You are on page 1of 9

Minerals Engineering 116 (2018) 143–151

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage:

Recovery potential of flotation tailings assessed by spatial modelling of T

automated mineralogy data

P. Büttnera, , I. Osbahra,b, R. Zimmermanna, T. Leißnerc, L. Satgea,d, J. Gutzmera,e
Helmholtz Zentrum Dresden Rossendorf, Helmholtz Institute Freiberg for Resource Technology, Germany
Deutschen Rohstoffagentur (DERA) in der BGR, Berlin, Germany
Institute of Mechanical Process Engineering and Mineral Processing, TU Bergakademie Freiberg, Germany
Department of Mineralogy, TU Bergakademie Freiberg, Germany


Keywords: The extraction of ore minerals from fine-grained flotation tailings is a commercially interesting but technolo-
Re-mining gically challenging endeavor that needs to be supported by a full technical and economic feasibility study. A
Remining novel approach to such an assessment is introduced here. It is illustrated by the example of a historic tailing
Tailing storage facility containing on average 0.2 wt% of Sn as cassiterite. Mineral processing test work identified flo-
3D modelling
tation as a suitable technology route to recover this cassiterite. The viability of flotation was attributed to three
Mine waste
material parameters, namely grade, liberation and particle size of cassiterite. These parameters were quantified
Zero waste approach for a set of ten exploration drill cores by chemical assay and mineral liberation analysis. For each of the three
MLA relevant parameters a suitable weighting function was defined that was applied to the entire data set. The data
Mineral liberation analysis was then geo-referenced and combined to construct a 3D model illustrating a depreciated grade, i.e., the amount
Processing of cassiterite-bound tin that can realistically be recovered from the tailings. Results of the case study illustrate
Re-processing the importance of combining chemical grade data with quantitative mineralogical and microfabric information
Reprocessing in any effort to objectively assess the residual value contained in industrial tailings or any other residue con-
sidered for re-processing.
Tin recovery
Spatial modelling
Automated mineralogy
Tailing storage facility

1. Introduction of industrial mining, such as many areas of Europe. Exploitation of TSFs

will make at least some contribution to meet the demand for metals and
Fine-grained residues hosted by industrial tailing storage facilities minerals of European industry and it is the second pillow of the Eur-
(TSFs) often contain significant amounts of minerals of commercial opean Raw Materials Initiative to foster the supply of such materials
interest (Macri et al., 2015; Louwrens et al., 2015). Such valuable from European sources (EU Commission, 2008, 2011). The drivers for
fractions may either comprise residual amounts of those minerals that research on TSFs in Europe, however, can never be reduced to resource
were the primary target of industrial exploitation but that escaped exploitation, but need to integrate concerns of environmental impact
previous beneficiation processes – or they may represent minerals that and sustainable rehabilitation of affected landscapes.
were previously not considered of commercial interest. TSFs should Anthropogenic ore deposits such as TSFs are readily exposed at
thus be regarded as low-grade ore deposits of anthropogenic origin. surface, have a well-defined tonnage and the raw material is already
This is of particular relevance to regions that have an extensive record crushed and milled –factors that will reduce financial risk and

Corresponding author.
E-mail address: (P. Büttner).
Received 27 April 2017; Received in revised form 27 August 2017; Accepted 16 September 2017
Available online 16 October 2017
0892-6875/ © 2017 Elsevier Ltd. All rights reserved.
P. Büttner et al. Minerals Engineering 116 (2018) 143–151

