an integrated steady-state - International Geothermal Association

13th New Zealand Geothermal Workshop 1991 161
AN INTEGRATED STEADY-STATE
SIMULATION AND ANALYSIS PACKAGE
CALUM GUNN and DEREK FREESTON'
Geothermal Energy New Zealand Ltd, Auckland, New Zealand
'Geothermal Institute, University of Auckland, New Zealand
SUMMARY An wellbore simulationand analysis e is usual stead -state
wellbore simulators in that it also includes a number of that pvide anal tools the
engineer. These comprise: an automatic output curve prediction facility; a matching analysis m for comparin
simulated versus measured downhole pressure and temperature profiles; a module for
of
downholeprofile data lower bound and, a preprocessormodule to the steady-stateSimulator to ensure that the
wellhead or feedconditions entered into the ackage, such as fluidcomposition and enthalpy, are consistent with
to each other. e input and output stored a database, and of the results can be a
is applied to a case study of an optimal wellbore casing and size problem, and the
benefits of its are demonstrated.
INTRODUCTION
The main function of a wellbore simulator is, given wellhead
conditions or underground geothermal conditions at
the interfacewith wellbore, to estimatehow those conditions
change with depth down (or up) the wellbore when it is
discharging fluid. in the form of pressures,
temperatures, other are uced by the
at depths the Where the
wellhead conditions are the specified input conditions the
simulator calculates from the top of the well down to the
dee est from the reservoir. This is termed a
simulation.
where deepest feed
zone conditions are specified, the
a
deepest simulahon. A steady-state assumes
that the simulated conditions do not significantly v with
respect to time, and thus the discharge is
be
"stable".
Wellbore simulatorscan be used as tools for predicting the
behaviour of geothermal wellbore flow by allowing the user
to
associated with the well's
and mass flowrates,and feed zone
a series over a range discharging
mass can the prediction a well's
characteristic output or roduction curve (the of mass
flowrate However, s a
knowledge of the
far the well,
express mass flowrate as a function of the undisturbed
reservoir pressures and pressures within the wellbore at each
feed zone.
GUM and Freeston (1991) have that the validity of
the predicted well output curve is dependent on three man
considerations:
the accuracy of the wellbore simulator
the accuracy of the parameters entered as input data by the
user intothe
and,, the- applicability of the chosen drawdown
The accuracy of the wellbore simulator depends on the
validity of the correlations coded into the program. These
correlations typically include expressions for:
fluid in the presence of non-condensable gases
and solids;
transition boundaries;
void (vapour volume fraction in two-phasefluid);
and
friction factor.
The accuracy of the input parameters entered into such a
steady-state is dependent on:
the amount of data available to accurate1 build a
model the example, correct
location of feed zones and the fluid parameters at those
the accuracy of any availablemeasured data (for exam le,
the ranges of measurement error for any
deepest feed pressures, temperatures,enthalpies, fluid
impurity concentrations, and mass flowrates), and,
whether the well had reached a stable flow when the
were and
the of any estunated data (for example,
and liner roughness, and in cases diameter
may have beenreducedby
The greater the amount of accurate measured data that is
available, then the better is the model of the well that can be
into the simulator. Such models are often verified
h matching analysis, the comparison of simulated
and
profiles againstmeasured
an actualdischargeof the well. Various
mput parameters may require
in the simulator
before simulated behaviour may match data. For
example, often non-condensablegases and dissolved solids
content are represented by an equivalent and NaCl
content respectively. This means that a measured value of
or NaCl might not be the input value
for those parameters to adequately match the observed
conditions. parameters may require calibration until a
suitable value is found.
examplewould in well Casing and liner
diameterswhere the amount of unknown, but other
meters are available with a hig de of accuracy.
as Gunn et al. there is
little pomt matchmg simulated data measured data to
calibrate suchparameters when there are notable errors in the
measureddata.
This paper presents a computer package that integrates a
steady-state wellbore simulator with a number of other
analysis tools. These tools are designed to:
aid the in the suitabilityof measured data
far matchmg
allow user to perform matching analysis for
p
to
curves, and in to this procedure to
unknown parameters of the deepest feed zone's
drawdown where necessary.

162
Although the e cannot in itself address concerns about NaCl or content values to be entered into the simulator.
the a particular drawdown it
the user with a choiceof such relahonshipsto apply.
