Top Banner
International Journal of Mineral Processing, 29 (1990) 249-265 Elsevier Science Publishers B.V., Amsterdam 249 Simulation - the modem cost-effective way to solve crusher circuit processing problems R.P. King Department of Metallurgyand MaterialsEngineering, University of Witwatersrand, 1Jan SmutsAvenue, Johannesburg (SouthAfrica) (Received April 11, 1989; accepted after revision January 30, 1990) ABSTRACT King, R.P., 1990. Simulation- the modern cost-effective way to solve crusher circuit processing prob- lems. Int. J. Miner. Process., 29: 249-265. Simulation is an effective technique for the improvement of crusher plant performance, and it is now used routinely by some crusher manufacturers for both plant design and trouble shooting. MOD- SIM is probably the most versatile ore dressing-plantsimulator in general use in the mineral-process- ing industry today. A case study using MODSIM applied to the crusher circuit of a major uranium producer is described in this paper. This study demonstrates the effectiveness of simulation to im- prove plant performance when reliable and effective models of the unit operations are available. The study was commissionedto investigate a 2000qon/h 4-stage crusher plant to identify a strategy to increase production at a finer product size. After an intensive technical audit on the plant, success- ful simulation was achieved for the existing operating conditions. The simulator was then used to identify the production bottlenecks and to establish plant modifications to meet the required produc- tion objectives in a cost-effective manner. INTRODUCTION Plant simulation techniques are becoming increasingly effective and there- fore more frequently used as tools to assess and improve plant performance. This is particularly so with crushing plants because the unit operations of crushing and screening can be described by reliable and accurate models. The study reported here was undertaken to establish cost-effective modifications to the fine crushing plant of Rossing Uranium Ltd., and the application of simulation to address some of the possibilities is described. The study was undertaken by a team consisting of personnel from Nordberg (Pty) Ltd., the Department of Metallurgy and Materials Engineering of the University of Witwatersrand, and Rossing Uranium Ltd. The objectives of the study were: ( 1 ) to investigate and establish all process and operating parameters under the current operating conditions; (2) to es- 0301-7516/90/$03.50 © 1990-- Elsevier Science Publishers B.V.
17

King's Simulation

Oct 23, 2016

Download

Documents

it4ujonly
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: King's Simulation

International Journal of Mineral Processing, 29 (1990) 249-265 Elsevier Science Publishers B.V., Amsterdam

249

Simulation - the modem cost-effective way to solve crusher circuit processing problems

R.P. King Department of Metallurgy and Materials Engineering, University of Witwatersrand,

1 Jan SmutsAvenue, Johannesburg (South Africa)

(Received April 11, 1989; accepted after revision January 30, 1990)

ABSTRACT

King, R.P., 1990. Simulation- the modern cost-effective way to solve crusher circuit processing prob- lems. Int. J. Miner. Process., 29: 249-265.

Simulation is an effective technique for the improvement of crusher plant performance, and it is now used routinely by some crusher manufacturers for both plant design and trouble shooting. MOD- SIM is probably the most versatile ore dressing-plant simulator in general use in the mineral-process- ing industry today. A case study using MODSIM applied to the crusher circuit of a major uranium producer is described in this paper. This study demonstrates the effectiveness of simulation to im- prove plant performance when reliable and effective models of the unit operations are available.

The study was commissioned to investigate a 2000qon/h 4-stage crusher plant to identify a strategy to increase production at a finer product size. After an intensive technical audit on the plant, success- ful simulation was achieved for the existing operating conditions. The simulator was then used to identify the production bottlenecks and to establish plant modifications to meet the required produc- tion objectives in a cost-effective manner.

INTRODUCTION

Plant simulation techniques are becoming increasingly effective and there- fore more frequently used as tools to assess and improve plant performance. This is particularly so with crushing plants because the unit operations of crushing and screening can be described by reliable and accurate models. The study reported here was undertaken to establish cost-effective modifications to the fine crushing plant of Rossing Uranium Ltd., and the application of simulation to address some of the possibilities is described. The study was undertaken by a team consisting of personnel from Nordberg (Pty) Ltd., the Department of Metallurgy and Materials Engineering of the University of Witwatersrand, and Rossing Uranium Ltd.

The objectives of the study were: ( 1 ) to investigate and establish all process and operating parameters under the current operating conditions; (2) to es-

0301-7516/90/$03.50 © 1990-- Elsevier Science Publishers B.V.

