Beruflich Dokumente
Kultur Dokumente
Saudi Aramco
THE SAUDI ARAMCO JOURNAL OF TECHNOLOGY
Journal of Technology
A quarterly publication of the Saudi Arabian Oil Company
Additional articles that were submitted for publication in the Saudi Aramco Journal
of Technology are being made available online. You can read them at this link on
the Saudi Aramco Internet Website: www.saudiaramco.com/jot
The Saudi Aramco Journal of Technology is EDITORIAL ADVISORS (CONTINUED) PRODUCTION COORDINATION
published quarterly by the Saudi Arabian Oil
Company, Dhahran, Saudi Arabia, to provide Omar S. Al-Husaini Richard E. Doughty
the company’s scientific and engineering General Manager, Drilling & Workover Operations
communities a forum for the exchange of Abdul Hameed A. Al-Rushaid DESIGN
ideas through the presentation of technical Chief Drilling Engineer
information aimed at advancing knowledge Graphic Engine Design Studio,
in the hydrocarbon industry. Nabilah M. Tunisi Austin, Texas, U.S.A.
Chief Engineer
Complete issues of the Journal in PDF format
are available on the Internet at: Ammar A. Nahwi
http://www.saudiaramco.com Manager, Research and Development Center
(click on “publications”). Ali A. Meshari
Manager, EXPEC ARC
SUBSCRIPTIONS
ISSN 1319-2388.
Send individual subscription orders, address CONTRIBUTIONS
changes (see page 73) and related questions to: Relevant articles are welcome. Submission © COPYRIGHT 2016
guidelines are printed on the last page. ARAMCO SERVICES COMPANY
Saudi Aramco Public Relations Please address all manuscript and editorial
Department ALL RIGHTS RESERVED
correspondence to:
JOT Distribution No articles, including art and illustrations,
Box 5000 in the Saudi Aramco Journal of Technology
Dhahran 31311, Saudi Arabia EDITOR
except those from copyrighted sources,
Website: www.saudiaramco.com/jot William E. Bradshaw may be reproduced or printed without the
The Saudi Aramco Journal of Technology written permission of Saudi Aramco. Please
EDITORIAL ADVISORS C-11B, Room AN-1080 submit requests for permission to reproduce
North Admin Building #175 items to the editor.
Ibraheem M. Assa´adan Dhahran 31311, KSA
Vice President, Exploration Tel: +966-013-876-0498 The Saudi Aramco Journal of Technology
E-mail: william.bradshaw.1@aramco.com.sa gratefully acknowledges the assistance,
Ahmad O. Al-Khowaiter contribution and cooperation of numerous
Chief Technology Officer Unsolicited articles will be returned only operating organizations throughout the
when accompanied by a self-addressed company.
Abdullah M. Al-Ghamdi envelope.
Executive Director, Gas Operations
Amin Nasser
Khalid M. Al-Abdulqader President & CEO, Saudi Aramco
General Manager, Unconventional Resources
Nasser A. Al-Nafisee
Vice President, Corporate Affairs
Abdulla I. Al-Isa
General Manager, Public Affairs
Winter 2016
Saudi Aramco
THE SAUDI ARAMCO JOURNAL OF TECHNOLOGY
Journal of Technology
A quarterly publication of the Saudi Arabian Oil Company
Contents
ABSTRACT INTRODUCTION
Economic recovery and production of nonassociated gas Some gas-bearing clastic reservoirs in Saudi Arabia are tight
from tight clastic reservoirs faces some challenges from the and so can be produced at an economical gas rate only by
reservoirs’ high bottom-hole temperatures, high in situ stress, applying hydraulic fracturing. These tight reservoirs present
low permeability, and high Young’s modulus and rock com- multiple challenges to the successful achievement of hydraulic
pressive strength. Hydraulic fracturing, which is needed to fracturing operations.
exploit tight reservoirs, is constrained in these cases by high
fracture initiation and treating pressure, risk of premature • High bottom-hole static temperatures (270 °F to 325
screen out and conductivity degradation with time. A con- °F) can significantly reduce the stability and viscosity
ventional fracturing treatment in such a challenging envi- of the fracturing gel. So high polymer concentrations,
ronment requires a higher polymer loading in the fracturing 40 lb to 50 lb per 1,000 gal of fluid, are needed to
fluids to stabilize fluid viscosity and the use of smaller size maintain fracturing fluid viscosity at an acceptable level
proppant at low concentrations to ensure proppant place- during the entire treatment. This high gel concentration
ment. This results in a shorter effective fracture half-length can leave a large amount of residue, causing significant
and low fracture conductivity. The reduced contact area in proppant pack damage and consequently reducing effec-
turn decreases production potential. tive fracture half-length.