The resultant lower bound profiles can be displayed
Lower bound analysisis described in
Gunn et al. (submitted).
pre of e is of an ongoing study
it is to extend the
scope of the package to include the of transient
STEADY-STATE SIMULATION FEATURES
effects.
The module of the packa e is the steady-state simulation
SIMULATION AND ANALYSIS MODULES
includes the
features.
liquid, two-phase, and superheated steam flow.
The packa e includes five main simulation and analysis
Incorporates dissolved solids content by an equivalent
modules, are as follows.
percent of NaCl.
non-condensable gas content by an
Discharge Test Simulation
This is the core module of the packa e. It allows the
simulation of steady-state flow in a well at a
single mass flowrate, The features, assumptions and
module are discussed more
later of paper.
Preprocessing Fluid Composition and Properties
The input parameters for a discharge test include fluid
and fluid sition, some of which ma be
or other measurements.
values will not all be consistent with to
each other, and in some cases various parameters will be
Inconsistenciescan be due to measurement error,
but typical1 measured enthal y values may not be
appropriate to the resence or dissolved solids.
Inconsistencies may result related measurements
having been observed at different times. Wellhead and
deepestfeed conditions into the simulationmodule are
preprocessed to ensure that a consistent and/or
complete set of input is used in the discharge test
simulation. A is not allowed to proceed without
such a step having first been performed
Output Test Simulation
In the real world a number of actual discharge tests on a
geothermal well can be performed to the well's
characteristic output or
this is an
expensive o package a module for
such a series of discharge test
simulations produce an estimateof the output curve. Such
a procedure termed an out ut test As part of the
simulation the module can requested to calculate suitable
drawdown parameters at the dee feed zone.Results can
be presented
in a text format. An
example of procedure is the case study in the
final section of
Matching Analysis
As above analysis is a tool can be
used to calibrate the mput parameters to the simulator.
Measured downhole pressure and
can be
entered into the package to allow their subsequent comparison
against simulated results. The profiles can be compared
allow ' matching, or a simple
matchmg report can be produced.
Lower Bound Analysis
As noted above, observed downholeprofiles in discharging
wells can be subject to significant measurement error.Lower
bound anal is a for estimating the of
measured for to matching analysis.
method works by simply analysing the consistency of
measured pairs of two-phase pressure and temperature
measurements taken at intervals down the wellbore. The
analysis estimates the minimum C02 or NaCl content
for each pair to satisf saturation condition, If these
minimum values, or lower have considerable
the then the data can considered
to be of use for or flwd parameters.
In some cases such analysis can aid the
or
verification of feed zone locations, and also suggest calibrated
equivalentweight percent of
Estimates the effectof both NaCl and
on fluid liquid
of on fluid vapour phase
and on superheatedsteam properties.
Allows or deepest calculation.
Models secondary feed zones, and calculates fluid
compositionat feeds.
Allows linear, quadratic, or superheated steam
drawdown relationshipsto be specified at each
calculating mass flowrate.
Allows multiplechangesof casin and liner diameter,
roughness to be (A of standard casing
and liner sizes is included).
Models the effect of deviations in the wellbore from
vertical on the pressure,
Models the rock formation
and well
Provides results at each depth of interest for:
pressure; temperature; enthalpy; dryness; wei ht percent
of NaCl the liquid phase; pressure
flowrate; flow regime or fluid.type; and, frictional,
and
pressure
Provides
of simulated downhole pressure
and
SIMULATION ASSUMPTIONS AND
CORRELATIONS
The stead -state simulation module of the has been
and su uently the two
and by
ad n for superheated steam flow.
programsit has been assumed that:
flow is steady and
are
constant a depth
and
dissolved solids and non-condensable gases can be
resented by an equivalent NaCl content and eqwvalent
C content
Fluid Properties
The formulae evaluating fluid are those outlined
in Hadgu (1989). is considered toaffect:
overall fluid pressure, and the partial pressure of is
determined from the relationship by Sutton (1976) and
from Henry's Law (the reason for using Sutton's
relationship as a ainst that found
Dalton's
Law is GUM and Freeston (m
of the vapour phase only (from et
enthalpy of both liquid and vapour phases (from Hadgu,
1989);and
viscosity of the vapour phase only (from Pritchett et al.,
1981).
