Professional Documents
Culture Documents
Permeability Estimation Based On Cokriged Porosity Data: September 2015
Permeability Estimation Based On Cokriged Porosity Data: September 2015
net/publication/281769705
CITATIONS READS
0 231
1 author:
Anne Sirait
Lehigh University
2 PUBLICATIONS 0 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Anne Sirait on 15 September 2015.
1. Background
Porosity and permeability distributions are two indicators needed for petroleum
important role. An accurate estimation of the porosity changes will give information
about changes in total reserves. In addition to being porous, a reservoir rock must have
the ability to allow petroleum fluids to flow through its interconnected pores. That is, the
pores of the rock must be connected together so that hydrocarbons can move from one
pore to another. The rock’s ability to conduct fluids is referred as permeability. The
applied to quantitatively relate well and seismic data and quatitatively assess map
accuracy (Wolf et. al., 1994). Cokriging method is chosen to integrate the well data and
seismic data. The cokriging method considers the well data as the primary variable and
All methods described above discussed how to estimate porosity distribution over
a large area. While to estimate the permeability distribution is still challenging to do;
this research will focus on how to estimate the permeability distribution over a large
area based on the cokriged porosity data. The goal of this research is to integrate the
porosity data from well and seismic data using cokriging method. The porosity
distribution produced from the cokriging method will be used to estimate the
procedure mentioned above will be the final step to determine the permeability
2. Purpose
The purpose of this research is to estimate the permeability based on the cokriged
porosity data. First the porosity data is extracted from the well and seismic data, then
both data are integrated using cokriging method. The well data is considered as
primary data and seismic data is considered as secondary data. Then, we will use the
permeability map and finally, analyze the value of permeability distribution based on
3. Methodology
Method used in this research are geostatistics and inversion methods. Both
methods were used to predict porosity and permeability distribution at Boonsville area.
Both methods were used to integrate the seismic and well log data to predict the
In the kriging method, the porosity distribution within the area of interest was
derived from one variable only, which is the porosity data form the well. The porosity
the well such as neutron log, sonic log and resistivity log. Density log measured the
bulk density of rock formations; the bulk density is the function of matrix density,
porosity and density of fluid in the pores. The neutron log was not chosen because the
neutron log measured the hydrogen content in rock formation not the porosity of rock
formation. Neutron log will give wrong impression within shale formations which have
The porosity derived from density log is averaged below MFS 90 for about 30 ft,
because the Gamma Ray logs at that depth show the low value which is concluded as
the sandstone area. However, the density log gives a quite high density value at that
area and the porosity log gives a very low porosity value. Based on the log description,
we can conclude that the area 30 ft below MFS 90 is a tight sandstone area.
The averaged porosity data is derived form 14 wells where the porosity value
ranges between 0.8 % - 11.57 %. This porosity value should be cross-evaluated using
neutron porosity and the condition of well based on its caliper value. The variation in
caliper represent the condition of well, it could be rough or flat. Based on the cross –
evaluation of the neutron log and caliper it can be conclude that only 8 wells are able to
be used to build the porosity model. The wells are ASHE C6, BY11, BY13, BY15,
In order to get the porosity map where the porosity value within the area is not
covered good enough with well data, the variogram model is required to give weighing
value. The weighing value is needed to help well data estimate the porosity value
outside the well location. The weighing method is derived from variogram model such
as spherical, exponential and Gaussian model. As the result, the spherical model is the
best method that can describe the well data. Spherical model is the best variogram
model because it gives the biggest range in value among other model. The bigger
range value is able to cover non data area from well location rather than the smaller
values outside the well location, kriging method was applied to get the kriging map of
porosity data. The kriging map stills is not able to show variation of porosity data within
the area of interest. The variation of porosity value still concentrated surrounding well
location while in the other parts there is still no variation displayed. This happens
because the well data is not distributed well within the area of interest. The kriging error
map also confirms this assumption, as the kriging error surrounding the well is smaller
compared to kriging error at the area outside the well location. This means the
estimation works best surrounding the well location, but does not work well outside the
well location.
Because the kriging method is not able to properly display the porosity distribution
within the area of interest, thus the cokriging method is needed. Cokriging method
combines two kinds of data, well data as the primary dataset and seismic data as the
secondary datasets. Seismic data is chosen because it is able to cover a large area,
but the data is still in inexact. The secondary dataset needs to be extracted from the
seismic data.
secondary dataset. Acoustic impedance is extracted from seismic data using the model
based inversion method. Model based inversion method tries to match the synthetic
data produced from the acoustic impedance from log to real seismic data. The method
to match the synthetic and real data is based on squeeze and stretch method until the
correlation achieves maximum value. In this method, the wavelet is considered as the
zero-phase wavelet.
