InllM~lOllm#OURllmW

mlnlRIIl
PROI:E|51U6
ELSEVIER Int. J. Miner. Process. 49 (1997) 185-192

Study of Merrill-Crowe processing. Part II:

Regression analysis of plant operating data
Gexia Chi *, Maurice C. Fuerstenau, Peter A. Anderson
University of Nevada-Reno, Mackay School of Mines, Reno, NV 89557, USA

Abstract

The recovery of precious metals from alkaline cyanide solution is accomplished effectively
utilizing the well-established Merrill-Crown process. Plant data from three silver operations were
analyzed in terms of zinc efficiency, and mathematical models were generated with multiple
regression techniques. Projections of the mathematical model are in good agreement with actual
plant performance. The results of the regression analysis showed that the silver concentration in
the feed is the dominant operating variable in determining zinc efficiency in Merrill-Crowe
processing, and this is especially true for operations with low-pregnant grades. Pregnant grade and
flow rate are the two operating variables which are beneficial to zinc efficiency. Increases in
sodium cyanide concentration and zinc addition result in reduced zinc efficiency. Finally, the
mathematical models indicate that modest economic improvement in Merrill-Crown processing
can be realized with relatively minor modifications to the circuits.

Keywords: zinc efficiency; mathematical model; regression analysis

1. Introduction

As an extension of the laboratory investigation presented in Part I of this paper, plant

data from three silver operations were analyzed by regression analysis. Table 1 lists the
operating parameters from the three plants studied and the corresponding zinc efficien-
cies on an annual average basis. Zinc efficiency is defined as the stoichiometric zinc
requireraent for precious metals precipitation divided by the total amount of zinc added.
Based on plant practice, it is expected that zinc efficiency in Merrill-Crowe precipita-
tion is strongly dependent on these operating variables and varies significantly from
plant to plant.

* Corresponding author.

0301-7516/97/$17.00 Copyright © 1997 Elsevier Science B.V. All rights reserved.

PII S 0 3 0 1 - 7 5 1 6 ( 9 6 ) 0 0 0 4 2 - 7
186 G. Chi et al. / Int. J. Miner. Process. 49 (1997) 185-192

Table 1
Plant operating parameters (annual average in 1990)
Plant Pregnant grade, Ag Flow rate Zn addition NaCN Zinc efficiency
(g/t) (gpm) (kg/day) (g/t) (%)
1 39.18 1934 322.06 815 62.1
2 13.04 4733 575.62 665 20.2
3 4.05 1797 176.45 305 8.2

The investigation described in this paper was aimed at effecting improvements in

Merrill-Crowe precipitation by optimizing zinc efficiency within the scope of practical
possibility. Multiple regression techniques were used as a means of identifying which
variables exert a significant influence on zinc efficiency to provide quantitative esti-
mates of zinc efficiency associated with these variables.

2. Merrill-Crowe precipitation

A simplified block diagram of the flowsheet for precious metal recovery by zinc
cementation is shown in Fig. 1 [1]. The pregnant leach solution is clarified by filtration,
and the clarified solution is de-oxygenated in a vacuum tower. Zinc dust is added to a
small mixing cone through which an auxiliary stream of barren cyanide solution passes,
carrying the zinc dust into the feed pipe. The cementation reaction starts at this point and
is completed as the zinc particles are collected and held in the filter press. In a

/ SOLUTION
FROMTHE VACUUM
I LrmCHTANKS ........... PUMP

[1[[[ [ I[[I (U BELTFEEDERU):::i:i

::~:':: I CYANIDE
SOLUTION TO
SP~LGuTNAc~T J~l @ ~ILL 1
TANKS

FILTERS TOWER BARREN

IrllrIJ][ll SOLUTION

PRECIPITATED
METALSTO
KEVERBERATORYFURNACE
Fig. 1. Flowsheet for gold and silver recovery by the Merrill-Crowe Process [1].
 G. Chiet al. / lnt. J. Miner. Process. 49 (1997) 185-192 187

