Volume

[Craig SS]

I have created a stratigraphic model with Boreholes interval quality data. The strat-model is simple layer cake. The stats of the modeled ore layers equal 1.6 million cu yds. Modeling the interval data yields a much bigger volume, 776M cu ft or 28M cubic yards, constraining this to the top and base of the strat model grids reduces the I data volume to 7 M yards, still too large.

 

It is puzzling the volumes do not match exactly.  Dimensions and nodes are same for the Strat Model and I Data.  The 3D models, profiles and cross sections are similar. Models total 5 MB so too big to send here.

 

Hope you have a suggestion to resolve this volume disparity.

 

Thanks 

 

Craig

 

[JimReed]

Hi Craig,

The volumetrics for a stratigraphic diagram based on grids are computed by;

(1) subtracting the subface (base) elevation from the corresponding superface (top) elevation,

(2) multiplying by the result by the cell x-size and y-size (i.e. base * width * height = volume within that "column"), and

(3) summing the results for all columns.

Conversely, the volumetrics for a block model are computed by summing the volumes for all voxels that meet some criteria (e.g. all voxels with a node value > 3.2 and <3.9).

Now, let's say that we create a voxel model based on grid surfaces (this is what the Stratigraphic Model / Save Numeric Model option does).

The program will assign the voxel values based on the bounding grids that lie above and below the voxel midpoint.

Given that a voxel midpoint may be just below the corresponding, overlying grid node, 49% of the voxel space could actually be above the superface grid node.

The same applies to voxel midpoints that are just above the corresponding underlying grid node - 49% of the voxel could actually be below the subface grid node.

This is why the volumetrics for a block model will never match the volumetrics for a grid-based model.

As you reduce the z-spacing for a block model the volumetrics will start to converge, but they'll never be exactly equal and the model generation will take longer.

The trick is to establish a z-spacing in for which the differences between the volumetrics is acceptable.  If the voxel-based profiles don't look right, then I suspect that the z-spacing is too large.

Best regards,

Jim Reed.

 

 

[Craig SS]

Jim

 

Great suggestion. I reduced z spacing to 1 foot and the volume is acceptable. Now the block model can be tweaked to understand interval grades factors. Appreciate the time you took to write a solution.

 

Best

 

Craig 

[Craig SS]

Hi Tom

 

I understand and will answer to complete my process clarity; before using Jim's solution, for the forum. This may help future users.

 

I constrained the model with a polygon clip and grids of the upper and lower ore surfaces. However the last action - reducing the z from 5 feet to 1 foot, eliminated the volume differential within acceptable limits. The model processing still functioned comfortably timewise. 

 

Really appreciate the forum and valuable inputs

 

Thank you

 

Craig