Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

SPE 134265

A Comprehensive Model of High-Rate Matrix Acid Stimulation for Long


Horizontal Wells in Carbonate Reservoirs
K. Furui, SPE, R.C. Burton, SPE, D.W. Burkhead, SPE, and N.A. Abdelmalek, SPE, ConocoPhillips, and A.D. Hill,
SPE, D. Zhu, SPE, and M. Nozaki, SPE, Texas A&M University

Copyright 2010, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Florence, Italy, 19–22 September 2010.

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
Matrix acidizing models have traditionally under-predicted acid stimulation benefits due to under-prediction of wormhole
penetration and corresponding completion skin factors in vertical wells. For long horizontal wells drilled in carbonate reservoirs,
productivity enhancement is a function of acid placement and effective wormhole penetration. However, prediction of wormhole
penetration requires more effective analysis than provided by current industry models. This paper presents results of matrix acid
modeling work for horizontal wells and describes a practical engineering tool for analyzing the progress of matrix acid stimulation
for cemented and un-cemented horizontal well completions typically employed in carbonate reservoirs.
An integrated flow model has been developed to predict the wellbore pressure profile and wormhole distribution by tracking
the movement of the acid in the wellbore and the formation. Analysis of injection rates and pressures during acid treatment
provides engineers with a way to determine the varying injectivity and tubing friction as stimulation proceeds. The model
presented here can be used as a forward model for analyzing real-time treatment rate and pressure histories and can also be used to
review past treatments to improve future treatment designs.
The wormhole growth model is based on Buijse and Glasbergen’s empirical correlation augmented by the effect of formation
heterogeneity and scale-up procedures that extend the wormhole geometry and penetration from laboratory flow tests on small
cores to field-size treatments.
Application of this modeling shows acid wormholing through carbonate formations can provide significant stimulation
resulting in post-stimulation skins as low as -3.5 to -4.0 vs. previously predicted values in the -1.0 to -2.0 range. Using actual field
stimulation data, we also discuss key elements to successful stimulation planning and the diagnosis of matrix acid treatments to
achieve effective wormhole coverage for horizontal completions in carbonate formations.

Introduction
Over the past several decades, horizontal wells have become an important and widely used completion technology around the
world. They have proven to be excellent producers and injectors for thin reservoirs or for thicker reservoirs with good vertical
permeability. The larger reservoir contact area provided by the horizontal completion interval improves well performance by
increasing well productivity for a given drawdown. This type of high productivity typically results in improved long-term sweep
and recovery of reserves. However, horizontal well completion operations such as perforating, zonal isolation, and stimulation are
often more expensive and complicated than those performed in comparable vertical wells.
Successful acid stimulation of long horizontal intervals in carbonate reservoirs, whether using matrix or acid fracturing
treatments, requires effective acid distribution along the entire reservoir length. In attempting to achieve effective placement, acid
is generally distributed across the zones of interest using various diversion techniques such as: rate diversion using limited entry
perforating, injection/washing using coiled tubing, injection with or following a diverting material, use of sliding sleeves and
isolation packers, and injection from acid jetting tools. In addition to achieving effective placement of acid along the entire length
of the horizontal interval, it is important to be able to inject acid at sufficiently high rates to drive the acid wormholes deep into the
reservoir to attain low completion skin values. Determining required acid injection rates and volumes to achieve desired
stimulation/skin results is therefore a key design variable in horizontal completion design. Unfortunately, there is a great deal of
uncertainty surrounding currently available theoretical models and associated design procedures.
Buijse and Glasbergen (2005) recently presented a semi-empirical wormhole growth model to calculate wormhole penetration
in carbonate acidizing. This model can be applied to various combinations of acid and rock types and injection conditions by
fitting the model leak-off parameters to the results of laboratory experimental data. One of the major benefits of this model is that
2 SPE 134265

it is expressed by a relatively simple form as it requires only two empirical parameters, the optimum interstitial velocity, vi,opt, and
the optimum pore volumes to breakthrough, PVbt,opt which are commonly available from existing references or can be readily
obtained through laboratory testing. As can be imagined, this is a great advantage when planning field acid stimulation treatments.
Offsetting the simplicity of the Buijse and Glasbergen model, many researchers have reported that the optimum injection rate
measured in a linear acid flooding test cannot be translated easily to field conditions as evidenced by the core size dependence of
wormholing (Buijse 2000). These core size effects further complicate the application of the linear coreflooding test results to
typical well-scale acid flow geometries characterized by radial and spherical flow. In fact, theoretical and laboratory-based matrix
acidizing models have traditionally under-predicted acid stimulation benefits due to under-prediction of wormhole penetration and
corresponding completion skin values.
Fig. 1 shows historical carbonate matrix stimulation data obtained from conventional vertical and inclined wells completed in
various carbonate reservoirs in the Middle East and North Sea fields. This experience indicates good stimulation results, negative
skins in the -3 to -4 range, are commonly achieved across a range of conditions. Based on assessments of skin factors from a series
of buildup test analyses, wormhole lengths on the order of 10 to 20 ft at the average acid injection volume of 75 gal/ft are
commonly achieved. Alternatively, application of conventional matrix acidizing models available in the industry today predict skin
values in the -1 to -2 range, which is equivalent to wormhole penetration lengths of 1 to 3 ft. The large discrepancy between the
field data showing skins in the -3 to -4 range with corresponding 10 to 20 ft stimulation radii vs. current industry matrix acid
stimulation models which predict skin values in the -1 to -2 range with wormhole penetration radii of 1 to 3 ft is striking.
Obviously, prediction of wormhole penetration requires more effective analysis than provided by current industry models.

Carbonate Matrix Acidizing Results


Skin Data Measured in Pressure Build-Up Tests After Acid Stimulation
0
Middle Eastern Limestone Reservoirs
-1
North Sea Chalk Reservoirs
-2

-3
Skin Value

-4

-5

-6

-7

-8
0 10 20 30 40 50 60 70 80 90 100
Cumulative % of Wells with Skin Worse Than Stated Value

Fig. 1⎯Field post-stimulation buildup test data for carbonate matrix acidizing.

In this study, we propose a new scale-up procedure that extends linear flow wormhole models to radial/spherical flow
geometries considering the symmetrical wormhole patterns formed in radial and spherical pressure fields, appropriate fluid loss
from the walls of wormholes, and competition among wormholes for injection fluid in the field-scale treatments. Semi-analytical
flow correlations that estimate the in-situ acid injection velocity at the tip of the dominant wormholes are developed based on
extensive 3D finite element method (FEM) model flow simulations.
Prediction of wormhole penetration and corresponding completion skin factor relies on analysis of downhole injection
pressures at the formation face. This requires a detailed review and understanding of tubing friction changes during treatment. In
many cases, job friction must be predicted from surface treating pressures and ISIP data. In other cases, a downhole pressure
gauge may be available, however, this downhole gauge is typically set above the production packer, several thousands of feet
shallower than the injection interval for a long horizontal well. This tubing and liner length between the downhole gauge and the
completion interval also requires friction analysis. In addition, perforation friction between the inside of the production casing or
liner and the formation face must be analyzed to arrive at the correct reservoir treating pressure. When evaluating pipe flow
friction, the effects of friction reducer must be accounted for. Friction reducer has been widely used in high-rate acid treatments,
drastically reducing Newtonian pipe friction and allowing operators inject fluid at much higher rates. However, the effect of the
friction reducer also changes due to variations in concentration of friction reducer during the course of the acid treatment whether
by design or through pump control variations.
Fig. 2 shows acid injection rate and pressure measured for a typical matrix acidizing job through 4-1/2” 12.6 lb/ft tubing set at
11202 MD-ft. A downhole gauge was set at 10939 MD-ft above a production packer and 28% HCl acid was injected from the
surface at t = 0 minutes. The acid was initially injected at a relatively low rate to displace less dense, resident tubing fluid into the
formation. As the acid front moves down the tubing, the tubing pressure drop, calculated by subtracting the wellhead gauge
SPE 134265 3

pressure (WHP) from the bottomhole gauge pressure (BHP), starts decreasing due to the increased hydrostatic pressure of the acid
column. Once the acid reaches the reservoir, the injection rates are increased to 60 bpm (t = 50 minutes) followed by large WHP
and BHP increases. Notably this excess tubing pressure drops shortly after the injection rate increases (t = 65 minutes). The
pressure increase from t = 60 to 65 minutes was caused by an insufficient amount of friction reducer being pumped as the slurry
rate increased, resulting in sub-optimal friction reducer performance in the tubing. Often these tubing events mislead engineers
into thinking other well and reservoir problems such as fracturing, plug leak, or formation/perforation plugging etc are occurring.
For the example shown, once the friction reducer concentration reached the proper level, the tubing friction stabilizes. However, at
t = 77 minutes, the friction reducer pump encountered a problem resulting in significant increases in the wellhead injection
pressures. This pressure increase caused the operator to reduce the injection rate from 60 bpm to 53 bpm. At t = 88 minutes, less
dense, overflush fluid is injected from the surface. The tubing pressure increases as the overflush fluid is moving down the tubing.
As shown in this example, the tubing pressure varies significantly and it is very important to filter-out these hydrostatic and
friction events prior to evaluation of wormhole length and skin factor.

10,000 160

8,000 140

6,000 120

4,000 100

2,000 80

0 60

-2,000 40

-4,000 20

-6,000 0
0 20 40 60 80 100 120

Fig. 2⎯Tubing friction changes for a typical matrix acidizing treatment job.

In this paper, a wellbore-reservoir coupled flow model has been developed to predict the wellbore pressure profile and
wormhole length distribution by tracking the movement of the acid in the wellbore and the formation. Analysis of injection rates
and pressures during acid treatment provides engineers with a way to determine the varying injectivity and tubing friction as
stimulation proceeds. The integrated flow model presented in this work can be used for analyzing real-time treatment rate and
pressure data, and can also be used for post-job treatment evaluation.

