Professional Documents
Culture Documents
SPE-113650-MS - A New Method For Gas Wells Deliverability Potential Estimation Using Mini DST and Single Well Modelling
SPE-113650-MS - A New Method For Gas Wells Deliverability Potential Estimation Using Mini DST and Single Well Modelling
SPE-113650-MS - A New Method For Gas Wells Deliverability Potential Estimation Using Mini DST and Single Well Modelling
Introduction
Abstract
Deliverability testing of gas wells is based on theory of
This paper presents techniques for interpretation of Mini- transient and pseudosteady flow of gases (Lee, 1982).
Drill Stem Test (MiniDST) for establishing commingled Traditionally, different testing procedures like flow-after-
Absolute Openhole Flow Potential (AOFP) in deep water flow, isochronal and modified isochronal are used to
exploration wells in India. These gas bearing reservoirs estimate parameters required to provide deliverability
are vertically heterogeneous with high permeability. estimates.
MiniDST’s are conducted using the inflatable straddle The turbulent or non-Darcy flow effects close to the
packer system of wireline formation tester. A MiniDST wellbore, which appear as rate-dependent or non-Darcy
transient sequence consists of a single or multiple flow skin, requires gas wells to be tested at a number of rates
periods, induced using a downhole pump, followed by a with the above mentioned tests so as to be able to estimate
pressure buildup. The objectives of a MiniDST are the non-Darcy flow coefficient by separating the
sampling, estimation of reservoir properties such as mechanical skin component from the total skin factor (st ).
permeability (k), skin(s), radial extrapolated pressure (p*) All these multirate methods of interpretation require well
and estimating AOFP. AOFP is an important gas well tests of quite long durations (Horne and Kuchuk, 1988).
flow parameter and is used to determine the
commerciality of discovered prospects. We use a two step Kabir (2006) suggested a two step approach based on
approach in establishing commingled AOFP of gas wells. multirate transient drawdown tests, followed or preceded
First, we conduct a multiple station MiniDST run and by a buildup. Firstly, he estimates reservoir parameters (k,
interpret the data to estimate reservoir parameters (k, s, s, D and p*) with transient data, rather than doing the
and p*). We also compute non-Darcy flow coefficient (D) traditional deliverability calculation with four points.
using Swift & Kiel expression and then use an analytical Then he uses these parameters to predict future
pseudo-steady state equation to establish single point deliverability by forward simulations with an analytic
AOFP for each of the tested zones. Second, we extend tool.
routine forward modeling and incorporate features such as
scaled permeability data, rock types and hydraulic flow As opposed to multirate testing, a number of authors have
units through interpretation of Nuclear Magnetic shown that a single transient such as a flow or buildup test
Resonance (NMR) and wireline petrophysics, into a can be used to estimate reservoir parameters which can be
model. The model is built in two different ways. One is used to generate a simulated deliverability characteristic
based on numerical simulator and another based on of the well. Horne and Kuchuk (1988) concluded that
cumulative permeability-thickness product for the gas using automated type-curve matching on gas well single
bearing zones, using average reservoir pressure and drawdown or buildup test, in which simultaneous flow
temperature for the whole zone of interest. The success of rate and pressure measurements are available, can result
single well simulation has given us the capability to in determination of non-Darcy skin parameter along with
forecast total AOFP for multiple zones using commingled other reservoir parameters. Mishra and Caudle (1984)
2 SPE 113650
proposed a new method for deliverability calculations of MiniDST operations can be carried out both in open-hole
gas wells, which eliminated multi-rate testing by and cased-hole completions. In a cased-hole MiniDST
developing dimensionless Inflow Performance curves operation, the test objects have to be isolated by cement
(IPR) for stabilized non-Darcy flow in unfractured gas from adjacent formations to prevent crossflow (Hurst,
reservoirs. They characterized turbulent flow effect by a 2000) and casing-cement integrity has to be verified from
constant D-coefficient, evaluated using the expression cement bond quality logs. Then a depth correlation run
given by Swift and Kiel (1962). needs to be done. A 0.33m to 1 m interval within the test
object is perforated, preferably using deep penetrating
In this work we use a two step approach to establish gas charges. After this, the packers are set across the
well deliverability. First, we use single point transient perforated depth interval and then the flow and buildup
testing consisting of single flow and buildup, to establish operations follow.
