Professional Documents
Culture Documents
GoldTailingsWB 2011
GoldTailingsWB 2011
Report to the
Water Research Commission
by
April 2011
DISCLAIMER
This report has been reviewed by the Water Research Commission (WRC) and approved
for publication. Approval does not signify that the contents necessarily reflect the views of
the WRC, nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
ii
EXECUTIVE SUMMARY
There is a regular need to undertake studies on the pollution potential of mine waste residue deposits in the
gold and coal mining industry. Such studies require the integration of water balance models with
geochemical models. Although mines do have rudimentary water balances for their waste residue deposits,
on close inspection it is almost always found that these water balances contain numerous unmotivated or
unvalidated assumptions. While these inaccuracies may be acceptable for an operational mine where the
water balance is being actively managed on a daily basis, these inaccuracies seriously hinder the ability to
make reliable predictions of post-closure pollution potential and to properly evaluate environmental
management / rehabilitation strategies.
The project was, therefore, structured to address the following three major objectives:
1. To develop a procedure and methodology that can be used by mines, researchers and consultants in
developing water balances for gold mine waste residue deposits.
2. To ensure that the outcome of the research is transferred to persons active in the field of pollution
assessment from gold mine residue deposits.
3. To provide material that can be used by tertiary educational institutions to incorporate into the standard
courses, thereby ensuring that new graduates have the knowledge to undertake accurate water
balances of such facilities.
The outcome of this proposed research project will, therefore, significantly enhance the ability to undertake
reliable and credible assessments of long-term pollution potential from the range of mine waste residue
deposits encountered in the gold mining industry.
SITE SELECTION
Potential tailings dams and waste rock dumps to be visited as part of the site selection process were
identified by approaching several mining companies. Although not all the mining companies approached
were willing to participate, Anglogold and Goldfields agreed to make a number of their facilities available.
While it was originally planned to visit 6 potential sites to select 3 for detailed study purposes, the project
team visited 10 tailings disposal facilities and 2 waste rock dumps.
From the perspective of this research project, the project team could distinguish the following different types
of deposits that should be included in the research project:
1. Gold tailings dam constructed with traditional day/night paddocks and beaching to a central penstock.
2. Gold tailings dam constructed through cycloning.
3. Gold waste rock dump constructed by end-tipping of waste off a conveyor.
The following criteria were used for a mine residue deposit to be selected as a study site:
In evaluating the suitability of various sites, preference was given to sites that already have installed
piezometers in their residue deposits with good historical records. At the outset of the project it was intended
to select 3 sites representing the 3 different types of facilities listed above, with facilities that have recently
been decommissioned. However, suitable type 2 tailings dams were not available and hence the project
team had no choice but to use only the type one tailings dam and two of these were selected. The following
sites were eventually selected for further study in the research programme:
iii
West Wits Old North Complex tailings dam (Anglogold)
Driefontein No 3A tailings dam (Goldfields)
Kopanang waste rock dump (Anglogold)
The following tasks which were in line with the agreed experimental set-up, data collection and monitoring
program and the additional tasks specified in the revised scope which was presented to and agreed by the
WRC Reference Group and the WRC have been implemented:
Sampling and analyses: All the required samples including undisturbed columns of samples from all
experimental sites were collected and all samples were characterised for their hydraulic properties.
Instrumentation: Installation of different probes and apparatus were completed successfully and data
collection was successful.
In-situ measurements of saturated and unsaturated hydraulic conductivity measurements at depths of
2000, 1500, 1000, 500, and 250 mm in all of the experimental sites were completed for both Driefontein
and Old North Complex tailings dams. Poor tensiometer data generation which was initially encountered
was addressed by installing the following additional equipment:
Mini tensiometers together with the lysimeter at the Driefontein site,
Six surface and side runoff plots at both the tailings dams,
Four deep tensiometer nests (500 mm, 1000 mm 4000 and 6000 mm long tubes)
Laboratory analyses: The laboratory measurements of hydraulic properties such as hydraulic
conductivity and water retention characteristics were completed.
Tailings evaporation: The actual evaporation from the surfaces at both tailings dams and transpiration
at the Old North Complex are clearly dominant factors in the water budget. This part of the study was
enhanced by:
Installing manual readable lysimeters,
Collection of a large vertical sample for laboratory testing of infiltration/redistribution behaviour and
Scintillometer measurements during the 2004/2005 wet season.
Rapid assessment of the infiltration redistribution process was instigated by introducing runoff plots
on the surfaces and slopes of the tailings dams as well as extracting large samples for testing in the
laboratory.
Substitution of some of the tensiometers with WaterMark sensors for monitoring of the energy status of
the tailings was implemented successfully and the response of the watermark sensors to capillary
pressure of the tailings was assessed.
Volumetric water contents were measured successfully with TDR100 in the tailings materials. However
the TDR100 apparatus was damaged and time constraints prevented installation of TDR probes.
The experimental and data collection task was successfully implemented. The data generated by the
instruments and laboratory analyses were collated into a database for use in the project. The relationship
between the datasets from the different instruments have been analysed and used in the development of the
water balances.
A tailings storage facility (TSF) consists of variably unsaturated porous media with saturated and
unsaturated zones. The proportion of the saturated zone decreases and may even disappear with time in
decommissioned dams. The hydrological system of gold TSFs for operational and non-operational tailings
dams are conceptually different and as such should be studied differently. Decommissioned tailings dams
have the following main characteristics:
They have spatially varying unsaturated zones, which makes simple physical water balance calculations
inappropriate.
Unlike the operational dam where the saturated zone is near the surface, the entire exposed surface of
the non-operational dam is variably unsaturated.
iv
In both decommissioned and operational tailings dams, there would be a spatial variation of the water
balance components from the edge to the centre of the dam. This is mainly attributed to the moisture
available in the profile, which is governed by the presence and depth of the phreatic surface. Thus infiltration
at the edge of the pool is maximum and minimum at the centre of the pool whereas evaporation would be
maximum at the centre of the pool and minimum at the edge of the pool. The depth of the phreatic surface
and rate of drop of this surface is difficult to estimate accurately.
Any attempt to determine the water balance of a decommissioned tailings dam should establish itself largely
on the unsaturated flow processes and must account for the impacts of surface flux boundary conditions.
The objectives of such studies should be to predict the actual infiltration rates and subsequently the seepage
quantity from the tailings dam through, for example, dam toe drains. The unsaturated zone in a tailings dam
is characterised by significant spatial variability and relies heavily on the unsaturated flow principles. This, in
turn, requires understanding and solving of the surface flux boundary problem – the result of which should be
applied to predict the water balance of the entire tailings dam which is a saturated-unsaturated medium.
Full understanding of the following aspects of the hydrological process in a tailings dam is required:
All these components were studied in this research project and the report presents detailed
recommendations on how these studies should be undertaken for cases where a screening-level water
balance is required and for cases where a detailed and accurate water balance is required. For a detailed
water balance, some form of numerical modelling will be required and the report provides guidance on what
models could be used, what parameters need to be determined and how the required data should be
collected.
However, the collected data was also evaluated in order to extract simple ‘rules of thumb’ or ‘general rules’
that could be used to more simply determine the facility water balance when a screening-level water balance
is deemed appropriate. These general rules that have been derived are listed below.
Event analysis:
No runoff for rainfall event less than 5 mm
Runoff = 0.235*Rain*(1+(API-7)/100)
Event analysis
Evaporate at potential rate until 40 m pore water tension then
Reduce to zero after a further 7 days without precipitation
v
GENERAL RULE FOR PERCOLATION
Long term analysis (Witwatersrand):
Percolation = 0.17*Rainfall
Event analysis
Derive response from pore water balance
Event analysis
No drainage for rainfall events less than 15 mm
Drainage = 1.1*Rain for events with high antecedent conditions
Drainage = 0.8*Rain for events with low antecedent conditions
The research project was successfully completed and a significant amount of data was collected that is
applicable to mine residue deposits in South Africa and that can be used to significantly improve the
accuracy and reliability of water balances provided for such facilities. As a water balance is an essential
precursor to a geochemical assessment, the water balance information provided in this research report will
significantly improve the reliability of impact assessments and predictions undertaken for South African mine
residue deposits.
Historically, there has been a need to use international research data to enable the apportionment of the
various components of a mine water balance, leading to inaccuracies in the water balance and in impact
predictions. While the values for infiltration (and subsequent reporting to seepage) of rainfall into tailings
disposal facilities is higher than values often and typically used in water balances prepared in South Africa, it
is recommended that the values reported in the ‘general rules’ shown above be used as default values in all
screening level assessments to be undertaken for mine residue deposits, unless site=specific data collection
exercises (done in accordance with procedures set out in this report) provide solid evidence for the use of
alternative site-specific values.
vi
ACKNOWLEDGEMENTS
The research in this report emanated from a project funded by the Water Research Commission and entitled:
Development of water balances for operational and post-closure situations for gold mine residue
deposits to be used as input to pollution prediction studies for such facilities
The financing of the project by the Water Research Commission and the contribution of the members of the
Reference Group are gratefully acknowledged.
The Reference Group responsible for this project consisted of the following persons:
vii
TABLE OF CONTENTS
viii
3.4 STORAGE DYNAMICS AND DRAINAGE 3-3
3.4.1 Tailings dams 3-3
3.4.1.1 Tensiometers 3-3
3.4.1.2 Tensiometer data 3-4
3.4.1.3 WaterMark sensors 3-4
3.4.1.4 Lysimeters 3-4
3.4.1.5 Tailings dam runoff plots 3-4
3.4.1.6 In-situ hydraulic conductivity measurement 3-5
3.4.2 Waste rock dump 3-5
3.4.2.1 Waste rock lysimeter 3-5
3.5 SAMPLING AND MATERIAL CHARACTERISATION 3-7
3.5.1 Sampling 3-7
3.5.1.1 Tailings dams 3-7
3.5.1.2 Kopanang waste rock dump 3-7
3.5.2 Laboratory measurements for physical characteristics 3-7
3.6 SUMMARY 3-8
ix
6 DEVELOPMENT OF WATER BALANCE MONITORING AND MODELLING
METHODOLOGIES 6-1
6.1 APPROACH 6-1
6.2 DATA REQUIREMENTS 6-1
6.2.1 Meteorological data 6-1
6.2.1.1 Basic meteorological data 6-1
6.2.1.2 Potential evapotranspiration 6-2
6.2.2 Material characteristic data 6-3
6.2.2.1 Particle size distribution (PSD) 6-3
6.2.2.2 Bulk density and porosity 6-4
6.2.2.3 Water retention characteristics 6-4
6.2.3 Hydrological response monitoring 6-4
6.2.3.1 Hydraulic conductivity 6-5
6.2.3.2 Moisture status monitoring 6-5
6.2.3.3 Runoff dynamics and infiltration 6-7
6.2.3.4 Evaporation and deep drainage (recharge) 6-7
6.3 TSF WATER BALANCE DETERMINATION 6-8
6.3.1 Conceptual model 6-8
6.3.2 Physical water balance calculation 6-9
6.3.3 Long term water balance prediction 6-10
6.3.3.1 Conceptual model 6-10
6.3.3.2 Model simulation 6-10
6.3.4 Screening or detailed model analyses 6-11
6.3.4.1 Water balance for an operational TSF 6-12
6.3.4.2 Water balance for a decommissioned TSF 6-12
6.4 WASTE ROCK DUMP WATER BALANCE DETERMINATION 6-14
6.4.1 Conceptual model 6-15
6.4.2 Screening or detailed model analysis 6-15
6.5 LONG TERM MONITORING 6-16
x
LIST OF FIGURES
3.1 Schematic experimental set-up for water balances of gold tailings dams 3-1
3.2 Schematic section of the lysimeter installation (measurements in mm) 3-5
3.3 Layout of the Kopanang waste rock dump, showing typical cross-sections for the location of
the lysimeter 3-6
3.4 Approximate position of the waste rock lysimeter at Kopanang 3-6
3.5 Cross-section of the lysimeter installation in the Kopanang waste rock dump 3-7
4.1 Daily rainfall data from the Driefontein tailings dam, October 2003 to May 2005 4-1
4.2 Daily rainfall data from the ONC tailings dam, October 2003 to May 2005 4-2
4.3 Daily rainfall data from the Kopanang waste rock dump, February 2004 to May 2005 4-2
4.4 Cumulative rain data for the Driefontein and ONC tailings dams and the Kopanang waste
rock dump, February 2004 to May 2005 4-3
4.5 Cumulative rain data for the Driefontein tailings dam and local met stations (Potchefstroom –
0437104; Fochville – 0474899; Carltonville – 0474680) for February 2004 to May 2005
4-3
4.6 Minimum and maximum daily temperatures for Potchefstroom (February 2004 to May 2005
4-4
4.7 Relative humidity (08H00 and 14H00) for Potchefstroom (February 2004 to May 2005) 4-4
4.8 Average wind speed (08H00, 14H00 and 20H00) for Potchefstroom (February 2004 to May
2005) 4-5
4.9 Cumulative reference potential evaporation (ETo) based on the Linacre 1977 equation
(Potchefstroom data) and Driefontein rainfall (February 2004 to May 2005) 4-6
4.10 Particle size distributions for the ONC tailings material for the penstock site at depths
0-600 mm (above) and 600-2100 mm (below) 4-7
4.11 Particle size distribution for the ONC tailings material: edge site 4-7
4.12 Particle size distribution for the Driefontein tailings material: edge (above) and
penstock (below) 4-8
4.13 Particle size distribution for the Kopanang waste rock, sand size fraction: surface 4-8
4.14 Water retention characteristics of the Driefontein tailings material: edge 4-10
4.15 Water retention characteristics of the Driefontein tailings material: middle 4-10
4.16 Water retention characteristics of the Driefontein tailings material: penstock 4-11
4.17 Water retention characteristics of the ONC tailings material: edge 4-11
4.18 Water retention characteristics of the ONC tailings material: middle 4-12
4.19 Water retention characteristics of the ONC tailings material: penstock 4-12
4.20 Saturated hydraulic conductivity for the Driefontein tailings (top) and ONC tailings material
(bottom) 4-14
4.21 Unsaturated hydraulic conductivity for the Driefontein tailings: edge, rep 1 (above) and rep 2
(below) showing Ksat 4-15
4.22 Unsaturated hydraulic conductivity for the Driefontein tailings: penstock 4-16
4.23 Diurnal fluctuation of the tensiometer response at the Driefontein tailings dam, middle site (4
to 11 December 2003) 4-17
4.24 Tensiometer and rainfall response at the Driefontein tailings dam, penstock site during
January 2005 4-18
4.25 Calibration relationship for the WaterMark sensors 4-18
4.26 WaterMark responses for the Driefontein site (middle) during January 2005 4-19
4.27 TDR measurements at ONC (above) and Driefontein (below) December 2003 4-20
4.28 Lysimeter drainage from the Driefontein tailings (January-May 2005) 4-21
4.29 Lysimeter drainage from the ONC tailings (January to May 2005), showing the tensiometer
response inside the lysimeter at 700 mm below surface 4-21
4.30 Lysimeter drainage from the Kopanang waste rock (April 2004-May 2005) 4-22
xi
4.31 Lysimeter drainage from the Kopanang waste rock in response to rainfall 4-23
4.32 Cumulative lysimeter drainage from the Kopanang waste rock in response to rainfall during
April 2005, following on a high antecedent moisture condition 4-23
4.33 Cumulative lysimeter drainage from the Kopanang waste rock in response to rainfall during
December 2004, following on a low antecedent moisture condition 4-24
4.34 Driefontein tailings dam side slope runoff and rainfall, October 2004 to May 2005 4-24
4.35 ONC tailings dam side slope runoff and rainfall, October 2004 to May 2005 4-25
4.36 ONC tailings dam top surface runoff (edge and middle/penstock) and rainfall, October 2004
to May 2005 4-25
4.37 Water content response during a wetting/drying event in a controlled evaporation experiment
in a column of ONC tailings 4-26
4.38 Estimation of evaporation from the surface of the ONC tailings column 4-26
5.1 Schematic of TSF water balance components for decommissioned dams 5-2
5.2 Schematic of the water balance for a decommissioned tailings dam 5-4
5.3 Relationship between rainfall and runoff for the side slope plot 1 on the ONC tailings dam
5-5
5.4 Relationship between rainfall and API-7-runoff for the side slope plot 1 on the ONC tailings
dam 5-5
5.5 Hydraulic gradients (positive for downwards flow) in the Driefontein tailings:
20 January 2005 to 20 May 2005 5-6
5.6 Schematic representation of the tailings dam water budget 5-8
5.7a Cross-section of the Driefontein tailings dam showing the approximate positions of the
monitoring stations 5-8
5.7b Initial boundary conditions for the Driefontein tailings dam 5-8
5.7c Distribution of material types (edge[red], middle[green], penstock[blue]) through the 5-9
Driefontein tailings dam
5.8a Variation of volumetric water content in the Driefontein tailings dam during 16 months of
simulation 5-9
5.8b Summary of the cumulative water balance for the Driefontein tailings dam during 16 months
of simulation 5-10
5.9a Finite element mesh with arbitrary observation nodes, material distribution and initial
condition (top to bottom respectively) for the Driefontein No 3 dam 5-10
5.9b Section of the finite element mesh, showing the finer mesh spacing in the near surface layers
and the near surface observation nodes for Driefontein No 3 dam 5-11
5.10 The position of the phreatic surface one year (top) and 5 years (bottom) after
decommissioning of ONC tailings dam 5-12
5.11 Summary of the cumulative water balance for the ONC tailings dam during 365 days for year
1 (top) and 5 years (bottom) of simulation after decommissioning 5-13
5.12 Schematic representation of a waste rock dump 5-15
6.1 Flow diagram showing the relationship between data collection, water balance determination 6-1
and monitoring for mine residue deposits
6.2 Schematic of data requirements for a detailed water balance of a gold tailings dam 6-2
6.3 Example of calculated cumulative reference potential evaporation (ETo) 6-3
6.4 In-situ water content measured by TDR at ONC (top) and Driefontein (bottom), December
2003 6-6
6.5 Schematic section of the lysimeter installation (measurements in mm) 6-7
6.6 Generalised flowchart for the water balance determination of a TSF 6-8
6.7 Schematic of water balance components for decommissioned TSFs 6-9
6.8 Schematic of the hydrological system constrained by the upper and lower boundary surfaces
of a TSF long after closure 6-11
6.9 Schematic of operational TSF water balance components 6-12
6.10 Schematic of decommissioned TSF water balance components 6-13
6.11 Schematic of waste rock dump water balance components 6-14
xii
LIST OF TABLES
5.1 Long term tailings dam runoff and infiltration (October 2004-May 2005) 5-5
5.2 Summary of the water balance for the 1st year and 5th year after decommissioning ONC No 3
dam 5-13
5.3 Water balance results from monitoring and experimental data (as percentage of annual
rainfall) 5-14
6.1 Data requirements for water balance determination for mine residue deposits 6-2
6.2 Hydrological response measurements and monitoring data 6-5
6.3 Water balance approach for an operational TSF dependent on level of detail required 6-12
6.4 Water balance approach for a decommissioned TSF dependent on level of detail required 6-13
6.5 Water balance approach for a waste rock dump dependent on level of detail required 6-16
xiii
ABBREVIATIONS
xiv
CHAPTER 1
INTRODUCTION AND BACKGROUND
1.1 MOTIVATION FOR THE PROJECT
There is a regular need to undertake studies on the pollution potential of mine waste residue deposits in the
gold and coal mining industry. Such studies require the integration of water balance models with
geochemical models. Extensive reviews of local and international literature on this topic have also been
previously undertaken and it can therefore be stated, with confidence, that there is a lack of knowledge and
data on how to accurately develop the water balances for the various mine waste residue deposits.