Development of Acidizing Model for High-Rate Matrix Stimulation in Carbonate Formations


Wormhole Growth Model
As noted above, most laboratory experiments for wormholing are performed in linear core flow cells while radial and spherical
flow geometries are more common in the field applications. There are many laboratory experimental results published in the
literature showing PVbt,opt values ranging from approximately 0.5 to 1.0 for relatively homogeneous calcite samples under
relatively high acid concentration and temperature conditions. Fig. 3 shows linear acidflooding test results for high porosity
outcrop chalk. The results show that PVbt,opt ranges from 0.3 to 0.6 for 1” diameter × 6” long core samples. These results indicate
slightly more effective wormholing than those seen in typical calcite core test results (e.g., Indiana limestone). However, these low
PVbt,opt values are not quite efficient enough to explain the large stimulation radii seen in the field stimulation jobs.
4 SPE 134265

Linear Acidflooding Test Data for High Porosity Chalk


k =1 to 2 md, φ = 30 to 35%
100
15 wt% HCl @ 150 F (1"x6")
28 wt% HCl @ 150 F (1"x6")
15 wt% HCl @ 200 F (1"x6")

Pore Volumes To
10 28 wt% HCl @ 200 F (1"x6")

Break Through
28 wt% HCl @ 150 F (4"x20")

0.1
0.1 1 10
Interstitial Velocity, cm/min
Fig. 3⎯Linear core flooding experiment results for high porosity outcrop chalk samples.

For example, according to the field data shown in Fig. 1, effective stimulation radius is typically in the 10 to 20 ft range for
field matrix acidizing treatments with acid injection volumes of 75 gal per ft (10 ft3 per ft). For a 15 ft-radius stimulated zone, the
pore volume of the stimulated region is calculated to be 424 ft3 per ft for 30% porosity rock. Thus, the pore volume to
breakthrough value estimated in the field is calculated to be PVbt = 0.047. This is at least an order of magnitude smaller than those
estimated from the linear acid flooding test results presented in Fig. 3. Using the stoichiometric volume of rock dissolved by acid,
0.16 cu ft CaCO3 per cu ft 28% HCl, the injected acid dissolves only 0.32% of the rock within the effective 15 ft stimulation
radius. As evidenced by many laboratory experiments, once dominant wormholes form in carbonate rocks, the wormhole channels
provide negligible resistance to flow and carry essentially all of the injected fluid with limited fluid loss at near optimum injection
conditions. These highly conductive flow paths, created in relatively low matrix perm carbonate rocks, provide significant flow
concentration near the wormholed region, particularly near the tip of the dominant wormholes. As a result, wormholes disturb the
near-wellbore pressure field and create a complex flow field significantly different from simple radial and spherical flow
geometries during stimulation.
It is widely known that wormhole growth rates, optimum injection rates, and breakthrough volumes depend on the core size
used in the laboratory flow tests. There is a consistent trend showing that wormholing in larger core samples appears more
efficient than that measured in smaller cores even for relatively homogeneous rocks. Core test results presented in Fig. 3 indicate
that 4” diameter core has PVbt,opt = 0.132, which is approximately 4 times smaller PVbt,opt value obtained for 1” diameter core using
28% HCl at 150° F. A similar observation was discussed by Buijse (2000). These core size dependencies are primarily due to flow
concentration at the wormhole tip and the boundary effects of the linear core flooding experiments. The axially elongated core
significantly limits fluid loss from the side wall of the dominant wormhole and forces injection fluid to reach the wormhole tip.
Therefore, for a given injection rate at the core inlet, the convergent flow at the tip of the wormhole becomes larger for a large
diameter core experiments.
Intrinsic carbonate heterogeneity such as natural fractures and vugs is also a very important factor in wormholing. Izgec et al.
(2008) presented large core (4” diameter) experiments with vugular limestone samples showing an order of magnitude lower PVbt
values than relatively homogeneous Indiana limestone data obtained from 1” diameter core samples. Their results show that large
scale heterogeneity (vugs) greatly enhances apparent wormholing efficiency. Although quantitative understanding of large scale
heterogeneity effects on wormholing efficiency is not fully understood, it is very important to separately measure the core size
scale effect for relatively homogeneous carbonate rocks.
In this study a 3D FEM flow model was developed to simulate laboratory linear coreflooding experiments. Fig. 4 shows
pressure and flux contour plots for 1” diameter × 6” length core flow test. A constant injection rate is applied to provide an average
interstitial core flow velocity, vi = Q/(Aφ), of 1 cm/min at the core inlet while a constant pressure of 1000 psia is applied at the
outlet of the core. No-flow boundary conditions are applied at the confining wall. The wormhole diameter and length are assumed
to be 0.1” and 2”, respectively. The simulation results show very small pressure drops and fluxes near the base of the wormhole
while very high flow concentration occurs at the tip of the wormhole. Calculated wormhole tip velocities are calculated to be as
high as vi = 10.7 cm/min.
SPE 134265 5

No flow boundary at the


Inlet side wall of the core
(b.c#1: q = qinj)

Outlet
(Constant pressure b.c.)
Wormhole

Pressure, psia

1000 1835

Interstitial Velocity, cm/min

0 10.7
Fig. 4⎯3D FEM simulation results for 1” diameter linear coreflooding experiment (vi = 1 cm/min).

Fig. 5 shows the interstitial velocity at the tip of the wormhole simulated for 1” and 4” diameter core tests. During initial
injection prior to wormholing, the interstitial velocity is equal to 1 cm/min for both the 1” and 4” diameter tests. Once a dominant
wormhole starts developing in the core, flow concentration at the tip of the wormhole continues developing as the wormhole
grows. While maintaining the constant injection rate in the test cell, the interstitial velocity at the tip of the wormhole, vi,tip
increases until the wormhole length reaches to Lwh = dcore. At this time, the interstitial velocity at the tip of the wormhole becomes
relatively constant. For a given size of the wormhole diameter, the in-situ interstitial velocity at the tip of the wormhole for the 4”
diameter core test is approximately 4 times higher than that seen in the 1” diameter core test. This high wormhole tip velocity
explains why the apparent wormholing efficiency is greater for larger diameter samples and may help explain the large
discrepancies between field stimulation radii and laboratory core flood test data.
Based on these observations, vi, which is an averaged interstitial velocity relative to the core diameter, dcore, can be
approximately correlated to vi,tip for a particular wormhole diameter, dwh;
⎛d ⎞
vi = ⎜⎜ wh ⎟⎟vi ,tip .................................................................................................................................................................. (1)
⎝ d core ⎠
As noted above, Buijse and Glasbergen (2005) recently presented an empirical wormhole growth model. According to their
model, the wormhole growth rate can be approximated by
2
⎛ vi ⎞ ⎛ vi ⎞
−γ ⎧ ⎡ ⎛ ⎞
2
⎤⎫
vwh =⎜ ⎟×⎜ ⎟ × ⎪⎨1 − exp ⎢− 4⎜ vi ⎟ ⎥ ⎪⎬ ....................................................................................................... (2)
⎜ PVbt ,opt ⎟ ⎜ vi ,opt ⎟ ⎢ ⎜⎝ vi ,opt ⎟ ⎥⎪
⎝ ⎠ ⎝ ⎠ ⎪⎩ ⎣ ⎠ ⎦⎭
where γ is the coefficient that represents the effect of the fluid loss limited wormholing. Note that when γ = 1/3, then Eq. 2 is
equivalent to the linear coreflooding experiment results as shown in the original Buijse-Glasbergen model. The first, second, and
third terms on the right hand side of the equation represent the optimum wormhole growth rate, the effect of tip splitting and side
branching, and the diffusion limited wormholing processes. Eq. 2 is based on macroscopic wormhole growth characteristics
referenced to vi. The term vi is the average interstitial velocity, Q/(Aφ), with respect to the cross sectional area of the core diameter,
dcore.
6 SPE 134265

Fig. 5⎯Effect of core size on the injection velocity at the tip of the dominant wormhole (FEM simulation results).

Hung et al. (1989) presented a mechanistic model of wormhole propagation based on acid transport and fluid loss from a single
wormhole, which is written in terms of the acid capacity number NAc as
⎛ Ctip ⎞
vwh = vi ,tip ⎜⎜ ⎟⎟ N Ac ........................................................................................................................................................... (3)
⎝ C0 ⎠
where C0 and Ctip are the injection acid concentration and the actual acid concentration at the tip of the wormhole. The acid
capacity number is the ratio of the amount of mineral dissolved by the acid occupying a unit volume of rock pore space to the
amount of mineral present in the unit volume of rock. For limestone, the acid capacity number is given by
φβC0 ρ acid
N Ac = ................................................................................................................................................................ (4)
(1 − φ ) ρ F
Comparing to Eqs. 2 and 3, wormhole growth rate at the optimum injection conditions can be estimated by vi/PVbt,opt and
vi,tipNAc for core-size scale and single-wormhole scale, respectively. Also the diffusion loss term in Eq. 3 may be approximated by
2
⎧ ⎡ ⎛ 2 ⎫

⎛ Ctip
⎞ ⎪ v ⎞ ⎪
⎜⎜ ⎟⎟ ≈ ⎨1 − exp ⎢− 4⎜ i ⎟ ⎥ ⎬ ....................................................................................................................................... (5)
⎝ C0 ⎠ ⎪ ⎢ ⎜⎝ vi ,opt ⎟⎠ ⎥ ⎪
⎩ ⎣ ⎦⎭
Considering the effect of tip splitting and side branching for fluid-loss limited wormholing region, Eq. 2 can be rewritten in
terms of vi,tip as
2
⎛ vi ,tip ⎞
−γ ⎧ ⎡ ⎛ v ⎞
2
⎤⎫
vwh = vi ,tip N Ac × ⎜ ⎟ × ⎪⎨1 − exp ⎢− 4⎜ i ,tip ⎟ ⎥ ⎪⎬ .................................................................................................... (6)
⎜ vi ,tip ,opt ⎟ ⎢ ⎜⎝ vi ,tip ,opt ⎟ ⎥⎪
⎝ ⎠ ⎪⎩ ⎣ ⎠ ⎦⎭
where the optimum injection velocity at the tip of wormhole is denoted by vi,tip,opt. Using Eq. 1,
⎛d ⎞
vi ,tip ,opt = vi ,opt ⎜⎜ core ⎟⎟ ........................................................................................................................................................... (7)
⎝ d wh ⎠
where the wormhole diameter can be approximated by the effective wormhole diameter, de,wh as
d e,wh ≈ d core PVbt ,opt N Ac ........................................................................................................................................................ (8)

Substituting Eq. 8 into Eq. 7 to give


vi ,opt
vi ,tip ,opt = for tip splitting and side branching ................................................................................................... (9)
PVbt ,opt N Ac

The reaction rate, which is defined by the change in the number of moles of acid, is given by
dM
RHCl = = rHCl s f .......................................................................................................................................................... (10)
dt
SPE 134265 7

where rHCl and sf are the area specific reaction rate and the surface area. Assuming that the effective surface area available for acid
reaction is proportional to the wormhole penetration length, rwh, the acid flowing velocity through the wormhole tunnel needs to
increase with time to maintain a sufficient acid concentration at the wormhole tip since the acid to formation contact time becomes
larger as wormholes grow. The minimum injection velocity required for the fluid loss limited wormholing will increase as
wormholes grow as;
vi ,opt Lcore
vi ,tip ,opt = for the diffusion limited wormholing ..................................................................................... (11)
PVbt ,opt N Ac rwh

where Lcore is the core length. Substituting Eqs. 9 and 11 into Eq. 6 gives the modified Buijse-Glasbergen wormhole growth model
as
2
⎛ vi ,tip PVbt ,opt N Ac ⎞
−γ ⎧ ⎡ ⎛ v PV ⎞
2
⎤⎫
⎟ × ⎪⎨1 − exp ⎢− 4⎜ i ,tip bt ,opt Ac wh ⎥ ⎪⎬ .................................................................... (12)
N r
vwh = vi ,tip N Ac × ⎜ ⎟
⎜ vi ,opt ⎟ ⎢ ⎝ ⎜ vi ,opt Lcore ⎟ ⎥⎪
⎝ ⎠ ⎪⎩ ⎣ ⎠ ⎦⎭
This is expressed in terms of the in-situ interstitial velocity at the tip of the dominant wormholes, vi,tip.