development time (Büttner et al., 2016) The fine-grained nature of the 3. Methods
raw material may be beneficial as less energy will be required for ex-
ploitation and comminution, but it may also have considerable detri- In a first step, all available historic data were compiled. Site visits
mental effects, as currently available (physical) beneficiation processes were undertaken to identify suitable drill locations. Two drilling cam-
struggle to be highly selective for very fine-grained particles. For his- paigns were executed; the first one to obtain a single drill core that was
toric TSFs this detriment is usually less problematic than for modern used to explore tailings characteristics and suitable beneficiation
TSFs. Former beneficiation processes were less effective than the state routes. The second campaign yielded nine additional cores that pro-
of the art technologies, it can thus be expected that historic tailings vided insight into the spatial distribution of tangible compositional
contains still higher concentrations of target minerals in suitable par- attributes for the resource potential model that was constructed using
ticle size fractions. The resource potential of TSFs may indeed be sig- simple yet robust linear interpolation methods.
nificant, but any exploitation will require quantification by a thorough
technical and economic feasibility study. 3.1. Drilling, sampling and sample preparation
It is important to stress here that every TSF has its own depositional
history and it will have its own, unique internal structure and compo- The surface of the Tiefenbachhalde has been modified considerably
sition. Relevant factors that will define the composition of the tailings subsequent to its closure in 1966. The surface was covered with soil,
include the primary characteristics of the ore processed, the processing cohesive material, construction waste and an asphalt layer mainly in
technology applied as well as the method of tailings deposition the southern half of the TSF (BIUG GmbH, 2013). Therefore, suitable
(Redwan et al., 2012). Every TSF will thus require exploration and drill sites could only be selected on the northern half of the TSF and it
technical feasibility studies similar in detail to those commonly applied was not possible to set up a regular drill grid pattern. Instead, it was
to primary ore deposits. Feasibility studies will depend heavily on a decided to drill along three main lines in irregular distances, in order to
sound understanding of relevant material characteristics, including constrain possible variations in the material composition in different
grade, particle and mineral grain size, liberation and deportment. spatial ranges (Fig. 2).
Furthermore, the spatial distribution of such characteristics within the The first drill hole was placed near the center of the TSF using a
volume of any investigated TSF needs to be well understood as TSF’s liner drilling method whereas the other nine holes were realized by an
cannot be regarded being of uniform composition. The latter task is of ultrasonic drilling technology. In the absence of a basal sealing layer
particular relevance because well-known spatial approaches to resource each drill core reached the topsoil immediately underlying the tailings
modeling (Kriging, etc.) should not be used, given that the assumption material, assuring that the total thickness of the tailings at each drill
of stationarity is in general not appropriate for tailings. point could be accurately constrained. Core samples were collected in
It is thus important to collate a comprehensive dataset for feasibility all cases in plastic liners of one meter length and a diameter of 11 cm.
studies to lower the risk of investment prior to embarking into com- Drilled material was exposed by cutting a window along the whole
mercial tailings reclamation. In this study, an innovative approach to length of the plastic liners. The cores were then photographed, logged
the definition of the resource potential of a TSF is presented in a case and channel sampled. The tailing material was found to be uniform in
where a physical beneficiation process (flotation) could be applied to macroscopic appearance. It was thus decided to always combine two
extract residual amounts of Sn (as cassiterite, SnO2) from a silicate-rich meters of drill core material into one sample, generating a total of 92
residue. Results were obtained as part of a publically funded research samples for analysis. The samples were air dried, because no changes by
project aimed to quantify the resource potential of large TSF’s from oxidation where expected as primary ores were virtually devoid of
historic mining operations in Saxony/Germany (Büttner et al., 2016; sulfides. The dry samples were carefully de-agglomerated by hand with
Büttner and Gutzmer, 2016). an agate mortar. Each sample was split by a Quantachrome rotary
powder sample splitter (Sieving Riffler™) until there was an adequate
2. Study area amount for preparation of polished grain mounts. These were used for
Mineral Liberation Analysis (MLA). Aliquots of samples were used for
The TSF “Tiefenbachhalde” in Altenberg/Saxony was established in wet sieve analysis and laser granulometry, and also for further grinding
1952 by Volkseigener Betrieb (VEB) Zinnerz Altenberg. Tailings origi- to analytical fineness for chemical assaying.
nated by processing of ores from the Altenberg Sn deposit, a world-class
greisen deposit located in the Eastern Erzgebirge, Saxony, Germany
(Weinhold, 2002). A 60 m high tailing dam wall was constructed in a 3.2. Surveying of tailings and data preparation for 3D modelling
steep valley to establish the Tiefenbachhalde. Two small streams (Tie-
fenbach and Schwarzwasser) were captured and canalized in tunnels 3.2.1. Topography
that run below the TSF. The TSF was in active use between 1953 and A historic topographic map of Altenberg (1:25.000; 1912; in Gauss-
1966 as storage facility for the residues of density separation processes Kruger Germany, zone 4) was digitized and used as pre-mining re-
generated during cassiterite concentrate production. The surface area of ference. The current topography was obtained via a drone-borne pho-
the Tiefenbachhalde measures nearly 350 meters from west to east and togrammetric survey using a fixed wing system equipped with a stan-
600 m from north to south. Being placed into a natural valley the dard RGB camera and GPS. Resulting nadir, geotagged aerial images
thickness of the TSF varies greatly, locally exceeding 30 m. Two sche- (ground resolution of 3.88 cm) were deployed to calculate a Digital
matic cross sections of the TSF are shown in Fig. 1. In the literature the Surface Model (DSM), ground resolution of 15.55 cm/px) using
total volume of the tailings deposited has been reported as 1.95 Mio m3 Structure-from-Motion (SfM) photogrammetry workflow implemented
(Weinhold, 2002 and references therein). Even average particle sizes as in Agisoft Photoscan. Further, an orthomosaic (3.88 cm/px resolution)
well as average mineralogical and chemical composition of the tailings was calculated from the images. The computed total XYZ error is given
material contained in the Tiefenbachhalde have been reported with 1.12 px (or ∼16cm).
(Weinhold, 2002 and references therein) (Table 1). Of particular re-
levance to the present study is the notion that only 50% of the Sn 3.2.2. Drill hole location
contained in the ore from the Altenberg deposit was recovered – with Locations of drill holes were surveyed using a Global Navigation
on average 0.22 wt% Sn lost to flotation tailings (Weinhold, 2002 and Satellite System (GNSS; Trimble R4-2) with a local base station
references therein). There is no conclusive information in literature (Trimble R5) setup. Surveyed points were base-line corrected to achieve
concerning the deportment of Sn, particle sizes of Sn-bearing minerals precise location within ± 2 cm. As reference system WGS84/UTM zone
and their liberation in the Tiefenbachhalde. 33N with the GRS80 ellipsoid was chosen.

