Вы находитесь на странице: 1из 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267459122

Well Test Rate Conversion to Compositional Wellstream

Article · March 2013


DOI: 10.2118/164334-MS

CITATION READS

1 731

2 authors, including:

Curtis Whitson
Norwegian University of Science and Technology
105 PUBLICATIONS   1,443 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Curtis Whitson on 15 November 2014.

The user has requested enhancement of the downloaded file.


SPE SPE-164334-MS

Well Test Rate Conversion to Compositional Wellstream


Mohammad Faizul Hoda, SPE, Petrostreamz AS, Curtis Hays Whitson, SPE, NTNU, Petrostreamz AS

Copyright 2013, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Middle East Oil and Gas Show and Exhibition held in Manama, Bahrain, 10–13 March 2013.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Well test rates are measured with test separators operating at varying pressure and temperature during initial production
tests for a new well and during periodic tests used to monitor well performance and allocate multi-well production separator
rates.
Converting test separator gas and oil volumetric rates to a common (fixed) set of surface separator conditions is useful to
ensure consistent history matching and rate allocation, development of valid inflow performance relations, and correlating well
performance changes over time. This paper provides a method to convert reported test rates to a molar compositional
wellstream rate. This compositional wellstream will exactly reproduce the reported test rates at the separator conditions
prevailing when rates were measured. The compositional wellstream rate can then be re-processed through a fixed set of
separator conditions to provide total surface gas and stock-tank oil rates.
The requirements for converting test rates to a compositional wellstream include: (1) an appropriate EOS model, (2) an
estimate of the wellstream composition – the “seed feed”, (3) test separator volumetric rates, and (4) test separator conditions
of pressure and temperature. The seed feed is usually the previously-determined wellstream composition from an earlier test.
The seed feed is flashed at test separator pressure and temperature, resulting in equilibrium gas and equilibrium oil
compositions. These equilibrium compositions are recombined in a ratio that yields exactly the test separator gas-oil ratio with
the EOS model, thereby yielding wellstream composition. Properties of the separator phases are used to convert separator
volumes to moles and thereby wellstream molar rate.
Sometimes a similar conversion is needed for volumetric gas and oil rates resulting from a multi-stage separation process.
An important example is conversion of black-oil rates to compositional wellstreams, where the black-oil PVT tables have been
generated from a multi-stage separation process.
Finally, we discuss the processing of compositional wellstreams through a higher-level (group, field, or asset) process
facility that takes into account the total feed composition of the process facility, and how to back-allocate well contributions of
total product rates.

Introduction
Well test rates are standard measurements used by operators to quantify the performance and behavior of individual wells.
Individual wells do not flow directly into field processing facilities, rather many individual wells flow into common flowlines
and pipelines that gather the streams and feed them to input manifolds for further processing. Well-group separation units may
send gas and liquid products separately to higher-level gathering systems for further processing, with an almost endless
labyrinth of stream splitting possibilities, mixing, and processing to the final sales products – gas and oil (and possibly gas
liquids).
Wells commingle at inlet manifolds, with each wellstream contributing its own complex mixture of oil, gas & water.
Continuous rate measurement at the well is usually restricted to multi-phase flow meters, providing estimates of volumetric
rates for gas, oil, and water. The metered volumes must be converted to molar compositional streams which can be used in
three ways: (1) re-processing the wellstreams to “equivalent” processed gas and oil rates for a well-defined separation process,
(2) combining with other wellstreams to provide higher-level compositional molar streams for processing at higher-level
processing facilities, and (3) back-allocating the higher-level gas and oil volumetric rates back to individual wells.
Multi-phase meter volumetric rates are notoriously uncertain, and compositional molar rates based on metered rates are
proportionately uncertain. To minimize this uncertainty most operators use periodic separator tests of individual wells, routing
a well’s production to a test manifold with a dedicated multi-phase separator unit used to measure gas, oil, and water well rates
2 SPE SPE-164334-MS