Fluid Loss and Competition For Injection Fluid in Radial and Spherical Flow Geometries
Gdanski (1999) and Glasbergen et al. (2005) discussed symmetry of wormholing under radial flow conditions and concluded that
wormholes grow in a certain pattern in both angular and axial directions. Huang et al. (1999) presented that the wormhole
population density (the number of major wormholes created per unit area of rock surface) under radial flow conditions is about 12
wormholes per ft for a typical carbonate reservoir. Their work indicates that there are about 4 wormholes at the same cross-section
of the wellbore because the region affected by one wormhole extends about 90° around the wellbore. Although the exact wormhole
pattern will be strongly affected by the permeability anisotropy and heterogeneity (Pichler 1992), a few dominant wormholes are
expected to grow symmetrically in the angular direction for a relatively isotropic and homogeneous permeability rock (Gdanski
1999). As a result, wormholes are expected to be spaced apart longitudinally with varying lengths along the wellbore.
In this study, a 3D FEM model is built for an openhole well with radially growing, highly symmetrical wormholes in
homogeneous and isotropic permeability formation. Note that the FEM model is not an acid-carbonate chemical reaction model.
The model calculates steady-state pressure and flux distributions for a given pattern of wormholes. The wormhole separation angle
in the angular direction is assumed to be 60° while five different wormhole lengths ranging from 0.5 ft to 1 ft to 2 ft to 4 ft to 8 ft
are located along the wellbore in the axial direction periodically. The effective wormhole diameter is 0.2 ft. The following
reservoir and well data are used in the simulation:
• Wellbore radius = 0.354 ft
• Injection rate = 60 bpm per 1000 ft length
• Injection pressure = 7000 psia at formation face
• Formation permeability = 2 md
• Formation porosity = 35%
• Fluid viscosity = 0.5 cp

Fig. 6 shows injection velocity and pressure profiles for a section of the wormholed formation (1/12th symmetry section) in the
near-wellbore region. The velocity contour plots show that significant flow concentration occurs near the tips of the longest
wormholes while the injection velocity is much lower for shorter wormholes near the wellbore. It is anticipated that the majority of
fluid loss occurs near the tips of the dominant wormholes, causing faster wormhole branching in this particular section.

Fig. 6⎯3D FEM simulation results for radially extending wormholes assuming an idealized wormhole pattern.
8 SPE 134265

Fig. 7 shows the detailed fluid loss profile for various length wormholes. The results indicate that the injection velocity for 8
foot long wormholes becomes as large as vi,tip = 15.8 cm/min while the injection velocity for 0.5 foot long wormholes is vi,tip = 0.32
cm/min. Clearly the short wormholes go into the low-velocity, diffusion-limited, wormholing region while the longer wormholes
continue staying in the high-velocity, fluid loss-limited, wormholing region. This example demonstrates the injection fluid
competition mechanisms for radial flow geometry during acid injection.

rwh = 8 ft
Interstitial Velocity, cm/min

rwh = 0.5 ft
rwh = 4 ft

rwh = 1 ft
rwh = 2 ft

Radial Distance from the Center of the Wellbore, ft


Fig. 7⎯Injected fluid competition among wormholes propagating with different lengths.

A semi-analytical correlation is derived for estimating interstitial velocities at the tip of the dominant wormholes based on a
number of 3D FEM simulation analyses considering various wormholing patterns. Appendix A includes details of the derivation of
the radial flow upscaling model and, also the verification of the models compared with the numerical simulation work. For radial
flow, the interstitial velocity is approximated as:

q ⎡ 1 ⎛ 1 ⎞⎤
vi ,tip = ⎢(1 − α z ) +αz⎜ ⎟⎥ for radial flow ................................................................................... (13)
φh πmwh ⎢⎣ ⎜d ⎟⎥
d e,wh rwh ⎝ e,wh ⎠⎦

where mwh and αz denote the number of the dominant wormholes along the angular direction and the wormhole axial spacing,
ranging from 0 to 1. When αz = 0, the dominant wormholes are closely spaced along the axial direction (2D radial flow geometry).
The injection velocity at the wormhole tip declines proportionally to 1/rwh0.5. This extreme case represents formations where the
permeability in the longitudinal direction is significantly lower than that in the radial direction. When αz = 1, the dominant
wormholes are sparsely spaced along the axial direction with very little effect on low-length/stalled wormholes. In this case, the
injection velocity at the wormhole tip doesn’t decline as rwh increases. For typical acid stimulation design, it is recommended to
use αz = 0.25 to 0.5 for vertical wells (ka < kr) and αz = 0.5 to 0.75 for horizontal wells (ka ≈ kr). Also it is important to note that
substituting mwh = 4π, de,wh = rwh, and αz = 0 into Eq. 12 reduces to the conventional radial flow equation:
q
vi ,tip = ................................................................................................................................................................. (14)
2πφhrwh
Fig. 8 shows example wormhole penetration calculations for openhole completions. This plot can be used for determining the
optimum acid injection rate for particular reservoir conditions.
SPE 134265 9

Wormholing in Radial Flow Geometry


V =1.5 bbl/ft, h =1000 ft, m wh =6, d ewh =15 mm, γ =0.333
PV bt,opt =0.416, v i,opt =1.4 cm/min, φ =35%,
30

Wormhole Penetration Length, ft


25

20

15

10

0
0 10 20 30 40 50 60 70 80 90 100
Injection Rate, bpm

Fig. 8⎯Wormhole penetration lengths in the radial flow geometry using the upscaling method developed in this study.

Spherical flow is another important flow geometry encountered in the near-wellbore region during acid treatments. For
example, horizontal cased hole perforated wells completed using limited entry perforation techniques have typical perforation
spacings in the order of one shot per 10 ft to one shot per 20 ft. Partially completed horizontal wells with short, high shot density,
perforation clusters spaced 100 to 200 ft apart have been widely used in North Sea chalk fields. These completions will have
spherical flow geometry in the near-perforation region during acid injection until the pressure interference between perforation
clusters becomes important. Since wormholes extend following the flow field, wormholes can be expected to extend in the
spherical coordinate system. A hypothetical wormhole pattern is assumed based on the pressure symmetry in the spherical
coordinate system (Appendix B). Such wormhole geometry may be characterized by an icosahedral symmetry in a relatively
homogeneous, isotropic permeability rock. In this case, the injection velocity at the tip of the dominant wormhole can be
approximated as:
q
vi ,tip = for spherical flow ................................................................................................................................ (15)
4πφd e,wh rwh

where the injection velocity declines proportionally to 1/rwh for the spherical flow geometry. It is important to note that substituting
de,wh = rwh into Eq. 15 reduces to the conventional spherical flow equation:
q
vi ,tip = .................................................................................................................................................................... (16)
4πφrwh
2

Fig. 9 shows example wormhole penetration length calculations for perforation cluster completions. In this example, the fluid
loss coefficient is varied for comparison. As shown in the figure, the optimum injection rate depends on the selection of γ. If the
fluid loss is greatly reduced (γ < 1/3), the wormhole penetration length can increase with increasing pump rates.

Wormholing in Spherical Flow Geometry


V =150 bbl, d ewh =15 mm
PV bt,opt =0.416, v i,opt =1.4 cm/min, φ =35%
70
Wormhole Penetration Length, ft

60

50

40

30

20

10

0
0 2 4 6 8 10 12 14 16 18 20
Injection Rate, bpm

Fig. 9⎯Wormhole penetration radius in the spherical flow geometry using the upscaling method developed in this study.
10 SPE 134265

Wellbore-Reservoir Coupled Flow Computational Method


Greyvenstein and Laurie (1994) presented a steady-state pipe network model based on node equations that utilizes the continuity
equation to establish a mass balance at each node of the network. These equations are then solved simultaneously to yield element
flow rates. The pressure change over each element is provided via a pressure drop-flow relationship, which follows from the
momentum balance equations. The nodal aspect of the formulation makes the network definition much easier and yields a sparse
matrix, while preserving good convergence characteristics. Their model is often referred to as the hybrid pipe network method
(Pretorius et al. 2008). This hybrid method can further deal with mixed boundary conditions consisting of either pressure or flow.
This is a great advantage since high-rate acid stimulation jobs are often constrained by either the maximum injection pressure (i.e.,
surface pump pressure or fracturing pressure) or target injection rate (i.e., maximum pump rate).
A network consists of a number of oriented elements connected to nodes as shown in Fig. 10. Elements are connected to one
another by nodes and each element is associated with an upstream and a downstream node. Numbers are used to identify each node
and element. Node i is connected through elements eij to neighboring nodes nij, with j = 1, 2, … J. J is the number of branches
associated with node i. The indices i, nij and eij are global indices while j is a local index. If the assumed positive flow direction for
element eij is towards node i, eij is assigned a positive value in the element continuity matrix E. Otherwise it is assigned a negative
value.

Fig. 10⎯Graphical presentation of a wellbore-reservoir flow network.

