About the BallParam_Open Spreadsheet ...

Scope : The BallParam_Open spreadsheet was designed to be used - in conjunction with the Ball Mill Grinding Simulators BallSim_Direct and BallSim_Reverse - in the "tuning" of such simulators to any specific actual grinding system, via the estimation of the various Model Parameters that characterize the grindability of any given ore. In other words, the attached spreadsheets provide an effective algorithm to search for the set of parameter values that best approximate the model response to the actual experimental measurements available ( obtained from Continuous, Pilot or Full Plant Scale Mills), based on a typical non-linear, least-squares criterion.

Theoretical Framework : The reference ball mill model is described with further details in the About ... worksheets of the BallSim_Direct or BallSim_Reverse files. Such model is based on the so called Modern Theory of Comminution. This theory introduced two new sets of parameters : the Selection Function S and the Breakage Function B. The first set also referred to as Grindability relates to the grinding kinetics of each independent particle and the second set also referred to as Distribution of Primary Fragments characterizes the size distribution of the fragments produced as a result of breakage events. The size-dependence of the Selection and Breakage Parameters is represented by the following relationships : - For the Selection Function : SiE = a0 (di*)a1 / [ 1 + (di* / dcrit)a2 ] with : di* = (di di+1)0.5 = average particle size of the i-th fraction. An expanded form of this expression - also available in this simulation routine - is given by : SiE = [1/(1+a02/a01)] { a01 (di*)a11 / [ 1 + (di* / dcrit)a2 ] + a02 (di*)a12 } - For the Breakage Function : Bij = b0 (di/dj+1)b1 + (1- b0) (di/dj+1)b2 (2) (1e) (1)

An expanded form of this expression - also available in this estimation routine - is obtained by replacing b0 in Equation 2 by : b0j = b00(dj+1/1000) -b01 , never > 1 (2e)

These expanded forms for Si and Bij are intended to provide the model greater descriptive flexibility when tuning the model to actual specific grinding systems but, since they increment the total number of ore characteristic parameters to be estimated, its use should be avoided whenever possible . In any case, the proposed expanded forms reduce to the normal forms if a02 and b01 are set equal to zero.

Moly-Cop Tools / 137862878.xls.ms_office

Page 1

4/3/2013 / 7:23 PM

On the basis of Pilot or Industrial Scale Data, like those typically obtained from plant sampling or audit campaigns, the BallParam_Open routine allows for the calculation of all a's and b's (including dcrit) that minimize the least-squares Objective Function :

f = S wi [(fi - fi)/fi]2
i=1

(3)

where the fi's represent the experimental size distribution of the mill product (as % retained on screen 'i' ) and the fi's represent the model response for each corresponding fi, for a given set of model parameters. The a's and b's (including dcrit) that yield the minimum possible value of f are so considered to be representative of the particular ore under analysis. Also in Equation 3, the wi's represent user defined Weighting Factors that quantify the relative quality and reliability of each particular mesh value with respect to the other screens data. Relatively high values of such weighting factors indicate more reliable measurements. At the extreme, a wi factor equal to zero means that this particular measurement is not being included in the Objective Function. The minimization problem stated above may be readily solved with the aid of the Excel Subroutine Solver.

Data Input and Program Execution : Most of the data required by the algorithm must be defined in each corresponding unprotected Turquesa background cell - inside the red double-lined border - of the here attached Data_File worksheet. blue background cells contain the results of the corresponding formulas there defined and are protected to avoid any accidental editing. The remaining information required to run the program is entered in the Control_Panel worksheet, where the user is requested to provide initial guesses of the grinding parameters listed above. To run the program, select the objective function Cell E27 in Control_Panel and then, from the Tools Menu, select Solver ..., then Min and then By Changing any combination of Cells E10:E24 (see Useful Hints below). Clicking on the Solve button will execute the desired calculations. Important Notice : Solver ... must be run every time any element of input data gets to be modified . Otherwise, the current outputs are not valid. Calculation results are summarized in the Reports worksheet. An interesting feature of this routine is that the user has the option to save for later reference every analyzed data set by copying the Data_File worksheet into as many as required Test 1, Test 2, etc. worksheets. For reprocessing these data, simply copy the information back to the Data_File and re-run the Solver routine. New Moly-Cop Tools users are invited to explore the brief comments inserted in each relevant cell, rendering the whole utilization of the worksheets self-explanatory. Eventually, the user may wish to remove the view of the comments by selecting Tools / Options / View / Comments / None.

