Technical Papers

Optimization of flotation
plant performance using
micro-price analysis
by G.H. Luttrell and A. Noble

Abstract ■ Flotation circuits play a key role in determining the profitability of base metal
concentrators. Many companies have recognized this and have opted to install complex control
systems that are designed to optimize performance based on one or more objective functions,
such as circuit yield, grade or other measures of efficiency or cost. Unfortunately, field studies
indicate that this generalized approach to optimization often does not provide a true financial
optimum and, in fact, can often lead to worse performance in terms of overall profitability.
Moreover, this blind approach to optimization provides little or no insight that can be used by
plant operators on a day-to-day basis to improve concentrator performance. A more effective
approach is to use micro-price analysis to identify and formulate optimization protocols for the
concentrator. The micro-price concept assigns unit values to each particle (or groups of similar
particles) passing through production units. Optimal performance occurs whenever all particles
with positive value are recovered and passed to market, and all negative value particles are
rejected as waste. This paper describes the micro-price concept as applied to froth flotation
circuits in base metal concentrators. A hypothetical case study is provided to demonstrate
the economic gains that can be realized through the application of this flexible optimization
concept.

Mining Engineering, 2017, Vol. 69, No. 2, pp. 34-40.

Official publication of the Society for Mining, Metallurgy & Exploration Inc.
https://doi.org/10.19150/me.7264