P. Büttner et al. Minerals Engineering 116 (2018) 143–151

Fig. 1. Location of the Tiefenbachhalde in Germany (a), satellite image of the surface (b) and schematic cross sections (c) and (d); (Modified after Google Inc. (2017) and BIUG GmbH

3.2.3. Drill hole data 3.2.4. Data formats and projection systems
The drill hole data were gathered and converted into a text file In order to use the available data in the 3D model, data harmoni-
format. For each drill core sample a drill hole marker was created with zation was required. All data were converted into WGS84 UTM33N
an absolute elevation called “Marker Depth”. Table 2 shows the geographic reference system. The Digital Surface Model (DSM) TIFF file
column-based file of drill hole A1 taken as an example. Drill hole data was converted in a text file with the NUMWRIT algorithm of the
text files were converted and adjusted in order to import the properties Geomatica© software with a spatial resolution of 1.25 m. Due to dif-
in the 3D modeling software package GOCAD. ference reference ellipsoids between Gauss-Kruger and UTM projection,

P. Büttner et al. Minerals Engineering 116 (2018) 143–151

Table 1 Wet sieving was done using standard 200 mm analytical sieves of
Average composition of the tailing material in the Tiefenbachhalde, Altenberg (Weinhold, 0.315, 0.5 and 1.0 mm mesh size. The < 0.315 mm fraction was then
2002 and references therein).
used for laser diffraction measurement by a Sympatec HELOS instru-
Mineral wt.% Element wt.% Element wt.% ment equipped with a SUCELL wet dispersing unit. De-agglomeration of
fine particles was achieved by 3 min ultrasound treatment of the sus-
Quartz 50–55 Si 30–35 Sn 0.22 pension prior to laser diffraction measurement.
Feldspar 5–7 Fe 4–7 As 0.1
Mica 20–30 Al 2–8 Bi < 0.02
Topaz 8–15 Ca 0.5 Mo < 0.01
Clay minerals 2 Mg 0.1 S 0.02 3.3.2. Mineral liberation analysis
Hematite 5 Ti 0.1 F 1 For quantitative mineralogical and microfabric analysis of the
granular samples, a Mineral Liberation Analyzer (MLA) was used based
on a FEI-Quanta 650 F field emission scanning electron microscope (FE-
all elevation values measured were corrected by subtracting 48.966 m. SEM) in combination with two Bruker Quantax X-Flash 5030 energy
Thus, the elevation of the drill holes’ head and the DSM became co- dispersive X-ray spectroscopy (EDS) detectors. Relevant instrument
herent. In order to do so, the correction was applied manually in the parameters are shown in Table 3. Data were acquired on the FEI MLA
attribute table of the GPS Points Shapefile and automatically by the suite 3.1.4 software. More detailed information about the MLA system
QGIS raster calculator for the DSM TIFF file. Fig. 2 shows the drill hole can be found in Fandrich et al.(2007) and Gu (2003).
(here borehole) locations in the DSM. Chemical analyses by X-ray fluorescence (XRF, see Uhlig et al.,
2016) where used for cross-validation of Sn concentration from MLA
3.2.5. Discrete smooth interpolation calculated assays. Deviations between XRF assays and MLA calculated
Because the infill of the TSF is anisotropic with respect to its geo- assays data were found to be less than 20%.
metry, the prerequisites for standard interpolation algorithms (e.g.,
Kriging) like stationarity and isotropic behavior of parameters are not
valid. Thus, the application of the Discrete Smooth Interpolation (DSI) 3.3.3. Density measurements
method is only valid here. Density measurements were carried out using a micromeritics
AccuPyc II 1340 gas pycnometer. Prior to analysis the samples were
3.3. Analytical methods dried, de-agglomerated in a mortar and split into subsamples of ade-
quate size using a rotary sample divider. Two subsamples were then
3.3.1. Particle size analysis used for density measurement.
Particle size distributions were determined by combining wet sieve The resultant data set provides a comprehensive account of the
and laser diffraction analyses. Sieving was necessary due to the broad composition of the tailing material. Data are available in the electronic
particle size distribution, which exceeded the range for laser diffraction. supplementary to this manuscript (Appendix A).

