Professional Documents
Culture Documents
150 Õ Vol. 123, JUNE 2001 Copyright © 2001 by ASME Transactions of the ASME
Overview of Simulator
Based on the recommendations of Forsdyke 关16兴 and Apte 关17兴,
a computer program based on the molecular diffusion theory was
developed for simulating paraffin deposition in multiphase flow
environments. An overview of this simulator presented in the
forthcoming includes coupling of multiphase fluid flow, solid-
liquid-vapor thermodynamics, multiphase heat transfer, and flow
pattern-dependent paraffin deposition. This steady-state simulator
is modular in structure, and each module makes computations for
a particular aspect. The modular structure of the simulator will
permit easy modifications. For example, other multiphase flow
simulators like OLGA™ could be implemented.
The simulator consists of four main modules that perform the
following tasks:
• multiphase flow hydrodynamic calculations,
• multiphase flow heat transfer calculations,
• wax and fluid thermodynamic calculations, Fig. 1 Flow chart for the wax deposition prediction program
• wax deposition rate calculations.
Modern hydrodynamic multiphase mechanistic models are used
tic model by Kaya is used to predict flow patterns, pressure gra-
to calculate the pressure gradient and to identify the flow pattern
dients, and liquid holdups for pipe inclination angles from ⫹30 to
existing in a section of the pipeline or wellbore. Depending on the
⫹90 deg from horizontal.
flow pattern predicted by these mechanistic models, the heat trans-
In the absence of an accurate mechanistic model for inclination
fer module then predicts fluid temperatures along the pipe seg-
angles from ⫹15 to ⫹30 deg, the mechanistic model developed
ment. The thermodynamic module developed by MSI 关18兴 is used
by Kaya has been chosen for prediction of flow patterns, pressure
to predict the wax concentration and properties of existing phases
gradients, and liquid holdups in the simulator.
at specified temperatures and pressures. The wax deposition mod-
ule then predicts the wax deposition rates for the prevailing flow Heat Transfer Calculation Module. The thermal environ-
and temperature conditions. A time step is then taken and the ment and the temperature gradient in a hydrocarbon-carrying
thickness of wax deposited in each pipe segment is calculated at pipeline are of paramount importance in predicting wax deposi-
the end of that time step. This calculation sequence is repeated, tion rates. The heat transfer calculation module utilizes a steady-
taking into consideration the newly deposited wax layer at the end state, one-dimensional flow, energy conservation principle. The
of each time step. The calculation procedure is continued until the two-phase mixture of natural gas and crude oil 共also referred as
given simulation time is reached. The overall flow chart for this ‘‘multiphase mixture’’ hereafter兲 is considered homogenous, re-
program is given in Fig. 1. Each calculation module is explained gardless of the flow pattern. An overall heat transfer coefficient
in detail in following sections. relationship is used to quantify the heat loss to the surroundings.
In the absence of external work, the differential form of the
Hydrodynamic Calculation Module. The pipeline or well-
energy equation for a steady one-dimensional flow for any process
bore under consideration is divided into a number of pipe seg-
can be expressed as
ments and the flow conditions in each of these segments are
evaluated. The hydrodynamic calculation module uses two dq⫹dh⫹ v d v ⫹gdz⫽0 (2)
mechanistic models developed at the Tulsa University Fluid Flow
Projects, viz., Xiao et al. 关19兴 for horizontal and near-horizontal Expressing specific enthalpy h, and heat-added q as heat per
pipelines, and Kaya 关20兴 for inclined and vertical wellbores. unit mass, and elevation z in terms of pipe length L and inclination
The flow patterns encountered during multiphase flow of liquid angle , the enthalpy gradient along the pipe length can be written
and gas in horizontal and near-horizontal pipes are categorized as as
stratified 共smooth and wavy兲, intermittent 共elongated bubble and
slug兲, annular, and dispersed bubble. The mechanistic model by
Xiao et al. is used to predict flow patterns, pressure gradients, and
dh
dL
⫽ ⫺冉 WT 冊冉 冊
U i d i 共 T o ⫺T g 兲
⫺ v
dv
dL
⫺ 共 g sin 兲 (3)
liquid holdups for pipe inclination angles from ⫺15 to ⫹15 deg. The change of enthalpy with respect to length is a function of
The flow patterns encountered during multiphase flow in in- the enthalpy change due to the heat loss to the surroundings, ac-
clined and vertical wells are categorized as bubbly, dispersed celeration, and elevation change, respectively. The acceleration
bubble, intermittent 共slug and churn兲, and annular. The mechanis- term is neglected. Slippage does not have an effect on the mixture
冋 冉 冊 冉 冊 册
⫽⫺ (7) calculated using the equations described in the foregoing
dr 1 dw di dw do dw two modules. Figure 2 shows the algorithm used for these
ko ⫹ ln ⫹ ln ⫹
␣ TP 2k w dw 2k p di ␣ od o calculations.