Page 2: King's Simulation

250 R.P KING

tablish the complete mass balance and size distribution flowsheet for the pres- ent configuration and for the plant under various proposed alternative flow- sheet configurations; (3) to estimate energy, steel, and other cost reductions expected from any proposed plant modifications; and (4) to back up all pro- posals by basic engineering information and detailed flowsheet calculations.

Simulation effectively addresses the second and fourth of these objectives and provides the necessary information to permit the calculation of energy and other cost savings to meet the third of the objectives. Nordberg Inc. of Milwaukee, have pioneered this approach and their Circuit Analysis Program (CAP) is now in use worldwide for the development and analysis of crusher circuits. Significant improvements in worker proouctivity and the quality of flowsheet design have been reported (O'Bryan, 1987). The application of simulation to crusher flowsheet design has been well described by Mage- rowksi and Karra (1982) and there is no doubt that simulation techniques will play an ever-increasing role in the future. This study afforded the oppor- tunity to use both MODSIM and CAP within the context of a real major plant analysis. The application of MODSIM to this problem is discussed in detail in this paper. The first objective was addressed by undertaking a detailed technical audit on the plant which included the measurement and recording of all relevant engineering parameters, together with tonnages and size distri- butions of key process streams.

The most important plant improvement required from this study was the reduction of the final product size from 80% passing 10.5 mm as in the exist- ing circuit to 80% passing 7 mm.

DATA COLLECTION

The data collection was undertaken over a period of five days during which the plant was operated sunder conditions close to normal. Production was, however, interrupted to permit the necessary sampling to be undertaken. Great care was taken to ensure that all samples were representative of normal oper- ating conditions.

The key process variables measured were the tonnages and size distribu- tions in those streams that were diagnostic of the operation of each of the units in the plant. The size distributions and tonnages were measured by stop- ping the appropriate conveyor belts in the plant in the plant and carefully cutting 1-m sections from the belt load. Total mass and size distributions of these samples were determined in the usual way.

The existing plant flowsheet is shown in Fig. l, and the streams from which samples were taken are identified in Table I. When necessary, flow of material in portions of the plant was stoppexi to allow the sampling of only one of the parallel streams. For example, both east and west secondary crusher products

Page 3: King's Simulation

S I M U L A T I O N O F C R U S H E R C I R C U I T P R O C E S S I N G P R O B L E M S 251

,l::t

tm e~

:::t

~D

O

e~

O

O

e~

eL

Page 4: King's Simulation

252

TABLE I

Stream identification and sample point location

R,P. K I N G

Stream Identification Comments number

1 2 3 4 5

31 35 37 25 45 46 36 10 17 40

Coarse ore reclaim conveyor Secondary crusher feed - west Secondary crusher feed - east Secondary crusher product - west Secondary crusher product - east Tertiary screen under size - west Tertiary screen under size - east Tertiary crusher product - west Tertiary crusher product - east Quaternary crusher product Quaternary screen underflow Final mill-feed product Tertiary screen top-deck overflow Tertiary screen lower-deck overflow Quaternary screen overflow

No flow through east secondary No flow through west secondary No flow through east tertiary No flow through west tertiary

TABLE 1I

Important operating variables in existing plant

Unit Type Operating parameters

Secondary crushers

Tertiary screens

i ¢

Tertiary crusher

Quaternary crushers

Quaternary s ~ ' e e n s

7-ft standard Symons cone crushers

Double-deck polyurethane

7-ft short-head cone crushers

7-It short-head cone crushers

6 X single-deck polyurethane

Closed-side settings: west 49.1 ram; east 51.7 mm

Topdeck: 49-ram square opening 4 l -ram ribs 23.4% open area 5.486 m × 2 . 1 3 4 m 18 ° inclination

Lower deck: 15.2-mm square opening 10.3-ram ribs 25.01% open area 6.094 m × 2,134 m 18 ° inclination

Closed-side settings west:No. 1 13.1 mm, No. 2 11.9ram east: No. 3 13.7 ram, No. 4 12.6 mm

Closed-side settings No. 1 8.17 ram, No. 2 8.6 mm

20.0-mm square opening 11.5-mm ribs 27.7% open area 5.48 m X 2.0 mm 18 ° inclination

