International Journal o f Mineral Processing, 5 (1979) 321--334
321
© Elsevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands
FLOTATION CIRCUITS FOR POORLY FLOATING COALS
BRUCE A. FIRTH, ANDREW R. SWANSON and STUART K. NICOL The Broken Hill Proprietary Company Limited, Central Research Laboratories, Shortland, 2307, N.S.W. (Australia)
(Received July 10, 1978; revised and accepted October 6, 1978)
ABSTRACT Firth, B.A., Swanson, A.R. and Nicol, S.K., 1979. Flotation circuits for poorly floating coals. Int. J. Miner. Process., 5: 321--334. A number of flotation circuits for the recovery of a poorly floating coal were investigated by laboratory batch testing. Analysis of the size and ash distributions of the products and tailings showed that the circuits which would allow an equitable distribution of collector between the coarse and fine size fractions were superior. These circuits were two-stage reagent addition, reflotation of classified tailings and split feed flotation. The reflotation of classified tailings circuit not only gave the best metallurgical performance, it also was the least affected by pulp density variations and imperfect size classification.
INTRODUCTION Coal f l o t a t i o n circuits, u p to the p r e s e n t time, have generally e m p l o y e d a single-stage r e a g e n t a d d i t i o n within, or p r i o r to, a b a n k o f f l o t a t i o n cells. H o w e v e r , t h e r e n o w a p p e a r s to be a case f o r considering alternative flotation circuits, p a r t i c u l a r l y as the need to t r e a t p o o r l y floating coals increases as a result o f e n v i r o n m e n t a l pressures and the necessity to i m p r o v e overall coal r e c o v e r y . V a r i a t i o n s in the f l o a t a b i l i t y o f coals f r o m d i f f e r e n t seams and even f r o m the same s e a m can o f t e n o c c u r and have b e e n a t t r i b u t e d to d i f f e r e n c e s in the size d i s t r i b u t i o n o f the feed (Firth et al., 1 9 7 8 ) a n d diff e r e n c e s in h y d r o p h o b i c c h a r a c t e r . This latter v a r i a t i o n can be caused b y d i f f e r e n c e s in r a n k ( B r a d y a n d Gauger, 1940), d i f f e r e n c e s in p e t r o g r a p h i c c o m p o s i t i o n ( B r o w n , 1962), d i f f e r e n c e s in the degree o f o x i d a t i o n (Sun, 1954), or t h e p r e s e n c e o f slime c o a t i n g on the surface o f t h e coal particles ( J o w e t t et al., 1956}. Earlier w o r k on the f l o t a t i o n o f p o o r l y f l o a t i n g coal has s h o w n t h a t a s u b s t a n t i a l increase in coal yield can be o b t a i n e d b y staged r e a g e n t a d d i t i o n ( F i r t h et al., 1978), and t h a t in general the k e y to successful f l o t a t i o n o f p o o r l y f l o a t i n g coals is t h e e q u i t a b l e d i s t r i b u t i o n o f c o l l e c t o r to all size fractions. In the p r e s e n t w o r k , an Australian coal w i t h p o o r f l o t a t i o n characteristics was used t o investigate a n u m b e r o f possible f l o t a t i o n circuits w h i c h i n c o r p o r a t e d this f e a t u r e . T h e s e circuits w e r e s i m u l a t e d on a l a b o r a t o r y scale.
322
The experimental scheme was chosen so that the circuits could be tested in the first instance under controlled conditions, that is, constant pulp density, perfect size classification, and identical feed characteristics. The flotations were taken to completion so that residence-time effects were eliminated. In an operating context, of course, constant pulp density and perfect size classification are unlikely to be achieved and for this reason some consideration of the effects of random fluctuations in these parameters are also presented. EXPERIMENTAL
Materials
The sample of Australian coal used in this study originated from the Bowen Basin in Queensland. The sample was wet screened on a 0.5 mm wedge wire screen and the undersize material used as the feed to flotation. This material was then air dried at room temperature to facilitate subsampling. Tests showed that this procedure did not have any effect on the flotation response of the coal. The size and ash distribution of this material is given in Table I. Pro cedu re
The flotation tests were performed with a Denver Laboratory Sub-AFlotation machine (Model-D-l). During all tests the impeller speed was fixed at 1200 rpm and an aeration rate of 7 1/min was used. A pulp density of 10% w/w was used throughout unless otherwise stated. To perform a TABLE I Size and ash d i s t r i b u t i o n for t h e f e e d t o f l o t a t i o n Size
Weight %
Ash (% db)*
27.0
14.3
--500 -~ 18.9 + 250 ~m J
14.1
--250 "-~ 17.0 ÷ 125 ~rn J
18.0
--125 } + 63 u m
12.9
20.7
- - 63 ~m
24.1
24.6
fraction
+ 500 ~ra
Feed a s h = 18.2. * d b -- d r y basis.