during a period of six to 24 hours. Regulatory bodies often require periodic individual-well testing, with test frequency ranging
from monthly to annually – mostly depending on the time variation of gas-oil ratio and water cut. Well tests provide invaluable
data to check and correct in-line multi-phase metering rates. Metering corrections are often assumed invariant from one test to
the next when forecasting rate behavior and performing daily or monthly well-rate allocation.
Several measurement devices are installed on the gas, oil and water outlets of test separators, most importantly those
providing volumes of gas, oil and water, including separator temperatures and pressures. Some additional test separator
properties that may be measured are summarized in Table 1. Most measurements are recorded at intervals of 15 to 30 minutes,
while some properties are measured only once during the well test. More-frequent measurements may be reported during well
tests on an exploration well.
The shrinkage factor (SF) is the fraction of metered separator oil rate that remains (transforms into) stock-tank oil after
further processing down to standard conditions of 1 atm and 60oF. The shrinkage factor may be measured, albeit with some
uncertainty in the field, or it may be estimated by correlation using the separator pressure and temperature – e.g. using
Standing’s saturated oil formation volume factor (FVF) correlation. The magnitude of shrinkage factor can range from 0.99 to
<0.75, depending mainly on the separator pressure and temperature and API gravity of the stock-tank oil. Particularly on
exploration well tests, the reported oil rates may not be labeled clearly, and it is uncertain whether they represent separator or
stock-tank volumes. Always ask the testing company to report both separator and (corrected) stock-tank oil rates, with proper
unit labeling to avoid uncertainty – e.g. separator-bbl/day and STB/day, instead of the ambiguous bbl/day often used.
An additional “meter factor” may also be used and reported, where this term is a mechanical correction and should be
based on meter tests. The meter factor is usually 1±0.05.
Fig. 1(a) shows example well test rate data, where gas and oil rates are plotted vs. time. Test separator temperatures and
pressures are shown in Fig. 1(b). Reported test oil rates may represent separator oil rate metered at separator pressure and
temperature, or they may have been “corrected” by the testing company to a stock-tank oil rate using a “shrinkage factor”.
Fig. 2(a) shows the computed gas-oil ratio (GOR) for the well test. The efficiency of separation of oil and gas in a
separator is dependent on the separator condition – i.e. the operating temperature and pressure, and incoming stream
composition. Changes in separator conditions can lead to significant changes in GOR, even for a constant incoming stream
composition. Changes in both separator conditions and incoming composition can lead to GOR variations, but may also lead to
a fortuitously-constant GOR. Disecting the effect of separator conditions and incoming stream composition is important, but is
often ignored.
Fig. 1(b) shows separator conditions varying considerably during the test. Part of the GOR variation is due to the changing
separator conditions. We would like to know the GOR during testing if separator conditions had remained invariant. Several
other reasons exist for expressing GOR in terms of a common, fixed surface process:

1. GOR reflects the producing wellstream (for constant separator conditions). Change in GOR indicates a change in
wellstream composition, which may indicate that the reservoir is (a) experiencing drawdown below the in-situ
saturation pressure, (b) producing from several zones, each zone containing a different in-situ fluid, or (c) coning
of gas into the perforated interval.
2. Monitor changes in well GOR over longer periods of time for history matching and production performance
forecasting.
3. Comparing well GORs throughout the field, indicating differences in reservoir performance and/or in-situ fluid
spatial variations.
4. Allocation of gas and oil to individual wells.
5. Compare well rates and well GORs processed individually, with rates and GOR “contributions” to the field
performance after field-wide commingled processing.
6. Study the sensitivity of surface process on rates and GOR.

This paper discusses each of the issues above and reviews recommended practices. We show examples using a consistent
method to convert measured test oil and gas rates to flowing compositional wellstream. When used with the appropriate EOS
model and well test conditions, this wellstream reproduces exactly the measured test rates, and it can be used to calculate gas
and oil rates for alternative surface separation processes. The wellstream compositional molar rates can be mixed, processed,
split and re-processed throughout the entire field complex, keeping track of each well’s contribution to total products produced
at any process facility within the asset.