Although mines do have rudimentary water balances for their waste residue deposits, on close inspection it
is almost always found that these water balances contain numerous unmotivated or unvalidated
assumptions. While these inaccuracies may be acceptable for an operational mine where the water balance
is being actively managed on a daily basis, these inaccuracies seriously hinder the ability to make reliable
predictions of post-closure pollution potential and to properly evaluate environmental management /
rehabilitation strategies.
For fine waste residue deposits, some form of water balance management normally occurs as water is
actively reclaimed and returned to the beneficiation plant and there is normally a need to monitor water
levels for dam safety reasons. However, there is no reliable information to describe how water flows
horizontally and vertically within the dams after decommissioning, or how rainfall is partitioned between
runoff, infiltration and evaporation, or how water tables within the dams change with time. A number of
geochemical assessment studies previously undertaken have shown that an accurate post-decommissioning
water balance is critical in obtaining accurate pollution predictions and that the general “rules of thumb”
approaches result in unacceptable errors in the assessment.
For coarse waste residue deposits, the state of knowledge is even poorer as water balances are not
traditionally undertaken for such facilities. These coarse waste residue deposits are generally in the
unsaturated state and exhibit preferential path flow. There is no reliable information available on how rainfall
is partitioned into runoff, infiltration and evaporation for these facilities, thereby again influencing the
accuracy of post-decommissioning pollution prediction studies.
The outcome of this proposed research project will, therefore, significantly enhance the ability to undertake
reliable and credible assessments of long-term pollution potential from the range of mine waste residue
deposits encountered in the gold mining industry.
The project was structured to address the following three major objectives:
1. To develop a procedure and methodology that can be used by mines, researchers and consultants in
developing water balances for gold mine waste residue deposits.
2. To ensure that the outcome of the research is transferred to persons active in the field of pollution
assessment from gold mine residue deposits.
3. To provide material that can be used by tertiary educational institutions to incorporate into the standard
courses, thereby ensuring that new graduates have the knowledge to undertake accurate water
balances of such facilities.
1-1
1. Task 1. Identify study sites: Provision was made for consultation with mining companies and DWAF and
site visits to 6 mines to evaluate potential mine residue deposits that would serve the purpose of the
research project.
It is known that the water balance of a mine residue deposit is influenced by the depositional history of
the deposit and its physical characteristics (dimensions, particle size distribution, differential particle
deposition arising from different operational strategies, etc.). From the perspective of this research
project, distinction was made between the following different types of deposits that would be
endeavoured to be included in the research project:
Gold tailings dam constructed with traditional day/night paddocks and beaching to a central
penstock.
Gold tailings dam constructed through cycloning.
Gold waste rock dump constructed by end-tipping of waste off a conveyor.
In evaluating the suitability of various sites, preference was given to sites that already had installed
piezometers in their residue deposits with good historical records. It was intended to select 3 sites
representing the 3 different types of facilities listed above, with facilities that have recently been
decommissioned. This task is described in Chapter 2.
2. Task 2. Implement monitoring & experimental programmes: This is the biggest and most important
component of the project as it provides the critical data required to meet the primary project objective.
Provision was made for experimental work on site, residue deposit sampling and monitoring of
piezometers.
On the basis of the survey undertaken in Step 1, appropriate monitoring and experimental programmes
were designed for each specific site. Although the precise requirements for each site depended on the
site characteristics, it was deemed necessary to obtain information on the following parameters:
Design and construction details for each site, including details for any underdrainage systems that
are installed.
Construction history for each facility, including different deposition techniques that may have been
applied during the life of the facility.
Physical dimensions and volumes for each facility.
Representative samples to be taken for particle size analyses.
In-situ permeability tests in pre-selected locations.
Volume of seepage from underdrains or seepage drains.
Rainfall records at each facility.
Water levels in and around each facility.
Specialised test programmes aimed at simulating different rainfall records and recording runoff
and infiltration values on test plots.
3. Task 3. Assess data and run models: This task was largely office based and involved ongoing critical
review and assessment of collected data. Water balance modelling was undertaken and calibrated using
site data.
The data obtained during Step 2 was collated and assessed on an ongoing basis. The assessment was
undertaken in different ways such as:
1-2
This task is described in Chapters 4 and 5.
4. Task 4. Develop water balance & monitoring methodologies that should be applied for the different types
of mine residue deposits. On the basis of the work undertaken in Steps 1, 2 and 3 above, it was intended
to develop a methodology that sets out the following:
A methodology that can be used by mines / regulators / consultants to assess the water balance
(operational and post-closure) for the different types of mine residue deposits.
Recommendations on data collection requirements for water balance modelling and long-term
monitoring for validation purposes.
5. Task 5. Prepare lecture modules: This task is largely office based and was planned to involve a major
input from the University of Venda academic staff. The results of the research project were intended to
be used to develop a lecture series on how to prepare water balances for different types of mine residue
deposits.
This task was not met due to budgetary constraints and lack of appropriate expertise at Venda University
who were planned to undertake this task. Funds were reallocated (with agreement from the Project
Reference Group and the WRC) to extend the monitoring programme. Whereas it was originally
intended to complete monitoring by September 2004, funds were reallocated to enable monitoring to
continue till May 2005.
6. Task 6. A one-day technology transfer workshop was originally planned to convey the outcome of the
research to interested stakeholders. This workshop was not held (with agreement by the WRC, the funds
were reallocated to enable monitoring to continue till May 2005).
7. Task 7. Prepare project reports: The findings of the project have been documented in various project
progress reports submitted to and discussed at Reference Group meetings as well as in this final
research report.
A number of progress reports have been submitted to the WRC. This report is the final detailed research
report that presents all the collected data and the recommended water balance and monitoring
methodologies. This report addresses all the various tasks and deliverables as described in Section 1.3
above.
The University of Venda for Science and Technology (UNIVEN) is one of PHD’s major collaboration
institutions for capacity building in South Africa. One student – Mr B Maiyona from UNIVEN used this
opportunity to complete a Master degree through this project. Dr Bisrat Yibas and Professor Jason Ogola
were the supervisors.
Additional capacity building initiatives that were included and met in this project are the following:
This project required a fair amount of field and laboratory testwork to generate data to support the
fundamental modelling studies. This created an ideal opportunity for capacity building involving a full-time
post graduate student. Through this process a student (Mr Bethania Maiyona) was recruited to fulfil the
requirements of certain aspects of this project. Mr Maiyona was housed at PHD’s offices in order to provide
him the full opportunity to interact with the professional staff and to also obtain exposure to the wide range of
research and consulting activities undertaken by PHD.
1-3
CHAPTER 2
SITE SELECTION
2.1 OBJECTIVES
The prime objective of the site visits is to evaluate the suitability of various gold mine residue deposits and
identify sites that meet the criteria as stipulated in the research proposal for use in the research programme.
It is known that the water balance of a mine residue deposit is influenced by the depositional history of the
residue deposit and its physical characteristics (dimensions, particle size distribution, differential particle
deposition arising from different operational strategies, etc.). From the perspective of this research project,
the project team could distinguish the following different types of deposits that should be included in the
research project:
4. Gold tailings dam constructed with traditional day/night paddocks and beaching to a central penstock.
5. Gold tailings dam constructed through cycloning.
6. Gold waste rock dump constructed by end-tipping of waste off a conveyor.
In evaluating the suitability of various sites, preference was given to sites that already have installed
piezometers in their residue deposits with good historical records. At the outset of the project it was intended
to select 3 sites representing the 3 different types of facilities listed above, with facilities that have recently
been decommissioned. However, suitable type 2 tailings dams were not available and hence the project
team had no choice but to use only the type one tailings dam.
2.2.1 Criteria
It was proposed that six potential sites were to be identified and visited, from which 3 sites representing the 3
different types of facilities listed in section 2.1 would then be selected.
The following criteria were used for a mine residue deposit to be selected as a study site:
On the basis of this site selection exercise, appropriate monitoring and experimental programmes were later
designed for each specific site. Although the precise requirements for each site depend on the site
characteristics, it was necessary to obtain information on the following parameters and preference was given
to sites that had data available:
Design and construction details for each site, including details for any underdrainage systems that were
installed.
Construction history for each facility, including different deposition techniques that may have been
applied during the life of the facility.
Physical dimensions and volumes for each facility.
Volume of seepage from underdrains or seepage drains.
Rainfall records at each facility.
Historical water levels in and around each facility.
Potential tailings dams and waste rock dumps to be used for this project were identified by approaching
several mining companies. Although not all the mining companies approached were willing to participate,
Anglogold and Goldfields agreed to make some of their facilities available.
2-1
The members of the project team visited the available sites while the sites that were visited and the dates of
visits are listed in Table 2.1.
The site visits were conducted to get an understanding of the area and the specific facility as well as to find
out what data are available for each facility and its surrounding area. The following issues were addressed
for each site:
Familiarise the project team with the sites and the mine personnel responsible for the sites.
Conduct a preliminary assessment of the sites with respect to:
the nature of the residue site,
the depositional history
the physical characteristics of the facility (size, height, volume/mass deposited per day)
Identify data availability or gaps regarding:
the underlying soil and rock strata;
the presence of clay layers underlying the tailings;
water level data from boreholes surrounding the facility;
the physical characteristics within the residue deposit (grain size distribution, porosity, moisture
content);
the surface topography of the surrounding area and beneath the deposit;
precipitation and evaporation;
infiltration, run off, leakage into deeper soil layers, seepage;
volume of water returned from the residue deposit;
hydrogeological features such as seepage points;
piezometers – distribution and historical records;
presence and position of underdrains; and
Identify possible sites for experimental set-up and measurements
The location of the sites that were visited as well as the contact persons at each site are listed in Table 2.2
and shown in Figure 2.1.
2-2
Table 2.2: Location of visited gold residue deposit sites and contact persons
Mining company Site Name Nearest Large Contact person and responsibility
Town
Tailings Dams
Goldfields, Beatrix No 1 Welkom Hennie Pretorius, Environmental Manager
Welkom
Anglogold, Vaal West Extension Klerksdorp Pieter Labuschagne, Environmental Manager
River Operations West Extension Dries Terblanche, Data
East Dam Harry Scramm, GIS data
Anglogold, West Old North Complex Carletonville Simone Barnard, Environmental Manager
Wits Operations
Anglogold, ERGO Brakpan Brakpan Pieter Labuschagne, Environmental Manager
Brakpan Fred Rautenbach, Dam Foreman
nd
Hennie Barrington,2 Foreman
Con Sebbagha, Engineering Manager
Hennie Geldenhuys, Environmental
John Austin, Manager, Knight Piesold Engineer
Goldfields Kloof No 1 Westonaria Phillip Jacobs, Environmental Manager
Venterspost No 1 Phillip van Rensburg, Frazer Alexander (data)
Venterspost No 2 George Jones, ECMP (data)
Driefontein No 3A Philip Woodhouse, Environmental Manager, Tom
Badenhorst, Dam Foreman
Waste Rock Dumps
Anglogold, Vaal Tao Lekoa Klerksdorp Pieter Labuschagne; Environmental Manager
River Operations Kopanang Dries Terblanche, Data
Harry Scramm, GIS data
During the site visit and discussions with mine personnel it was found that cycloning is not widely used as a
method of deposition of gold tailings in South Africa. The Goldfields group did not have any tailings dams
available that were constructed through the cycloning method while Anglogold only had two tailings dams
that had been constructed by means of cycloning method for the entire life of the tailings dams.
2-3
Discussions with Anglogold mine staff and tailings dam specialists indicated that the current practice was to
change to the cycloning method as this method provides more stability due to the separation of fine and
super fine particles. The trend is to deposit, by way of cycloning, on top of an existing dam that has been
constructed in the traditional manner. This means that the lower portion of the tailings dam, which has been
constructed by the paddock deposition method, will have a different particle size distribution than the upper
portion of the tailings dam constructed by cycloning. These differences need to be considered in the water
balance if such a site is studied.
Tailings dams that have been constructed using a combination of the two methods of deposition are:
The ERGO Brakpan tailings dam is the largest gold tailings dam in the world and has been operational since
1985. The footprint of this tailings dam is 800 ha in size and the perimeter is 11 km. The top surface of the
dam is approximately 550 ha, while at the highest point the dam is 80 m above surface. Figure 2.2 shows
the ERGO Brakpan tailings dam as viewed from the stormwater dam.
Figure 2.2: The ERGO Brakpan tailings dam as viewed from the stormwater dam
Approximately one third of the dam on the western side is situated on dolomite, while the rest of the dam is
underlain by weathered dolerite and dolerite sill. The lower portions of the side slopes are vegetated for
most of the tailings dam perimeter (Figure 2.2).
Adjacent to the dam is the footprint of the old Withok tailings dam which is currently being rehabilitated.
Crushed dolerite from this footprint is mixed with turf and used as cladding on the slopes of the ERGO
Brakpan tailings dam. It is reported that vegetation grows easily on this cover material.
Water is pumped from the pool of the dam to various other sites for different purposes. The pool also
receives clean water from the plant from time to time after storms when the plant has an excess of water.
This makes calculation of the water balance for the ERGO Brakpan tailings dam very complicated.
According to the information obtained from the mine personnel, tailings are deposited on the dam at a rate of
3
approximately 120 Mℓ/day. The pool area on the dam is estimated to contain 20 million m of water. The
tailings dam has no underdrains. There are pollution paddocks to catch spills and toe drains below the
pollution paddocks. The toe drains intercept shallow seepage and transport the seepage to the evaporation
dam and the return water dams. The cycloning method of deposition was used for the entire life of the ERGO
Tailings Dam.
2-4
Available data
Table 2.3 gives information obtained during the site visit and indicates what data are available for the ERGO
Brakpan tailings dam.
Table 2.3: Available data for the ERGO Brakpan tailings dam
Flow from the toe drains into the evaporation dam is measured as well as water distributed to various other
areas like the plant, ERPM, Daggafontein, Withok and to evaporation spray bars.
The West Extension tailings dam is situated on the border of two distinct geological areas. The eastern part
of the tailings dam is situated on chert rich dolomite where the water level of the surrounding area is
approximately 20 m below surface. The western part of the tailings dam overlies bedrock dolomite and the
water level in this area is shallow.
Although the dam was originally constructed using the traditional Paddock method, it was later changed to
the cycloning method (Figure 2.3).
2-5
Figure 2.3: Cycloning method of deposition used at West Extension tailings dam
Available data
All data are available from the mine and Table 2.4 gives information obtained during the site visit and
indicates what data are available for the West Extension tailings dam.
Table 2.4: Available data for the West Extension tailings dam
2-6
2.3.3 Gold tailings dams constructed by the traditional paddock method
The Beatrix No 1 tailings dam has been in operation for 20 to 25 years. A new tailings dam has been
constructed adjacent to the Beatrix No 1 tailings dam. The new dam has been in operation for the last two
years and since then the Beatrix No 1 tailings dam has received smaller amounts of tailings. The Beatrix No
1 tailings dam will be operational for the next 5 years in conjunction with the new dam. Figure 2.4 shows the
northwestern corner of the Beatrix No 1 tailings dam.
Figure 2.4: Beatrix No 1 tailings dam showing the vegetated western slope (right) and the
unvegetated northern slope
The Beatrix No 1 tailings dam has been vegetated on the southern and western slopes, but due to
instabilities on the west slope, irrigation was terminated. Diffuse seepage is observed on the western side of
the tailings dam. It is possible that the dolerite sill underlying the western side of the dam prevents
infiltration, thereby resulting in the seepage observed (Pretorius, Personal communications).
Return water from the penstock is discharged directly into the evaporation dam
Available data
The mine agreed to make data available once the selection of this site for the project is confirmed. Table 2.5
gives information obtained during the site visit and indicates what data are available for the Beatrix No 1
Tailings dam.
2-7
Table 2.5: Available data for Beatrix No 1 tailings dam
The mine estimates that 30% of the water deposited on the tailings dam is currently discharged to the
evaporation dam. It may be possible to calibrate the pipe from the penstock to calculate amounts and to
build structures to measure flow from the two sides of the solution trench.
Three tailings dams and two waste rock facilities were visited at Anglogold’s Vaal River Operations. These
facilities are the West Complex tailings dam, the West Extension tailings dam, the East tailings dam, the Tao
Lekoa waste rock dump and the Kopanang waste rock dump (Figure 2.5).
The West Complex tailings dam consists of several compartments at different heights, the highest of which is
approximately 55 m. The tailings dam has been vegetated on the slopes and one of the decommissioned
compartments is fully vegetated. Most of the tailings dam is situated on bedrock dolomite, but the western
side of the tailings dam overlies lava.
2-8
The toe paddocks from where the toe seepage water and side slope runoff evaporates. No water was
observed in the paddocks during the site visit.
