 2008) but not, as yet, in the case of waste degradation. However, substantive changes to this part of the model are not possible within the scope of this project.


Golder Associates

0

5

10

15

20

25

30

35

40

45

0 1000 2000 3000 4000

Time [days]

Elev

atio

n [m

abo

ve b

ase

liner

]

Figure 39: Outline HBM Model Surface Elevation Predictions for NENT

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0 1000 2000 3000 4000 5000 Time [days]

Measured Data 6001 (Λ=-0.2)6003 (Λ =-0.4)

Strain

(-)

Figure 40: Comparison of HBM Model Output with NENT Data Presented as Surface Displacements Showing Influence of Λ (= -0.2 and -0.4). 8.2.3 UK, SITA #2

UK, SITA #2 has been modelled with a solid degradable fraction of 45% and a dry unit weight of 0.7, and has been considered with zero days, 180 days and 365 days filling prior to post completion settlement (shown in Figure 41). The waste depth is 13.8 m.


Golder Associates

0.000

0.020

0.040

0.060

0.080

0.100

0.120

0.140

0 200 400 600 800 1000 1200 1400 1600

Time [days]st

rain

(-)

Survey

modelled - no filling

modelled 180 daysfillingmodelled 365 daysfilling

Figure 41: Modelled and Measured Settlements at UK, SITA #2 Landfill In this case the settlement profile has been estimated from survey data and is known to include a filling event of at least 0.4 m, which coincides with the reported data on filling. This is believed to represent the placement of restoration soils and the accelerated settlement following this event could be as a result of primary compression, which is not included in the modelled traces shown in Figure 41. The data for this settlement analysis were derived from survey records and it is difficult to define the exact time of filling, filling height and settlement magnitude.

8.2.4 UK, BRE #4

A series of settlement data were acquired for UK, BRE #4 Landfill representing a cross section running from North-East (P1) to South-West (P18) of Pit 4 and west-southwest (P19) to east-northeast (P22 and BH 2000) in Pit 5. The monitoring data have been normalised to the end of filling, so that the data for different locations can be compared on a single plot, see Figure 42.


Golder Associates

Figure 42: UK, BRE #4 Landfill Modelling against HBM Model Predictions

0.00

00

0.05

00

0.10

00

0.15

00

010

0020

0030

0040

0050

0060

0070

0080

00

Tim

e (d

ays)

Strain (-)

0.20

00

0.25

00

mod

elle

d: L

ambd

a =

-0.3

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10

P11

P12

P13

P14

P15

P16

P17

P19

P20

P21

P22

BH(2

000)

mod

elle

d: D

efau

lts


Golder Associates

The settlement traces derived from the UK, BRE #4 Landfill settlement data show significant variability. Points 17 and 18 have significantly steeper settlement profiles than indicated by other points. This could be due to their location on the relatively steep perimeter of the slope, thus the 3-dimensional geometric effects exacerbate the settlement; alternatively, material may have been placed at the edge of the cell later than at the centre, yet they were assigned the same timings as the centre of the cell. Points P19 and P20 represent the geometry at the edge of Pit 5, which is less steep than the edge of Pit 4, this being reflected in the less pronounced steep settlement curve.

No information has been gained pertaining to the waste type and composition. However, the default waste parameters still represent the correct settlement profile shape, whilst the magnitude of the settlement is controlled by the material properties, especially the material degradation induced settlement parameter and solid degradable fraction. Whilst attempts could be made using the model to fit to an individual settlement trace, this would be somewhat academic without additional information on waste types.

The UK, BRE #4 data reportedly have between 1000 and 4000 days filling, however, at 3000 days filling, the modelled values are at the upper bound of the reported modelled data (see Figure 43), indicating the default model input parameters, when used in conjunction with a filling sequence of this magnitude, do not adequately represent the dataset. As suggested previously, a filling sequence in the order of 300 days has little effect on the long-term settlement magnitude. The model has also been run with a filling event of 3000 days, and then a lambda of -0.3 (see Figure 43). This shows a more acceptable correlation with the overall measured data, however, several of the values show significantly more settlement than a lambda of -0.3 and solid degradable fraction of 0.4 suggests.


Golder Associates

With

fillin

g (3

000

days

) + la

mbd

a =

-0.3

Figure 43: Settlement Plots Including Filling

0.00

00

0.05

00

0.10

00

0.15

00

010

0020

0030

0040

0050

0060

0070

0080

00

Tim

e (d

ays)

Strain (-)

0.20

00

0.25

00

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10

P11

P12

P13

P14

P15

P16

P17

P19

mod

elle

d: L

ambd

a =

-0.3

P20

P21

P22

BH(2

000)

mod

elle

d: D

efau

lts

With

fillin

g (3

000

days

)


Golder Associates

8.2.5 UK, SITA #3

UK, SITA #3 is a steep sided landfill site, modelled with a mesh 40 m height and 50 m wide. The waste characterisation has estimated a solid degradable fraction of approximately 46%.

A period of settlement monitoring from November 2002 to June 2003 has been carried out by survey. The site has been described by operators as being capped quickly and filled quickly.

A basic model was run for UK, SITA #3 assuming a full filling height from the beginning of the model with zero days filling (shown in Figure 44). The model has also been run with a reported fast filling sequence of 300 days, and a maximum 1500 days filling derived from surveys of the cell area.

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045

0.05

0 100 200 300 400 500 600 700

Time [days]

Settl

emen

t (-)

zero days filling

Settlement data

1500 days filling

300 days filling

Figure 44: UK, SITA #3 Post Filling Settlement The modelled output shows good correlation with the measured data from the settlement traces, particularly when the 300 day infilling period is considered prior to settlement occurring. The actual filling sequence contains a series of pauses and filling events.

8.2.5.1 UK, SITA #3 Model – Full Filling Sequence

The actual filling sequence at UK, SITA #3 consisted of (at least) three distinct filling phases. The estimated filling sequence from survey data is shown is summarised in Table 13.


Golder Associates

Table 13: UK, SITA #3 Filling Sequence

Time Filling Stage Filling Rate 0 – 270 days Filling 0 -27 m 0.1 m.day-1 270 – 940 days Settlement phase 0 940 – 1000 days Filling 27 - 33 m 0.1 m.day-1 1000 – 1720 days Settlement phase 0 1720 – 1790 days Filling 33 – 40 m 0.1 m.day-1 1790 onwards Settlement phase 0

Assumed rates Heights from survey

Figure 45, shows the settlement for the Λ = -0.3 model with the three settlement events visible on the upper line (representing the upper model surface). It can also be seen that the settlement at the end of the 3rd filling stage give the corrected final filling height of 40 m. Figure 46 shows the post-filling (measured and modelled) settlement profiles when considering the full filling sequence. The stages of non-filling are prescribed in the model by identifying a filling sequence event to an area located outside of the limits of the modelling grid. This technique allows the duration of the settlement event to occur over a known time period, without causing additional filling to occur within the model. The model has been run with the default decomposition induced void change parameter of -0.65, as well as with a value of -0.3, in line with the parameters suggested for the analysis for UK, BRE #4 Landfill. The second of the two models (run UK, SITA #3_005) was initially run with a 40 m height but, due to the increased settlement, settlement of 7.5 m had occurred before the final filling, and hence no filling occurred in the final stage. To account for this, the model height was increased to 48 m prior to any settlement, which resulted in a final filling height of 40 m following the filling sequence.

filling events

Figure 45: Settlement Profiles for UK, SITA #3 with 3 Filling Events


Golder Associates

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045

0.05

0 100 200 300 400 500 600 700

Time [days]

Settl

emen

t (-)

Settlement data

3 stage filling

3 stage filling sequence, total 40 m, lambda= -0.3

Figure 46: UK, SITA #3 Post-Filling Settlement Including Full Filling Sequence Figure 46 shows a good correlation between the modelled and measured settlement data when a value of -0.3 is used for the decomposition-induced void change parameter. Figure 43 and Figure 46 show the importance of specifying the correct filling sequence.