reservoir parameters k, s and p*. This is done by
conducting Wireline Formation Tester (WFT) based one- Basic objectives of MiniDST are estimating reservoir
meter straddle-packer MiniDST. D-coefficient for tested parameters (k, s, and p*) and collecting reservoir fluid
zone is calculated using the Swift and Kiel expression. samples. But in addition to these, any MiniDST (for this
AOFP for the tested zone is then calculated as the positive matter, any testing operation) in exploratory gas wells has
root of the Forchheimer’s pseudo-steady state quadratic to be conducive to estimating AOFP. But in MiniDST,
equation, whose coefficients are functions of these flow periods are conducted using downhole pumps in
reservoir parameters, and drainage area and drainage which flowrate variations and the maximum flowrate
shape factor. Second, we build a well model to predict inducible (i.e. the maximum pumping speed) are limited.
combined well deliverability of all the gas bearing Therefore conducting a multi-pumping rate MiniDST (for
objects. We build the model in two different ways. One is estimating Non-Darcy coefficient, D) on the lines of
based on a numerical simulator and the other is based on stabilized multirate gas well testing is not feasible using
cumulative permeability-thickness product for the gas downhole pumps currently being used in WFT’s,
bearing zones, using average reservoir pressure and especially in the case of high permeability gas zones.
temperature for the whole zone of interest. These
techniques were implemented in deepwater exploratory Therefore, in applying MiniDST techniques for gas well
gas wells, where conventional four-point deliverability deliverability estimation, we avoid the method of
test is highly expensive due to exorbitant rig costs. stabilized multirate gas well testing and in place, apply
single point AOFP technique (Lee and Wattenbargar,
Single Point AOFP from MiniDST - Theory 1996).
Modern WFT technology is designed around a modular Single Point AOFP technique, as opposed to multirate
principle, i.e. different functional parts of the formation testing, requires either single flow period or a flow period
tester tool are divided into separate and independent followed by a buildup, depending on whether flow period
modules. Thus, there may be a single probe module, a transient can be interpreted or not. Flow period pressures
downhole pumping module, a fluid analyzer module, an and derivatives can be extremely noisy in gas reservoirs,
inflatable straddle packer module and various types of especially those having high flow capacity. We have seen
sampling modules. that while conducting Single Point AOFP tests using
MiniDST, flow periods are generally noisy; therefore
The straddle packer module has two packer elements that flow-buildup sequence is used. Many a times, two
are inflated to isolate a borehole interval of approximately sequences of flow-buildups are carried out for data quality
1 m to 3m. The entire borehole wall is open to the control purposes.
formation, so the fluid flow area is nearly thousand times
larger than with conventional probes (Joshi et al. 2007). The buildup transient is interpreted to estimate reservoir
parameters (k, s, and p*). The interpretation methodology
MiniDST operation using straddle packer comprises of a of a MiniDST pressure transient is same as that of any
cleanup period and an extended flow period, using other testing technique – flow regime identification based
downhole pump, followed by a build up period. Cleanup on rate normalized pressure derivative and a conventional
period is monitored using the downhole fluid analyzer Horner analysis followed by analytical modeling. The
module of WFT tool string. The extended flow period is pressure transient data is analyzed using a commercially
conducted at a constant pumping speed. After the buildup available well testing software package.
period, flow is again carried out and then samples are
collected. Buildup Pressure-derivatives are monitored and A single-point AOFP is estimated using the reservoir
interpreted in real time so as to determine onset and parameters,obtained from analysis of the MiniDST
development of radial flow regimes. transient, in the pseudo-steady state expression with
pseudo-pressure approximation for gas flow rate (Lee
and Wattenbargar, 1996), given as follows:
SPE 113650 3
q=
(
703 × 10 −6 kh p 2 − p wf
2
) bounded cylindrical reservoir under pseudosteady-state
flow contains the term pi initially and this term is
μ T z ⎡⎢ln⎛⎜ re r ⎞⎟ − 0.75 + s + Dq ⎤⎥ replaced by p only as an approximation (Lee, 1982).
