THE UNIVERSITY OF DODOMA

COLLEGE OF EARTH SCIENCE AND ENGINEERING

DEPARTMENT OF MINING AND MINERAL PROCESSING

PROGRAMME: BACHELOR OF SCIENCE IN MINING ENGINEERING

COURSE NAME: GEOSTATICS AND ORE RESERVE ESTIMATION
COURSE CODE: MN 319
COURSE INSTRUCTOR: ENG. GEORGE BENNET

NAME REG. NO.

KAIZA MAPUNDA T22-03-13866

LEVIATUS FABIUS T22-03-01068

JOSHUA SELEMIA T22-03-10781

KELVIN SWAI T22-03-11871

ISDORY ZAKAYO T22-03-13235

PETER MASHAURI T22-03-03666

KIRAKA CHARLES T22-03-09938

JOJI AMONI T22-0311872

IMPORTATION OF DATA.

The project was provided with 2 sets of data, i.e. the data and the grid data which were excel type data.
The data provided for the project was supposed to be imported on the SGeMs software so that
interpolation can be done.

Figure 1; importation of data

The first data which was imported was the data data. The data was put into point set because the data
provided was in 2-D. The reason of putting the data in point set were

. to manage multiple data set from drill holes.

.to compare different sampling campaign.

Figure 2; point set

NAMING OF DATA.

The data provided both data and grid data were named first and their coordinate of points were
specified into x, y and none since the data were in 2-D.
Figure 3; naming of the data

Then after the naming of the data provided, the data set had to be made visible and distinguishable
between “data” data and “grid” data. And so, on the “data” data, the preferences were changed to be
suitable in the visual and so the point size were increased to 10 scale and the were painted into color
map of rainbow.

GENERATION OF THE HISTOGRAM.

After the changing of the texture of the “data” data, then the histogram was generated by using the
data analysis. The histogram was made were the object used was “data” and the property used was
“grade”. The number of bins was 8 because it provided good spatial data distribution of the sample data
provided.
Figure 4; a histogram

VARIOGRAM MODEL GENERATION.

The variogram was generated from scratch since we didn’t have a loaded variogram. The grid name and
its properties were filled. The grid name was DATA, the head property was GRADE and tail property was
GRADE. Then after filling next was clicked.

Figure 5; a variogram modeler

Then the parameters of the variogram were filled. The favorable number of number of lags was 8
because it shows the good spatial data distribution. The lag separation was 25, in order too many data
have to be included because the more the data the more accurate the variogram. also, the lag tolerance
was 12.5 because it had to be at least half of the lag separation.

The number of directions given were three which were east-western direction, north-southern
direction, northwest-southeast diagonal. The azimuth was 0, 90, 45 degrees because it helps to study
anisotropy because it helps in variogram.

The dip was 0 degrees . The tolerance was 15, 15 and 20 degrees for the diagonal to ensure the vertical
continuity is considered and slight wider tolerance to capture diagonal continuity. And bandwidth was ½
of spacing which are 50,50 and 75ft since too high would mix unrelated points from different locations.
And too low may reduce the number of pairs. the next was clicked.
Figure 6; parameters of a variogram

After the variograms were drawn, the other data was filled to perfect the variogram. The nugget effect
was 0.05 because data has good spatial continuity (less noise). The number of structures is one because
the nested structure was used. Sill is obtained by using variance minus nugget effect (13.9371-
0.05=13.8871). this indicate 99.64% of data were used which means the ore grade change in predictable
way in distance. Then the 3 models were obtained i.e. spherical, gaussian and exponential after
balancing the ranges i.e. max 200, min 0, med 200.
Figure 7; exponential variogram

Figure 8; spherical variogram

Figure 9; gaussian variogram

KRIGING INTEROLATION.

CREATING OF CARTESIAN GRID.

The new cartesian grid was created by using grid data with the purpose of regular spacing and simplicity,
efficient interpolation and kriging and finally better visualization and easy resource estimation. Where
we named it GRID2, then we filled the number of cells in x which was 240 and in y was 150.
Figure 10; a new cartesian grid

RESOURSE ESTIMATION.

The resource was estimated by using kriging (ordinary and simple kriging) and the data were filled from
histogram and variogram above. The estimation was done by using all the types of variogram.
Figure 11; importation of data of spherical SK

THE VISUALIZATION OF DATA.

After the data being filled in both simple kriging and ordinary kriging the data was visualized in 2-D for
the purpose of better understanding of spatial pattern (identify trends, cluster and anisotropy),
detection of anomalies and errors and finally more accurate resource estimation and block modelling for
effective communication and decision making in mining and metallurgical works.

Figure 12; visualization of spherical simple kriging.

Figure 13; Visualization of exponential SK

Figure 14; visualization of gaussian SK

Figure 15; visualization of spherical OK

Figure 16; visualization of exponential OK

Figure 17; visualization of gaussian OK

INTERPRETATION OF DATA.

The color distribution reference from the preference in the data;

 Red color shows the super high grade (44% of Fe ore)

 Yellow color shows the medium grade (38.67% of Fe ore)
 Pale yellow shows medium low grade (36% of Fe ore)
 Light blue color shows subgrade (33.33% of Fe ore)
 Dark blue color shows the subgrade (30.67%-28% of Fe ore)
 Pale red color shows high grade (41.33% Fe ore)

From the data provided, iron was graded from high grade to subgrade on the direction of NORTHWEST
to SOUTHEAST with continuity of the grade from southwest to northeast. This indicate that in order to
get maximum profit the pit design should lie along NW-SE to follow the highest-grade zone and prioritize
high grade zone for area extraction to maximize economic returns by using cut of grade optimization to
adjust mining sequence.

Implementing blast hole sampling, RC drilling or in-pit sampling to refine grade boundaries. Use GPS for
accurate positioning of the sample.

Therefore, based on the two kriging, ordinary kriging is more appropriate and unbiased because our
data show local variation so we don’t depend on mean.