Available data
The project team was informed that all required data are available from the mine. Table 2.6 gives information
obtained during the site visit and indicates what data are available for West Complex tailings dam.
Figure 2.5: Location of the Vaal River Operations facilities that were visited
Table 2.6: Available data for the West Complex tailings dam
2-9
2.3.3.3 East tailings dam, Vaal River Operations (Anglogold)
Available data
All data are available from the mine. Table 2.7 gives information obtained during the site visit and indicates
what data are available for the East tailings dam.
The Kloof No 1 Tailings Dam has been decommissioned since 1997. This tailings dam is of a type called
"side-hill ring-dyke" which is constructed on a hillside so that the closing side of the impoundment is formed
by the slope of the natural ground. This means that water will flow from the higher side out towards the other
sides of the tailings dam, which is different from the flow paths in most other tailings dams which are
commonly a proper ring-dyke type impoundment. Kloof No 1 Tailings Dam covers a surface area of
approximately 112 ha and is vegetated on all sides and on the top of the dam.
Available data
All data for the Kloof No 1 tailings dam are available from Mr George Jones of Environmental Civil Mining
Projects (ECMP). Table 2.8 gives information obtained during the site visit and indicates what data are
available.
2-10
Table 2.8: Available data for the Kloof No 1 tailings dam
Venterspost No 1 tailings dam is a relatively small dam that has been decommissioned many years ago
when the No 2 dam was started and the dam is vegetated with grass. Water level data are not available for
this facility. Since there are data available for the nearby Venterspost No 2 tailings dam,
it was decided not to spend any further time or effort on the No 1 tailings dam.
The Venterspost No 2 tailings dam is approximately 60 m high and is vegetated on the slopes and on top.
The penstock is still in place to redirect storm water. The water level in this dam has dropped below the
piezometer level, but historical data are available.
Available data
All data for the Venterspost No 2 tailings dam are available from the mine. Table 2.9 gives information
obtained during the site visit and indicates what data are available.
2-11
Table 2.9: Available data for the Venterspost No 2 tailings dam
The Old North Complex tailings dam consists of six compartments at different levels. Water level data are
not available for this facility. The oldest of these compartments dates back to 1959. From 1976 to 1997, the
No 2 compartment was used as a return water dam. The No 3 compartment is the largest and highest.
During December 1996 a cut-off trench was installed along the southern toe of No 2 compartment and the
amount of seepage is measured. The different compartments of the Old North Complex tailings dam have
been decommissioned at different times.
Available data
All data for the Old North Complex tailings dam are available from Anglogold. Table 2.10 gives information
obtained during the site visit and indicates what data are available.
2-12
Table 2.10: Available data for the Old North Complex tailings dam
The Driefontein No 3A Tailings Dam is approximately 30 m high, is vegetated on the slopes and the mine is
in the process of establishing vegetation on top of the dam (Figure 2.6). The penstock has collapsed since
decommissioning and the water levels in this dam have dropped below the piezometer level. Historical
piezometer data are available.
Available data
Data for the Driefontein No 3A Tailings Dam are available from the Environmental Manager. Table 2.11
gives information obtained during the site visit and indicates what data are available.
2-13
Table 2.11: Available data for the Driefontein No 3A tailings dam
2.3.4.1 Tao Lekoa waste rock dump, Vaal River Operations (Anglogold)
The Tao Lekoa waste rock dump is situated west of Orkney in close proximity to the Vaal River. The
geology consists of lava overlain by alluvium clays. This waste rock dump is watered for stability and for dust
prevention. The amounts of water used on the dump can be obtained from the shaft engineer. Seepage
water is observed on the southern and eastern parts the dump. These areas are close to the Vaal River and
topographically low.
Water level data is available from the four boreholes surrounding the dump. The regional groundwater level
is shallow at approximately 2 m below surface. Figure 2.6 shows a part of the Tao Lekoa Waste Rock dump.
The Kopanang Waste Rock dump is located at No 9 shaft on the Free State side of the Vaal River. This
dump is not watered but the waste rock is damp when deposited. No seepage was observed during the site
visit, but it has occurred during the wet season. The dump is approximately 80 m high. Large erosion gullies
are observed on the slopes of the older parts of the dump. There is a visible grading of material with fine
material at the top and boulders at the foot of the dump. Four boreholes in the vicinity of the dump are
measured for water levels. The regional groundwater level is at approximately 10 m below surface. Figure
2.8 shows the Kopanang waste rock dump.
2-14
Figure 2.7: Tao Lekoa waste rock dump showing how the dump is watered
The original proposal proposed to use three facilities, i.e. two tailings dams and one waste rock dump as
study sites for water balance studies. The two tailings dams to be used would represent the two most
commonly employed construction methods of tailings facilities, namely the cycloning method of construction
and the traditional day /night paddock method of construction.
During the discussions with the gold mining companies and subsequent site visits to identify suitable sites it
was found that cycloning is not widely used as a method of deposition of gold tailings in South Africa. The
Goldfields group does not have any tailings dam available that are constructed through cycloning while
Anglogold only has two tailings dams that have been constructed by means of the cycloning method for the
entire life of the tailings dam.
The current trend is to change the method of deposition from the traditional paddock method to the cycloning
method, by the deposition of new slimes on top of an existing dam that has been constructed in the
2-15
traditional paddock manner. This is currently being done at Vaal River Operations for the West Extension
tailings dam and this process will start soon for the Mispah dam. The West Extension tailings dam is a
possible study site, but is not ideal because the two different methods that have been used to deposit tailings
will complicate modelling of the water balance. Furthermore, the site is currently in operation and does not
comply with the requirement that decommissioned sites should be studied.
The ERGO Brakpan tailings dam has been constructed through cycloning. Although this facility has a good
record of water level data and flow data, the water balance is complicated by the distribution of water from
the pool area to various other facilities. This facility is the largest gold tailings dam in the world and due to the
size of the dam, there is a possibility that the water balance cannot be compared to that of smaller facilities.
Part of the dam is constructed on a previous tailings dam, which further complicates the water balance for
this facility and makes it different from other tailings dams. The ERGO Brakpan Tailings Dam is therefore not
an ideal site for developing a water balance and additionally, the site is currently in operation which does not
comply with the requirement that decommissioned sites are studied.
Although Anglogold has another tailings dam at Daggafontein on the East Rand which is constructed by the
cycloning method, this facility was not made available to the project team. There are, therefore, no
decommissioned tailings dams constructed by the cycloning method in South Africa that could be used for
this investigation.
Six gold tailings dams constructed by the traditional paddock method were visited, of which two are
operational. Although the proposal document stipulated that decommissioned sites should be chosen, the
mines made their operational tailings dams available and only after more negotiations were the
decommissioned dams also made available. Only the decommissioned tailings dams are considered as
potential study sites for this project.
Based on the stipulated selection criteria and information gathered during the site visits, the West Wits Old
North Complex (Anglogold), the Driefontein No 3A (Goldfields) Tailings Dams, and the Kopanang waste
rock dump (AngloGold Ashanti) have been selected for the study. A summary of the information for the
selected sites is given below.
Although this is a complex dam composed of a number of compartments, it is envisaged that the current
water table below this complex is in equilibrium and can be assumed as one hydrological system. This
complex has information regarding the history of the residue deposit but has no water level data available.
However, Anglogold is willing to install piezometers to support the project.
The Driefontein No 3A Tailings Dam is the most recently decommissioned dam (since 2000), which makes it
attractive for studying the water balance. Although the water level is currently below the piezometer depths,
there is a possibility that the water level is still decreasing and has not yet reached equilibrium. Driefontein
mine agreed to consider installing piezometers to support the project. Due to the relatively recent
decommissioning, this site is one of the best potential study sites.
The two waste rock dumps visited differ in the fact that one is watered and the other is a dry waste rock
dump. From conversations with Environmental Managers at different mines, it seems that most waste rock
dumps are not watered. It was therefore decided that the dry Kopanang waste rock dump be used for the
water balance study.
2-16
CHAPTER 3
MONITORING AND EXPERIMENTAL PROGRAMMES
3.1 INTRODUCTION
The monitoring and experimental program discussed herein is the most important component of the project
as it provides the critical data required to fulfil the primary project objective. Originally the completion of this
component of the research was scheduled for September 2004. However, because not all the objectives of
the task were met due to technical difficulties, the task was extended until March 2005 to cover the
2004/2005 wet season.
3.2. METHODOLOGY
The experimental set-up, data collection and monitoring programme is subdivided into the following sub-
tasks:
Experimental site selection within each selected gold mine residue deposit.
Design a detailed experimental set-up, monitoring and sampling program.
Sampling.
Installation of probes and apparatus.
In-situ testing.
Laboratory tests (testing probes and sample testing).
Data analysis (see Chapter 4).
Conducting a proper and holistic water balance of any porous media such as mine residue deposits requires
measurements of unsaturated and saturated hydraulic conductivity, water retention characteristics, particle
size distribution of the material, matrix suction and weather data such as rainfall, wind speed and direction,
temperature, solar radiation and humidity.
Rainfall
Evaporation;
Transpiration
Storage Drainage
dynamics Tipping bucket
Lateral flow Lysimeter and soak-away
Runoff plot
Drainage
Figure 3.1: Schematic experimental set-up for water balances of gold tailings dams
3-1
3.3. DATA REQUIREMENTS AND INSTRUMENTATION
The flux direction, whether downwards drainage to the phreatic surface or groundwater or upwards uptake
back into the water balance zone, will be inferred from the hydraulic gradients measured near the 10 m
depth. In order to have confidence in the measured data, all the different measurements were performed on
a specific spot, where possible near a piezometer.
rainfall,
evaporation (indirectly),
storage dynamics,
lateral flow,
drainage, and
run-off measurement on the side slopes.
It is essential to have weather data for the period over which other measurements are made. At the
beginning of the project it was planned to install meteorological stations within a secure area at either
Goldfields Driefontein or Anglogold West Wits as close to the dam as possible and another one at the
Anglogold Vaal River Operations, if there is no local station in that area. However since there are
meteorological stations near these project sites it was decided to install only automated rain gauges on each
of the facilities and to make use of the weather stations of the mines and Weather Bureau data.
3.3.1.1 Rainfall
Three MC Systems tipping bucket rain gauges were installed for on-site measurement of rainfall as rainfall is
very site specific. Other rainfall records in the area were obtained and used for in filling, checking and
extension of the record.
A full weather station is situated on the West Wits Mine within 2-3 km of each dam. Data have been
requested and obtained from the operator and have been deemed sufficient for temperature, humidity, wind
speed and direction and solar radiation.
3.3.1.2 Evaporation
Actual evaporation is estimated from lysimeters as well as from the simulated water balance, calibrated
against observed soil water status.
Solar radiation, temperature, humidity, wind speed and direction have also been obtained from the South
African Weather Bureau and from the mines' weather stations.
3.3.2. Runoff
Simple runoff plots with flow concentration panelling at the downstream end were provided at each of the
tailings dams and at the waste rock dump. Measurement of discharge was effected by collection of the
runoff in a drum. A tipping bucket mechanism was provided at one of the sites to assess the
infiltration/runoff dynamics.
3-2
3.4. STORAGE DYNAMICS AND DRAINAGE
The following tasks were planned for measurement of storage dynamics and drainage of the tailings dams:
Monitor the energy status of water within the profile using nests of probes at proposed depths of 200,
500, 1000, 2000, 8000 and 10000 mm.
Two nests near the surface and one nest on the side slopes would be installed to assess both vertical
and lateral gradients.
Collect sufficient sample material to determine the hydraulic characteristics of the materials in the
laboratory.
These characteristics, together with direct tests on the probes would be used to determine the best
sensors for monitoring the profile water status.
Test the different probes to be used including the following:
TDR and capacitance probes (volumetric water content),
Tensiometers,
CS229 heat dissipation, and
WATERMARK resistance probes (pore water tension).
Provided the energy status is determined to be within range, the probes at 8000 and 10000 mm would
comprise the deep tensiometers developed at the University of Natal.
Data recording is automated with HOBO loggers and was initially downloaded using a laptop computer.
Downloading of the data was thereafter done automatically using a HOBO shuttle.
Examination of material characteristics and deposit surface conditions during the initial visit to determine
the method of installation of simple lysimeters on the top of the deposits.
Install lysimeters by one of the following methods:
Excavate and install a tank (approximately 2 m in diameter) with base, repack material into the tank
and measure outflow locally with a tipping bucket mechanism and soak-away or manual pumping.
Excavate around a centre column of undisturbed material, encase the column in plastic and brace
the column. Slide a stainless steel base sheet under the column and seal to the sides and to the
base. Raise the column if possible and install an outflow base of coarse material and geofabric.
Measure outflow as above.
Drive a large open-ended tank into the profile and insert a base as above.
Install a simple lysimeter, comprising a plastic water tank (2 m x 2 m) for assessing the evaporation
and drainage components.
Drainage outflow would be measured with a tipping bucket mechanism on an adjacent side slope or
below the tank.
The lysimeter would also be installed with TDR probes for manual readings of volumetric water content.
Soil temperature up to 100-200 mm depth would be measured using temperature loggers. Four
temperature sensors per facility were planned.
3.4.1.1 Tensiometers
The tailings were instrumented with pore water suction sensors (tensiometers) that are used for soil moisture
sensing. A total of 8 shallow tensiometer nests were installed in the two tailings dams. Each nest consists of
2000 mm, 1000 mm, 500 mm and 250 mm long automatic recording tensiometers. The tensiometer nests
were installed as follows:
3 tensiometer nests at the 3 experimental sites (edge, middle and ex-pool area) in the Old North
Complex tailings dam; and
3 tensiometer nests, comprising four automatic recording tensiometers, at the 3 experimental sites
(edge, middle and ex-pool area) in the Driefontein No.3A tailings Dam.
The three sensor stations at each dam were located in a profile along the beach, from the outer edge
near the day wall to the penstock. At each nest, the tailings pore water capillary pressure was recorded
3-3
at 2000 mm, 1000 mm, 500 mm and 250 mm below the surface. The tensiometer pressure transducers
were connected to a logger.
A nest of 4 mini tensiometers was installed in Driefontein tailings dam to monitor the energy status
outside and inside of the lysimeter.
As part of the revised proposal a total of 4 deep tensiometer nests were installed at the edge and middle
sites of the two tailings dams. Each nest consisted of 500 mm, 2000 mm, 4000 mm and 6000 m long
tensiometers.
Tensiometers were installed at three stations on each tailings dam surface in a transect from the edge to the
penstock. The tensiometers comprise a porous ceramic cup, filled with de-aired water in hydraulic contact,
via the ceramic, with the pore water of the tailings material. The water inside the ceramic reduces in pressure
in response to the pore water suction. This suction is measured using a differential pressure transducer,
hydraulically connected to the ceramic water via a 6 mm hydraulic hose. The hose is encased in a PVC tube
and the ceramic and tube assembly is inserted into the tailings by augering a 22 mm diameter hole.
The tensiometers were installed to depths of 2 m below surface. During October 2004, these tensiometer
nests were supplemented with deep tensiometers, which reached a depth of 6 m below surface.
In order to measure soil water tensions beyond the range of tensiometers, WaterMark sensors were tested
and calibrated against tailings material in the laboratory. Initial tests were performed on a tailings material in
which both WaterMark and tensiometer sensors were inserted. The WaterMark sensor output was closely
correlated to that of the tensiometer.
3.4.1.4 Lysimeters
Two lysimeters, one at each dam, were installed to assess drainage and evapotranspiration fluxes. The
value of lysimeter measurements to assess the evapotranspiration and drainage fluxes is dependent on the
ability of the lysimeter profile to mimic that of the in-situ profile. In the material, the depth of the lysimeter and
the hydraulic characteristics of the material determine the ability of the lysimeter. Mini tensiometer nests
were installed to monitor the energy status in and outside of the lysimeters at Driefontein and Old North
Complex.
The methodology and procedure of the lysimeter installation is according to the design given in Figure 3.2. A
340 mm diameter plastic tube was pressed into an excavated column of tailings and a base installed on the
resultant column. A 22 mm diameter outflow pipe was connected between the base and a drainage collector
tube of 110 mm diameter. Miniature tensiometers were installed into the lysimeter column and into the
undisturbed profile adjacent to the lysimeter (Figure 3.2).
Runoff plots 10 m long and 2.25 m wide were installed on the tailings dams during October 2004. At the Old
North Complex tailings site, two runoff plots were installed on the side slopes and two on the top surface.
The side slope plots were equipped with a tipping bucket mechanism to record runoff rates automatically to a
HOBO event logger. The beach slopes on the top surface, however, were not steep enough to allow for the
installation of runoff tipping bucket mechanisms. Hence, the runoff plots on the tailings top surface were
equipped with flow splitters and a tank to collect a fixed proportion of the runoff volume. At the Driefontein
dam two runoff plots were installed on the slope surface. No runoff plot was installed on the top surface of
the Driefontein dam as the surface is furrowed by the mine for dust prevention purpose.
3-4
Mini tensiometers
Plastic
lysimeter Excavation
Tailings
Galvanised base
Coarse sand
Geotextile
Outflow pipe
Measurement of in-situ saturated and unsaturated hydraulic conductivity at four top surface sites ranging
from the edge to the centre of the deposit at each tailings dam was planned. Tension infiltrometers as well as
Guelph permeameters were used for measurement of unsaturated hydraulic conductivity and double ring
infiltrometers for the saturated hydraulic conductivity measurement.
The Kopanang waste rock dump was instrumented with a drainage lysimeter of 8x8 m in plan and 2 m deep.
The lysimeter base outflow discharged into a tipping bucket mechanism, which recorded the outflow. An
automatic rain gauge was installed on the surface of the waste rock dump to determine the input to the
lysimeter surface.
For the waste rock dump, a lysimeter was initially designed to comprise 18 of 2x1 m pans located on a 6x6 m
base at 2 m below the surface (Figure 3.5). This system would cost approximately R10 000 to install and
would allow for preferential flow assessment.
3-5
The outflow pipe was led to the outer slope of the waste rock dump, where a tipping bucket gauge was
installed to measure the drainage outflow.
An event logger was installed on the tipping bucket to record the outflow rate. A tipping bucket rain
gauge was installed on the dump surface next to the lysimeter. The approximate location of the
lysimeter on the dump surface is show in Figures 3.3 and 3.4.
