A Simulator for Crushing- Screening Plants

George Grandy Director Metallurgy - Mineral Processing Technology Kvaerner - Metals E&C 1265 7 Alcosta Blvd. Sun Ramon, CA 94583 USA Cristian Araya Process Specialist, TechnologyDepartment Kvaerner - Metals E& C Nva. Tajamar 481, p.19 Sun tiago, Chile ABSTRACT Kvaerner experience is being used to develop software to design or simulate crushing and screening plants. The model techniques discussed are presently being used for open-circuit plants and are being extended to closed circuit plants. The program contains a library of routines, each representing a piece of equipment or unit operation, which are assembled into one of several open-circuit plant configurations. Routines include gyratory primary crushers; standard secondary and shorthead tertiary cone crushers; single, double or triple deck screens; and stockpiles, feed bins and belt conveyors. The program uses steady-state models to simulate the performance of crushers and screens that have small residence times and quickly reach equilibrium when operating conditions and/or parameters are changed. The dynamic aspect of the total circuit model concerns the storage units within the circuit, i.e. stockpiles and bins, and, in the closed circuit configuration, circulating loads and size distributions. In many cases, just vendor's catalog data are necessary for a pre-feasibility simulation; in other cases, actual plant test data are required to obtain more accurate simulations. The entire plant is shown in a single graphic screen so the user can see all the relevant information he might want, and soft-keys are provided for the user to make online parameter changes and observe the effects on storage elements, equipment feed rates and the final product size distributions. The model also reports equipment utilization in percentage, power consumption, and other related operating information, and soon, will be able to report estimated capital and operating costs.

Proceedings of Copper 99Cobre 99 International Environment Conference Volume 11-Mineral Pmcessing/Enviroment, Healthmd Safe1 Edited by B.A. Hancock and M.R.L. Pon The Minerals, Metals & Materials Society, 1999

162

VOLUME II

Engineering firms are Wuently asked to look at many alternate crushing plant circuit configuratiions and incorporate alternate equipment selections. For each of these configurations, the operation of the proposed circuit is required, as well as the ultimate circuit capacity and the predicted size distribution of the final products. Tradionally, these alternatives are investigated as "paper' or trade-off studies that include the engineer's calculations and input h m the the various manufactures that supply crushing plant equipment. This procedure is time consuming, expensive and often inadequate when screening many plant design alternatives. Most equipment manufixtures have proprietary simulation sofware that predicts steady-state circuit performance that is- valuable in the design of crushing circuits. However, these simulations are not always compatible- when comparing, different suppliers. They also do not provide the engineer with an opportunity ,to compare circuit performances when mixing .alternativeequipment within a proposed circuit. In addition, they do not cover the wide range of operating conditions that may occur in the proposed plant and they are not capable ofsimulating dynamicconditions. To be useful for design, the models used must be capable of handling a wide range of size distributions, h m coarser ROM feed to final crusher products, including the production of minus 100 mesh material. This is very useful in the design or optimization of crushing plants for heap leaching operations, in which control of h e r particles has an important role in the metal recovery and fluid (both liquid and air) percolation. In fact, the driving force to develop the crushing plant simulator was the need to develop a h n t end module for the dynamic leach and SX/EW simulator as discussed by Araya ( 1). It is for these reasons that Kvaerner decided to develop a simulator that is both simple and reliable, and capable of handling many brands of equipment. A user friendly interface is required so that the designer can make on-line parameter changes and quickly observe changes in the results, including final product size distributions. The simulator should predict equipment utilization, power consumption and other related operating parameters. Also, the simulator should be useful for existing operations where changes can be investigated before implementing them into the plant. EQUIPMENT MODELS There are two unit operations that require mathematical models, i:e. crushers and screens. These models mustibe general enough to make the simulator fit a variety of plant designs, circuit configurations and equipment selections. For the crusher model, it was

MINERAL PROCESSINGIENVIRONMENT, HEALTH AND SAFETY

decided to use only one general model to describe the behavior of all.crushers of interest, i.e. gyratory and cone crushers. The screen model is based on defining a simple. classification curve that partitions the screen feed into oversize and undersize product streams. Crusher performance (ke. power and. utilization) is obtained h m the crusher model and plant feed rate. The screen performance (i.e. efficiency and bed depth) is back calculated:with formulas used for sizing and selecting screens.