Literature and Software Survey


One of the first SPE publications using a computer was by Muskat and McDowell (1949), showing how to do equilibrium
flash calculations for separator design optimization. Portman and Campbell (1956) discuss the effect of pressure and
temperature on the quantity of stabilized separator fluid for a given wellstream composition. They give a series of correlating
charts to predict the amount of stock tank oil from a given fluid. They acknowledge that the calculations may be solved by
computers but complain that such devices are not readily available and do not provide results fast enough. Statoil (1984)
provide detailed documentation on a procedure to perform “split-phase” process simulation to obtain accurate equilibrium
SPE SPE-164334-MS 3

phase data under precise pressure/temperature conditions and, from recombination calculations, to produce detailed and
representative wellstream compositions.

Table 1 — List of typical measurements made during a well test.

Purpose Location Number of measurements

Well head pressure and temperature Well head Usually every 15 minutes
Oil API Stock Tank Once
Choke Size Choke Manifold Once; more if changed during test
Separator pressure & temperature Separator Every time a rate measurement is made
Net/Gross Oil Rate Separator Oil outlet Usually every 15 minutes
Oil Meter Factor Separator Oil outlet -do-
Oil Shrinkage Stock Tank Infrequent
Gas Rate Separator Gas outlet Usually every 15 minutes
Residue/Free water Water outlet -do-
Water Rate Water outlet -do-
BSW Oil outlet -do-
Water Chloride Water outlet Once or less frequently
Water pH Water outlet Once or less frequently
Gas Composition Gas Outlet Infrequently /special request
Oil Composition Oil Outlet Infrequently /special request
Gas Gravity Gas Outlet Once

Strong, Thomas, and Bennion (1993) and a host of others discuss the recombination technique, but focus primarily on lab
recombination and on obtaining insitu representative fluid compositions. Singh et. al (2011) discuss the use of EOS to obtain
fluid compostions of different zones and layers without actual samples being available; measured oil-gas ratios (OGR) are
used to estimate wellstream compositions. Whitson and Sunjerga (2012) use an EOS-based recombination procedure in their
paper describing an integrated approach to modeling PVT in liquid-rich shales. They use surface separator information and an
EOS model coupled to the reservoir model to solve the problem of estimating in-situ fluid composition.
PhazeComp is a state-of-the-art, equation-of-state program for compositional phase behavior modeling and fluid
characterization. Its flexibility allows an unprecedented set of operations on fluid samples including mixing (recombining) of
equilibrium phases in a ratio that can be optimized to match the measured GOR (entered as data in PhazeComp). A data set
can be set up to use a feed estimate to simulate a well test (EOS Flash) at the conditions of the well test (p, T). It performs
regression to find the correct recombination ratio to obtain calculated rates that match measured data. More complicated
surface processes can be set up in PhazeComp. This procedure is ideal for computing the wellstream composition for a few
measured rates but is cumbersome if hundreds or thousands of rates from a database are to be converted.
PVTSim is a widely used EOS program in the industry. They have developed an Excel spreadsheet based solution driving
the PVTSim engine to correct a user provided fluid sample to the user provided measured oil & gas rates (GOR). Its
documentation suggests it is a regression based solution. The user provides the 1st stage separator well test conditions while the
2nd stage is assumed to be at standard conditions. The program allows upto five data points (measured oil and gas rates and the
corresponding separator (p,T) conditions).
Pipe-It/Streamz is software for integrated modeling and optimization. A standard module is used to perform the conversion
from well test rates to compositional wellstream described in this paper. This solution has been used extensively by the authors
to perform well test conversions since 2001. It is a non-regression procedure, described below, and the implementation allows
hundreds and thousands of compositional wellstream computations in a matter of seconds. All of the examples presented in
this paper were run using Pipe-It/Streamz.
Fig. 2(b) shows the excellent agreement between PhazeComp & Pipe-It, giving credence to the non-regression method of
Pipe-It/Streamz.
Vendors of multiphase meters often embed solutions for calculating “allocated” rates from well test rates, sometimes
without rigorous documentation. They might use an approach similar to that proposed here, but with a fixed “seed feed”
composition; else they might use empirical “conversion” factors to correct well test oil and gas rates.
4 SPE SPE-164334-MS

