VANCOUVER 2015
1 | P a g e
9/5/2015
EXPLORING THE EFFECT OF ENERGY RECOVERY POTENTIAL
ON COMMINUTION EFFICIENCY: THE GLENCORE RAGLAN MINE CASE
*Peter Radziszewski1 and David Hewitt
2
1 Metso Minerals Canada
795 George V
Lachine, Quebec, CANADA, H8S 2R9
(*Corresponding author: [email protected])
2 Glencore Raglan Mine
120 avenue de l’Aéroport
Rouyn-Noranda, Québec J9Y 0G1 Canada
VANCOUVER 2015
2 | P a g e
9/5/2015
EXPLORING THE EFFECT OF ENERGY RECOVERY POTENTIAL
ON COMMINUTION EFFICIENCY: THE GLENCORE RAGLAN MINE CASE
ABSTRACT
Typically, comminution efficiency is cited as being less than 1%. As a result, much effort is going
into increasing that efficiency through the development of new and innovative equipment as well as
through the means to measure and quantify changes in comminution efficiency. However, what about the
other 99%? The focus of this paper is to explore the impact of the other 99% of the energy input into a
given mill on comminution efficiency. Specifically, using, mill and comminution circuit temperature data
from the Raglan Mine SAG and ball mills, this paper will first revisit a thermodynamic model of
comminution processes in order to illustrate the potential and means to energy recovery. This will be
followed by an investigation into the sources of energy loss including conduction/convection and radiation
losses as well as mass transfer. Finally, a discussion will explore the potential impact of energy recovery
on comminution efficiency.
INTRODUCTION
Efficiency, expressed as a percentage, is defined as a dimensionless ratio of the work produced
over the energy consumed or input in order to produce that work. In the case of comminution processes,
the work produced is defined by the new surface energy produced in grinding. Typically, that is cited to be
less than 1% (Lowrinson, 1974) indicating that comminution processes are very inefficient.
As a result, much effort is going into increasing that efficiency through the development of new
and innovative equipment and processes such as blast design (fig. 1a), HPGRs circuits (fig. 1b), flanged
rolls (fig. 1c) and stirred mills (fig. 1d) as well as through the means to measure and quantify changes in
comminution efficiency (Efficiency, 2015; Rowland and McIvor, 2008).
VANCOUVER 2015
3 | P a g e
9/5/2015
a) blast design - upto 15% increase in comminution efficiency
(Hart et. al.,2006.)
b) HPGR ball mill circuit - 25% increase in efficiency over a
SABC circuit (Wang et al. 2013)
c) HRC ® - flanged design produces potential 15% increase in
efficiency over non-flanged HPGR (Knorr et al., 2013)
d) VERTIMILL ® - upto 50% increase in efficiency over ball mills
for regrind applications (Merriam et al., 2015)
Figure 1 - Energy Efficiency Processes and Equipment
However, what about the other 99% of the energy input?
Typically, it is accepted that the other 99% of the energy input is transformed into heat
(Radziszewski, 2013) which indicates that comminution processes are actually very efficient in producing
heat. However, the value of heat is negligible unless, as it cannot be currently, it can be converted into a
more value added product or commodity such as electrical energy.
The focus of this paper is to explore the other 99% of the energy input into a given mill.
Specifically, using temperature, mill and circuit data for the Raglan Mine SAG and ball mills, this paper
will first revisit a thermodynamic model of comminution processes in order to illustrate the potential and
means for energy recovery. This will be followed by an investigation into the sources of energy loss
including conduction/convection, radiation and mass transfer as well as exploring avenues into possible
means to maximise energy recovery. Finally, a discussion will explore the potential impact of energy
recovery on comminution efficiency.
THE RAGLAN MINE CONCENTRATOR
Glencore’s Raglan Mine is located along the northern limit of Quebec’s territory, along the 62nd
parallel (see Figure 2). The average annual temperatures are about -10C with lows in winter below -40C
(Isolated, 2013) and average ambient temperatures underground around -15C. Raglan Mine’s property
spans 70 km consisting of 4 operating underground nickel mines, a port facility on Deception Bay,
accommodation and administrative facilities at Katinniq, an airport at Donaldson and the necessary
infrastructure to connect these installations. It is important to note that Raglan Mine is not connected to the
Hydro-Quebec grid due to its remote location. As a result, all electrical energy has been until recently
produced by diesel electric generation. In 2014, a wind energy project including an energy storage system
VANCOUVER 2015
4 | P a g e
9/5/2015
was constructed in order to capture the local wind energy potential and off-set diesel fuel costs, this system
now accounts for up to 5% of Raglan Mine’s total energy production requirements. As of June 2015, it has
produced 6.76 GWh of electrical energy, displacing 1.734 million litres of diesel fuel and reducing
greenhouse gas emissions by 4838 Tonnes.