This modelling exercise has shown the importance of over-estimating the initial model height when a complex filling sequence is required. Previously, the filling had been assumed to be to the correct height and corrected through normalisation of the settlement to percentage strain, which can be directly compared regardless of ultimate filling height. However, where distinct filling events occur, height must be available for the filling events to occur, otherwise final filling events can be too small or totally omitted as occurred in run UK, SITA #3_005.

8.3 Summary of Site-Specific Validation

The settlement model has been shown to be capable of reproducing the settlement patterns recorded and measured from site-specific settlement monitoring and survey data.

There is, in most cases, limited information surrounding the timing of filling, capping, and monitoring. The model is sensitive to the representation of the filling sequence. Whilst parametric variations are best assessed using a zero filling time for simplicity, for predicting landfill settlement and time to completion, it is imperative that the filling sequence is taken into account. This is discussed in greater detail in Section 8.3.3.

The generic input parameters have been used in most cases, only the geometry, filling time and solid degradable fraction have been altered, yet the model gives a response which is still representative of the actual data. A more detailed suite of site-specific input parameters and filling sequence would allow a fuller validation of the model predictions.


Golder Associates

8.3.1 Decomposition Induced Void Change Parameter

The decomposition induced void change parameter is a key controlling factor in the predicted settlement by the HBM model. Indicative results for UK, BRE #4 , NENT , and UK, SITA #3 indicate that a value of -0.3 may be more appropriate than the suggested default input parameter of -0.65. Definition of other controlling input parameters, particularly the filling sequence is required to accurately estimate Λ. This is demonstrated for the UK, SITA #3 dataset where the measured data could be matched with Λ = -0.65 and 300 days filling or Λ = -0.3 and 1500 days, 3 stage filling sequence.

This is discussed as an area of further research in Section 12.0.

8.3.2 Parametric Discussion

Parametric variations in several input parameters have been addressed in this report. Time dependency of parameter effects is important. When carrying out parametric studies using the model, a parameter may not have a short-term effect; for example, solid degradable fraction (e.g. see Figure 21) as the influence of this parameter is limited by the biodegradation reaction rate, and the short-term and long-term influence of the parameters may not be proportional.

The hydraulic conductivity was shown to have very little effect, however, a hydrostatic profile was defined, with no filling sequence and no infiltration, hence there is unlikely to be any significant fluid migration through the waste mass.

The digestibility is shown to have significant influence on the time to completion but ultimately will have no effect of the total degradation. Relating the digestibility to the actual waste stream is very important in order to be able to fully validate the model behaviour. In the short-term, guidance on acceptable digestibility parameters for differing waste streams should be provided as part of the HBM guidance documentation.

8.3.3 Influence of including Filling Sequence

Inclusion of the filling sequence, will generally reduce the predicted time to completion as degradation has occurred before the filling is complete. When looking at the settlement trace without including the filling sequence, there is an initial increase in the settlement rate as a result of settlement as degradation begins and the VFAs are reduced. The maximum settlement rate is followed by a slowing rate as degradation becomes limited by the availability of solid degradable fraction in the waste.

When the filling sequence is included, each individual element may be at any given stage in the acceleration, or deceleration of the settlement rate. This alters the overall settlement profile, depending on the time elapsed prior to filling.


Golder Associates

Using the default parameters, the maximum settlement rate is reached after approximately 1 year. Where the filling sequence is greater than 1 year, lower post-filling settlements are predicted at a given time.

8.4 Issues Arising in Validating

The lack of specific information on the waste makes validation challenging; however, this investigation has shown that using default values, combined with limited site-derived information, acceptable correlations between the measured and predicted settlement traces can be achieved.

The scatter of measured data even for a single site can be large. The settlement monitoring data from UK, BRE #4 Landfill Site and UK, Viridor #1 Landfill Site show this to be the case. Whilst attempts can be made to distinguish between the various settlement recordings based on time of filling and commencement of monitoring, the settlement data cannot be accurately located in time – strain space, due to a lack of knowledge on the amount of settlement that has already occurred. This highlights the importance of systematic settlement monitoring from immediately after the cap has been completed. In periods when no waste placement is occurring and temporary capping is present, temporary settlement monitoring points would allow the interim settlement to be assessed and the settlement to completion to be measured.


Golder Associates

9.0 SETTLEMENT IMPLICATIONS TO STAKEHOLDERS

9.1 Operators

The implications of landfill settlement to landfill operators, both during filling and after completion of filling, are numerous. From the acquisition and compilation of the settlement data from operators, and the data affecting settlements, it is clear that a greater knowledge of actual settlements and understanding of the settlement process can lead to significant commercial, planning and operational benefits to landfill companies. Benefits are summarised as:

1. Several operators expressed the need for at least quarterly surveys to enable new remaining void estimates to be calculated from the surveys and the pre-settlement levels agreed at planning. The void estimates were then fed to the company management for financial reporting purposes; data on tonnages should also be available for landfill tax purposes.

2. One operator uses quarterly level surveys to compute and track with time the waste

density which, after the accumulation of a substantial database of these values, enables the operator to assess compaction performance and to identify effects of changing waste streams.

3. More importantly for the same operator, the frequent level data and waste density

calculations, together with knowledge of the waste types, enables estimates to be made, using an in-house calculation method, of future post-filling settlements based on target filling completion dates. These estimates, in turn, can allow more efficient distribution of waste disposal between different cells to maximise void space, this being a dynamic assessment because the remaining void space increases with time due to the on-going settlement of the emplaced waste.

4. The settlement predictions made by the same operator also provide the technical

justification for negotiating and agreeing, during the filling stage, revised pre-settlement levels with the planning authority on a cell-by-cell basis. As shown by the HBM analyses, the rate of filling is a critical factor in post-closure settlement, and quarterly measurements provide the information to facilitate technically defensible settlement predictions. These negotiations occur before completion of filling to the previously agreed pre-settlement levels and this can avoid the disruption and costs involved in the placement and subsequent removal of temporary or permanent caps and gas collection systems.

5. Even if not used in this way, the settlement records can be important after landfilling is

complete to the prediction of long-term post-closure settlement, the updated estimation of the post-settlement contours and the timing of final capping and restoration.


Golder Associates

6. With the recent Court of Appeal judgement allowing landfilling to take place based on the principle of “piggy-backing”, where a new landfill is wholly or partially built on top of pre-existing landfilled waste, the importance of reliable estimation of settlements of the old landfilled wastes underlying the new landfilling will be critical. The regulator can be expected to require demonstration that adequate functionality, stability and durability of the new engineered structure, including the lining and emission control systems, can be assured, all of which rely on reasonably accurate predictions of the future settlements of the old waste as well as the new wastes.

7. Predictions of post-filling settlement with time can be useful in agreeing a programme

of final capping with the Environment Agency or restoration with the planning authority, where delaying these activities would allow a substantial proportion of settlement to occur. The risk of settlement-induced damage to the capping or to the restoration e.g. cracking, tears, settlement hollows, surface water drainage disruption and gas collection system deterioration would lessen and the extent of remediation works would be reduced.