Introduction erational parameters in response to

Froth flotation circuits are widely disturbances in ore characteristics and
G.H. Luttrell, member SME, is the used throughout the mineral process- plant flows. These early online con-
E. Morgan Massey Professor in the ing industry for the concentration of trols typically involved feedback loops
Department of Mining and Minerals fine particles. It had been estimated designed to stabilize cell performance
Engineering, Virginia Polytechnic that nearly 2 Gt (2.2 billion st) of ores through the automatic manipulation of
Institute and State University, are treated globally each year using process variables such as pulp levels,
Blacksburg, VA, USA, and this versatile separation process (Fuer- gas rates and chemical dosages. The
A. Noble, member SME, is assistant stenau, 1999). Due to the high feed ton- set-points for these controllers were
professor in the Benjamin M. nage, a large economic incentive exists often set by plant operators based on
Statler College of Engineering and for flotation plant operators to opti- one or more operator objectives, such
Mineral Resources, West Virginia mize the performance of this critical as maximizing throughput, recovery
University, Morgantown, WV, USA, processing circuit. Prior to the 1950s, or grade. More recently, these regula-
email: [email protected]. the optimization of flotation circuits tory controllers have been interfaced
Paper number TP-16-017. primarily involved the manual adjust- with high-level data display and man-
Original manuscript submitted ment of operating variables based on agement systems that provide opera-
March 2016. Revised manuscript visual observation and sporadic sam- tors with the ability to better optimize
accepted for publication September pling (Lynch, Harbort and Nelson, flotation plant performance based on
2016. Discussion of this 2010). During the next several decades, mathematical models and rule-based
peer-reviewed and approved however, the advent of improved on- algorithms (Greet, 2010).
paper is invited and must be line instrumentation coupled with While tremendous strides have
submitted to SME Publications the rapid development of high-speed been made during the past several
by May 31, 2017. microprocessors allowed the indus- decades, even the most modern flota-
try to develop more advanced control tion plants are still faced with the long-
systems that could quickly adjust op- standing question: “What are we trying
34 february 2017    Mınıng engıneerıng www.miningengineeringmagazine.com
 Technical Papers
to optimize?” From an operator perspective, some common point of micro-pricing as applied to a flotation concentrator.
responses include “Maximize tonnage of saleable concen-
trate” or “Maximize metals recovery at a customer-specified What is our inventory of products? The trade-off between
market grade.” These replies may or may not reflect those concentrate yield and quality is well recognized in mineral
of management, who often focus on economic goals such as processing. For example, Table 1 shows the cumulative mass
“Maximize profits for each operational unit” or “Maximize and grade of copper (Cu) recovered from a bank of nine
rate of return for all corporate assets.” Unfortunately, these flotation tanks in series. The cumulative yield of concentrate
high-level objective statements do not always translate well increases as the feed slurry passes down the bank. Likewise,
into practice due to the disconnect between plant operat- the cumulative concentrate grade deteriorates down the
ing protocols and product value. Attempts have been made bank as larger numbers of incompletely liberated particles
by industry to overcome this issue using various types of enter the froth product. The middlings population includes
optimization routines such as linear programming or other poorly liberated particles and/or groupings of entrained
optimization engines. While this approach does have some gangue intermixed with floatable minerals that report as a
merit, these blind routines require considerable input data unit to the froth. In practice, the incremental grade of the
and interpretational expertise and provide little or no fun- particles recovered, or rejected, by a flotation bank can be
damental insight that can be used to drive improvements in obtained from the cell-to-cell yield-grade curve. The incre-
concentrator performance. mental grade is calculated from a mass balance involving the
total mass of solids and the total mass of a specific contained
Micro-price optimization component before and after an incremental change in per-
An alternative approach to plant improvement called formance. For example, consider a hypothetical separation
micro-price optimization has been developed to translate curve illustrated in Fig. 1. In this case, the incremental grade,
the economic objectives of a mineral producer down to the ΔG, of particles recovered by the fifth cell, and rejected by
operational level. This technique assigns monetary value to the fourth cell, can be estimated from:
each particle, or groups of particles, passing through produc-
(1)
tion units based on market-driven cost information. Opti-
mization occurs whenever (1) the maximum number of par-
ticles of positive value are generated through liberation, re-
(2)
covered and passed to market and (2) the minimum number
of particles with negative value are recovered and shipped. in which Y and G are the cumulative mass yield and grade.
This concept, which was originally conceived in the coal in- According to Table 1, the cumulative grade decreased from
dustry (Luttrell, 2013; Luttrell, Noble and Stanley, 2014; Lut- 11.08 percent to 10.79 percent Cu as the cumulative yield in-
trell and Noble 2015), is now being applied to a variety of creased from 6.57 percent to 7.47 percent. Based on Eq. (2),
mineral applications including the optimization of flotation the incremental grade of solids rejected by the fourth cell
banks (Luttrell and Noble, 2013). and recovered by the fifth cell is 8.74 percent Cu. By the time
The micro-pricing approach treats mineral producers no the ninth cell is reached, the incremental grade is reduced
differently than any other manufacturer of consumer goods. to 6.16 percent Cu for a final froth concentrate containing a
In the mineral industry, raw materials (ores) are converted cumulative content of 9.95 percent Cu. (Note that to be pre-
(manufactured) into valuable products (mineral particles) cise, the instantaneous incremental grade at any point along
through the process of grinding, classification, concentra-
tion and dewatering. A pure mineral particle produced by
this process would have the highest value in the product Table 1
line, while composite (middling) particles would have lower Cumulative and incremental yield-grade data for a copper
values depending on the proportions of intermixed valuable rougher flotation bank.
mineral and gangue in each recovered increment. Wastes
generated during the process consist of gangue particles that Cell Cum. Cum. Incr. Incr.
contain insufficient mineralization to warrant recovery. If no. mass % % Cu mass % % Cu
inadvertently misplaced into the product, these gangue par-
ticles would incur a cost penalty that would lower the mar- 0 0.00 − − −
gin realized by the mineral producer. As such, micro-pricing 1 2.28 12.11 0.00 12.11
dictates that the recovery and shipment of negative-value
particles must be minimized to realize optimal economic 2 4.06 11.73 2.28 11.25
performance. The micro-pricing concept forces mineral pro- 3 5.45 11.39 1.77 10.39
ducers to address several key questions about their business 4 6.57 11.08 1.40 9.55
(Luttrell, Noble and Stanley, 2014). Often, the three most
important of these include: 5 7.47 10.79 1.11 8.74
6 8.20 10.54 0.90 7.99
• What is our inventory of products?
7 8.81 10.32 0.73 7.31
• What is the worth of our products?
• How should our inventory be managed? 8 9.31 10.12 0.61 6.69
9 9.73 9.95 0.50 6.16
The following addresses these questions from the view-

www.miningengineeringmagazine.com Mınıng engıneerıng    february 2017 35

Technical Papers

Table 2
Micro-price calculation for particles containing various amounts of chalcopyrite and siliceous gangue.