Raglan Mine has been in operation since 1997. Originally designed for 800 000 tonnes of ore
annually, Raglan’s Concentrator now operates at the maximum 1 320 000 M tonnes annually as per The
Raglan Agreement signed with the government and local Inuit communities in 1995.
Figure 2 - Location of Glencore’s Raglan Mine (Raglan Mine, 2015)
THERMODYNAMIC COMMINUTION MODEL REVISITED
Thermodynamic model development (Radziszewski, 2013) starts with the definition of a control
volume around a piece of equipment or circuit followed by establishing all of the input and puts as
illustrated in Figure 3.
a) control volume around a SAG mill circuit b) control volume around a ball mill
Figure 3 - Defining a control volume around a comminution circuit and equipment
VANCOUVER 2015
5 | P a g e
9/5/2015
Once the control volume and all input and outputs are determined, an energy balance can be
prepared:
�̇�𝑐.𝑣. − �̇�𝑙𝑜𝑠𝑡 = �̇�𝑠𝑙(ℎ2 − ℎ1) (1)
where:
�̇�𝑐.𝑣. - work input to the control volume [kJ/s or kW],
�̇�𝑙𝑜𝑠𝑡 - heat lost to the environment [kJ/s or kW],
�̇�𝑠𝑙 - slurry mass flow rate [kg/s],
h2, h1 - slurry discharge and feed enthalpy respectively [kJ/kg].
In order to simplify the relationship, the work input consider as only the heat input into the control
volume which for ball mill grinding is assumed to be 99% of the mill input power.
Assuming constant pressure, an incompressible fluid and solid in the slurry, it is possible to
expand the energy balance to the following:
�̇�𝑐.𝑣. − �̇�𝑙𝑜𝑠𝑡 = (�̇�𝑜𝑟𝑒𝑐𝑜𝑟𝑒 + �̇�𝑤𝑎𝑡𝑒𝑟𝑐𝑤𝑎𝑡𝑒𝑟)(𝑇2 − 𝑇1) (2)
where:
core, cwater - specific heats of the ore and water [kJ/kg-K],
T2, T1 - discharge and feed slurry respective temperatures [K].
Using the thermodynamic model, the definition of Carnot efficiency along with anecdotal and
published grinding circuit data, it was demonstrated that energy recovery potential for a temperature
difference of 30C between a discharge slurry temperature and a cold water source was up to 10% of the
energy used in comminution.
Knowing the energy recovery potential of comminution systems, it is then possible to look at how
this energy can be converted to a usable form such as electricity. There are mechanical means such as a
Sterling cycle engine or a organic Rankine turbine that approach Carnot efficiency. However, such systems
require a significant infrastructure. On the other hand, there are electrical means that can be used to
accomplish this too. One such electrical system is a thermoelectric generator as illustrated in Figure 4.
With a 30C temperature difference, the expected energy conversion performance of this particular
system would be 1.27 kW/m2.
As a point of comparison, acoustic energy potential is estimated to be in the hundredths of a W/m2
(Sound, 2011), wind energy potential is estimated to be around 1 W/m2 (current wind technology potential
(Wind, 2012) divided by earth’s surface) and solar energy potential is estimated to be 300 W/m2 (average
solar energy at Earth’s surface (680 W/m2) x solar cell efficiency (Sun, 2014)).
VANCOUVER 2015
6 | P a g e
9/5/2015
a) general structure of a TEG (TEG, 2014) b) power output for eTEG HV56 thermoelectric generator
(reproduced from Nextreme, 2012b)
Figure 4 - Structure and performance of a thermoelectric generator
Although quite promising, this initial study also emphasized the need to address a number of
questions especially related to the underlying assumptions through the use of industrial data before any
industrial implementation can be envisioned.
THERMAL DATA COLLECTION
The ore processed is a nickel bearing ore coming from 4 different mines in the area. The average
ore composition is illustrated in Table 1 along with the respective mineral’s heat capacity determined from
Waples and Waples (2004). The concentrator has a SAG mill circuit feeding a ball mill circuit. The general
specifications of the SAG and ball mills can be found in Table 2.
Table 1 - Heat Capacities Table 2 - Mill Data
One control volume was defined around the SAG mill circuit and another control volume was
defined around the ball mill. Temperature measurement points were determined at the control volume
boundaries as illustrated in Figure 5. All other data was collected from the plant’s data historian.
Temperature data was collected on four different occasions using both an immersion and an
infrared thermometer (see Figure 6) and recorded manually.