Page 5: King's Simulation

SIMULATION OF CRUSHER CIRCUIT PROCESSING PROBLEMS

TABLE III

Measured size distributions (percent passing indicated size)

2 5 3

Size ( r a m )

Stream Stream Stream 1 Stream Stream Stream Stream Stream Stream Stream Stream Stream 2 3 reeonstr. 4 5 31 35 37 25 45 46 36

from 2&3

150 88.3 81.7 85.2 100.0 . . . . . . . 100 76.1 61.1 69.1 100.0 99.6 . . . . . . . 70 66.2 51.6 59.4 91.9 94.8 . . . . . . . 63 63.5 48.6 56.6 87.9 90.9 . . . . . . . 50 57.3 44.6 51.4 75.5 78.0 - - 100.0 . . . . 37.5 48.1 38.0 43.4 59.6 64.3 - - 99.7 100.0 100.0 100.0 - 25 40.5 33.0 37.0 48.9 52.3 100.0 - 93.6 94.4 99.5 99.0 - 19 35.1 29.8 32.6 42.9 47.3 99.6 100.0 80.7 81.4 95.2 96.5 100.0 12.7 29.4 25.6 27.6 35.6 40.0 97.5 99.3 55.7 53.5 70.4 79.6 88.3 6.4 23.1 22.3 22.7 26.9 30.7 80.1 82.1 37.2 30.0 39.4 50.0 58.0 4.7 20.8 20.0 20.4 24.2 27.5 71.2 73.2 29.4 25.6 33.1 43.0 50.6

Flowrate 1071 933 2004 1150 970 366 366 856 724 677 1403 2270 (tons/ + 3 5 2 + 7 2 4

h) 718 1580

were sampled on the conveyor indicated as stream 35 in Fig. 1. Only one of the parallel crusher lines was operated while loading the belt for sampling.

Crusher gaps were determined by leading, and when necessary, gaps were adjusted to ensure uniform operation during the entire the entire audit pe- riod. Current drawn by the crusher motors under operating load, as well as under no-load conditions, was determined from existing plant instrumentation.

Screen dimensions and screen apertures were measured by direct observa- tion. Screen vibration amplitudes and motions were recorded for each screen. Samples of screen overflows were taken by the manual lunge method and are consequently less reliable than the belt samples taken.

Conveyor belt speeds were determined from measurements of the drive drum diameters and rotational speed.

The key operating variables that were measured are listed in Table II. The measured size distributions and tonnages are given in Table III.

S I M U L A T I O N O F T H E E X I S T I N G F L O W S H E E T

MODSIM is a modular simulator that can simulate any or dressing-plant flowsheet. Models are required for the description of the operation of each unit in the plant. For use within MODSIM, a model must be capable of ac- cepting as input the complete description of the feed to the unit and calculat- ing the nature of the product stream in detail sufficient for the needs of the simulation. The parameters that described the size and operating conditions

Page 6: King's Simulation

254 R P KING

of each unit must be supplied to the simulator. When simulating an existing plant, these operating parameters must match the values actually set up on the plant.

For the Rossing crusher circuit, models were required for the segregating action of the 1000-ton coarse ore bin, for the standard and short-head cone crushers, and for the double- and single-deck screens.

A model was developed to describe the measured segregation in the coarse ore bin because the segregation was too large to be ignored. Tow effects were modelled: the unequal discharge rates caused by variations in the operation of the two apron feeders under each discharge and the tendency of coarser particles to discharge preferentially to the east side stream because of segre- gation in the bin. The measured flowrates and size distributors in the prod- ucts permitted the development of a simple empirical model. The fine mate- rial (up to 10 mm) splits in direct proportion to the overall mass flows. Intermediate sizes ( 10 mm to 50 mm) favoured the west side to the extent of 20%. Larger sizes (over 50 mm) favoured the east side increasingly as size increased. The recovery to the west side decreased by 7.2% as the size in- creased by a factor of 2.