Particle type
(Gold = CuFeS2)
(Gray = SiO2)

Percent CuFeS2 100 75 50 37.5 25 12.5 0

Percent Cu 34.6 25.95 17.3 12.975 8.65 4.325 0
Base payment* $1,309 $982 $655 $491 $327 $164 $0
Payment reduction −$38 −$38 −$38 −$38 −$38 −$38 −$38
Treatment charge −$90 −$90 −$90 −$90 −$90 −$90 −$90
Refining charge −$74 −$55 −$36 −$26 −$17 −$7 +$2
Freight/Misc. charge −$40 −$40 −$40 −$40 −$40 −$40 −$40
Total $1,068 $759 $451 $297 $143 −$11 −$166
*Sales contract: Base payment = $3,784/t Cu per LME after deduct from agreed copper assay 1 unit, treatment charge = $50/t concentrate,
refining charge = $220/t of payable copper, freight/handling and other sales-related charges of $40/t of concentrate.

the flotation bank should be determined from the yield- (Wills and Munn, 2006). The contract requires a deduction
grade curve as ΔY→0.) The tabular values of incremental from agreed copper assay of 1 unit with the remainder paid
mass and grade shown in Table 1 represent the “inventory” at the London Metal Exchange (LME) price, which was es-
of products that, if recovered, are available for sale by the timated to be $3,784/t ($3,432/st) of copper at the time this
mineral producer. article was prepared. The contract also includes treatment
and refining (T/R) charges based on delivered tonnages of
What is the worth of our products? Once the invento- concentrate at $50 per dry ton and payable copper at $220/t
ry of products is known, the next step is to determine how ($200/st), respectively. An additional charge of $40/t ($36/st)
much each saleable product is worth. Under micro-pricing, of concentrate is also included to cover freight and handling
this is achieved by assigning prices to each particle passing and other sales-related costs. Other payments and charges
through the concentrator. As with any commercial business, may also be applied in a detailed analysis to account for
this calculation uses data from customer contracts and is in- other payable metals or customer costs. For example, value
dependent of the inventory of available products. For exam- terms could be added to account for payable components
ple, Table 2 shows a simplified smelter contract for a copper such as precious metals or other desirable sulfides present
concentrator. The contract offered by the smelter is prorated in the ore. Penalty terms could also be included to account
based on the measured copper content of the concentrate for negative charges associated with unwanted components,
such as pyrite or other gangue minerals, which may inad-
vertently be passed into a recovered increment. Only two
Figure 1 components, a valuable mineral and unwanted gangue, were
included in the current analysis to simplify the description
Estimation of incremental grade from yield-grade data for a of the micro-pricing approach. The use of actual field data
cell-to-cell flotation bank. in the micro-pricing calculations was not possible since min-
eral producers do not openly share corporate costing and
customer pricing data with external entities. However, the
terms required in the micro-pricing analysis are typically
available, albeit not always in the most usable format, to the
management and metallurgists working at mineral produc-
tion facilities.
Based on the contract described above, the net value of
each unit of tonnage passing through the production pro-
cess can be calculated as shown in Table 2. This information
indicates that pure particles of chalcopyrite, with 34.6 per-
cent Cu, would carry a price tag of $1,068/t ($969/st) after
applying the appropriate payments and charge deductions.
Likewise, middlings containing 50 percent chalcopyrite, with
17.3 percent Cu and 50 percent siliceous gangue would carry
a lower price tag of only $451/t ($409/st). A further reduc-

36 february 2017    Mınıng engıneerıng www.miningengineeringmagazine.com

 Technical Papers
tion in mineralization below 12.5 percent chalcopyrite, with Figure 2
4.33 percent Cu, yields a product with a net negative value of
−$11/t ($10/st). The largest negative value is associated with Linear correlation between micro-prices and incremental
pure gangue, with 0 percent Cu, representing a cost of $166/t copper grades (yellow = CuFeS2).
($151/st) of gangue shipped to the smelter. In other words,
$166/t ($151/st) is lost for each ton of this material inadver-
tently misplaced into the final concentrate. These price tags
represent the value of each particle, or class of particles, that
can be offered to the market by the mineral producer. As
such, the inherent value of each particle remains the same
regardless of whether they are recovered or rejected.