Figure 3.3: Layout of the Kopanang waste rock dump, showing typical cross-sections for the
location of the lysimeter
3-6
19mm aggregate wrapped in geofabric
8000mm
Tipping bucket flow gauge
Plastic liner
Waste rock
Figure 3.5: Cross-section of the lysimeter installation in the Kopanang waste rock dump
3.5.1. Sampling
Samples from the two tailings dams were collected for studies of hydraulic properties and physical
characterisation of the tailings materials. Undisturbed vertical samples were also taken at every site and level
at which in-situ hydraulic conductivities were measured.
Tailings samples, extracted using thin walled tubing, were prepared for laboratory saturated and unsaturated
hydraulic conductivity measurements. Vertical and horizontal undisturbed and disturbed samples were
collected from the exact locations of the in-situ tests identified at each dam. These samples were then
submitted for laboratory analysis of physical properties (particle size distribution, bulk density), water
retention and saturated and unsaturated hydraulic characteristics in the UKZN soil Laboratory.
Samples were collected from the Kopanang waste rock dump to assess the particle size distribution and
hydraulic properties. The following 3 sets of samples were collected from the Kopanang Waste Rock Dump:
All the samples were sent to the Hydrology Laboratory at the School of Bioresources Engineering and
Environmental Hydrology (SBEEH), University of KwaZulu-Natal (UKZN), Pietermaritzburg where selected
samples were tested for particle size distribution at the UKZN laboratory as follows:
Determine particle size distribution (PSD) by sieving and hydrometer analysis
Bulk density,
Determine the water retention characteristics (WRCS) using an unconsolidated sample controlled
outflow cell together with conventional pressure pot extraction.
Conduct saturated hydraulic conductivity in vertical permeameter tests and the unsaturated hydraulic
conductivity using vertical short column analyses and Bruce-Klute diffusivity tests.
The characterisation of the materials was performed with the following objectives:
3-7
To develop pedotransfer functions to relate simple characteristics, such as the texture, bulk density and
disposal method, to the hydraulic characteristics.
Provide details of the variation of hydraulic characteristics from the edge to the centre of the deposit.
The hydraulic characteristics could be used for the following:
As input data in the HYDRUS2D model to determine the water balance of the unsaturated and
saturated components of the deposit.
To compare simulations with observations in order to verify the physics of the water balance in the
vertical and lateral directions.
To determine the dominant processes in the water balance.
To define areas for future research and to extend the water balance using long-term climate records.
3.6 SUMMARY
The following tasks which were in line with the agreed experimental set-up, data collection and monitoring
program and the additional tasks specified in the revised scope of Task 2, presented to and agreed by the
WRC Reference Group have been implemented:
Sampling and analyses: All the required samples including undisturbed columns of samples from all
experimental sites were collected and all samples were characterised for their hydraulic properties.
Instrumentation: Installation of different probes and apparatus were completed successfully and data
collection was successful.
In-situ measurements of saturated and unsaturated hydraulic conductivity measurements at depths of
2000, 1500, 1000, 500, and 250 mm in all of the experimental sites were completed for both Driefontein
and Old North Complex tailings dams. Poor tensiometer data generation which was initially encountered
was addressed by installing the following additional equipment:
Mini tensiometers together with the lysimeter at the Driefontein site,
Six surface and side runoff plots at both the tailings dams,
Four deep tensiometer nests (500 mm, 1000 mm 4000 and 6000 mm long tubes)
Laboratory analyses: The laboratory measurements of hydraulic properties such as hydraulic
conductivity and water retention characteristics were completed.
Tailings evaporation: The actual evaporation from the surfaces at both tailings dams, and transpiration
at the Old North Complex are clearly dominant factors in the water budget. This part of the study was
enhanced by:
Installing manual readable lysimeters,
Collection of a large vertical sample for laboratory testing of infiltration/redistribution behaviour and
Scintillometer measurements during the 2004/2005 wet season.
Rapid assessment of the infiltration redistribution process was instigated by introducing runoff plots
on the surfaces and slopes of the tailings dams as well as extracting large samples for testing in the
laboratory.
Substitution of some of the tensiometers with WaterMark sensors for monitoring of the energy status of
the tailings was implemented successfully and the response of the watermark sensors to capillary
pressure of the tailings was assessed.
Volumetric water contents were measured successfully with TDR100 in the tailings materials. However
the TDR100 apparatus was damaged and time constraints prevented installation of TDR probes.
The experimental and data collection task has been successfully implemented. The data generated by the
instruments and laboratory analyses were collated into a database for use in the project. The relationship
between the datasets from the different instruments have been analysed and the use of this data for the
water balance study is discussed in Chapters 4 and 5.
3-8
CHAPTER 4
DATA ASSESSMENT
4.1.1 Rainfall
Rainfall data from the two tailings dams have been recorded continuously using the rainfall gauges installed
during October and November 2003. The rainfall data at each of these sites is available as automatic rainfall
records of 0.2 mm increments. Daily data are presented in Figures 4.1 to 4.3.
80
70
60
Daily Rainfall (mm)
50
40
30
20
10
0
Oct-03 Dec-03 Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
Figure 4.1: Daily rainfall data from the Driefontein tailings dam, October 2003 to May 2005
At the Driefontein site, missing data from 13 November 2003 to 14 February 2004 occurred due to an
extreme rainfall event flooding the equipment in November 2003. The record and all analyses assume a
February 2004 start.
4-1
80
70
60
Daily Rainfall (mm)
50
40
30
20
10
0
Oct-03 Dec-03 Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
Figure 4.2: Daily rainfall data from the ONC tailings dam, October 2003 to May 2005
At the Old North Complex (ONC) site, missing data occurred from 13 November 2003 to 15 February 2004
and 17 April 2004 to 26 June 2004. These data have been patched with records from the Potchefstroom
meteorological station.
80
70
60
Daily Rainfall (mm)
50
40
30
20
10
0
Oct-03 Dec-03 Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
Figure 4.3: Daily rainfall data from the Kopanang waste rock dump, February 2004-May 2005
The Kopanang rainfall station was established on 14 February 2004 and has missing data from 23 May 2004
to 13 July 2004. These data were patched from the Potchefstroom meteorological station data.
4-2
ONC
1000
Kopanang
Driefontein
800
400
200
0
Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
Figure 4.4: Cumulative rain data for the Driefontein and ONC tailings dams and the Kopanang waste
rock dump, February 2004 to May 2005
Meteorological data have been received for September 1997 to April 2002 from the Anglogold operated
station between the Driefontein and ONC tailings dams. This record was not continued for the period of the
experimentation at the tailings dams due to technical faults with the instrumentation. The measurements
were revived in February 2005 and these data were used to compare against the latter part of the
experimentation record. Additional data were requested from the SA Weather Services and compared
against the data recorded at the two tailings dams and the Kopanang waste rock site for the period January
to June 2005, (Figure 4.5). The ONC and Driefontein sites experienced higher rainfalls than were observed
at Kopanang for this period. The Carltonville rainfall closely correlated to the Kopanang data with the record
at Potchefstroom being the lowest recorded rainfall. The weather data for the Potchefstroom site includes
daily rainfall, maximum and minimum temperatures, humidity and wind speed. A MAP of 601 mm at the
Potchefstroom station was determined for the 28 year record. The meteorological data are presented in
Figures 4.6 to 4.8.
Potchefstroom
1000
Fochville
Carltonville
800
Driefontein
Cumulative rain (mm)
600
400
200
0
Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
Figure 4.5: Cumulative rain data for the Driefontein tailings dam and local met stations
(Potchefstroom – 0437104; Fochville – 0474899; Carltonville – 0474680), for February
2004 to May 2005
4.1.2 Temperature
Daily maximum and minimum temperature data for the Potchefstroom meteorological station are shown in
Figure 4.6 for the period that corresponds with the research project.
4-3
40
35
30
25
o
C)
20
Temperature (
15
10
0
Maximum daily
-5
Minimum daily
-10
Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
4.1.3 Humidity
The recorded values for morning and afternoon humidity at the Potchefstroom meteorological station are
shown in Figure 4.7.
100
90
80
70
60
RH (%)
50
40
30
20
08H00
10
14H00
0
Feb-04 A pr-04 Jun-04 A ug-04 Oct-04 Dec-04 Feb-05 A pr-05
Average wind speed data for the research period at the Potchefstroom meteorological station are shown in
Figure 4.8 below.
4-4
9
7
Average Daily Wind Speed (m/s)
0
Feb-04 A pr-04 Jun-04 A ug-04 Oct-04 Dec-04 Feb-05 A pr-05
Figure 4.8: Average wind speed (08H00, 14H00 and 20H00) for Potchefstroom
(February 2004 to May 2005)
Reference potential evaporation data were derived using the Linacre 1977 estimate for potential evaporation
(Schulze and Kunz, 1994) and are shown in Figure 4.9:
4-5
2200
2000 ETo
1400
1200
1000
800
600
400
200
0
Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
Figure 4.9: Cumulative reference potential evaporation (ETo) based on the Linacre 1977 equation
(Potchefstroom data) and Driefontein rainfall (February 2004 to May 2005)
Bulk densities have been determined on samples extracted with the thin walled tubing and analysed for
water retention. A selection of the results is shown in Table 4.1. Porosities have been estimated from the
bulk densities, using a value for particle density of 2.65 kg/m3.
PSD samples have been analysed for the ONC tailings and are presented in Figures 4.10 to 4.12. The silt
fraction is low, approximately 20%, while the clay size fraction is less than 5% in all samples. There are no
discernible differences with depth for the samples from the 0 to 600 mm depths of the ONC1 station.
However, there appears to be a small increase in the fine fractions from the edge to the middle station.
The PSD determinations show some disparity between sieve and hydrometer results over the silt range,
which is most likely caused by incomplete wet sieving through the two finest sieves. The silt and clay
fractions are therefore best derived from the sieve data alone, where these data sets do not match up at the
smallest sieve size. It should be noted that the more samples analysed, the better understanding of particle
size distribution across the dam and with depth.
4-6
100
0-120
120-220
80 220-600
60
% Finer 40
20
0
0.001 0.010 0.100 1.000 10.000
Particle Size (mm)
100
80
60
% Finer
600-800
800-1000
40 1000-1300
1300-1500
1500-1700
20
1700-1900
1900-2100
0
0.001 0.010 0.100 1.000 10.000
Particle Size (mm)
Figure 4.10: Particle size distributions for the ONC tailings material for the penstock site at depths 0-
600 mm (above) and 600-2100 mm (below)
100
400-500
550-700
850-1000
80
1000-1150
60
% Finer
40
20
0
0.001 0.010 0.100 1.000 10.000
Particle Size (mm)
Figure 4.11: Particle size distribution for the ONC tailings material: edge site
4-7
100
0-100
400-650
80 900-1000
1500-2000
60
% Finer
40
20
0
0.001 0.010 0.100 1.000 10.000
Particle Size (mm)
100
200-500
500-1000
1000-1500
80
1500-2000
60
% Finer
40
20
0
0.001 0.010 0.100 1.000 10.000
Particle Siz e (mm)
Figure 4.12: Particle size distribution for the Driefontein tailings material:
edge (above) and penstock (below)
100
sieve
hydrometer
80
60
% Finer
40
20
0
0.001 0.010 0.100 1.000 10.000
Particle Size (mm)
Figure 4.13: Particle size distribution for the Kopanang waste rock, sand size fraction: surface
The rock sample comprises 57% rock, 23% sand and 20% silt. The sand and silt fraction is classified as a
sandy loam.
4-8
4.2.2 Water retention characteristics
Water retention characteristics have been determined using a controlled outflow cell apparatus for capillary
pressure head range between 0 and 10000 mm and standard pressure pot extraction for the range between
10000 mm and 150000 mm according to the procedures described in Lorentz et al., 2001.
The van Genuchten relationship was fitted to water retention characteristic data (capillary pressure head
versus volumetric water content). The relationship is given by:
m
1
Se n
1 h
where
r
Se = effective saturation, and
s r
Apart from the higher porosity of the materials at the edge station compared to those at the middle station,
the retention characteristics of all the tailings samples are similar, with high residual water content ranging
from 0.10 to 0.20 (Table 4.2).
These values have been used in the simulation exercise using the HYDRUS 2D.
Typical water retention characteristics are shown for the ONC edge site for a sample taken between 0 and
120 mm in Figure 4.14.
Table 4.2: Summary of van Genuchten curve fit parameters for water retention characteristics
-1
Site (rep) Depth (mm) Porosity Residual WC α (mm ) n
ONC EDGE (1) 0-120 0.480 0.00 0.00100 1.30
ONC EDGE (1) 120-220 0.462 0.20 0.00010 1.95
ONC EDGE (1) 220-600 0.422 0.15 0.00010 2.00
ONC MIDDLE (1) 0-100 0.348 0.00 0.00050 1.20
ONC MIDDLE (2) 0-100 0.318 0.18 0.00012 4.40
DRIE EDGE (1) 0-100 0.452 0.12 0.00030 1.80
4-9
160000
0 mm
250 mm
140000
500 mm
2000 mm
Capillary pressure head (mm) 120000
100000
80000
60000
40000
20000
0
0.00 0.10 0.20 0.30 0.40 0.50
Volumetric water content
Figure 4.14: Water retention characteristics of the Driefontein tailings material: edge
20000
0 mm
18000 250 mm
500 mm
16000 2000 mm
14000
Capillary pressure head (mm)
12000
10000
8000
6000
4000
2000
0
0.00 0.10 0.20 0.30 0.40 0.50
Volumetric water content
Figure 4.15: Water retention characteristics of the Driefontein tailings material: middle
4-10
160000
250 mm
500 mm
140000
1000 mm
2000 mm
Capillary pressure head (mm) 120000
100000
80000
60000
40000
20000
0
0.00 0.10 0.20 0.30 0.40 0.50
Volumetric water content
Figure 4.16: Water retention characteristics of the Driefontein tailings material: penstock
160000
0-120 mm
140000 120-220 mm
250 mm
500 mm
120000
220-600 mm
Capillary pressure head (mm)
100000
80000
60000
40000
20000
0
0.00 0.10 0.20 0.30 0.40 0.50
Volumetric water content
Figure 4.17: Water retention characteristics of the ONC tailings material: edge
4-11
160000
140000 0-100 mm
100000
80000
60000
40000
20000
0
0.00 0.10 0.20 0.30 0.40 0.50
Volumetric water content
Figure 4.18: Water retention characteristics of the ONC tailings material: middle
160000
500 mm
1000 mm
140000
2000 mm
120000
Capillary pressure head (mm)
100000
80000
60000
40000
20000
0
0.00 0.10 0.20 0.30 0.40 0.50
Volumetric water content
Figure 4.19: Water retention characteristics of the ONC tailings material: penstock
4-12
4.2.3 Hydraulic conductivity characteristics
In situ measurements
Saturated and unsaturated hydraulic conductivity measurements have been conducted at the tailings dams
using double ring and tension infiltrometer apparatus. The hydraulic conductivities for the 250 mm and 500
mm depths at the Driefontein site are not significantly different. Conductivities range from 2.5 mm/h at 5 mm
capillary pressure head to 0.4 mm/h at a capillary pressure head of 190 mm.
4-13
Ksat (mm/h)
1 10 100
0
0.5
1
Depth (m)
1.5
2
Penstock
Middle
Edge
2.5
Ksat (mm/h)
1 10 100
0
0.5
1
Depth (m)
1.5
Penstock
Middle
Edge
2.5
Figure 4.20: Saturated hydraulic conductivity for the Driefontein tailings (top)
and ONC tailings material (bottom)
Ksat of the waste rock was determined in a permeameter column as 60.8 mm/h.
4-14
100.0
0mm
250mm
500mm
1000mm
0.1
1 10 100 1000
capillary pressure head (mm)
100.0
0mm
250mm
500mm
1000mm
hydraulic conductivity (mm/h)
2000mm
10.0
Ks 0mm
Ks 250mm
Ks 500mm
Ks 1000mm
Ks 2000mm
1.0
0.1
1 10 100 1000
capillary pressure head (mm)
4-15
100.0
250mm
500mm
Ksat 250mm
1.0
0.1
1 10 100 1000
capillary pressure head (mm)
Figure 4.22: Unsaturated hydraulic conductivity for the Driefontein tailings: penstock
4.3.1.1 Tensiometery
The tensiometers comprise a porous ceramic cup, filled with de-aired water in hydraulic contact, via the
ceramic, with the pore water of the tailings material. The water inside the ceramic reduces in pressure in
response to the pore water suction. This suction is measured using a differential pressure transducer,
hydraulically connected to the ceramic water via a 6 mm hydraulic hose. The hose is encased in a PVC tube
and the ceramic and tube assembly is inserted into the tailings by augering a 22 mm diameter hole.
The tensiometer data has been assessed and frequently shows evidence of a diurnal pattern, most probably
caused by loss of contact between the ceramic cup and the tailings pore water. This behaviour is illustrated
by the data shown for the penstock site at the Driefontein dam in Figure 4.23, where capillary pressure head
(pore water suction) is represented as a positive capillary pressure head. In the cooler parts of the day and
overnight, the hydraulic contact prevails and measurements of suction are obtained. This pattern was
recorded for the period from 4 to 11 December 2003. The highest tension recorded is about 5800 mm nearer
the surface (at 250 mm deep) compared to deeper in the profile (at 1000 mm below surface), where tensions
are approximately 4200 mm. However, during the day (typically from 08:30 to 18:30), the ceramic looses
hydraulic contact and suctions drop, until the recorded signal reflects the height of the column of water
between the ceramic and pressure transducer.
Tensiometer response to a rainfall event is presented in Figure 4.24 for a rainfall event of some 120 mm.
The responses clearly show the rapid drop in tension in the near surface (250 mm) tensiometer, while those
installed deeper (500 mm and 1000 mm) reveal a drying trend 6 days after the event. The deep tensiometer,
installed at 6000 mm below surface shows no response to the event.