continuous operation, a process will have several filter presses in parallel so that one
filter can be cleaned while the others are kept in operation, typically on a five to
seven-day cycle. When a cycle is completed, the cake is blown dry with air; the press is
opened, and most of the cake is discharged into a wheeled tray placed beneath the press.
The remainder of the cementation cake is removed with scrapers.
It is evident that the cementation reaction occurs both in the feed pipe and in the
plate-and-frame filter press. Of course, most of the time the zinc particles are stationary
in the press while they react. Parga, Wan and Miller [2] have shown that about 10% of
the silver is precipitated in the pipe, while the remaining 90% is removed in the filter
presses as the solution passes through the cake

3. Mathematical modeling

The SPSS/PC ÷ is a powerful software package for microcomputer data management

and analysis, which is designed for personal computers in the IBM PC and IBM P S / 2 TM
lines with a 10 MB or larger hard disk. This program was used for regression analysis of
the routine operating data from the three silver cyanidation plants. The model devised
covered the period from January 1990 to December 1990. Four plant variables were
included in the analysis, and operating data for these variables were taken on a monthly
average over a 12-month period. The four process variables are listed below:
Grade: Grade of silver in pregnant solution (g/ton);
Add: Daily zinc addition (g);
F~[ow: Flow rate (gpm);
NaCN: Concentration of sodium cyanide (g/ton).

Zinc efficiency (ZE) was selected as the dependent variable and is defined, again, as:
Stoichiometric zinc required for precious metals precipitation
ZE(%) = Total amount of zinc added X 100
The mathematical models which characterized the three Merrill-Crowe circuits in
terms of the above parameters are given in Table 2.

Table 2
Results ef multiple regression
Equation Multiple R 2
Plant 1 ZE = (38.34 X Grade) + (5.02 x 10- 2 X F l o w ) - (59.05 X NaCN) 0.9175
- ( 1 . 0 6 X 10 -1 ×Add)+91.89

Plant 2 ZE = (44.19 X Grade) + (4.44 x 10- 3 × Flow) - (5.99 X NaCN) 0.9915

-(1.68 X 10 -2 ×Add)+8.31

Plant 3 ZE = (55.96 × Grade) + (3.99 × 10- 3 × Flow) - (0.24 × NaCN) 0.9915

- ( 1 . 9 × 10 -2 ×Add)+0.64
188 G. Chi et al. l i n t . J. Miner. Process. 49 (1997) 185-192

100
Medal prediction
• Plant data
8O

60

40
Plant 2
N 20
Plant 3

L t i 1 i i

0 2 4 6 8 10 12
Month
Fig. 2. Prediction o f plant performance with mathematical models.

The coefficient of multiple regression, R 2, shows that 92% of the operating data can
be represented by Eq. (1) for plant 1, and 99% of the operating data fit Eqs. (2) and (3)
for plants 2 and 3, respectively. Non-linear behavior of the independent variables and
interactions between them, of course, are to be expected, but they were not specifically
sought in the initial analysis.
The comparisons between the predicted zinc efficiency and the actual plant data are
shown in Fig. 2. It can be seen that the agreement is good.

4. Interpretation of modeling results

As might be expected, the mathematical model for each plant reflects, for the most
part, plant operating experience. The influences of the four operating variables on zinc
efficiency are determined by their corresponding correlation coefficients with the
dependent variable. All three models show a consistent effect of these four parameters
on zinc efficiency.