How should our inventory be managed? The values from

Table 2 are replotted in Fig. 2 for each unit of saleable prod-
uct in the concentrator feed inventory. As shown, the copper
grade and dollar-per-ton values are linearly correlated for
this simplified example with a regression line given by:
Value {$/t} = 35.64 (percent Cu) – 165.60 (3)
Note that detailed calculations for real-world cases may
or may not be linear due to site-specific cost factors such as
variable sales-related costs. Nevertheless, for this example,
the critical assay at which the incremental tonnage goes Figure 3
negative is ∼4.6 percent Cu. This particular composition rep- Hypothetical flotation circuit consisting of nine roughers, six
resents the zero-value particle present in the flotation feed. scavengers, seven cleaners and one regrind mill (Lynch et al.,
As such, the optimal operating point for the concentrator 1981).
would be to recover all particles in the feed containing more
than 4.6 percent Cu and to reject all particles containing less
than 4.6 percent Cu. This optimal incremental grade is the
economic driver that dictates how the flotation plant should
be controlled. Interestingly, a rule-of-thumb often quoted in
the minerals industry is that all particles with an incremental
grade higher than the feed grade should be recovered since
at this point it would be better to add feed to the concentrate
than to recover lower quality middlings. However, as shown
in Fig. 2, the cut-off grade may be much lower or higher than
the feed grade depending on market conditions and down-
stream charges. hours in a year = $1,377,800. Therefore, rather than focus-
Up to this point, no mention has been made regarding ing on grouped lots/piles/shipments of concentrate, micro-
the populations of products that are available to sale un- pricing forces plant operators to place value on each particle
der the micro-pricing approach. The populations of mineral moving through the concentrator.
particles with different compositions can be altered by the
mineral producer by liberating valuable minerals through Micro-price case study
size reduction. In light of the micro-pricing structure, the The best way to demonstrate the potential impacts of
liberation should be performed in a manner that produces micro-price optimization is through a case study. The hypo-
a population of positive value particles that when com- thetical data used in this study were generated by numeri-
bined have a cumulative grade that is capable of meeting cally simulating the flotation circuit shown in Fig. 3. The
the smelter sales contract. If the liberation is inadequate to simulation data were used in place of actual plant perfor-
attain this constraint, then the concentrator will be forced to mance data to protect the confidentiality of groups that have
reject positive value particles until the contract specification been involved in detailed micro-price analyses. The simula-
is met. This undesirable situation results in a less favorable tions were performed using the kinetic modeling approach
economic situation for the concentrator since payable prod- described by Lynch et al. (1981) in their landmark book. The
ucts have been lost to waste. Likewise, if the flotation feed is simulation module was developed in a spreadsheet-based
overliberated, then grinding costs will have been expended format to provide an easy interface for user input and tabu-
unnecessarily to make room within the shipped concentrate lar or graphical output. The mass percentages, mineral con-
for negative value gangue particles that should not be passed tent and rate constants for the fast-, slow- and non-floating
to the smelter. For the case shown, each ton per hour of si- components are listed in Table 3. The hypothetical values
liceous gangue that is inadvertently misplaced to the final used in the simulations closely mimic the values published
froth concentrate will incur a cost of $166 per hour to the in the Lynch et al. (1981) book. The first set of values were
mineral producer. Over the run of a year, this would equate selected to represent a good feed ore, while the second set
to a loss of nearly $1.4 million per year, as $166/h × 8,300 of values were de-rated to represent a poor feed ore. The
www.miningengineeringmagazine.com Mınıng engıneerıng    february 2017 37
Technical Papers

Figure 4
Cumulative and incremental yield-grade curves for (a) good feed ore conditions and (b) poor feed ore conditions.
(a) (b)

Table 3 production calculations were performed for a hypothetical

100-kt/d (110,000-stpd) concentrator cycling between these
Simulation input data for (a) good and (b) poor feed ore two types of ore. Product qualities were controlled by vary-
flotation characteristics. ing the magnitudes of the stated rate constants up or down
as required to achieve the desired grade of combined total
Fast- Slow- Non- Total concentrate.
floating floating floating Figures 4a-b provide the cumulative yield and grade
(a) Good ore curves for the range of good and poor ores evaluated in the
case study. The first set of calculations was performed such
Mass (%) 2.0 3.0 95.0 100.0 that the flotation bank maintained a constant cumulative
CuFeS2 (%) 100.0 10.00 0.00 2.30 grade of 27.5 percent Cu for the froth concentrate regardless
Cu (%) 34.62 3.46 0.00 0.80 of feed type. Based on these separation plots, a 27.5 percent
copper concentrate would result in a mass yield of 2.60 per-
Rate (1/min) 0.20 0.03 0.005 − cent when processing the good feed and 1.2 percent when
(b) Poor ore processing the poor feed. The hypothetical 100-kt/d (110,000-
Mass (%) 1.0 4.0 95.0 100.0 stpd) concentrator cycling between these types of ore would
produce 637,468 t/a (703,000 stpy) of mineral concentrate,
CuFeS2 (%) 90.0 9.00 0.00 1.26 with 27.5 percent Cu, valued at $814/t, thereby generating
Cu (%) 31.16 3.12 0.00 0.44 revenue of $519.2 million per year. These production calcu-
lations were then repeated for a range of target concentrate
Rate (1/min) 0.20 0.03 0.005 − grades ranging from 24.5 percent to 30.5 percent Cu. The re-
sults, which are provided in Table 4, provide some interesting