Fig. 2. DSM acquired by drone showing the drill hole (here borehole) locations within a former topographic map of 1912 as background. The red line shows the representative cross
section used in Fig. 6; Geographic coordinate system and Datum: WGS84 UTM33N.

P. Büttner et al. Minerals Engineering 116 (2018) 143–151

Table 2
Drill holes location (WGS84/UMT33N), Top and bottom information relative to the Amsterdam Mean See Level (AOD).

Drill hole ID* A1 A3 A4 A5 A6

Location WGS84/UTM33N X 413399.122 413501.191 413469.726 413432.216 413399.219

Y 5623909.313 5623849.278 5623866.988 5623896.599 5623929.506

Elevation, Amsterdam Ordnance Datum Top Z 699.406 698.663 698.568 697.677 697.127
Bottom Z 668.406 673.663 673.568 683.677 687.127

Drill hole ID* A7 A8 A9 A10 A11

Location WGS84/UTM33 N X 413562.683 413539.355 413424.991 413398.63 413485.805

Y 5623919.949 5623686.111 5623735.317 5623806.407 5623714.617

Elevation, Amsterdam Ordnance Datum Top Z 699.105 698.718 696.523 696.999 697.151
Bottom Z 677.105 661.718 684.523 691.999 672.151

* Drill hole A2 was stopped after 2 m because of technical complications.

Table 3 extracted from chemical assay, particle grain size and MLA data. The
Summary of MLA instrument parameters. weighted parameters then provided the base for establishing a resource
potential model – essentially a block model of depreciated (recover-
SEM parameters MLA parameters
able) grade. This block model was constructed by linear interpolation.
Voltage [kV] 25 Scan speed 16
Working distance [mm] 13 Resolution [px] 1000 × 1000
Probe current [nA] 10 Pixel size [µm/px] 0.5 4.1. Parameter selection
Spot size 5.66 Acquisition time [ms] 5
Horizontal field width [µm] 500 GXMAP step size [px] 6 Assay data reveal an average Sn concentration of about 0.22 wt%,
Brightness 93.55 Minimum grain size [px] 4 with individual samples ranging from < 0.1 to 0.4 wt% Sn content.
Contrast 24.14
BSE calibration Cu 251
Cassiterite was identified as the only significant Sn bearing mineral,
rendering deportment assessments obsolete. Cassiterite is typically fine-
grained but well-liberated. Fig. 3 illustrate relevant particle size data
4. Results (X10/X50/X90) from drill core A1 in combination with concentration
data for Sn, As and S. This indicates that the Sn concentration is higher
In order to develop a resource potential model to illustrate the in finer particle size classes, which confirms previous statements
amount of cassiterite recoverable by flotation in the Tiefenbachhalde it (Gruner, 1981; Fahning et al., 1983). Also, the expected layered
was necessary to identify the most relevant material parameters for structure as a result of the deposition process is well visible in the data.
beneficiation. Weight factors were proposed for the parameters Fig. 4 shows the liberation classes of cassiterite and their frequency
in all 92 samples studied. Evidently, fine-grained cassiterite is

Fig. 3. Particle size distribution of drill core A1 and con-

centration of As, Sn und S by depth, analyzed within the
first drilling campaign on a logarithmic scale.