4-16
2 50m m
1 000 0
2 50m m
800 0
400 0
200 0
-200 0
04 -Dec 05 -Dec 06-Dec 0 7-De c 08 -Dec 09 -Dec 10-Dec 11-Dec
5 00m m
1 000 0
5 00m m
800 0
Cap illa ry p re ssu re h ead (m m )
600 0
400 0
200 0
-200 0
04 -Dec 05 -Dec 06 -Dec 07 -Dec 08-Dec 09 -Dec 10-Dec 11-Dec
1 000m m
1 000 0
1 000 m m
800 0
Cap illa ry p re ssu re h ead (m m )
600 0
400 0
200 0
-200 0
04 -Dec 0 5-De c 06-Dec 07 -Dec 0 8-De c 09-Dec 1 0-De c 11-Dec
4-17
6000 140
250mm
500mm
5000 120
1000mm
6000mm
Capillary pressure head (mm)
4000 100
Rain
2000 60
1000 40
0 20
-1000 0
20-Jan-05 22-Jan-05 24-Jan-05 26-Jan-05 28-Jan-05 30-Jan-05
Figure 4.24: Tensiometer and rainfall response at the Driefontein tailings dam,
penstock site during January 2005
Due to the difficulty in maintaining hydraulic contact with the tensiometers, a set of WaterMark sensors was
installed in the Driefontein and ONC tailings. Calibration of these sensors is presented in Figure 4.25 and
applied to the signals recorded at the tailings sites.
40000
CH1 (PP)
35000 CH2 (PP)
CH3 (PP)
30000 Regress Ch1
Capillary Pressure Head (mm)
Regress Ch2
20000
15000
10000
5000
0
0 0.5 1 1.5 2 2.5
Response (V)
Typical responses presented as 24 hour average values in Figure 4.26, show significant responses in the
upper 500 mm, but lagged and dampened responses at the 1000 mm depth. Clearly only large events (144
mm on 20 January), impact at the 1000 mm level and the peak wetting at this level occurs 11 days after the
rainfall event.
4-18
14000 1000
900
12000
800
Capillary Pressure Head (mm)
10000 700
4000 300
Watermark (250mm)
Watermark (500mm) 200
2000 Watermark (1000mm)
Rain 100
0 0
20-Jan-05 09-Feb-05 01-Mar-05 21-Mar-05 10-Apr-05 30-Apr-05 20-May-05
Figure 4.26: WaterMark responses for the Driefontein site (middle) during January 2005
Water contents (WC) were measured in-situ using Time Domain Reflectometery (TDR100) probes at
different depths in all the experimental sites of the two dams.
The TDR100 system of Campbell Scientific records the volumetric water content and the bulk electrical
conductivity by generating and interpreting the electromagnetic wave form.
This apparatus was tested for use in the tailings material in the top 2000 mm of the profile. A 2000 mm deep
trench was excavated and the measurements were made at 2000 mm, 1500 mm, 1000 mm, 500 mm and
250 mm, 150 and 100 mm depths. Water content and bulk electric conductance (EC) were measured in all
the experimental sites (Figure 4.27). Approximately 30 mm of rain had fallen over the three days prior to the
measurement.
At the Old North Complex tailings dam (Anglogold), the variation in volumetric water content down each
profile is less than the variation between stations. The water content measurements showed that the upper
100 mm has higher water content than the deeper profile. Below the 500 mm depth, the water content is
uniform. However, the water content at the edge and middle station (~0.16) are significantly less than those
at the station near the penstock (~0.25). At the penstock station the water content at 100 mm below surface
(~0.316) is also the maximum recorded at all stations and depths. Porosities measured from the horizontal
samples extracted from the excavations indicate that the saturation of the edge and middle profile ranges
between 0.3 and 0.5, while the saturation at the 2000 mm depth at the penstock site is approximately 0.9.
4-19
Volumetric Water Content
0.000 0.050 0.100 0.150 0.200 0.250 0.300 0.350
0
200
400
600
800
Depth (mm)
1000
1200
1400
1600
edge
1800 middle
penstock
2000
200
400
600
800
Depth (mm)
1000
1200
1400
1600
edge
1800 middle
penstock
2000
Figure 4.27: TDR measurements at ONC (above) and Driefontein (below), December 2003
At the Driefontein site, the water content variation along each profile is high and the profile near the penstock
is not significantly wetter than those at the edge and middle stations.
4-20
4.3.1.4 Lysimeter monitoring
Tailings
No drainage was observed from either lysimeter until 11 February 2005. Subsequent drainage, between
February and June 2005 comprised 24% of the rainfall at Driefontein and 21% (February to May 2005) at
ONC as illustrated in Figures 4.28 and 4.29 respectively.
600
Driefontein
Cumulative rainfall and drainage (mm)
500 Rain
400
300
200
100
0
Jan-05 Feb-05 Mar-05 Apr-05 May-05
Figure 4.28: Lysimeter drainage from the Driefontein tailings (January-May 2005)
4500 400
inside 700mm
4000 rain 350
drainage
3500
300
3000
250
2500
200
2000
150
1500
100
1000
500 50
0 0
20-Jan-05 30-Jan-05 09-Feb-05 19-Feb-05 01-Mar-05 11-Mar-05
Figure 4.29: Lysimeter drainage from the ONC tailings (January-May 2005), showing the tensiometer
response inside the lysimeter at 700 mm below surface
These drainage data will be used to assess the water balance during the simulation phase. Although
miniature tensiometers were installed into the lysimeters and into the tailings profile adjacent to the
lysimeters, these data have significant periods of disconnection with the tailings material or capillary
pressure heads outside the range of the tensiometers. The miniature tensiometers inserted in the tailings
lysimeter respond better than the standard tensiometer nests, but the data are regularly interrupted by
tensiometers loosing contact with the tailings through drying or excessive tensions in the material. A
4-21
successful series of tensions in the ONC lysimeter 700 mm depth probe have been recorded for a series of
rainfall events during February 2005, (Figure 4.29).
Waste Rock
A plastic liner lysimeter was installed in the Kopanang waste rock dump at a depth of 2 m below the surface.
The 8 m by 8 m liner buried in the waste rock material is connected to an outflow pipe which discharges into
a tipping bucket mechanism to record the outflow rate. A single event of 56.4 mm during December 2003
resulted in a drainage outflow of 15.2 mm, some 27% of the inflow. The drainage started 10 hours after the
start of the rain event and ceased within 15 hours of the cessation of rain. The bulk of this event and all other
smaller events are apparently stored in the fines in the first 2 m thickness of the waste rock. The response
for the full recorded period (Figure 4.30) is analysed for the rainfall-runoff relationship in Figure 4.31. Events
for a wet antecedent condition and dry antecedent condition are presented in Figures 4.32 and 4.33
respectively.
700
Drainage
Drainage2
600 Rain
Cumulative drainage and rain (mm)
500
400
300
200
100
0
Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
Figure 4.30: Lysimeter drainage from the Kopanang waste rock (April 04-May 2005)
4-22
90
80
70
60
Drainage (mm)
23-Dec-2004
50
40 1-Apr-2005
30
20
10
0
0 10 20 30 40 50 60 70 80 90
Rain (m m )
Figure 4.31: Lysimeter drainage from the Kopanang waste rock in response to rainfall
70
60
Cumulative drainage and rain (mm)
50
40
30
20
10
0
31-Mar-05 01-Apr-05 02-Apr-05 03-Apr-05 04-Apr-05 05-Apr-05 06-Apr-05 07-Apr-05
Figure 4.32: Cumulative lysimeter drainage from the Kopanang waste rock in response to rainfall
during April 2005, following on a high antecedent moisture condition
4-23
70
60
40
30
20
10
0
23-Dec-04 25-Dec-04 27-Dec-04 29-Dec-04 31-Dec-04 02-Jan-05 04-Jan-05 06-Jan-05 08-Jan-05
Figure 4.33: Cumulative lysimeter drainage from the Kopanang waste rock in response to rainfall
during December 2004, following on a low antecedent moisture condition
Rainfall and runoff plot data have been resolved for the side slope runoff plots at Driefontein and ONC. The
surface runoff plots were periodically flooded from runoff outside the plot perimeter. This was rectified in
January 2005.
At Driefontein the runoff from the side slopes varied between 17% (side slope 2) and 38% (side slope1) of
the rainfall between 7 October 2004 and 20 May 2005 (Figure 4.34). At the ONC site the runoff from the
side slopes varied between 5% (side slope 2) and 17% (side slope1) of the rainfall between 7 October 2004
and 20 May 2005 (Figure 4.35). This lower runoff from the ONC side slopes is possibly due to the
vegetation on these slopes.
800
rainfall
700
runoff S1
cumulative rainfall and runoff (mm)
runoff S2
600
500
400
300
200
100
0
Oct-04 Nov-04 Dec-04 Jan-05 Feb-05 Mar-05 Apr-05 May-05
Figure 4.34: Driefontein tailings dam side slope runoff and rainfall, October 2004 to May 2005
4-24
800
rainfall
700
runoff S1
runoff S2
500
400
300
200
100
0
Oct-04 Nov-04 Dec-04 Jan-05 Feb-05 Mar-05 Apr-05 May-05
Figure 4.35: ONC tailings dam side slope runoff and rainfall, October 2004 to May 2005
800
rainfall
700 runoff edge
runoff middle/penstock
cumulative rainfall and runoff (mm)
600
500
400
300
200
100
0
Oct-04 Nov-04 Dec-04 Jan-05 Feb-05 Mar-05 Apr-05 May-05
Figure 4.36: ONC tailings dam top surface runoff (edge and middle/penstock) and rainfall, October
2004 to May 2005
4.3.3 Evapotranspiration
A large column sample of 330 mm diameter and 950 mm long was extracted from the ONC tailings dam and
instrumented for laboratory testing of wetting and drying cycles. Miniature TDR probes and miniature
tensiometer probes were inserted into the sample. Temperature sensors were installed in the near surface
and at 100 mm below surface, air temperature was observed and a solar radiation sensor was used to
record incoming radiation. Figure 4.37 shows the response of the TDR to the application of simulated rainfall
of 120 mm to the surface. Significant wetting and drying is evident in the top 300 mm, but very retarded
responses below this level.
4-25
0.4
100mm
0.35 300mm
500mm
Volumetric water content (m/m)
0.3 750mm
0.25
0.2
0.15
0.1
0.05
0
01 June 2005 06 June 2005 11 June 2005 16 June 2005 21 June 2005 26 June 2005
6 800
Evaporation
Solar radiation 700
5
600
Average Daily Evaporation (m/m)
)
2
4
3 400
300
2
200
1
100
0 0
04-Jun-05 09-Jun-05 14-Jun-05 19-Jun-05 24-Jun-05
Figure 4.38: Estimation of evaporation from the surface of the ONC tailings column
Rate of water loss during the evaporation phase is determined by simulating the water content responses.
4-26
CHAPTER 5
DATA EVALUATION AND MODELLING
A tailings storage facility (TSF) consists of variably unsaturated porous media with saturated and
unsaturated zones. The proportion of the saturated zone decreases and may even disappear with time in
decommissioned dams.
The hydrological system of gold TSFs for operational and non-operational tailings dams are conceptually
different and as such should be studied differently. Decommissioned tailings dams have the following main
characteristics:
Have spatially varying unsaturated zones, which makes simple physical water balance calculations
inappropriate.
Unlike the operational dam where the saturated zone is near the surface, the entire exposed surface of
the non-operational dam is variably unsaturated.
In both decommissioned and operational tailings dams, there would be a spatial variation of the water
balance components from the edge to the centre of the dam. This is mainly attributed to the moisture
available in the profile, which is governed by the presence and depth of the phreatic surface. Thus infiltration
at the edge of the pool is maximum and minimum at the centre of the pool whereas evaporation would be
maximum at the centre of the pool and minimum at the edge of the pool. The depth of the phreatic surface
and rate of drop of this surface is difficult to estimate accurately.
Any attempt to determine the water balance of a decommissioned tailings dam should establish itself largely
on the unsaturated flow processes and must account for the impacts of surface flux boundary conditions.
The objectives of such studies should be to predict the actual infiltration rates and subsequently the seepage
quantity from the tailings dam through, for example, dam toe drains (Rykaart et. al., 2003). The unsaturated
zone in a tailings dam is characterised by significant spatial variability and relies heavily on the unsaturated
flow principles. This, in turn, requires understanding and solving of the surface flux boundary problem – the
result of which should be applied to predict the water balance of the entire tailings dam which is a saturated-
unsaturated medium.
Full understanding of the following aspects of the hydrological process in a tailings dam is required:
Non-operational TSFs are characterised by the presence of a saturated zone whereby the phreatic zone
could be near surface (with or without a pool at the centre) and the unsaturated zone above the phreatic
surface. The position of the phreatic surface is unknown for most pre-closure TSFs. In TSFs where the pool
5-1
is not dried up (operational and pre-closure TSFs), the monitoring of the pond /pool water level can be used
to calculate changes in the tailings pond storage volume (∆Spool) as shown in Figure 5.1.
Q3 Q7
Q6 Q10
Unsaturated Q9
Zone (UZ)
Q11
Q8 Q12
Saturated
Zone (SZ)
Figure 5.1: Schematic of TSF water balance components for decommissioned dams
It is feasible that over time, the phreatic surface will retreat into the tailings and that a pool can establish on
the surface of the TSF after a significant precipitation event. In such a case, the pool will be underlain by an
unsaturated zone. Water balance equations for Figure 5.1 can be constructed as follows:
This equation implies that ∆S has a negative value when there is a decline in storage over the assessment
period and has a positive value when there is an increase in storage.
It can be reasonably assumed that over a period of time, infiltration flows from the beach and side slope
areas report to the saturated zone and are not stored in the immediate beach or slope zone, then ∆Sbeach
and ∆Sside slope are zero and Equations 3 and 4 can be rewritten as follows:
As it is not reasonable to assume that ∆Spool will be zero over a period of time and as it is easily measurable,
Equation 2 can be rewritten to solve for infiltration as follows:
Q9 = Q1+Q7-Q4-∆Spool Equation 8
Where
Q1 = volume of water from precipitation falling on the pool (Pp)
Q2 = volume of water from precipitation falling on the beach area (Pb)
Q3 = volume of water from precipitation falling on the side slope area (Ps)
Q4 = Volume of actual evaporation from the pool area (Ep)
Q5 = Volume of actual evaporation from the beach area (Eb)
Q6 = Volume of actual evaporation from the side slope area (Es)
Q7 = Volume of runoff from beach area (Rb)
Q8 = Volume of runoff from side slope area (Rs)
5-2
Q9 = Volume of infiltration from pool area (Ip)
Q10 = Volume of infiltration from beach area (Ib)
Q11 = Volume of infiltration from side slope area (Is)
Q12 = volume of water released from the saturated zone
Q13 = total volume of discharge/decant/seepage from TSF
The water balance for the complete TSF can be constructed as follows:
The first step in the long term prediction of a decommissioned water balance should be developing a method
to calculate the physical water balance. Two main scenarios can be considered for pre-closure TSFs:
There is little literature available on prior investigations to establish the phreatic surfaces in decommissioned
dams although techniques such as direct mapping, photo interpretation, thermal line scanning and
measurements of silt content have all been used.
The rate of water movement through tailings and into the underlying geological materials can be estimated:
a) from the physical properties of the tailings; b) can be calculated by combining point measurements of
hydraulic head and hydraulic conductivity; and c) using geochemical characteristics of the tailings pore water
(Blowes, et al. 2003).
Blight and Du Preez (1997) estimated the retreat rate of the phreatic surface to be 0.3 m/y or more. Using
the concentration of tritium in the tailings pore water, Coggans et al. (1999), estimated that annual recharge
rate into the impoundment of a copper tailings was 25 mm/year. A similar rate is also presented by Johnson
et al. (2000). A recharge rate of 75 mm/year was estimated for the tailings impoundment at the Campbell site
near Red Lake, Ontario.
Phreatic surface monitoring data from the piezometers at ONC No. 3 showed the rate of phreatic surface
retreat for the first 9 years is 4.03, 2.81 and 2.87 m/year in average at the edge, middle and ex-pool part of
the dam, respectively. The rates have dropped to 0.97 and 0.25 m/y for the middle and ex-pool parts of the
dam. If we extrapolate these data, the current approximate position of the phreatic surface would be at about
45 m, 31 m and 27.3 m at the edge, middle and ex-pool part of the dam respectively. At Driefontein No 3
5-3
dam the retreat rate of the phreatic surface at the edge of the dam has dropped from 8.82 m/y to 0.88 m/y in
2004/2005.
The physical water balance of a pre-closure TSF can then be calculated using the above equation provided
the following parameters can be obtained:
Rainfall input, runoff, infiltration, evaporation and evapotranspiration loss and percolation below 0.7 m depth
are derived directly from observed data for the ONC and Driefontein TSFs discussed in this report (Figure
5.2). The data are further used to develop rules for evaluating typical responses of the components.
Rainf all
Transpiration Transpiration
f rom side slopes
Evaporation
f rom pond
Runof f f rom Evaporation
dry beach Pond storage f rom dry beach Evaporation
f rom side slopes
Storage
dynamics Runof f f rom
Lateral f low
Drawdown of the side slopes
Percolation phreatic surf ace
Figure 5.2: Schematic of the water balance for a decommissioned tailings dam
The total rainfall for the analysis period was 722 mm at the Driefontein tailings site and 677 mm at the ONC
site. The potential evapotranspiration for the analysis period, 1207 mm, has been derived from the
Potchefstroom meteorological station.
The runoff plot responses (Figures 4.34 to 4.36) were analysed for total rainfall, runoff and infiltration over
the period. This analysis is summarised in Table 5.1. The total side slope runoff from each of the four sites,
varied between 16% and 19% of the rainfall, except at the second side slope at ONC. This site was noted to
have well established vegetation and effective terracing. The Driefontein side slope which yielded 39% runoff
was steep and denuded. The only runoff plots to operate successfully on the tailings top surface were those
on the ONC tailings dam. Here, the long term runoff was 16% and 19% of the total rainfall for the period for
the Edge and Middle/Penstock Site respectively. These general runoff and infiltration volumes (Table 5.1)
can be used for long term water balance estimates for summer rainfall.