Proposed Method
The idea proposed here is to estimate the compositions of equilibrium separator gas ( ) and separator oil ( ) phases from
an isothermal flash calculation of some estimated “seed feed” wellstream composition ( ). Once estimated, the equilibrium
compositions and are recombined in some ratio of moles separator gas to wellstream moles ( / ) that yields the
actual test wellstream composition 1 . The choice of is made to ensure that the same GOR measured in
the actual well test is simulated when flashing composition . All flash calculations are made using an appropriate equation
of state (EOS) model.
Because equilibrium ratios / are not very dependent on overall composition at separator conditions,
, , , , the seed feed estimated composition is not so important to the accuracy of this method. The seed feed
can represent any reasonable composition of what is produced from the well, e.g. the initial reservoir composition. Typically
we use the most-recent measured or calculated wellstream as the seed feed.
The well test molar component rate is found by scaling the computed seed-feed separator gas mole fractions by the ratio
of measured gas rate to calculated seed-feed gas molar volume , and adding to the computed seed-feed oil mole
fractions scaled by the ratio of measured oil rate to seed-feed calculated oil molar volume :

(1)

with / and ∑ = the well test molar rate. The actual wellstream gas mole fraction is given by
/ . Molar volumes and are evaluated at the conditions where rate is measured – e.g. oil rate at separator
condition (p,T) and gas rate at standard conditions. Eq (1) holds exactly. There is no iteration involved.
The measured test separator oil rate must be reported at separator conditions1, and measured test separator gas rate does not
include additional gas released during depressurization of the separator oil to stock-tank conditions.
This method should use a reliable EOS model, and preferably one that applies to all wells in a field (for consistency).

Generalized Conversion for Rates from Any Surface Process. The same method, including Eq. 1, can be used to convert
reported gas and oil rates and that result from any surface process to a compositional wellstream. The seed feed is
processed using any calculation method that represents the surface process associated with the reported volumetric gas and oil
rates. The molar compositions and represent the gas and oil products2. The molar volumes and are evaluated at the
conditions where rates are measured (normally standard conditions).
A simple but common example is well test rates being reported at standard conditions, where gas rate is only from the
primary separator and oil rate is after separator oil has flashed down to stock-tank conditions in a second stage of separation.
A more complex example would be a field stream entering a complex process facility. Fiscal gas and oil rates are reported
daily at standard conditions. The surface process model might be Hysys and would provide the gas and oil product
compositions and , and molar volumes and . Eq. 1 would then be used to back-calculate incoming field stream
compositions and molar rates on a daily basis.

Seed Feed Dependency. The choice of seed feed estimate does not affect converted well test rates significantly, i.e. the
process of converting well test rates to gas and oil rates with a fixed separator process. Any reasonable seed feed estimate can
usually be used. The equilibrium phase compositions are most dependent on the separator conditions (p,T), so an accurate
record of pressure and temperature is needed together with rates reported at the same time.
Seed feed composition may be more important when (1) wellstream compositions have relative amounts of light and
intermediate components that change significantly, e.g. due to injection gas breakthrough, and (2) calculated compositions are
used for component tracking and modeling large process facilities that are sensitive to accurate compositional stream
description – e.g. modeling CO2 extraction and gas-liquid plants. These situations, however, would normally necessitate the
use of compositional reservoir simulation to make proper production forecasts.
One way to improve seed feed estimation uses compositional reservoir simulation results. When available, we use the
most-recent compositional run to generate a table of wellstream compositions, by well, versus producing OGR (for the surface
process used in the simulator). The tables provide seed feed estimates for doing well test rate conversions, or for converting
black-oil reservoir simulation well rates to compositional wellstreams.
Fig. 19(b) illustrates an example seed feed table created from a compositional reservoir simulator forecast. Usually only a
few points are needed because compositions often vary linearly with OGR, compared with GOR as in Fig. 20(a).

1
If separator oil carry over is suspected in the separator gas, this must be corrected for. If the stock tank oil rates are reported
instead of the separator rates, they need to be converted to separator rates using the reported shrinkage factor.
2
These surface gas and oil products will likely not be in equilibrium, so their mole fractions do not calculated true equilibrium
ratios but rather “process pseudo-K-values”.
SPE SPE-164334-MS 5

(a) Example well test rate data (b) Example test separator temperature & pressure

Fig. 1 — Examples Well Test data.

