Transcript Diapositiva 1 - CETEM - Centro de Tecnologia Mineral

Experimental Design to evaluate a reagent
system for a nickel ore flotation
Authors
• Jean Louzada; Ronald Hacha; Marisa Monte
and Mônica Cassola
• (a) CETEM – Centre for Mineral Technology,
Rio de Janeiro, Brazil
• (b) Clariant S.A, São Paulo, Brazil
MOTIVATION
•The optimization of flotation conditions is a complex
task because many process variables can affect
flotation responses.
•It is not uncommon for multiple interactions to occur
between independent variables;
•The identification of these interactions play an
important role in advancing our understanding of the
chemistry of such system in plant operations.
OBJECTIVES
• To employ a factorial design to investigate the
effect of chemical variables on the flotation
performance of dithiophosphates for a nickel
ore;
• To optimize these variables for maximum
nickel recovery and grade.
EXPERIMENTAL
– REAGENTS
– Ethyl secbutil sodium dithiophosphates (Hostaflot E501) and
sodium dialkyl dithiophosphate (Hostaflot M92) were supplied
by Clariant;
– Polypropylene glycol methyl ether (CH3(OC3H6)n-OH and
other consisting of a mixture of aliphatic alcohols, ethers and
esters. The two frothers were supplied by Clariant .
– The activator and the depressant used were copper sulfate
and carboxymethyl cellulose, respectively
EXPERIMENTAL
– A nickel ore sample from Minas Gerais, Brasil, was
completely characterized for mineralogical and
chemical compositions:
• Mineralogical composition, associations and liberation
were measured in a FEI Quanta 400 SEM with the
Mineral Liberation Analyzer (MLA) software.
• Chemical analysis were carried out in a PanAnalytical
Epsilon 3 X-ray Fluorescence machine.
EXPERIMENTAL
– Factorial Design;
• Only factors which influenced in the recovery of nickel
by flotation will be presented here.
• The factorial design was implemented with two levels
and six factors resulting in thirty two experiments.
• The Statistics software was used for the regression
analysis, statistical and optimization calculations.
Table 1. Factors and levels applied in 26-1 fractional
factorial design. Flotanol (low: 20; high: 40); Montanol
(low: 20; high: 80).
Parameters Values
Factors
Symbols
Units
Low level (-1)
High level (+1)
Collector Type
A
--
Hostaflot M92
Hostaflot E501
Collector
Concentration
B
(g/t)
80
200
Frother Type
C
--
Flotanol
Montanol
D
(g/t)
Low
High
E
(g/t)
50
100
F
(g/t)
50
100
Frother
Concentration
Dispersant
Concentration
Depressant
Concentration
Experimental
• Flotation Tests
•
•
•
•
•
•
•
The samples were ground in a rod mill, to which were added the dispersant and
the activator at pH 6.0.
Immediately after grinding, the material was deslimed and, subsequently, the
sample was transferred to a cell with two liters.
The pulp was kept under stirring at 1400 rpm and the pH was adjusted to 9.5 with
a solution of NaOH 10% (p/v).
pH adjustment was immediately followed by depressant addition, carboxymethyl
cellulose, and conditioning for 4 minutes.
Afterwards, the collector was added and conditioned for 30 seconds.
Finally, the frother was added to the system and conditioned for 1 minute.
The pH was kept at about 9.5 during the conditioning with all reagents. The
flotation time was 4 min.
Experimental
– Curve Fitting and Statistical Analysis
• The important response variable chosen in this study
was nickel recovery
• The statistical significance of effects and interactions
between processes and the response variable was
determined using the F-test.
• Probability (P) values larger than 0.05 were indicative
of a measured effect being statistically significant at a
confidence level  95%
RESULTS AND DISCUSSION
Mineralogical characterization
• These studies showed that the major minerals are
talc, hornblende, ilmenite, pyrite and pyrrhotite.
• The results revealed that talc is not the predominant
magnesium carrier mineral
• Hornblende is present and predominates over talc
in all ranges of particle size.
Figure 1 Mineralogical Composition as a function of the particle size
Talc
Apatite
Chlo rite1
Ilmenite
Pyro xmangite
Titanite
Pentlandite
Calc ite
Do lo mite
K_feldspar_1
Pyro xmangite_Pyro xf
Magnesio ho rnblende
Carbo nato FeMg
FeO
Plagio c lase1
Pyrrho tite
Ankerite
Chalc o pyrite
Ferro silite
Pyrite
Quartz
100%
90%
80%
70%
Massa (%)
60%
50%
40%
30%
20%
10%
0%
-300+210
-210+150
-150+106
-106+75
Fração (um)
-75+53
-53+38
Figure 2 Minerals associated to pentlandite in the particle size range
between 210 and 38 μm.
Talc
12%
Magnesiohornblende
3%
Ankerite
2%
Free Surface
52%
Calcite
4%
CarbonatoFeMg
2%
Chlorite1
3%
Dolomite
0%
Ferrosilite
5%
Pyrite
3%
Quartz
1%
Pyrrhotite
6%
Pyroxmangite_Pyroxf
5%
Figure 3 - Synthesis of the results obtained for theoretical recovery and
grades of pentlandite in the concentrate at different size ranges.
Pareto Chart
Pareto Chart of Standardized Effects; Variable: Recuperation of Ni (%)
6 factors at two levels; MS Residual=12.37447
DV: Recuperation of Ni (%)
B
D
B*D
A*C
C*E
C
C*D
A*C*D
E
A*B*C
B*C
C*F
B*C*E
A*F
A
A*B*E
A: Type of colletor
B: Concentration of
C:Type of frother
D: Concentration of
E: Concentration of
F: Concentration of
A*B*F
B*C*D
B*E
B*C*F
A*B*D
p=.05
Standardized Effect Estimate (Absolute Value)
colletor
frother
disperst
depressor
Analysis of Variance
These variables and their interactions presented higher probabilities:
• B: Concentration of collector ;
• D: Concentration of frother
• BD: interactions between them
• In other cases, the null hypothesis is rejected because the estimated values
of p-levels (Test P) are smaller than 0.05, i.e., the effects have a probability
smaller than 5% so they represent only of noise.
Response Surface
.
Recu
peration
of Ni (%)
(%)
Nickel
Recovery
Fitted Surface; Variable: Reuperation
Nickel Recovery
of Ni
(%)(%)
6 factors at two levels; MS Residual=14.1346
DV: Recuperation
Nickel Recovery
of (%)
Ni (%)
Co
nc
Fern t
ortha
et iro
n
C
oonf
cfer
ont r
hae
trio
(gn (
/t)g/
t)
tt))
(g
(g/ /
r
n
o
ctti o
lltera
o
n
c
f ce
nCoo n
o
i
r
t
tracto
eonll e
c
nC
Co
CONCLUSIONS
• The results of these studies showed that the
main factors that influence more significantly
the nickel recovery are the collectors and
frothers concentrations.
• The differences between the collectors are the
alkyl chains do not influence the recovery.