XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 415
novel and the successful results achieved are believed to be
unprecedented at this scale.
Static Testing
One or more pre-characterised samples are measured
repeatedly to characterise MR measurement precision and
accuracy.
Procedure
MRA measurements are produced for each sample and
assays are performed on subsamples thereof. The MRA pro-
duces a measurement for 100% of the sample material in
each case, whereas standard assay techniques are performed
on samples typically of only a few grams. Proper sampling
and assaying methodologies must be maintained to mini-
mize error introduced as a result of measuring different
samples.
To reduce the potential for error, ore samples were
crushed finely and pre-characterized by Atomic Absorption
Spectrometer (AAS) assay at the KMP site laboratory.
MRA Sample Measurement Procedure
Each sample was made up of three bags which
were laid out along the 1.1 m length of the sen-
sor to approximately replicate typical ore loading
of the sensor. While the conveyor belt was stopped,
and with a NextOre representative either remotely
or physically present, the MRA was disabled and
samples positioned manually inside the MRA
sensor opening area. The MRA was then restarted
and operated for approximately five minutes to
collect measurements of the sample in 4 second
intervals for approximately 300 individual mea-
surements per sample.
Static Testing Results
Figure 9 shows results of static measurements for accuracy
with an R2 value of 0.98.
One low-grade outlier was removed from the data
set. Two high grade outliers are observed but have been
retained in the results. It is observed that some particularly
high conductivity ore produces an attenuation effect in
response signal at high grades. Pods of feed grade exceeding
4.0% copper were regularly observed during operation of
the MRA at Kansanshi. These high grades exceed sorting
cut-off grades by a sufficiently large margin that the attenu-
ation does not impact sorting results.
Figure 10 shows signal magnitude of MRA measure-
ments over time during the sample testing program. Data
in the chart is made up of successive 4-second measurement
intervals and the observed noise gives measurement preci-
sion. When converted from magnitude to copper grade
percentages, this grade variability represents a 1-sigma stan-
dard deviation of 0.014% copper and a 0.045% copper
lower detection limit.
Figure 9. Measurement of signal magnitude for crushed ore samples compared to the KMP
metallurgical laboratory
novel and the successful results achieved are believed to be
unprecedented at this scale.
Static Testing
One or more pre-characterised samples are measured
repeatedly to characterise MR measurement precision and
accuracy.
Procedure
MRA measurements are produced for each sample and
assays are performed on subsamples thereof. The MRA pro-
duces a measurement for 100% of the sample material in
each case, whereas standard assay techniques are performed
on samples typically of only a few grams. Proper sampling
and assaying methodologies must be maintained to mini-
mize error introduced as a result of measuring different
samples.
To reduce the potential for error, ore samples were
crushed finely and pre-characterized by Atomic Absorption
Spectrometer (AAS) assay at the KMP site laboratory.
MRA Sample Measurement Procedure
Each sample was made up of three bags which
were laid out along the 1.1 m length of the sen-
sor to approximately replicate typical ore loading
of the sensor. While the conveyor belt was stopped,
and with a NextOre representative either remotely
or physically present, the MRA was disabled and
samples positioned manually inside the MRA
sensor opening area. The MRA was then restarted
and operated for approximately five minutes to
collect measurements of the sample in 4 second
intervals for approximately 300 individual mea-
surements per sample.
Static Testing Results
Figure 9 shows results of static measurements for accuracy
with an R2 value of 0.98.
One low-grade outlier was removed from the data
set. Two high grade outliers are observed but have been
retained in the results. It is observed that some particularly
high conductivity ore produces an attenuation effect in
response signal at high grades. Pods of feed grade exceeding
4.0% copper were regularly observed during operation of
the MRA at Kansanshi. These high grades exceed sorting
cut-off grades by a sufficiently large margin that the attenu-
ation does not impact sorting results.
Figure 10 shows signal magnitude of MRA measure-
ments over time during the sample testing program. Data
in the chart is made up of successive 4-second measurement
intervals and the observed noise gives measurement preci-
sion. When converted from magnitude to copper grade
percentages, this grade variability represents a 1-sigma stan-
dard deviation of 0.014% copper and a 0.045% copper
lower detection limit.
Figure 9. Measurement of signal magnitude for crushed ore samples compared to the KMP
metallurgical laboratory