(a) Example well test GOR (b) Feed correction by two different programs

Fig. 2 — (a) An example of well test GOR. (b) Regression and non-regression based feed correction comparison

Examples
Examples given in this section are a combination of research / academic cases and adapted industry cases.

(a) Compositional wellstream; correction of seed-feed (b) GOR of wellstream processed via reference separator

Fig. 3 — Increasing observed GOR is acutally a decreasing GOR if the effect of changing sperator temperature is eliminated.
6 SPE SPE-164334-MS

Exampe 1: Synthetic reservoir simulation model (FBHP<dewpoint). Example 1 is a reservoir model producing a gas
condenstae below the dewpoint, with in-situ fluid showing compositional variation with depth. The model simulates a surface
test separator with a continuously increasing temperature, Fig. 3(b). If temperature variation was the only property changing
during the test, a continuously-increasing GOR would be expected. Compositional wellstream is calculated from an estimate
(constant, initial fluid) to match the reported GOR using the approach outlined in this paper. Fig. 3(a) shows C1 and C7p of the
computed wellstream as a function of the test time. Different adjustments to the constant seed-feed is made corresponding to
the variation in the reported GOR, Fig 3(b). When the effect of the changing temperature is eliminated by processing this
wellstream through a reference separator at fixed temperature and pressure, the GOR is actually falling with time. This is an
example of application of the compositional wellstream calculation.

Exampe 2: Synthetic reservoir simulation model (FBHP>dewpoint). Example 2 is again a reservoir model producing a
Gas Condenstae, but this time above the Dew point. The model simulates a surface test separator with a continuoiusly
increasing temperature, Fig. 4(b).The surface well test rates display increasing GOR, with a periodically changing Shrinkage
factor, Fig. 4(b). Compositional wellstream is calculated from an estimate (constant, initial fluid) to match the reported GOR
using the approach outlined in this paper as shown in Fig. 4(a). When the effect of the changing temperature is eliminated by
processing this wellstream through a reference separator at fixed temperature and pressure, the GOR is constant with time.
Without such an adjustment using the computed compositional wellstream would have led to misinterpretation of the reservoir
dewpoint.

(a) Compositional wellstream; correction of feed estimate (b) GOR of wellstream processed via reference separator

Fig. 4 — Increasing observed GOR is acutally a constant GOR if the effect of changing seperator temperature is eliminated.

Exampe 3: Monthly Well Tests, Well 8, 6 & 7. Monthly well tests is common in most operating fields. This example
discussed a highly under-saturated fluid from a reservoir with high pressure and low temperature. Fluids display unusual near-
critical phase behavior. There is a large GOR variation vs. depth. Monthly well test data for three wells are available for a
duration ranging from one to two years. Duration of tests vary from a few hours to days to over a week. Fig. 5(a) shows the
abruptly and continuously changing separator condition during testing of well 8. Compositional wellstreams are computed
where the correction to C1, C3-6 and C7p of the seed-feed is shown in Fig. 5(b). Fig. 6(a) shows the GOR before and after
correction to the reference separator conditions in Fig. 5(a). Less variation in the GOR is observed after the effect of changing
separator conditions is eliminated. Fig. 6(b) shows the reported shrinkage factor, which changes abruptly at irregular intervals.
It is clearly inconsistent compared with EOS-calculated shrinkage.
The compositional wellstream computed for this well was processed via a “super efficient” C4-/C5+ process. Fig. 7(a)
shows the improved oil recovery from such a process (greater than 5% oil gain). Fig. 7(b) shows the GOR from such an
efficient process. Allocation of well test rates to a more efficient, realistic process is now possible.
SPE SPE-164334-MS 7

(a) Actual and Reference Separator Conditions, Well 8 (b) Compositional wellstream; correction of seed-feed Well 8

Fig. 5 — Varying Separator conditions at a real well test; Adjustments to the seed-feed to obtain the final compositional wellstream.