The model used for the cone crushers is based on the well-known classifi- cation and breakage-zone developed by Whiten at the Julius Kruttschnitt Mineral Research Centre (Whiten, 1973; Whiten et al., 1979 ). Karra ( 1982 ) has demonstrated that this model can be used to describe the operation of the short-head cone crusher as well as the standard cone crusher. The behaviour of the crusher is modelled through a classification function and a breakage function. The classification function defines the chance that a particle of a given size will actually be crushed during the nipping period of the crusher cycle. Thus the function C(x) is the fraction of material of size x that will be crushed during a nip. Material that is not crushed, is presumed to be dis- charged directly through the crusher into the product stream. Material that is crushed, produces an entire spectrum of particle sizes. The size distribution of the products of breakage is described by a breakage function B (x,y) which is defined to be the fraction of daughter particles smaller than size x that re- sult from the breakage of particles on size y. The daughter particles are them- selves subject to further classification to select those that will be broken fur- ther during subsequent crusher nips. The operation of the crusher can be completely described by the classification function and the breakage func- tion. The forms of these functions used are given by:

C(x) =0

=1 kd,

= 1 . 0

x~d~

dl <<. x <~ d2

x> d2

(1)

Page 7: King's Simulation

SIMULATION OF CRUSHER CIRCUIT PROCESSING PROBLEMS 255

and:

B ( x , y ) = G ( x / y ) "l + ( 1 - G ) ( x / y ) n2 (2)

The parameters in these functions must be related to the way the crusher is set up. In eq. 1, dl represents the smallest size that the crusher can nip and therefore break, while d2 represents the largest particle that can pass through the crusher during the fully open part of the cycle, dl and dE are affected largely by the closed-side setting of the machine. Whiten et al. (1979) and Karra (1982) have found a linear dependence between those variables. Careful ex- perimentation by Whiten et al. (1979) has revealed a weak inverse depen- dence of dE on the feed rate to the crusher, but our data were not sufficiently comprehensive to reveals this effect. We have accordingly modelled dl and dE by:

dl =0.653 CSS (3)

dE = a CSS (4)

The value of a was found to vary with crusher type, being close to 1.7 for the standard cone crushers used as secondaries, and 3.5 for the short-head cone crushers used in the tertiary and quaternary stages. A value of n = 1 was used for the standard cone crushers, and n = 3 for the short-head crushers.

The parameters in the breakage function eq. 2 are also machine-specific, n l and n2 were fixed at the values suggested by Whiten et al. (1979) for the standard cone crusher and by Karra ( 1982 ) for the short-head crusher. Thus:

n, = 0.5 standard

= 0.518 short-head

and:

n 2 = 4.5 standard

= 2.475 short-head

The parameter G represents the fractional production of fines from single- particle breakage events within the crusher. The value of this parameter is assumed to be a function of the machine type and of the ore, but is assumed to be independent of the crusher setting.

Values of a and G were found for each crusher in the circuit to match the measured size distributions in the streams that were sampled. No special pa- rameter estimation procedures were used. The parameters were simply varied until a good fit was obtained with the measured data. MODSIM does not offer an automatic parameter estimation mode although its companion pro- gram MICROSIMI~ does

The values used are given below:

Page 8: King's Simulation

256 R.P. KING

Secondary crushers West G = 0.15, a = 1.8 East G=0.20, a = l . 6

Tertiary crushers G=0.32, a=3 .5 Quaternary crushers G = 0.32, a = 3.5

These parameters are entirely reasonable, and the correspondence between the predicted size distributions and those measured can be seen in Fig. 2. The agreement can be considered to be very satisfactory, and the crusher model used can be regarded as reliable.

The model used to describe the operation of the screens in the plant was adapted from a model described by Karra ( 1979 ). This is a predictive model of screen behaviour and is based on the conventional description of screening behaviour through a set of capacity factors which depend on the tonnage and size distributions of the material fed to the screen and on the nature of the screen itself. Karra's model is based on considerable operating data and was developed to provide a description of screening behaviour that is as close as possible to conventional industrial practice for the design and assessment of screening performance. Consequently, Karra's model can be related directly to well-established industrial practice. The model can be evaluated using well- understood procedures. Karra was able to establish the parameters in the model using the accumulated experience of screen performance available within his company. The model is considered to be robust and reliable and is

' Z D' / 1 "'0 .-/" ,//, / '" / Z /

l

° ' ° .... : : : : : 4 ~, ,~ ..............