NaCl is considered to affect
pressure (from Haas, 1976);

163
4
Flow Regimes
Where:
wn factor, or steam
action, Friction
Calculations
Tests
DATA STRUCTURE
Well
e various
tions
described
Feed Zone

as this involves calculating two unknown drawdown
parameters (the "a" and parameters in equation (2)).
Measured Static Temperature Profiles
A single static downhole temperature profile, taken from
shut-in measurements,can be assigned toany well. Thisdata
can be used to takeaccount of heat in a dischargetest
simulation if the user so requires. Apart from the downhole
pairs of depths and temperatures, the static temperature
profile data type also includes heat transfer parameters such
as: rock ; rock and, casin thermal
e and,the to
the
Measured Discharge Profiies
Measured downhole pressure and tem profiles at
different discharging mass
measurements
taken an actual dischar e test, be to
well. This can be used matching analysis
simulated profiles calibrate discharge test, feed zone, or
geometry configurahonparameters.
SIMULATION METHODOLOGY
Wellbore simulators are likeany other computer program that
attempts to simulate some sort of complex physical
In the required there are a
number of assumptions have to be the program
itself such as the correlationsdiscussed above, and a number
of assumptions that need to be made in selecting the
a input parameters., The e
user. with tools for whether its
to the well bem modelled, and whether the
input parameters are like$ to be appro riate. Matching
analysis can enable user to veri$ the
and to calibrate input parameters.
measured data is available for a articular well then it
be used to gau e the
performance. The
of measured data should be
the package's lower bound analysis facility however.
In man cases measured pressure and temperature rofiles
under e conditions may not yet be The
simulator even be applied to a field where no wells
used to estimate discharge shut-in
measurements, and to assess the relative benefits say in
increasing the diameters of subsequently drilled wells.
However, as more and more measurements available,
these can be buildup the model of the
well and its associated parametersas entered into the ackage.
For le, secondary feed zone
more as may a drawdown expression
for the primary feed.
When no
data is available for corroborationof the
package's predictions it is important to that the power
of the simulator then lies in the differences between the
"what-if' that are being than in
the actual of those This is because
each will be based on the set of underlying
assumptions.
For exam le, a fictitious representative well can be easily
modelled a field for instance. "What-if' studieson such a
representativewell can aid decisions as tooptimal casing and
liner sizes, t other considerations. In such a case there
of course no substantiating data to assess the
rformance. However, the package can be
to the relative difference between a number
scenarios (as illustrated in the case study at the end of this
POTENTIAL APPLICATIONS OF THE PACKAGE
Because the package integrates a number of simulation and
analysis modules it has numerous applications. These
can be grouped into three
164
Modelling Existing Systems
These applicationsare generally relevant toinvestigating well
behaviour where a reasonable amount of corroborative
measured data is available, particularly actual downhole
dischar e profiles. This data could be used for calibrating
such parameters as secondary feed conditions or
drawdown parameters, and involve matching analysis. The
resultant well model could then be used to examine the
existing performance of the well. A t pical reason for
an existing system would be the
point depth. This is important for the
depth for chemical toreduce
at the flashpoint de th. The flashing depth is hi hly
dependent on the sition. The sensitivity
depth to levels of can be
Modelling Changes to Existing Systems
These a lications generally investigatechan es a system
where metershave been calibrated,
an attempt to predict the effects of a physicalchange in any of
those parameters. This could involveperforming ou ut test
simulations increases or decreases in
due to a number of factors, or example
casing and liner diameter reductions resulting from a
workover, or changes due
assing". This
latter case can occur which have a high
non-condensable gas content. This content may decrease
rapidly after the well is discharged for the first time, and the
simulator could estimate the
in
through an a
analysis of feed input
parameters to flud
Modelling Representative Systems
As mentioned above, a "fictitious" well, representative of all
or part. of a field's, reservoir conditions. (dependin on
homo eneity) can be modelled in the
Sensiuvity can then be performed to determine
which in ut parameters are the significant factors to any
in output. a
demonstrated in the case study a
well.