(a) Reported GOR corrected to a reference conditions (b) Reported shrinkage compared to EOS calculated

Fig. 6 — GOR correction to reference condition. Consistency of shrinkage factor compared to EOS calculated shrinkage.

(a) Well Test stock tank oil compared to oil from a more (b) Corresponding GOR correction for the efficient process.
efficient process

Fig. 7 — Application of computed compositional wellstream – increase in oil product from a real, more efficient process.
8 SPE SPE-164334-MS

Fig. (8) and (9) are similar plots for well 6 as discussed previously for well 8.

(a) Actual and Reference Separator Conditions, Well 6. (b) Compositional wellstream; correction of seed feed, Well 6

Fig. 8 — Calculation of Compositional wellstream for Well 6 of Example 3.

(a) A large GOR correction corresponding to the difference in (b) Shrinkage factor associated with Well 6.
actual and reference separator conditions

Fig. 9 — Compositional wellstream calculation for Well 6 of Example 3.

Figs (10) and (11) are similar plots for well 7 as discussed previously for well 8.

(a) Actual and Reference Separator Conditions at Well 7 Test (b) Compositional wellstream; correction of seed feed, well 7.

Fig. 10 — Compositional wellstream calculation for Well 7 of Example 3.


SPE SPE-164334-MS 9

(a) GOR correction for well 7. (b) Shrinkage factor for Well 7.

Fig. 11 — Compositional wellstream calculation for Well 7 of Example 3. Note that the combined effect if changing separator
conditions and wellstream composition gives a fortiously similar corrected GOR, for a period. Shrinkage factor analysis in (b).

Example 4: 3 DSTs. This example discusses conversion of 3 DST data to compositional wellstreams. Fig. 12(b), 13(b) &
14(b) show the result of correcting a constant (different for each case) seed-feed composition to the measured well test rates.
Correction is about 1 mole % for C1 in DST1 while it is close to 3% for DST 2 & 3. The corrected wellstream is processed via
a (common) reference separator operating at 60 psia & 60 F. The corresponding operating conditions during each test, the
reference conditions, the original GOR and the corrected GOR are shown in Fig. 12(a), 13(a) & 14(a). The Test GOR for
DST1 shows a decreasing trend, while the correct GOR is more constant. DST3 is a multi-rate test where the rate variation has
the effect of changing the operating conditions, Fig. 14(a). The (slightly) decreasing GOR is corrected to a more steady one.

(a) Actual and Reference Separator Conditions; GOR (b) Compositional wellstream; correction of feed estimate
correction.

Fig. 12 — Interpretation of DST1 of Example 4 by first removing the effect of varying separator conditions. Calculation of
compositional wellstream from well test rates is the first step in achieving this correction.
10 SPE SPE-164334-MS

(a) Actual and Reference Separator Conditions; GOR (b) Compositional wellstream; correction of feed estimate
correction.

Fig. 13 — Interpretation of DST2 of Example 4 by first removing the effect of varying separator conditions. Calculation of
compositional wellstream from well test rates is the first step in achieving this correction.

(a) Actual and Reference Separator Conditions; GOR (b) Compositional wellstream; correction of feed estimate
correction

Fig. 14 — Interpretation of DST3 (multi-rate) of Example 4 by first removing the effect of varying separator conditions. Calculation of
compositional wellstream from well test rates is the first step in achieving this correction.

Exampe 5: Synthetic reservoir simulation case. Example 5 is from a synthetic reservoir case to analyse the behavior of
well tests in a major North Sea field.

(a) Well Process Rates and GOR (b) Effect of Separator conditions on Process Oil Rates

Fig. 15 — Temperature conditions may have a huge effect on calculation of Process Rates.
SPE SPE-164334-MS 11

The purpose is to convert well test rates to actual process rates with the objective of allocation of total field or platform
production and to obtain correct production rates for history matching. Well tests typically use a single-stage separator, while
the full-process is typically very complex. Test separator conditions change for each well or for the same well at different
times. Originally-used conversion factors used by the operator were complex functions of psp, Tsp and the producing
wellstream compositions. Fig 15(a) shows the field production rates and GOR from the E300 simulator using GP-Tables. Fig.
15(b) plots the process rates for each test separator rate for different test separator operating conditions. Originally a table
covering the expected range of operating conditions was used to compute the conversion factor, by linear interpolation. Also
this table was based on a single compostion. Fig. 16(a) shows that a large error in oil rate calculations is possible when using
the constant “shrinkage” factor method. Fig. 16(b) shows the effect of varying temperature, where the use of common
shrinkage factor based on the initial composition could result in a 50% error.
The proposed method circumvents these problems by accurately computing the compositional wellstream and correcting
for both the varying temperature and processing through the actual process (GP-Table in this case, simulating the real process).

