7
and red days represent higher than average for that sensor.
From this panel of the figure, it can be seen, that there is
a clear pattern of overall dust level increase for all sensors
starting in early November 2022 compared to September
and October. Additionally, weekends, not surprisingly, can
be seen as prominently blue vertical stripes down the right
side of each heatmap which are more apparent in the screen
house than in the dry house. Dry 2 has a grey portion of the
data from September-December where the original sensor
installed had failed and a new one had yet to be installed
resulting in a loss of around 500,000 timestamps (Table 1).
Because each heatmap is normalized to its yearly average
these types of visualizations allow for a more comprehensive
perspective on the day-to-day changes that go on through-
out both buildings where the sensors are deployed.
Using the same TWA information, we next sought to
parse the data by day of the week to determine if the day of
the week had any impact on overall dust levels (Figure 3B).
In general, the highest days in terms of concentration levels
tended to be earlier in the week with Friday, Saturday, and
Sunday exhibiting the lowest dust levels. When observing
the hourly respirable dust data and organizing by hour in
the day, an additional pattern of elevated levels from 6 a.m.
to 6 p.m. can be seen, which follows along with the stan-
dard work shifts as reported by the mine (Figure 3C). From
panels B and C, the overall dust levels can be discerned as
Dry 2 having the highest overall dust levels while Dry 1 had
the lowest, and additionally, Dry 1 was not as impacted by
day of the week or hour of the day as much as other sensors.
After using the LCDM system for one year, we met
with the mine to get their feedback on the pros and cons of
the system and how the data could better serve their needs.
We have synthesized our interactions with this mine into
three major takeaways from this case study:
1) The live feed of the data is very helpful, but
often too much information to be monitored continu-
ally. After a period of around 2–3 months, the industrial
hygienist at the mine felt that there was a high enough level
of understanding of the data live stream that they would
prefer to get weekly summary reports of the dust levels on
site. From that point on, the weekly reports were moni-
tored, and when a high dust level day occurred, they felt
that looking back seven days to identify a potential source
of the dust was not too large of a burden. Monthly reports
were also implemented but were too infrequent and hard to
identify the source of dust when looking 20–30 days into
the past. A combination of the weekly reports to monitor
overall trends and the live feed to monitor specific short-
term dust levels was ultimately used by the on-site indus-
trial hygienist.
2) The first hour of startup is always the highest dust
levels. This pattern was first identified by the industrial
hygienist on-site through the observation of the early morn-
ing cyclical high concentration events that only occurred
on production days. Once the mine figured out this pat-
tern, they began to alter their start-up routines to purposely
avoid overlapping with times when workers will be near
the machines. This was fully identified, assessed, and put
into operation by the industrial hygienist on-site with no
suggestions or interventions from the NIOSH researchers
and is a perfect example of empowering industrial hygien-
ist professionals to make “best practice” decisions for their
workers by using data generated from these sensors.
3) Mobility of the system is important. At this spe-
cific site, there were both power and Wi-Fi limitations as
to the potential location of the sensors which limited the
functionality of the system. The mine has expressed interest
in purchasing more sensors and running power and ether-
net ports to new locations to install more. The industrial
hygienist on site wanted to be able to use the sensors to
identify problematic dust areas, create mitigation strategies,
confirm the efficacy of the intervention, and move the sen-
sors to a new location to repeat the process but were unable
to fully carry out the full scope of this plan due to sensor
location restrictions.
DISCUSSION AND CONCLUSION
This case study depicting the interactions between a sand
mine in Wisconsin and the researchers at NIOSH lays the
groundwork for the future use and implementation of low-
cost dust monitors in the mining environment. Through
this publication, we have discussed what is needed to set up
a system of LCDMs, what the expected outcomes could be,
and how the data can be used. At NIOSH, we see the need
for wider adoption of sensors and more data collection but
believe that the value added should be on pace with the
increased burden on the industrial hygienist professionals
who will be responsible for carrying out the actions. These
types of data and in general data of the future is only grow-
ing more and more complex, so having effective ways to
convey the meaning of complex data to the right people at
the correct time is extremely important. Through this col-
laboration, we at NIOSH have been able to answer many
questions about the viability of using LCDMs in an opera-
tional environment, but it also brought up many questions
that remain to be answered.
One of the major questions that remains unanswered
is how many sensors are needed to accurately capture the
variability within a given location. The answer to this ques-
tion is probably a balance between increasing the resolution
and red days represent higher than average for that sensor.