Composition Heat Capacity
[%] [kJ/kg-K]
Pentlandite: 10 0.576
Infotherm (2015) :
0.445 kJ/mol-K x
0.77194 kg/mol
Chalcopyrite: 5 0.534
Pyrrhotite: 15 0.594
Gabbro: 70 0.825average of 0.65 and
1.0 kJ/kg-K
average ore 0.7509
Water 4.183
Soda Ash 1.06
CommentMineral Parameter Ball mill SAG mill
Diameter [m] 4.27 7.3
Belly length [m] 6.4 3.2
Shell thickness [m] 0.075 0.1
Liner (average)
thickness[m] 0.25 0.185
Liner material rubber steel
Thermal
conductivity[W/m-K] 0.13 43
Estimated forced
convection coef.[W/m
2-K] 200
VANCOUVER 2015
7 | P a g e
9/5/2015
Figure 5 - Raglan Mine’s SAG and ball mill circuit with temperature measurement points
a) measuring temperature at the
sump
b) temperature sensor used at the
sump
c) infrared sensor used to capture
ball mill discharge temperature
Figure 6 - Temperature measurement system and locations
DETERMINING HEAT LOSSES
Heat losses for the Raglan Mine’s ball mill and SAG mill circuit can be determined using
appropriately defined energy balances for each context.
Ball mill heat loss
In the ball mill case, the energy balance found in equation (2) is sufficient as all the ore and water
components of the slurry at the feed and discharge are at a similar temperature of T1 and T2 respectively.
Solving equation (2) for energy lost �̇�𝑙𝑜𝑠𝑡 and using the data provided in Table 3, it is possible to estimate
the average portion of energy input into ball mill grinding lost to the environment which in this case is 30%
or 681 kW.
Using mill make-up water (16C) as the cold source and the average slurry temperature (34.4C) as
the hot source, the energy capture potential efficiency is estimated to be defined by the Carnot efficiency.
For this particular case, the Carnot efficiency is 5.99% which represents some 132.7 kW.
VANCOUVER 2015
8 | P a g e
9/5/2015
Table 3 - Ball Mill Thermal Quantities
SAG mill circuit heat loss
Examining the temperature data for the SAG mill circuit found in Table 4 indicates that the three
input streams (rock, water and soda ash) all have different temperatures. In addition, for water, the input
stream is actually two input streams both at the same temperature. Further, energy input into the control
volume comes from two principle sources which are the SAG mill and the crusher. As a result, it becomes
necessary to expand further the energy balance described in equation (2).
With respect to the control volume approach, it should be noted that only the control volume
inputs and the outputs are considered and not the location in the circuit of these inputs and outputs.
Assuming a similar circuit discharge temperature (T2) for all components, the resulting expanded energy
balance for the SAG mill circuit is:
�̇�𝑐.𝑣. − �̇�𝑙𝑜𝑠𝑡 = �̇�𝑟𝑐𝑟(𝑇4 − 𝑇𝑟𝑖𝑛) + �̇�𝑤𝑐𝑤(𝑇4 − 𝑇𝑤𝑖𝑛
) + �̇�𝑠𝑎𝑐𝑠𝑎(𝑇4 − 𝑇𝑠𝑎𝑖𝑛) (3)
where the subscripts “r”, “w”, “sa” designate rock, water and soda ash respectively and the sub-
subscript “in” indicates input.
Defining �̇�𝑐.𝑣. for this SAG mill circuit requires assuming the amount of energy converted to heat
for both the SAG mill and the crusher. This can be accomplished by referring to the work of Nadolski et.
al. (2014) who defined a “Benchmark Energy Factor” (BEF) which is the “ratio of actual equipment energy
to a minimum practical energy” required to break ore to a desired target size. Using the BEF for a ball mill
circuit (1.71) and the BEF for a SAG mill circuit (2.46) which includes a pebble crusher and assuming that
the efficiency of the ball mill circuit is indeed 1%, it is possible to estimate the efficiency of the SAG mill
circuit as being (1.71/2.46) 0.7%.
Therefore, the work input (as heat) into the SAG mill circuit control volume is defined as:
�̇�𝑐.𝑣. = 0.993(�̇�𝑆𝐴𝐺 + �̇�𝐶𝑟𝑢𝑠ℎ𝑒𝑟) (4)
Due to the short contact time at the screens, the water addition there is assumed to be outside of
the control volume. As a result, the temperature of the slurry and the rock leaving the SAG mill circuit
control volume is considered to be the same.
The addition of soda ash (sodium carbonate) to the SAG mill feed is a means to control slurry pH
by neutralizing the sulphuric acid generated by the oxidation of the massive sulphide ore being ground.