CASE STUDY APPLICATION TO ESTIMATING
OPTIMAL CASING AND LINER SIZES
This case study the application of current
simulator casing and for a
representatwe well. The study does not attempt to actually
optimizewellbore fluid production, but serves todemonstrate
how a wellbore simulatorcan be a useful part of the overall
process. The application of simulators to
the effect of wellbore on output curves
been examined by a number of authors Hadgu
and Freeston and Barnett (1989).
Representative Well Input Parameters
The raw data is taken the examplein the Tutorial Manual
of the current acka e (GENZL and University,
1991). is representahve a near steam
well. Wellhead conditions given below.
Pressure: 12bara
Dryness: 98%
Gas Content 1% by weight of total fluid
NaCl Content None
The actual well casing and liner confi on associated with
the above wellhead conditions is as fo ows.
Casing 9 (vertical)
620 - Casing 9 vertical)
1145- Liner 7" from
Because the wellhead conditions are almost-dry, a steam
drawdown
assumed for the major feed at the

well bottom (ie equation (3) .above. The undisturbed
pressure at depth (ie 128 m) is assumed to be
36.5 bara.
Two larger casing and liner
the urpose of im roving
the
are as
are suggested for
from similarwells to be
0 Casing 13 (vertical)
620 - Casing 13 from vertical)
1145-
10 from vertical)
Casing 18 (vertical)
400 - 620m Casing 13 (vertical)
620 - Casing 13 from vertical)
1145- Liner 13 (30"from vertical)
Output Curve Prediction for the Actual Casing and
Liner Diameters
The overall analysis procedure firstly involves fully
specifying the charactenstic drawdown relationship for the
major feed zone at In a standard wellbore simulator
this would normally be achieved by performing a
discharge test simulation based on the above wellhead
conditions and the associated casing and liner confi
resultantcalculated in the at e
the primary then manually used in
(3) to calculate the unknown drawdown factor . (The
drawdown exponent is typically assumed to be some
re resentative value. For exam le, Hadgu 1989) used a
of 0.75, as is used here. a real stu y it would be
to a sensitivity analysis on this parameter and
e most conservative results when any
subsequentconclusions).
At this with the drawdown relationship fully
a set zone ressure and mass flowrates could be
manually produced satisfy the expression over the range
of flowrates of interest. A series of discrete feed
simulations could then be individual1 rformed, and the
estimated conditions to the
to produce the output curve by using a separate graphing
The currentsimulationpackage inte allthese steps and
then them automatically. steps occur at
the input data stage rather than at the stage. These
steps are as follows.
Enter the well
data.
Assign the initial and liner configuration tothe well
eometry configuration) selecting the appropriate
given above from the standard library of sizes.
Assign a discharge test to the well encom the
specified conditions from and
referencmg the geometry
Preprocess the wellhead todetermineunknown
parameters such as enthal y and
step
performed by the ackage when the
test to
Assign an output test to the we . This data.
comprises: the.appropriate relationship
steam); the
pressure at the deepest
feed; the range of mass flowratesof interest; the
of ints required on the output curve; and, it
references discharge test assigned to the well in the
revious step.
output
is done simply
by the key from the output
test data entry screen).
The package then the initial
test and uses the results to determine
missing
packa e continues the
by a new test of
165
deepest feed for each of the mass
flowrates as part of the output test data.
Each of these new discharge tests is individually
processed to the consistencyof the conditions.
packa e letes the by running each of
these test and the wellhead
results are saved as an output curve within the database.
Output Curve Predictions for the Larger Casing
and Liner Diameters
Now that the well's drawdown relationshi is
various wellbore parameters canbe drawdown
relationshi s de ndency on wellbore is assumedto
procedure for determining the new
output curves is as follows, and for each of the
two
geometry configuration to the with a
new set of and liner sizes.
Copy one of new discharge tests created by the
previous output test simulationand edit this to reference
or an theSent in the
using one the tests
y the ou ut test simulation is that it already
a set
zone
wellhead
could be copied, but more ta would be required
torelate it feed conditions.
the output test to a new output for the
we
included in this test. The new test the
previous step is now referenced, and the output test
. simulation rformed over a new range of
flowrates. this-range would
wide, as simulahon is performed it not be
known what the upper for mass flowrate might be.
r limit occurs when the flow "chokes" the
packa e the current discharge test
simulation
Results for the Representative
The first output test simulation for the representative well
using the original smallcasing and liner sizes was perfomed
over of mass flowrates from 10 to22.5 and six
discrete points on the output curve were requested. The two
ut test simulations using the larger sizes were both
over a flowrate range of 10 to 40 and
seven pomts on the output curve were requested.
