Solutions to common problems encountered in Minex resource reporting using SQL
Date 12 September 2013 Prepared for Gemcom Community Conference Author M Davis
Page 2
Gemcom Community Conference | September 2013 Introduction Solutions are presented to four common problems encountered in Minex complex resource reporting using grid and borehole SQL. These four common problems include: Quantifying linear core recovery where coreloss is only recorded through lithology variables, rather than in sample intervals. Defining a point of observation with multiple criteria Setting a meaningful minimum thickness for resource estimation in a ply model (where plies have coalesced), without the need to generate a working section. Setting resource limits with multiple dependant conditions to meet reasonable prospects for eventual economic extraction Examples of improvements to reliability of resource estimation using these tools are presented. These solutions were found to be very effective in large multiseam deposits, generally focussed on opencut mining.
1 Quantifying linear recovery The 2003 Aust Coal Guidelines defines a point of observation (PoO) for use in resource classification as allowing the presence of coal to be unambiguously determined. A Point of Observation for coal quality evaluation is normally obtained by testing samples obtained from surface or underground exposures, or from bore core samples having an acceptable level of recovery (normally >95 per cent linear recovery). The linear recovery of a coal seam is a key factor in determining whether the data can be used as a valid point of observation or not. Core loss is frequently (incorrectly) only recorded in the lithology codes, rather than between sample intervals. In Minex software linear recovery is assessed by comparing the sample intervals and seam intervals. A minimum sample percentage is set in the borehole database settings; and seams with less than the defined sample percentage for a given variable will not allocate a weighted value of that variable for the seam.
Page 3
Gemcom Community Conference | September 2013
Figure 1 Typical borehole with core loss recorded only as lithology, rather than through sample intervals If core loss is recorded through lithology only, rather than sample intervals, it will be ignored by the software, and the linear recovery of a seam may be overestimated. Whilst other software packages (such as Minescape) have built in functionality to deal with this recording of core loss, Minex does not. A method for quantifying core loss logged only through lithology codes (instead of sample intervals) using borehole SQL is outlined below: 1. A new SQL variable is created in the lithology datatype 2. A borehole SQL will allocate a value of 0 for all rocktypes representing core loss (eg KL, NR etc), and all other rocktypes are assigned the value 1. 3. A weighted average of the values is then calculated for the seam using the default seam weighting function. The weighted sum for the seam will show the linear core recovery. Care should be taken if an RD weighting is also assigned, as this may impact the composited seam value 4. Compare the linear recovery to borehole profiles and sample intervals to further support the numerical analysis Care should be taken if an RD weighting is also assigned, as this may impact the composited seam value.
Page 4
Gemcom Community Conference | September 2013
Figure 2 Linear recovery calculated using Minex default and LINREC borehole SQL This technique was applied to a Hunter Valley dataset comprising 549 boreholes with coal quality samples. The minimum sample percentage in the borehole database properties for this dataset was set to 90%. The table below summarises the sample data for the dataset, including samples excluded through the use of min sample percentage setting (104 samples) and the further samples identified through the use of the SQL (a further 102 samples). Table 1 Summary of use of coreloss SQL Use Number of seams Comment Excluded through use of min sample percentage setting 74 Less than 90% of the seam was sampled (due to heat affected coal or intrusions), but no core loss 30 More than 10% core loss, recorded in sample intervals
Page 5
Gemcom Community Conference | September 2013 Excluded through use of SQL, but not min sample percentage setting 102 More than 10% core loss, recorded in lithology, but no sample intervals
2 Point of observation with multiple criteria An Indonesian project has a dedicated Competent Person who has derived hierarchy of points of observation to classify confidence in the resource estimation based upon borehole type, linear core recovery, presence of geophysical logging (Table 2). The combination of the points then form the basis of the resource classification. Table 2 Criteria for classes of points of observation Class Description A Coal Quality with < 10% core loss and valid geophysics
B Coal Quality with >10% core loss with valid geophysics Open hole with valid geophysics Cored hole with 100% recovery and no valid geophysics C Coal quality with >10% core loss with no valid geophysics Open hole with no valid geophysics
The original method of applying the point of observation class included: Assignment of the PoO class by the field geologist and database manager; The modelling geologist running numerous borehole selection parameters to create distance grids based on the PoO class. In addition to the process being tedious, issues in the system arose in circumstances:
Page 6
Gemcom Community Conference | September 2013 Where the top of a borehole was geophysically logged, but the base was not (most likely due to blockage of the tool down the borehole). If a seam had been sampled, and indicated as sampled by the geologist, but the assay data had not been returned from the laboratory. Seams that had previously been uncorrelated had not been assigned a classification. A borehole SQL was written to account for the complex system of PoO definition, using the criteria in the table above. Results of the SQL showed that 11% of the seam intervals had incorrectly been assigned Class A status in the original classification (for the reasons outlined above). The total number of intervals classified increased by 2% (Figure 3).