323 flotation test, the pulp was first conditioned for one minute without aspiration or reagent addition. The flotation reagents, which consisted of a collector, kerosene (specific gravity 0.79 at 15°C and viscosity 1.2.1.2 .3 Ns/m 2 at 22°C (1.20 cP}), and a frother, methyl isobutylcarbinol (MIBC), mixed in the ratio of 10:1, were then added. In all the circuits the total reagent addition was 2 kg/tonne unless otherwise stated. The product and tailing from each circuit were wet screened on woven-wire at 500, 250, 125 and 63 p m and analysed.
Flotation circuits Single-stage flotation. This simple flotation circuit mentioned in the introduction is shown diagrammatically by Circuit A in Table II. Its history extends from the beginning of coal flotation (Bacon and Hamor, 1919) TABLE II YIELD PRODUCT % by wt ash adb
FLOTATION CIRCUIT
Single Stage Rotation
PI~)I~JCT { ~~TA,UNGS
F&O B)
TAIUNGS
~
:D) FEED ~
G
5
27'0
50"0
8"2
313
230
51.0
13"0
23"5
90
65'0
12"8
28"6
145
75.2
11"Z
39"5
265
79.6
11'0
43 6
315
81.0
9"9
58-0
475
P'ROOI.JCT ~ ~~TAIUNG$
Splrl Feed Flotation :F)
PRODU~CT
Two Stage Reagent Addition P E E ~ U N Reflototien of Classified failings
115
PRODUCT
lwo Stage Conditioning
:G )
14.0
59"6
PROOUCT
Desliming C) Separate Cenditioningof the Course end Fine Size Fractions F
TAILINGS ash adb
GS P~OOUCr,
324
till the present day and it has been extensively studied by many research workers (Brown, 1962; Glembotskii et al., 1963). This circuit was simulated by a single batch flotation with a reagent addition of 2 kg/tonne. To illustrate the poor flotation characteristics of this coal, a number of single-stage flotations were performed at different levels of reagent addition.
Desliming. In this circuit the finest size fractions (e.g.--63 pm) are removed by classification and the oversize presented to the flotation cells (see circuit B, Table II). The removal of this part of the flotation feed facilitates the recovery of the remaining coal in the feed, but may involve the overall loss of a significant portion of the coal. The latter situation, while acceptable in the past, would seldom be deliberately adopted in the present day. This circuit was one of the earliest attempts at altering the flotation circuit to improve the recovery of coal from this process. An example of its use is found in the coal washeries on the Australian Bowen Basin coal field, where the flotation of the complete feed has proved difficult (Williamson and Arnold, 1977). This circuit was simulated using a feed that had been deslimed by wet screening at 63 t~m and a single batch flotation performed with a reagent addition of 2 kg/tonne on a total undeslimed feed basis at a pulp density of 10%.
Separate conditioning of the coarse and fine size fractions. In this circuit, prior to the flotation process, the feed is classified into two size fractions and the collector is added separately to each fraction. After a suitable conditioning time the two fractions are mixed together and presented to the flotation cells. This circuit is shown in section C, Table II. By this treatment the collector is more likely to be equitably partitioned between the two size fractions, so the flotation response of the coarse size fraction should not be depressed by any disproportionate adsorption of the collector by the fine size fractions. Glembotskii et al. {1974) have tested this circuit in a Russian coal washery and claim it achieved an improved performance over singlestage flotation. However, the data which they presented show that this improvement was only marginal. To model this circuit, the feed to flotation was wet screened at 250 t~m and the oversize and undersize were separately conditioned with equal amounts of collector. The pulps, thus obtained, were recombined and a batch flotation performed. Preliminary experimentation had indicated that flotation performance was deficient in the +250 t~m sizes so classification was made at 250 t~m.
