1046 XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3
METHODOLOGY
A detailed sampling campaign is conveyed on the shal-
low zone (3 m) of a historical TSF located in Freiberg,
Germany. All 68 samples are characterized with X-ray fluo-
rescence (XRF) and a subset of 36 samples is also character-
ized with the Mineral Liberation Analyser (MLA). Samples
are clustered based on these preliminary characterization
results, dividing the TSF into four regions. Composite
samples of each of these regions as well as a global compos-
ite sample are used for flotation tests with the goal of bulk
sulfide recovery. The processing products of each of these
tests are characterized with XRF and MLA to investigate
the flotation behavior of material from the different parts
of the TSF. Besides, the MLA data collected from the pro-
cessing products of the global composite is used for train-
ing a particle-based separation model, which can estimate
the potential sulfide recovery of all samples from the TSF
that were analyzed with the MLA. A recovery model of the
TSF is then computed by interpolating sulfide recovery
throughout the TSF with ordinary kriging.
Sampling and Clustering
The Davidschacht TSF, located in Freiberg, Germany, is
the focus of the current investigation. This TSF was built
between 1944 and 1969 with tailings slurry from the
processing of polymetallic hydrothermal vein ores of the
Freiberg mining district to produce galena, sphalerite, and
pyrite concentrates. After the deposition phase, a layer of
coarse sand, topsoil, and plants was deposited on top of
the TSF. Building rubble and household waste were also
dumped later on the southern part of the TSF, where sam-
pling was avoided (Blannin et al., 2022).
The sampling protocol and analytical scheme were pro-
posed by Blannin et al. (2022), who provide further details
on these procedures. Figure 1 shows an aerial view over the
TSF and marks the location of the 68 samples, taken with a
jackhammer drill with 1 m-long cylinder attachments and
extension bars. The shallow zone (3 m) was sampled on a
30 m grid with additional nested grids of ca. 15 and 7.5 m,
random samples, and twin samples. Sampling was limited
to the region where human activities were limited after the
closure of operations. An 8 cm-wide core bit was used for
sampling the first meter, while a core bit of 4 cm width was
used for the subsequent meters. Soil from above the tailings
material was removed and samples were collected in inter-
vals of 1 m. In locations where the depth of soil above the
tailings material approached 1 m, samples were collected
down to 4 m depth to ensure that 3 m of sample material
was collected at every point. Given the high and inhomo-
geneous level of oxidation of the first meter sampled in each
point, it was left out of this study. A 1–3 m blend sample
was produced for each point by drying, homogenizing, and
splitting the samples from the 1–2 m and 2–3 m intervals
to aliquots of 100–250 g and mixing equal splits of both
depth intervals.
The 68 samples collected were analyzed as pressed
pellets of Ø40 mm with XRF at the Helmholtz Institute
Freiberg for Resource Technology (HIF) using a PANalytical
Axiosmax Minerals wavelength-dispersive XRF. Major ele-
ments (Na, Mg, Al, Si, S, K, P, Ca, Fe, Ti) were evaluated
using the semi-quantitative Omnian program and trace ele-
ments (As, Ba, Cu, F, Mn, Ni, Pb, Rb, Sn, Sr, Zn) using
the fully calibrated Protrace program. Twenty samples were
analyzed at Actlabs in Canada with Inductively Coupled
Plasma Optical Emission Spectroscopy (ICP-OES) and
Inductively Coupled Plasma Mass Spectroscopy (ICP-MS)
for validation purposes.
A hierarchical clustering method is used for clustering
the samples into regions of similar chemical composition.
Figure 1. Satellite view of the TSF with marked sampling
points
METHODOLOGY
A detailed sampling campaign is conveyed on the shal-
low zone (3 m) of a historical TSF located in Freiberg,
Germany. All 68 samples are characterized with X-ray fluo-
rescence (XRF) and a subset of 36 samples is also character-
ized with the Mineral Liberation Analyser (MLA). Samples
are clustered based on these preliminary characterization
results, dividing the TSF into four regions. Composite
samples of each of these regions as well as a global compos-
ite sample are used for flotation tests with the goal of bulk
sulfide recovery. The processing products of each of these
tests are characterized with XRF and MLA to investigate
the flotation behavior of material from the different parts
of the TSF. Besides, the MLA data collected from the pro-
cessing products of the global composite is used for train-
ing a particle-based separation model, which can estimate
the potential sulfide recovery of all samples from the TSF
that were analyzed with the MLA. A recovery model of the
TSF is then computed by interpolating sulfide recovery
throughout the TSF with ordinary kriging.
Sampling and Clustering
The Davidschacht TSF, located in Freiberg, Germany, is
the focus of the current investigation. This TSF was built
between 1944 and 1969 with tailings slurry from the
processing of polymetallic hydrothermal vein ores of the
Freiberg mining district to produce galena, sphalerite, and
pyrite concentrates. After the deposition phase, a layer of
coarse sand, topsoil, and plants was deposited on top of
the TSF. Building rubble and household waste were also
dumped later on the southern part of the TSF, where sam-
pling was avoided (Blannin et al., 2022).
The sampling protocol and analytical scheme were pro-
posed by Blannin et al. (2022), who provide further details
on these procedures. Figure 1 shows an aerial view over the
TSF and marks the location of the 68 samples, taken with a
jackhammer drill with 1 m-long cylinder attachments and
extension bars. The shallow zone (3 m) was sampled on a
30 m grid with additional nested grids of ca. 15 and 7.5 m,
random samples, and twin samples. Sampling was limited
to the region where human activities were limited after the
closure of operations. An 8 cm-wide core bit was used for
sampling the first meter, while a core bit of 4 cm width was
used for the subsequent meters. Soil from above the tailings
material was removed and samples were collected in inter-
vals of 1 m. In locations where the depth of soil above the
tailings material approached 1 m, samples were collected
down to 4 m depth to ensure that 3 m of sample material
was collected at every point. Given the high and inhomo-
geneous level of oxidation of the first meter sampled in each
point, it was left out of this study. A 1–3 m blend sample
was produced for each point by drying, homogenizing, and
splitting the samples from the 1–2 m and 2–3 m intervals
to aliquots of 100–250 g and mixing equal splits of both
depth intervals.
The 68 samples collected were analyzed as pressed
pellets of Ø40 mm with XRF at the Helmholtz Institute
Freiberg for Resource Technology (HIF) using a PANalytical
Axiosmax Minerals wavelength-dispersive XRF. Major ele-
ments (Na, Mg, Al, Si, S, K, P, Ca, Fe, Ti) were evaluated
using the semi-quantitative Omnian program and trace ele-
ments (As, Ba, Cu, F, Mn, Ni, Pb, Rb, Sn, Sr, Zn) using
the fully calibrated Protrace program. Twenty samples were
analyzed at Actlabs in Canada with Inductively Coupled
Plasma Optical Emission Spectroscopy (ICP-OES) and
Inductively Coupled Plasma Mass Spectroscopy (ICP-MS)
for validation purposes.
A hierarchical clustering method is used for clustering
the samples into regions of similar chemical composition.
Figure 1. Satellite view of the TSF with marked sampling
points