Figure 3 Change in the number of points of observation in each class with Borehole SQL method The criteria for the use of a point of observation in the resource classification for the project is defined in Table 3. Circles were drawn around the points of observation, using the distances based upon the recommendations of the 2003 Aust Coal Guidelines. The circles were then rationalised at the discretion of the competent person, with further reference to the geological interpretation. Figure 4 shows an
Page 7
Gemcom Community Conference | September 2013 example where Class B points of observation were used to support continuity of quality and thickness between Class A points of observation for an Indicated classification. Table 3 General principles of resource classification based upon a hierarchy of points of observation
Figure 4 Distances around points of obervation class A and class B
Page 8
Gemcom Community Conference | September 2013 The benefits of using this method include: increased confidence in resource classification by using more than one criteria; the ability to utilise data that may not meet Class A criteria as supporting data for resource classification, and the potential to improve indicated and inferred resources through this.
3 Meaningful minimum thicknesses Ply models are constructed to represent geological variations through a coal seam in a stratified model. Significant value can be added to a model when plies are defined well, with appropriate samples and parting bands identified. Working sections can then be compiled from the ply model when careful consideration is paid to potential mining methodology and coal quality. But often at the resource stage (especially for fledgling projects), the ply model is not converted to working sections, which presents an issue for using a minimum thickness cut-off for resource estimation. Setting a minimum thickness cut-off using the Minex detailed resource reporter function will look at the thickness of the ply in isolation, which can lead in an underestimate of the total coal tonnages if individual plies are thin, but the combination of coalesced plies is greater than the desired minimum thickness cut-off. A common work-around for this issue is to use no minimum thickness setting, which can also lead to an overestimate of the volume of coal. A method that assesses seam thickness in the context of the ply relationship to other coal units is required.
Figure 5 Example of typical habit of a coal seam with ply splitting and coalesced.
Page 9
Gemcom Community Conference | September 2013 The solution is presented by using an SQL to create a recovery grid, which is then used in the detailer resource reporter function in the place of a Mine Recovery grid. The SQL to calculate the recovery grid uses the following steps: Recovery SQL 1. Determine maximum parting thickness for plies to be considered coalesced or split (Max Parting) 2. Test whether the ply interburden (IB) and underburden (UB) is greater than or less than Max Parting 3. If IB and UB is greater than Max Parting, the ply is assumed to be isolated and a minimum thickness test is applied 4. If the IB or UB is less than Max Parting, the ply is assumed to be coalesced with the plies above or below, and no minimum thickness test is applied. 5. If the Min Thickness test is passed, the Recovery grid is set to 100%, otherwise, recovery is set to 0%. Seam specific criteria or further conditions, such as seam/interburden ratio, can be included in the script if required. For plies where the coalesced unit is known to be less than Min Thickness, a further individual seam group condition should be applied. This method is highly effective in large multi-seam deposits with complex splitting, such as open cut mines exploiting the Whittingham Coal Measures. An example is presented of a Hunter Valley mine with three coalesced plies that form one seam group. Three estimates were run, using minimum thickness settings of: no minimum thickness, minimum thickness of 0.3 m, and minimum thickness using a recovery grid. The no minimum thickness option obviously recorded the largest estimate, and this formed a basis for comparison for the other two estimates. When a minimum thickness of 0.3 m was applied, the estimate was reduced by 4.6 Mt. Using a recovery grid to define minimum thickness based on the plies proximity to other seams resulted in a reduction in the estimate of only 0.4 Mt. This difference of 4.2 Mt difference between the two methods of defining limits of reasonable prospects for eventual economic extraction may have a significant impact upon the overall assessment of a resource, especially when compiled for many seam groups.
Page 10
Gemcom Community Conference | September 2013
Figure 6 Comparison of tonnes reporting for three coalesced plies using different Min Thickness methods The use of a recovery grid will not only increase the accuracy and efficiency of the resource estimation, it also provides a grid which clearly shows the areas excluded from the estimate. This grid can be rationalised to remove isolated points of exclusion, and provide a more sensible resource estimation.
4 Resource limits based on multiple dependant criteria The recovery grid method can be extended further to potentially complex conditions to meet the JORC requirement of reasonable prospects of eventual economic extraction. If a limit utilises more than one variable, or uses dependant variables, simple SQL IF statements can be written to adjust the recovery grid. For example, a greater strip ratio may be feasible in domains of lower (Figure 7), or a resource limit may be set on a combination of coal quality parameters. This recovery grid method can also be used for seam specific areas of exclusion, without the need to generate exclusion polygons for each seam, and calculate seams individually. Examples of individual seam cut-offs may include areas of heat affected coal, or off-sets from fault zones.
Page 11
Gemcom Community Conference | September 2013 Again, the recovery grid method provides a transparent record of areas of exclusion and a grid that can be rationalised to a sensible area of resource estimation.