The wellbore flow model is based on continuity and pressure drop-flow rate equations assuming a steady state, incompressible,
isothermal fluid network. Appendix C presents details of the wellbore flow model employed in this work. The continuity condition
can be expressed in discretized form as
J


j =1
Qij ζ ij = −d i .................................................................................................................................................................. (17)

where ζij = 1 if eij > 0, ζij = -1 if eij < 0 and di is the external flow into node i accounting for wellhead injection flow and flow into
the formation. The pressure drop-flow rate relationship for any of the branch elements of node i can be expressed as
Δpij = pnij − pi = ζ ij H ij g ij Qij2 ............................................................................................................................................ (18)

where gij represents various flow friction loss elements such as pipe, choke, perforation hole, or inflow control devices and Hij =
Qij/|Qij|.
A model to track the interfaces created between the various injected fluids is also included in the wellbore fluid model
(Eckerfield et al. 2000). Fluid friction loss coefficients gij are changed depending on the type of fluids flowing for element eij.
The first step in the computational scheme is to guess initial pressures at all nodes. These values are tentative values and
denoted by pi*. Next Eq. 18 is solved for Qij using the tentative pressures. These values are also tentative values and denoted by
Qij*.
A field of pressure and flow rate corrections is calculated in such a way that the corrected values better satisfy Eqs. 17 and 18.
The corrected values are calculated as follows:
p = p * + p′ ....................................................................................................................................................................... (19)

Q = Q * +Q′ ...................................................................................................................................................................... (20)

where p′ and Q′ are the pressure and flow rate corrections, respectively. From this, we obtain an equation which can be solved for
p’. The result is
⎡J ⎤
pi′ = ⎢∑ (cij p ′nij ) + bi ⎥ / cii , i = 1, 2, …, I .......................................................................................................................... (21)
⎢⎣ j =1 ⎥⎦
SPE 134265 11

where
J ⎧ ⎫⎪
⎪ 1
cii = ∑ ⎨ * ⎬
+ a Jx ................................................................................................................... (22)
⎩ H ij [2 g ij Qij + Qij (∂g ij / ∂Qij )] ⎪⎭
j =1 ⎪
* *2

1
cij = ................................................................................................................................... (23)
H ij [2 g ij Qij* + Qij*2 (∂g ij / ∂Qij* )]

J
bi = qi + a Jx ( pres − pi* ) + bJx + ∑ Qij*ζ ij ............................................................................................................................ (24)
j =1

aJx and bJx are parameters that define transient reservoir flow defined by
4.91816 × 10 −6 kΔx
a Jx = .............................................................................................................................................. (25)
μ[ p D (t Dn − t Dn−1 ) + s ]

⎡ n −1 k ⎤
⎢∑ (q r ,i − qr ,i ) p D (t Dn − t D j −1 )⎥ − qr ,i p D (t Dn − t Dn−1 )
k −1 n −1

bJx = ⎣ ⎦
k =1
.............................................................................................. (26)
[ p D (t Dn − t Dn−1 ) + s ]

The dimensionless variables employed in Eqs. 25 and 26 are chosen for different reservoir flow conditions. For relatively long
horizontal wells, the conventional transient radial flow equations are used while spherical flow equations are used for short, widely
spaced, perforation clusters. The dimensionless groups and validity of the reservoir flow equations are presented in Appendix D.
In Eqs. 25 and 26, initial completion skin factors are calculated for different completion types and formation and perforation
damage conditions (Furui et al. 2003; Furui et al. 2005; Furui et al. 2008). Once acid injection starts, skin factors are re-calculated
based on wormhole geometry and penetration depths. For radial flow conditions, the wormhole skin factor is approximated by
s = − ln(rwh / rw ) ................................................................................................................................................................ (27)
Strictly speaking, the skin factor for radially extending wormholes depends on the wormhole phasing number, mwh. However as
long as wormholes propagate with a good circumferential coverage, such as mwh = 6 (60° phasing), the resultant skin factor can be
closely approximated by Eq. 27. According to our previous work (Furui et al. 2008) on perforation skin factor analysis, the skin
factor approaches those estimated by Eq. 27 as long as mwh ≥ 4 (90° phasing).
For perforated wells, wormholes propagate spherically from the perforation tip resulting in a complex flow field in the near-
wellbore vicinity. Appendix E shows a detail of the skin factor calculation for spherically extending wormholes. In this case, the
skin factor can be obtained from:
h perf ⎛ h perf ⎞
s= − ln⎜⎜ ⎟ − 1 for rwh < hperf/2 ............................................................................................................................ (28)

2rwh ⎝ 2rw ⎠
where hperf is the perforation spacing. Once rwh reaches to hperf/2, wormholes grow radially and then the skin factor can be
calculated using Eq. 27.

Field Case Study


In this section, we present a post-job evaluation for matrix acidizing in a carbonate reservoir using actual field data. A high-rate
matrix acid stimulation was performed for a cased and cemented horizontal well that was stimulated in four stages using the
limited entry techniques which rely on perforation friction to provide fluid diversion. Each stage was perforated using TCP guns
loaded at one shot per 10 feet, with perforations oriented to the high side of the wellbore and each perforation having a 0.21 inch
entrance hole diameter. Individual perforation and associated acid treatment stages were separated by composite plugs set on
coiled tubing. Plugs were milled at the end of the operation prior to placing the well on production. In this work, we present the
post-job evaluation result for the first zone’s treatment. Fig. 11 presents slurry rates and pressures measured by the surface and
downhole gauges. Other well and stimulation data are summarized in Table 1.
12 SPE 134265

High-Rate Matrix Acidizing Treatment Data


10,000 100

8,000 80

Slurry Rate, bpm


Pressure, psig
6,000 60

4,000 40

2,000 20

0 0
0 50 100 150 200
Elapsed Time, min
Fig. 11⎯Horizontal well matrix acid stimulation data.

The stimulation program consists of the pre-stimulation step-rate injection test, main acid injection stage, and overflush
injection stage. The initial step-rate injection test was performed to estimate fracture extension and closure pressures and measure
the pre-stimulation perforation pressure drop. The main acid injection stage was 28% HCl at an injection rate up to 56 bpm. The
average acid injection volume during this stage was 1.65 bbl/ft. The maximum injection rate of 60 bpm was achieved during the
overflush injection stage, and the average overflush injection volume was 2.65 bbl/ft.
Tubing and perforation friction between the downhole gauge and the reservoir-face are subtracted from the gauge data to arrive
at the true injection pressure at the reservoir-face. As noted earlier in this paper, friction reducers are widely used in high-rate acid
treatments. These additives significantly reduce pipe friction, allowing higher rates of injection than would be attainable for
standard, untreated/Newtonian, fluids. However the effect of the friction reducer can also change during the course of the job due
to insufficient control of the friction reducer volume as the injection rate varies during treatment. Best practice for evaluating the
effectiveness of the friction reducer during the stimulation job is to calibrate tubing friction curves using actual pressure data
measured by both surface and downhole gauges. In this work, a wellbore-reservoir flow coupled model is built considering varying
friction reducer efficiency during treatment. The network model developed here consists of 24 nodes and 23 elements including 10
reservoir nodes.

Table 1 - Summary of Well and Stimulation Data


Well Data
Completion Cased, cemented and perforated
Tubing Size 4-1/2" (ID=3.958")
Production Liner Size 5" (ID = 4.044")
Stimulation Interval Length 970 ft
Perforation Diameter 0.21 inch
Perforation Shot Density 0.1 SPF
Downhole Gauge Setting Depth 10374 MD-ft (8880 TVD-ft)
Production Packer Depth 10607 MD-ft (8863 TVD-ft)
Top Perf Depth 15173 MD-ft (9632 TVD-ft)

Stimulation Data
Acid Concentration 28 wt %
Acid Volume 1.65 bbl per ft
Overflush Volume 2.65 bbl per ft

Fig. 12 shows history matching results for tubing friction. The pressure differentials between the surface and downhole gauges
are calculated as: ΔPgauge = WHP - BHP. This is subsequently matched by adjusting the friction multiplier in the wellbore flow
model developed for this study.
SPE 134265 13

3,000 160
dP
2,000 140
Slurry Rate
1,000 120

Friction Reducer Rate, gpm


WH_Frictio

Friction Multiplier, %
Pressure Drop, psi
Frict Rate

Slurry Rate, bpm


0 100

-1,000 80

-2,000 60

-3,000 40

-4,000 20

-5,000 0
0 20 40 60 80 100 120 140 160 180 200
Elapsed Time, min
Fig. 12⎯Tubing friction matching using the actual pressure gauge data.

Table 2 summarizes ISIP (Instantaneous Shut-In Pressure) analysis results for pre-stimulation and post-stimulation falloff data.
The total friction pressure losses occuring below the downhole gauge are estimated by subtracting ISIP from BHP just before shut-
in. The pipe friction loss between the downhole gauge and the reservoir-face is extrapolated by the wellbore flow model shown by
Fig. 12. The average perforation friction losses are then calculated to be 971 and 709 psi for pre- and post-stimulation falloff data,
respectively.

Table 2 - Tubing and Perforation Friction Analysis Summary


BHP ISIP Rate ΔPfriction ΔPpipe ΔPperf
Falloff Data PSIG PSIG BPM PSI PSI PSI
Pre-Stimulation 7783 6200 45.0 1583 613 971

Post-Stimulation 7200 5700 60.0 1500 791 709

The perforation pressure drop is often expressed by the following formula (White 1986):
ΔPperf = 1.98244 × (q / N perf ) 2 γ /(Cv × d perf
2
) 2 ..................................................................................................................... (29)

where γ and Cv are fluid specific gravity and flow coefficient. q and dperf are in bpm and inches.
For design purposes, the most critical parameter is perforation diameter. Service company literature and testing reports are
available and should be requested to estimate effective perforation size for different gun types and perforation conditions. This is
particularly important when the limited entry technique is used for acid diversion. In critical wells, perforation coupon tests should
be performed using specified charges and pipe to determine the perforation diameter and flow coefficient to be expected for the
target well. In general, the flow coefficient for new perforations is estimated at 0.6, which represents a sharp-edged orifice (Crump
and Conway 1988). The flow coefficient can be as large as 0.89 for rounded-corner orifices/eroded perfs. We recommend flow
coefficients of Cv = 0.6 to 0.8 for most acid stimulation data analysis (no proppant injection conditions). In this work, we used Cv =
0.8. The result shows that there are 63 and 98 holes open out of the total of 98 perforations shot into the pipe for pre- and post-
stimulation periods, respectively. This indicates that all of the perforations are open at the end of the acid job, which is a good
indicator of effective acid diversion during stimulation.
For high-rate matrix acidizing jobs, slurry injection rates are often ramped up quickly during the main acid injection stage. This
is likely to exceed the formation fracturing pressure. Although the main purpose of the stimulation job is not to fracture the
formation hydraulically, the initial perforation breakdown and resulting small fractures, could help improve fluid diversion during
early time injection period. Therefore it is crucial to understand the formation stresses and fracture pressures (extension and
closure) to assess how long fracturing takes place prior to the matrix acidizing stage. If the fracturing pressure and formation
stresses in a region of interest are not obvious, a pre-stimulation step-rate injection test should be performed as a part of the
stimulation program. This information is a key component of the post-job evaluation.
Fig. 13 shows the step rate test plot using corrected reservoir-face pressures based on tubing and perforation friction analysis
results. The fracture extension pressure is determined to be 6241 psi at reservoir depth. Fracture closure (minimum in situ stress) is
commonly determined from pressure falloff data following injection. Several analytical techniques are used to determine closure
(Barree et al. 2009). The most common analysis procedure is to plot pressure versus the square root of shut-in time. For this
particular case, the fracture closure stress is determined to be 5745 psi based on the square root time plot.
14 SPE 134265

7,000

6,500

Reservoir-Face Pressure, psi


6,000

5,500

5,000

4,500

4,000
0 5 10 15 20 25 30 35 40 45 50
Slurry Rate, bpm

Fig. 13⎯Pre-stimulation step rate injection test result using the corrected reservoir-face pressures.