8. Waste settlements can cause damage to side slope lining systems by downdrag stresses,

disruption of elements of the lining system and physical damage from differential movement of adjacent elements of the lining system. Reducing waste settlements adjacent to side slopes can be achieved by placement of inert wastes or other methods, and monitoring of settlements in these areas provides information on the performance of the installation compared to the design. Similarly, other components of the landfill engineering, including leachate and gas wells, will be affected by waste settlements. Knowledge of the settlements of the waste during and after filling, relative to the wells will provide information on impending failure of wells, and data of use in future well design.

9. In the long term, when an operator wishes to cease active management of the landfill, a

sound understanding of landfill settlement will facilitate agreement with the regulator that completion has been achieved.

Following completion of filling at a site (or part of a site), settlement will be most rapid in the first five years. Several of the benefits set out above can be gained from more accurate post-filling settlement monitoring on fixed survey markers at quarterly rather than annual readings. Based on these more frequent readings, selecting the time to extend the survey interval to yearly can be made with the benefit of a sound understanding of the settlement behaviour.

9.2 Planning Authorities

Planning authorities grant permission for the post-settlement and usually pre-settlement levels for landfills, as well as agreeing timetables for completion of filling and restoration, and often find this is a challenging responsibility. This aspect was discussed at some length in the Workshop (Appendix 1). Several of the concerns are seen as:

1. How does settlement affect the visual impact of individual landfills with time and what are reasonable timescales for restoration.


Golder Associates

2. How does settlement affect regional waste capacity and future planning for regional waste disposal?

3. Planning authorities can, in effect, only control pre-settlement levels, with the post-

settlement levels eventually being achieved after many years. 4. If excessive settlement occurs, then a new planning application is typically required to

permit new filling back up to the post-settlement levels (or more to allow for continuing settlement). The need to come back to re-fill sites because of underestimated settlements is a serious issue with the planning authorities.

5. The planning authority needs to be convinced that proposed pre-settlement will achieve

to desired post-settlement profiles but concern was expressed on the range of methods of calculation accompanying planning applications, the lack of supporting data or methodology justification, and the absence of guidance to assist their assessment.

6. Waste density at the time of disposal and then the increasing density with time, which

is used in applications to assess permitted tonnages for defined waste volumes in proposed landfills, can vary considerably between applications.

7. There is no published guidance to assist planning authorities on assessing the various

approaches in estimating pre-settlement contours and settlement used by applicants and their consultants. This lack of guidance is considered to be an important defect.

8. It would be beneficial if one calculation method was used by all applicants or if all

applications were assessed using one method (e.g. similar to the use of LandSim and GasSim).

Traditionally, a single percentage settlement of the waste depth from the pre-settlement levels to achieve the desired post-settlement levels (or alternatively, an uplift or surcharge on top of the post-settlement levels calculated as a single percentage of the waste depth from the post-settlement levels) has been used in planning applications which gives a smooth pre-settlement profile. This will generally result in post-settlement profiles which are not correct and instead, pre-settlement levels should be carried out for each cell but this will often lead to uneven pre-settlement levels. Planning authorities have accepted irregular, even unsightly, pre-settlement surfaces where these applications have been supported by sound technical justification, increasingly with the inclusion in the permission of a condition requiring the applicant to re-assess the settlement and update settlement predictions every three to five years.

9.3 Environmental Regulation Agencies

Several regulatory implications are seen to be of primary importance to the environmental regulation agencies and to operators, which can be summarised as:


Golder Associates

1. There is no regulation or guidance specifically related to settlement prediction and there is a recognised need for this, although the currently available mechanisms for preparing a guidance document are unclear. Settlement models currently submitted with planning applications are variable in technique, often simply generic, and are seldom supported with site-specific information.

2. There are now an increasing number of planning applications for vertical and lateral

(with overlap) (piggy-back) extensions. The requirement that all operating landfills are to be Landfill Directive compliant by 2009 will be a significant issue if present basal engineering for a vertical extension is not compliant.

3. There is a need to be able to assess reliably and differentiate between the settlement of

both the old waste and the new waste in piggy-back extensions. 4. Regulators are concerned on the effects of waste settlement on containment

engineering, and on leachate and landfill gas management. 5. The problem of re-contouring old sites where the settled profile is unacceptable is also

of concern to the regulators. 6. It is recognized that there are commercial drivers which have caused operators to gather

more settlement monitoring data but these additional data are not being made available to the regulators.

7. It is apparent that currently operators have to supply monitoring data in different

formats for each regulatory region and a consistent format in data presentation throughout the country would remove unnecessary work in data presentation.

Regulators can only expect to receive settlement data as required by the landfill permit. However, the improved knowledge of the settlement performance of the landfill gained with more frequent monitoring and better data presentation, especially when considered in conjunction with leachate quality and gas monitoring data, will give the opportunity to develop a more comprehensive understanding of the operational and post-completion behaviour of the landfill wastes.


Golder Associates

10.0 PROPOSED LANDFILL SETTLEMENT PROTOCOL

10.1 Introduction

This section describes a proposed protocol for the monitoring, analysis and prediction of landfill settlement, as developed from the findings of this project. The proposed protocol sets out the key steps involved in each part of the process to enable reasonably reliable predictions of landfill settlement to be made, based on (i) systematic site and waste-specific data recording and on (ii) a fundamental analytical approach. Improvements in the understanding of certain parameters used in the model and in the ease of use of the model will provide the opportunity to make the proposed protocol more robust and available for wider use.

10.2 Data Collection and Monitoring Framework

Table 14 contains the desirable site, operational and waste information and monitoring data for landfill settlement prediction. Table 15 shows additional monitoring which would be beneficial to validating the model performance. Whilst this list of requirements has been developed by using the HBM model, these are considered to be universally applicable requirements for the reasonably accurate prediction of landfill settlement behaviour by any method. Additional monitoring as itemised would also be useful for gaining a better understanding of the waste degradation behaviour and to utilise the gas generation prediction function of the HBM model.

It should be noted that there is an emphasis on the filling sequence because identifying the relative spatial and chronological position of individual elements of waste allows much more accurate representation of the behaviour of the overall waste mass.


Golder Associates

Table 14: Required Site Specific Information for Waste Settlement Prediction

Site, Operational and Waste Information Side slope geometries Site Geometry Maximum filling depths Start of filling Location (within cells) of filling Approximate filling rate Compaction used Breaks in filling End of filling Time and type of temporary capping/s Time and type of permanent capping

Filling Sequence Definition of the filling sequence is required to identify chronological position in the degradation sequence.

Types of daily cover used Dry unit weight as placed. This can be based on bulk unit weight and (assumed) moisture content. Waste composition (European waste classification as a minimum requirement, with MSW breakdown) Biodegradable content

Waste classification. Classification of the waste in terms of chemical and physical components.

Digestibility Monitoring

3 monthly surveys during filling phase including during filling breaks

Settlement monitoring Settlement monitoring is required to validate model performance and calibrate forward predictions.