P. Büttner et al. Minerals Engineering 116 (2018) 143–151

4.2. Weighting functions

A weighting function (called index here) was created for each

parameter to describe the influence of the different material parameters
on the success of cassiterite flotation. The Cassiterite Index of
Concentration (IC) is a linear function of the cassiterite concentration
measured in weight percent (Fig. 5a). The Index of Liberation
(0 < IL < 1) is a non-linear function that scales with the actual de-
gree of surface liberation as determined by MLA (Fig. 5b). The Index of
Particle Size (0 < IP < 1, Fig. 5c) considers the particle size range of
the material favorable for flotation between 10 and 45 μm (Miettinen
et al., 2010).
Fig. 4. Histogram of the average frequency of liberation classes from 92 samples (com-
bined data from all 10 drill holes). The three indices IC, IP, and IL were then combined into a weighting
function named “LCF”. This acronym refers to “Liberation”,
“Concentration” and “Flotation”. LCF is calculated as a simple product:
particularly well liberated (> 80%). This suggests that such fine grains
were well liberated during comminution but not well concentrated by LCF = IC·IL·IP
flotation. Also, there is an abundance of poorly liberated (< 20%) Although this approach is admittedly simplistic, it is possible to define
cassiterite grains. Recovery of such poorly liberated (by surface) cas- those regions within the Tiefenbachhalde that are most suitable for the
siterite grains would require regrinding. In drill cores in the center of recovery of cassiterite by flotation. In a region where IL and IP are
the TSF the histogram shows fewer particles in the poorly liberated highly favorable, the LCF calculation is roughly equal to IC, i.e., grade.
classes, in drill holes closer to the border fewer in the well-liberated In regions where either IL or IP (or both) are very unfavorable, LCF is
classes. The reason is obviously gravity separation during the spilling notably lower than IC, i.e., the grade and cassiterite content are
(deposition) process, where fine-grained particles are transported to- markedly depreciated because of unfavorable physical properties that
wards the center of the tailing dam (lake) whereas coarse-grained will negatively affect recoveries.
particles are retained close to the spill point.
In agreement to literature data (Weinhold, 2002) major gangue 4.3. Densities
minerals were identified as quartz, feldspar, topaz and muscovite.
Considering these characteristics of the tailings material, flotation was Density determination is needed to translate the depreciated grade
readily identified as the most suitable technology to produce a cassi- information in combination with volume information into tonnages as
terite concentrate. The three main material parameters, which de- part of the resource potential model. Table 4 shows the mineral density
termine the success of the envisaged flotation process include grade, (values taken from FEI MLA software), area and weight percentage data
particle size and cassiterite liberation (Fahning et al., 1983; Gruner, for the main tailing minerals based on MLA analysis. Density mea-
1981, 1982). Quantitative values for these particular parameters are surements by pycnometer measurements on two individual samples
available from the data set obtained by XRF, MLA and laser granulo- yielded an average density (φ∅1) of 2.84 g/cm3 for the tailing material
metry. (Table 5). The measured density is in excellent agreement with the
average density (φ∅2 is 2.85 g/cm3) that has been calculated from

Fig. 5. Graphs of the different weighting functions used for (a) Index of Concentration IC, (b) Index of Liberation IL and (c) Index of Particle Size IP.

P. Büttner et al. Minerals Engineering 116 (2018) 143–151

Table 4 top soil below the TSF, drill core intercepts could be used to create the
Average composition of 92 samples from the Tiefenbachhalde for the main minerals; the tailing’s bottom surface. This was compared to the available Digital
content (by weight percent) is calculated from area % obtained from MLA analysis,
Terrain Model (DTM) of the former topography (see Section 3.2 for
multiplied by average mineral density, values taken from FEI MLA software.
details) – and minor deviations corrected. The drill core data and the
Mineral Weight Area Density interpolation results can be easily compared along a N-S section (Fig. 2)
% % g/cm3 through the 3D model (Fig. 6a –c). The results concerning the corre-
lations between cassiterite and hematite/topaz/bismuth minerals be-
Quartz 53.04 56.84 2.65
Kaolinite 7.76 8.56 2.60 fore and after the DSI was calculated by the rlm{MASS} R function
Muscovite 7.99 8.07 2.83 using the Mass package (Venables and Ripley, 2002). Parameters along
Biotite 8.12 7.52 3.10 the drill holes served as hard constraints for interpolation. Interpolation
Zinnwaldite 6.91 6.60 3.00 quality was evaluated by comparing the interpolation results to drill
Topaz 7.17 5.77 3.55
core data by successively removing single drill holes from the inter-
Hematite (Fe-Oxide) 4.32 2.34 5.30
Albite 1.22 1.34 2.62 polation.
Plagioclase 1.24 1.33 2.69
Fluorite 0.78 0.71 3.13 5. Discussion

