4
a better forward outlook is required if the goal is to ensure
availability with the least amount of inventory. The full data
set implies that forward looking forecasting should allow
for up to ±60% monthly demand variability and quarterly
about ±40%.
INVENTORY MODELING
To develop better forecasting, the data was used to model
inventory levels. Figure 6 illustrates the monthly inventory
levels for three representative low, medium and high vol-
ume bits. It appears like a “random walk” as consumption is
sometimes above and other times below average. The con-
sumption rate is from actual data whereas the bit inventory
is modeled, starting with two months of safety stock and
replenished at the monthly average, following the old rule
of thumb. The purpose of this model is to show how often
manual intervention is required to prevent outages. A nega-
tive inventory is obviously not possible but is used to count
the occurrences of emergency sourcing.
In Figure 6, one of 64 data points is negative, giving
a 1.6% outage rate. A different safety stock would simply
shifts the curves up or down, affecting the percentage of
outages. Although just examples, the curves also illustrate
how lower volume bits have higher variability. Following
the old rule of thumb, the high volume bit appears over-
stocked while the low and medium volume bits appear
understocked. In this illustration, the medium volume bit
stocked out in one month, while the low volume bit almost
stocked out over several months.
Based on all 60 SKUs across the thirteen mines, differ-
ent safety stock levels were experimented with to evaluate
the effects both on outages and total inventory volume. The
base case is to have two months of safety stock for all bits –
low, medium, and high volume. As modeled, the base case
results in outages 3.5% of the time. The logic is the same
as in Figure 6, except with 60 curves. The corresponding
inventory level is normalized to 100%. Not surprisingly,
running the inventory leaner, basically shifting the curves
down, increases the outages, while excess inventory reduces
the outages. One month safety stock results in 10.0% out-
ages, while three months safety stock reduces it to 1.3%.
The balanced case simply cuts the safety stock of high
volume bits from two to one month while increasing the
safety stock of low volume bits from two to three months.
Medium volume bits are left with two months of safety
stock. The result is 14% less inventory while reducing the
outages by 30% (from 3.5% to 2.5%). These percentages
are directionally meaningful but should not be taken as
real-world outages in any particular mine.
It can be argued that the high volume bits are most
important to mine operations. It therefore may seem coun-
ter-intuitive to shift safety stock to the low volume bits.
Nevertheless, this is what the data tells us if we want to
balance availability and inventory. The new rule of thumb
then becomes:
• Low volume – 3 months’ safety stock
• Medium volume – 2 months’ safety stock
• High volume – 1 month safety stock
Although this “rule” is directionally supported by large
amounts of data, it should be adapted for each site, e.g.:
• The general variability of a mine may be higher or
lower
• The cut-offs of 50 and 200 bits per month may be
changed
• Longhole bits may be reclassified for less safety stock
• Development bits may be reclassified for more safety
stock
Figure 6. Model of inventory levels based on select bits and
actual consumption data
Table 2. Impact on outages and inventory with selective
safety stock levels for high, medium, and low volume bits
a better forward outlook is required if the goal is to ensure
availability with the least amount of inventory. The full data
set implies that forward looking forecasting should allow
for up to ±60% monthly demand variability and quarterly
about ±40%.
INVENTORY MODELING
To develop better forecasting, the data was used to model
inventory levels. Figure 6 illustrates the monthly inventory
levels for three representative low, medium and high vol-
ume bits. It appears like a “random walk” as consumption is
sometimes above and other times below average. The con-
sumption rate is from actual data whereas the bit inventory
is modeled, starting with two months of safety stock and
replenished at the monthly average, following the old rule
of thumb. The purpose of this model is to show how often
manual intervention is required to prevent outages. A nega-
tive inventory is obviously not possible but is used to count
the occurrences of emergency sourcing.
In Figure 6, one of 64 data points is negative, giving
a 1.6% outage rate. A different safety stock would simply
shifts the curves up or down, affecting the percentage of
outages. Although just examples, the curves also illustrate
how lower volume bits have higher variability. Following
the old rule of thumb, the high volume bit appears over-
stocked while the low and medium volume bits appear
understocked. In this illustration, the medium volume bit
stocked out in one month, while the low volume bit almost
stocked out over several months.
Based on all 60 SKUs across the thirteen mines, differ-
ent safety stock levels were experimented with to evaluate
the effects both on outages and total inventory volume. The
base case is to have two months of safety stock for all bits –
low, medium, and high volume. As modeled, the base case
results in outages 3.5% of the time. The logic is the same
as in Figure 6, except with 60 curves. The corresponding
inventory level is normalized to 100%. Not surprisingly,
running the inventory leaner, basically shifting the curves
down, increases the outages, while excess inventory reduces
the outages. One month safety stock results in 10.0% out-
ages, while three months safety stock reduces it to 1.3%.
The balanced case simply cuts the safety stock of high
volume bits from two to one month while increasing the
safety stock of low volume bits from two to three months.
Medium volume bits are left with two months of safety
stock. The result is 14% less inventory while reducing the
outages by 30% (from 3.5% to 2.5%). These percentages
are directionally meaningful but should not be taken as
real-world outages in any particular mine.
It can be argued that the high volume bits are most
important to mine operations. It therefore may seem coun-
ter-intuitive to shift safety stock to the low volume bits.
Nevertheless, this is what the data tells us if we want to
balance availability and inventory. The new rule of thumb
then becomes:
• Low volume – 3 months’ safety stock
• Medium volume – 2 months’ safety stock
• High volume – 1 month safety stock
Although this “rule” is directionally supported by large
amounts of data, it should be adapted for each site, e.g.:
• The general variability of a mine may be higher or
lower
• The cut-offs of 50 and 200 bits per month may be
changed
• Longhole bits may be reclassified for less safety stock
• Development bits may be reclassified for more safety
stock
Figure 6. Model of inventory levels based on select bits and
actual consumption data
Table 2. Impact on outages and inventory with selective
safety stock levels for high, medium, and low volume bits