Moly-Cop Tools / 137862878.xls.ms_office

Page 2

4/3/2013 / 7:23 PM

Useful Hints : The proper utilization of the parameter estimation spreadsheets - like BallParam_Open - is by far the most complex task to be undertaken with Moly-Cop Tools. There is no set of firm recommendations to obey, but the following hints may help guiding the user in the search of the most representative set of grinding parameters : 1. It is not required to run the Solver search for all parameters at the same time. In fact, this is strongly not recommended, as the search algorithm will most likely fail to find the minimizing optimum when dealing with too many parameters at once. To exclude any given parameter off the search, remove the corresponding cell reference (in the E10:E24 range in Control_Panel) off the By Changing list of cells in Solver and that parameter value will then remain constant and equal to its original guess during the whole search. 2. When multiple experimental data sets are available (the most desirable condition) for various ore types (hard, soft, etc.) or grinding conditions (mill filling, ball size, % solids, etc.), always attempt to obtain a reasonable model fit of these data with the same average parameter values for all sets. If that is not acceptable, proceed to allow differences in the most critical parameters in the following sequence : a0, dcrit, b0, a1and b1. Keep in mind that a larger number of adjustable parameters in the search (that is, included in the By Changing list) will necessarily yield lower Objective Function (Cell E27) values, but the values of the parameters that achieve such minimum will be less reliable in statistical terms and therefore, from one data set to another, these parameters will exhibit significant variability and covariance amongst them, rendering them meaningless. 3. An advisable starting run of the algorithm for each independent available data set should considered a search for parameters a0, a1, dcrit and b0; leaving the others fix at nominal values : a2 = 3.0, b1 = 0.5 and b2 = 4.0. Use the green background table below the Control_Panel area to summarize the results of your search with the various available data sets. Be careful to follow the instructions provided at the bottom of such table. If the model fits appear to be reasonable, then you should attempt to reduce the number of adjustable parameters. Look for the one parameter that exhibit the least Coefficient of Variation (the ratio of its standard deviation to its average value, for all data sets) and set it fix at its average value and equal for all data sets, for the next runs of the search. Continue removing the remaining adjustable parameters off the By Changing list, one at a time, until the model fit is no longer judged acceptable. The concept behind this parameter screening process is to lump the impact of whatever ore properties or process variable are being investigated into the least possible number of parameters in order to improve the statistical significance of the estimations and facilitate the derivation of valid conclusions regarding such effects. For instance, it is believed that the distribution of ball sizes in the mill only affects the parameters a0 and dcrit. Similarly, hardness variations of the same ore source may be well described by changes in the a0 parameter only. Not all grinding systems require a finite value of dcrit; whenever the estimated value of this parameter tends to be greater than the top mesh opening, set it fix at 100,000 and remove it from the By Changing list. 4. If the step above does not provide satisfactory model fits, then you should add b1 to the list of adjustable parameters and eventually, include b01, a02 and a12 (one at a time, in that order). If the model fits are still not satisfactory, there is a good chance the data contains significant experimental errors and should be discarded. 5. Never run a search for a2 or b2, as these parameters have little effect on the objective function and may just confuse the search algorithm. When in doubt, simply try assigning them values between 2 and 4 and observe the response of the Objective Function (Cell E27). In other words, these two parameters are better optimized by a very basic trial-and-error procedure, not via the Solver routine.

Moly-Cop Tools / 137862878.xls.ms_office

Page 3

4/3/2013 / 7:23 PM

Moly-Cop Tools / 137862878.xls.ms_office

Page 4

4/3/2013 / 7:23 PM

Moly-Cop Tools / 137862878.xls.ms_office

Page 5

4/3/2013 / 7:23 PM

inimizing optimum when dealing with too many parameters at once.

(hard, soft, etc.)

If the model fits appear

have little effect on the objective function and may just confuse the

Moly-Cop Tools / 137862878.xls.ms_office

Page 6

4/3/2013 / 7:23 PM

Moly-Cop Tools

TM

BALLPARAM_OPEN : Estimation of Grinding Parameters from Plant Scale Data

100

SELECTION FUNCTION : alpha0 alpha1 alpha2 Dcrit Expanded Form alpha02 alpha12 0.009180 0.650 2.5 6532 0.0000000 1.000

% Passing

10

BREAKAGE FUNCTION : beta0 beta1 beta2 Expanded Form beta01 0.20000 0.250 4.0
Feed Discharge (Exp.) Discharge (Adjusted) SiE * 10

1
0.000
10 100 1000 Particle Size, microns 10000 100000

Objective Function

0.00

Note : Current calculations are not valid, if SOLVER has not been run after the last data modification.

(*) Suggested Default Values : alpha02 = 0 and beta01 = 0.