= ......... / / ' z * ~ > - - 1 o 4o ...... .,,<,, ~ v , . . ..... , ~ p / ' o . . l < _ _ _ = = , . . . , s c_ i < / ® / l i . + . . . . " ~ . ~ / ~> ~ _ _ s l , . , , , , ,,7 I

" X t ~ i ~ . . ....... * e : ~ v . . . . s= . , , , , 4

,..~..; ' . ~ , , , . ~ " + .............. st,. . , , , , I 20 ~ t , , -= '= ' ' - & ~ ~ s ~ ' e u a "]

~ ~ S l : r i e u 3 i

I 0 E] - - . - - S l P i l i 1

/ 0 . . . . . . . . I I

iO tOO t O 0 0

P a r t i c l e s i z e mms

Fig. 2. Simulated and measured size distributions in several streams in the plant. The stream numbers correspond to those in Fig. 1. The size distributions that were measured during the plant audit are shown by means of the plotted symbols. The correspondence between the measured and simulated size distributions is good.

Page 9: King's Simulation

SIMULATION OF CRUSHER CIRCUIT PROCESSING PROBLEMS 2 5 7

used in MODSIM without change. It was found to be entirely satisfactory for the present study and was very effective in allowing a useful characterization of the behaviour of the screens in the Rossing circuit.

A brief description of the model is given here, and those aspects of the model that can be specially exploited by MODSIM are highlighted. The model is based on the well-known procedure for assessing the capacity of a vibrating screen through the basic capacity factor, A, which defines the tonnage of un- dersize that a particular screen can transmit per unit screen surface area. This basic factor is increased or decreased depending on the nature of the feed and conditions on the screen. A number of capacity factors allow for the amount of oversize in the feed (factor B), the amount of half-size in the feed (factor C), the check location (factor D), wet or dry screening (factor E) , and ma- terial bulk density (factor F) . These factors all have a value of unity at the nominal standard operating condition and move down or up as the screen duty becomes more or less arduous.

Thus the model makes use of an index that defines the duty of the screen in its position in the flowsheet. This index which Karra designates as K is de- fined as:

tons of undersize in the feed/unit area of screen K - ABCDED ( 5 )

Karra gives formulas for the calculation of each of the factors A to F, and these are programmed into the MODSIM screen model. These formulas re- late the factors to the physical characteristics of the screen and to the nature of the feed. Thus Karra's model can be selected for use in a MODSIM simu- lation only if the physical characteristics of the screen have been completely specified. MODSIM supplies all information about the feed for every screen in the circuit.

Karra has found that the capacity of a screen is reduced if there is a consid- erable quantity of near-size material in the feed. He defines an additional near-size capacity factor Gc and calculates it from

X. ~o51, Gc =0.975(1 - 1--~/ (6)

where Xn is the percentage of near-size material in the feed. Thus the theoret- ical amount of undersize that can be transmitted by the screen is given by:

Th=A.B .C .D.E .F .G¢ X screen area

A screen will be well designed to handle its duty in the circuit if Th is approx- imately equal to the quantity of undersize in the feed.

In practice, not all of the undersize is transmitted because of various phys- ical factors that impair the efficiency of the screen. This effect is described by

Page 10: King's Simulation

258 kp KIN(;

the screen partition function. Several standard functional forms are available to describe this effect, and Karra uses the function:

Partition factor = 1 - exp [ ( - 0.693 (d/dso) 5.9 ] ( 7 )

The parameter that will determine the screening efficiency is dso. Values of dso smaller than the screen mesh size will lead to low efficiencies, and values ofdso greater than the mesh size give high efficiencies.

The actual dso achieved will depend primarily on the effective throughfall aperture of the wire mesh used on the screen. The throughfall aperture is in turn related to the actual mesh size, h, by:

h-r = ( h+ dw )cosO-dw (8)

where 0 is the inclination angle of the screen, and dw is the wire diameter. Karra's analysis of experimental data from an extensive test program pro- duced the following relationship for the prediction of dso:

dso Gc hv - K 0"148 (9)

A further refinement must be added to Karra's model before it can be used as a general simulation model for operating screens. Karra makes no allow- ance for the percent open area of the screen cloth used. His formula for the calculation of capacity A is based on industrial light-medium woven wire mesh. For other screen cloths and surfaces, A must be adjusted in proportion to the open area. The percent open area for l ight-medium wire mesh is related to the mesh size h by:

OA=21.5 logl0h+ 101.5 (with h in meters) (10)

Thus capacity A must be adjusted to:

A X actual % open area OA

The model provides a simulation of the actual performance of the screen in the circuit. This performance can be compared with the design capacity of the screen and the screen performance evaluated. In particular, the actual oper- ating efficiency can be calculated from:

Simulated efficiency = tonnage in underflow tonnage of undersize in feed

The effective utilization of the screen area can be calculated from:

Area utilization factor = AUF

tonnage of undersize in feed - theoretical ability of the screen to pass undersize

Page 11: King's Simulation

SIMULATION OF CRUSHER CIRCUIT PROCESSING PROBLEMS 259

An AUF equal to uni ty indicates that the screen capacity is exactly balanced to the required duty. AUF ~< 1 indicates that the screen is under loaded, while AUF >I 1 indicates that the screen is overloaded.