An analysis of the event responses from the side slope runoff plot 1 at the ONC site was used to evaluate
event responses. Discrete events were identified from the rainfall data set as those which comprised
5-4
continuous rainfall, interrupted by a minimum of 4 hours without precipitation. The runoff events were
selected as any event which generated overland flow. A comparison of the rainfall and runoff responses
shows a general increase in runoff for increasing rain (Figure 5.3). The slope of the regression is 28%,
significantly higher than the 17% derived for the long term response. A 7-day Antecedent Precipitation Index,
API-7 (total rainfall during the 7 days prior to the event), was extracted for each event and used in a further
comparison (Figure 5.4). The results are a slight improvement (R2=0.60) and can be used to derive event
rainfall-runoff-infiltration fluxes. It is noted that no runoff was recorded for rainfall events less than or equal to
5 mm, irrespective of the antecedent conditions.
Table 5.1: Long term tailings dam runoff and infiltration (October 2004-May 2005)
25 25
y = 0.2351x - 1.1151
R 2 = 0.6031
20 y = 0.2794x - 0.8795 20
R2 = 0.5317
15 15
Runoff (mm)
Runoff (mm)
10 10
5 5
0
0
0 10 20 30 40 50 60 70
0 10 20 30 40 50 60 70 80 90
Rain (mm)
Rain*(1+API-7/100) (mm)
Figure 5.3: Relationship between rainfall and Figure 5.4: Relationship between rainfall and API-7-
runoff for the side slope plot 1 on the ONC tailings runoff for the side slope plot 1 on the ONC tailings
dam dam
Runoff = 0.17*Rainfall, except in well vegetated areas or on steep bare side slopes
Event analysis:
No runoff for rainfall event less than 5 mm
Runoff = 0.235*Rain*(1+(API-7)/100)
5-5
5.1.3.3 Evapotranspiration
The observations from the column experiment showed that evaporation occurred from the tailings surface at
the potential rate for 14 days, after which the evaporation decreased. Since this was a thoroughly wetted
upper profile and the winter evaporation demand was low, it is expected that this might be a maximum
period. The water content was 0.13 when the evaporation rate declined. This water content corresponds to a
pore water tension of approximately 40 m and could be used in simplified water balance estimates.
The in-situ lysimeter responses yielded a total of 120 mm at Driefontein and 100 mm at the ONC tailings
dam. If it is assumed that the remainder of the infiltrated volume evaporated over this period, then the
evaporation from the tailings surface is 468 mm. This comprises 39% of the potential evaporation and 69%
of the rainfall. However, reversal in hydraulic gradient (implying upwards flow between 1000 mm and 500
mm depth) are indicated by the WaterMark data as shown in Figure 5.5.
20 1000
900
15 800
700
10 600
Gradient 250-500mm
300
0 200
20-Jan-05 09-Feb-05 01-Mar-05 21-Mar-05 10-Apr-05 30-Apr-05 20-May-05
100
-5 0
Event analysis
Evaporate at potential rate until 40 m pore water tension then
Reduce to zero after a further 7 days without precipitation
5.1.3.4 Percolation
The percolation is estimated from the in-situ drainage lysimeters where 120 mm drained from 700 mm below
surface at the Driefontein tailings dam and 100 mm drained from the same depth in the ONC material. These
amounts comprise 17% and 15% of the rainfall for the Driefontein and ONC sites respectively.
5-6
GENERAL RULE FOR PERCOLATION
Event analysis
Derive response from pore water balance
In the water balance equation above where percolation (infiltration) should report as decant by way of
penstock and seepage hence:
Mines usually measure penstock decant volumes using different methods such as return dam level or on
demand of the mine to use return water. In the absence of such mine measurement data, penstock decant
volumes can be modelled as part of the water balance component.
Seepage volume could be obtained from historical data where seepage is commonly collected in a drainage
trench around the TSF perimeter. However seepage rate could not be physically measured at every point
around the perimeter. Seepage can be estimated theoretically using Darcy’s equation (Holtz and Kovacs,
1981) for all sections of the seepage drain and calibrated using actual seepage measures if available.
In order to evaluate the long term water balance, predictive numerical modelling is required. Components of
the balance derived from the observed data discussed in the preceding section could be used as long term
balance inputs. The HYDRUS2D model (Simünek et al.) was used to estimate the long term closure water
balance of decommissioned TSFs.
Elements of the water balance include rainfall, runoff, and evaporation from the surface, seepage from the
phreatic surface and change in storage of the profile. The HYDRUS-2D model allows for detailed simulation
of flow in the porous medium, dimensioned as a cross section of the transect of the tailings dams through
which the instruments were installed (Figures 5.6 and 5.7).
The HYDRUS-2D model allows for atmospheric input of rainfall and potential evaporation. Actual
evaporation from the tailings surface is calculated by the model, dependent on atmospheric demand and
hydraulic characteristics of the tailings material at the surface.
Lateral and vertical variation of tailings properties such as particle size distribution, water content variation,
bulk density, etc., are taken into account to classify the tailings deposit into 3 and 6 material types laterally
and vertically, respectively. The Driefontein tailings dam was modelled as a 2 dimensional cross sectional
porous medium, which transects the main monitoring stations as shown in Figure 5.7a.
The rate of phreatic surface drop in Driefontein dam is given to be 8.82 m/yr in the first few years since it was
decommissioned (2000) and dropped to 0.88 m/yr in 2004/5. This implies that the TSF has become
practically unsaturated during a period of 5 years. Thus, to specify zero head boundary condition near the
base of the impoundment to represent the phreatic surface as initial boundary conditions would be
appropriate (Figure 5.7b). Three material types were specified to represent hydraulic conditions in the edge,
middle and penstock zones (Figure 5.7c). The finite element mesh was constructed to be denser near the
surface where large hydraulic gradients are expected. Observation nodes allow output water contents and
capillary pressure heads to be observed. These were specified at the surface, 120 mm, 500 mm, 1000 mm
and 15000 mm.
5-7
Rainfall
Evaporation;
Transpiration
Storage
dynamics
Lateral flow
Drainage
Phreatic surface Drains/seepage
1670
1650
1630
elevation (m)
1610
1590
profile
1570 stations
ngl
1550
0 200 400 600 800 1000 1200
distance (m)
h (mm)
0
10 000
20 000
30 000
40 000
50 000
60 000
70 000
Figure 5.7b: Initial boundary conditions for the Driefontein tailings dam
5-8
Figure 5.7c: Distribution of material types (edge [red], middle [green], penstock [blue]) through the
Driefontein tailings dam
The volumetric water contents of the observation nodes, located near the centre of the tailings dam are
shown in Figure 5.8a for a typical 16 months period. The water contents at the surface have maximum
variation, ranging from 0.12 to 0.47 and respond rapidly to rainfall events. These variations decrease with
depth, with the variation at 1 m depth ranging slowly between 0.16 and 0.30. The volumetric water contents
at 15 m below the surface do not change at all throughout the 16 months simulation.
The effective water balance was constructed from the flux output of the model and is shown in Figure 5.8b.
There is a net positive input to the tailings dam of 90 mm over the simulation period. The only runoff resulted
from an event of 68 mm rain and comprised 18 mm of runoff.
The actual evaporation from surface (498 mm) comprises approximately 50% of the potential evaporation
and is a fundamental component of the water balance. Although this is higher than that estimated from the
monitored period (39%), it confirms that the losses due to evaporation from the surface are a considerable
component of the water balance.
The seepage from the phreatic surface is taken up into the tailings during the simulation period and is a
result of initial capillary pressure head conditions not being in equilibrium with the gravitational potentials at
certain locations. Clearly an initial simulation period of two or three months is required before the water
balance is meaningful.
0.5
surfac e
0.45
120
500
0.4 1000
15000
volu m etric w a te r c on te nt
0.35
0.3
0.25
0.2
0.15
0.1
Jan-01 Mar-01 May-01 Jul-01 Sep-01 Nov-01 Jan-02 Mar-02 May-02
5-9
800
600
rain
400
Inflow/Outflow (mm)
0
runoff
seepage
-200
actual evaporation
-400
-600
potential evaporation
-800
-1000
-1200
Jan-01 Mar-01 May-01 Jul-01 Sep-01 Nov-01 Jan-02 Mar-02 May-02
2 3 1
5
9 7
4 8 6
Figure 5.9a: Finite element mesh with arbitrary observation nodes, material distribution and initial
condition (top to bottom, respectively) for the Driefontein No 3 dam
5-10
1
42
3
6
Figure 5.9b: Section of the finite element mesh, showing the finer mesh spacing in the near surface
layers and the near surface observation nodes for Driefontein No. 3 dam
5.1.5 Modelling the long term water balance for ONC Dam No 3
The simulation performed for the ONC Dam No.3 assumed an initial condition of saturation immediately after
decommissioning. Therefore, seepage and free drainage was allowed for the first two years at the penstock.
An hourly rainfall and evaporation data for 365 days has been used for the 5 year simulation period. The first
year and the fifth year simulation result is presented and discussed herein. The water balance after one year
of simulation is characterised by intense inflow and outflow peaks at 50, 120, and 300 days resulting in
infiltration during the 50-120 days and evaporation thereafter.
A sectional view of the water content of the entire dam is also shown where the position and shape of the
phreatic surface is depicted (Figure 5.10). It should be noted that the phreatic surface is roughly indicated by
that portion of the dam that has a water content greater than 0.35. By the end of year 1 the saturated zone
dropped by 25 m at the penstock and by 13 m at the edges of the dam (Figure 5.10). Phreatic surface
monitoring data from the piezometers at ONC No. 3 showed the rate of phreatic surface retreat for the first 9
years is 4.03, 2.81 and 2.87 m/year in average at the edge, middle and ex-pool part of the dam, respectively.
The rates have dropped to 0.97 and 0.25 m/y for the middle and ex-pool parts of the dam.
By the end of year 5 the net outflow/inflow pattern has changed in that the net inflow increased and outflow
decreased (Figure 5.10). This pattern is mainly attributed to unsaturated drainage and reduced water input
(vegetation growth in the upper 1 metre may also play a smaller role) from year 2 onwards (Figure 5.10). The
shape and position of the saturated zone has also changed with a significant drop of the saturated zone in
the western part of the dam.
The summary of the water balance for the first year and the fifth year simulation period is given in Table 5.2.
Figure 5.11 and Table 5.2 show the summary of the water balance simulation results. Due to its
configuration, HYDRUS reports surface fluxes in terms of either evaporation or infiltration (not both),
transpiration (root uptake) and also reports boundary fluxes such as seepage and deep drainage, although it
does not simulate for runoff separately. TSF water loss/gain can also be calculated. Since the runoff from the
top surface reports to the pool/ex-pool area it will ultimately be available for infiltration and /or evaporation
5-11
fluxes. However the runoff from the slope surfaces should not be available for infiltration/evaporation surface
fluxes unless the slope surface is densely vegetated and is rugged.
Figure 5.10: The position of the phreatic surface one year (top)
and 5 years (bottom) after decommissioning of ONC tailings dam
5-12
8
Rainfall
6 Seepage
Inflow/outflow (cm x 10 )
Infiltration
6
2
4 Evaporation
-2
-4
-6
0 40 80 120 160 200 240 280 320 360
Days
7
Rainfall
6 Transpiration
Seepage
5 Evaporation
4 Infiltration
Inflow/outflow (cm X10 )
6
3
2
2
1
0
-1
-2
-3
0 40 80 120 160 200 240 280 320 360
Days
Figure 5.11: Summary of the cumulative water balance for the ONC tailings dam during 365 days for
year 1 (top) and 5 years (bottom) of simulation after decommissioning.
Table 5.2: Summary of the water balance for the 1st year and 5th year after decommissioning ONC No
3 Dam
5-13
The overall water balance of the TSF is given in Table 5.2 where a breakdown of the contributions from
rainfall and the TSF water loss to the usual water balance components is presented. The simulation has
shown that evaporation was over 121% of rainfall by the end of the first year of simulation and progressively
decreased to 62% by the end of the 5th year mainly as transpiration (T). Simulation after Year 1 was run with
root uptake as part of the atmospheric boundary fluxes and seepage and discharge were allowed to continue
as deep drainage boundary fluxes. By the end of the simulation for Year 5, transpiration (T) largely replaced
evaporation and infiltration into the TSF was significant as result of TSF moisture loss due to deep drainage
(seepage).
As runoff couldn’t be obtained directly from the simulation results, runoff plot data from the research project
was used to give an average of 17% runoff (Tables 5.2 and 5.3). Verification and validation of the simulation
results against the measurement and experimental data presented elsewhere in this report is summarized in
Table 5.3.
Table 5.3: Water balance results from monitoring and experimental data (as percentage of annual
rainfall)
The simulation results presented in Table 5.2 were entered into the general water balance equation
(Equation 10 in Section 5.12) to give the results presented below.
Where
Pp, Pb, Ps = Precipitation on pool, beach and slope respectively
Ep, Eb, Es = Evaporation from pool, beach and slope respectively
Rs = Runoff from side slopes
Change in TSF storage is negative when water is lost and positive when water is gained
Therefore, using the data in Table 5.2 for year 1, the following is obtained:
169.6 = 456 – 552 – 77.5 – (-457.6 + 103.3) = 180.8 (Error = -11.2 mm = -6.2%)
The -6.2% error in the water balance for the Year 1 simulation may be the portion of the TSF water loss that
is evaporated from the upper part of the TSF and/or the pool. The other possible area of uncertainty is the
fact that runoff could not be independently simulated by the model and verified from the runoff plot data. This
is unlike infiltration and evaporation for which comparison of experimental measurements and simulation
results has been possible as shown in Table 5.3.
A single event of 56.4 mm during December 2003 resulted in a drainage outflow of 15.2 mm, some 27% of
the inflow. The drainage started 10 hours after the start of the rain event and ceased within 15 hours of the
cessation of rain. The bulk of this event and all other smaller events are apparently stored in the fines in the
first 2 m thickness of the waste rock.
5-14
Over the period April 2004 to June 2005 the drainage comprised 87.2% of the rainfall as shown in Figure
4.30. Analysis of the events indicates that drainage is minimal for events of 15 mm or less while most major
events produced drainage at or above the rainfall for that event. Clearly additional drainage occurred from
previous events during these major events (Figure 4.31). Event drainage is obviously influenced by the
antecedent moisture conditions as illustrated by the high drainage/rainfall ratio of a high antecedent moisture
condition during April 2005 (Figure 4.32) and the low drainage/rainfall ratio of a low antecedent moisture
condition for an event during December 2004 (Figure 4.33).
Rainfall
Evaporation
Transpiration from slopes
from slopes
Runoff
Runoff from top from side
slope Preferential slopes
flow Storage
dynamics
Percolation Dynamics of the
phreatic surface
Seepage
Event analysis
No drainage for rainfall events less than 15mm
Drainage = 1.1*Rain for events with high antecedent conditions
Drainage = 0.8*Rain for events with low antecedent conditions
For situations that represent high antecedent conditions, the drainage value of 1.1 x rainfall is explained by
the fact that drainage from the preceding event has a lag time, resulting in a measured drainage for the
actual event + the preceding event.
5-15
CHAPTER 6
DEVELOPMENT OF WATER BALANCE MONITORING AND
MODELLING METHODOLOGIES
6.1 APPROACH
The chapter presents a guideline on water balance methodology that can be used by mines / regulators /
consultants to assess the water balance (operational and post-closure) for the different types of mine residue
deposits, although it is drawn from studying the water balance of gold TSFs in the Witwatersrand. The
optimal data requirement for the estimation of water balance of a mine residue deposit forms the central
basis for the guideline. The preceding chapters have shown that the data required for tailings storage
facilities and waste rock dumps and even for operation and non operational TSFs is markedly different.
Figure 6.1 shows a generalised flowchart to be followed in collection of the required data, determination
and/or prediction, monitoring and validation phases.
Figure 6.1: Flow diagram showing the relationship between data collection, water balance
determination and monitoring for mine residue deposits
The data required for a water balance determination of a mine residue deposit could broadly be categorised
into a) meteorological (climatic) data, 2) material characteristic data and 3) hydrological response data (see
Table 6.1).
A comprehensive meteorological data set includes daily automated rainfall records, maximum and minimum
temperature, humidity, wind speed as well as solar radiation. Whereas daily rainfall over years is evident to
be used as precipitation input, data for wind speed, temperature and humidity data is critical for the
derivation of potential evapotranspiration. It is preferable that a complete and dedicated meteorological
station be installed on each mine residue facility for which a water balance determination is required.
However, climatic data sets from nearby Weather Bureau stations are useful to complement and compare
the climatic data collected at the site.
6-1
Table 6.1: Data requirements for water balance determination for mine residue deposits
Rainfall
Evaporation;
Transpiration
Storage Drainage
dynamics Tipping bucket
Lateral flow Lysimeter and soak-away
Runoff plot
Drainage
Figure 6.2: Schematic of data requirement for a detailed water balance of a gold tailings dam
Reference potential evaporation data could be derived using the Linacre 1977 estimate for potential
evaporation (Schulze and Kunz, 1994) as follows:
where
ETo = Reference potential evaporation (mm)
Ta = mean air temperature (oC)
z = altitude (m)
6-2
Φd = latitude (o)
uL = Linacre wind factor (regional monthly values)
Ta – Td = 0.0023z + 0.37Ta + 0.53*(Tmx – Tmn) + 0.35Tru – 10.9
Tmx = maximum daily temperature (oC)
Tmn = minimum daily temperature (oC)
Tra = range of mean air temperature between warmest and coolest month (oC)
2200
2000 ETo
1400
1200
1000
800
600
400
200
0
Feb-04 Apr-04 Jun-04 Aug-04 Oct-04 Dec-04 Feb-05 Apr-05
The hydrological response of a material depends a great deal on the hydraulic (physical) characteristics such
as particle size distribution. The variability of the hydrologic response of tailings material is governed by, for
example, the grain size variation across the profile. Therefore, it is critical to obtain hydraulic properties data
for as many samples as possible, both laterally and vertically, to better constrain the water flux boundary
across the TSF’s profile. Particle size distribution, bulk density, porosity and water retention characteristics
are the most critical hydraulic characteristic data required for a water balance calculation (Table 6.1).
The particle size distribution across a TSF is governed by the hydraulic depositional characteristics of the
slurry along the slope gradient which could also vary with time. There is a general trend whereby the fine
proportion of the tailings material increases from the edge (embankment) to the pool (ex-pool area) of a TSF.
Although there is evidence of increase of the fine proportion with depth in some TSFs (Yibas et al., 2010),
there is no clear systematic grain size variation vertically. This may be due to the variation of the slope
gradient from the edge to the middle of a TSF and due to the hydraulic dynamics of each cycle of deposition
which complicates the lateral and vertical particle size distribution.