⎣ ⎝ w ⎠ ⎦
In using Eq. (1), we assume a pseudosteady state flow
…………………………. (1) with a prior knowledge of the drainage radius, re, of the
well. With MiniDST in exploratory wells, these
Gas viscosity, μ and gas compressibility factor, z in conditions may not be met. Lee (1982) has discussed that
equation (1) are evaluated at average reservoir pressure, even in cases where the drainage radius of the well is
p and average reservoir temperature T . unknown; Eq. (1) can be applied assuming a drainage
radius. The drainage radius, re appears as a logarithmic
term in Eq. (1). Therefore drainage radius does not have a
The non-Darcy flow coefficient, D, is calculated based on major effect on flow rate computation. This has also been
the expression given by Swift and Kiel (1962): shown in a later example in the paper.
∑ kh = ∑ k i hi
i =1
.........................…................... (9)
Pressure [Psia]
h
Where,
n
To estimate composite AOFP, we will again use Eqs. (1) Equation (15) is then solved as a quadratic equation in a
through (8). Therefore we also need to compute average similar manner as discussed earlier and the positive root
pressure and average temperature for the zone of interest. of this equation for flowing pressure of 14.7 psia gives
composite AOFP for the entire zone of interest.
Formation pressures obtained from WFT single-probe
pretests at various depths across the zone of interest were An important assumption here is that damage skin, s, is
used in calculating the average formation pressure for the constant throughout the zone of interest. Also, drainage
zone based on the expression, radius may have to be assumed especially in an
exploration scenario. Therefore it is prudent to estimate
p=
∑ pΔh ……………………………………… (13)
composite AOFP for a range of possible damage skin
factors. Similarly, sensitivities may also be carried out on
∑ Δh drainage radii of the well.
q=
2
(
703 × 10 −6 (∑ kh ) p 2 − p wf ) thickness product obtained from the MiniDST stations as
shown in Table 2.
μ T z ⎡⎢ln⎛⎜ re r ⎞⎟ − 0.75 + s + D q ⎤⎥
⎣ ⎝ w ⎠ ⎦
Depth Interval Scaling factor
……………………………… (15) XX34.8 – XX77.2 m 1.25
coefficient for the zone of interest and is estimated using XX86.5 – XX99.5 m 1.25
net hydrocarbon pay thickness (h), net-thickness based
average permeability (k ) and net-thickness based Table 2: Scaling factors for NMR derived
average porosity ( φ ), as follows: permeability log for different intervals in the gas
bearing zone of well ‘A’
6 SPE 113650
Fig. 3 shows a section of the gas bearing interval with the 7500
NMR permeability log, scaled permeability log and re = 1177.5 feet
Elemental log analysis (ELAN) volumes (lithology and 6500
fluid saturations). Secondly, an industry standard well-log
AOFP [MMscfd]
based reserve summation application was used for 5500
calculating net-thickness, net-thickness based average
A
permeability and net-thickness based average porosity for 4500
the gas bearing zones. Only those zones were considered
for summation which had an effective porosity greater
3500
than 8% and non-clay intergranular water saturation lower
than 90% and scaled permeability greater than 1mD. B
Based on these cut-off parameters a net hydrocarbon pay 2500
thickness of 52.12m, average permeability of 413.5 mD
and average porosity of 28.5% were obtained. 1500
0 5 10 15
WFT single-probe pressure and temperature Damage skin [dimensionless]
measurements were used in estimating an average
reservoir pressure of 3406 Psia and average temperature Fig 4: Composite AOFP plotted as a function of
of 621.6°R based on Eqs. (13) and (14). Damage Skin, with and without rate dependent
skin. Curve ‘A’ shows AOFP calculated without
A D-coefficient of 2.4E-6 (Mscf/D)-1 was calculated accounting for rate dependent skin and curve ‘B’
based on Eqs. (16) and (17). Since damage skin varied shows AOFP calculated considering effect of
highly in this zone as obtained for the five MiniDST rate dependent skin.
stations therefore composite AOFP for the zone was
estimated as a function of damage skin ranging from an Composite AOFP was calculated both with and without
optimistic value of 0 to quite a high value of 15 (Fig. 4). rate dependent skin to understand the effect of this skin on
AOFP values for a 100 Acre drainage area (equal to a
drainage radius of 1177.5 feet). Fig. 4 shows that for
KTIM Scaled - KTIM ELAN damage skin values from 0 to 5, there is a steeper decline
Volumes in AOFP values when calculated without accounting for
rate dependent skin, as compared to AOFP values
calculated accounting for rate dependent skin.
s=0
3400
AOFP [MMscfd]
3350
3300
XX90 m
3250
3200
500 1000 1500 2000 2500 3000
Fig. 3: A section of the gas bearing interval of Fig 5: Composite AOFP plotted as a function of
well ‘A’ with KTIM permeability log, scaled KTIM drainage radius
permeability log and ELAN volumes.