Different to Louwrens et al. (2015) who used classified data to

Table 5 conduct a block modelling approach for resource estimation in a TSF,
Results of the density measurement from two representative samples from the we prefer a direct statistical interpolation and thresholding method in
GOCAD. This novel approach is much better suited to reconstruct the
Sample Mass Volume Density layered geometry that is inherent to TSFs. Our approach is also more
suited to explain subtle compositional variations of natural sedi-
Average Deviation Average Deviation mentation layers and, thus, lead to a more precise resource estimation.
g cm3 cm3 g/cm3 g/cm3 Based on our approach areas in the tailing can be defined more properly
A1 31.5536 11.1244 0.0016 2.8364 0.0004 for selective mining and processing.
A2 33.4478 11.7945 0.0032 2.8359 0.0008 Fig. 6 illustrates the 3D visualizations of the three single parameters
cassiterite concentration (a), cassiterite liberation (b), particle size
distribution in the tailing (c) and the combined LCF factor model (d)
averaged quantitative mineralogical information (Table 4) from MLA along the chosen NW-SE section across the TSF (see Fig. 2 for re-
(assuming that area% yielded by MLA data corresponds to Vol%). ference). The combination of the three indices into the unitless LCF
However, these two values describe the average mineral grain parameter (Fig. 6d) clearly highlights two areas within the TSF of
density without considering porosity. For porosity, it is assumed that particular resource potential. Comparison to sections that illustrate only
the fine-grained tailings comprise of a close-packing of equally-sized one of the three parameters (Fig. 6a–c) clearly document the advantage
spheres using the formula: to assess resource potential by taking into consideration all relevant
π parameters that influence the recoverability of cassiterite from the
≅ 0.74048 ≈ 74%
3√2 tailing residue.
The LCF value denotes the recoverable concentration of cassiterite
Finally, it is assumed that the available pore space is filled with air from a specific volume of the TSF. Results illustrate that the distribution
(φP = 0.0 g/cm3), resulting in an average density of 2.10 g/cm3 (φR) of the LCF value (recoverable concentration of cassiterite) is not uni-
for the tailing material stored in the Tiefenbachhalde. This density was form. Instead, a systematic picture emerges with two large lenses with
then applied to construct the resource potential model. The formula high concentrations of recoverable cassiterite (up to 0.15 wt% Sn).
which was used is: Every voxel, which is part of the interpolated volume, was assigned a
φR = φ∅·0.74 + φP ·0.26 LCF value. If that assigned value exceeded a chosen threshold value
(0.07), the cassiterite concentration was considered together with the
volume in the calculation for recoverable SnO2.
4.4. Volume calculation It was further important to determine the correct TSF volume and
contained tonnage, to calculate a reliable estimate of the total amount
The GOCAD© software uses a geologic grid, which corresponds to of recoverable tin in the Tiefenbachhalde TSF. The determination of the
Stratigraphic Grids (SGrid). It is a flexible 3D grid that can model a total volume of tailing material contained in the TSF was found to be
volume between two boundary surfaces, here the drone-borne DSM and the greatest challenge. Comparison between the volume reported in
the former topographic surfaces. According to the layered structure of literature (Weinhold, 2002) and the interpolated volume is shown in
mining residues resulting from the deposition technology (Redwan Table 6. Our interpolation suggested a TSF volume of 2.7 mm3. This
et al., 2012), a structured grid with irregular geometry was constructed volume was determined by taking into account drill hole data, original
(vertical section: same number of cells, parallel to top and the bottom topography as well as the DEM. This is considerably larger than the
surfaces). In this study, a vertical resolution of 1 m (self-centered SGrid) reported volume of 1.95 mm3 of tailings deposited in the TSF from
and a lateral resolution of 4 m were selected. The drill core data were production records from the former mining company VEB Zinnerz Al-
imported into Gocad as PointsSets. The Discrete Smooth Interpolation tenberg (Weinhold, 2002). Evidently, the volume obtained in our as-
was given preference, because it is the only interpolation method that sessment was affected considerably by the fact that the TSF was covered
respects the sub horizontal bedding of the material. The IC, IL and IP with cohesive material, construction waste and soil for remediation
properties held by the PointsSets object were used as hard constraints after active deposition of tailings between 1967 and 1975 and ad-
on the SGrid in order to interpolate them. The interpolated tailing vo- ditionally by construction material by the new owner after 1990 (BIUG
lume is 2.7 mm3. It represents about 2/3 of today’s total volume GmbH, 2013).
(3.17 mm3), calculated from the remote sensing data by using the DEM The actual shape and volume obtained in our assessment is thus a
as upper border. The total tailing volume cannot be interpolated, be- combination of the cassiterite containing tailing material from the
cause more drill core data would be necessary. former process and the deposited construction/cohesive material at the
Because most of the drill cores (except A2) intersected the former surface. The liner samples from our drill campaign intersected a cover