Table 4
Production calculations for simulations constrained to a constant cumulative concentrate grade.

Total (100%) 50% good ore 50% poor ore

Conc. Conc. Value Value Cum. Incr. Incr. Cum. Incr. Incr.
(% Cu) (t/a) ($/t) ($million/a) (% Cu) (% Cu) ($/t) (% Cu) (% Cu) ($/t)
24.5 731,968 $708 $517.9 24.5 2.92 −$62 24.5 4.65 $0
25.5 698,496 $743 $519.1 25.5 3.05 −$57 25.5 5.35 $25
26.5 667,155 $779 $519.6 26.5 3.33 −$47 26.5 6.74 $75
27.5 637,468 $814 $519.2 27.5 3.74 −$32 27.5 9.54 $174
28.5 608,452 $850 $517.2 28.5 4.21 −$16 28.5 12.66 $286
29.5 577,953 $886 $511.9 29.5 5.05 $14 29.5 18.08 $479
30.5 536,628 $921 $494.4 30.5 6.55 $68 30.5 26.60 $782

38 february 2017    Mınıng engıneerıng www.miningengineeringmagazine.com

 Technical Papers
insight when viewed from an incremental grade and micro- Figure 5
price perspective. For the 27.5 percent Cu production run, a
close examination of the data indicates that the incremental Total annual value resulting from operation over a range of
qualities of the last particles added to the combined concen- marketed copper grades.
trate were very different at 3.74 percent Cu and 9.54 percent
Cu, respectively, for the blends of good and poor feeds. As
a result, the micro-price of the last particles recovered for
the good feed was slightly negative at −$32/t ($29/st), while
those recovered from the poor feed were large positive at
+$174/t ($158/st). The recovery of negative value particles
for the good ore and loss of positive value particles for the
poor ore show that this operating point is not optimal from
an economic perspective.
The second group of simulations was carried out for the
case in which the cumulative concentrate grade was allowed
to independently vary for each type of feed. The simulations
were constrained so that a combined final concentrate grade
of 27.5 percent Cu was achieved while holding the incre-
mental qualities identical for both good and poor feeds. The
calculations were then repeated for other combined target
grades ranging from 24.5 percent to 30.5 percent Cu. As
shown in Table 5, the 27.5 percent Cu production run pro- a target grade of 29.5 percent Cu.
duced 640,420 t/a (706,000 stpy) of froth concentrate, with The data in Fig. 5 also shows that maximum value is gen-
27.5 percent Cu, worth $814/t ($738/st) for the same 100- erated when targeting a cumulative concentrate grade of 27.5
kt/d (110,000-stpd) concentrator. For this case, the annual percent. This finding was expected in light of the incremental
revenue for operation at a constant incremental grade was micro-prices discussed previously, which indicated that the
$521.6 million, compared with $519.2 million when operat- zero-value particle for this particular ore occurs at an incre-
ing at a constant cumulative grade. The net gain of more mental grade of 4.65 percent Cu. Table 5 shows that opera-
than $2.4 million is due to the selection of optimal operating tion at a cumulative grade of 27.5 percent Cu corresponds to
points that are designed to hold identical incremental grades this critical copper content and produces a zero-value incre-
regardless of the type of feed ore processed. For reference, ment for both the good and poor feeds. Operation below a
the estimated revenue generated for each scenario is plot- cumulative grade of 27.5 percent Cu, with incremental grade
ted in Figure 5 over the range of target grades used in the = 4.65 percent Cu, would place negative value particles into
simulations. This plot shows that (1) operation at a constant the product, while operation above 27.5 percent Cu would
incremental grade always produces greater value than the result in the loss of positive value particles to waste.
standard industry practice of operating flotation banks at
a constant cumulative grade, and (2) operation at a higher Implications of micro-pricing
cumulative grade creates a larger gap between optimal and There are several important implications that arise from
suboptimal operation due to the sharp decline in yield as the the micro-pricing concept. First and foremost, the concept
ultimate concentrate grade is approached. A delta of more shows that the optimum operating point for a concentrator
than $7.4 million exists between operating at optimal and is determined not by the characteristics of the ore, but by the
suboptimal conditions when pushing the concentrator up to specifications outlined in the sales contract. The contract is

Table 5
Production calculations for simulations constrained to a constant incremental concentrate grade.