Settlement monitoring on survey stations immediately following completion of permanent capping and at 3 monthly intervals for the first 5 years, annually thereafter (minimum 1 point/hectare at large sites but preferably 2 or 3/hectare at smaller sites and at landfill margins)

Table 15: Beneficial Additional Monitoring

Additional Monitoring Leachate quantity on a cell specific basis Leachate quality on a cell specific basis Details of any recirculation carried out including time, volumes and distribution. Gas quantity by cell or, if not available, by other defined areas

Leachate and Gas Production Comparison between the measured and modelled gas and leachate production allows calibration of the model to site specific characteristics, thus allowing more accurate forward predictions to be made. Gas quality by cell or, if not available, by other

defined areas

Waste decomposition Whilst it is not regarded to be a primary settlement parameter, it will contribute to calibration of the model and to a better understanding of the waste degradation process at the site

Physical sampling during any intrusive investigation/installation (e.g. retro-fitting wells) and laboratory testing to assess mass degraded and further potential biodegradability


Golder Associates

10.3 Settlement Prediction Protocol

A hierarchy of settlement prediction methodology is proposed in Figure 47. This identifies the increasing complexity from time dependant to full process modelling.

Simple time dependant methods identify a simple mathematical function to “best fit” previously observed settlement behaviour and, whilst methods that include filling will improve accuracy and acknowledge the differing behaviour of waste throughout the waste profile, they still do not fully address the actual processes occurring within the waste.

Current practice is typically towards the lower end, with consideration of time dependency, but not of the varying waste age or actual settlement processes.

Simple time dependant methods (e.g. Cα, log time, power law relationships, see

Section 2.2.1)

Time dependant modelling acknowledging filling (e.g. ISP method, see Section 2.2.2)

Full representation of mechanical, hydraulic and biodegradation processes (HBM model)

Assigned settlement percentage. No consideration of time to completion

Modelling settlement only

Modelling of processes causing settlement

Figure 47: Settlement Prediction Model Hierarchy


Golder Associates

10.4 Simplified Settlement Prediction Protocol

For any settlement prediction method, an understanding of the age of the waste is essential. For Cα based methods and similar, generic Cα values are typically applied irrespective of waste type and age of waste, and this leads to errors in the prediction of settlement behaviour.

For settlement prediction using these techniques, the filling sequence and settlement monitoring should still be recorded as suggested in Table 14. It is also important to record the waste composition as this assists in applying an appropriate Cα value.

10.5 HBM Model Settlement Prediction Protocol

The Settlement Prediction Protocol must recognise the range of HBM model input parameter types. There exists within the model a default parameter set, which means that the model can be run with minimal input information, i.e. data relating to the geometry of the site only. At the other end of the spectrum, where quantitative accuracy is necessary, landfill behaviour can be controlled using geometry, waste type, detailed operational circumstances and fundamental process data.

It is neither likely, nor desirable, to sample and test for each and every HBM model input parameter. Some are fundamental parameters controlling the simulation of microbial processes and unlikely to be sensitive to the ‘waste’ domain and the macro-scale. The type or role of input parameter should be recognised and efforts to validate input data concentrated accordingly. The validation effort may be further fine-tuned according to any specific interest, e.g. hydraulic, biodegradation and/or mechanical.

10.5.1 Role of HBM Input Parameters

Figure 48 (Figure 7 repeated) shows the function and determinability of all HBM input parameters and identifies the three distinct types of input parameter – site-specific, waste-specific and generic, as discussed in Section 3.1.1. These are shown as the apexes of a triangular diagram. Many inputs will show characteristics of more than one parameter type including combined site-specific and waste-specific, and combined waste-specific and generic.


Golder Associates

Figure 48: HBM Input Protocol


Golder Associates

11.0 CONCLUSIONS

11.1 Landfill Settlement Monitoring

The settlement of completed landfills has to be monitored on an annual basis as a permit requirement both during the operational filling stage and in the aftercare period, implementing this aspect of the Landfill Directive. This frequency is insufficient to evaluate the settlement performance of the landfill, particularly during filling but also in the early years after closure. No method of settlement monitoring is specified in the relevant regulations, so general topographical surveys, which are suitable during the filling stage, are also often carried out after closure which allows only a coarse and inaccurate evaluation of settlement across the landfill surface, compared to repeat measurements on settlement markers or stations installed in the landfill surface.

Operators have different approaches to settlement monitoring in accordance with the importance they place on the need to have settlement data for their forward operations, or financial reporting or forecasting. The frequency can range from fortnightly to annually. The benefits to operators of frequent level monitoring and settlement prediction during the filling stage do not appear to be widely appreciated across industry and their broader acceptance should be advantageous to the waste industry.

11.2 Collection and Assembly of Data other than Settlements

In addition to settlement data, the Landfill (England and Wales) Regulations 2002 require operators to provide information on an annual basis during the operational phase on the structure and composition of the landfill body. More specifically, the required information is defined as:

• The surface occupied by waste; • The volume and composition of the waste; • The methods of depositing; • The time and duration of depositing; and • The calculation of the remaining capacity still available at the landfill. Further expansion on the expected or required level of detail or reporting format of these types of data is not given in the above Regulations. No copies of the annual reports as required by the Regulations were examined in this project. The data required by the Regulations (including gas and leachate monitoring records in addition to those listed above) effectively covers all the elements of data for application in the HBM model.

The data set sought from the operators comprised information gathered either for their own operational purposes or for compliance with permit conditions. As for settlement records, instead of annually, more frequent recording of the data during landfilling and for several


Golder Associates

years post-completion would provide a better understanding of landfill behaviour, this being beneficial to landfill operators, planning authorities and regulators.

No new types of data were requested for this project; however, the data were, of course, not specifically being collected or assembled by landfill operators for use in the HBM model. Acquiring the data for the model could involve compiling data from different parts of an operator’s organisation. Where company ownership had changed, personnel continuity had been disrupted, or offices holding relevant records relocated, shut or new premises opened, then data completeness and continuity were often compromised. Over time, the data collection undertaken, method of assembly or means of storage have changed and generally improved. For longer running landfills, this inevitably made data retrieval more complex and time-consuming and the support of the contributing data providers, despite these practical difficulties, is gratefully acknowledged.

11.3 Data Collection and Assembly

It is apparent that as the data requirements in EU countries are set by existing directives and national legislation, a standardised format could be introduced in the UK so that the data are collected, assembled, reported and stored in common compiled manner. Consistent reporting across all areas of the country could then be achieved instead of highly variable standards of reporting developed on an area or operator basis. This would facilitate review by regulators as well as enabling operators to make use of the data for their operational and commercial benefit. Currently, it is apparent that the detailed scope of work for the reporting required by the regulations is uncertain.

It is concluded that clear guidance on these data reporting requirements is prepared and issued as a matter of some urgency to remove any current uncertainty caused by lack of detail of the requirements in the Landfill (England and Wales) Regulations 2002 or permits, or inconsistent requirements in permits for different landfills. Written guidance could be supported or extended by the preparation of data management software in the form of a database, following consultation with industry, and made available to industry.

It is recommended that the required frequency of level surveys and data collection is increased to a minimum of quarterly during the operational phase and, for the first five years following completion of landfilling in an area, that level surveys on survey markers are also recorded quarterly, increasing to annually thereafter. While this is an increase above the statutory requirements (although some landfills do already have quarterly reporting requirements and others carry this out for their own purposes), many of the benefits of this additional monitoring accrue commercially to the landfill operators. Importantly, acquisition of these data will provide a better understanding of the behaviour of the landfilled wastes and will enable more reliable projections of future settlements to be made and updated with time.


Golder Associates

The HBM model not only computes future settlement but can also be used to forecast gas generation rates, enabling a more holistic understanding of the degradation behaviour of the landfilled wastes to be developed. This is important in the procedure to assess when landfill completion has been attained whereby the landfill site is no longer a threat to the environment and that all management can cease.