Ci-usher Model
The crusher model selected is. based on population balance midel concepts for ball mill grinding. It is proposed that the .same modeling concepts can be applied to crushers when the parameters are defined to give ,the characteristic crusher product size .distributions. The restrictive (or simp1e)solution to the general model for batch grinding has been discussed several times in the literature, including Austin, et ,al. (2,. 170) and Yap, et al. (3, 198). A typical form is as follows: (1 - Pi) = (1 - Fi) exp( -Si E) where Fi Pi Si E
- .. Feed cumulative percent passing size Xi Product cumulative percent passing size Xi Crushing function for size Xi Specific crusher energy for the ore being processed, kWh/t Particle size, microns

.-..

.~

Xi

In the above equation it is further assumed that the-feed material to the ciusher.

has been scalped, i.e. the fines ,havebeen removed. For secondary and tertiary crushers
thisis generally the case a s the crusher is fed screen oversize material. However, in the case of a primary crusher, feed scalping is generally not part of the flowsheet. For this case a hypothetical scalping of the feed at the open side .setting (OSS) is calculated .and model equation applied to the resulting oversize material. Final primary crusher product is detennined,bycombiningthe hypothetica1,screenundersize with the crusher product. Data h m m operating crushers was used to establish a .mathematical equation to define. the: crushing function. The form was established1 after noting that the crusher product resulting h m screening a scalped feed is of the. form shown in Figure 1. The equation. selected reflects the. fact that the crusherproduct has a steeper slope in the coarse sizes and a finer slope in the h e r sizes. It is a convex linear combination of two terms that define ,the slopes in the coarse and h e ranges o f a typical crusher product. The.parameterranges are indicated afterzthe definition of the.terms.

VOLUME 11

where

K 1
Wi
a

a
Xo

Calibration factor for ore breakage (3 -30) Crushing work index, metric Fraction contribution from the fines (0.01 - 0.25) Slope of the fine range of the size distribution (0.3 - 0.8) Slope of the coarse range of the size distribution (1 - 5) Crusher setting, m i c r o n s

10000

Particle:Size, microns Figure 1 - Typical Secondary and Tertiary Crusher Products The crusher setting (Xo) is equal to the open side setting (OSS) for primary crushers and two times the closed side setting (CSS) for secondary and tertiary cone crone crushers. The shape of size distribution is altered by adjusting the parameters a, a and p in the crusher function, and the fineness of crush is modified with the breakage calibration factor, K1. The crusher specific energy term (E) is based on the analysis of catalog crusher data andlor field performance of specific crushers. The absolute value depends on the crusher manufacture, model, eccentric throw for primary crushers and crusher bowl selection for cone crushers. The mathematical relationship developed is as follows:

E = K2(Wi I 10) (Gx,/ Mx )

(3)

MINERAL PROCESSING/ENVIRONMENT, HEALTH AND SAFETY

165

where K2 Wi Mx Gx

Calibration factor for specific energy Crushing work index, metric Maximum crushing rate, t/h Average operating,crusherpower at Mx, Kw

The energy calibration factor, K2, depends on the crusher type and is adjusted together with the breakage calibration factor, K1,to obtain the target specific energy and product size distribution. The overall aim is to obtain a primary crusher product that is 95-98 percent passing the open side setting (OSS), a secondary cone crusher product that is 60-70 percent passing the closed 'side setting (CSS) and a tertiary cone crusher product that is 70-80 percent passing the closed side setting. For most applications the K2 factor is equal to 1.0, but may vary between 0.8-1.0 depending on the selection of the average operating crusher power, Gx. The crusher average power (Gx) is best established from field data, but it can also be estimated from catalog data by taking 80-90 percent of the connected motor power. Flavel (4, 373) stated that the power varies from 50-60 percent for secondary crushers without automatic control to above 90 percent for crushers with automatic feed and setting control. Tertiary crushers vary h m 80 percent to above 95 percent depending on the degree of control. Crusher maximum rate (Mx) is either taken h m field measurements or catalog data. A review of catalog data indicates that the following linear relationship is useful for primary, secondary and tertiary crushers. Mx = where
X

x Vx = x [

b (crusher setting) + c ]

(4)