Additionally, high polymer loading is not easily breakable
• High in situ stresses — 0.9 psi/ft or above — and a high
and may cause major damage to the proppant pack, thereby
Young’s modulus — 5 to 10 × 106 psi — significantly
substantially reducing fracture conductivity. These challenges
increase the risk of screen out due to the creation of
are compounded when the reservoir is relatively tight and so
thin fractures. For certain challenging fields, a high
often cannot generate enough energy to clean up the injected
screen out ratio was seen at 35%1. To mitigate this risk,
fracturing fluids. To overcome these challenges, channel
a low final proppant concentration, measured as pound
fracturing was introduced. In this kind of fracturing, prop-
of proppant additive (PPA) per gallon of clean fluid,
pants are added in pulses in the fracturing fluids along with
ranging from 4 PPA to 6 PPA is typically used, which
dissolvable fibers, creating stacks of pillar-like structures
negatively impacts fracturing width and conductivity.
inside the fracture. These proppant pillars stay suspended
and are kept intact by the fibers during the treatment. Once • Low quality reservoirs in heterogeneous formations re-
pumping is stopped, the fracture closes on the pillars, and quire the creation of fractures that penetrate all potential
the fibers degrade under formation temperature. Because the pay zones and have a long fracture half-length with suffi-
pillars holding the fracture open act as columns, the voids cient conductivity to maximize reservoir contact and pro-
surrounding them are essentially stable channels that extend vide a conductive pathway for the gas to flow across the
along the entire geometry of the fracture. The channels pro- entire created fracture. This is not feasible with a conven-
vide an open pathway for hydrocarbon flow, resulting in a tional fracturing approach due to the fracture conductiv-
near-infinite conductivity environment. Channel fracturing ity damage caused by the high gel concentrations and low
has other benefits: The technology reduces the amount of proppant concentrations that this approach requires, as
proppant pumped compared to a conventional fracturing well as proppant damage due to crushing or embedment.
treatment, and the pulsation of the proppant during pump- The situation is made worse when screen outs occur and
ing reduces the risk of an early screen out. the desired fracture geometry cannot be placed properly.
PROPPED
Fig. 1. Representation of fractures achieved with channel fracturing and conventional proppant fracturing, and a schematic indicating different fracture lengths.
Fig. 1. Representation of fractures achieved with channel fracturing and conventional proppant fracturing,
and a schematic indicating different fracture lengths
substitutes the usual homogeneous proppant pack in the CHANNEL FRACTURING IN SAUDI RESERVOIRS
fracture with a heterogeneous structure containing a network
of open channels, Fig. 1. The key difference in pumping To date, more than 50 stages have been successfully treated
schedules is that channel fracturing, unlike conventional using channel fracturing in Saudi Arabian reservoirs. While
proppant fracturing, adds proppant in short pulses, sep- around 37% of stages experience screen out with conventional
arated by pulses of clean fluid. This discontinuity in the fracturing in some tight reservoirs, no screen out occurred
proppant addition helps reduce the net pressure during when similar wells were treated using channel fracturing
treatment and ensures reliable placement of the proppant technology. Typically, channel fracturing treatments were
with minimal, if not totally eliminated, risk of screen out. pumped in wells where high bottom-hole treating pressures
Statistics on 10,000 treatments show an average 99.9% suc- (BHTP) were expected, based on mechanical properties and
cess rate in proppant placement through channel fracturing2. a mini-fracture analysis. If conventional fracturing were to be
Abdelhamid et al. (2013)3 reports increased well productivity used in such wells, there is a very high probability of screen
and total elimination of screen out after implementation of out due to limited fracture width and premature proppant
channel fracturing in the Silah field in Egypt, while screen packing. In the case of channel fracturing, however, prema-
out occurred in more than 45% of the standard fracturing ture proppant packing is eliminated with the pulsed addition
treatments, according to the statistics over a two-year period. of proppant. Figures 2 and 3 show examples where the BHTP
Successes in faster gel recovery and improved productivity (indicated by the gray lines) was kept just below the comple-
are reported by Kayumov et al. (2014)4 and Valenzuela et al. tion limit of 16,000 psi during channel fracturing treatments.
