ASTRAD is implemented as a set of Excel functions. A wizard is available for automating the process of configuring input and output data tables and the ASTRAD functions. This is the recommended way to proceed.

The jig test data is entered into Excel, and these ASTRAD functions are entered to predict the jig result, based on model parameters.

These parameters may be fitted using the Excel solver function to best match the model prediction to the laboratory test data.

The fitted parameters may then be used to predict various jig results obtainable in full sized jigs of various sizes.

Typical uses:

1)      Fitting and mass balancing experimental data to extract jigability parameters for the feed.

2)   Predicting feed washability from batch jig results.

2)      Evaluating jig data to help determine problem areas in operation

3)      Designing jig length, width to best suit a feed type and rate

4)      What-if scenarios to estimate effects of ore type variability or planned throughput changes on plant output

 ASTRAD is implemented with the following data tables and functions

Input Parameters

1)      Jig definition table (bed dimensions and operating conditions)

2)      Feed definition table (Size, Density, Assay, Jigability for each feed component)

3)      Bed Response Table (Transport and Bulk properties at each bed layer of the simulation)

4)      Bed Split Table (Number and mass fraction of layers to split the bed)

Output data

1)      Bed response table

·         (Volume fraction of each feed component at each layer in the bed)

·         Cumulated grade and recovery

·         Fraction of each component to feed and product

2)      Bed Layer table (Mass fraction of each component in each user defined layer)

3)      Partition Curve Table (partition fraction of each component to light/heavy.

Function List

1)      _AJ_ASTRAD (The ASTRAD model taking input tables 1-4 and producing output tables 1,2

2)      _AJ_RESULTS (massages output tables into bulk properties per layer

3)      _AJ_TOTALS (Totals bulk properties of Feed, Product and Reject

The ASTRAD Jig Model Wizard

A Wizard is available to aid in setting up the table dimensions and entering the required functions.

The wizard is started from Start->All Programs->Andrew Jonkers->JigModelWizard. It is hoped at some point in the future that this will be implemented as a menu item in Excel, but it is currently difficult to do this AND support both Excel XP and Excel 2003.

The unit of simulation is a separation zone and a splitting zone. Multi-compartment jigs and/or multiple jigs each require a separate model, however multiple stratification compartments before a gate can be simulated with one model provide the fluid mechanical conditions and bed depth don’t vary widely over the length of the jig.

There is nothing however to prevent the input tables of sequential models (such as a jig with middlings and product) being linked by Excel formulas.

Jig Definition Table

This Table contains all the general parameters required for the simulation that don’ logically fit anywhere else.

Jig Length (m)

The length of the jig or jig compartment being simulated range typically 0-6m Small values can be used to estimate bed composition near the feed end of the compartment.

Density of Jigging Fluid

The fluid density of the jigging medium (usually 1000=water)

Jig Bed Depth

The depth of the bed (typically 0.2 to0.5 m)

Splitting Height

This is the percent of the depth at which to split the bed to product and reject. Typical range 5 to 95 (higher numbers splitting lower in the bed.

Because material flow is volumetric and non-plug flow, there is no a-priori relationship between split height and mass fraction to product. If you wish to calculate result for a particular mass fraction you can either interpolate the yield vs height table in Output table 1, or you can use the built in excel solver to adjust split height until the split achieves the specified mass fraction.

Splitting Efficiency

An indication of how accurately the bed is Split. Percent of bed height which represents one standard deviation of the variation of the splitter about the average during the splitting action.

Calculation

A flag to indicate if the model should be calculated or not. Often while the model is being set up you may not wish the model to be calculated every time you change an input parameter

Jig Width

The width of the jig in meters. Required in a continuous jig to relate the feed rate to the transport velocity of material down the jig.

Jigging Time

In a batch test, the time over which the test is to be conducted (range typically zero to 600 seconds)

If this time is set to non-zero, the feed rate, jig width and length are ignored, and the batch mode simulation is done (all species have the same residence time0. In continuous mode 9Batch time = 0, the residence time of the species is defined by the overall volumetric balance of feed and jig cross-sectional area and the transport profile

AstradID

Any number to identify the model run. This value is not used by the model and is purely to allow the user to identify the model run if required.

Feed definition

The feed is defined by a user defined number of unique feed components whose combined  mass fractions add to one. Typically at least four to five density fractions and two to three size fractions are a good compromise between availability of experimental data and enough resolution to produce useful model output. The model requires a minimum of two components, and will happily simulate a binary separation, but if the feed is actually more complicated than a binary system don’t expect miracles in the output.

Each feed component has a proscribed density, size, mobility, diffusion coefficient, mass fraction and associated assay.

Density

The average density of the component. Note that if the component represent a large component of the feed, this component may represent a moderately wide density range of material. The model output represents the spread of all of this material in the bed as if it was a single density component. This is why it is important to define the feed as finely as practical.

Size

The average size of the component. This is usually the geometric mean of the size fraction, but may be weighted if the mass distribution with size is known.

Mass Fraction

The fraction of the component in the feed. Mass fractions of all components must add to one. If they do not the model normalises the sum to one and continues with the calculation.

Diffusion Coefficient

The tendency for this species to remix in the face of turbulent forces. Typically the same value may be set for all similar components, but will typically vary with size.

Mobility Coefficient

The tendency of this component to separate when surrounded by a bed with unity density gradient. This value is typically set the same for all feed components

Assay

Up to three assay values may be defined for each component. These values actually take no part in the model calculation but are carried through with each species in the model run so as to calculate grade information for the model output.

Bed Table

This table defines the conditions for the bed

Relative transport velocity

While the overall velocity and residence time in the jig is defined by the feed rate, cross sectional area and void space in the jig bed, different layers in the bed transport at different rates. Top layers typically transport several times faster than lower layers due to lighter density making them fluidise more easily, and the boundary condition in the to of the bed being a fast moving layer of water, compared to the fixed bed plate, and the flow rate constraints of the gate and weir bar at the end of the compartment.

Void Space

The volumetric fraction of space in the bed not taken up by particles. This is used to calculate the volumetric flow and velocity of feed down the bed which determines residence time. Typically set to a bulk value at all heights, but may be adjusted due to size distribution variation from the bottom to the top of the bed.

Relative Diffusion

This is a relative value: each species has its own diffusion coefficient, but this may change with depth in the bed due to different average fluid conditions. Typically increase slowly towards the bottom of the bed, but good model fits can usually be achieved even if this effect is ignored since the lower layer typically reports in total through the gate so how mixed it is relative to the layer above it has no direct bearing on product.

Relative mobility

Similar concept to the relative diffusion except it is the relative tendency to separate with height. Typically set to one for all layers.

Bed Split Table

The ASTRAD model simulates the separation at quite a fine vertical resolution in the bed. Normally bed data is collected by manually extracting layers from the bed, and it is difficult to extract material to a predefined volume or mass profile. Typically the mass fraction in each layer is calculated after the extraction and only approximately aimed for during the extraction process. It is useful therefore to be able to cumulate the ASTRAD results to a custom profile.

Number of layers

How many layers are extracted in a real test, or the detail in which you want to view the predicted result (in the case of no data being available)

Mass fraction in each Layer

Total of layers should add to one, otherwise you are free to chose (Typically set to actual mass fractions extracted during an experiment).