Two-stage conditioning. An intensive conditioning of the feed prior to the flotation cells is a feature of this circuit. Part of the collector is added in a conditioner and the remainder of the collector is added just prior to the first flotation cell, see circuit D, Table II. Burdon et al. (1976) have claimed a significant improvement in flotation yield (22% by weight), with this
325 circuit at an Australian Southern NSW coal washery. In the case in question it should be noted that the total collector addition was only 0.3 kg/tonne, indicating that the particular coal must be classified as a readily floating coal. It was not known what role, if any, this multiple conditioning would play in dealing with a poorly floating coal. The circuit model adopted consisted of a 1 kg/tonne addition of collector, one minute conditioning time, another I kg/tonne addition and finally, a further one minute conditioning time before aeration commenced.
Split feed flotation. This circuit is a natural development from feed desliming in which an attempt is made at recovering the fine reject coal. Accordingly, the feed to flotation is classified into two size fractions and each size fraction treated in its own bank of flotation cells, see circuit E, Table II. This circuit was advocated in the early 1960's by Bearce (1961, 1962) and recently by Williamson and Arnold (1977) who claimed good recoveries of an Australian Bowen Basin coal with this circuit. This circuit was simulated by wet screening the feed at 250 #m and the oversize and undersize were floated separately with the reagent addition equally divided between the two streams. In the present work it was assumed that the overflow stream would be thickened prior to passing to the ultrafine flotation bank. Studies have shown (Firth et al., in prep.) that no metallurgical penalty would be suffered through performing the flotation at this higher pulp density.
Two-stage reagent addition. For a poorly floating coal, a single-stage flotation can be regarded as a form of classification {Firth et al., 1978). Hence the staged addition of reagents is equivalent to a fines classification step followed by the flotation of the coarse fraction in the second stage, see circuit F, Table II. This circuit has been used in a number of coal washeries in Europe and the United Kingdom with little c o m m e n t in the literature. Lewis gave a resume of the coal flotation practice in the United Kingdom in 1961, and praised the use of staged reagent addition for the improved collection of the coarse size fractions. The model for this circuit consisted of adding 1 kg/tonne of reagent, performing a flotation and then repeating the procedure on the tailings from the first flotation. Re flotation of classified tailings. In this circuit, the tailings from the first bank of flotation cells are classified prior to passing to a second bank of cells, see circuit G, Table II. By removing much of the fine mineral matter from the first stage tailings the a m o u n t of entrained mineral matter in the product is minimised. This modification does not appear to have any previous history in the coal-flotation literature. The model consists of a flotation at 1 kg/tonne reagent addition followed by screening the tailings at 125 pm and performing another flotation on the oversize using a reagent addition of 1 kg/tonne (initial feed basis).