Anglo American (1995) recognised four types of tailings with respect to their grain size distribution namely,
FC/CF, CC/CF, FC/FF and CC/FF, with C representing coarse grain size, and F representing fine grain size.
The authors described the Witwatersrand gold tailings as FC/CF type and relatively uniformly graded.
Comparison of the grain size distribution of less oxidized tailings such as studied in this project with the other
dams with mature oxidation zones suggests that the grain size distribution of the latter has been reworked by
oxidation and weathering processes (Yibas et al., 2010). Thus it is critical to obtain sufficient grain size
variation data across the profile as well as vertically for each TSF for use in water balance simulation and
prediction.
6-3
A combination of sieve and hydrometer analyses is the best option to obtain the best continuum of the grain
size distribution, since tailings grain size ranges from sand through silt to clay, although the percentage of
clay seldom reaches 5%, making the TSF either silty sand or sandy silt.
Bulk densities should be determined for laterally and vertically representative samples extracted with thin-
walled tubes. The same samples should be analysed for water retention characteristics (WRCS). Porosities
3
would then be estimated from the bulk densities, using a value for particle density of 2.65 kg/m for Wits
gold tailings.
The quantification of fluid flows in unsaturated porous materials is determined by the relationship between
the volumetric fluid flux through the media and the gradient of the driving force causing the flow (Darcy,
1856). The gradient of the water retention curve is used to determine the unsaturated conductivity (Ku) at
specific water contents.
Water retention characteristics could be determined using a controlled outflow cell apparatus for capillary
pressure head range between 0 and 10000 mm and standard pressure pot extraction for the range between
10000 mm and 150000 mm according to the procedures described in Lorentz et al., 2001. The van
Genuchten relationship could be fitted to water retention characteristic data (capillary pressure head versus
volumetric water content). The relationship is given by:
m
1
Se n
1 h
where
r
Se = effective saturation, and
s r
= volumetric water content,
= residual volumetric water content,
s = saturated volumetric water content,
= air entry parameter, (mm-1),
h = capillary pressure head, (mm),
n = pore size distribution parameter and
m = pore connectivity parameter and is taken as 1-1/n.
In this study it has been shown that apart from the higher porosity of the materials at the edge station
compared to those at the middle station, the retention characteristics of all the tailings samples are similar,
with high residual water content ranging from 0.10 to 0.20. This is consistent with grain size variation of the
tailings across the profile. These values can be used in the simulation of the water balance for the TSF in
question.
Hydrological response monitoring includes in situ hydraulic conductivity, water content measurements,
energy status monitoring, infiltration and runoff dynamics as well as monitoring of drainage (recharge) at
depth. The different methods of hydrological response monitoring is summarised in Table 6.2.
6-4
Table 6.2: Hydrological response measurements and monitoring data
The hydraulic conductivity of a soil and other porous media is a measure of its ability to transmit water in
response to a gradient in potential energy during saturated and unsaturated flow. Saturated and unsaturated
hydraulic conductivity measurements should be conducted for the tailings materials using double ring and
tension infiltrometer apparatus at different positions both laterally and vertically. In this study these
measurements have been conducted at surface, at 250 mm and 500 mm depths at the edge, middle and ex-
pool area. It is recommended that measurements beyond this depth should also be conducted. These
measurements also assist in determining evaporation from the surface of the porous medium, the rate of
phreatic surface movement as well as analysis of solute transport. Seepage conditions and consolidation
rate around the pond can be analyzed from hydraulic conductivities measured with a certain degree of
confidence.
In-situ volume water content can be measured by Time Domain Reflectometry (TDR100) system of
Campbell Scientific which also records the bulk electrical conductivity. TDR100 was tested in the tailings
material in the top 2000 mm of the profile in this study. A 2000 mm deep trench was excavated and the
measurements were made at 2000 mm, 1500 mm, 1000 mm, 500 mm and 250 mm, 150 and 100 mm
depths.
At the Old North Complex TSF, the variation in volumetric water content down each profile was less than the
variation between stations. The water content measurements showed that the upper 100 mm had higher
water content than deeper in the profile. Below 500 mm depth the water content was uniform. However, the
water content at the edge and middle stations (~0.16) were significantly less than those at the station near
the ex-pool area (~0.25). At the ex-pool station the water content at 100 mm below surface (~0.316) was
also the maximum recorded at all stations and depths. Porosities measured from the horizontal samples
extracted from the excavations indicate that the saturation of the edge and middle profile ranges between 0.3
and 0.5, while the saturation at the 2000 mm depth at the penstock site is approximately 0.49 (Figure 6.4).
This observation confirms the inverse relationship between evaporation and infiltration from the edge to the
ex-pool area of a tailings profile.
6-5
The in-situ volumetric water content should be used to constrain and calibrate the long term water balance
prediction.
200
400
600
800
Depth (mm)
1000
1200
1400
1600
edge
1800 middle
penstock
2000
200
400
600
800
Depth (mm)
1000
1200
1400
1600
edge
1800 middle
penstock
2000
Figure 6.4: In-situ water content measured by TDR at ONC (top) and Driefontein (bottom), December
2003
6-6
6.2.3.3 Runoff dynamics and infiltration
Surface runoff and slope run-off at the TSF can be measured using simple runoff plots with flow
concentration panelling at the downstream end equipped with a drum to collect and measure. A tipping
bucket mechanism is suitable to assess the surface infiltration/runoff dynamics.
The runoff volume is dependent on the slope gradient physical properties such as grain size distribution of
the material and the presence or absence of vegetation. At Driefontein, the runoff from the side slopes varied
between 17% and 38% of the rainfall. At the ONC site, the runoff from the side slopes varied between 5%
and 17% of the rainfall. This lower runoff from the ONC side slopes is possibly due to the vegetation on
these slopes.
Lysimeters are used to assess drainage and evapotranspiration fluxes. The value of lysimeter
measurements to assess the evapotranspiration and drainage fluxes is dependent on the ability of the
lysimeter profile to mimic that of the in-situ profile. In the material, the depth of the lysimeter and the
hydraulic characteristics of the material determine the reliability of the lysimeter. Mini tensiometer nests
should preferably be installed to monitor the energy status in and outside of the lysimeters. The methodology
and procedure of the lysimeter installation is according to the design as discussed in detail in chapter 3 and
is shown again in Figure 6.5. These drainage data can be used to assess the water balance during the
simulation phase.
Mini tensiometers
Plastic
lysimeter Excavation
Tailings
Galvanised base
Coarse sand
Geotextile
Outflow pipe
6-7
6.3 TSF WATER BALANCE DETERMINATION
A generalized flowchart for the water balance determination of a TSF is given below.
Figure 6.6: Generalized flowchart for the water balance determination of a TSF
A tailings storage facility (TSF) is variably unsaturated porous medium with saturated and unsaturated zones
separated by the phreatic surface. There is a spatial variation of the water balance components from the
edge (embankment) to the pool (or ex-pool) area of a TSF. This is mainly attributed to the moisture available
in the profile, which is governed, by the presence and depth of the phreatic surface. Thus infiltration at the
edge of the pool is maximal and is minimal at the centre of the pool whereas evaporation would be maximal
at the centre of the pool and minimal at the edge of the pool.
Decommissioned (and post-closure) TSFs have the following main characteristics when compared to an
operational TSF:
1. Spatially varying unsaturated zone wherein the depth of the phreatic surface and rate of drop of this
surface is difficult to estimate accurately unless measured using piezometers installed across the profile
of the TSF. Unlike the operational TSF where the saturated zone is near the surface, the entire exposed
surface of the non-operational dams is variably unsaturated.
2. Any attempt to determine the water balance of a decommissioned TSF should establish itself largely on
the unsaturated flow processes and must account for the impacts of surface flux boundary conditions.
The objectives of such studies should be to predict the actual infiltration rates and subsequently the
seepage volume from the tailings dam.
The conceptual model for operational and non-operational TSFs differs, since their hydrological systems are
conceptually different insofar as the operational TSF has a pool and other water balance components in
addition to the water balance components common to both. The unsaturated zone in a tailings dam is
6-8
characterised by significant spatial variability and relies heavily on the unsaturated flow principles. Whereas
simple physical water balance calculations are appropriate for the determination of the water balance for
operational and pre-closure TSFs, determination of a water balance for decommissioned TSFs requires
understanding and solving of the surface flux boundary problem – the result of which should be applied to
predict the water balance of the entire TSF.
All components of a TSF water balance, including any water pumped from or back into the TSF, in addition
to the classical water balance components, are considered in the physical water balance calculation of a
TSF. Operational TSFs are characterised by the presence of a saturated zone (with a pool at the centre)
whereby the actual evaporation equals the potential evaporation.
Two main scenarios can be considered in conceptualisation of a pre-closure TSFs water balance:
In South Africa, most TSFs are equipped with a penstock system. Moreover, most pre-closure TSFs are
characterised by the presence of penstock decant and toe drains to manage decant flow and seepage
volumes and as such the water balance of such facilities requires consideration of these two aspects.
Q3 Q7
Q6 Q10
Unsaturated Q9
Zone (UZ)
Q11
Q8 Q12
Saturated
Zone (SZ)
The physical water balance of a TSF can be calculated using the equations shown in Figure 6.7 and
expanded in Section 5.1.2 of the report, provided the following data set is obtained:
For a pre-closure TSF where the pool is not dried up, the monitoring of the pond /pool water level can be
used to calculate changes in the tailings pond storage volume (∆Spool) or for seepage as follows:
Based on the extensive hydrological response data collected for the two Witwatersrand decommissioned
TSFs (ONC and Driefontein) used in this study, simple rules have been developed to calculate the physical
water balance as follows when simple screening-level water balances are desired:
The 66% evaporation value is the minimum evaporation value and assumes the 17% runoff from exposed
tailings surface to the ex-pool area is decanted out of the system as penstock decant. If the penstock is
closed, the evaporation could be as high as 83% of the annual rainfall. Moreover, reversal of hydraulic
gradient, implying upward flow of moisture from 50 cm and 100 cm below surface is observed from the
Watermark data (Chapters 4 and 5).
Water balance calculations for post-closure TSFs require long term water balance prediction using numerical
models. However, the lack of a refined numerical model to accurately solve the flux boundary of the
saturated-unsaturated system characteristics of tailings storage facility makes such prediction less reliable
(see for example, Rykaart et. al., 2003).
The main purpose of the conceptual model is to understand and constrain the upper and the lower
boundaries of the cross-section to be modelled.
In order to evaluate the long term water balance, predictive numerical modelling is required. Components of
the balance derived from the observed and measured data discussed in the preceding section would be
used in the long term prediction. In this study the HYDRUS2D model (Simünek et al.) was used to estimate
the long term closure water balance of decommissioned TSFs. Other numerical models such as SoilCover
could also be used (Rykaart et al., 2002, 2003). However, the requirements and level of confidence of the
selected prediction model must be known.
6-10
Controlled by
material properties
e.g. anisotropy, PSD
and is unique to
MOVES DOWNWARDS
each facility
Affected by site-
specific factors, e.g.
geology, groundwater,
underdrain design,
seepage trenches
1. Geometric data – HYDRUS-2D models allows for detailed simulation of flow in the porous medium,
dimensioned as a cross section / transect of the tailings dam.
2. Atmospheric boundary conditions – Rainfall and potential evaporation could be used as atmospheric
boundary conditions. Simulation should be performed for mean, dry and wet annual rainfall data for
vegetated and non vegetated options. Actual evaporation from the tailings surface should then be
determined by the model dependent on atmospheric demand and hydraulic characteristics of the tailings
material.
3. Root water uptake if the surface of the TSF is vegetated to simulate actual transpiration (T)
4. Water flow parameters such as water content (saturated and residual), saturated hydraulic conductivity
(Ks)
5. Period of simulation.
Output
A set of output data includes potential and actual boundary water fluxes such as actual evaporation,
seepage and drainage fluxes, total water balance as well as water content at any point along the cross-
section. These output data will then be converted to the usual water balance components namely
precipitation (P), Evaporation (E), transpiration (T), Evapotranspiration (ET) and net infiltration (NI) in terms
of percentage of annual precipitation (rainfall). E will be equal to ET if the TSF surface is not vegetated.
Model calibration
The components of the physical water balanced derived from the observed and measured data such as
actual infiltration, runoff, water content values, measured seepage volumes, etc., should be used to calibrate
the predicted long term water balance.
The nature of the water balance to be prepared will depend on the intended use of the water balance. Where
simple assessment of water balance issues is adequate, a screening level analysis may be appropriate
whereas a water balance in support of a mine closure plan will require a detailed water balance. The
difference in approach for each scenario is shown in the sections below.
6-11
6.3.4.1 Water balance for an operational TSF
The schematic water balance for an operational TSF is shown in Figure 6.9 below and the different
approaches for a screening or detailed water balance assessment are set out in Table 6.3 below.
Rainf all
Slurry water
Transpiration Evaporation
Table 6.3: Water balance approach for an operational TSF dependent on level of detail required
The schematic water balance for a decommissioned TSF is shown in Figure 6.10 below and the different
approaches for a screening or detailed water balance assessment are set out in Table 6.4 below.
6-12
Rainf all
Transpiration Transpiration
f rom side slopes
Evaporation
f rom pond
Runof f f rom Evaporation
dry beach Pond storage f rom dry beach Evaporation
f rom side slopes
Storage
dynamics Runof f f rom
Lateral f low
Drawdown of the side slopes
Percolation phreatic surf ace
Coarse waste residue deposits typically provide an environment that is very conducive to the production of
acid mine drainage. This is primarily due to the fact that coarse waste deposits are very permeable and allow
the ingress of water and oxygen to considerable depths. However, this is counteracted by the fact that, per
definition, coarse waste has a large particle size, thereby ensuring that a large amount of the minerals that
contribute to the change in water quality, are locked up within the larger particles.
In these coarse waste deposits, water may flow in channels and macropores somewhat independently of the
hydraulic conditions in the smaller pores. The saturated hydraulic conductivities of the macropore region can
be up to several orders of magnitude higher than the hydraulic conductivity of the porous matrix (López et al,
1997).
Newman et al. (1997) suggest that samples where the portion of fine material in a waste rock dump was
larger than 40% were capable of retaining water under negative pore-water pressures. This capacity to
retain water was very small for samples with less than 40% fine material.
6-13
The driving forces and processes that characterise the hydrological behaviour of a coarse waste residue
deposit and that need to be explicitly considered when assessing the water balance for such a waste deposit
are shown schematically in Figure 6.11 below.
Rainf all
Evaporation
Transpiration f rom slopes
f rom slopes
Runof f
Runof f f rom top
f rom side
slope Pref erential
Storage slopes
f low
dynamics
Percolation Dynamics of the
phreatic surf ace
Seepage
Figure 6.11 shows that the following macro-scale driving forces need to be considered in a coarse waste
residue deposit:
The phreatic surface within a coarse waste residue deposit is dependent on the volume of water ingress on
the dump surface and the permeability of the underlying weathered material/aquifer relative to that of the
waste material. It is anticipated that the underlying lithologies generally have a lower permeability than that of
coarse waste residue deposits.
Hydraulic conductivity of a waste rock dump can be up to several orders of magnitude higher than the
hydraulic conductivity for the porous matrix (López et al, 1997). Todd (1980) suggests hydraulic conductivity
values of 45 m/day and 150 m/day for coarse sand and coarse gravel respectively. The hydraulic
conductivity of most waste rock dumps can be assumed to be within this range.
Since the permeability of the coarse waste material is relatively high, one would expect the phreatic surface
within the dump to be relatively flat and located within the lower portion of the dump. This depressed water
table has two very important effects that need to be considered:
Water flows vertically down through the dump along preferential flow paths – this means that some
particles are fairly isolated from the water flow path.
Oxygen penetrates deep into the dump and the spaces between particles are typically filled with air (and
oxygen), while particles are surrounded by a thin layer of water.
Leachate from a coarse waste residue deposit is most likely to seep from the face of the dump in close
proximity to the toe rather than percolating into the underlying aquifer. This is attributed to the relatively low
permeability of the underlying material which cannot accommodate the seepage volume from the coarse
waste residue for any given rainfall event.
Uncontrolled leachate that emanates from the coarse waste residue as surface runoff will flow down gradient
until it percolates into the weathered soil profile. The distance that such a surface runoff covers before
seeping into the underlying aquifers is dependent on the seepage volume, the permeability of the underlying
6-14
material and the topographic slope in the immediate vicinity of the coarse waste residue. The vertical
percolation of the leachate into the underlying aquifers may be hindered by the upward convergence of
groundwater in the immediate vicinity of water courses. However, relatively clean groundwater could come
into contact with mine waste residue as a result of the convergence process. This contaminated groundwater
will usually enter the water courses as base flow.
A “rule of thumb” conceptual model has previously been developed by the researchers to calculate
evaporation from a waste rock dump. The following assumptions were used and were derived from simple
research experiments:
Water available for evaporation occurs in the top 200 mm of the waste rock dump.
Any water that infiltrates deeper than 200 mm will ultimately report as seepage.
Assume that surface moisture that can be retained in the 200 mm layer is equivalent to 6.5% of the
mass of waste rock in this layer.
From the equation below, the mass of water available for evaporation was calculated. This mass was
3
converted to a volume of water (1000 kg of water = 1 m ). By dividing this volume of water by the footprint
area of the waste rock dump, the mm of rain required to produce the volume of water stored in this 200 mm
layer can be calculated.
Using this method, the amount of rainfall needed to fill the storage space that can contribute to evaporation
is determined to be around 15 mm. The following procedure was used to calculate infiltration from daily data:
Based on the data collected in this research project (and prior research undertaken by the researchers for
confidential clients), the following important statements can be made:
1. If the percentage of fine material (<2 mm) in the waste rock dumps as determined from sampling and
particle size determination is less than 40%, the water retaining capability of thee dumps is considered to
be small.
2. This implies that infiltration will be relatively high and runoff very low. The research data collected
suggests that over the long term 87% of the rainfall will percolate into the dump.