SPE 113650 7
In a deepwater exploration scenario where rig days and Well ‘B’ encountered multiple gas-bearing objects. A
associated service cost are important components of MiniDST was conducted for testing the potential of one
exploration project cost, a single well modeling, based on of the objects. A DST was planned but could not be
well logs, and simulation can help in predicting flow carried out due to some operational reasons. However, it
potential of the well. was necessary to estimate flow potential for all the objects
of this well in order to determine the commercial viability
In the method discussed previously for composite AOFP
of the reserves.
estimation from cumulative-permeability thickness
product, simplified assumption was made in respect of We use the application described earlier in building a
using average pressure for the whole zone of interest. single well model. The application was carried out in the
Moreover multiphase flow could not be handled. Using a following broad steps:
numerical simulator helps in handling variations in
1. 1D modeling from ELAN and NMR log
pressure and multiphase flow. Multiphase flow is handled
through the use of relative permeability curves for 2. Permeability calibration
different phases present in the reservoir.
3. 3D static modeling
Moreover rock typing concepts can be used in
4. Dynamic flow simulation and AOFP estimation
establishing different average porosity and permeability
for different rock types rather than using a single average 5. Production modeling and well deliverability
value. Also, different sets of capillary pressure and estimation
relative permeability curve can be established depending
on rock-types.
1D modeling from ELAN and NMR log: Petrophysical
We use an interactive, workflow driven application to
interpretation results (porosity, permeability and fluid
build 1D petrophysical and 3D simulation model from
saturations) from ELAN were used to define hydraulic
wellbore data. The simulation model could be either
flow units and rock types within the interval of interest.
numerical (Schlumberger, 2006) or semi-analytical
Fig. 6 shows the well log section and rock types. The rock
(Gilchrist et.al. 2007). We simulate these models to first
types and flow units were identified based on the neural-
optionally calibrate the models with well test or
network technique given by Amaefule et al. (1993). Total
production data and then predict the well’s future
four rock types were identified based on this technique.
performance. Integration of existing core, pressure and
Rock types were identified as 0, 1, 2 and 3. Rock type 0 is
production/testing data in the model building and
the best quality reservoir rock, having highest sand
calibration phases provides a critical link between the
content. Rock type 3 has the highest clay content and is
wellbore and the reservoir, as well as validation.
set as non-reservoir rock. Rock types 1 and 2 are
Petrophysical interpretation results (total porosity, intermediate quality rocks in that order. Capillary
permeability, total fluid saturations and irreducible water pressure and relative permeability were not used as this
saturation) are used to define hydraulic flow units and was a dry gas reservoir with a thick gas-shale contact
rock types within the defined interval of interest. scenario.
Capillary pressure and relative permeability are input
Permeability calibration: In the absence of core derived
from either core measurements, or derived and calibrated
absolute permeabilities, the NMR based KTIM
internally using logs or empirical relationships.
permeability curve was calibrated to WFT measured
Reservoir fluid properties are based on PVT analyses if permeability in the following steps. MiniDST transient
available. Alternatively, standard correlations built into was analyzed to estimate effective-permeability to gas for
the application can be used. the tested zone. A scale factor was derived as a ratio of
this MiniDST effective-permeability to the WFT single
Building the 3D static model requires definition of the
probe drawdown mobility measured in the same zone.
expected drainage area, vertical interval of interest, rock
This scaling factor was multiplied to drawdown mobility
type based layering, and location of existing completions.
measured at different depths across the zone of interest.