(a) Degree of potential error from use of contant “shrinkage” (b) Degree of potential error from use of constant “shrinkage”

Fig. 16 — The potential error obtained from use of a constant conversion factor to obtain process rates from test rates. The % error
increases alarmingly as a function of GOR. Although the quantity of oil may be smaller, a bigger chunk is lost.

Exampe 6: Online Meter application. This example is conversion of 1 month data from an online meter. The data is
recorded every 15 mins. Over 2500 data points (streams) with corresponding measured rates are converted in matter of
minutes to corresponding compositional wellstreams. Fig. 17(a) shows the reported GOR while Fig. 17(b) shows the C1 & C7p
molar percent of each corresponding stream. A single fixed fluid estimate is used and the degree of correction is also shown in
Fig. 17(b). These wellstreams are readily converted to any user specified (e.g. multi-stage) process or can even be sent to a
sophisticated process simulator for accurate product calculations.

(a) Reported GOR from a automated measurement system. (b) Compositional wellstream; correction of feed estimate

Fig. 17 — Not only periodic well tests but robust online rates measurement systems can utilize this method. Here almost three
thousand measured rates are converted to equivalent compositional wellstreams.
12 SPE SPE-164334-MS

Exampe 7: Adapted from Multi-Field Asset Allocation. This example deals with a two year historical well test data
involving close to a hundred wells. Multiple fields have dedicated EOS models. As part of a consistent well allocation solution
it is required to obtain compositional wellstreams for each well test. All possible data linked to a well is carried forward with
each well test stream, including well test rates. Fig. 18(a) shows the GOR for each well test covering a range of low to high
GORs. The method proposed in this paper is used to obtain the compositional wellstreams. A filtering scheme organizes each
well test processed with a different EOS-model, specific to the field to which the well belongs. Fig. 18(b) and 19(a) show the
C1 and C5p content of each wellstream, before and after correction. The deviation from the diagonal is an indication of degree
of correction. Each symbol in these two figures indicates the different field and consequently the different EOS models used.
In addition to using different EOS for different fields, each field also uses a specific lookup table to obtain the initial seed-
feed compostion, based on the well test OGR. This is setup once and used forever. It may be noted that this seed-feed estimate
is just the initial value that is used in the EOS Flash to duplicate the well test. The values correspond to the x-axis in Fig. 18(b)
and 19(a). The corrected compostions, from the method proposed, represent the values on the y-axis. To obtain this lookup
table the wellstreams of compositional reservoir simulation for each field were processed through a reference separator. The
composition was plotted against the resulting OGR values (circles) as shown in Fig. 19(b). Also plotted on this figure is a
piece-wise linear curve covering the average values (squares for C1 and triangles for C5p). This is done for each component in
the EOS model. This piece-wise linear approximation is prepared as the lookup table mentioned earlier. The reason for
choosing OGR instead of GOR is apparent in Fig. 20(a). Wellstream composition is a more linear function of OGR than of
GOR of the stream. We always recommend using an OGR-based table for use as a seed-feed approximator.

(a) Historical well test GORs for a multi-field asset. (b) Compositional wellstream; correction of seed feed.

Fig. 18 — Implementation of proposed method in a multi-field asset. Different symbols in Fig (b) represent different fields.

(a) C5p of Compositional wellstream; correction of seed feed (b) Generation of OGR based seed feed table.