-. -. -

vx

Material bulk density, tlm3 Maximum crushing volumetric rate, m3 /h

The crusher setting depends on the type of crusher and/or crusher configuration. For primary crushers, the open side setting (OSS) is used; however, for secondary and tertiary crushers, the closed side setting (CSS) is applied. The relationship must be determined for each crusher application since the linear-equation parameters b and c depend on the crusher manufacture, crusher size, eccentric throw (primary crushers) andlor the crusher bowl (cone crushers) selected. In addition, the type of material being crushed and, in some cases, the feed-size distribution influences the maximum crushing rate as discussed by Flavel (3, 365). Catalog data is generally conservative and there is justification for revising the stated crusher capacity upward by multiplying the equation with a constant factor.

166

VOLUME II

The crusher power draw (G, kW) is obtained by multiplying the crusher feed rate by the specific crushing energy (E) and the crusher power utilization is obtained by dividing the power draw by the maximum average power (Gx). Similarly, the crusher capacity utilization is obtained by dividing the crusher feed rate by the maximum crushing rate (Mx). The work index (Wi) appears in the both the crusher function (Si, denominator) and the specific crushing energy ( E ,numerator), but has no affect on the product size Table 1 - Crushing Plant Measured Data

MINERAL PROCESSING/ENVIRONMENT, HEALTH ANDSAFETY

distribution. This is by design in the model. By incorporating the work index in this manner the crusher power increases or decreases with ore hardness without affecting the size distribution of the crusher product. To demonstrate the validity of the crusher model, the data shown in Table 1 was used. The data was collected from a crushing plant with the secondary crusher operating in open circuit on scalping screen oversize and the tertiary crusher operating in closed circuit with screens. The ore is oxidized and secondary material mined from a porphyry copper deposit. The data was used to estimate most of the parameters in the crusher model. The specific energy, E, was calculated directly from the feed rate and measured crusher power. To estimate the parameters in the in the crushing function, Si, the function SiE was calculated using Equation 1 and the size distribution data. The predicted SiE function was calculated using Equation 2 and the specific energy, E. The Solver routine in Microsoft Excel was used to estimate the parameters ( a, a,P and K1 ) by minimizing a cost function defined as follows:

[ (SiE measured - SiE predicted) / SiE predicted j2 ,for all i cost = C

(5)

Figure 2 shows a comparison of the predicted' and measured SiE function for the data in Table 1 and Figure 3 shows the comparison of the simulated and measured s u e distribution. These two figures demonstrate the ability of the model to simulate typical crusher s u e distributions using Equations 1 and 2. In these calculations the crushers were considered independent of each other. However, if the parameters in Equation 2 are considered.fundamental .propertiesof the ore, only-one set of parameters is possible. The same calc~lationswere.~ade but with a combinedcost function and the results presented in Figures 4 and 5. With this combined condition the simulation deteriorates, but it is still very good. The parameters calculated for the three optimizing conditions are summarized in Table 2. To predict the other parameters in the model it is necessary to go to the published data for the Hydrocone crusher. Figure 6 presents the capacity curves for the mediurncoarse and medium-fine liner bowls at the maximum of the capacity range. From these curves, the maximum crushing rate, Mx, at the closed side setting is calculated. The crusher power, Gx, at the maximum crushing rate is determined after h t assuming that the calibration parameter K 2 is equal to 1.00. The calculations indicate that the crusher power at Mx is 77.6% of the motor power for the secondary crusher and 93.2% for the tertiary crusher. On the other hand, the measured power is 52.3% of the comected power for the secondary crusher and 67.6% for the tertiary crusher. This indicates that both crusher powers are underutilized by approximately 35%, i.e. the feed rate to the crushing plant could be increased by that amount. The feed rate to Mx ratio indicates the same potential increase in crusher capacity.