Once the 3D grid is defined, different property-population
This resulted in discrete effective-permeability value at
techniques can be used for populating all of the cells in
respective depths. These WFT-effective permeabilities
the model with properties defined for each rock type. The
were used in calibrating the KTIM permeability log. Thus
application passes the 3D, property populated model to a
a continuous log of effective permeability to gas was
simulation engine where optimization (calibration) and
obtained (Fig. 6). Other calibration techniques are also
simulation are performed.
possible where core data is available.
8 SPE 113650
3D static modeling: A 3D model was built based on an 3. Since MiniDST is carried out at low flowrates,
assumed drainage radius of the well, rock type and the effect of turbulence is minimized. Therefore,
hydraulic flow units based layering, and location of the data is more amenable to interpretation.
perforations for local grid refinement. These were used to
4. Single well modeling technique in combination
define a rectangular grid around the well, with
with production logs can help in predicting the
logarithmic growth (Fig. 7). Once the 3D grid was
potential of even those zones which could not
defined, properties were populated in all the cells in the
contribute to flow during a conventional testing
model based on flow units identified earlier. Variable
operation due to operational reasons like
lateral and vertical properties were assigned within each
incomplete cleanup of completion fluid sump in
flow unit based on stochastic distribution of given
the wellbore, off-depth perforation, sanding etc.
property for each rock type.
Dynamic flow simulation and AOFP estimation: The
Nomenclature
application passed the 3D, property-populated model to a
numerical simulator in this case. Further calibration of D = non-Darcy flow coefficient, (Mscf/D)-1
this model with production data was not carried out due to
lack of sufficient production data. Moreover, the model D= non-Darcy flow coefficient based on average rock
has already been calibrated with MiniDST in step 2. properties, (Mscf/D)-1
Damage skin obtained for one of the MiniDST tested dm(p) = gas pseudo-pressure , Psi2 / cP
zone, was used throughout the wellbore for all the dm(p)’ = gas pseudo-pressure derivative, Psi2 / cP
perforations. For each connection in the well, a D- Bg = gas formation volume factor, cf/scf
cf = formation compressibility, Psi-1
coefficient is calculated by the simulator based on the
h = net pay thickness, ft
permeability and porosity of the connected grid block
k = permeability, mD
together with the fluid properties of the wellbore gas and
ki= permeability of ith layer, mD
is based on Swift and Kiels’s method described earlier. A
commingled AOFP of 149.5 MMscfd was obtained for k = average permeability based on net-thickness, mD
the well as the maximum gas flow rate prediction at a Σ kh = cumulative permeability-thickness product, mD-ft
bottomhole pressure target of 14.7 psia (Fig. 8). M = molecular weight of gas (gas gravity x 28.96)
pi= initial reservoir pressure, psia
Production modeling and deliverability estimation: p = average reservoir pressure, psia
Production scenarios were simulated from the 3D model Psc = standard pressure condition, 14.7 psia
under different bottomhole pressure targets to generate pwf= flowing bottomhole pressure, psia
the IPR curve for the well. This along with the tubing p* = extrapolated radial pressure, psia
intake curves for different choke sizes obtained from a q = gas flow rate, Mscf/Day
commercial pipeline simulator was used to perform nodal re = drainage radius, ft
analysis. A well deliverability of 27 MMscfd was rw = wellbore radius, ft
obtained for 1000psi pipeline pressure, 32/64 inch choke s = formation damage skin, dimensionless
and 3.5inch tubing (Fig. 9). T = average reservoir temperature, °R
Tsc = standard temperature condition, 520 °R
T1 = formation temperature at the top of the zone of
Conclusion interest, °R
The objective of well testing includes determination of a) T2 = formation temperature at the bottom of the zone of
reservoir parameters b) well potential c) well interest, °R
deliverability and d) PVT samples. We have outlined a z = real-gas deviation factor, dimensionless
methodology that successfully uses MiniDST to achieve β = gas turbulence factor
the above goals. This technique has a wide variety of β = gas turbulence factor based on average rock
practical field applications and benefits. These include: properties
γ = gas gravity
1. MiniDST technique is non-destructive to the
environment and can result in substantial cost φ = effective porosity, fraction
savings especially in deepwater offshore context. φ i = effective porosity of ith layer, fraction
2. Additionally, the risk due to possible failure is φ = average porosity based on net-thickness, fraction
considerably mitigated.
