6
finally using the line charts, the mine personnel could go
through the data minute by minute and see how the dust
data correlated with the production data or other factors to
determine why that day may have failed.
When a high dust day was identified, we found that
working backward from the calendar view, to the hourly
heatmap, then to the line charts not overly cumbersome
and overall, less effort than trying to constantly monitor
the live feed data for the same information. The result of
these interactions with the mine resulted in weekly reports
for the last seven days being automatically emailed out to
everyone on Friday mornings so that the industrial hygien-
ist at the mine had time to see the data before the end of the
week and make any necessary changes. Additionally, at the
end of each month, a monthly report like Figure 1B and
1C would be sent out to view the month, which is good for
keeping records as well as comparing to previous months’
reports.
Well before we had an established system to display the
data and aggregate the information in the weekly reports,
we saw the first practical “data-to-action” application for
using the deployed sensors. Within the first few weeks, the
foreman on site had identified a pattern of high dust con-
centration when one of the conveyor belts that fed a specific
screen would start up, and upon investigation substantial
spillage was found from the backside of the belt, and it
was replaced with an auger lift system. Figure 2A depicts
the week leading up to the change, where evident days of
production can be seen in the dust levels, as well as low
levels over the weekends. Figure 2B shows the seven days
following the change where the dust levels rarely exceed
weekend dust levels. With this simple change, using parts
the mine personal already had on-site, mine personnel were
able to reduce the dust level around that monitor 3-fold
(Figure 2C). It can also be seen that the weekend dust lev-
els were similar before and after the change (Figure 2D)
indicating that the observed change is explicitly linked to
a reduction in airborne dust during production. This type
of confirmation has great utility in ensuring that the engi-
neering control is performing in the way expected, not only
after initial installation but also after continued use.
One of the major benefits of an LCDM system like
the one described here is access to high-resolution histori-
cal data. Using the full year of data, we were able to discern
seasonal, weekly, and hourly trends in the respirable dust
levels within the plant. Figure 3A depicts a Z-score heat-
map of daily TWA respirable dust data from all sensors for
365 days. A Z-score is a numerical value that represents
how far a given data point is from the mean of that sen-
sor, measured in terms of standard deviations. It is a way
to standardize data for comparison across different sen-
sors. Functionally, blue days represent lower than average,
Figure 2. Respirable dust concentrations following the change from a belt-fed to an auger-
fed size-selective screen on 9/27. A. Scatter plot representing the respirable dust levels for
7 days before the change in blue and (B) the 7 days following the change in orange. C. Bar
graph representing the average respirable dust concentration before and after the change.
D. Depicts the same data but over a weekend showing minimal difference.
finally using the line charts, the mine personnel could go
through the data minute by minute and see how the dust
data correlated with the production data or other factors to
determine why that day may have failed.
When a high dust day was identified, we found that
working backward from the calendar view, to the hourly
heatmap, then to the line charts not overly cumbersome
and overall, less effort than trying to constantly monitor
the live feed data for the same information. The result of
these interactions with the mine resulted in weekly reports
for the last seven days being automatically emailed out to
everyone on Friday mornings so that the industrial hygien-
ist at the mine had time to see the data before the end of the
week and make any necessary changes. Additionally, at the
end of each month, a monthly report like Figure 1B and
1C would be sent out to view the month, which is good for
keeping records as well as comparing to previous months’
reports.
Well before we had an established system to display the
data and aggregate the information in the weekly reports,
we saw the first practical “data-to-action” application for
using the deployed sensors. Within the first few weeks, the
foreman on site had identified a pattern of high dust con-
centration when one of the conveyor belts that fed a specific
screen would start up, and upon investigation substantial
spillage was found from the backside of the belt, and it
was replaced with an auger lift system. Figure 2A depicts
the week leading up to the change, where evident days of
production can be seen in the dust levels, as well as low
levels over the weekends. Figure 2B shows the seven days
following the change where the dust levels rarely exceed
weekend dust levels. With this simple change, using parts
the mine personal already had on-site, mine personnel were
able to reduce the dust level around that monitor 3-fold
(Figure 2C). It can also be seen that the weekend dust lev-
els were similar before and after the change (Figure 2D)
indicating that the observed change is explicitly linked to
a reduction in airborne dust during production. This type
of confirmation has great utility in ensuring that the engi-
neering control is performing in the way expected, not only
after initial installation but also after continued use.
One of the major benefits of an LCDM system like
the one described here is access to high-resolution histori-
cal data. Using the full year of data, we were able to discern
seasonal, weekly, and hourly trends in the respirable dust
levels within the plant. Figure 3A depicts a Z-score heat-
map of daily TWA respirable dust data from all sensors for
365 days. A Z-score is a numerical value that represents
how far a given data point is from the mean of that sen-
sor, measured in terms of standard deviations. It is a way
to standardize data for comparison across different sen-
sors. Functionally, blue days represent lower than average,
Figure 2. Respirable dust concentrations following the change from a belt-fed to an auger-
fed size-selective screen on 9/27. A. Scatter plot representing the respirable dust levels for
7 days before the change in blue and (B) the 7 days following the change in orange. C. Bar
graph representing the average respirable dust concentration before and after the change.
D. Depicts the same data but over a weekend showing minimal difference.