P. Büttner et al. Minerals Engineering 116 (2018) 143–151

Fig. 6. Spatial distribution of index figures along

a representative N-S cross section through the
Tiefenbachhalde. [a] Cassiterite concentration
(red: 0.4 wt% Sn), [b] Mineral liberation (red:
80%), [c] Mean particle size (red: 30 µm), [d] LCF
value (red: LCF > 0.07).

Table 6 1.95 mm3 of process residues reported by (Weinhold, 2002). Assuming

Comparison between literature reported and calculated tailing volume and estimation of a density of 2.1 g/cm3 this volume translates into a total tonnage of 4.1
total and recoverable SnO2 content (interpolated from GOCAD model). Mt (Table 6).
Source Weinhold (2002) GOCAD Model

Volume in m3 1,950,000 2,700,000

6. Conclusions
Calculated Mass in t (φR = 2.10 t/m3) 4,095,000 5,670,000
Total (interpolated) SnO2 content in t n.a.* 4970 The approach suggested in this study is best described as a geo-
Recoverable SnO2 content in t n.a.* 2038 metallurgical 3D-model to illustrate the resource potential of fine-
(LCF > 0.07)
grained material contained in TSFs. This assessment is based on a
* Not available from literature data.
combination of chemical analysis as well as quantitative mineralogical
data and an accurate assessment of the geometric shape, volume and
layer of up to 5 m thickness above the actual tailings – and the presence tonnage of the TSF. By introducing weighting indices, the approach
of construction material heaps up to 7 m high is evident. All of this has quantifies inevitable losses that are due to unfavorable mineralogical or
an obvious influence on our volume calculations (Fig. 7). As there is no physical (i.e. liberation or particle size) attributes of the ore minerals to
available documentation regarding the amount of material deposited on be recovered. The approach may be considerably refined by taking into
the TSF subsequent to its active use, we decided to use the volume of account the results of actual beneficiation test work when constructing
the weighting function. Based on analytical data and subsequent 3D

P. Büttner et al. Minerals Engineering 116 (2018) 143–151

Fig. 7. Tailing volume information from (a) GOCAD model

calculated with the actual DEM surface, (b) Interpolated
drill hole (well) data (ground level surface), and (c)
Literature information (Weinhold, 2002).