100

90

80

70

m 60

O. 50

4o L

~- ao

20

0

"/7 ~ ' "/'® , / / / / I i i 1 ° i °

i

/ / /o A

i~ i/° /l /

// /° • /

o® / ¢

. . . . . . ~ _ _ _ . . 4 . . . . . . . , 1 0 100

Particle size mms

A - - __St rsam 4 0

O - - __Stream 17

_ _ _ _ _ S t r e a l l t t 0

tO00

Fig. 3. The simulated and measured size distributions in the overflow streams from the top deck of a tertiary screen (stream 10) the lower deck of a tertiary screen (stream 17 ) and the overflow from the quaternary screens.

TABLE IV

Simulated ~rformance of crushers in the existing flowsheet

Unit Design Simulated Installed Measured Simulated tonnage tonnage net power net power net power

(kW) (kW) (kW)

Secondary crushers: West 1000 1069 171 126 130 East 1000 935 158 136 167

Te~iarycrushers: West l 445 354 151 108 102 West2 440 354 108 113 East l 464 289 145 86 76 East2 450 289 96 84

Quaternarycrushers: N o . l 390 235 158 34 47 No. 2 396 235 148 80 42

Page 12: King's Simulation

260

TABLE V

Simulated performance of screens in the existing flowsheet

RP. KING

Unit Simulated Simulated Calculated Simulated tonnage tonnage to dso efficiency (tons/h) underflow (mm) (%)

(tons/h)

Area utilization factor

Tertiary screens: West top deck 534 317 34.9 82.7 1.60

lower deck 317 180 12.4 94.0 1.14 East top deck 468 293 35.6 84.0 1.40

lower deck 293 179 12.8 95.8 1.04

Quaternary screens 293 214 15.8 86.8 0.94

A particular advantage of the Karra screen model is that no free parameters are required to be estimated from operating data. The adequacy of the model for the screens in the Rossing plant can be judged by comparing the measured and predicted size distributions in the screen overflow streams as shown in Fig. 3. With the exception of the overflow from the lower deck of the second- ary screens, the agreement is remarkably good.

A summary of the simulated behaviour of all of the units in the flowsheet is given in Tables IV and V.

ASSESSMENT OF EXISTING FLOWSHEET OPERATION

The models used for the crushers and screens in the simulation required very little tuning to achieve a very good match between the simulator output and the observed behaviour of the plant. The models provide he necessary information that is required to diagnose the plant bottlenecks and to suggest modifications to enable the plant to meet its operating objective of 80% pass- ing 7 mm in the final product. The simulation of the existing flowsheet pre- dicted an 80% passing size of 10.4 mm against the measured value of 10.5 mm.

The summarized simulator data in Tables IV and V identify the important operating characteristics immediately. The power drawn by each crusher was calculated by the simulator using the known impact work index of the ore (9kWh/ton) using formula:

Power= TW, ( 1 IP~so - 1 IF~so)

where T= tonnage handled by the crusher, IV~ the impact work index,/'so the 80% passing size in the product from the crusher and Fso the ~ passing size in the feed to the crusher. It is evident from Table IV that all the crushers in

Page 13: King's Simulation

SIMULATION OF CRUSHER CIRCUIT PROCESSING PROBLEMS 261

the plant are operating at or within their installed design capacity. The sec- ondary crushers are working just within capacity, the tertiaries on the east side at about 65% of capacity.

Table V shows that the tertiary screens are overloaded, particularly on the upper decks. The calculated area utilization factors are 1.6 and 1.4 on the top decks and this indicates that the screens are being required to transmit 60% more material than their design capacity. Screening efficiencies are relatively low. The lower decks are operating in a slightly overloaded condition, but in spite of this, the simulated efficiency is quite high. However, the measured size distribution of the lower deck overflow, as shown in Fig. 3, indicates some carry over of undersize material, and that would be consistent with the sim- ulated overloaded conditions.