It is recommended that gas monitoring on a cell-by-cell basis is investigated. The benefit would be that gas monitoring data would become much more useful in developing the understanding of waste degradation and gas generation within landfills, which can also be linked to settlement records and predictions.

11.4 Changing Waste Streams

As the HBM code works from a fundamental process-driven perspective, not a curve fitting process based on previously obtained waste settlement data, it is ideally suited to adaptation for changing waste streams. It is therefore important that research into input parameters for new waste types is carried out at a sound theoretical level, as this will allow an understanding to be developed of the changes that will occur in future settlement behaviour as waste types change.

11.5 Conclusions on Modelling

The model provides a theoretically robust approach with consideration of the waste mechanics and chemical processes which control settlement. The HBM model is unique in that it provides a coupling between the, hydraulic, biodegradation and mechanical components of waste behaviour.

Attempts have been made to compare the model to measured landfill settlement data and it has been shown that the model can be used to reproduce measured settlement traces for a number of landfills with markedly different waste and site characteristics. A lack of complete data sets (with waste composition, filling sequence, filling rate, capping date, and subsequent full settlement record, cell specific gas production, leachate production and quality) does mean that a fully rigorous validation is currently beyond the scope of the available data sets.

The model is complex in its formulation and despite a very effective graphical interface, it remains a relatively complex tool to implement. As with all finite element software, particularly bespoke codes such as this, it cannot be treated as a “black box”, and a sound understanding of the theory is required in order to implement the code. In its present form, it is considered to be an extremely useful research tool. However, users with the requisite experience and theoretical understanding can currently apply the model with relatively little effort to real cases and obtain settlement predictions with a sounder technical basis than the presently used time-dependant methods. Sensitivity and parameter analyses can be readily incorporated in the analyses.


Golder Associates

A simplified version of the model could in future be developed to be used in conjunction with the landfill settlement monitoring protocol and an improved waste classification scheme. The theoretically complex parameters could be hidden from the user. In their place qualitative input values could be developed in lieu of the absolute values.


Golder Associates

12.0 FUTURE DEVELOPMENTS

12.1 Data Collection and Recording

Waste and site-specific data are already collected by landfill operators as part of their permit conditions. It is apparent from this project that the potential benefits from a more consistent and thorough method of recording and compiling the data are currently being missed in many cases. Development of guidelines or permit conditions for data collection and reporting would encourage or require the collection and reporting of the data in an integrated and much more usable form, with little additional effort in collection and reporting. More frequent data measurement or recording (e.g. quarterly settlement surveys) would entail additional effort but would provide the opportunity of material benefits to operators, and improve the understanding of landfill behaviour in the short and long-term for all stakeholders.

12.2 HBM Model Parameter Research

Whilst a greater understanding of waste mechanics in general is still required, several key parameters have been identified. The decomposition induced void change parameter (Λ) has a direct controlling influence on the biodegradation parameter, and the solid degradable fraction and digestibility can affect the settlement magnitude and time to completion.

12.2.1 Decomposition Induced Void Change Parameter (Λ)

This parameter controls the relationship between decomposition induced volume change and the resulting settlement. Where decomposition occurs, it results in a loss of the solid degradable fraction, and an increase in void. McDougall and Pyrah (2004) discuss the theoretical derivation and application of Λ. However, McDougall (2007) acknowledges that there are little data currently available for quantification of Λ. This primarily impacts the magnitude of settlement experienced by the waste. In practice, this may be further complicated by collapse events and the waste structure becoming looser and weaker due to degradation, with particle reorganization and void reduction then occurring driven by overburden loading.

This is an area which requires further investigation in order to fully understand the link between biodegradation induced solid phase loss and the resulting, if any, settlement that occurs. Whilst physical sampling would be a costly exercise, it may be possible to carry out such investigations using samples collected when retro-drilling installations through the waste mass. This is a recommended topic for a future research and development project.

12.2.2 Digestibility and Solid Degradable Fraction

The amount of degradable material and the rate at which a waste mass degrades are fundamental factors in controlling the time to completion of a landfill site. The digestibility


Golder Associates

of components should relate to the degradable percentage and type of material. A classification framework is required to allow operators to classify their waste and to derive the solid degradable fraction and digestibility. Each category in the Consolidated European Waste Catalogue could be assigned a digestibility rate factor and, from the percentage concentrations, both the overall solid degradable material and digestibility (rate) can be assigned to the waste mass.

Biodegradation testing is carried out as part of the Landfill Allowance Trading Scheme (LATS), and this will provide additional information on the waste going to landfill. The actual solid degradable fraction of the waste may be lower, as all organic material may not degrade. Additionally the daily cover content should be considered.

Similar to the decomposition induced void change parameter, this is an area for additional research.

12.3 HBM Model Development Requirements

12.3.1 Definition of Filling/Non Filling Events

At present, periods of non-filling (i.e. just settlement) can be specified by identifying a filling region outside of the limits of the defined model.

Whilst effective, this was not an obvious feature in the model. It is suggested that the model GUI should be modified to allow periods of non-filling, and whilst the code may assign a filling sequence to an area outside of the model limits, it would allow the user easier access to this feature.

12.3.2 Infiltration Events

An issue was noticed whereby when more than 30 infiltration events were defined, only the first 30 were transferred to the input data file. This should be rectified, or if 30 is a limiting number of events, the GUI should only allow 30 events to be defined.

12.3.3 Applied Loads

It would be a useful addition to the model to allow for additional loading increments to the upper boundary of the model. This would allow the application of capping, restoration soil loading and additional stockpiles thus allowing them to be sequenced within the correct timeframe. This is particularly important in light of the deformation of the mesh that occurs during filling and the lower than specified height being achieved if filling is specified, as a material cannot easily be specified in the correct position to represent the capping layer.


Golder Associates

12.3.4 Settlement of Deactivated Zones

As settlement occurs during the model filling stage, the upper surface of the modelling mesh also settles as a result of movement in the underlying elements. Even though the upper zones are yet to be disclosed, the movement still occurs. This is a result of modelling using finite element techniques, which makes re-meshing of the geometry challenging and very time consuming; hence the initial mesh with associated deformations is retained through the model. Whilst it is not necessarily suggested that the model be revised to allow filling to full height and re-formation of the mesh, it is suggested that this issue is addressed in greater detail in the users’ literature, and techniques for approaching the filling height are identified.

Several methods of approaching this issue have been identified:

1. Use a mesh specified to the actual height, and correct the filling to the lowered height by normalising the post-filling settlement to strain (-), not an absolute measurement of settlement (m). Care must be taken when using this technique to take the post-filling settlement from the actual end of filling, which can be identified by a deviation in the settlement traces of the upper two elements within the column.

2. The mesh can be specified to a height greater than the initial height of the model. This

can be an iterative process and the model settlement may alter as a function of the additional height. Any significant parametric variations will result in the mesh dimensions requiring re-defining.

3. An alternative method is to define a mesh significantly taller than required, even by the

parametric variation with the largest resultant settlement (particularly with the highest solid degradable fraction and Λ). Filling events can then be defined within this mesh and periods of settlement can be defined by specifying filling events outside of the mesh area, as suggested in Section 12.3.1.

12.3.5 Determination of End of Filling

This specifically relates to modelling option (1) from Section 12.3.4 above, where the filling occurs within a settling mesh that has post-settlement smaller than the specified filling height. As filling will reach the full height of the mesh before the date calculated from the initial mesh filling height and the filling rate, the end of filling is required in order to extract data for the post-filling settlement. The end of filling can be identified by the onset of settlement in the upper surface of the waste, which is in turn identified by non-parallel settlement traces for the upper element in the model and the element directly below it. Information on the filling sequence should be included in the HBM model literature.