From this panel of the figure, it can be seen, that there is
a clear pattern of overall dust level increase for all sensors
starting in early November 2022 compared to September
and October. Additionally, weekends, not surprisingly, can
be seen as prominently blue vertical stripes down the right
side of each heatmap which are more apparent in the screen
house than in the dry house. Dry 2 has a grey portion of the
data from September-December where the original sensor
installed had failed and a new one had yet to be installed
resulting in a loss of around 500,000 timestamps (Table 1).
Because each heatmap is normalized to its yearly average
these types of visualizations allow for a more comprehensive
perspective on the day-to-day changes that go on through-
out both buildings where the sensors are deployed.
Using the same TWA information, we next sought to
parse the data by day of the week to determine if the day of
the week had any impact on overall dust levels (Figure 3B).
In general, the highest days in terms of concentration levels
tended to be earlier in the week with Friday, Saturday, and
Sunday exhibiting the lowest dust levels. When observing
the hourly respirable dust data and organizing by hour in
the day, an additional pattern of elevated levels from 6 a.m.
to 6 p.m. can be seen, which follows along with the stan-
dard work shifts as reported by the mine (Figure 3C). From
panels B and C, the overall dust levels can be discerned as
Dry 2 having the highest overall dust levels while Dry 1 had
the lowest, and additionally, Dry 1 was not as impacted by
day of the week or hour of the day as much as other sensors.
After using the LCDM system for one year, we met
with the mine to get their feedback on the pros and cons of
the system and how the data could better serve their needs.
We have synthesized our interactions with this mine into
three major takeaways from this case study:
1) The live feed of the data is very helpful, but
often too much information to be monitored continu-
ally. After a period of around 2–3 months, the industrial
hygienist at the mine felt that there was a high enough level
of understanding of the data live stream that they would
prefer to get weekly summary reports of the dust levels on
site. From that point on, the weekly reports were moni-
tored, and when a high dust level day occurred, they felt
that looking back seven days to identify a potential source
of the dust was not too large of a burden. Monthly reports
were also implemented but were too infrequent and hard to
identify the source of dust when looking 20–30 days into
the past. A combination of the weekly reports to monitor
overall trends and the live feed to monitor specific short-
term dust levels was ultimately used by the on-site indus-
trial hygienist.
2) The first hour of startup is always the highest dust
levels. This pattern was first identified by the industrial
hygienist on-site through the observation of the early morn-
ing cyclical high concentration events that only occurred
on production days. Once the mine figured out this pat-
tern, they began to alter their start-up routines to purposely
avoid overlapping with times when workers will be near
the machines. This was fully identified, assessed, and put
into operation by the industrial hygienist on-site with no
suggestions or interventions from the NIOSH researchers
and is a perfect example of empowering industrial hygien-
ist professionals to make “best practice” decisions for their
workers by using data generated from these sensors.
3) Mobility of the system is important. At this spe-
cific site, there were both power and Wi-Fi limitations as
to the potential location of the sensors which limited the
functionality of the system. The mine has expressed interest
in purchasing more sensors and running power and ether-
net ports to new locations to install more. The industrial
hygienist on site wanted to be able to use the sensors to
identify problematic dust areas, create mitigation strategies,
confirm the efficacy of the intervention, and move the sen-
sors to a new location to repeat the process but were unable
to fully carry out the full scope of this plan due to sensor
location restrictions.
DISCUSSION AND CONCLUSION
This case study depicting the interactions between a sand
mine in Wisconsin and the researchers at NIOSH lays the
groundwork for the future use and implementation of low-
cost dust monitors in the mining environment. Through
this publication, we have discussed what is needed to set up
a system of LCDMs, what the expected outcomes could be,
and how the data can be used. At NIOSH, we see the need
for wider adoption of sensors and more data collection but
believe that the value added should be on pace with the
increased burden on the industrial hygienist professionals
who will be responsible for carrying out the actions. These
types of data and in general data of the future is only grow-
ing more and more complex, so having effective ways to
convey the meaning of complex data to the right people at
the correct time is extremely important. Through this col-
laboration, we at NIOSH have been able to answer many
questions about the viability of using LCDMs in an opera-
tional environment, but it also brought up many questions
that remain to be answered.
One of the major questions that remains unanswered
is how many sensors are needed to accurately capture the
variability within a given location. The answer to this ques-
tion is probably a balance between increasing the resolution