The reason for the overloading of the front end of the circuit is not difficult to see. The plant feed on the coarse ore reclaim conveyor has a large quantity of fine material with aproximately 50% smaller than the closed-side setting of the secondary crushers. All of this material must be handled by the secondary crushers and the tertiary screens before leaving the plant in the tertiary screen underflow. All material that leaves in the tertiary screen underflow reduces the load on the tertiary crushers and the quaternary circuit. The calculated load on the quaternary circuit was found to be 1286 tons/h with a circulating load of 37%.

Two strategies for the improvement of plant performance immediately sug- gest themselves.

S T R A T E G I E S F O R I M P R O V E M E N T

Strategy A. Increase the circulating load in the quaternary circuit and uti- lize the existing quaternary screening and crushing capacity more effectively to produce a fine product.

Strategy B. Install secondary screens ahead of the secondary crushers to remove final product-size material before it enters the crusher circuit and uti- lize the released crushing and screening capacity to produce a finer product.

Strategy A

The circulating load in the quaternary circuit is increased if the mesh size of the quaternary screens is decreased. Even modest increases in the circulat- ing load lead to an overload condition on the quaternary screens. This effect is shown by the simulator, and the pertinent data are summarized in Table VI in lines 1, 2 and 3, for simulations of the existing flowsheet with the qua- ternary screen mesh reduced from 20 mm to 17.5 mm and 15.2 ram. The latter condition increases the circulating load from 37% to 89%, and the qua- ternary screens and quaternary crushers become overloaded. It is necessary

Page 14: King's Simulation

262

"FABLE VI

Simulation of effect of finer crushing in the quaternary circuit

R P KIN(I

Simulation Quaternary Quaternary Circulating % - 7.41 Nr, screen crusher CSS load mm in final

aperture (ram) (%) product (mm)

Comments

1.1 20 8.17 and 8.6 37 63 2 17.5 8.17 and 8.6 56 68 3 15.2 8.17 and 8.6 89 74

4 20 6.5 28 66 5 17 6.5 40 72 6 15.2 6.5 48 75 7 15.2 6.5 45 74

Screens are overloaded Screens and crushers are overloaded Screens are overloaded No overloads Screens are overloaded No overloads. Quaternary screens increased from 6 to 8

*~Existing flowsheet base case.

to reduce the closed-side setting of the quaternary crushers ot keep the qua- ternary circuit in balance, and the effect of this can be seen in rows 4, 5 and 6 of Table VI. To relieve the overload on the quaternary screens, a further two screens can be installed in the quaternary circuit, and a product having 74% of material ~< 7.4 mm is produced by the circuit.

Strategy B

A modified circuit having additional secondary screening capacity installed is shown in the flowsheet in Fig. 4. These screens remove product-size mate- rial, allowing the secondary crusher to be set finer. The settings chosen for each unit in the modified plant are summarized in Table VII, and the simu- lator confirmed that no unit was overloaded in the circuit.

MODSIM provides detailed operating information on the operation of the screens so that it is easy to select screens appropriate to the duty to be per- formed. In this case two pairs o f double-deck screens 6.1 m × 2.13 m are re- quired, the apertures and other information are specified in Table VII. The apertures were chosen to balance the load on each deck so that the AUF is approximately equal to 1.0 for each.

The addition of the secondary screens reduced the load on the quaternary circuit from 1286 to 1186 tons/h, so that a product having 75%~< 7.4 mm can be produced without additional screening capacity in the quaternary circuit.

Page 15: King's Simulation

S I M U L A T I O N O F C R U S H E R C I R C U I T P R O C E S S I N G P R O B L E M S 263

.7,

b -

'X~ 0¢

,~ e'~

O O *-'

a a ~ ¢ a " ~

o

Page 16: King's Simulation

264

TABLE VII

Settings o f operat ing pa ramete r s in modi f ied flowsheet

RIL KING

Uni t Type

Secondary Double-deck po lyure thane screens

Secondary 7-ft standard Symons cone crushers c rushers

Tertiary screens

Double-deck polyurethane

Tertiary 7-fi short-head crushers

Topdeck: 7 0 - m m square opening 5 0 - m m ribs 6.1 m × 2 . 1 3 m 18 ° incl inat ion