(2012)5 in gas fields in Russia and Mexico, respectively. Overall, production results after channel fracturing are
Channel fracturing does not depend on the proppant pack more consistent compared to those after conventional prop-
to provide conductivity; rather it depends on the open flow pant fracturing. For the fracturing treatments where the data
channels, which achieve conductivity that is a few orders of for offset wells were available, production results after chan-
magnitude higher compared to the proppant pack conduc- nel fracturing show more consistency and better performance
tivity. Gillard et al. (2010)6 indicates that pressure buildup compared to those after conventional proppant fracturing.
(PBU) tests conducted on wells in Argentina’s Loma La Lata The next section provides several case studies illustrating the
field after channel fracturing confirmed that fracture conduc- application of channel fracturing and the results obtained
tivity was above a threshold value and might be considered with that approach.
Fig. 2. Channel
effectively fracturing
infinite for treatment
given reservoir on Well-S04.
conditions. Long-term
performance data for the same wells support the notion that Case Study #1: Well-S01
the structure of the created channels within the proppant
pack remained stable for no fewer than two years during the Vertical Well-S01, located in a recently developed area,
production period. Furthermore, even when highly concen- underwent the first fracturing operation in this field. The
trated fracturing gel is pumped, it can be easily flowed back formation is deep sandstone, exhibiting a high initial reser-
through the open channels, thereby minimizing damage to voir pressure, a thick net pay interval and low permeability
the fracture and providing effective fracture half-length near- — less than 0.1 millidarcy (mD). Channel fracturing tech-
ing the created hydraulic fracture half-length. nology was selected to minimize the risk of screen out and
maximize reservoir contact. A mini-fracture operation was
used to compute fracture characteristics and reservoir flow
Fig. 2. Channel fracturing treatment on Well-S04. Besides the good production achieved in Well-S01, the
entire proppant amount was successfully placed. A con-
ventional fracturing treatment pumped on an offset well,
Well-S05, resulted in screen out with only 55,000 lb of
proppant placed into the formation. As a result, Well-S05
did not meet production targets, and flow back data analysis
confirmed much lower PI compared to Well-S01. Another
well, Well-S02, was later conventionally fractured success-
fully in the same field with 100,000 lb of proppant at a max-
imum proppant concentration of 5 PPA. A greater proppant
amount or higher proppant concentration was not possible
Fig.
Fig. 3. 3. Channel
Channel fracturing
fracturing treatment
treatment on Well-S39.
on Well-S39. due to the high BHTP this would have required to break
down the formation and propagate the fracture, thereby
parameters. It revealed significant fracture height growth
Fig.risking
3. Channel
thefracturing treatment
completion on Well-S39.
integrity. The calculated PI after frac-
able to cover the entire pay zone. The main fracturing treat-
turing was lower compared to the PI for Well-S01, although
ment was successfully conducted as per design with 130,000
Saudi Aramco: Company
reservoir General
properties Use wells were very similar. Figure
for both
lb of proppant pumped at a maximum concentration of 7
2
PPA. Due to the pulsation in proppant pumping, the created
fracture geometry
1 to date has not been compromised, and
it matches with the geometry achieved by pumping a higher
proppant amount — about 215,000 lb — in a conventional 2
2
1
Fig. 7.7.
Fig. Comparison of reservoir
Comparison parameters
of reservoir and post-fracture
parameters andproductivity for
post-fracture
Fig. 6. Nodal analysis for Well-S04. Well-S04 and Well-S25.
Fig. 6. Nodal analysis for Well-S04.
CONCLUSIONS
ACKNOWLEDGMENTS
REFERENCES
ABSTRACT 20 years near the injection wellbore and in the mixing zones
due to the formation of calcium carbonate and calcium sul-
Water injection is commonly used in oil fields for pressure fate scales, respectively. A negative feedback loop develops
support and productivity enhancement. Injection water, if between the initial exploitation of preferential flow paths
incompatible with formation water, can induce scale pre- and subsequent reservoir quality deterioration in these paths.
cipitation and impose a significant threat to economic and The kinetic approach described in this article offers a signif-
efficient field operation. Reliable scale prediction is of critical icant improvement in scale prediction, providing better results
importance in the selection of suitable injection water and than the traditional approach and leading to enhanced field de-
the development of an effective scale management strategy. velopment and a more cost-effective scale management strategy.
