You are on page 1of 19

Special Topics in Reservoir Engineering

MSc in PETROLEUM ENGINEERING

Estimation and Prediction of various


Reservoir features using Mbal Software

Prof: V. Gaganis Student: E. Pafra


July 2021
Project description
An initially undersaturated reservoir is to be modeled using the material balance
method. The production data collected over a 6.5 years period is available. The initial
estimates of the reservoir data is also given in the Table below.
1. Get the best possible estimates of the reservoir reserves, the recovery factor at
the end of the 6.5 years production period, the water influx (if any) and the
phase relative permeabilities.

2. Use the history matched model to predict future production of oil, gas and
water for the next 6 years. The reservoir is scheduled to produce at a constant
rate of 3,000, 2,000 and 1,000 stb/d for three consecutive 2 year periods. Draw a
chart of the predicted gas production and comment on that.

3. The reservoir porosity and water saturation are assumed not to be known
exactly but to be governed by normal distributions with standard deviation of
0.03 (fraction) and 0.02 (fraction) respectively. The fluid GOR value is governed
by a uniform distribution with an average value equal to the given one and a
span of ±10%. Estimate the oil and the gas zone volumes at initial conditions so
that the reserves estimated in question 1 correspond to P10, P50 and P90 of the
estimated reserves distribution respectively.

The material balace method


• Get reservoir, fluid and recorded production data

• Identify (or verify) the drive mechanism


• Tune the material balance equation against this data (history matching)

• Get optimized values for crucial parameters


• Original Oil Initially In Place –OOIIP, gas cap size, etc

• If the equation reproduces accurately the recorded history, it can be safely used
to predict future reservoir production
Part 1
To proceed with estimating the required parameters, we need to start building our model in
Mbal according to our given dataset (No 2)

The following figures show a step-by-step initialization of our problem.

File->New
Tool->Material Balance (check)

Options:

PVT->Fluid Properties (Insert available data and then click Match)


We then introduce the measured PVT Data and click “Match” again

Mbal applies a linear transformation to each correlation...


...from which, after we click on “Calc” and then “Match Param” we will get to choose the one
that gives the best approximations for the two parameters, plus the smallest deviation! Glaso
and Petrosky et al seem to be the best correlations in our case.
Going back to the main scree, we select Input->Tank Parameters
and fill in our data

We assume that no aquifer is present, which practically means no water drive, to check the
behavior of the reservoir with this set up and we take into account the rock compaction as the
only driving mechanism for our undersaturated reservoir.

Τhen, we fill up the matrix of the relative permeability tab according to our data.
Then, we press “Plot” and we get our relative permeability curves.

We load our given model at the “Production History” tab and convert gas production to GOR,
the behavior of which is shown at the plot below. We notice that solution GOR follows the
trend, while the producing GOR starts rising from the middle of 2004 because bubble point
pressure has been reached and after a while the gas has reached critical saturation and starts
flowing.
We go on selecting “History Matching” to check the validity of our model and start our check
with the graphical method.
And we also check the analytical method.

We can conclude that our model deviates from the production data, and we need to introduce
some pressure support in order to achieve a match.

A first approach in matching our models is to apply “best fit” on Campbell plot (first graphical).
We can see a better fit, but still not satisfying. Our reserves shifted from 186 MMStb to 326.75
MMstb
We can see that the analytical method is improved but we still need to check on that pressure
support. It can be no other than a water aquifer, since a gas cap is not present anyway in our
case. So we select “Water influx” and introduce our aquifer.
After trying various models, (Small Pot, etc.) the best image was given by applying Hurst-van
Everdigen modified model.

We go back to checking our plots. With the OOIIP at 326.75 MMstb and the presence of the
specific aquifer, we get the following image.
So, what we can do, is to go back to the initial 187 MMstb OOIIP and check the effect of just the
addition of the aquifer...
which indeed looks much better, but still, we need to regress on our parameters to achieve a
good match.
Checking back at our plots:

So the final estimation of our reserves seem to be at 151 MMstb.


The next step is to run simulation on history matching, to check the validity of our model and
get the solution on the asked data for the first part of our problem.

We observe a quite decent match at the pressures.

The Oil Recovery Factor is 12.56%


The aquifer influx at the end of the 6.5 years production period is around 10.3 MMstb.

In order to check the change in the relative permeabilities, we need to fine tune the fractional
flow for water and go from the first image to the second by regressing on the parameters
suggested by the system.

Then, after saving the changes, we head back to the relative permeabilities and check the new
numbers.
Part 2

We have to use the history matched model to predict future production of oil, gas and water
for the next 6 years while the reservoir is scheduled to produce at a constant rate of 3,000,
2,000 and 1,000 stb/d for three consecutive 2 year periods.

First we went to Production prediction and we select Production setup. We start prediction
from the end of production history and end it after 6 years

We select Production and constrains, we add the dates and the average oil rates of 3,000, 2,000
and 1,000 stb/d for three consecutive 2 year periods and we are done.

We run production prediction and plot our results.

From point of prediction the reservoir seems to be perfectly capable to produce for 6 years
with the given oil flowrates without complete depletion. Due to changing oil flowrates every
second year we can notice the change of incline of gas production curve because of milder
reservoir depletion. The graph below shows the tank pressure and the cumulative gas
production.
Part 3

In order to proceed with the third part, we use will use Monte Carlo tool, which is included in
our software.

We go to Tool and we select Monte Carlo.


We are introducing again our oil data.
From Input we choose Distributions and we open the table shown below where we add our
available data and the distributions.

By trying out different values for the bulk volume in order to get the estimates calculated at the
first question (151 MMstb) we end up with the following tables/numbers:

P90
Bulk Volume: 7530 MMft3
P50
Bulk Volume: 5862 MMft3
P10
Bulk Volume: 4822 MMft3

You might also like