12.3.6 Moisture Content – Biodegradation Relationship

The model currently uses a linear relationship between moisture content and decomposition, with no limiting moisture content value on the rate of decomposition. Additionally, increased moisture content has the same effect of different waste types (e.g. food waste, paper, wood)


Golder Associates

represented through the digestibility parameter. Refinements could be made to the relationship between moisture content and biodegradation.

12.3.7 Boundary Conditions

At present, the mechanical boundary conditions do not represent the frictional interfaces found at the edges of landfills in worked out mineral quarries. Ultimately coupling an accurate hydro-bio-mechanical settlement model with lining system modelling, such as those described by Villard (1996) and Fowmes et al (2006) would allow not only the settlement but also the settlement predictions to be defined. The mechanical component of the model is designed to represent vertical settlement and the horizontal stress state may not be appropriate for this modelling.

In addition to the boundary condition model, the edge effects of landfill sites should be considered. In this research project, a column of waste near the centre of the waste mass has been considered, hence the effects of waste thinning towards the edges on the behaviour of the waste have been omitted. Landfills typically have sloping sideslope liner and capping system geometries, and this may potentially influence the settlement characteristics of the waste mass. Additionally, landraise-type landfills can be affected by lateral spreading and these effects should also be investigated. It is a recommended subject of further work to compare the behaviour of the HBM model to settlement data from across the full landfill geometry.

12.4 Settlement Prediction Guidance

Currently there are no regulations or guidance on the procedures or methodology to be used in predicting settlements. This causes difficulties for planning authorities, regulatory agencies and operators in agreeing predicted settlements as different methods of varying assumptions and complexity (or, more usually, simplicity) and site-specific nature are submitted with planning applications. It is expected that this deficiency will become more acute when applications for vertical extensions are being considered.

It is concluded that guidance on settlement prediction should be prepared and issued for the benefit of these stakeholders.

12.5 Gas Generation Prediction

Further work and validation on the gas generation function of the HBM model would be beneficial, together with assessing the potential to linking the model with other gas generation assessment models e.g. GasSim2. Data on gas generation on a cell basis would facilitate this research.


Golder Associates

12.6 Probabilistic analyses

The HBM model is currently a deterministic model. It could be adapted to include probabilistic analyses, however, this would require a significant increase in available processing power as, at present, a single run using a 10 x 10 finite element mesh for a 25 year analysis period, takes approximately 2 hours (using an Intel Core Duo 2.00 GHz CPU.).


Golder Associates

13.0 REFERENCES

Barlaz M.A., Ham R.K. and Schaefer D.M. (1989) Mass balance analysis of anaerobically decomposed refuse. A.S.C.E., J. Env. Eng. Div, 115, 6, 1088-1102.

Beaven, R.P.(1999) The hydrogeological and geotechnical properties of household waste in relation to sustainable landfilling. PhD thesis. University of London.

Beaven, R.P. (2000) The hydrogeological and geotechnical properties of household waste in relation to sustainable landfilling. PhD Thesis, Queen Mary and Westfield College, University of London

Barlaz M.A., Ham R.K. and Schaefer D.M. (1989) Mass balance analysis of anaerobically decomposed refuse. A.S.C.E., J. Env. Eng. Div, 115, 6, 1088-1102

Bjarngard, A. and Edgers, L. (1990) Settlement of solid waste landfills. Proc 13th Annual Madison Waste Conf., Wisconsin, 192-205.

Bookter T.J. and Ham R.K. (1982) Stabilization of solid waste in landfills. A.S.C.E., J. Env. Eng. Div, 108, EE6, 1089-1100.

Cecchi F., Traverso, P.G., Claney, J. and Zaror, C. (1988) State of the art of R and D in the anaerobic digestion process of municipal solid waste in Europe. Biomass, 16, 257-284.

Cooke, S.D., Walker, N. and Thomas, R.L. (2007) Calibrated Waste Settlement Prediction Data From Numerical Modelling Of Waste Processes At Some UK Landfill Sites. Proceedings 11th Int. Waste Management Symp. Sardinia, 1 - 5 October 2007.

Dixon, N. and Langer, U. (2006) Mechanical properties of MSW: Development of a classification system. International Journal of Waste Management Science and Technology, 26, 3, 207-318.

Edil, T.B., Ranguette, V.J. and Wuellner, W.W. (1990) Settlement of municipal refuse. Geotechnics of waste fills: Theory and practice STP 1070, eds. Landva, A and Knowles, G.D., AST, Philadelphia, 225-239

El-Fadel, M, Findikakis, A.N. and Leckie, J.O. (1996) Numerical modelling of generation and transport of gas and heat in landfills I. Model formulation. Waste Man. and Res. Volume 14.483-504.

El-Fadel, M. and Khoury, R. (2000) Modelling settlement in MSW landfills: a critical review. Critical reviews in Environmental Science and Technology, CRC Press, 30(3), 327-361.


Golder Associates

Environment Agency Wales (2007), Determination of the Biodegradability of Mixed Industrial and Commercial Waste Landfilled in Wales, 402-0341-00011, I&C Waste Analysis Study, November 2007

Fowmes, G.J., Dixon, N., Jones, D.R.V. and Cowland, J. (2006). Modelling of Lining System Integrity. Proceedings 8th International Conference on Geosynthetics, Yokohama, Japan.

Fowmes, G.J., Dixon, N. and Jones, D.R.V. (2007). Landfill stability and integrity: the UK design approach. ICE J. of Waste and Resource Management. May 2007, WR2, 51-61.

Fricke, K. Muller, W., Bartetzko, C., Einzmann, U., Franke, J., Heckenkamp, G., Keller-Aschenbrenner, K., Kolbl, R., Mellies, R., Niesar, M., Wallmann, R. and Zipfel, H. (1999) Biological Pre-treatment of Waste for landfills: Stabilisation of residual waste by mechanical biological pre-treatment and effects on landfilling) 1480945, German Federal Ministry of Education and Research, Witzenhausen, Germany.

HMSO (1996) Waste Management Paper 26E, (1996) Landfill Restoration and Post Closure Management (Consultation draft August 1996). ISBN 0 11 753185 5.

Jessberger, H. L. Syllwasschy, O. and Kockel, R. (1995). Investigation of waste body-behaviour and waste-structure-interaction. Proc. 5th Int. Landfill Symp., S. Margherita di Pula 2, 731–743.

Jones, K.L. and Grainger, J.M. (1983) The application of enzyme activity measurements to a study of factors affecting protein, starch and cellulose fermentation in domestic refuse. European J. Appl. Microbiology and Biotech., Vol.18. 181-185.

Kazimoglu, Y.K., McDougall, J.R. and I.C.Pyrah (2005) Moisture retention and movement in landfilled waste. Proc. GeoProb2005. Int’l. Conf. on Problematic Soils, Eastern. Mediterranean University, North Cyprus, May 2005, ed. Bilsel, H. pp 307-314

Langer, U. (2005) Shear and compression Behaviour of undegraded municipal solid waste. Unpublished Ph.D. thesis, Loughborough University. November 2005.

Lee D. D. and Donaldson T. L. (1985) Anaerobic digestion of cellulosic wastes. Biotechnol. Bioengng. Symp. 15, 549–560.