μ = gas viscosity at average reservoir pressure, cP
μwf =viscosity at flowing pressure, cP
SPE 113650 9
AOFP = Absolute Openhole flow potential, Mscf/Day Kabir, C.S. 2006. What is the Real Measure of Gas Well
ELAN = Elemental Log Analysis Volumes Deliverability? SPEREE April 2006: 126-134; SPE-84469-
IPR = Inflow Performance Relationship PA.
KTIM = NMR based Timur-Coats permeability Lee, J. 1982. Well Testing. Dallas: SPE Textbook Series.
Lee, J. and Wattenbargar, R.A. 1996. Gas Reservoir
NMR = Nuclear Magnetic Resonance Engineering. Richardson: SPE Textbook Series, Volume 5.
WFT = Wireline Formation Tester Mishra, S. and Caudle, B.H. 1984. A Simplified Procedure for
Gas Deliverability Calculations Using Dimensionless IPR
Curves. Paper SPE 13231 presented at SPE Annual
Acknowledgements Technical Conference and Exhibition, Houston, Texas, 16 –
19 September.
This paper is based on a project supported by ONGC and Schlumberger 2006. Eclipse Reference Manual
Schlumberger.We wish to thank them for their support. Schlumberger 2007. SWPM version 2.0 Users Guide.
Swift, G.W. and Kiel, O.G. 1962. The Prediction of Gas Well
We also wish to gratefully acknowledge Dibyatanu
Performance Including the Effect of Non-Darcy Flow.
Kundu, Mihira Narayan Acharya and Kapil Seth for their Trans., A1ME 225791-798.
contribution. Whittle, T. M., Lee, J., and Gringarten, A.C. 2003. Will
Wireline Formation Testers Replace Well Tests? Paper
We wish to clarify that the views expressed by the authors SPE 84086 presented at the SPE Annual Technical
in this paper are their own and do not necessarily reflect Conference and Exhibition held in Denver, Colorado,
any position of the companies concerned. USA, 5 – 8 October.
References
y e [ ]
1726
MD
Completions 0.00 PHIT 0.40 0.1 WFT-PERM 10000.0 Moved hydrocarbon RT_NN_HFU T2_DIST
0.1 KTIM 10000.0 Bound water
0.1 PERM_CAL 10000.0 Water
Gas
Special mineral 2
Special mineral 1
Biotite
Muscovite
17280
Orthoclase
Quartz
Clay 2
Clay 1
Montmorillonite
Kaolinite
Illite
17300
17320
Horizon 1
Horizon 1 Horizon 1
17340
rock 3
Horizon 2
Horizon 2 Horizon 2
rock 1
Horizon 3
Horizon 3 Horizon 3
rock 2
Horizon 4
Horizon 4 Horizon 4
17360
rock 3
Horizon 5
Horizon 5 Horizon 5
rock 0 Horizon 6
Horizon 6 Horizon 6
rock 3 Horizon 7
Horizon 7 Horizon 7
rock 0 Horizon 8
Horizon 8 Horizon 8
17380
rock 2
Horizon 9
Horizon 9 Horizon 9
rock 1
Horizon 10
Horizon 10 Horizon 10
rock 3
Horizon 11
Horizon 11 Horizon 11
17400
rock 0
Horizon 12
Horizon 12 Horizon 12
17420
7425
Fig 6: Shows well log section and rock types in the gas bearing zone in Well ‘B’. Figure total porosity
(PHIT), KTIM Permeability, Calibrated KTIM (PERM-CAL) based on WFT permeabilities, ELAN volumes
(lithology and fluid-saturations), Rock-types based on neural network and NMR T2 distribution.
SPE 113650 11
Rock-0
Rock-1
Rock-2
Rock-3
Fig 7: Rock types and logarithmic gridding in the 3D model of well ‘B’
Inflow
Performance
Curve
24’’ 32’’
40’’
Tubing Intake
Curves
Fig.8: AOFP prediction from single well model of Fig.9: Well deliverability from Nodal Analysis.
well ‘B’ with bottomhole pressure target of Figure shows Inflow performance curve and
14.7psia. tubing intake curves for different choke sizes of
well ‘B’. The intersection of these two curves
gives the well deliverability at respective choke
size.