Lower deck: 16-mm square opening 12-mm ribs 6.1 m x 2 . 1 3 m 18 ° incl inat ion

Closed-side setting: 35 m m

Topdeck: 15-mm square opening 10-mm ribs 6.1 m X 2 . 1 3 m 18 ° inclination

Lower deck: 12 -mm square opening 8 - m m ribs 6.1 m X 2 . 1 3 m 18 o inclination

Closed-side setting: 8 m m

Quaternary 6b'ingle-deck polyurethane 15.2-mm square opening screens 10 .3-mm ribs

5.46 m × 2.0 m 18 ° inclination

Quaternary 7-ft shor t -head crushers

Closed-side setting: 6.5 m m

D I S C U S S I O N

The main observation to be made from the simulations is that the product size distribution is determined primarily by the set-up of the quaternary crushing circuit. It is necessary to utilize both screening and crusher capacity in this section of the plant in an optimal fashion to achieve the best product. The simulations indicate that this can be done by appropriate choice of mesh size on the quaternary screens and settings in the quaternary crushers. The introduction of additional secondary screens has apparently very little poten- tial for increasing the capacity of the plant. The reduction in load on the qua- ternary circuit from 1286 to 1186 tons/h cannot be regarded as a very attrac- tive incentive for the introduction of four large double-deck screens. Improved

Page 17: King's Simulation

SIMULATION OF CRUSHER CIRCUIT PROCESSING PROBLEMS 265

plant operat ion should be sought first by exploiting the capacity of the qua- ternary circuit to the greatest possible extent before addit ional secondary screening capacity can be considered.

Space does not permit a comprehensive presentat ion of all the data ob- tained from MODSIM, which is very voluminous. A complete set o f data from the study is available in King, 1987b and c.

ACKNOWLEDGEMENTS

The opportuni ty to participate in this important plant s tudy was made pos- sible by Rossing Uran ium and by Nordberg Manufacturing Company. The expert assistance and co-operation of these companies is greatly appreciated. Continuing development of modelling and simulation techniques has been made possible by a grant from the Foundat ion for Research Development .

REFERENCES

Karra, V.K., 1979. Development of a model for predicting the screening performance of a vi- brating screen. CIM Bull., April 1979, pp. 167-171.

Karra, V.K., 1982. A process performance model for cone crushers. Proc. 14th Int. Mining Congr., Toronto, III: 6.1-6.14.

King, R.P., 1986. A User's Guide to MODSIM. University of the Witwatersrand, Department of Metallurgy and Materials Engineering, Johannesburg, Rept. GEN?l/86, 66 pp.

King, R.P., 1987a. MODSIM: A Modular Method for the Design, Balancing and Simulation of Ore Dressing Plant Flowsheets. University of the Witwatersrand, Department of Metallurgy and Materials Engineering, Johannesburg, Rept. GEN/2/83, 77 pp.

King, R.P., 1987b. The Fine Crushing Circuit of Rossing Uranium Ltd. Simulation of the Ex- isting Flowsheet. University of the Witwatersrand, Department of Metallurgy and Materials Engineering, Johannesburg, Rept. GEN/8/87, 262 pp.

King, R.P., 1987c. The Fine Crushing Circuit of Rossing Uranium Ltd. Simulation of Modified Flowsheets. University of the Witwatersrand, Department of Metallurgy and Materials En- gineering, Johannesburg, Rept., GEN/13/87, 262 pp.

Magerowski, A.J. and Karra, V.K., 1982. Computer-aided crushing circuit design. In: A.L. Mu- lar and G.V. Jorgensen (Editors), Design and Installation of Comminution Circuits. Society of Mining Engineers, New York, N.Y., p. 288.

O'Bryan, K., 1987. Crushing flowsheet simulation: increased productivity and improved flow- sheet design. APCOM 87. Proc. 20th Int. Symp. Application of Computers and Mathematics in the Mineral Industries, Vol. 2, Johannesburg, SAIMM, pp. 167-178.

Stange Wayne, 1989. Strategies for more flexible sequential-modular simulation. In: A. Weiss (Editor), Proc. 21st Int. Symp. Application of Computers and Operations Research in the Minerals Industry. Society of Mining Engineers, New York, N.Y., pp. 811-825.