The traditional (thermodynamic) approach for scale pre-
diction only provides precipitation potential at different INTRODUCTION
mixing ratios of injection water and formation water. While
it describes the initial and final states of the system, it omits In an effort to increase or maintain the oil production rate
some intermediate processes and products that may be im- in a reservoir, water injection has long been used in daily
portant for operations. In addition, it may overestimate the operations; however, any freshwater or saltwater injected
total amount of the scale forms. In this article, we describe a into a reservoir could disrupt its chemical equilibrium and
new approach, the kinetic approach, which facilitates a more induce reactions with both the formation fluids and the
accurate prediction of scale formation. reservoir rocks. These reactions would modify the chemical
We demonstrate the kinetic approach in a parametric composition of the injection and formation fluids, and affect
study of seawater injection into a carbonate reservoir in the chemical and physical properties of the reservoir rock.
Saudi Arabia. The study used reactive transport modeling
Furthermore, water injection generates a supersaturated wa-
(RTM), which simulates coupled multiphase fluid flow
ter due to water mixing, and a variety of scales may form as
and chemical reactions; it has the capability of predicting
a result. Scale formation can cause reservoir damage, flow
time-dependent mineral scale formation and associated reser-
line blockage and facility failure, presenting a significant
voir quality deterioration. The formation of calcium sulfate
threat to economic and efficient field operations.
scale in the reservoir and the near wellbore regions is a ma-
An evaluation of the scaling potential of water injection
jor concern when there is a high calcium concentration in the
formation water and a high sulfate concentration in the in- Aqueous Species Seawater (ppm) Formation Water
jected seawater. We built a 2D RTM using TOUGHREACT pH 8 6.6
v1.2, with GUI pre- and post-processor PetraSim v2015, to
Ca +
2
664 50,843
simulate scale formation over a 20-year time span. Reservoir
Mg2+ 2,130 5,530
heterogeneities were considered in our model. The impact of
scale formation on the reservoir and well properties during Na+ 17,885 34,987
waterflooding operations was evaluated. K+ 665 872
The results indicate that the migration path of the solutes Cl- 31,011 163,474
depends highly on the reservoir permeability distribution: HCO3- 150 9.6
preferential flow paths form in the high permeability layers. SO42- 4,321 147
At 20 years, the model shows up to 11.6% calcium sulfate
Sr2+ 12 2,108
scale in the mixing zones of the injection and formation
Br- 8e-6 8e-16
waters, and up to 18% calcium carbonate scale near the in-
jection wellbore — within a ~10 m radial distance from the Table 1. Concentration of aqueous components of seawater and formation water
in the simulations
wellbore. So a severe reduction in porosity is predicted at
Thermodynamic
Thermodynamic of over 500 barrels of fluid per day per ft1. Studies show that
more than 50% of the total flow of a well can be produced
Anhydrite
2
from a single super-K zone2. Therefore, a super-K zone may
form a critical layer to the flow and dominate the flow paths
in a reservoir. The problem arises in a reservoir undergoing
1
active waterflooding for pressure support; a super-K zone
may cause early water breakthrough and increased water
production (due to focused flow through this zone). Poor
0 sweep efficiency may result.
40% 20% 40% 60% 80% 100%
Kinetic, 1 day
Kinetic, This article describes the kinetic approach for scale pre-
Seawater Ratio
Seawater ratio
Kinetic, 10
Kinetic, day diction and provides a parametric study of waterflooding in
Kinetic, 30
Kinetic, 30 day
day a carbonate reservoir in Saudi Arabia. Formation of calcium
Fig. 1. Comparison of the amount of calcium sulfate (anhydrite or CaSO4)
Fig. 1. 3
Comparison of the amount
Kinetic, 1 yearof calcium sulfate (anhydrite or CaSO4) scale predicted to form at
Kinetic, seawater/formation
scale predicted to form at different 1 year water mixing ratios sulfate scale in this reservoir and in the near wellbore region
different seawater/formation water mixing ratios by thermodynamic and kinetic approaches. The kinetic
(g/L)
Thermodynamic
Thermodynamic scale formation information. Phreeqc
approach can provide time-dependent 25 was used for
time-dependent scale formation information. Phreeqc V2.025 was used for the
calculations in a batch system. the formation water and the high sulfate concentration in the
calculations in a batch system.