Finally, the downhole flowing pressure analysis plot is developed for the main acid stage as shown in Fig. 14. Using the
wellbore flow model developed in this work, the reservoir-face pressure is calculated from the downhole gauge pressure data
considering hydrostatic and friction (tubing and perforation) pressure drops. Acid injection started at the surface at t = 60 minutes.
The injected acid travels through surface lines and the tubing and arrives at the top perforations at t = 68 minutes. During the initial
tubing displacement, with fresh water entering the formation, the downhole treating pressure exceeded the fracture extension
pressure estimated from the step-rate injection test. Based on fracture modeling this created relative short fractures with a half-
length in the 5 to 10 ft range. At t = 70 minutes, fracture closure occurs approximately 2 minutes after acid enters the reservoir.
The downhole treatment pressure then stays below fracture closure pressure (the minimum horizontal stress) from t = 70 minutes
until the end of the job. The observations above suggest matrix acidizing (wormholing) is the key mechanism responsible for the
pressure decline and stimulation benefit seen.

Downhole Flowing Pressure Analysis Plot

PDPG Reservoir Face Slurry Rate


10000 120

9000 100
Slurry Rate, bpm

8000 80
Pressure, psig

7000 60

6000 40

5000 20

4000 0
60 65 70 75 80 85 90 95 100
Elapsed Time, min
Fig. 14⎯Reservoir-face injection pressure during the main acid injection.

Fig. 15 shows downhole injection pressure matching results during the main acid injection stage. The wormhole growth
parameters used in the calculation are PVbt,opt = 0.393 and vi,opt = 1.468 cm/min. First the simulation is performed with a wormhole
model with no upscaling. That means Eqs. 13 and 15 are used for the acid injection velocity calculations. As shown in the figure,
the downhole injection pressure becomes too high indicating that the calculated wormhole penetrtation depth is too short.
Next, the upscaling method is applied with mwh = 6 and dewh = 0.1”. az was changed until a good match to the measured pressure
data is obtained. For this particular case, az = 0.75 gives a good match.
SPE 134265 15

Downhole Injection Pressure Matching


PV bt,opt = 0.393, v i,opt = 1.468 cm/min
10,000 120

9,000 100

Slurry Rate, bpm


8,000 80

Pressure, psig
7,000 60

6,000 40

5,000 20

4,000 0
60 65 70 75 80 85 90 95 100
Elapsed Time, min
Fig. 15⎯Injection pressure history matching results during the main acid treatment.

Fig. 16 shows the resultant wormhole penetration and post-stimulation skin factor distribution along the treatment interval. The
wormhole penetration depth ranges from 18 to 28 ft. The corresponding local skin factor is therefore calculated to range from -4.1
to -4.6. Due to the variation of the formation permeability along the injection interval, pressure drops across perforation holes
during stimulation are varied and calculated to be 410 to 936 psi at the injection rate of 60 bpm. Although the tighter, low-
permeability interval located at 350 ft from the heel has the smallest wormhole penetration/largest local skin, the calculated
wormhole length of approximately 18 ft is sufficient to remove any partial penetration effects relative to the perforation-to-
perforation half-distance of 5 ft.
Finally, it is important to compare actual stimulation skin estimated from the post-job analysis with target skin to determine if
job design parameters, such as acid diversion technique, injection rate, acid volume etc, need to be improved for the next well or
series of wells. The estimated skin factor profiles show local skin factors better than the target skin factor of -4.0, corresponding to
our expected P50 skin value presented in Fig. 1. This leads to the conclusion that target skin values were achieved and that
effective acid diversion was achieved along the entire treatment interval of 970 ft using the limited entry perforation technique, and
thus the stimulation job was successfully executed to meet stimulation objectives.

0.0 50
1 2.0
1 50
rwh Skin PHI Perm
‐0.5 45
0.9 1.8
0.9 45
Wormhole Penetration Length, ft

‐1.0 40
0.8 1.6
0.8 40
Post‐Stimulation Skin Factor

‐1.5 35
0.7 1.4
0.7 35
Permeability, md

‐2.0 30
0.6 1.2
0.6 30
Porosity, %

‐2.5 25
0.5 1.0
0.5 25
‐3.0 20
0.4 0.8
0.4 20
‐3.5 15
0.3 0.6
0.3 15
‐4.0 10
0.2 0.4
0.2 10
‐4.5 5
0.1 0.2
0.1 5
‐5.0 00 0.0
0 0
0 100 200 300 400 500 600 700 800 900 1000
Distance from Heel, ft
Fig. 16⎯Simulated wormhole penetration and post-stimulation skin factor profile along a horizontal lateral.

Conclusions
This paper demonstrates the application of a new wormhole modeling approach for matrix acidizing in carbonate formations.
Semi-analytical flow correlations are developed for calculating the in-situ acid injection velocities at the tip of the dominant
wormholes and are validated against 3D FEM simulation analysis. The new scale-up technique developed in this work extends
linear acidflooding test results to radial and spherical flow geometries and account for core size dependencies seen in laboratory
acidflood experiments.
16 SPE 134265

We have shown that application of the model developed in this study shows acid wormholing through carbonate formations
can provide significant well stimulation benefits, resulting in post-stimulation skins in the -3.5 to -4.0 range. These skin results are
consistent with post-stimulation pressure buildup test results obtained from a wide range of field carbonate matrix acidizing data.
This work also shows that post-job evaluation is essential to evaluate the effectiveness of acid treatments. This allows well
design and operations teams to learn from any shortcomings and to improve designs and procedures for the benefit of future
stimulation jobs. It is also shown that it is important to compare actual stimulation skin estimated from the post-job analysis with
target skin to determine if job design parameters, such as acid diversion techniques, injection rates, acid volumes etc, need to be
improved for the next well or series of wells. The model presented here can be used as a forward model for analyzing real-time
treatment rate and pressure histories and can also be used for post-job evaluation analysis.

Acknowledgements
The authors are grateful to ConocoPhillips Company and Texas A&M University for permission to publish this work.

Nomenclatures
A = flow area, L2, ft2
B = formation volume factor, L3/L3, RB/STB
ct = total compressibility, Lt2/m, psi-1
C0 = acid concentration
Ctip = acid concentration at the wormhole tip
Cv = flow coefficient
dc = choke diameter, L, inch
di = external flow into node i, L3/t, RB/D [m3/d]
dperf = perforation diameter, L, inch
dwh = wormhole diameter, L, ft
de,wh = effective wormhole diameter, L, ft
eij = element associated with node i and branch j
f = friction factor
gij = friction term associated with node i and branch j
h = interval length, L, ft
hperf = perforation spacing, L, ft
hwh = wormhole axial spacing, L, ft
Hij = Qij/|Qij|
i = global node number
I = total number of nodes
j = local branch number
J = number of branches associated with a specific node
k = permeability, L2, md
ka = permeability parallel to the well axis, L2, md
kr = permeability in the radial direction, L2, md
Lcore = core length, L, inch [cm]
Lwh = wormhole length in linear flow, L, ft
mwh = number of dominant wormholes per 2D plane
nij = node associated with node i and branch j
NAc = acid capacity number
Nperf = number of perforations
p = pressure, m/Lt2, psia
pD = dimensionless pressure
pres = reservoir pressure, m/Lt2, psia
PVbt = pore volume to breakthrough
PVbt,opt = optimum pore volume to breakthrough
q = injection rate, L3/t, RB/D
qr = flow into reservoir, L3/t, RB/D
Q = wellbore flow rate, L3/t, RB/D
rHCl = area specific reaction rate for HCl, moles/tL2, moles/sec/m2
rw = wellbore radius, L, ft
rwh = wormhole penetration radius, L, ft
rws,eq = equivalent wellbore radius for spherical flow, L, ft
s = skin factor
sf = surface area, L2, ft2
t = time, t, min
SPE 134265 17

tD = dimensionless time
vi = interstitial velocity, L/t, cm/min
vi,opt = optimum interstitial velocity, L/t, cm/min
vi,tip = interstitial velocity at the wormhole tip, L/t, cm/min
vwh = wormhole growth rate, L/t, cm/min
V = acid injection volume, L3, bbl
αf = friction multiplier
αz = wormhole axial spacing coefficient
β = acid dissolving power, m/m, lb/lb, [kg/kg]
ε = error
φ = porosity
γ = parameter for fluid-loss limited wormholing
γ = specific gravity
μ = viscosity, m/Lt, cp
ρ = density, m/L3, lbm/ft3
ζ = 1 if element eij is directed towards node i, else -1
Δp = pressure drop, m/Lt2, psi
Δpperf = perforation pressure drop, m/Lt2, psi
Δx = segment length, L, ft