Moly-Cop Tools / 137862878.xls.ms_office

4/3/2013 / 7:23 PM

Moly-Cop Tools / 137862878.xls.ms_office

4/3/2013 / 7:23 PM

Moly-Cop Tools

TM

BALLPARAM_OPEN : Estimation of Grinding Parameters from Plant Scale Data

Mill Dimensions and Operating Conditions Diameter Length Speed Charge Balls ft ft % Critical Filling,% Filling,% 18.50 22.00 72.0 38.00 38.00 rpm 12.82

Interstitial Slurry Filling,% 100.00

Lift Angle, () 35.00

3348 0 536 3885 10.0 4316

Balls Overfilling Slurry Net Power % Losses Gross kW Mill Flowrate, tph (dry) Total Energy, kWh/ton Balls Energy, kWh/ton Objective Function 1622.8 2.39 2.06

% Solids (by weight) Ore Density, ton/m3 Slurry Density, ton/m3 Balls Density, ton/m3

72.0 2.80 1.86 7.75

Charge Volume, m3 63.76

Mill Charge Weight, tons Apparent Ball Slurry Density Charge Interstitial Excess ton/m3 296.48 47.48 0.00 5.395 Feed Size Distributions

Mill Feed i 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Mesh 1.05 0.742 0.525 0.371 3 4 6 8 10 14 20 28 35 48 65 100 150 200 270 400 -400 Opening 25400 19050 12700 9500 6700 4750 3350 2360 1700 1180 850 600 425 300 212 150 106 75 53 38 0 Mid-Size % Ret % Pass 100.00 100.00 97.65 91.00 85.02 80.19 75.76 71.36 66.87 61.49 55.67 48.60 40.41 31.68 23.97 18.22 14.31 11.73 9.95 8.74 0.00

Mill Disch. (exp) % Ret % Pass 100.00 100.00 98.88 96.32 93.81 91.53 89.16 86.44 83.28 79.16 74.30 68.10 60.53 51.77 42.92 35.18 29.08 24.57 21.24 18.84 0.00

Mill Disch. (adj) % Ret % Pass 100.00 100.00 98.88 96.32 93.81 91.53 89.16 86.44 83.28 79.16 74.30 68.10 60.53 51.77 42.92 35.18 29.08 24.57 21.24 18.84 0.00

%Ret. (smoothed) Exp Adj.

% Error

wi

wi ABS(error)

21997 15554 10984 7978 5641 3989 2812 2003 1416 1001 714 505 357 252 178 126 89 63 45 19

0.00 2.35 6.65 5.98 4.83 4.43 4.41 4.49 5.38 5.82 7.07 8.19 8.72 7.72 5.75 3.91 2.59 1.78 1.21 8.74

0.00 1.12 2.56 2.51 2.28 2.37 2.72 3.16 4.13 4.86 6.20 7.57 8.76 8.85 7.74 6.10 4.51 3.33 2.40 18.84

0.00 1.12 2.56 2.51 2.28 2.37 2.72 3.16 4.13 4.86 6.20 7.57 8.76 8.85 7.74 6.10 4.51 3.33 2.40 18.84

3.68 6.19 8.47 10.84 12.44 13.04 14.66 17.23 21.06 25.91 31.51 36.23 39.12 39.02 35.96 30.53 24.08 35.18 29.08 24.57

3.68 6.19 8.47 10.84 12.44 13.04 14.66 17.23 21.06 25.91 31.51 36.23 39.12 39.02 35.96 30.53 24.08 35.18 29.08 24.57

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) 0.00 (0.00) (0.00) (0.00) 0.00 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) sum

3 4 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 4 3 94.00

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Moly-Cop Tools / 137862878.xls.ms_office

4/3/2013 / 7:23 PM

Moly-Cop Tools

TM

Test N

BALL_PARAM
Conventional Grinding Model Parameter Estimator Remarks : Base Case Example

DESIGN AND OPERATING CONDITIONS

Particle Size Distributions (Cumm. % Passing) i Mesh Opening Feed Exp.