The soda ash - sulphuric acid reaction is exothermic and produces CO2 gas which should bubble off. As the
mass rate of the soda ash addition is quite small as compared to the rock and water, it will not be included
in the energy balance. Further, it will be assumed that the difference between the energy generated by the
Mill
Power
slurry
flow
rate
slurry
densityore density
point 1
temperature
point 2
temperature
Slurry
Specific
Heat
Heat
Input
(Ẇcv)
Shell
Temperature
[kW] [m3/hr] [% solids] [s.g] [C] [C] [kJ/kg-K] [kW] [C]
initial data 2240 173 71 3 33.8 37.4 1.75 2218
1 1746.9 250 71 3.35 28.8 33.8 1.75 1729 29
2 1730.5 208 71 3.26 28.1 32 1.75 1713 27
3 1755.3 166.4 71 3.3 25 31.9 1.75 1738 23
4 1794.5 77.1 71 3.19 31.5 39.4 1.75 1777 27
average 1853.44 174.98 71 3.2 29.4 34.9 1.75 1834.91 26.6
˙
VANCOUVER 2015
9 | P a g e
9/5/2015
exothermic reaction and that lost through CO2 mass transfer to the atmosphere is small and will not affect
the overall energy balance.
The resulting energy balance for the SAG mill circuit control volume is reduced to:
�̇�𝑐.𝑣. − �̇�𝑙𝑜𝑠𝑡 = �̇�𝑟𝑐𝑟(𝑇4 − 𝑇𝑟𝑖𝑛) + �̇�𝑆𝐴𝐺𝑐𝑤(𝑇4 − 𝑇𝑤𝑖𝑛
) (5)
The result is that the average portion of input energy lost to the environment is determined by
solving the energy balance (equation 5 in this case) for) for energy lost �̇�𝑙𝑜𝑠𝑡 and using the data provided in
Table 4. In the Raglan Mine SAG mill circuit case, the average energy lost to the environment represents
21.6% of the energy input to the circuit at a rate of 500 kW.
Using mill make-up water (16C) as the cold source and the average slurry discharge temperature
(26.2C) as the hot source, the energy capture potential efficiency is estimated to be defined by the Carnot
efficiency. For this particular case, the Carnot efficiency is 3.41% which represents some 79 kW.
Table 4 - SAG Mill Circuit Thermal Quantities
MODELLING SOURCES OF ENERGY LOSS
There are essentially three sources of energy loss which are related to conduction/convection, radiation and
mass transfer.
Conduction / Convection Losses
Estimating the effect of conduction/convection on the slurry heat losses requires the development
of a heat transfer model. In the literature, Kapakyulu and Moys (2007 a, b) presented a rather complete
heat transfer model between the mill charge and the mill exterior. However, for the present case, a lumped
parameter model will be defined for the simplified mill cross-section illustrated in Figure 7.
Figure 7 - Simplified tumbling mill cross-section
feed rate feed temp.feed rate
SAG
feed rate
screensfeed temp. feed rate feed temp.
[kW] [kW] [t/hr] [C] [m3/hr] [m
3/hr] [C] [l/min] [C] [C] [C]
1 2264.4 63.04 180 -14.5 39.14 17.28 16.6 28.77 18.9 28.4 28
2 2464.58 63.35 177 -20 32.28 22.5 16.2 59.03 16.5 28.4 30
3 1854 60 204 -1.2 40.6 23.2 12.6 30.1 12.6 21 21
4 2451 57.6 166 0 43.6 23.8 17.7 61.9 16.7 27 28
average 2258.50 61.00 181.75 -8.93 38.91 21.70 15.78 44.95 16.18 26.2 26.88
Rock Water Soda Ash
sample no.
Crusher
Power
SAG Mill
Power
Discharge
Temp.
Shell
Temp.
VANCOUVER 2015
10 | P a g e
9/5/2015
In the context described in Figure 7, part of the heat lost is through the mill shell by conduction
and then convection (Qcc). The relationship that describes this heat transfer context is defined as (Holman,
1976):
sossll
acc
hAAk
rr
Ak
rr
TTfQ
12312
1
lnln (6)
where:
T0 – average slurry temperature [C]
T1 – inside liner surface temperature [C]
T2 – liner/shell interface temperature [C]
T3 – outside shell temperature [C]
Ta – ambient temperature [C]
Al, As, Aso – surface area of liner, liner/shell interface, shell outer [m2]
r1 – radial distance to the average inside liner surface [m]
r2 – radial distance to the liner/shell interface [m]
r3 – radial distance to the outside shell surface[m]
kl, ks – thermal conductivity of liner and shell [kW/m-K]
h – convection coefficient [kW/m2-K]
f – adjustment factor (f = 1.5)
An adjustment factor was introduced into equation (6) in order to compensate for the fact that the
mill ends are not included in the surface area over which heat is lost to the environment. In this particular
case, it is assumed that the mill ends add another 50% to the conductive/convective heat losses.