Anhydrite
model using thermodynamic methods. 0.08 It deals with the impact of the potential scale formation on the reservoir and
Calcite or dolomite (g/L)
3 Calcite
initial and final states of the system, e.g., far from equilib- well properties, taking into account reservoir heterogeneities,
Anhydrite (g/L)
(g/L)
4 0.1 4 0.1
Anhydrite 1 year
Calcite
Calcite or dolomite (g/L)
3 3 Calcite 10 day
Calcite or Dolomite (g/L)
Anhydrite (g/L)
(g/L)
(g/L)
Dolomite 1 year
Dolomite
(g/L)
2 2
Anhydrite
0.04 0.04
1 1
0.02 0.02
0 0
0 0
0% 50% 100%
0% 50% 100%
Seawater Ratio
Seawater ratio Seawater Ratio
Seawater ratio
Fig. 2. Comparison of the amount of scales formed at different seawater/formation water mixing ratios using the thermodynamic approach (left) and kinetic approaches
Fig. 2. Comparison of the amount of scales formed at different
(right). Calcite will form first before converting into dolomite, but the thermodynamic method only provides the final status (dolomite formation) and ignores the
using the thermodynamic approach (left) and kinetic approaches
intermediate product (calcite). Phreeqc V2.0 was used for the calculations in a batch system. converting into dolomite, but the thermodynamic method only provides
formation) and ignores the intermediate product (calcite). Phreeqc
batch system.
SAUDI ARAMCO JOURNAL OF TECHNOLOGY WINTER 2016 11
Normalized Porosity Permeability RSA scale formation sequence. In Fig. 2, the two im-
Lithology portant scales formed, according to the thermody-
Depth (m) (%) (md) (cm2/g)
Mud-lean namic approach, are calcium sulfate and dolomite
0 17 100 500 (CaMg(CO3)2), whereas calcium sulfate and calcite
Packestone
Mud-lean (CaCO3) are the important scales in the kinetic
1 17 100 500
Packestone approach. That is because dolomite precipitation
Mud-lean is very slow5. It is not likely to precipitate a signif-
2 17 100 500
Packestone icant amount of dolomite at the production time
3 18 3 Packestone 3,750 scale. Instead, calcium carbonate will form first and
4 18 3 Packestone 3,750 eventually convert into dolomite. Therefore, the
thermodynamic approach ignores both the interme-
5 18 3 Packestone 3,750
diate product — calcium carbonate — and the more
6 18 3 Packestone 3,750
complicated scale formation sequence.
7 17 1 Wackestone 7,470
8 17 1 Wackestone 7,470
9 17 1 Wackestone 7,470 THE KINETIC APPROACH
10 17 1 Wackestone 7,470
11 17 1 Wackestone 7,470 To overcome the limitations of the thermodynamic
12 9 0.1 Mudstone 9,570 approach, the new model needed a mathematical ex-
pression to say how fast scale precipitates for a given
13 17 1 Wackestone 7,470
time interval. A reasonable assumption about the ki-
14 17 1 Wackestone 7,470
netics controlling mineral dissolution or precipitation
15 17 1 Wackestone 7,470 might be that the rate of dissolution/precipitation is
16 17 1 Wackestone 7,470 proportional to the degree of supersaturation or un-
17 15 250 Grainstone 100 dersaturation. A general form of this rate law6, 7 is:
Grainstone
17.3 18 1,200 100 r
(super-K) R kaHi ( fGr )
(1)(1)
18.3 15 250 Grainstone 100 s
19.3 17 1 Wackestone 7,470 where R is the dissolution/precipitation rate in
mol m-2 s-1 — positive values indicate dissolution
Table 2. Simplified rock properties of a carbonate formation in Saudi Arabia and negative values indicate precipitation — rwhere
r E 1 where 1 R is- i R
theR k exp arate
dissolution/precipitation ( inmol s-1 kgw) a ( fGr )
scale formation information. In the case of mixing seawater 1 s
; S is the specific reactive R perTkg of298
area
surface .15k is H
H2O;
and formation water in a carbonate reservoir, Table 1, at
the temperature-dependent rate constant in mol m-2 s-1; +
different ratios, the thermodynamic method provides the is the activity of H
is the activity of H+; and i is the empirical reaction order