4.1. Silver concentration in the feed of the pregnant solution

The influence of silver grade of the pregnant solution on the efficiency of Merrill-
Crowe processing is well recognized in plant operations. Inspection of Eqs. 1, 2 and 3
shows that grade of the pregnant solution is the only variable that exerts a similar effect
on zinc efficiency for the three different operations.
Zinc efficiency as a function of grade of pregnant solution using operating data from
plants 1, 2 and 3 is shown in Fig. 3. This figure demonstrates that grade of pregnant
solution is the major operating parameter in determining zinc efficiency in Merrill-Crowe
 G. Chi et a l . / Int. J. Miner Process. 49 (1997) 185-192 189

t00
Y = 3.97 + 1.289X, R^2 = 0.960

8O

~ 60

w 40

E
~ 20 Plant 2

I I I i i I I i I I

5 10 15 20 25 30 35 40 45 50 5
Pregnant Solution Grade (Ag), g/t
Fig. 3. Relationship between pregnant solution silver grade and zinc efficiency.

processing. An estimate of zinc efficiency can be made from Eq. (4) without considering
the operating environment and other variables:
y = 3.97 + 1.289x (4)
where x is the silver grade of the pregnant solution, and y is zinc efficiency.
As s]aown in Fig. 4, the agreement is generally good, but deviates at high grade of
pregnant solution. The deviation is due to the significant influence of the other operating
variables. It is, thus, apparent that the most promising parameter for improving zinc
efficiency in the Merrill-Crowe process is the grade of precious metals in the pregnant
solution.

100

8O

"
= 60

w 40
e- Plant 2
N
2(1
Plant 3
. u

0 I ! I I I I

0 2 4 6 8 10 12
Month
Fig. 4. Estimation of zinc efficiency over time as pregnant solution grade varies.
190 G. Chi et al. l i n t . J. Miner. Process. 49 (1997) 185-192

4.2. Flow rate

Eqs. (1), (2) and (3) show that increasing the flow rate should have a beneficial effect
on zinc efficiency, although it is likely that, above a certain limiting value, zinc
efficiency would decrease due to physical problems with filtration (cracking of the cake)
and inadequate residence time of solution, etc. This phenomenon supports Parga and
Miller's [2] observation that 90% of the silver reduction occurs in the plate-and-frame
press. Higher flow rates would result in an increase in the packed bed mass transfer
coefficient up to some point and, hence, reaction rate would increase with increased
flow rate. The extent of the beneficial effect of increasing flow rate is, of course,
dependent on the reasonable range of operating conditions.
Close examination of Eqs. (1), (2) and (3) indicates that the rate of increase of zinc
efficiency with flow rate rises with increasing grade of pregnant solution. At an average
grade of 39.18 g / t , the increase of zinc efficiency for every 100 gpm flow rate is 5%,
whereas at an average pregnant solution grade of 4.054 g / t , the increase is only 0.4%.
4.3. NaCN concentration

An increase in NaCN concentration in the precipitation circuit was found to be

detrimental in terms of zinc efficiency. This fact can be explained with the data
presented in fig. 5 of the companion paper (Part I)[3]. As shown, the solubility of zinc
dust increases significantly with an increase in NaCN concentration, which suggests that
high cyanide concentration is not always advantageous. When the cyanide concentration
is in excess of the value required to avoid any significant formation of zinc hydroxide,
and sufficient free cyanide is present to maximize the rate of precious metals precipita-
tion, then, excess zinc will dissolve, which results in a decrease of zinc efficiency. An
economic benefit could be obtained by decreasing the cyanide concentration, but caution
must be exercised to maintain the cyanide concentration sufficiently high to avoid
formation of zinc hydroxide and to maximize the precipitation rate.
4.4. Daily zinc addition

It is common to add zinc dust at a rate between 5 and 30 times the stoichiometric
precious metals requirement, depending on solution composition and operating effi-
ciency [4]. Eqs. 1, 2 and 3 show that any increase in daily zinc addition in Merrill-Crowe
processing will result in a decrease of zinc efficiency. In other words, the increased
addition will increase the consumption of zinc for the same amount of precious metals
precipitated. An improvement of zinc efficiency can be achieved by reducing daily zinc
addition with the extreme caution, though, that low barren solutions are still obtained.
Achieving a low barren is the overriding objective in the operation of Merrill-Crowe
circuits, and this is always achieved at the expense of excess zinc addition (and normally
excess NaCN).

