Modelling PVT properties of reservoir hydrocarbon fluids

while oil field development planning

A.I. Brusilovsky and A.N. Nugayeva

Sibneft

Understanding and development of approaches to reservoir fluids description are among

the most topical issues for the gas sector: typically, companies acquire both oil and gas
fields. This paper focuses on description of an efficient approach to preparation of data
used to generate reservoir fluid PVT relationships in "black oil" hydrodynamic models. Case
history of using this technique in Yarainerskoye field (West Siberia) is discussed here.
Several alternative models differing in PVT properties formation of reservoir fluids are used for
hydrodynamic simulation of hydrocarbons field development.
The use of "black oil" models presumes that properties of phases depend only on pressure at constant
temperature. Thus, two types of models are distinguished: the first one considers the solubility of oil
in gas phase (i.e. wet gas), and the second - that oil is missing in the gas phase (i.e. dry gas). Models
of the first type are used more widely while forecasting the development of oil/condensate deposits
and gas/condensate fields under natural depletion conditions.
Multi-component fluid flow models are basically applied when it is necessary to forecast reservoir
properties for the development of gas/condensate, oil-and-gas, and oil deposits using external gas
drive characterised by intensive inter-phase mass transfer. For these models, reservoir fluid properties
could not be represented (even approximately) as exclusively pressure-dependent, since at the same
pressure gas phases of different composition can be present within the reservoir bed.
I. Data used for PVT models development. Adequate reasoning using per-bed reservoir fluid
properties is required for reserves estimation and field development planning. The uniform (in terms
of both area and volume) characterisation of reservoir beds by representative fluid sampling is one of
the most important requirements. Thus, well-grounded presentation of reservoir fluid composition
could be established, as well as its physicochemical properties. In practice, two basic techniques are
used to produce representative fluid samples:

Downhole sampling

Separator sampling.

Now it would be pertinent to consider basic conditions underlying the representative reservoir fluid
samples.1
Downhole sampling. At least three samples should be recovered to provide adequate analysis of
reservoir fluid PVT properties. Pressure at the sampling point should exceed saturation pressure, i.e.
the recovered sample should be a single-phase one.
Separator sampling. When downhole sampling is not possible, examination of the recombined
specimen is carried out, which has been mixed using oil and gas samples recovered while separation

of the produced fluid. The measured GOR value is used while recombining. In case that the
recombined sample at reservoir conditions (P = const, T = const) forms a two-phase gas-liquid
system, the liquid phase (under-gas zone) and the gas phase (gas cap) should be examined
separately.
II. Theoretical basis of PVT properties modelling. The "black oil" model considers reservoir fluid
as a two-component mix. The "oil" component - stock tank oil, and the "gas" component - the
dissolved gas. According to Gibbs rule of phases, due to pseudo-binarity of the mixture, it would be
sufficient to use relationships between properties of fluid and vapour phases on pressure at reservoir
temperature. Density is set under standard conditions for both "oil" and "gas" components. The input
data set on reservoir fluid properties used in "black oil" models, includes formation volume factor, gas
content and dynamic viscosity vs pressure relationships. The structure of input data concerning vapour
phase properties depends on whether dry gas or wet gas model is applied. In both cases pressure
relationships are set vs pressure, formation volume factor and dynamic viscosity. In case of wet gas
model it is necessary to set data for "oil" component solubility in the vapour phase.

Figure 1. Reservoir oil saturation pressure vs. binary interaction coefficient between
CH4 and 7 + (a); stock tank
oil density vs. "shift parameter" of 7 + (b); reservoir oil formation volume factor vs.
methane "shift parameter" (c);
dynamic viscosity of reservoir oil vs. Vcr,C7+ parameter (d):
pc, of 7+: 1 - obtained by solution of the equation of state and Edmister relationship; 2 obtained from
correlations

Cubic equations of state are widely used in models applied to estimate PVT properties of natural
hydrocarbon systems. Peng-Robinson (PR) equation of state is the most spread. 2,3,4

The use of PR equation of state to describe density of fluid phase in natural hydrocarbon mixtures
results in major inaccuracy. At present, the approach based on "splitting" of phase equilibrium
simulation and phases density adjustment is widely used. This method is based on the correction
evaluation for the molar volume of the phase v calculated using the equation of state by value of c,
i.e. the "correct" molar volume is estimated as
using the following rule:

. For mixtures, this parameter is calculated

where N - number of components in the phase; xi - molar percentage of the i component in the
phase; ci - constant parameter for "pure" i component.
Please note that the phase volume correction keeps the result of modelled equilibrium phases
composition constant, i.e. the phase equilibrium should be calculated first, and then the values for
phase molar volumes are to be re-estimated.
Using the PR equation, Jhavery and Youngren5 recommend to calculate the value ci with the help of
the so-called "shift parameter", Si which is related to ci via bi factor in the following way: ci = Sibi.
Two methods are used to estimate critical pressure pc, critical temperature Tc, and "acentric
factor" of hydrocarbon fraction CN+ (where N = 6, 7,...):
1. Boiling point and density correlations under standard conditions. Here, Kessler and Lee correlations
are mostly used.
2. Equation of state to identify critical pressure. The idea is to use the available density data for the
given fraction under standard conditions psc, and its molar mass, M, with reference to the specific
equation of state. In other words, fraction density calculated according to the equation of state
at sc = 0.101 325 MPa and Tsc = 293.1 (or 288.6 in compliance with SPE standards), should be
equal to its measured value. This means that to estimate only two out of three values (i.e. Tc, pc, ),
empirical equations should be applied, and the third value is defined from the equation of state.
Kessler-Lee and Edmister correlations can produce sharp "acentric factor" values' estimates for all
groups of hydrocarbons. Correlations for pc estimation were obtained for hydrocarbons < C20, thus,
their extrapolation is less reliable than the use of corresponding relationships for T c. Therefore, the
Tc estimation is carried out by applying one of the above correlations. Both Edmister formula and the
mentioned Kessler-Lee correlation can be used to calculate the "acentric factor". P c and values are
also calculated from the following system of equations: the equation of state for psc, Tsc, vsc = M/psc,
and the relationship used for "acentric factor" estimation.
III. Reservoir fluid PVT properties modelling while field development planning. For oil
reserves estimation, the values of formation volume factor (FVF) at initial reservoir pressure and the
density of stock tank oil (STO) obtained after multistage separation are used. The significant task here
implies the exact reconstruction of these estimation parameters in the set of data required for
hydrodynamic modelling of field development. Besides, it is necessary to reproduce the experimental
values of reservoir fluid saturation pressure and its dynamic viscosity at initial formation pressure.
Sequential "tuning" of modelling parameters is performed to solve these problems.
1. The reservoir fluid saturation pressure matching.
The solution can be achieved by two methods:

using a unique relationship between saturation pressure and binary interaction coefficient
methane CN+ (N = 7,...) fraction KCN4-CN+: the increase of this value results in growing
reservoir fluid saturation pressure

using unique relationship between mixture saturation pressure and the CN+ (N = 7,...)
fraction boiling point: reservoir fluid saturation pressure grows with the increase in fraction's
boiling point temperature.

2. The STO density matching (a reserves estimation parameter).

Here authors employed the following methods:

using unique relationship between STO density and "shift parameter" value of CN+ fraction
Sr,: increase in "shift parameter" value results in augmentation of STO density

using unique relationship between STO density and CN+ (N = 7,...) fraction density under
standard conditions: the STO density grows with the increase in fraction's density.

3. The reservoir oil FVF at initial formation pressure matching (a reserves estimation parameter).
Since methane is dominating the dissolved gas, its "shift parameter" can be used to match the
reservoir oil FVF.
4. The reservoir fluid dynamic viscosity under initial formation pressure matching. To produce
relationships of reservoir fluid dynamic viscosity vs. pressure at various gas content, which are used in
modern hydrodynamic simulators, it is necessary to apply an integrated approach towards both
experimental and calculated data. The Lohrenz-Bray-Clark (LBC) method 6 is most commonly employed
for such estimates. The obtained results strongly depend on critical molar volumes, Vcr, for the CN+ (N
= 7,...) group of components. Thus the Vcr CN+ value is a reliable parameter which allows to
adequately describe the dynamic viscosity of reservoir fluid.

Figure 2. Relationships of reservoir fluid

volume factor at various values
ol gas content vs. reservoir pressure
(a); saturated reservoir
luid gas content vs pressure (b); and
reservoir fluid dynamic viscosity
vs. pressure at various gas contents
(c):
p and p - initial formation pressure and
saturation pressure at reservoir temperature;
experimental data are marked by circles

IV. Case history. As an example, let us discuss the data obtained while of reservoir fluid properties
and STO density matching for one of the bed of the Yarainerskoye field. The reservoir fluid composition
(molar percentage): C02 - 0.01; N2 - 0.55; CH4 - 34.07; C2H6 - 2.04; C3H8 - 3.20; i-C4H10 - 2.35; nC4H10 - 2.87; i-C5H12 - 2.01; n-C5H12 -2.11; C6 - 4.09; C7+ - 46.70. Reservoir pressure and temperature
are 24 MPa and 78 C, respectively. The experimental value of reservoir fluid saturation pressure at
this temperature is 12.7 MPa.
Figure 1 (a-g) shows the relationships illustrating the influence of various modelling parameters on
estimated reservoir fluid properties and STO density. The value points shown in these graphs
correspond to accurate matching of experimental saturation pressure at reservoir temperature,
volume factor and dynamic viscosity of reservoir fluid at initial formation pressure, as well as STO
density.
Figure 2 shows calculated relationships of volume factor, gas content, and dynamic viscosity of
reservoir fluid vs.pressure, which could be used in hydrodynamic simulation of reservoir depletion
using the "black oil" model.
Now the following conclusions can be made:

The PVT relationships construction method for "black oil"-type hydrodynamic simulators has
been developed, which is based on experimental and theoretical data generalisation

The algorithm for sequential identification of parameters for the calculated reservoir fluid
model was proposed.

The practical importance of this research lies in the precise matching of reserves estimation
parameters: the Formation Volume Factor of reservoir oil, and the Stock Tank Oil density.