3. Infiltration (and seepage) will occur for rainfall events that exceed 15 mm with drainage dependent on
antecedent rainfall conditions (0.8*Rain for low antecedent and 1.1*Rain for high antecedent conditions)
The nature of the water balance to be prepared will depend on the intended use of the water balance. Where
simple assessment of water balance issues is adequate a screening level analysis may be appropriate
whereas a water balance in support of a mine closure plan will require a detailed water balance. The
difference in approach for each scenario is shown in Table 6.5 below.
6-15
Table 6.5: Water balance approach for a waste rock dump dependent on level of detail required
Long and sustainable monitoring of the important and measurable water balance components is required in
order to validate and calibrate the long-term prediction results. The monitoring program must include but not
be limited to the following main parameters:
1. Seepage (underdrainage and loss into the foundation) (volume and rate),
2. Penstock decant/ discharge (volume and rate),
3. Runoff into toe paddocks (volume and rate),
4. Stormwater (if any)
5. Phreatic surface level (piezometer reading)
6. Any leakage (volume and rate)
7. Any extraneous pumping onto or from the TSF
8. Complete meteorological data to refine the runoff/infiltration dynamics for upper boundary fluxes.
Flow volumes and flow rates of runoff (from toe paddocks), seepage into the underdrainage system (if any),
which surround the TSF and which is fed back to the return water dam, seepage losses into the foundation
and any other drainages, including leakages, should be measured. Penstock decant to the return water dam
and outfall pipes, filter drain outlets and solution trenches, settling ponds and evaporation dams, .etc, if still
operational must be included in the monitoring program.
Any other component of the water balance such as stormwater data and other site-specific to the TSF such
as extraneous water pumped onto the surface must also be monitored. Phreatic water level should be
measured on a regular (monthly or quarterly) basis using piezometers installed across the profile of the TSF.
Complete meteorological data needs to be collected over several years and rainfall and runoff fluxes that
govern unsaturated flow should be monitored at the surface boundary. Monitoring should be performed for
sufficient duration and frequency to characterize the seasonal and long-term trends. Monthly or quarterly
monitoring over several annual cycles is normally necessary. The data set collected over several years from
the monitoring program should be used to calibrate the prediction and model and refine the accuracy of the
results.
6-16
CHAPTER 7
CONCLUSIONS AND RECOMMENDATIONS
The research project was successfully completed and a significant amount of data was collected that is
applicable to mine residue deposits in South Africa and that can be used to significantly improve the
accuracy and reliability of water balances provided for such facilities. As a water balance is an essential
precursor to a geochemical assessment, the water balance information provided in this research report will
significantly improve the reliability of impact assessments and predictions undertaken for South African mine
residue deposits.
Historically, there has been a need to use international research data to enable the apportionment of the
various components of a mine water balance, leading to inaccuracies in the water balance and in the
resulting impact predictions. The results from this research project can be used to replace the international
research data as set out in the following conclusions and recommendations:
1. It is concluded that the data collected in this research project and the general rules that were developed
can be used as default data for all tailings storage facilities constructed with the traditional day/night
paddocks and beaching towards a central penstock.
2. It is concluded that the tailings storage facilities constructed with the cycloning method will have a quite
different particle size distribution and will therefore also have a different apportionment of water to
evaporation, runoff and infiltration. The general rules developed in this research project cannot,
therefore, be applied to such facilities with confidence.
3. It is recommended that wherever possible, monitoring equipment similar to that used in this research
project be installed to generate site-specific data for use in water balance studies and in impact
prediction studies, especially when the study is being undertaken as part of a mine closure investigation.
4. It is recommended that where the general rules developed in this project are used in lieu of a site-
specific monitoring programme as input into an impact prediction exercise, it is important to collect as
much monitoring data as possible (see Section 6.5 of this report) in order to calibrate and validate the
models. If successful calibration cannot be achieved, then it may be necessary to replace the general
rules with site-specific data collected in accordance with methodologies set out in this research report.
5. It is recommended that impact prediction studies that incorporate a water balance and that have not
undertaken detailed site-specific data collection should use the general rules documented in this report
unless a clear motivation can be provided to use alternative values.
The general rules that have been derived are listed below.
Where
Pp, Pb and Ps are precipitation on pool area, beach area and side slope area respectively
Ep, Eb and Es are actual evaporation on pool area, beach area and side slope area respectively
Rs is actual runoff from side slopes
‘Change in facility storage’ is determined as change in volume of water stored in the saturated zone
(measured as phreatic surface change) + change in volume of water stored in pool on facility surface.
Event analysis:
No runoff for rainfall event less than 5 mm
Runoff = 0.235*Rain*(1+(API-7)/100)
7-1
GENERAL RULE FOR EVAPORATION FROM TAILINGS SURFACE
Long term analysis (Witwatersrand):
Evaporation = 0.69*Rainfall,
Event analysis
Evaporate at potential rate until 40 m pore water tension then
Reduce to zero after a further 7 days without precipitation
Event analysis
Derive response from pore water balance
Event analysis
No drainage for rainfall events less than 15 mm
Drainage = 1.1*Rain for events with high antecedent conditions
Drainage = 0.8*Rain for events with low antecedent conditions
7-2
CHAPTER 8
REFERENCES & BIBLIOGRAPHY
GENERAL
DARCY H (1856) Les Fontaines Publiques de la Ville de Dijon. Dalmont, Paris. 647pp
HOLTZ and KOVACS (1981) An introduction to geotechnical engineering, Prentice Hall, Inc., Englewood
Cliffs, N.J., 733 pp
LINACRE ET (1977) A simple formula for estimating evaporation rates in various climates, using temperature
data alone, Agricultural Meteorology, Volume 18, Issue 6, December 1977, Pages 409-424
LORENTZ SA, GOBA P and PRETORIUS JJ (2001) Hydrological Processes Research: Experiments and
Measurements of Soil Hydraulic Characteristics. Water Research Commission, Pretoria, Report 744/1/01,
69p.
WASTE ROCK
FALA O, AUBERTIN M, MOLSON J, BUSSIERE B, WILSON GW, CHAPUIS R. and MARTIN V (2003)
Numerical modelling of unsaturated flow in uniform and heterogeneous waste rock piles. 6th International
Conference on Acid Rock Drainage. Cairns, QLD, Australia, 12-18 July 2003:895-902.
FINES P, WILSON GW, WILLIAMS DJ, TRAN AB and MILLER S (2003) Field characterization of two full-
scale waste rock piles. 6th International Conference on Acid Rock Drainage. Cairns, QLD, Australia, 12-18
July 2003:903-909.
FRANKLIN M, FERNANDES HM and VAN GENUCHTEN MT (2008) Modelling the water flow in
unsaturated waste rock pipe: an important step in the overall closure planning of the first uranium mining site
in Brazil. In Merkel, B.J. and Hasche-Berger, A. 2008. Uranium Mining and Hydrogeology. Springer. 177-
186.
LEFEBVRE R, HOCKLEY D, SMOLENSKY J and GELINAS P (2001) Multiphase transfer processes in
waste rock piles producing acid mine drainage: Conceptual model and system characterization. Journal of
Contaminant Hydrology, 52: 137-164.
LÓPEZ DL, SMITH L and BECKIE R (1997) Modeling Water Flow In Waste Rock Piles Using Kinematic
Wave Theory. Fourth International Conference on Acid Rock Drainage. Vancouver.
MARCOLINE JR, BECKIE RD, SMITH L and NICHOL CF (2003) Mine waste rock hydrogeology – The effect
of surface configuration on internal water flow. 6th International Conference on Acid Rock Drainage. Cairns,
QLD, Australia, 12-18 July 2003:911-918.
MOLSON JW, FALA O, AUBERTIN M, and BUSSIERE B (2005) Numerical simulation of pyrite oxidation and
acid mine drainage in unsaturated waste rock pipes. Journal of Contaminant Hydrology, 78: 343-371
NEWMAN LL, HERASYMUIK GM, BARBOUR SL, FREDLUND DG and SMITH T (1997) The Hydrogeology
of Waste Rock Dumps and a Mechanism for Unsaturated Preferential Flow. Fourth International Conference
on Acid Rock Drainage. Vancouver.
NICHOL C, SMITH L and BECKIE R (2003) Time domain reflectometry of water content in coarse waste
rock. Can. Geotech. J. 40: 137-148
NICHOL C, SMITH L and BECKIE R (2003) Water flow in uncovered waste rock- A multi-year large lysimeter
study. 6th International Conference on Acid Rock Drainage. Cairns, QLD, Australia, 12-18 July 2003:919-926.
O’KANE M, STOICESCU J, JANUSZEWSK S, MCHAIN DM, and HAUG MD (1998) Design of field test plots
for a sloped waste rock surface. Paper presented at the 1998 National Meeting of the American Society for
Surface Mining and Reclamation, St. Louis, Missouri, May 16-21, 1998
TODD DK (1980) Groundwater Hydrology (second edition). John Wiley and Sons, Inc., New York
WEEKS B and WILSON GW (2003) The influence of solar radiation on evaporation for three-dimensional
flux boundary modeling of soil cover systems on waste rock dumps. 6th International Conference on Acid
Rock Drainage. Cairns, QLD, Australia, 12-18 July 2003:711-717.
WELS C, O’KANE M, FORTIN S and CHRISTENSEN D (2001) Infiltration test plot study for mine waste rock
piles at Questa mine, New Mexico. Paper presented at the National Meeting of the American Society for
Surface Mining and Reclamation, Albuquerque, New Mexico, June 3-7, 2001: 1-14.
8-1
MINE WASTE CHARACTERISATION
BLIGHT G and DU PREEZ J (1997) The escape of salt pollution from decommissioned gold residue deposits
in South Africa, The Journal of the South African Institute of Mining and Metallurgy, July/August 1997, pp
201-204.
BLOWES DW, PTACEK CJ and JUROVEC J (2003) Mill Tailings: hydrogeology and geochemistry. In:
Environmental aspects of mine water (Eds. Jambor JL, Blowes DW and Ritchie AIM), 31, p95-116.
COGGANS CJ, BLOWES DW, ROBERTSON WD and JAMBOR JL (1999) The hydro geochemistry of a
nickel-mine tailings impoundment, Copper Cliff, Ontario. In: The Environmental Geochemistry of Mineral
Deposits. Part B: case Studies and research topic (L.H. Filipek & G.S. Plumlee, eds.) Rev. Econ. Geol. 6B,
447-465.
FREDLUND DG and XING A (1994) Equations for the soil-water characteristic curve. Can. Geotech. J.
31:521-532.
FREDLUND MD, FREDLUND DG and WILSON GW (2003) Prediction of the soil-water characteristic curve
from grain-size distribution and volume-mass properties.
HOEPFNER U and FREDLUND MD (2005) The θ(ψ)- function of waste rock: measurements and
pedotransfer functions. Poster presented at Workshop: German Soil Science Soc. Feb 18-19, 2005.
JOHNSON RH, BLOWES DW, ROBERTSON WD & JAMBOR JL, 2000. Hydrogeochemistry of the Nickel
Rim Mine tailings impoundment, Sudbury Ontario, J. Contam. Hydrol. 41, 49-80.
ROSNER T, BOER R, REYNEKE R AUCAMP P and VERMAAK J (2001) A preliminary assessment of
pollution contained in the unsaturated and saturated zone beneath reclaimed gold-mine residue deposits.
WRC Research Report No.797/1/01
STOCKWELL J, BECKIE R and SMITH L (2003) The hydrogeological characterisation of an unsaturated
waste rock pile, Key Lake, Saskatchewan, Canada. 6th International Conference on Acid Rock Drainage.
Cairns, QLD, Australia, 12-18 July 2003:927-937.
SWANSON DA, SAVCI G, DANZIGER G, MOHR RN and WEISKOPF T (1999) Predicting the soil-water
characteristics of mine soils. Tailings & Mine Waste '99, 6th International Conference on Tailings & Mine
Waste, Fort Collins, CO. USA: 345-349.
YIBAS B, PULLES W and NENGOVHELA C (2010) Kinetic development of oxidation zones in tailings dams
with specific reference to the Witwatersrand gold mine tailings dams. WRC Research Report No.1554/1/10
8-2
RYKAART EM, WILSON GW, CURREY N and RITCHIE P (2003) Case study – Final closure water balance
of a tailings impoundment using direct vegetation. 6th International Conference on Acid Rock Drainage.
Cairns, QLD, Australia, 12-18 July 2003:363-371.
WELS C and ROBERTSON AMacG (2003) Conceptual model for estimating water recovery in tailings.
Proceedings of the 10th International Conference on Tailings and Mine Waste, Vail, Colorado, USA, 12-15
October 2003: 87-94.
WELS C, ROBERTSON AMacG, and MADARIAGA PM (2004) Water Recovery Study for Pampa Pabellon
Tailings Impoundment, Collahuasi, Chile. Proceedings of the 11th Annual Conference on Tailings and Mine
Waste, Vail, Colorado, USA, 10-13 October, 2004.
WILLIAMS B, NEIVVANDT R, O’KANE M, BROWNE D and WHITE C (2003) Laboratory calibration and
initial results of a soil moisture monitoring system at the Mt Leyshon tailings dam and waste rock cover
system. 6th International Conference on Acid Rock Drainage. Cairns, QLD, Australia, 12-18 July 2003:411-
417.
WILLIAMS DJ (2008) The influence of climate on seepage from mine waste storage during deposition and
Post-Closure. Mine closure 2008, ed Fourie, A., Johannesburg, South Africa
WOYSHENER MR and ST-ARNAUD L (1994) Hydrolological evaluation and water balance of a thickened
tailing deposit near Timmins on Canada. Paper presented at the International Land Reclamation and Mine
Drainage Conference and the Third International Conference on the Abatement of Acidic Drainage,
Pittsburgh, PA, 24-29 April 1994
EVAPORATION/TRANSPIRATION
BLIGHT GE (2002) Measuring evaporation from soil surfaces for environmental and geotechnical purposes.
Water SA 28(4):381-394.
BUTTERS GL and DUCHATEAU P (2002) Continuous flow method for rapid measurement of soil hydraulic
properties: І. Experimental consideration. Vadose Journal 1:239-251.
CORNELIS WM, RONSYN J, VAN MEIRWVENNE M and HARTMANN R (2001) Evaluation of pedotransfer
functions for predicting the soil moisture retention curve. Soil Sci. Soc. Am. J. 65:638-648
ENGELS J and DIXON-HARDY D (2008) Water management consideration for conventional storage.
http//www.tailings.info/waterhtm. Accessed, Dec 2008.
EVETT SR, MATTHIAS AD and WARRICK AW (1994) Energy balance model of spatially variable
evaporation from bare soil. Soil Sci. Soc. Am. J. 58(6):1604-1611.
FEDDES RA and RAATS PAC (2004) Parameterizing the soil water plant root system. In: Feddes R. A. et al.
Editors. Unsaturated Zone Modelling: Progress, Challenges and Applications. Wageningen Frontis Series. 6:
95-141.
FREDLUND MD, WILSON GW and FREDLUND DG (2002) Use of grain size distribution for estimation of
the soil water characteristic curve. Can. Geotech. J. 39:1103-1117.
HUPET F, LAMBOT S, FEDDES RA, VAN DAM JC and VANCLOOSTER M (2003) Estimation of root water
uptake parameters by inverse modeling with soil water content data. Water Resour. Res. 39(11):1312-1328.
HWANG SП and POWERS SE (2003) Lognormal distribution model fro estimating soil water retention
curves for sandy soils. Soil Science. 163(3):156-166.
HWANG SП and POWERS SE (2003) Using particle size distribution models to estimate soil hydraulic
properties. Soil Sci. Soc. Am. J. 67:1103-1112.
LIU S, GRAHAM WD and JACOBS JM (2005) The value of climate forcings to capture soil water dynamics
and evapotranspiration using daily potential evapotranspiration. J. of Hydrology. 309:39-52.
MENENTI N, JIA L and BASTIAANSEN WGM (2004) Energy and water flow through the soil vegetation
atmosphere system: the fiction of measurements and the reality of models. In Feddes R. A. et al. Editors.
Unsaturated Zone Modelling: Progress, Challenges and Applications. Wageningen Frontis Series. 6:211-
226.
PACHESKY YA and RAWLS WJ (1999) Accuracy and reliability of pedotransfer functions as affected by
grouping soils. Soil Sci. Am. J. 63:1748-1757.
8-3
PACHESKY YA, SMETTEM KRJ, VANDERBORGHT J, HERBST M, VEREECKEN H and WOSTEN JHM
(2004) Reality and fiction of models and data in soil hydrology. In Feddes R. A. et al. Editors. Unsaturated
Zone Modelling: Progress, Challenges and Applications. Wageningen Frontis Series. 6:231-251.
SADANI LK and MURTHY BS (2005) Estimation of potential evaporation from bare soil. Indian Institute of
Tropical Meteorology, NCL post, Pune 411 008. http://www.commonwealthknowledge.net. Accessed 2005
SIMÜNEK J, ŠEJNA M and VAN GENUCHTEN MT (1996) The HYDRUS-2D software package for
simulating water flow and solute transport in two dimensional variably saturated media. Version 1.0. IGWMC-
TPS-53. Int. Ground Water Modeling Ctr., Colorado School of Mines, Golden.
SIMÜNEK J, VOGEL T and VAN GENUCHTEN MT (1994) The SWMS-2D code for simulating water flow
and solute transport in two-dimensional variably saturated flow. U.S. Soil Salinity Laboratory, USDA,
Research Report No. 132: pp 196-197.
SNYDER RL, BALI K, VENTURA F and McPHERSON HG (2000) Estimation of evaporation from bare or
near bare soil. Journal of Irrigation and Drainage Engineering, 126(6): 399-403.
TOMASELLA J, PACHEPSKY Y, CRESTANA S and RAWLS WJ (2003) Comparison of two techniques to
develop pedotransfer functions for water retention. Soil Sci. Soc. Am. J. 67:1085-1092.
XINFA H, VIRIYASENAKUL V and DECHAO Z (1999) Design, construction and installation in drainage
lysimeter. Proceedings of 99 International Conference on Agricultural Engineering. Beijing, China,
December, 1999: 162-168.
YANFUL EK and MOUSAVI SM (2003) Quantifying evaporation from soils using experimental and
mathematical models. 6th International Conference on Acid Rock Drainage. Cairns, QLD, Australia, 12-18
July 2003:1185-1191.
8-4