Page 61 - Handbook Of Multiphase Flow Assurance

56 3. PVT and rheology investigation

Viscosity
A number of correlations exist for dead oil viscosity as function of temperature and density.
A summary overview of these correlations is provided in Bergman and Sutton (2007). Based
on over 9000 viscosity measurements from over 3000 oil samples, they proposed a correlation:
(
ViscositycP] = expexp 22 ( .33 − .194 ∗+ .00033 ρ 2 −( .32 − .0185 ρ)∗Ln (T LM + 310 ))) −1
)
[
∗
0
∗
ρ
0
0
ρ = density, [API°], T LM = log-mean temperature [°F] for fluid between inlet and outlet.
T LM = exp(average(Ln(T INLET ),Ln(T OUTLET ))).
A simplified correlation is proposed here for stock tank oil viscosity at 60 °F for initial esti-
mates, based on their correlation. Measured data should be used when available.
)
(
[
°
.
3
STOViscosity cP] = exp 194 . /ρ   API  /2747

Pseudocomponents and lumping
In order to speed up the compositional analysis reservoir simulation specialists lump mul-
tiple components together. For example, components C 12 through C 15 may be lumped into
a single pseudo-component C 12 –C 15 , etc. Properties of such pseudocomponents including
critical temperature, critical pressure, acentric factor, molecular weight, boiling temperature
etc. are calculated using the EOS tuning process. In early days of computer application for
reservoir simulation as few as three or four components were used, as C 1 –C 2 , C 3 –C 6 and C 7 +
in order to accelerate the vapor liquid equilibrium computation. Today using 15 pure compo-
nents and 10 pseudocomponents is not uncommon. When two or more zones produce into
the same well tubing, each zone gets characterized with a different set of pseudocomponents.
This progressively increases the number of pseudocomponents and decreases the speed and
accuracy of fluid property prediction. Typically five or more pseudocomponents, in addi-
tion to pure components (from C 1 to C 7 ) provide adequate ability to characterize a hydrocar-
bon fluid while maintaining reasonable computation speed. Usually software optimizes the
lumping and pseudocomponent selection automatically, but this can be changed if necessary.

Lumping for different fluids

It is preferred to have the same set of pseudocomponents for all fluids. Dedicated PVT
tools which are used for fluid characterization for reservoir simulation have the capability to
lump and tune different but similar fluids using the same set of pseudocomponents, which
is preferred because it improves accuracy of blended fluid properties and improves compu-
tation speed.
The accuracy of fluid behavior prediction improves substantially if the binary interaction
parameters or k ij are also supplied along with the properties of pseudocomponents and en-
tered in the PVT simulation software. The binary interaction parameters are the additional
adjustable coefficients in the equations of state which allow a more accurate prediction of
fluid properties in multi-component mixtures.
Flow assurance specialists usually receive fluid characterization information from the res-
ervoir engineers who used the fluid properties to model the multiphase flow in the reser-
voir. The range of temperatures of interest to reservoir engineers is usually different from
that of the flow assurance engineers. While XHPHT or extra high pressure high temperature