Testwork: Aggregating samples
Model Results Table
The list of samples that run through a circuit model is shown in any "model results" table. This is the list of samples that contain all of the necessary test results required to satisfy a particular model. For example, the "results" table for a Bond/Barratt model will show only those samples that have a crushing work index (WiC), Bond rod mill work index (WiRM) and a Bond ball mill work index (WiBM). Any sample that lacks any of these three values can not be run in the model and will not appear in the listing.
The samples that appear in the "results" table can not be directly manipulated. The list is generated automatically by combining individual test results by matching the sample name and program fields.
A default value can be entered in the testwork program editing dialogue box (see screenshot). If a sample is missing a particular test, then any default value specified for that program will be given for that test. This allows data sets that are missing a required test to be used in models with the default test value.
Example: Aggregation of data
The example data set from Los Bronces (from published papers by Oestreicher & Spollen, SAG 2006 and Becerra & Amelunxen, Procemin 2012) is summarized below. It contains samples that have been subjected to a variety of tests including the Bond work index series and the drop weight tests.
|Los Bronces A||13.3||9.4||8.97||55.4|
|Los Bronces B||14.9||11.2||8.83||44.4|
|Los Bronces C||14.5||13.1||9.46||33|
|LB 25th percentile||15.6||14.7||35.7|
|LB 75th percentile||17.4||18.5||27.1|
A Bond/Barratt SAG circuit model would generate a result table that returns only samples Los Bronces A, Los Bronces B and Los Bronces C. The other samples are missing one of the required test results (the crushing work index, WiC), so these are not available to the model.
In order to use those samples missing the WiC in the standard Bond/Barratt SAG circuit model, the testwork program can be edited to include a default WiC value (10 is used in the example screenshot). This gives a default value for a missing test results, the WiC in this case. This will satisfy the requirement to have a crushing work index value for the remaining samples, LB 25th to LB minimum, so all may now be used in the calculation.