5. Economic considerations

The data show that zinc efficiency can be increased by changing any of the four
parameters analyzed, that is by reducing zinc and cyanide addition or by increasing
 G. Chi et aL lint. J. Miner. Process. 49 (1997) 185-192 191

Table 3
Annual economic benefits derived from an increase of zinc efficiency by 15%
Operating parameter Cost saving (S/year)
Flow rate (gpm) NaCN (g/t) Zinc added (kg/day)
Plant 1 Operating range 1500-2350 650-1000 118-558 27.100
Modified value * 2150 725 277

Plant 2 Operating range 3600-5700 605-750 136-1270 48.900

Modified value * 5440 400 499

Plant 3 Operating range 1230-2010 285-360 45-227 14.350

Modified value* 2010 0 145

silver cc,ncentration in the feed and flow rate. Although an increase in pregnant solution
grade will result in an increase in zinc efficiency, this is not a parameter that is easily
changed and is usually a function of ore head grade. Solution flow rate is not easy to
change either. It is usually a function of type of process used for Ag (and Au) extraction,
wash ra~Lio's, leaching requirements, etc. Any increase in solution flow rate would be
achieved at the expense of pregnant solution grade. Table 3 indicates the cost savings
that coultd be achieved by increasing zinc efficiency by say 15%, and this strategy can be
fulfilled by modifications of either solution flow rate or zinc addition within the scope of
practicality.
Table', 3 also reveals that reduction in cyanide concentration is not a practical
approach to increase zinc efficiency, and in this case it is only achievable for plant 1.
The modified values in Table 3 were calculated based on Eqs. (1), (2) and (3) by
changinlg only one operating parameter each time while keeping the others constant.

6. Summary and conclusions

This investigation has examined the influence of several key operating variables on
zinc efficiency in the Merrill-Crowe process. The results have shown that the grade of
the pregnant solution (silver concentration in the feed) is the most significant parameter
controllJLng zinc efficiency in Merrill-Crowe precipitation. An increase in the grade and
in the flow rate of pregnant solution both increase zinc efficiency. An increase in NaCN
concentration is detrimental to efficiency due to excessive dissolution of zinc. A
reduction of daily zinc addition to a Merrill-Crowe circuit will certainly increase zinc
efficiency, but whether this is possible is determined by actual plant conditions. These
results ,;upport the theory and laboratory testwork observations discussed elsewhere in
the literature.
The mathematical models generated permit a quantitative estimate of the likely
benefits derived from minor modifications to plant practice. The models thus serve a
useful function as an aid to decision-making whether for short-term adjustments or
long-term plans for the plant operations.
192 G. Chi et al. / Int. J. Miner- Process. 49 (1997) 185-192

Acknowledgements

T h e a u t h o r s w i s h to a c k n o w l e d g e the f i n a n c i a l s u p p o r t p r o v i d e d b y the C e n t e r for

P r e c i o u s M e t a l s o f the M a c k a y S c h o o l o f M i n e s , U n i v e r s i t y o f N e v a d a , Reno.

References

[1] Parga, J.R., 1987. Analysis of the Zinc Cementation Reaction for Recovery of Precious Metals From
Cyanide Solutions, Ph.D. Dissertation Dept. of Metallurgy and Metallurgical Engineering, University of
Utah, Salt Lake City.
[2] Parga, J.R., Wan, R.Y. and Miller, J.D., 1988. Zinc Dust Cementation of Silver From Alkaline Cyanide
Solutions - Analysis of Merrill-Crowe Plant Data. Min. Metall. Processing, August, pp. 170-176.
[3] Chi, G., Fuerstenau, M.M. and Marsden, J.O., 1997. Study of Merrill-Crowe processing. Part I: Solubility
of zinc in alkaline cyanide solution. Int. J. Miner. Process., 49: 171-183.
[4] Marsden, J.O. and House, C.I., 1992. The Chemistry of Gold Extraction, Ellis Horwood Ltd., Chichester,
UK and Prentice-Hall, Englewood Cliffs, NJ.