Superscripts
* = tentative values
′ = corrections

References
Barree, R.D., Barree, V.L., and Craig, D.P. 2009. Holistic Fracture Diagnostics: Consistent Interpretation of Prefrac Injection Tests Using
Multiple Analysis Methods. SPE Prod & Oper 24 (3): 396-406. SPE-107877-PA. doi: 10.2118/107877-PA.
Brigham, W.E., Peden, J.M., Kitson, F. Ng., and Neil, O’Neill. 1980. The Analysis of Spherical Flow with Wellbore Storage. Paper SPE 9294
presented at the SPE Annual Technical Conference and Exhibition, Dallas, Texas, 21-24, September. doi: 10.2118/9294-MS.
Buijse, M.A. 2000. Understanding Wormholing Mechanisms Can Improve Acid Treatments in Carbonate Formations. SPE Prod & Fac 15 (3):
168-175. SPE-65068-PA. doi: 10.2118/65068-PA.
Buijse, M. A. and Glasbergen, G. 2005. A Semiempirical Model to Calculate Wormhole Growth in Carbonate Acidizing. Paper SPE 96892
presented at the SPE Annual Technical Conference and Exhibition, Dallas, Texas, 9-12, October. doi: 10.2118/96892-MS.
Crump, J.B. and Conway, M.W. 1988. Effects of Perforation-Entry Friction on Bottomhole Treating Analysis. J. Pet Tech 40 (8): 1041-1048.
SPE-15474-PA. doi: 10.2118/15474-PA.
Eckerfield, L.D., Zhu, D., Hill, D.A, Robert, J.A., and Bartko, K.M. 2000. Fluid Placement Model for Horizontal-Well Stimulation. SPE Drill &
Compl 15 (3): 185-190. SPE-65408-PA. doi: 10.2118/65408-PA.
Furui, K., Zhu, D., and Hill, A.D. 2003. A Rigorous Formation Damage Skin Factor and Reservoir Inflow Model for a Horizontal Well. SPE
Prod & Fac 18 (3) 151-157. SPE-84964-PA. doi: 10.2118/84964-PA.
Furui, K., Zhu, D., and Hill, A.D. 2005. A Comprehensive Skin-Factor Model of Horizontal-Well Completion Performance. SPE Prod & Fac 20
(3): 207-220. SPE-84401-PA. doi: 10.2118/84401-PA.
Furui, K., Zhu, D., and Hill, A.D. 2008. A New Skin Factor Model for Perforated Horizontal Wells. SPE Drill & Compl 23 (3): 205-215. SPE-
77363-PA. doi: 10.2118/77363-PA.
Gdanski, R. 1999. A Fundamentally New Model of Acid Wormholing in Carbonates. Paper SPE 54719 presented at the 1999 SPE European
Formation Damage Conference held in The Hague, The Netherlands, 31 May – 1 June. doi: 10.2118/54719-MS.
Glasbergen, G., van Batenburg, D., van Domelen, M., and Gdanski, R. 2005. Field Validation of Acidizing Wormhole Models. Paper SPE 94695
presented at the SPE European Formation Damage Conference held in Sheveningen, The Netherlands, 25-27 May. doi: 10.2118/94695-MS.
Greyvenstein, G.P. and Laurie, D.P. 1994. A Segregated CFD Approach to Pipe Network Analysis. Int. J. Num. Meth. Eng. 37 (21): 3685-3705.
Huang, T., Zhu, D., and Hill, A.D. 1999. Prediction of Wormhole Population Density in Carbonate Matrix Acidizing. Paper SPE 54723 presented
at the SPE European Formation Damage Conference, The Hague, Netherlands, 31 May – 1 June. doi: 10.2118/54723-MS.
Hung, K.M., Hill, A.D., and Sepehrnoori, K. 1989. A Mechanistic Model of Wormhole Growth in Carbonate Matrix Acidizing and Acid
Fracturing. J. Pet Tech 41 (1): 59-66. SPE-16886-PA. doi: 10.2118/16886-PA.
Izgec, O., Keys, R., Zhu, D., and Hill, A.D. 2008. An Integrated Theoretical and Experimental Study on the Effects of Multiscale Heterogeneities
in Matrix Acidizing of Carbonates. Paper SPE 115143 presented at the SPE Annual Technical Conference and Exhibition held in Denver,
Colorado, 21-24 September. doi: 10.2118/115143-MS.
Joseph, J.A. and Koederitz, L.F. 1985. Unsteady-State Spherical Flow with Storage and Skin. SPE J. 25 (6): 804-822. SPE-12950-PA. doi:
10.2118/12950-PA.
Lee, J., Rollins, J.B., and Spivey, J.P. 2003. Pressure Transient Testing. Textbook Series, SPE, Richardson, Texas 9: 225.
Mishra, V., Zhu, D., Hill, A.D., and Furui, K. 2007. An Acid-Placement Model for Long Horizontal Wells in Carbonate Reservoirs. Paper SPE
107780 presented at the SPE European Formation Damage Conference held in Scheveningen, The Netherlands, 30 May-1 June. doi:
10.2118/107780-MS.
18 SPE 134265

Pichler, T., Frick, T.P., Economides, M.J., and Nittmann, J. 1992. Stochastic Modeling of Wormhole Growth in Carbonate Acidizing with Biased
Randomness. Paper SPE 25004 presented at the SPE European Petroleum Conference held in Cannes, France, 16-18 November. doi:
10.2118/25004-MS.
Pretorius, J.J., Malan, A.G., and Visser, J.A. 2008. A Flow Network Formulation for Compressible and Incompressible Flow. Int. J. Num. Meth.
Heat & Fluid Flow 18 (2): 185-201.
Raghavan, R. and Clark, K.K. 1975. Vertical Permeability from Limited Entry Flow Tests in Thick Formations. SPE J. 15 (1): 65-73. SPE-4556-
PA. doi: 10.2118/4556-PA.
Rodriguez-Nieto, R. and Carter, R.D. 1972. Unsteady Three-Dimensional Gas Flow in Thick Reservoirs. SPE 4266 available from SPE,
Richardson, Texas.
White F.M. 1986. Fluid Mechanics. New York City: McGraw-Hill Book Co. Inc.

Appendix A –Development of A Flow Correlation Determining Interstitial Velocities at the Wormhole Tip in
Radial Flow Geometry
Most laboratory experiments for wormholing are performed in linear core flow tests. Macroscopic wormholing models applied to
field applications require some kind of a scale-up procedure. Reviewing previously published works, most of the existing
wormhole growth models assume the flow geometry in the near-wellbore vicinity is approximated by radial flow. This assumption
ignores complex fluid loss system and pressure symmetries created by multiple wormholes in the near-wormhole vicinity. They
simply calculate the acid injection velocity declines inversely proportional to the wormhole penetration radius, rwh. This is a
primary reason why most of the conventional wormholing models underpredict wormhole penetration.
Since flow from the side wall of the wormholes will be highly limited by the pattern and symmetry of the pressure field, the
actual (in-situ) injection velocity at the tip of wormholes will be much higher than those predicted by the simple radial flow
equation. In this work, we developed a 3D finite element model to evaluate the fluid loss from the side wall and injection velocities
at the tip of wormholes for different wormholing patterns. Fig. A.1 shows FEM gridding system for wormhole separation angles of
60° (mwh = 6). The following well and injection conditions are used for the simulation;
• rw = 0.3542 ft
• rwh = 2 ft
• de,wh = 0.02 ft
• q/h = 0.046 bpm per ft
The maximum injection velocity occurs at the tip of the wormholes and is calculated as high as vi,tip = 28.12 cm/min. Also the
simulation results indicate that fluid loss occurring near the base of the wormholes is significantly prohibited by the pressure
symmetry created by the wormhole pattern.

Fig. A.1⎯FEM gridding and velocity contour plot for highly symmetrical wormholes with mwh = 6.

Fig. A.2 compares tip injection velocities for different wormhole diameters and angular separation angles. The results show
that the acid injection velocities quickly increase as soon as wormholes form at the wellbore surface. As the wormholes get deeper
into the formation, the velocities decline proportionally to r0.5, which is much slower than those predicted by the conventional
radial flow theory. This is primarily due to the limited fluid loss induced by the symmetric wormhole pattern. The injection
velocities at the tip of the dominant wormhole depend on both mwh and de,wh. The simulation results show

vi ,tip ∝ 1 / mwh d e, wh ........................................................................................................................................................ (A-1)


SPE 134265 19

Fig. A.2⎯Injection velocities at the tip of the dominant wormholes closely spacing in the axial direction.

Based on these findings, the following empirical correlation can be obtained for calculating the in-situ acid injection velocity at
the tip of the wormhole for the 2D radial flow geometry;
q
vi ,tip = .................................................................................................................................................. (A-2)
φh πmwh d e, wh rwh

When wormholes grow sparsely in the axial direction, the wormhole tip injection velocity is even more pronounced. Fig. A.3
shows wormhole orientations employed in the FEM simulation analysis. For this particular case, the axial wormhole spacing is
fixed with hwh = rwh. No ceased wormholes that may exist between the dominant wormholes are modeled in the simulation model.
Other simulation conditions are the same as the previous radial flow case (mwh = 6, de,wh = 0.02’, rw = 0.354’, and rwh = 2’). The
flux contour plot obtained by the FEM model shows the flow concentration occurring at the tip of the dominant wormhole
becomes rather spherical and reaches as high as vi = 232.8 cm/min, which is significantly larger than those estimated by the 2D
radial flow model.

Fig. A.3⎯Wormholes sparsely growing along the wellbore axis.

Fig. A.4 compares injection velocities at the tip of the wormholes for different wormhole patterns. Once the dominant
wormholes form in the near-wellbore region, significant flow concentration occurs at the wormhole tip. Unlike the previous 2D
radial flow case, the injection velocities become relatively constant independent of rwh. Due to the nature of spherical flow, vi,tip
becomes highly dependent on de,wh. The results show
vi ,tip ∝ 1 /(d e, wh mwh ) ..................................................................................................................................................... (A-3)
20 SPE 134265

mwh=3, de,wh=0.02'
mwh=6, de,wh=0.02'
mwh=6, de,wh=0.04'
mwh=6, de,wh=0.06'
mwh=12, de,wh=0.02'

Fig. A.4⎯Injection velocities at the tip of the dominant wormholes sparsely spacing in the axial direction.

Based on these finings, the following empirical correlation can be obtained for the 3D radial flow geometry (dominant
wormholes only)
q
vi ,tip = for rwh >> rw ................................................................................................................................. (A-4)
φhd e, wh πmwh
Actual wormhole tip injection velocities depend greatly on the wormhole axial spacing along the borehole length. Ceased
wormholes that are unable to grow due to dominant wormholes also affect the overall fluid loss. In other words, injection
velocities at the tip of the wormholes should range between Eq. A-2 and A.4. Based on these observations, the following empirical
correlation can be obtained bounding two extreme cases given by Eq. A-2 and A-4;

q ⎡ 1 ⎛ 1 ⎞⎤
vi ,tip = ⎢(1 − α z ) +αz⎜ ⎟⎥ .......................................................................................................... (A-5)
φh πmwh ⎢ ⎜d ⎟⎥
⎣ d e, wh rwh ⎝ e, wh ⎠⎦

where αz is the wormhole axial spacing parameter defined by


0 .7
⎛h ⎞
a z = ⎜⎜ wh ⎟⎟ .................................................................................................................................................................. (A-6)
⎝ rwh ⎠
αz is set to be 0 for wormholes closely spaced in the axial direction while αz is set to 1 for wormholes sparsely oriented in the
axial direction. Injection velocities predicted by Eqs. A-5 and A-6 are compared with FEM simulation results in Fig. A.5. As
shown in the figure, the empirical correlation and the FEM simulation results show very good agreement. The average errors are 8,
11, and 16% for mwh = 3, 6, and 12, respectively. Strictly Eq. A-5 overestimates injection velocities for relatively short wormhole
lengths due to fluid loss from the open wellbore surface. However, these effects are usually very small since wormholes quickly
extend due to high injection velocities under normal stimulation conditions.
SPE 134265 21

* Total 192 data points


mwh = 3, 6, and 12

vi,tip Predicted by Correlation, cm/min


rw = 0.3542'
rwh = 0.5', 1', 2', and 5'
de,wh = 0.25", 0.5", 0.75", and 1"
hwh/rwh = 0, 0.25, 0.5, and 1

mwh = 12 ( = 8.1%)
mwh = 6 ( = 11.0%)
mwh = 3 ( = 16.3%)

vi,tip Simulated by 3D FEM Model, cm/min

Fig. A.5⎯Comparison of injection velocities at the wormhole tip predicted by Eq. A-7 and 3D FEM model (radial flow geometry).