1'68

VOLUME II

loo00

20000

30000

40000

50000

60000

P a r t i c l e Size, nlifmm
Figure 2 - Actual'and'Simulated. SiE Function with Separate Crusher Analysis

10000'

P a r t i c l e Size, microns
Figure 3:- Actual and Simulated Size Distributions with Separate Crushers Analysis

MINERAL PROCESSING/ENVIRONMENT,HEALTH AND SAFETY

loo00

25lmO

30000

40000

50000

60000

Particle Size, microns Figure 4 - Actual and Simulated SiE Function with &parate Crusher Analysis

1 m

ParticleiShe, microns Figure 5 - Actual and Simulated Size Distributions with .Separate:CrushersAnalysis

VOLUME 1 1

Closed Side Setting, mm

Figure 6 - Crusher Capacity, M x

Table 2 -Crushing Plant Calculated Information

MINERAL PROCESSING/ENVIRONMENT, HEALTH AND SAFETY

Screen Model The screen model uses the partition curve concepts for classifiers. A plot of the partition curve is also known as a Tromp curve. In the continuous size form it is defined .as follows: o(x) =.C(x) f(x) ,GI where C(x) Rx) O(X)
= = =

Fraction of size x to x + dx sent to the coarse stream Weight of screen feed material in size interval x to x + dx Weight of screen oversize material in size interval x to x + dx

For screens it is possibleto define the following conditions fora simple partition c w e , C(x):

where
Y
Z

Rr

Effective screen opening size, microns Lower size limit for "near size" bypass, microns Fines bypass below size z, microns

The lower size.limit and.the fines .bypass factors are established to give a .-get screen efficiency value, asdefined below:

Err =

Jvercent undersize in feed that actuallv vasses) (percent undersize in feed that should pass)

(6)

In the model the feed rate, size distribution, screen dimensions and screen cloth openings are defined. This information is used to determine the performance of the screen, i.e. screening efficiency and bed depth. The approximate efficiency is calculated with the screen area formula as defined by the Vibrating Screen Manufactures Association (VSMA)(5). The formula and calculation procedure is not presented here and the reader is referred to the reference for the details. The procedure described is the same form as used by most screen manufactures; however, $thevalues selected for the various "factors" (i.e. basic capacity, oversize, halfsize, deck location, open area, slot shape, efficiency, etc) are not universally the same. The VSMA values for the "factors" are in table form and these were converted to equations to simplify the programming.
It was determined that the following definitions for the lower size limit and fines bypass parameters give calculated screen efficiencies near the VSMA efficiency.

172

VOLUME II

However, it should be noted that the efficiency as calculated by Equation 6 depends on the feed size distribution as well as the definition of the partition curve.(Equation5).

For size discretized systems as used here the Ci term for each size interval is the geometric mean of the partition values calculated at the top and bottom of the size interval. The size intervals are generally selected using a square root of two criteria

SIMULATOR
Initially the computer and screen models were used to develop steady-state crushing plant simulations in a spreadsheet. Open circuit configurations are easy to program, but closed circuit configurationsrequire macros to calculate the recycle streams when circular references are eliminated. In the latter case, the recycle streams are "tom open" and an iterative copy and replace macro written to "recycle" products back to the head of the circuit. It has been found that the copy-replace routine converges in about 10 cycles, which is fast enough to eliminate the need to write more complicate search routines in order to speed up the convergence. While the spreadsheet simulation approach is sufficient for general circuit design and equipment selection, it is not convenient to use when the output must be linked to other simulation packages. On-line changing of input parameters is awkward and observing the results is not convenient. For these reasons it is better to develop a simulator in a higher level language, i.e. quick basic or visual' basic. The initial effort is more, but the resulting simulator is simple and easier for the operator to make changes without having to understand the underlying principles.

Simulator Input
Data is input to the simulator in one of two ways. Data that is. considered setup information (i.e. not likely to be changed after the original configuration)is entered into a table. This information includes the crusher selection, screen size (L x W), conveyor sizes and ore storage sizes. Other data that can be changed 'bbdynamically" are.input on the graphic -screen. representing the flowsheet schematic using function keys (soft-keys) to incrementally increasedor decrease:the settings; The input data, reference data and parameters for calculation routines, .and output data are.all handled in the simulatoras vectors.of information. In this way all flow rates, are stored as a matrix of bulk densities.and size distribution input data and~calculations

MINERAL PROCESSING/ENVIRONMENT, HEALTH AND SAFETY

173

information for all streams. Similarly, parameter data for crushers is stored in a matrix with all the elements defined in Equations 2 , 3 and 4 (i.e. a, b, c, a,P, K1, Kz, Mx,and Gx). Numerous crusher types, sizes and configuration can be stored in the matrix to be used by the program when selected by the operator. The screen partition curves are also stored in matrix format, with a vector of parameters for each screen in the plant. The other input data includes material properties (e.g. Wi), feed rates, crusher selection and settings (OSS and CSS), screen dimensions, number of screen decks, and' screen cloth opening size, slot type and percent open area Conveyor information can also be input to allow the program [to calculate conveyor loading and power demand. Maximum capacities of storage elements are included for completeness. In the future it is planned to include capital and operating cost data for each unit operation.