For the ball mill the conduction/convection losses are determined to be 0.84 kW which is 0.05%
of the energy input into the ball mill. For the SAG mill, the conduction/convection losses are determined to
be 48.1 kW or 2.07 % of the input energy.
As a validation of the conduction/convection model (equation 6), the mill shell temperature (T3)
was calculated with a modified form of equation (6) giving 27.7C for the ball mill and 25.3C for the SAG
mill. Comparing these calculated values with the average measured shell temperatures (see Tables 3 and 4)
on the Kelvin scale indicates that the calculated values are within 1% of the average measured shell
temperatures.
Radiation Losses
Heat transfer by radiation can be defined as follows (Holman, 1976):
soarad ATTfQ 44
3 (7)
where:
– surface emissivity [0.96 (painted surface)]
– Stefan-Boltzman constant [5.6703 x 10-11
kW/m2-K
4]
Using the mill shell temperature (T3) previous calculated, it was possible to estimate the ball mill
heat losses by radiation heat transfer as about 4.97 kW or some 0.27% of the ball mill input energy. For the
SAG mill the heat losses due to radiation are about 2.71 kW or some 0.12% of the input energy.
VANCOUVER 2015
11 | P a g e
9/5/2015
Evaporative Losses
Mass loss (�̇�𝑒𝑣𝑎𝑝) due to evaporation in a tumbling will be approximated by following
relationship developed to evaluate the evaporative mass loss of small water surface areas (Asdrubali, 2009,
Engineering, 2015, Rafferty, 1986):
3600xxAfm ssleevap (8)
where:
v1925 (9)
and:
v – velocity of air over water surface [m/s]
Asl – area of water surface [m2]
xs – humidity ratio in saturated air at the surface water temperature (Tw) [kg/kg]
x – humidity ratio in the air [kg/kg]
fe – correction factor [unity]
The humidity ratio for saturated air can be determined as follows:
wsawss pppx 621980. (10)
where:
pa – atmospheric pressure of moist air [kPa]
pws – saturation pressure of water vapour [kPa]
further:
28
723500560345077
1000 .
..
T
ep
TT
ws
(11)
The resulting energy lost by evaporation can now be determined as:
evapevapevap hmQ (12)
where: hevap – heat of vaporization of water [hevap = 2260 kJ/kg]
The slurry surface area for an overflow ball mill will be estimated by 1.25 Rmill Lmill which is the
radius of the mill plus the radius of a typical trunnion (0.25 Rmill) times the average length of the mill. As a
slurry pool is an undesired operating condition (Powell, Valery, 2006), the slurry pool over which energy is
lost by evaporation will be estimated to be 35% of a ball mill or 0.35 Rmill Lmill.
Table 5 summarised the energy lost to evaporation along with the associated results for
evaporation and those related to energy lost by conduction/convection and radiation. It should be noted that
the correction factor (fe = 1.53) was determined from the ball mill case and applied to the SAG mill case.
Table 5 - Estimated losses in Raglan Mine’s SAG and ball mills
Mill PowerConduction &
convection lossLoss by radiation
Loss by
evaporationOther losses Total lost
[kW] [kW] [kW] [kW] [kW] [kW]
Ball mill 1853.44 0.84 4.97 635.23 - 641.04
SAG mill circuit 2319 48.10 2.71 50.33 399.26 500.40
VANCOUVER 2015
12 | P a g e
9/5/2015
Based on these loss models, it is interesting to note that the majority losses in the ball mill case is
due to evaporation while in the SAG mill circuit it is relegated to other losses. This may be due to the
smaller “slurry pool” in the SAG mill as well as to higher losses around the circuit.
INVESTIGATING AVENUES TO MAXIMISE ENERGY RECOVERY POTENTIAL
Having determined the source of energy losses to the environment for these two particular cases, it
now becomes possible to expand equations (2) and (5) to include these losses as well as solve them for the
respective ball mill and SAG mill circuit discharge temperatures as illustrated:
Ball mill discharge temperature
𝑇2 =�̇�𝑐.𝑣.−(�̇�𝑐𝑐+�̇�𝑟𝑎𝑑+�̇�𝑒𝑣𝑎𝑝)
(�̇�𝑜𝑟𝑒𝑐𝑜𝑟𝑒 + �̇�𝑤𝑎𝑡𝑒𝑟𝑐𝑤𝑎𝑡𝑒𝑟) + 𝑇1 (13)
SAG mill circuit discharge temperature
𝑇4 =�̇�𝑐.𝑣.+�̇�𝑟𝑐𝑟𝑇𝑟𝑖𝑛
+�̇�𝑤𝑐𝑤𝑇𝑤𝑖𝑛−(�̇�𝑐𝑐+�̇�𝑟𝑎𝑑+�̇�𝑒𝑣𝑎𝑝)
�̇�𝑟𝑐𝑟+�̇�𝑤𝑐𝑤 (14)
This will allow the possibility to start exploring different means to increasing the energy recovery
potential as defined by Carnot efficiency. Knowing the Carnot efficiency, it becomes possible to estimate
the associated potential value.