Appendix B - Development of A Flow Correlation Determining Interstitial Velocities at the Wormhole Tip in
Spherical Flow Geometry
Spherical flow is another important flow geometry encountered in formations in the near-wellbore region. For example, horizontal
cased hole perforated wells completed by limited perforation technique have perforation spacing on an order of 10 ft. Partially
completed horizontal wells with perforation clusters spaced 100 to 200 ft apart have been widely used in the North Sea chalk field.
These completions will have spherical flow geometry in the near-perforation region until the pressure interference among the
perforations occurs.
Similarly to the radial flow geometry, it is assumed that wormholing of reactive fluids in carbonates forms a certain
symmetrical geometry in the spherical flow system. Assuming the wormhole symmetry pattern depends greatly on the pressure
field, the hypothetical wormhole geometry in the spherical flow coordinate system may be described by the icosahedral symmetry
in a relatively homogeneous, isotropic carbonate rock. A regular icosahedron has 60 rotational symmetries with 20 identical
equilateral triangular faces. Assuming that a single dominant wormhole can form per tetrahedron, there occurs 20 dominant
wormholes covering in the stimulated domain as shown in Fig. B.1.

Fig. B.1⎯Hypothetical dominant wormhole distribution in the spherical flow geometry.

Fig. B.2 shows FEM simulation results for injection velocities at the tip of wormholes having icosadedral symmetry at the
injection rate of 0.5 bpm. The results indicate that the injection velocity declines proportional to 1/rwh. Note that the conventional
spherical flow model assumes the injection velocity declines proportionally to 1/rwh2. The injection velocity decline during
wormhole propagation in the spherical flow geometry is much slower than those estimated by the conventional spherical flow
model.
22 SPE 134265

de,wh = 0.1'
de,wh = 0.15'
de,wh = 0.2'

Fig. B.2⎯Injection velocities at the tip of the dominant wormholes in the spherical flow geometry.

Based on these observations, the following empirical correlation is derived for estimating the injection interstitial velocity at
the tip of wormholes in the spherical flow geometry:
q
vi ,tip = ........................................................................................................................................................... (B-1)
4πφd e, wh rwh

Eq. B-1 is compared with the FEM simulation results as shown in Fig. B.3. The results show the average error of 2.9%.

20 wormholes
vi,tip Predicted by Correlation, cm/min

rwh = 1', 2', 3', 4', and 5'


de,wh = 0.1', 0.15', 0.2'

vi,tip Simulated by 3D FEM Model, cm/min

Fig. B.3⎯Comparison of injection velocities at the wormhole tip predicted by Eq. A-9 and 3D FEM model (spherical flow geometry).

Appendix C - Wellbore Flow Model for Advanced Completion Design Using One-Dimensional Pressure-Based
Nodal Method
The governing equations to be satisfied in a steady state, isothermal fluid network are the continuity, pressure drop-flow rate and
density equations. The latter is the equation of state in the case of gases. The strong form for the continuity equation for 1D flow,
as may be applied to a pipe network, is as follows:

( ρu ) = 0 ..................................................................................................................................................................... (C-1)
∂x
where ρ and u denote the cross-sectional average density and velocity at a point along a 1D pipe element, respectively. Assuming
incompressible fluid flow, the continuity condition can be expressed as
J

∑=
j 1
Qijζ ij = −d i , i = 1, 2,…, I ........................................................................................................................................... (C-2)
SPE 134265 23

in the discretized form. ζij = 1 if flow enters the node while ζij = -1 if the flow exits the node. The momentum equation, in the
absence of gravitational effects, reduces to the following:
Δpij = pni , j − pi = ζ ij H ij g ij Qij2 ......................................................................................................................................... (C-3)

where gij = gij(⏐Qij⏐) is a function of the absolute value of Qij and Hij = Qij/⏐Qij⏐. The g function accounts for various flow friction
losses obtained analytically or from experimental data. Friction losses can be due to flow in pipe, across a choke, perforation holes,
or ICD devices. For example,
f ( Qij ) ρΔxα f
g ij ( Qij ) = 1.52 for pipe segments .................................................................................................. (C-4)
( d1 − d 0 ) 3 ( d1 + d 0 ) 2
where f and αf are the friction factor and the friction multiplier.
ρ
g ij = for choke segments ............................................................................................................................ (C-5)
31.48Cv2 d c4
The first step in the computational scheme is to guess pressures at all nodes. These values are tentative values and denoted by
pi*. Next Eq. D-3 is solved for Qij using the tentative pressures determined at the previous step. These values are also treated as
tentative values and denoted by Qij* because it is based on tentative pressures. We calculate a field of pressure and flow rate
corrections in such a way that the corrected values better satisfy Eq. C-2 and C-3. The corrected values are calculated as follows:
p = p * + p′ ..................................................................................................................................................................... (C-6)

Q = Q * +Q′ .................................................................................................................................................................... (C-7)


where p′ and Q′ are the pressure and flow rate corrections, respectively. First we have to determine a relationship between pressure
correction and flow correction. If Eq. C-3 is differentiated with respect to Qij, we get
∂pni , j ∂pi ⎡ ∂g ij 2 ⎤
− = ζ ij H ij ⎢2 g ij Qij + Qij ⎥ ........................................................................................................................ (C-8)
∂Qij ∂Qij ⎣⎢ ∂Qij ⎦⎥
Because the partial differential term ∂() on the left-hand-side will become zero as convergence is reached, they may be replaced by
the correction term ()′:
pni′ , j pi′ ⎡ ∂g ij 2 ⎤
− = ζ ij H ij ⎢2 g ij Qij + Qij ⎥ ........................................................................................................................... (C-9)
Qij′ Qij′ ⎢⎣ ∂Qij ⎥⎦
Rearranging the terms of Eq. C-9, an equation for the flow correction in terms of pressure correction is now established as:
pni′ , j − pi′
Qij′ = .................................................................................................................................... (C-10)
⎡ ∂g ⎤
ζ ij H ij ⎢2 g ij Qij + ij Qij2 ⎥
⎢⎣ ∂Qij ⎥⎦
To complete the proposed algorithm, ∂gij/∂Qij is computed numerically to first order accuracy as follows:
∂g ij g (Qij + dQij ) − g (Qij )
= ....................................................................................................................................... (C-11)
∂Qij dQij

where dQij is set as


dQij = max(10 −3 Qij ,10 −5 Qijmax ) ...................................................................................................................................... (C-12)

to keep dQij from becoming zero.


We now turn our attention to the continuity equation. Substituting Eq. C-7 into Eq. C-2 to give

∑ (Q )
J
*
ij + Qij′ ζ ij = − d i .................................................................................................................................................... (C-13)
j =1

Substituting Eq. C-10 into the above equation, it reduces to


24 SPE 134265

⎧ ⎫
⎪ ⎪
J
⎪ * p′n,ij − pi′ ⎪
∑ ⎨Qij +
j =1 ⎪ ⎡ ∂g ij 2 ⎤ ⎪
⎬ζ ij = − d i ............................................................................................................... (C-14)
⎪ sij H ij ⎢2 g ij Qij + Qij ⎥ ⎪
⎢⎣ ∂Qij ⎥⎦ ⎭

For reservoir nodes, the external volumetric flow into node i can be give by
d i = qi + a Jx ( pres − pi ) + bJx .......................................................................................................................................... (C-15)
Using Eq. C-6,
d i = qi + a Jx ( pres − pi* − pi′ ) + bJx .................................................................................................................................. (C-16)

If Eq. C-16 is substituted into Eq. C-14 we obtain an equation which can be solved for pi′. The result is
⎡J ⎤
pi′ = ⎢∑ (cij pn′ ,ij ) + bi ⎥ / cii , i = 1, 2, …, I ...................................................................................................................... (C-17)
⎢⎣ j =1 ⎥⎦
where
J ⎧ ⎫⎪
⎪ 1
cii = ∑ ⎨ * ⎬
+ a Jx .............................................................................................................. (C-18)
⎩ H ij [2 g ij Qij + Qij (∂g ij / ∂Qij )] ⎪⎭
j =1 ⎪
* *2

1
cij = ............................................................................................................................... (C-19)
H ij [2 g ij Q + Qij*2 (∂g ij / ∂Qij* )]
*
ij

J
bi = qi + a Jx ( pres − pi* ) + bJx + ∑ Qij*ζ ij ........................................................................................................................ (C-20)
j =1

Eq. C-17 can be written in the matrix form:


Ap′ = b ......................................................................................................................................................................... (C-21)
with
aii = cii ............................................................................................................................................................................ (C-22)

ai,nij = - cij ....................................................................................................................................................................... (C-23)

Appendix D - Reservoir Flow Model


During the acidizing process, the wellbore rate and the reservoir inflow at any location are changing with time so transient effects
become dominant in the reservoir. Following our previous work (Mishra et al. 2007), we present dimensionless groups used for
flow of slightly compressible liquids that are being injected at constant rate. For radial flow, the dimensionless pressure is defined
by
kh( pi − p)
pD = ............................................................................................................................................................ (D-1)
141.2qBμ
The dimensionless time is defined by
4.395 × 10 −6 kt
tD = ......................................................................................................................................................... (D-2)
φμct rw2
where t is in minutes.
We assume that the well can be represented as a line source, in other words, the wellbore is infinitesimally small (rw →0). Also
we assume that the effects of the outer boundaries of the reservoir are not seen and the reservoir acts as if there were no boundaries
(i.e., the reservoir is infinite-acting). Based on these assumptions, the line-source solution in dimensionless variables is given by
1 1
p D (t D ) = − Ei(− ) ................................................................................................................................................... (D-3)
2 4t D
SPE 134265 25

The early radial flow regime is valid either until the transient reaches a vertical boundary (Lee et al. 2003) or when flow from
beyond the end of the wellbore becomes important; that time is given by
7500 Lwφμct
t end = (in minutes) ...................................................................................................................................... (D-4)
k
For a long horizontal well, tend calculated by Eq. D-4 becomes very large so that the early-radial flow assumption made in the
model is reasonable. To validate this statement, Eq. D-3 is compared with a commercial reservoir simulation software program.
Fig. D.1 shows boundary conditions and finite difference grid system employed in this work. A rectangular reservoir with the
dimension of 10000’ × 10000’ × 100’ is assumed. No flow boundary conditions are assumed at the reservoir boundaries in x-, y-, z-
directions. A horizontal well with the injection length of 1000’ is located in the center of the reservoir. In this study, we run two
cases as shown in the figure. Case 1 is full field simulation while Case 2 assumes no flow boundaries at the heel and toe of the
horizontal well.