326 Simulation o f working conditions
In the circuits involving a classification stage some imperfection in the sizing operation must be expected. The amount of misplaced material depends on the type of equipment used to perform the classification, e.g., hydrocyclone or sieve bend. The effect of imperfection in size classification on the overall flotation response was studied for the two circuits, G and E, by returning a representative portion of the undersize material to the oversize. A further practical problem of importance is the response of the chosen circuit to variations in the nature of the feed material. For example, the pulp density of the feed to flotation in most coal washeries is not controlled or even monitored, and it would be expected that the pulp density could vary over a considerable range of values owing to random variations in upstream operations. In a plant the reagent dosage would be fixed for a particular pulp density, hence if the pulp density were to double the effective reagent addition would be halved. The effect of this type of variation on the three circuits, single-stage flotation, two-stage reagent addition, and the reflotation of classified tailings was investigated. In the experimental system~ the reagent addition was fixed at 2 kg/tonne for a 10% pulp density. Criteria for circuit comparison
In coal washing there are three parameters of particular interest, yield Y, product ash Ap, and tailings ash A T. To facilitate the comparison between the above flotation circuits a simple convenient mathematical combination of the above parameters was required. One such parameter is the efficiency index, E (Tsiperovich and Evtushenko, 1959; Swanson et al., 1978), given by: Z = (Y X AT)/A p
It has been shown that by maximising the value E, the o p t i m u m combination of the three parameters is obtained and can be related to the financial return from the coal washing operation (Swanson et al., 1978). RESULTS AND DISCUSSION The response of the coal, selected for study, to single-stage flotation is shown in Fig. 1. Its poor flotation characteristics are demonstrated by the low maximum yield and the relatively high reagent consumption necessary to obtain this yield. The products and tailings from these flotations were sized and the low yield was shown to be due to the coal in the coarser size fractions reporting to the tailings (see Fig. 2, curve A}. The yields, product ashes, and tailings ashes for the seven flotation circuits are shown in Table II, along with the efficiency index (rounded to the nearest
327
5 unit) for each of the circuits. Figs. 2 and 3 show the calculated recoveries of carbonaceous material, as a function of particle size, obtained using each of the circuits described. Inspection of the data shows that the poor overall yields obtained with circuits A, C and D arose from their incapacity to float the coarse size fractions. This shortcoming is reflected in the poor quality of the tailings and low E values. The data for circuit B reveal that, once the finer fractions had been removed from the feed, the coarser fractions float readily but that such a procedure leads to an overall yield penalty. However, a high E value is obtained as a result of the very low ash content of the product. The non-zero recovery of --63 ~m coal shown in Fig. 2 is believed to arise from attrition during the flotation process. Single-stage flotation and two-stage conditioning, for the same total reagent addition, give significantly different results. This result implies that the collector addition, in the conditioning process, was non-additive. The reason for this behaviour is not clear at present, but a tentative explanation could be that the collector is distributed more efficiently throughout the pulp during the conditioning process. Satisfactory collection of coarse particles was achieved with circuits E, F and G as evidenced by the observed improvements in overall yield and E values shown in Table II. Circuit G would appear to be the outstanding system, having an E value of 475. Analysis of the products show that twostage flotation (circuit F) has a bias towards the complete flotation of the finer size fractions, whereas reflotation of classified tailings (circuit G) appears to be biased against the size fraction immediately below the size 80
60
YIELD % BY WEIGHT
40
20
I
J
I
I
1
2
3
4
REAGENT ADDITION
Kg/TONNE
Fig. 1. The effect of reagent addition on the yield f r o m a single-stage flotation.
328
"\
90 80
circuit isymbol\\~I
3[ 2C
D
-~---
-125.63 -250.125-500.250.500 SizeFrochon I
I
I
L
l
Fig. 2. Carbonaceous recovery as a function of size for circuits A, B, C and D. at which the classification is performed. In the case of circuit E, inspection o f the data shows that the --250/~m fraction flotation response is depressed compared to even single stage flotation but the recovery of the +250 pm material more than compensates for this loss. The reduced flotation efficiency of the fine fractions is believed to be due to excessive reagent consumption by the ultra-fine material (--63/~m). It must be n o t e d that circuit G will p r o d u ce a much higher volume of railings and may be more demanding on water supplies than circuits E and F. The reason for this is, that a large percentage of the feed water will be discarded during the intermediate size classification step, necessitating the use of make-up water to perform the ~econd flotation. This additional volume of tailings will mean that an increase d capacity is required for water clarification. For the particular coal in question, a possible m e t h o d of improving the
329 100
7 `0``` . 90
80
',, \ ..---\.o.
70
-
60 Iorcuit symbol 50
'~ 40
E
---o.
F G
--o
30 20 10
-63 -125"63 -250:125 -500÷250 "500 Size Frochon Fig. 3. The c a r b o n a c e o u s r e c o v e r y as a f u n c t i o n of size for circuits A, E, F a n d G.
efficiency of two-reagent addition is evident from consideration of Table III which shows the ash content as a function of particle size for the product of the second stage flotation. In principle therefore, it should be possible to improve the process efficiency by judicious removal of high ash size fractions. To select the o p t i m u m size for such a classification, the cut-point corresponding to the maximum value of the Efficiency Index E has been chosen. Thus Fig. 4 shows that the overall value of E can be increased by fifteen points if the second-stage product is classified at 63 ~m. However, this value of E is still lower than that obtained for reflotation of classified tailings (circuit G). Also, the practicality of classifying a flotation froth must be questioned.