In the case of the ball mill, leaving the mill shell un-painted (iron dark grey surface: = 0.31)
would reduce radiation losses by about two thirds to 1.48 kW. Reducing air flow through the mill by 80%
would drop evaporative heat loss by about two thirds also to 180 kW.
Similar type reduction can be expected for the SAG mill. Further, it is interesting to note that in
the case of the SAG mill conduction/convection losses could be reduced from 48.1 kW to 0.4 kW by
changing to a rubber liner.
Effect of warmer water addition
In the Raglan Mine context, it is difficult to envision a heat source with which to warm up the feed
water. However, through the sunny time of year, it is possible to consider a solar concentrator to increase
feed water temperature by 10C. The resulting effect on the ball mill slurry feed would be to increase it by
some 5C. In the case of the SAG mill circuit, 10C warmer make-up water would increase discharge
temperature by about 5C.
It should be noted that these changes in feed water temperature also affect heat losses due to
conduction/convection, radiation and mass transfer through evaporation. The net effect on discharge
temperature is captured by equations (13) and (14). In both cases, the resulting ball mill discharge and the
SAG mill circuit discharge temperature increase by about 5C. As a result, the associated increase in Carnot
efficiency (assuming a cold source temperature of 16C) is 7.18% from the baseline 6.14% for the ball mill
and 4.75% from the baseline 3.41% for the SAG mill circuit.
Effect of colder cold source
Considering that the average underground ambient temperature is around -15C, it is possible to
envision a geothermal propylene glycol (-59C freezing point) based system to provide a -15C cold source.
The resulting increase in energy recovery potential according to Carnot increases to 15.91% for the ball
mill and 13.69% for the SAG mill circuit.
VANCOUVER 2015
13 | P a g e
9/5/2015
Cumulative Effect
Graphically, the effects of all of these changes on heat loss are illustrated in Figure 8 along with
the cumulative result of all of the changes.
a) ball mill losses b) SAG mill circuit losses
Figure 8 - Effect of changes on comminution heat loss
Examining these results, one can notice that overall heat loss is in the order of 20% to over 40%
depending on the context. In the ball mill case, evaporation makes up the vast majority of the heat loss
followed by heat loss by radiation while conduction and convection represents a very small portion of the
total loss. This is undoubtedly the result of using rubber liners in the ball mill. In the SAG mill circuit case,
one more loss term is included which is “other” and represents the losses in the circuit as opposed to those
from the mill. In this case, evaporative losses are similar in terms of magnitudes to those through
conduction and convection. This is understandable as the SAG mill liners are steel having a rather high
thermal conductivity coefficient. Changing the liners to rubber or even poly-metallic liners would reduce
significantly such losses. Evaporative losses in both the ball mill and the SAG mill could potentially be
reduced by limiting air flow through these mills. In addition for the SAG mill circuit, if the “other” losses
can be reduced through various yet to be defined means, it is possible to bring SAG mill losses to a
minimum.
The effect of these changes on the slurry discharge temperature can be determined as described
previously. As a result, the associated Carnot efficiency can be determined and then used to estimate the
energy recovery potential for both the ball mill and the SAG mill circuit. In turn, the energy recovery
potential can be used to estimate the potential annual value of the energy of the slurry. Figure 9 illustrates
both the changes in the Carnot efficiency and the associated annual value of the energy in the slurry.
VANCOUVER 2015
14 | P a g e
9/5/2015
a) ball mill case b) SAG mill circuit case
Figure 9 - Carnot efficiency and annual value of energy in the slurry
In summary, if all of the measures were employed on the Raglan Mine SAG and ball mills the
resulting recovery potential would increase from the current 5.79% to 17.4% for the ball mill and from
3.32% to 16.02% for the SAG mill circuit. This represents an annual energy recovery potential (95%
equipment availability, $0.40/kWhr) for the current baseline case just under $0.7 million per year and $2.5
million per year for the cumulative improvement case.
DISCUSSION
In the paper Radziszewski (2013), five assumption were made which include slurry heat capacity,
portion of input energy converted to heat, portion of energy captured in slurry, the influence of water
addition to the grinding circuit and adiabatic/sealed operation conditions.