Fig. 19 — Implementation of proposed method in a multi-field asset. Different symbols in Fig (a) represent different fields. Fig (b)
shows the use of reservoir simulator (compositional) output, processed through a reference separator. C1 & C5p Mole% is plotted as
a function of OGR. A 5-node OGR vs. compostion table is generated for the seed feed estimation.
SPE SPE-164334-MS 13

(a) C1 & C5p vs. GOR is highly non-linear and not


recommended for the linear interpolated seed feed table

Fig. 20 — Implementation of proposed method in a multi-field asset. Fig (a) illustrates why GOR is not used as interpolating variable.

Conclusions
1. This paper presents a method to convert volumetric gas and oil rates measured during a well test into compositional
wellstreams and molar rates. The method requires an EOS model, a coarse estimate of the wellstream composition
(“seed feed”), measured gas and oil volumetric rates, and separator conditions of pressure and temperature at the time
rates are measured.
2. The same conversion method can also be used to convert volumetric gas and oil rates from any source (well tests,
reservoir simulators, IPR-based well models) resulting from a multi-stage process of arbitrary complexity.
3. Once wellstream compositions have been estimated from the proposed method, a “common” surface separation can be
made to yield volumetric gas and oil rates at a fixed set of separator conditions. The GOR calculated from volumetric
rates with a fixed process will vary only if wellstream composition varies. This allows time-varying GOR to be used as
an indicator of changing wellstream composition that may reflect behavior in the reservoir such as production below
the saturation pressure or commingled production from layers with varying in-situ compositions.
4. With wellstream compositions and molar rates available for each well in a large-scale petroleum asset, compositional
streams can be aggregated, split, and processed to higher-level well groups, fields, and total asset products. Individual
well allocation is then possible, by component and by product (e.g. CO2, sales gas, NGL, stock-tank oil, and water).

References
Muskat, M. and McDowell, J.M.: "An Electrical Computer for Solving Phase Equilibrium Problems," Trans., AIME (1949)
186, 291-298.
Portman, W.E. and Campbell, J. M., U. of Oklahoma, 1956. Journal of Petroleum Technology May 1956, Volume 8, Number
5 Pages 59-62.
Statoil report, 1984, The Composition of Statoil (Norway) Gas Well Troll 31/6-6. NPD website.
PhazeComp A state-of-the-art, equation-of-state program for compositional phase behavior modeling and fluid
characterization. www.zicketch.com
Calsep documentation spreadsheet “GOR Correction OS Sample.xls”, part of default PVTSim installation.
Pipe-It Software for integrated modeling and optimization. Streamz, a unique and powerful program for petroleum stream
management that serves as the core element of the Pipe-It. www.petrostreamz.com
Strong, J., Thomas, F. B., and Bennion, D. B. 1993, Reservoir Fluid Sampling And Recombination Techniques For Laboratory
Experiments
Thomas, F. B., Stepani, E., Imer, D., and Bennion, D. B., 2007, Representing Reservoir Oil Using Recombination Techniques.
Singh, K., PERA, Mantatzis, K., OMV, Whitson, C. H., NTNU/PERA, and Benjemia, R., OMV, 2011, Reservoir Fluid
Characterization and Application for Simulation Study
Whitson, C. H., NTNU & PERA and Sunjerga, S., PERA, 2012, PVT in Liquid-Rich Shale Reservoirs
14 SPE SPE-164334-MS

Nomenclature

mole fraction of the actual wellstream that ends up in the gas phase
actual gas volumetric rate measured during the well test
actual oil volumetric rate measured during the well test
component equilibrium ratios
total moles in the wellstream
moles in the gas phase
well test molar component rate
EOS calculated seed-feed separator gas molar volume
EOS calculated seed-feed separator oil molar volume
composition of equilibrium separator oil
composition of equilibrium separator gas
composition of the actual wellstream
composition of “seed feed”, an estimate of the actual wellstream composition used in the proposed method

BSW basic sediment and water


DST drill stem test
EOS equation of state
FVF formation volume factor
GOR gas-oil ratio
OGR oil-gas ratio
p pressure
PVT pressure, volume, temperature
T temperature
SF shrinkage factor, the decrease in volume of oil going from the separator conditions to Stock-Tank conditions

View publication stats

Вам также может понравиться