Response o f circuits to working conditions From the foregoing arguments it is considered that the preferred circuit,
330 TABLE III Size and ash distribution for the second-stage product from circuit F Size fraction
Weight %
+500um
44.6
4.5
~ 25.2
7.4
--500
Ash (% db)
+ 250 pm J --250 ~ 16.3 + 125 pm J
9.9
--125 ~ 7.0 + 63 pm J
16.2
6.9
41.4
--
63
um
340
320
30(3
28G
260
Z40 0
I
I
I
100
200
250
SIZE OF CLASSIFICATION OF SECOND STAGE PRODUCT
Fig. 4. Efficiency Index for two-stage reagent addition as a function of the size at which the second-stage product is classified.
would be either two-stage reagent addition (circuit F), reflotation of classified tailings (circuit G) or split feed flotation (circuit E). To expand the information base from which a decision between these options can be made, it is relevant to consider the response of these circuits to practical problems such as imperfect classification and random changes in feed pulp density.
331 10Q
80'
I REFLOTATION OF CLASSIFIED TAILINGS- CIRCUIT G
A
o
SPUT FEED CIRCUIT E
40 S
LIJ
20
,,
l
L
I
10
20
30
% -G3um HATERIAL REPORTING TO OVERSIZE FRACTION Fig, 5. F l o t a t i o n yield as a f u n c t i o n o f the a m o u n t o f --63 # m material r e p o r t i n g to the oversize f r a c t i o n for split feed and r e f l o t a t i o n o f classified tailings circuits.
In Fig. 5, the variations in the flotation yield with respect to the amount of --63 t~m material reporting to the oversize fraction for the split feed (circuit E) and refloating of classified tailings {circuit G) are reported. Inspection shows that the presence of very small quantities (< 5%) o f - 6 3 pm material has a deleterious effect on the flotation yield for the split feed circuit and that this effect is likely to represent a serious drawback, particularly with the use of cyclones, to this circuit. In contrast, the yield for reflotation of classified tailings was not affected by misplaced material. This phenomenon can be explained by a comparison of the material in the two circuits. In the split feed circuit, a major constituent of this material is likely to be fine hydrophobic coal particles. Such particles have been shown (Firth et al., in prep.) to be detrimental to the flotation owing to their ability to adsorb collector at the expense of coarser particles. Hence the loss in yield as a result of misplaced material, seen in the case of split feed flotation, seems likely to be due to this effect. In the case of reflotation of classified tailings these particles would report to the froth in the first flotation step and thereby be removed from the system. In the absence of strict process control measures, the usual effect of variations in the feed pulp density is to create an imbalance in reagent addition. For example, if the pulp density is halved the reagent addition (kg/ tonne) is effectively doubled. Fig. 6 shows the effect of pulp density on flotation yield for two-stage reagent addition (circuit F), reflotation of
332
classified tailings (circuit G) and, for comparative purposes, single-stage flotation (circuit A). The abscissa shows the effective reagent additions corresponding to a steady reagent addition of 2 kg/tonne at 10% pulp density. The data show that reflotation of classified tailings (circuit G) is insensitive to pulp density variations. Single-stage flotation shows a small decrease in flotation yield with increasing pulp density while two-stage reagent addition (circuit F) shows a dramatic decrease in flotation yield at high pulp densities. Size analyses of the product and tailings (Fig. 7) show that this latter effect is due to the increasing amounts of --63 pm material presented to the second-stage flotation as the pulp density is increased. CONCLUSIONS
The optimum flotation circuit for the washing of the coal in question, was the reflotation of classified railings (circuit G). It produced the best yield and gave the highest efficiency index value. Also, it was the circuit which was least affected by variations in working conditions.