The correlation of synthetic data to the real seismic data is based on three wells
(BY11, BY18D, and CY9), where the correlation coefficient ranges between 0.67 and
0.78. Those value can be considered good for seismic well tie procedure. The inversion
analysis of modeled synthetic trace from the impedance log showed a good correlation
with maximum correlation 0.99 and maximum error 0.07. It means that the modeled
The inversion result shows that the area of interest located below the MFS 90 has
a high impedance value range between 31800 ft/s.g/cc and 34100 ft/s.g/cc. High
impedance value show that this area is the area with dense rock formation. Although
based on Gamma Ray log this area is considered as the sandstone area. Futhermore,
it can be concluded that this area is a tight sandstone area because its high acoustic
impedance value. The acoustic impedance (AI) property is chosen because it derived
from rock density and wave velocity. The density of rock was formed based on the rock
matrix and density of fluid in the rock pore. The density of rock generally decreases as
porosity increases. As in the figure 2, there is AI variation on the map, either at the well
location or at the area outside the well location. The AI information covers the whole
The unique characteristics of well data which have exact information but limited
about the formation properties and the characteristics of seismic data which are able to
cover large are but inexact information of formation properties are combined together in
the cokriging method to improve the porosity map. In order to combine the seismic
model is chosen based on its components such as nugget, sill and range. The
appropriate variogram model for the seismic data (AI value) is spherical model because
the range value of this model is the highest among other model. The higher range
value covers non data area better than smaller range value.
The next step to combine both datasets is to find the appropriate variogram model
between well and seismic. The appropriate variogram model is also spherical model
based on its range value. The modeling variogram of well and seismic are chosen
based on their range value. Range value on the model was shown how far the model
able to reach out the area outside the well. Based on the range value from the model,
the estimation of reservoir properties can be done thoroughly. The cokriging map of
porosity data is displayed in the figure 3. The resulted map from cokriging method is
showing fine variation of interest area. The range of porosity value is between 0.1 % -
12%. The high porosity value is concentrated around ASHE C6, IG Yates 14, and B
Yates 13 well which are represented in color purple. While the low porosity value is
well show the variation of porosity value. The cokriging method which combines well
and seismic data able to improve the quality of porosity map rather than the kriging
map.
Although the map shows the distribution of low porosity value, this map able to
improve the quality of map and based on that the cokriging porosity map is chosen as
the input map for the permeability estimation process. The porosity value range from
0.5% - 12.2%, the vast range of porosity value is can be happen because the
Using the resulted map of porosity data based on cokriging method, permeability
within interest area are able to obtained. The porosity map from cokriging data is used
because it can show the variation of porosity properties of interest area. While the
porosity has a correlation with permeability, it is believed that by using the porosity map
The correlation between porosity and permeability data is obtained based on the
core analysis of two wells, Tarrant A4 and Sealy C2. The correlation between both
properties is based on the function shown in the graph in Figure 4. The function shows
correlation with one permeability value. So based on this correlation, the permeability
Figure 4. Porosity – permeability relation taken from two wells core based on linear regression.
The permeability map obtained from the porosity cokriged porosity data are shown
in the Figure 5. The permeability map obtains from cokriged porosity data able to show
a fine variation within interest area. The map shows low permeability value with the
highest value of permeability 0.85 mD. The low porosity value along with low
permeability value shows that the interest area is very tight zone.
Conclusions
using the kriging and cokriging methods. Based on the method that has been
1. Based on Gamma Ray, density and porosity log, the interest zone is located below
the MFS 90. It can be concluded that the interest zone is a tight sandstone
formation.
2. Porosity values are derived from the density log. It was chosen because it
describes the rock porosity well better than other logs such as neutron log. There
are 14 wells that contains density log, so based on that there were 14 wells with
porosity data.
3. The porosity value of all wells ranges between 0.5% and 12.2% which is quite low
4. Model based inversion has been applied to seismic data to obtain the AI value
5. Ordinary kriging method was chosen to obtain the porosity map. The map obtained
based only from well data and the map shows that there are variations around the
well, but the map was not able to show variations in area beyond the well.
Cokriging method was used as a way to improve the resolution of the porosity
map.
6. The spherical model was the best variogram model for the well data and seismic
data. The spherical model is also the best cross variogram model between well
7. The cokriging method was chosen to improve resolution of the porosity maps
8. The correlation between porosity and permeability was derived from two wells near
the interest area. Porosity and permeability was linearly dependent with correlation
9. The permeability map resulted from cokriged porosity data was derived form the
one on one correlation of porosity and permeability data. The resulted map
showed a fine variation of permeability compared to the map obtained from kriging
process.
Recommendations
1. Since the porosity data was only derived from density log, it is suggested to obtain
the porosity data from other log information such as neutron log, sonic log and
2. The porosity derived from density is needed to be cross – validate with neutron
3. Besides using the model based inversion, the other inversion method can be
3. The correlation between the porosity and permeability value should be derived
4. Other than using the AI inversion, the seismic attributes data can be used as the
5. The other geostatistic method such as Kriging with External Drift or Collocated