Discharge Adj. / / / / / / / / / / / / / / / / / / / / / 100.00 100.00 98.88 96.32 93.81 91.53 89.16 86.44 83.28 79.16 74.30 68.10 60.53 51.77 42.92 35.18 29.08 24.57 21.24 18.84 1273

1.05 0.742 0.525 0.371 3 4 6 8 10 14 20 28 35 48 65 100 150 200 270 400 D80, microns

25400 19050 12700 9500 6700 4750 3350 2360 1700 1180 850 600 425 300 212 150 106 75 53 38

100.00 100.00 97.65 91.00 85.02 80.19 75.76 71.36 66.87 61.49 55.67 48.60 40.41 31.68 23.97 18.22 14.31 11.73 9.95 8.74 4680

100.00 100.00 98.88 96.32 93.81 91.53 89.16 86.44 83.28 79.16 74.30 68.10 60.53 51.77 42.92 35.18 29.08 24.57 21.24 18.84 1273

MODEL PARAMETERS Selection alpha01 alpha02 alpha11 alpha12 alpha2 Dcrit 0.009180 0.0000000 0.650 1.000 2.500 6532 beta00 beta01 beta1 beta2 Obj. Function Breakage 0.20000 0.000 0.250 4.000 0.00
4/3/2013 / 7:23 PM

Moly-Cop Tools / 137862878.xls.ms_office

Moly-Cop Tools

TM

BALLPARAM_OPEN : Estimation of Grinding Parameters from Plant Scale Data

Mill Dimensions and Operating Conditions Diameter Length Speed Charge Balls ft ft % Critical Filling,% Filling,% 18.50 22.00 72.0 38.00 38.00 rpm 12.82

Interstitial Slurry Filling,% 100.00

Lift Angle, () 35.00

3348 0 536 3885 10.0 4316

Balls Overfilling Slurry Net Power % Losses Gross kW Mill Flowrate, tph (dry) Total Energy, kWh/ton Balls Energy, kWh/ton Objective Function 1622.8 2.39 2.06

% Solids (by weight) Ore Density, ton/m3 Slurry Density, ton/m3 Balls Density, ton/m3

72.0 2.80 1.86 7.75

Charge Volume, m3 63.76

Mill Charge Weight, tons Apparent Ball Slurry Density Charge Interstitial Excess ton/m3 296.48 47.48 0.00 5.395 Feed Size Distributions

Mill Feed i 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Mesh 1.05 0.742 0.525 0.371 3 4 6 8 10 14 20 28 35 48 65 100 150 200 270 400 Opening 25400 19050 12700 9500 6700 4750 3350 2360 1700 1180 850 600 425 300 212 150 106 75 53 38 Mid-Size % Ret % Pass 100.00 100.00 97.65 91.00 85.02 80.19 75.76 71.36 66.87 61.49 55.67 48.60 40.41 31.68 23.97 18.22 14.31 11.73 9.95 8.74

Mill Disch. (exp) % Ret % Pass 100.00 100.00 98.88 96.32 93.81 91.53 89.16 86.44 83.28 79.16 74.30 68.10 60.53 51.77 42.92 35.18 29.08 24.57 21.24 18.84

Mill Disch. (adj) % Ret % Pass 100.00 100.00 98.88 96.32 93.81 91.53 89.16 86.44 83.28 79.16 74.30 68.10 60.53 51.77 42.92 35.18 29.08 24.57 21.24 18.84

%Ret. (smoothed) Exp Adj.

% Error

wi

wi ABS(error)

21997 15554 10984 7978 5641 3989 2812 2003 1416 1001 714 505 357 252 178 126 89 63 45

0.00 2.35 6.65 5.98 4.83 4.43 4.41 4.49 5.38 5.82 7.07 8.19 8.72 7.72 5.75 3.91 2.59 1.78 1.21

0.00 1.12 2.56 2.51 2.28 2.37 2.72 3.16 4.13 4.86 6.20 7.57 8.76 8.85 7.74 6.10 4.51 3.33 2.40

0.00 1.12 2.56 2.51 2.28 2.37 2.72 3.16 4.13 4.86 6.20 7.57 8.76 8.85 7.74 6.10 4.51 3.33 2.40

3.68 6.19 8.47 10.84 12.44 13.04 14.66 17.23 21.06 25.91 31.51 36.23 39.12 39.02 35.96 30.53 24.08 35.18 29.08

3.68 6.19 8.47 10.84 12.44 13.04 14.66 17.23 21.06 25.91 31.51 36.23 39.12 39.02 35.96 30.53 24.08 35.18 29.08

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) 0.00 (0.00) (0.00) (0.00) 0.00 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

3 4 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 4

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Moly-Cop Tools / 137862878.xls.ms_office

4/3/2013 / 7:23 PM

21

-400

19

8.74

0.00

18.84

0.00

18.84

0.00

24.57

24.57

(0.00)

0.00

Moly-Cop Tools / 137862878.xls.ms_office

4/3/2013 / 7:23 PM

Moly-Cop Tools / 137862878.xls.ms_office

4/3/2013 / 7:23 PM

Moly-Cop Tools / 137862878.xls.ms_office

4/3/2013 / 7:23 PM