With respect to slurry heat capacity, in the Raglan Mine case no assumption was made as the
general ore composition was used to determine the slurry heat capacity.
It was assumed that 99% of the input energy is converted to heat. This was somewhat modified for
the SAG mill circuit using the “Benchmark Energy Factor” (Nadolski et.al., 2014). However, the basis of
this assumption is that all energy loss paths eventually lead to generating heat.
No assumption was made on how much energy was captured in the slurry. This was calculated
directly from the model with the use of the feed and discharge temperature measurements.
No assumptions were made on the influence of water addition at the cyclones and circuit in
general. All water inputs to the circuit were taken into account. However; it was assumed that energy value
of soda ash addition was equal to the energy value loss of carbon dioxide gas.
No assumptions were made on adiabatic and sealed conditions for the mills. Instead models for
heat and mass transferred were developed which required the establishment of a few more assumptions and
a couple of factors.
Namely, for conduction and convection as lumped parameter model was used to estimate heat lost
through the mill shell. Comparing the average measured shell temperature with the calculated estimate
indicates that this model is within 1% of the measured value. However, it should be noted that the surface
over which conduction convection occurs is assumed to be 50% greater than just the circumferential mill
shell area.
VANCOUVER 2015
15 | P a g e
9/5/2015
With respect to radiation loss, the same 50% greater surface area was used to estimate the surface
over which radiation loss occurred. No validation of the resulting radiation loss was made.
With respect to mass transfer, the model used assumes that evaporative heat loss is a function of
“pool” area, the speed of air over that surface and the relative humidity around the mill. In this case, “pool”
area was estimated for both the ball and SAG mills as a function of the mill dimensions while air speed
was estimated and relative humidity was taken from local humidity records. Further, a correction factor
was used.
To summarise, the results presented here are the product of a reduced number of assumptions due
to the use of industrial data. However, in so doing, a few more assumptions were made related to heat
transfer and mass loss models used to estimate sources of energy loss.
Efficiency
It was described that energy efficiency, expressed as a percentage, is defined as a dimensionless
ratio of the work produced over the energy consumed or input in order to produce that work. In the case of
comminution processes, the work produced is defined by the new surface energy produced in grinding and
is typically considered to be less than 1% of the input energy. Therefore comminution efficiency (comm) is
defined as the efficiency of producing new surface energy (nsa):
𝜂𝑐𝑜𝑚𝑚 = 𝜂𝑛𝑠𝑎 (15)
However, if comminution were to be seen as a process that produces two usable products (ground
ore and energy), it may be possible to redefine comminution efficiency as:
𝜂𝑐𝑜𝑚𝑚 = 𝜂𝑛𝑠𝑎 + 𝜂𝐶𝑎𝑟𝑛𝑜𝑡 (16)
where:
Carnot – Carnot efficiency defining potential energy recovery from the mill slurry.
If this were the case, the potential efficiency of the Raglan Mine’s SAG mill circuit would be
about 17% while for the ball mill the efficiency would be about 18%.
On the other hand, it is important to include the efficiency of a particular technology (tech) to
meet the efficiency defined by Carnot which suggests that comminution efficiency should actually be
defined as:
𝜂𝑐𝑜𝑚𝑚 = 𝜂𝑛𝑠𝑎 + 𝜂𝑡𝑒𝑐ℎ 𝜂𝐶𝑎𝑟𝑛𝑜𝑡 (17)
In either case, the potential to recover comminution energy should be included in the definition of
comminution efficiency in order to underline that comminution processes have potentially two products.
CONCLUSIONS
An initial study of the comminution energy recovery potential indicated that there is potential
value that can be capture from a mills slurry through a few possible technological means. However, a
number of assumptions were used in that development that could only be addressed using industrial data.
In examining these assumptions in the case of Glencore Raglan Mine’s SAG and ball mill circuits,
the thermodynamic analysis indicates that the current energy recovery potential is in the order of 190 kW
(ball mill and SAG mill circuit) or 4.6% of the total comminution grinding energy. Exploring possible
means of improving this recovery potential indicates that it can potentially be increased to 684 kW (ball
mill and SAG mill circuit) or 16.6%.
VANCOUVER 2015
16 | P a g e
9/5/2015
In working to estimate Raglan Mine’s comminution energy recovery potential, it was possible to
make a number of observations:
Average energy lost from the Glencore Raglan Mine ball mill is 34.6%
Average energy lost from the Glencore Raglan Mine SAG mill circuit is 21.6%
Conduction/convection represents a small amount of energy lost for the two cases examined.
Conduction/convection losses can be reduced by the use of rubber. Although it was not estimated, one
could expect similar reductions with polymetallic liners.