modelling, the calculated recoverable cassiterite content from the tail- und Mineralien aus sächsischen Bergbauhalden. [Hrsg.] Christian Peter Brandstetter,
Gudrun Gräbe, Lars Rentsch Anke Dürkoop. Innovative Technologien für
ings is 2038 t (Table 6) of a total of 4970 t of SnO2 contained in the TSF. Ressourceneffizienz - Strategische Metalle und Mineralien. Fraunhofer Verlag,
This, in turn, would translate into a recovery of 41%. Freiberg, 23, S. 345–360.
The method introduced here will be further developed by including EU Commission, 2011. Tackling the Challanges in Commodity Markets and on Raw
Materials, Brussels. Communication from the Commission to the European
geophysical data in order to maximize the quality of the resource model Parliament, The Council, the European Economic and Social Committee and the
and minimize analytical cost. Further, more research in understanding Committee of the Regions: COM, 25 final.
the depositional characteristics of tailing material is needed to refine EU Commission, 2008. The raw materials initiative — meeting our critical needs for
growth and jobs in Europe, Brussels. Communication from the Commission to the
modeling approaches. European Parliament and the Council – COM, 699 final.
Fahning, E., Rössel, H., Weigel, S., 1983. Zinn aus Haldensanden, Freiberg. Bergarchiv
Acknowledgements Freiberg ZE-Alt THA-Nr. III Ae60, lfd. Nr. 319.
Fandrich, R., Gu, Y., Burrows, D., Moeller, K., 2007. Modern SEM-based mineral libera-
tion analysis. Int. J. Miner. Process. Bd. 84 (1) S. 310–320.
The authors gratefully acknowledge funding of the SMSB research Google Inc., 2017. Google Maps. [Online]. [Cited: February 9, 2017.] < https://www.
project by the German Federal Ministry of Education and Research > .
(BMBF) as part of the r3 research program (FKZ: 033R095). We ap- Gruner, H., 1982. Zinn aus Haldensanden, Aufbereitung der Haldensande.
Forschungsinstitut für Aufbereitung, Freiberg, S. 60.
preciate discussion and support by research partners at the TU Gruner, H. 1981. Zinn aus Haldensanden, Aufbereitung der Haldensande, Bericht zur
Bergakademie Freiberg, G.E.O.S. Ingenieurgesellschaft mbH, SAXONIA Leistungsstufe V2. Forschungsinstitut für Aufbereitung. Freiberg, S. 34.
Standortentwicklungs- und verwaltungsgesellschaft mbH (Freiberg) Gu, Y., 2003. Automated scanning electron microscope based mineral liberation analysis.
J. Miner. Mater. Charact. Eng Bd. 2 (1) S. 33–41.
and AKW Apparate + Verfahren Anlagenbau GmbH & Co. KG Louwrens, E., Napier-Munn, T., Keeney, L., 2015. Geometallurgical characterisation of a
(Amberg). tailings storage facility – a novel approach, pp. 125–131.
Macri, D., Graedel, T., Reck, B., 2015. Applying Circular Economy Principles to the Mine
and Mineral Processing Industry: Reunsing and Designing the Disposal of Mine Waste
Appendix A. Supplementary material for Future Metal Recovery. Environmental Management, Yale University. New
Haven, Connecticut, United States of America. Circular Economy Innovation Project,
Supplementary data associated with this article can be found, in the Schmidt-MacArthur Fellowship.
Miettinen, T., Ralston, J., Fornasiero, D., 2010. The limits of fine particle flotation. Miner.
online version, at Eng. 23 (5), 420–437.
Redwan, M., Rammlmair, D., Meima, J.A., 2012. Application of mineral liberation ana-
References lysis in studying micro-sedimentological structures within sulfide mine tailings and
their effect on hardpan formation. Sci. Total Environ. 414, 480–493.
Uhlig, S., Möckel, R., Pleßow, A., 2016. Quantitative analysis of sulphides and sulphates
BIUG GmbH, 2013. Tiefenbachhalde in Altenberg – Ergänzende Erkundung, by WD-XRF: capability and constraints. X-Ray Spectromet. 45, 133–137 January 27.
Standsicherheitsberechnung, Nutzungsempfehlungen. Freiberg. Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics with S, fourth ed. Springer,
Büttner, P., Gutzmer, J., 2016. Re-Mining – Gewinnung strategischer Metalle und anderer s.l. ISBN 0-387-95457-0.
Mineralien aus sächsischen Bergbauhalden. In: Thomé-Kozmiensky, Karl J. (Ed.), Weinhold, G., 2002. Bergbau in Sachsen – Die Zinnerzlagerstätte Altenberg/
Mineralische Nebenprodukte und Abfälle, Band 3. TK Verlag Karl Thomé- Osterzgebirge. [Hrsg.] Sächsisches Oberbergamt Sächsisches Landesamt für Umwelt
Kozmiensky, s.l., pp. 383–393. und Geologie. Freiberg, Bd. 9.
Büttner, P., Osbahr, I., Luhmer, R., Pilz, C., Uhlig, S., Leißner, T., Pätzold, C., Scheel, M.,
Jahns, C., Martin, M., Gutzmer, J., 2016. SMSB – Gewinnung strategischer Metalle