The ackage allows the results of a number of output test
to be lotted against each other on the same
ph. The mass
feed zone pressure (ie the
relationship) can also be plotted along with the
output curves. The results as produced b the packa e are
shown in Figure 1. The choked in caseat
Interpretation of the Results
The improvement
between the original casing
and liner and the first set of gested new
sizes is readily apparent. At a typical
of
around 15bara the increasein mass is close to 50%.
By contrast the production increase between the first set of
new sizesand the est sizes is only an 5%or so.
Should these results associated an anal of
then could be shown this
increase
worth the
The package's presentation of the mass flowrate against feed
zone on the same as the ou ut curve is a
as tothe ible gains in Where
the pressure difference the output curve and feed
zone is relatively small, as is the case when comparin the
two sug
this example,
cannot %e a large in The feed zone
pressure curve provides an upper bound to the ou ut curve,
As and diameters there be a cutoff
point above which no more may be extracted
the well.

EX1 Actual size output curve:
size out ut curve:
+ Largest size curve:
o - Feed zone moas flowrate pressure
Figure 1: Comparison of the representative well's drawdown relationship against output curves for various
geometry configurations (Graph from WELLSIM Version 1.0)
This study demonstrates thepotential use of thepackage
to highli hting where an additional increase in wellbore
only a marginal effect on the production. As
noted above, the featuresfor such a representative
well study is to look at the relativeproduction increases rather
than the magnitudes. Although only two sets of geometry
configurations have been considered in this example, the
number of alternatives that the package can examine is not
CONCLUSION
A new computer package has been developed, designed to
incorporate a steady-state simulator with various analysis
tools to aid the user in assessing the simulators rformance
and assessing the consistency of any entered
into the program. Three major application areas for the
package have been identified terms existin
systems, changes to existing systems,
modelling or systems. The facility
of ou ut curve has been illustrated for
an application o the latter type, and the has been
shown be a powerfulaid for scenarios
parameters.
ACKNOWLEDGEMENTS
The behind the develo ment of the simulation and
anal sis modules discussed in aper has been supported
by from the New Zealand foundation for Research,
Science and Technology. Grateful thanks are extended to
Energy New Zealand Limited for the permission
to publish this paper.
REFERENCES
Barnett, B. (1989). A theoretical stud of the effect of bore
diameter on well outputs. Proc. th Workshop,
Auckland University,pp. 181-187.
Bjornsson, G. and Bodvarsson, G.(1987). A multi-feedzone
wellbore simulator. Geothermal Resources Council Trans.,
11, 503-507.
GENZL, and Auckland University (1991). WELLSIM
TutorialManual.
C. and Freeston, D. (1991). of
geothermal inflow performance and
relationshipsto wellbore
Resources Council Trans.,
ut curve prediction. Geothermal
01. 15, (in press).
Gunn, C. and Freeston, D. (in pre The a plicability of
two standard for the of on
geothermal fluid properties.
C., Freeston, D. and Had T. submitted).
Principles for wellbore simulator
calibration
using matching analysis. Submitted to
Haas Jr. J. 1970). An equation €or the of vaporsaturated
Na& solutions from 75°C to 325 C. Am.Jnl.
of Vol.
Haas Jr., J. (1976). Physical properties of the coexisting
phases and
of the H20 component
of boiling NaCl solutions. U.S.Geological Survey Bull.,
1421-A.
,T. and Freeston,D. (1986). Application of wellbore
to management problems for a eothermal field
case studies. Proc. 9th Australasian Mechanics
Auckland University,
Had T. and Freeston, D. 1990). A multi- urpose
simulator. Geothermal Trans., 14,
1279-1286.
Kanyua, J. (1985). The Vertical Flow
Liquid Separator and its A plication to the Gases
from Geothermal
thesis, University of
Auckland.
Pritchett, J., Rice, M.
Statefor Water-CarbonDioxide
Baca Reservoir. Report for
of New Mexico and
Mexico.
Sutton, F. (1976). Pressure-temperature curves for a
hase mixture of water and carbon dioxide. Jnl. of
19,297-301.