Fig. D.1⎯Grid system and boundary conditions for numerical simulation.

Fig. D.2 shows comparison of injection pressures calculated by Eq. D-3 and the numerical simulation results (FDM). The
reservoir, well, and fluid data used in this calculation are listed as follows:
• Injection rate = 10 bpm
• Wellbore radius, rw = 0.354’
• Viscosity = 0.5 cp
• Total compressibility = 1.2×10-5 1/psi
• Permeability = 3 md
• Porosity = 0.38
• Skin = 1.2

The results indicate the analytical and numerical calculation results show a very good match at t = 0.1 to 10 hr. After t = 10
hrs, the effects of the top and bottom boundaries of the reservoir become significant. However, in most stimulation jobs, the
duration of acid injection is much less than 10 hrs. So the application of Eq. D-3 will give a reasonable approximation. In addition,
comparing Case 1 and Case 2 results, the effect of the end flow will be insignificant for relatively long horizontal well for t < 10
hrs.

8,000
Analytical (Radial)
7,000 FDM (Full field)
Injection Pressure, psia

FDM (Segmented)

6,000

5,000

4,000

3,000
0.01 0.1 1 10 100
Time, hr

Fig. D.2⎯Effect of the reservoir boundary in radial flow.


26 SPE 134265

On the other hand, for relatively short injection intervals (e.g., cluster perforation completions), the radial flow approximation
may not be valid due to very small Lw. Similarly to limited entry flow tests, the non-radial flow regime may be developed in the
vicinity of the wellbore at early times. The early-time pressure behavior from the short perforation interval can be adequately
described by a spherical source rather than a cylindrical source. For spherical flow, the dimensionless pressure is defined by
krws ,eq ( pi − p)
pD = ........................................................................................................................................................ (D-5)
141.2qBμ
and the dimensionless time is defined by
4.395 × 10 −6 kt
tD = ......................................................................................................................................................... (D-6)
φμct rws2 ,eq
where rws,eq is the equivalent spherical radius that has potential strength equal to the cylindrical wellbore. According to Rodriguez-
Nieto and Carter (1972), the equivalent spherical radius can be expressed as
Lw
rws ,eq = ..................................................................................................................... (D-7)
⎛ (1 − z ) + (1 − z ) 2 + (r / L )2 ⎞
⎜ w w ⎟
ln⎜ ⎟⎟
⎜ − z + z + (rw / Lw )
2 2
⎝ ⎠
Fig. D.3 shows a comparison of Eq. D-7 and numerical simulation results for the well length of 20’ and the wellbore radius of
0.175’. The results show that the equivalent spherical radius with z = 0.15 show good agreement with the numerical simulation
results. Although some literature (Rodriguez-Nieto and Carter 1972; Raghavan and Clark 1975) suggests the use of z = 0.5, we use
z = 0.15 in Eq. D-7 to calculate the equivalent spherical radius.

13,000

11,000
Injection Pressure, psia

9,000

7,000

5,000

3,000
0.01 0.1 1 10 100
Time, hr

Fig. D.3⎯Determination of the equivalent radius for spherical flow.

Finally, the dimensionless pressure for spherical flow is calculated by (Brigham et al. 1980; Joseph and Koederitz 1985)

( )
p D (t D ) = (1 − e −1/ tD ) t D / π + erfc 1 / t D ...................................................................................................................... (D-8)

Fig. D.4 shows a comparison of Eq. D-8 and numerical simulation results for the perforation cluster with Lw = 10’ and rw =
0.354’. The injection rate is 1 bpm. The results indicate Eq. D-8 and numerical simulation results (Case 1) match very well at t <
10 hrs. Note that after t > 10 hrs, the transient reaches the upper and lower reservoir boundary from the well (same as Fig. D.2).
Case 2 forces flow to be radial with no-flow boundaries at the end of the wellbore. Clearly, Case 2 overpredicts the injection
pressure. Application of radial flow assumption to the cluster perforation causes larger pressure drops in the system. This is why a
pseudo-skin factor was introduced to the radial flow model as shown in the literature (Mishra et al. 2007). In this study, Eq. D-8 is
used for cluster perforation completions. The initial equivalent spherical radius can be obtained by Eq. D-7. Then once wormholes
start growing, the spherical flow radius is set to rwh.
SPE 134265 27

14,000

12,000

Injection Pressure, psia


10,000

Analytical (Spherical, z=0.15)


8,000 FDM (Full field)

FDM (Segmented)

6,000
0.01 0.1 1 10 100
Time, hr
Fig. D.4⎯Effect of the reservoir boundary in spherical flow.

Appendix E - Skin Factor Model for Wormholing Wells in Spherical Flow Geometry
A simple analytical model is developed for calculating convergent flow skin for a partially stimulated horizontal interval. The
model is correlated with well and stimulation dimensions such as wellbore radius, rw, perforation interval length, hperf, and
wormhole penetration radius, rwh. The analytical model is further validated with a series of 3D FEM simulations.
Fig. E.1 shows an illustration of a fully perforated horizontal well stimulated by matrix acidizing with limited perforation
technique. Wormhole penetration radius is measured from the center line of the wellbore. In this study, we assume the distance
between each perforation is far enough so that wormholes extend spherically from each perforation.

Fig. E.1⎯Partially stimulated long horizontal wells.

An analytical skin factor equation can be derived as follows. Consider idealized flow geometries in the near wormholed region as
shown in Fig. E.2.

Radial flow

Transition zone from


radial to spherical flow
rwh
rw Spherical flow

hperf
Fig. E.2⎯Idealized flow geometry for wormholes extending from a single perforation tunnel.

There exists no flow boundary at the half-distance between each perforation hole. Flow into the wormholed region can be
characterized by spherical flow. The spherical flow occurs approximately from r = rwh to r = hperf/2. The spherical flow disappears
and approaches to radial flow in the region far away from the perforation due to the nature of the perforation pattern and well
geometry. Assuming that the radial flow takes place at r = hperf/2, the following formula can be derived. The total pressure drop in
the system is calculated by

qBμ ⎛⎜ 1 2 ⎞ ⎛ ⎞
⎟ + qBμ ln⎜ rb ⎟ .................................................................................................. (E-1)
pb − p wf = −
4πk ⎜⎝ rwh h perf ⎟ 2πkh perf ⎜ h perf / 2 ⎟
⎠ ⎝ ⎠
where rb and pb are the outer boundary of the system and the pressure at the outer boundary. Using the definition of the skin factor,
Eq. E-1 can be rewritten as
28 SPE 134265

h perf ⎛ 1 ⎞ ⎛ ⎞ ⎛ ⎞
s= ⎜ −
2 ⎟ + ln⎜ rb ⎟ − ln⎜ rb ⎟
⎜ ⎟ ⎜ h perf / 2 ⎟ ⎜r ⎟
2 ⎝ rwh h perf ⎠ ⎝ ⎠ ⎝ w⎠
............................................................................................................... (E-2)
⎛ h perf ⎞ ⎛ 2r ⎞
= ⎜⎜ − 1⎟⎟ + ln⎜ w ⎟
⎜ h perf ⎟
⎝ 2rwh ⎠ ⎝ ⎠
or
h perf ⎛ h perf ⎞
s= − ln⎜⎜ ⎟ − 1 .................................................................................................................................................. (E-3)

2rwh ⎝ 2rw ⎠
The analytical model presented in Eq. E-3 is based on an assumption of no wellbore blockage effect. For short wormholes, the
wellbore blockage effect can be significant. However, practically speaking, once wormholes become sufficiently large such that
rwh >> rw, the effect can be negligible. Eq. E-3 is compared with a 3D FEM simulation results in Fig. E.3. The perforation spacing
is assumed hperf = 10 ft.

14
3D FEM (no wellbore)
12
3D FEM (No flow at wellbore)
10 3D FEM (flow at wellbore)
8 Analytical (No wellbore)
Skin Factor

Analytical (flow at wellbore)


6

-2

-4
0.1 1 10
Wormhole Penetration Radius, ft

Fig. E.3⎯Comparison of the analytical model with 3D FEM simulation results.

The results indicate that Eq. E-3 (labeled by blue) matches very well with the 3D FEM simulation results without the existence
of the wellbore. The 3D FEM simulation results with no flow at the wellbore show that the effect of the wellbore blockage is only
important for very short wormhole penetration lengths, which can be negligible for practical applications.
The third case represents openhole liners with a good filter cake removal allowing additional flow from the wellbore surface.
As shown in the simulation results, the skin factor becomes favorable, particularly for low wormhole penetration lengths.
However, for sufficiently long wormhole penetrations (rwh > 2 ft at the perforation spacing of 10 ft), the effect of wellbore inflow
is insignificant. These observations help an engineer decide whether more perforations should be added to reduce the convergent
flow skin. Similarly to Eq. E-3, an analytical equation for openhole cases can be derived by introducing an equivalent wormhole
length as
⎛ ln[h perf /( 2rw )] ⎞ 2 h perf / 2
rwh , eq = ⎜ ⎟r + ..................................................................................... (E-4)
⎜ (h perf / 2){1 + ln[h perf /( 2rw )]} ⎟ wh 1 + ln[h perf /( 2rw )]
⎝ ⎠
and
h perf ⎛ h perf ⎞
s= − ln⎜⎜ ⎟ − 1 ................................................................................................................................................ (E-5)

2rwh,eq ⎝ 2rw ⎠

You might also like