Total (100 percent) 50 percent good ore 50 percent poor ore

Conc. Conc. Value Value Cum. Incr. Incr. Cum. Incr. Incr.
(% Cu) (t/a) ($/t) ($million/a) (% Cu) (% Cu) ($/t) (% Cu) (% Cu) ($/t)
24.5 732,988 $708 $518.6 25.9 3.16 −$53 22.0 3.16 −$53
25.5 699,940 $743 $520.2 27.2 3.50 −$41 22.6 3.50 −$41
26.5 669,241 $779 $521.2 28.1 3.87 −$28 23.7 3.87 −$28
27.5 640,420 $814 $521.6 29.1 4.65 $0 24.7 4.65 $0
28.5 613,045 $850 $521.1 30.0 5.91 $45 25.8 5.91 $45
29.5 586,313 $886 $519.3 30.9 7.60 $105 26.9 7.60 $105
30.5 558,988 $921 $515.0 31.8 10.66 $214 28.0 10.66 $214

www.miningengineeringmagazine.com Mınıng engıneerıng    february 2017 39

Technical Papers
used to identify the zero-value particle, which in turn is used alleviated through the successful implementation of the mi-
to establish the appropriate incremental quality that should cro-pricing approach to concentrator optimization.
be targeted by the grinding and flotation circuits within the
concentrator. Once determined, this critical incremental grade Summary
would not change as ore type varies. The critical incremental The profitability of most base metal concentrators is
grade may change in response to market conditions, but gen- highly dependent on the day-to-day performance of their
erally only over the long term to avoid unwanted disturbances flotation circuits. Unfortunately, traditional online control
created by day-to-day fluctuations in market pricing. systems that are designed to stabilize the cumulative con-
A second notable implication of micro-pricing relates centrate grade of the froth product generally provide subop-
to the monitoring and control of flotation circuits. Micro- timal economic returns. This strategy also provides little or
pricing suggests that automatic control systems designed no fundamental insight that can be used by plant operators
to provide froth concentrates of constant cumulative grade for day-to-day plant management. To address this shortcom-
throughout the production cycle will result in a less than ing, a new approach to optimization based on micro-price
optimal revenue stream. These types of suboptimal control analyses has been developed. Micro-pricing assigns price
systems respond to variations in feed quality and separation tags to each particle passing through the concentrator. Op-
characteristics by varying the incremental grade as neces- timal performance occurs whenever all particles with posi-
sary to maintain a constant cumulative grade. As described tive value are generated through liberation, recovered and
above, this will not provide an optimal operating point in passed to market. The most important insight derived from
terms of revenue. In contrast, an optimal control system micro-pricing, however, is that the optimum operating point
would monitor and hold constant the incremental grade of for a flotation bank is fixed by the sales agreement and not
the last grouping of particles recovered/rejected by the flota- by the feed characteristics or separating behavior of the in-
tion circuit. While an exact measurement is difficult in actual dividual feed ores passing through the concentrator (assum-
practice, field studies indicate that tracking of incremental ing a fixed degree of liberation). For a hypothetical copper
quality can be accomplished by online sampling and analysis flotation circuit evaluated in the present study, the micro-
of the grade of the froth product generated by the last cell in pricing concept suggests that operation a constant incre-
a flotation bank. This measurement would track the grade of mental grade of 4.6 percent Cu will maximize revenue gen-
the last recovered increment reporting to froth product. Em- erated by the concentrator, as this critical grade corresponds
pirical correlations can then be used to estimate the incre- to the increment with a zero-dollar price tag. Operation at
mental grade. Commercial flowsheet simulation programs this optimal point provided a $2.4 million gain in value for
can be used to help establish these correlations (Noble et the concentrator when compared with operation at a con-
al., 2014). Optimal operation would require that this incre- stant cumulative grade. The data also suggest that new con-