Calculations
Several calculation subroutines are included, one for the crusher, another for the. VSMA screen calculations and another for the screen classification. Other subroutines are available to interpolate the oversize and ,halfsize information required by the VSMA routine when the screen opening size is not the same as the d e h e d size intervals in the size distribution .matrix. Some calculations are made directly and stored in. the appropriate database.

Simulator Output and Control

As stated previously, function keys are available to incrementally increase or decrease certain variables, including the feed rates, crusher settings (OSS and CSS) and screen openings. A counter representing a time interval (say, one hour) is provided. AU calculations are performed for each interval and displayed on the screen within a few seconds. When changing an online parameter (e.g. feed rate) the operator immediately sees short-term effects on the process equipment and the longer-term effects on storage elements and size distributions. In this way the operation is pseudo dynamic. In the closed circuit configuration, circulating loads will be monitored.

A typical screen output is shown in Figure 2, where it is noted that the graphic output is similar to a process control output screen for the same plant. To make a change, the operator highlights the selected equipment or element, followed by manipulation of the appropriate function keys to increase or decrease an input value or setpoint. After making a change the program continues to update the information at the

,1174

VOLUME 11

selected time interval d e h e d for the counter. Because the p r o m updates every second or so (depends on computer speed) and the counter time interval can be set at 1 hour, the operator is able to rapidly observe the plant behavior over a longer time period, even days, within a few minutes. After making numerous changes the operator can reset the program to the default conditions and start another change. Also, the infomation can be recorded and a trend line produced.

Figure 2 - Output Screen for Large Open-Circuit Crushing Plant

MINERAL PROCESSlNG/ENVIRONMENT, HEALTH AND SAFETY

175

The simulator contains a library of routines, each representing a piece of equipment or unit operation, which are assembled into one of several open-circuit plant configurations. Routines includes gyratory primary crushers; standard secondary and shorthead' tertiary cone crushers; single, double or triple deck screens; and stockpiles, feed bins and belt conveyors. The program uses steady-state models to simulate the performance of crushers and screens which have small residence times and quickly reach equilibrium when operating conditions andlor parameters are changed. The dynamic aspect of the total circuit model concerns the storage units within the circuit, i.e. stockpiles and bins, and, in the closed circuit configuration, circulating loads and size distributions. In many cases, just vendor's catalog data are necessary for a pre-feasibility simulation; in other cases, actual plant test data are required to obtain more accurate simulations. The entire plant is shown in a single graphic screen so the user can see all the relevant information, and soft-keys are provided for the user to make on-line parameter changes and observe the effects on storage elements, equipment feed rates and the final product size distributions. The model also reports equipment utilization in percentage, power consumption, and other related operating information, and soon, will be able to report estimated capital and operating costs.

REFERENCES
1. 2. Cristian Araya, A Dvnamic Simulator for LM-SX-EW Plants, Proceedings of Copper 99 -Cobre 99,1999. L.G. Austin, R.R. Klimpel and P.T. Luckie, Process Engineering of Size Reduction: Ball Milling, Society of Mining Engineers of AIME, New York, NY, USA, 1984,165-180. R.F. Yap, J.L.. Sepu1veda:andR. Jawequi, "Detemination~of the Bond Work Index Using an Ordinary Laboratory Ball Mill", Design and Installation of Comminution.Circuits,k L Mular and G. V. Jergensen,H, Eds., Society of .MiningEngineers of AJME, New York, NY,USA, 1982,176-203. M.D. Flavel, "Selection and Sizing of Crushers", Design and Installation of C o m m i n u ~A.L , Mular and G. V. JergenseqII, Eds., Society of Mining Engineers of AIME, New York, NY, USA, 1982,343-392.
'

3.

4.

5.

VSMA Vibrating Screen Handbook. Vibrating ScreenManufacturesAssociation, Stanford, Connecticut, USA, 1980, Chapter 5.