"''-
I) ~
REFLOTATION OF CLASSIFIED TAIIJNGs - CIRCLNT G
~
80
0
0
m
0
\
.~ 7o
TWO STAGE~ R EAGENT M]DfflON CIRCUIT F
>-
60 ×
--" X
A
SIN6LE STAGE- CmCUfT A
I
5
10 PULP DENSITY ( wt % ) ~ENT
I
15
X I
2O
ADDITION Kg/tonne
Fig. 6. F l o t a t i o n yield as a f u n c t i o n o f feed pulp density for circuits A, F and G. (Reagent addition is fixed at 2 k g / t o n n e for 10% pulp density. Thus a m o u n t present at o t h e r 10% densities = 2 × ) actual % pulp density
333 20
\ 8 ~ z o
RECOVERY OF * 500 /Jm COAL FROM THE SECOND STAGE
\
15
\
o
\
o
== LL
\
i Z
\
10
\
O.
\ t=
\
Z
/
5
~ "
8
f
/ 0
9/ 5
/
f
f
f
J
f
~"
- 6 3 j u r a COAL PRESENTED TO THE SECOND STAGE
/
I 10
PULP
I 15 DENSITY
I 20
J 25
Fig. 7. The effect of ultra-fines on the recovery of coarse coal as a function of pulp density. (The presence of--63 ~m coal in the second stage due to the increase in pulp density has severely depressed the flotation of the +500 pm coal.) The generalisation of this result to other coals w o u l d be d e p e n d e n t u p o n their h y d r o p h o b i c i t y and size distribution, but it w o u l d be difficult to believe that this result is specific to the coal used in this work. For an existing washery the easiest circuit to install and the one which w o u l d give a significant i m p r o v e m e n t in yield, w o u l d be two-stage reagent addition (circuit F). ACKNOWLEDGEMENT The authors w o u l d like to express their appreciation to Mr. T. Roberts and Mr. G. Burgin for their assistance with the experimental work and w o u l d like to thank The Broken Hill Proprietary C o m p a n y Limited for permission to publish this work.
334 REFERENCES Bacon, R.F. and Hamor, W.A., 1919. Problems in the utilisation of fuels. J. Soc. Chem. Ind., 38: 161. Bearce, W.A., 1961. Mechanics and control in coal flotation. Colo. Sch. Mines Q., 56: 371. Bearce, W.A., 1962. Progress in froth flotation. Min. Congr. J. Wash., 48: 37. Brady, G.A. and Gauger, A.W., 1940. Properties of coal surfaces. Ind. Eng. Chem., 32: 1599. Brown, D.J., 1962. Coal Flotation. In: D.W. Fuerstenau (Editor), Froth F l o t a t i o n 50th Anniversary Volume. AIME, New York, N.Y., 518 pp. Burdon, R.G., Booth, R.W. and Mishra, S.K., 1976. Factors influencing the selection of processes for the beneficiation of fine coal. In: A.C. Partridge (Editor), Proceedings of the Seventh International Coal Preparation Congress, E.1. Firth, B.A., Swanson, A.R. and Nicol, S.K., 1978. The influence of feed size distribution on the staged flotation of poorly floating coals. Proc. Australas. Inst. Min. Metall., 267: 49. Firth, B.A., Swanson, A.R. and Nicol, S.K., in prep. Glembotskii, V.A., Klassen, V.I. and Plaksin, I.N., 1963. Flotation, part 3, chap. II. (Primary Sources: New York). Glembotskii, V.A., Zaikin, S.A., Panova, A.A. and Rubinshtein, Yu.B., 1974. Research into the possibilities for separate conditioning in coal-slurry flotation. Koks Khim., 6: 8. Jowett, A., EI-Sinbawy, H. and Smith, H.G., 1956. Slime coating of coal in flotation pulps. Fuel, 35: 303. Lewis, J.L., 1961. Present flotation practice in the coal industry in the United Kingdom. Colo. Sch. Mines Q., 56: 333. Sun, S.C., 1954. Effects of oxidation of coals on their flotation properties. Trans. AIME, 199: 396. Swanson, A.R., Firth, B.A. and Nicol, S.K., 1978. A convenient index for the assessment of coal cleaning processes. Proc. Australas. Inst. Min. Metall., 268. Tsiperovich, M.V. and Evtushenko, V.Ya., 1959. Preparation and Carbonisation of Coals, vol. 1, Sverdlovsk, Metallurgizdat, 72. Williamson, M.M. and Arnold, J.J., 1977. The application of bore core data to coal preparation plant design. Aus. I.M.M. Symposium on Coal Borehole Evaluation, 114.