mental grade be held constant at all points in time and, ide- figurations of online control measurements and downstream
ally, held at the critical zero-value incremental grade set by blending systems may be required to fully take advantage of
micro-pricing. Automatic adjustment of reagent dosages, or this optimization strategy. ■
other controllable parameters, can then be used within the
control routine to maintain this optimal operating condition. References
A third implication of micro-price optimization is the Fuerstenau, D.W., 1999, “The froth flotation century,” Advances in FlotationTechnology,
B.K. Parekh and J.D. Miller, eds., Society for Mining, Metallurgy & Exploration,
need for downstream blending facilities. For fluctuating feed Inc., Englewood, CO, 463 pp.
ores, operation an optimal incremental grade implies that Greet, C.J., 2010, “Flotation plant optimisation: A metallurgical guide to identifying
the cumulative grade generated by the concentrator will and solving problems in flotation plants,” Australian Institute of Mining and
Metallurgy, 210 pp.
vary during the production cycle. In most cases, this variation Luttrell, G.H., 2013, “Coal Micro-Pricing: A Strategy Learned from Working Between
may or may not be acceptable to the downstream consumer. Mines and Markets,” Luncheon Keynote, 2013 Society for Mining, Metallurgy &
Therefore, the concentrator must install blending systems Exploration/Pittsburgh Coal Mining Institute of America (SME/PCMIA) Annual
Joint Meeting, Canonsburg, PA, Oct. 24-25, 2013.
to ensure that a constant and homogeneous concentrate is Luttrell, G.H., and Noble, C.A., 2013, “Optimization of Flotation Plant Performance
available for shipment. In practice, this can be achieved by Using Incremental Quality Analysis,” FLSmidth Flotation Conference, Valparaiso,
placing higher-grade concentrates in one pile/bin and low- Chile, May 23-24, 2013.
grade concentrates in a second pile/bin. Blending from these Luttrell, G.H., and Noble, C.A., 2015, “Micro-Price Optimization: An Industrial Case
Study for Coal Processing Facilities,” 2015 SME Annual Conference & Expo,
piles can then be controlled using an automatic ratio feeder, Denver, CO, Feb. 15-18, 2015, Society for Mining, Metallurgy & Exploration, Inc.,
or by manual control, to ensure that a consistent on-spec Englewood, CO.
product is produced that can meet the target smelter con- Luttrell, G.H., Noble, A., and Stanley, F., 2014, “Micro-Price Optimization of Coal
Processing Operations,” 2014 SME Annual Conference & Expo, Salt Lake City, UT,
tract. In essence, the variations in bin/pile levels are used to Feb. 23-26, 2014, Society for Mining, Metallurgy & Exploration, Inc., Englewood,
absorb unwanted fluctuations in feed grade and separating CO, Preprint No. 14-127, 5 pp.
characteristics that otherwise would have been corrected by Lynch, A.J., Harbort, G.J., and Nelson, M.G., 2010, “Circuit control,” Chapter 10,
History of Flotation, Australian Institute of Mining and Metallurgy, Carlton, Victoria,
adjusting set-points for the flotation banks. As micro-pricing Australia, 348 pp.
tends to support consistent and stable set-points, most flota- Lynch, A.J., Johnson, N.W., Manlapig, E.V., and Thorne, C.G.., 1981, “Mineral and
tion circuit operators tend to find this approach easier to ac- coal flotation circuits – their simulation and control,” Developments in Mineral
Processing, Vol. 3, Elsevier Publishing, New York, NY, 291 pp.
commodate in an industrial plant environment. Richer feeds Noble, A., Bratton, R.C., Luttrell, G.H., and Wiseman, D., 2014, “Solving Process
tend to produce richer concentrates and leaner feeds tend Engineering and Flowsheet Design Challenges with LIMN,” 2014 SME Annual
to produce leaner concentrators. Attempts by operators to Conference & Expo, Salt Lake City, UT, Feb. 23-26, 2014, Society for Mining,
Metallurgy & Exploration, Inc., Englewood, CO, Preprint No. 14-070, 7 pp.
level out these routine peaks and valleys are often a source Wills, B.A., and Munn, T.J., 2006, Mineral Processing Technology, Butterworth-
of constant frustration for plant operators that can be partly Heinemann, Burlington, MA, 456 pp.

40 february 2017    Mınıng engıneerıng www.miningengineeringmagazine.com