Ling, H.I., Leshchinsky, D. Mohri, Y. and Kawabata, T. (1998) Estimation of municipal solid waste landfill settlement. ASCE, J. Geotech. Geoenv. Engrg, 124(1), 21-28.

Marques, A.C.M., Filz, G.M. and Vilar, O.M. (2003) Composite compressibility model for muinicipal solid waste. ASCE, J. Geotech. Geoenv. Engrg, Vol 129(4), 372-378.


Golder Associates

McDougall, J.R. (2005) HBM Graphical User Interface Manual. P5.

McDougall, J.R. (2007) A hydro-bio-mechanical model for settlement and other behaviour in landfilled waste. Computers and Geotechnics, Special Issue: Chemo-Mechanical Interaction in Geomaterials, 34(4), 229-246.

McDougall, J. (2008) Science and engineering of landfilled waste mechanics. In press, ASCE GeoCongress.

McDougall, J.R. and Philp, J.C. (2001) Parametric study of landfill biodegradation modelling: methanogenesis and initial conditions. Sardinia 2001, 8th Intl Waste Man. and Landfill Symp, S.Margherita di Pula, eds. Christensen, Cossu and Stegmann, CISA Cagliari, Vol 1, pp 79-88.

McDougall J.R., and Pyrah I.C. (2004). Phase relations for decomposable soils. Geotechnique, Vol 54, No7, pp 487-494.

McDougall, J.R. and Hay, J. (2005) Hydro-bio-mechanical modelling of landfilled waste: formulation and testing. Proc. Int’l. Workshop on Hydro-Physico-Mechanics of Landfills, University of Grenoble, March 2005.

Morris, J.W.F., Vasuki, N.C., Baker, J.A. and Pendleton, C.H. (2003) Findings from long-term monitoring studies at MSW landfill facilities with leachate recirculation. Waste Management, 23, 653-666.

National Assembly for Wales (2002) Pilot Study on Municipal Waste Composition in Wales, AEAT/ENV/R/0901.

Needham, A.D, Jones, D.R.V, McDougall, J, Dixon, N, Braithwaite, P and Rosevear, A. (2007) Assessment of Landfill Settlement Data for Evaluation of a Hydro-Bio-Mechanical Settlement Model. Proceedings Sardinia 2007, Eleventh International Waste M, 26, 3, p171.

Needham. A.D., Fowmes, G.J., McDougall, J., Boniface, T., Dixon, N. and Braithwaite, P. (2008) Predicting long-term settlement of landfills using a fundamental model of waste behaviour. Proc. Waste 2008, Stratford-upon-Avon, England (in publication).

Olivier, F., Gourc, J.P., Munoz, M.L., Budka, A. and Denecheau, P. (2003a) Validation of an incremental waste settlement prediction model with surface survey data. Sardinia 2003, 9th Intl Waste Man. and Landfill Symp, S.Margherita di Pula, eds. Christensen, Cossu and Stegmann, CISA Cagliari, CD only.

Olivier, F., Gourc, J.P., Lopez, S., Benhamida, S. and Van Wyck, D. (2003b). Mechanical behaviour of solid waste in a fully instrumented prototype compression box. Sardinia 2003,


Golder Associates

9th Intl Waste Man. and Landfill Symp, S.Margherita di Pula, eds. Christensen, Cossu and Stegmann, CISA Cagliari, CD only.

Olivier F. and Gourc J-P. (2006) Hydromechanical behaviour of municipal solid waste subject to leachate recirculation in a large-scale compression reactor cell. Waste Management, 2006, 27, No 1, 44-58.

Olivier, F. and Gourc, J.P. (2007) Hydro-mechanics of MSW subject to leachate recirculation in a large-scale compression reactor cell. Waste Management Journal.

Park, H.I. and Lee, S.R. (1997) Long-term settlement behaviour of landfills with refuse decomposition. J Solid Waste Technology and Management, Vol 24(4), 159-165.

Powrie, W., Richards, D.J. and Beaven, R. (1998) Compression of waste and implications for practice. In Geotechnical Engineering of Landfills, Proc. Symp. East Midlands Geotechnical Group, eds. Dixon, Murray and Jones, Thos Telford, 3-18.

Powrie, W. and Beaven, R.P., (1999) Hydraulic properties of household waste and implications for liquid flow in landfills. Proceedings of the Institution of Civil Engineers, Geotechnical Engineering, October, 1999.

Qian, X., Koerner, R.M. and Gray, D.H. (2002). Geotechnical Aspects of Landfill Design and Construction, Prentice-Hall, Inc. ISBN 0-13-012506-7.

Sharma, H.D. and De, A. (2007) Municipal solid waste landfill settlement: post-closure perspectives. ASCE, J. Geotech. Geoenv. Engrg, Vol 133(6), 619-629

Sowers, G.F. (1973) Settlement of waste disposal fills. Proc 8th ICSMFE, Moscow, Vol.2, 207-210

Straub, W.A. and Lynch, D.R. (1982). Models of landfill leaching: moisture flow and inorganic strength. J. Env. Eng., Am. Soc. Civ. Eng. 108, EE2, 231-250.

Thomas, R.L. and Cooke, S.D. (2007) Landfill Waste Settlement Prediction By Numerically Modelling Waste Processes. Proceedings 11th Int. Waste Management Symp. Sardinia, 1 - 5 October 2007.

Villard, P. (1996). Modelling of interface problems by the finite element method with considerable displacements. Computers and Geotechnics, Vol. 19, No. 1, 2345.

Viturtia, A., Llabres-Luengo, P., Cecchi, F. and Mata-Alvares, J. (1995) Two-phase kinetic model fitting in a two-phase anaerobic digestion of highly biodegradable organic matter. Env. Technology, Vol.16. pp379-388.


Golder Associates

Wald S., Wilke C.R. and Blanch H.W. (1984) Kinetics of the enzymatic hydrolysis of cellulose. Biotechn. and Bioeng., 26, 221-230.

Wang Z. and Banks C.J. (2000) Accelerated hydrolysis and acidification of municipal solid waste in a flushing anaerobic bio-reactor using treated leachate recirculation. Waste Man. and Res., 18, 215-223.

Watts K and Charles J.A. (1990). Settlement of recently placed domestic refuse landfill. Geotechnical Engineering, Proc. ICE. Vol 88, No1.

Williams, E. (1991) A model for long term settlement in landfills MS Thesis, Department of Civil and Environmental Engineering, Tufts University, Medford, MA.

Yen, B.C., Scanlon, B., (1975). Sanitary landfill settlement rates. ASCE, Journal of the Geotechnical Engineering Division 101, 475–487.

Yuen, S.T.S. “Bioreactor Landfills Promoted by Leachate Recirculation: A Full-Scale Study” PhD Thesis, Department of Civil Engineering, University of Melbourne, UMI Dissertation Services, Michigan, Catalogue No. 9948381.

Yuen, S.T.S. and McDougall, J.R. (2003) Effect of enhanced biodegradation on settlement of municipal solid waste landfills. Australian Geomechanics, 38,(2), 17-28.

Zekkos, J.D., Bray, J.D., Kavazanjian, E., Matasovic, N., Rathje, E., Reimer, M. and Stokoe II, K.H. (2005). Framework for the estimation of MSW unit weight profile, 10th Intl Waste Man. and Landfill Symp, S.Margherita di Pula, eds. Cossu and Stegmann, CISA Cagliari,

Zhang, B. (2007). Constitutive modelling of municipal solid waste. Unpublished PhD thesis. Loughborough University, UK.