Possible means to reduce radiation losses are related to mill paint can coatings.
Mass transfer due to evaporation was found to be the major cause of heat loss to the environment.
Possible means to reduce mass transfer heat loss is through sealing the comminution circuit.
As a result, it is possible to consider that energy is a potential second product of comminution.
Accepting this observation means that comminution efficiency can be redefined as the sum of new surface
energy produced and energy recovered divided by energy input.
One final concluding remark:
Lord Kelvin stated: “If you cannot measure it, you cannot improve it”.
Our response: “So measure it!”
ACKNOWLEDGEMENTS
The authors would like to thank Glencore’s Raglan Mine and Metso for granting permission to
publish and present this work and associated data and results.
REFERENCES
Asdrubali, F. (2009). A scale model to evaluate water evaporation from indoor swimming pools, Energy
and Buildings, 41, 311-319.
Efficiency (2015). Industrial Comminution Efficiency, Global Mining Standards and Guidelines Group,
retrieved from http://www.globalminingstandards.org/working-groups/2135-2/.
Engineering (2015). Evaporation from water surfaces, retrieved from http://www.engineeringtoolbox.com
/evaporation-water-surface-d_690.html
Hart, S., Valery, W., Clements, B., Reed, M., Song, M., Dunne, R. (2001). Optimisation of the Cadia Hill
SAG mill circuit, SAG2001, Vancouver, I-11 – I-30.
Holman, J.P. (1976). Heat Tranfer, McGraw-Hill, Toronto, 530 pages.
Kapakyulu, E., Moys, M.H. (2007a). Modeling of energy loss to the environment from a grinding mill:
Part II Motivation, literature survey and pilot plant measurements, Minerals Engineering 20, 646-
652.
Kapakyulu, E., Moys, M.H. (2007b). Modeling of energy loss to the environment from a grinding mill:
Part II Modelling the overall heat transfer coefficient, Minerals Engineering 20, 653-661.
Knorr, B., Hermann, V., Whalen, D. (2013). HRC: taking HPGR efficiency to the next level by reducing
edge effect, Procemin 2013, pp19.
Lowrinson, G.C., (1974). Crushing and Grinding, Butterworths, London.
VANCOUVER 2015
17 | P a g e
9/5/2015
Merriam, K., Rotzinger, R., DeHart, I., Radziszewski, P., McDonnell, J. (2015). Assessing Vertimill
ultrafine grinding performance – the Gibraltar Mine case, CMP Ottawa, 66-72.
Nextreme (2012, July 31). eTEG HV56 Power Generator Data Sheet, retrieved from
http://www.nextreme.com/media/techspecs/Nextreme_eTEG_HV56_Data_Sheet.pdf
Powell, M., Valery, W. (2006). Slurry pooling and transport issues in SAG mills, SAG2006, Vancouver,
I-133 – I-152.
Radziszewski, P. (2013). Energy recovery potential in comminution processes, Minerals Engineering, 46-
47, 83-88.
Rafferty, K. (1986). Pond heat loss, Geo-Heat Center Quarterly Bulletin Vol. 9, No.4, 15-17.
Rowland, C.A., McIvor, R.E. (2009). The Bond standard for comminution efficiency, SME Annual
Meeting, 08-016,
Sound (2011). Can sound be converted to useful energy?, Ask an engineer – MIT School of Engineering,
retrieved from http://engineering.mit.edu/ask/can-sound-be-converted-useful-energy
Sun (2014). French-German collaborators claim solar cell efficiency world record, EE Times Europe,
retrieved from http://www.electronics-eetimes.com/en/french-german-collaborators-claim-solar-
cell-efficiency-world-record.html?cmp_id=7&news_id=222923159
TEG (2014). Thermoelectric energy generation takes flight for aircraft and spacecraft monitoring, Digi-
Key Article library, retrieved from http://www.digikey.com/en/articles/techzone/2014/apr
/thermoelectric-energy-generation-takes-flight-for-aircraft-and-spacecraft-monitoring
Wang, C., Nadolski, S., Mejia, O., Drozdiak, J., Klein, B. (2013). Energy and cost comparison of HPGR
based circuits with the SABC circuit installed at the Huckleberry mine, CMP Ottawa, 121-135.
Waples, D., Waples, J. (2004). A Review and Evaluation of Specific Heat Capacities of Rocks, Minerals,
and Subsurface Fluids. Part 1: Minerals and Nonporous Rocks, Natural Resources Research, Vol.
13, No. 2, 97-122.
Wind (2012). The Earth has enough wind energy potential to power all of civilisation, Business Insider,
retrieved from http://www.businessinsider.com/the-earth-has-enough-wind-energy-potential-to-
power-all-of-civilization-2012-9