The temperature at this interface is derived from the following
overall heat balance at the interfacet: Table 3 Test section input data
U id i
T w ⫽T o ⫺ 共 T ⫺T g 兲 (8)
␣ TPd w o
The effective inner pipe diameter, d w , is then calculated with
the knowledge of the wax thickness, ␦, using
d w ⫽d t ⫺2 ␦ (9)
Table 1 identifies the correlations chosen to determine ␣ TP for
Fig. 4 Comparison of wax thickness buildup profile in test section for Test A
Thermodynamic Calculation Module. The thermodynamic a look-up table for the transport properties of the fluid phases
module was developed by MSI 关18兴. It calculates the solid-liquid- under investigation.
vapor phase behavior of hydrocarbon systems and predicts cloud The pressure and temperature iteration loops shown in Fig. 2
point and solid ‘‘wax’’ fraction as a function of temperature and are executed to obtain the pressure and temperature at a given
pressure. It is based on the method described by Erickson et al. node. Wax concentration at that node is then obtained from
关1兴 and is fully compositional. The method assumes that the wax a paraffin look-up table, generated using the thermodynamic
phase is comprised essentially of normal paraffins. Output from module.
this module is in the form of a two-dimensional table of paraffin
concentration gradients (dC w /dT) as a function of pressure and Wax Deposition Calculation Module. In the absence of ad-
temperature. The thermodynamic module is also used to generate equate understanding of deposition physics during multiphase
Fig. 6 Comparison of wax thickness buildup profile in test section for Test B
flow, molecular diffusion is considered the primary mechanism. tional to the concentration gradient of wax in solution (dC w /dT)
Based on the recommendations of Forsdyke 关16兴 and Apte et al. and the radial temperature gradient at the wall (dT/dr). The con-
关14兴, the single-phase approach given by Eq. 共1兲 is used as a first centration gradient of the wax phase is obtained from the thermo-
approximation towards predicting paraffin deposition buildup dynamic module. The radial temperature gradient at the wall is
rates in multiphase environments. obtained from the heat transfer module. The diffusion coefficient
It is assumed that all the wax that moves to the pipe wall by the is calculated using the empirical correlation of Wilke and Chang
diffusion mechanism deposits on the wall. The pipe wall is as- 关4兴.
sumed fully wetted, except in the case of stratified flow, where
deposition is assumed to occur only on the wetted inner surface of Incorporation of Findings of Lund †9‡. Lund 关9兴 showed
the pipe. that important factors that must be considered during simulation
According to Eq. 共1兲, the rate of deposition of wax is propor- of wax deposition buildup rates are shear stripping, changes in oil
Fig. 8 Comparison of wax thickness buildup profile in test section for Test C
content of the deposit 共which changes the deposit thermal conduc- Simulation of Experimental Tests
tivity兲, and pipe wall roughness. Lund concluded that for simula-
tion of wax deposition during single-phase flow, the commonly The computer program was used to simulate the preliminary
used diffusion coefficient correlations 关4,5兴 significantly underpre- wax deposition tests conducted during the experimental phase of
dict deposit thickness for high flow rate cases. Diffusion coeffi- the JIP 关29兴. The oil/gas mixture inlet temperature in the test
cients must be multiplied by 5 to match experimental data. Creek section was 105°F and the glycol inlet temperature was 60°F. The
关6兴 accounted for shear stripping by using the shear dispersion duration of all the tests was 24 h. The predictions of the simulator
formulation with a negative dispersion coefficient. were tuned to the preliminary experimental results. Using the
The foregoing findings were incorporated into the multiphase compositions reported by Marathon Oil Company 关30兴, specific
simulator by providing options for the user to input multipliers for input files for the test fluid were created.
the oil fraction in the deposit, thermal conductivity of the deposit, The test section used in this study was simulated as a set of five
pipe wall roughness, and a choice for a shear-stripping coefficient. pipe segments using the details shown in Table 3. The outside