July 2008 06529217.502 Landfill Settlement Version A.2 Final Report

Golder Associates

APPENDICES


Golder Associates

APPENDIX 1

NOTES ON LANDFILL SETTLEMENT WORKSHOP BIRMINGHAM, 29TH APRIL 2008

29 April 2008 - 1 - 06529217.450 Landfill Settlement Workshop Birmingham

Golder Associates

LANDFILL SETTLEMENT WORKSHOP: SUMMARY

Present:

Peter Braithwaite (DEFRA/EA), Rob Marshall (EA), Nicola Ingrey (EA), Richard Moss (EA), Kevin Clarke (Biffa), Stuart Hayward-Higham (SITA), Chris Myers (WRG), Chris Ratcliffe (WRG) Richard Terry (Veolia), Andy Mackintosh (Hertfordshire CC), Roy Romans (Bedfordshire CC) Roy Leavitt (Essex CC) Tim Grindell (Cory Environmental), Nick Walker (Veolia), Neil Dixon (Loughborough), Asif Siddiqui (Southampton), Ken Watts (BRE), John McDougall (Napier), Adrian Needham, Gary Fowmes, Russell Jones (Golder).

Presentation 1: Landfill Settlement: Overview of processes and implications (AN)

• There is confusion over terminology used by various consultancies, e.g. surcharge allowance and settlement percentages. It would be useful to include agreed definitions in a guidance document.

• The fear of overfilling that leaves the incorrect profile is the biggest issue for planners. Can the planners trust the submissions made by the operators? Further, do the operators follow the recommendations made by their own consultants?

• Is pre-settlement important to planners? Local authorities can only control pre-settlement profiles. They need to be convinced that the consultants’ predictions are correct.

Presentation 2: Landfill Settlement: The planner’s perspective (RR)

• There is currently no guidance for assessing various approaches to landfill settlement used by consultants/operators.

• It would be useful to have guidance on waste densities for future planning. Note that there is a difference between densities used for regional planning compared with the densities for waste in various parts of a specific landfill.

• Some sites take a long time to fill and so predictions made 5 or 10 years ago may no longer be valid. The methodology used and future predictions should be reviewed say every 5 years.

• As biodegradable content of landfilled waste decreases, there will be a reduction in void space and revisions will be required to operators’ business models.


Golder Associates

• It was noted that a recent planning permission made by the local authority inserted a requirement to assess the settlement and settlement prediction every 3 years.

Presentation 3: Landfill Settlement: An operator’s perspective (SHH)

• In complex sites with different waste deposited in different areas will have very “untidy” pre-settlement contours to enable a smooth post-settlement profile. Would such an “untidy” surface be acceptable to the planner? It was felt that planners would accept this approach as long as it was treated as an no-going process with regular review and updating.

• It was noted that the biodegradable reduction targets are only on municipal waste and that the non-domestic waste stream will still have a large biodegradable content.

Presentation 4: Regulators issues (RM)

• Landfill settlement models submitted to the Environment Agency vary significantly and generally do not relate to any site specific information.

• There is a dearth of useful data available in the public domain. There is probably information available in filing cabinets throughout the country but it is generally not been made available.

Presentation 5: DEFRA funded study: Overview (AN/GF)

• The collection of gas production data on a cell-by-cell basis would be extremely useful to validate sophisticated waste process-based landfill settlement models.

Presentation 6: Landfill settlement: Interpretation and data obtained (JM)

• Experience has shown that biodegradation can start to happen very quickly and is not just something that happens after the capping has been placed.

• Any conclusions on leachate recirculation and affect on landfill settlement? Not from the field data obtained. After two years of recirculation, there was no evidence of increased rate of settlement. There is some lab data that may show this.


Golder Associates

Presentation 7: HBM model: Overview (JM)

• When sites are retro-drilled, it would be useful to take samples and measure the amount of degradation that has taken place.

Presentation 8: HBM model: User’s perspective (GF)

• Is the model a probabilistic simulation model? No it is deterministic. Could it be done? Yes, it is a sensible way forward.

• How do the planners cope with different settlement models being used by different consultants/operators? It is very difficult for planners; it would be beneficial if everyone used the same model.

Presentation 9: Mass loss and settlement (JM)

• Industry just starting to understand landfill gas production based on different waste constituents. In other words, it is much more complex than a one material degradation.

• There was a consensus in the recent waste workshop in New Orleans that this methodology was the best way of modelling other mechanical behaviour in addition to settlement.

Presentation 10: Improving settlement prediction and insights into landfill behaviour (AN/GF)

• Increasing the amount of monitoring would probably be resisted by the operators. However, the benefits of this additional monitoring would be better predictions. More regular data should improve confidence in the operators proposals.

• Currently the national operators are required to supply monitoring data in different formats in different regions and any national requirement may be easier for the operators to comply with.


Golder Associates

Future research needs

Data collection guidance:

• Cell specific gas production data. This cannot currently be carried out remotely with any accuracy, and so would incur additional costs.

• A major operator is about to start a five year data gathering information including live cell monitoring. This will include the measurement of the compression of each 6 m thick layer. Will this be made available to the public?

• Waste characterisation is important, need to relate physical properties of the waste constituents. It is up to the waste industry to take things forward and start describing the waste accurately.

Environment Agency guidance:

• There is a need to establish guidance or a series of case histories, but who should do this? Should Waste Management Paper 26B be updated to include this? Maybe ESA or CIWM could be tasked with producing the guidance? Such development would need a motivated group.

• Could the planning officers organisation produce some guidance? Possibility of setting up a joint planning officers/Environment Agency working group on landfill settlement.


Golder Associates

APPENDIX 2

HBM GRAPHICAL USER INTERFACE EXAMPLE

July 2008 - 1 - 06529217.502 Landfill Settlement Appendix 2 Final Report

Golder Associates

HBM MODEL GRAPHICAL USER INTERFACE EXAMPLE

This appendix shows a simple graphical representation of the HBM model, presenting the pre-processor, model calculation stage and post-processor outputs. This is not intended to provide a guide to applying the HBM model, but gives an example to aid the reader’s understanding of the model from a user’s perspective. A selection of screenshots with commentary are provided below.

Pre-Processor

Screen 1: Allows the user to define the Job title, select the analysis stage (moisture or transient), the time of the model and the length of a time step.


Golder Associates

Screen 2: Used for geometric inputs. The dimensions of the finite element mesh are defined, and the corner points of the model are used to define the geometry.

Screen 3: Used for the input of material properties. Advanced access must be selected to allow access to all properties. Additional materials can be added with varying material properties. The materials are assigned to the mesh in the Boundary Conditions screen.


Golder Associates

Processor

During the model processing the elapsed time steps and numerical iterations are displayed as shown below.

Post-processor

Screen 1. Shows the model filling height and moisture profile at specified days within the model run.


Golder Associates

Screen 2. Allows detailed analysis of the moisture content profiles to be shown at specified days within the filling sequence.

Screen 3: Shows the VFA concentrations at specified locations within the waste.


Golder Associates

Screen 4: Shows the Methanogenic Biomass at specified locations within the waste.

Screen 5: Shows the remaining solid degradable fraction at specified locations within the waste.


Golder Associates

Screen 6: Gives a detailed analysis of the phase relationship for a single location within the waste mass.

Screen 7: Shows the settlement at specified locations within the waste.

Golder Associates (UK) Limited Attenborough House Browns...

Documents

Transcript of Golder Associates (UK) Limited Attenborough House Browns...