How does the Block Factor option work (in the estimation functions?)

stuart masters 9 months ago in Resource Estimation updated by Сергей Дроздов 9 months ago 4

Hi all,

MM has an option for a "Block size factor' during estimation.  The online help states the following about it:

"The block size factor allows a user-defined block size to be used for the interpolation of the grades to the block model. If specified, the block factor will be a (2, 3, 4, etc.) multiple of the parent block size."

If, for example, I have 15x15x5m parent blocks and set the factor to 2, will it estimate the blocks as if they were 30x30x10m and should my QKNA and subsequent discretisation and search parameters be based on that expanded block size?  Or does MM just estimate 15x15x10m blocks, in this case 8 (2x2x2), and average the estimated grades of such?



Hi Stuart,

Although I haven't needed to use this particular part of the estimation functions, my understanding is that your first example is correct.  Two things suggest this 1 - the option to enter a block size factor is only available when you select "interpolate parent blocks only", and 2 - the help file description states "multiple of the [parent block] size" (the term parent block here is misleading I think, it should probably say "input block size").    As far as I can tell using this option is a sort of inverse-discretisation.  Perhaps it is useful to people when you've been told to produce a block model with 2x2x2 blocks, but you know that the geostatistical best block size to estimate is far bigger than that.  I would therefore also expect that you would want to use geostatistical inputs that are suited to the larger block size, not the smaller.  I also assume in that situation using this option would produce a quicker estimation also as opposed to running the estimation for every single block.

Your second example, about discretisation, is all handled on the "modelling parameters" tab.  Whereby you define the number of subdivisions of your input block size under the Discretisation menu,



Thanks Jon,

That's my understanding too and that's how I'm using it (as a "quick and dirty" pseudo larger parent block size estimate for sensitivity testing purposes).

Regarding the number of discretisation points, although they are specified under the modelling parameters tab, they should partly be determined from the QKNA.



Here's the response from Micromine

I've added some comments in [ ].

One of our developers has provided some information about the original improvement request that got submitted. I'll copy and paste the suggestion verbatim below to ensure I don't mess up the explanation

“Estimation block size” gives the opportunity to input any user value for block size (in one [each] direction, the interpolation block size should be bigger than cell size and be divisible) for interpolation of the grades to block model.

  • The analysis of volume variance
  • To estimate few domains (across the thickness of ore body) in one model
  • To take into account the different drilling pattern in block model

So the function Estimation block size gives us ability to interpolate small blocks as one big block (this option works only in one side [I think this means that the factor is applied to all dimensions, i.e. we can't have different factors for each dimension]). If we have model with cell 5*5*5, we can use Filter condition like “Z>100” or Polygon restriction (or flag field) to interpolate some regions of this model with the block size 50*50*50 for example (the value should be divisible to the cell size). So all the original blocks in this region will have the one grade value (Block kriging should use the new block size for this region).

In the last part of the explanation, it does state that further work / kriging settings should use the new block size as things have been re-factored / re-sized.

Colleagues, hi

I often use this function to etstimate sites with different densities of the exploration network. For example: the parameters of the main grid are 50x50 m, the detail area is 10x10 m, and the flanks are 200x200 m. I allocate domains with wf and use different factor for each with the same block size of the model.