maximum amounts of calcium sulfate scale expected to form
accounting r for catalysis by H+ in the solution. is
at each seawater/formation water mixing ratio, Fig. 1. The R ka i
( fG ) G (1)
f (Gon
the rate sdependence ) 1 chemical
r
the exp driving
r
force — how(3) su-
H
greatest amount of calcium sulfate scale (~3.8 g/L) forms at
a 90%:10% ratio of seawater to formation water. But when
persaturated and/or undersaturated RT
is thesolution — of the
r reaction,i ΔGr. The overall rate law becomes:
chemical kinetics is considered, the amount of calcium sul- R kaH ( fGr ) (1)
s r E
( fGra)( G1r ) a(1)
1
fate scale that is formed is significantly different at different i 2
R
R ka
k exp i
( fGr ) (2) (2)
reaction times, Fig. 1. In addition, the greatest amount of ss f (Gr ) R
H
exp
T 298.15 1H (4)
calcium sulfate scale predicted to form on the first day oc- RT where
curs at a 70%:30% seawater:formation water ratio. After a r Ea 1 1 i where
R where
k exp ( ) aisH a (function
fGr ) using (2)
year, the amount of scale predicted to form using the kinetic s r R T EG a 1
298 .15 1
r (T the
i
Rf ( G 1kexp ) a H ( fGr ) (2)
s )energy,
exp
approach is close to that predicted from the thermodynamic
activation Ea, toRcalculate 298 . 15
rate (3)
constant at the tem-
approach; however, this presumes a batch system without
RT aGibbsC freeenergyC faction of C
mass transport. If fluid flow and transport is considered, it
perature of interest.
the reaction based tG on
We used
(Ci ) D i q i 8-10 i
x
the Transition
x Theoryx(TST) :k t k
State
may take a much longer time to reach equilibrium and formf (G ) 1 exp r G 2(3)
ff (( RT Gr r 1
G Gr)) 1exp
scale. Therefore, use of the thermodynamic approach may
exp (3) (4) (3)
overestimate the scale formation problem. RT RT 11
Saldidoes
et al.not
+(2009)
Another serious limitation of the old approach is that it
Ca
G
2+ + HCO
2
3 =the
- CaCO 3 (calcite) + H fit
r
Saldi et al. (2009) 11
finds that first-order dependence
does not fit
may give the wrong impression as to which scale(s) may formf (G )
r the
on exp
saturation 1used for dissolution
state 2 (4)does not fit with
RT Gr
at the production time scale because it does not consider the
(Gr )
the fsecond-order exp
dependence 1
observed (4)
for precipitation. (5)
2++ C
DRT Ci Ci
Ca
(Ci ) SOi 42-=q CaSO
4 (anhydrite)
(7)
12 WINTER 2016 SAUDI ARAMCO JOURNAL OF TECHNOLOGY
t x x x k t k
(5)
C C C
r
R ka
Furthermore, thei nonlinear
H
( fGr ) (1)
s
rate law, Eqn. 4, is based on
the Burton-Cabrera-Frank
(BCF) theory for crystal
E 1
growth12r, which describes 1
R k exp a ( ) a Hi ( fGr ) (2)
s
the experimental ofR T 298.15
data
magnesite precipitation with
great success. The BCF-type
rate law has also success-
G r
f (simulated
fully G) 1 secondary
exp r (3) R kaHi ( fGr ) (1)
s
mineralization in feldspar RT Fig. 3. Simplified conceptual model for a water injection well in the Middle Jurassic carbonate reservoir.
hydrolysis experiments13.
2 predecessor of the current TOUGHREACT is the multiphase
Gr fluid andr heat flow code Ea 1 TOUGH2. 1 TOUGHREACT
f (Gr ) exp 1 (4) (4) R k exp called ( ) a Hi ( fGr )
RT was builts upon R T 298.15 chemical specia-
TOUGH2, adding a general
The results of scale formation in a batch system when tion and reaction progress package based on EQ3NR3, 14, 15. In
using the kinetic approach for modeling are quite different this program, an integral finite difference approach is used to
from those using the thermodynamic approach. obtain
(5) space discretization; flow, transport and geochemical
Gr using a sequential iteration
C C C f (G)
reactions are 1 exp
solved separately (3)
t
2DRTM
(Ci ) D i q i i
x x x t
k
approach 3
.
k RT
RTM is a tool that applies kinetic theory to understand the Governing Equations 2
Gr
dynamics
Ca2+ +ofHCO physical
3 =and
- chemical
CaCO processes
3 (calcite) + in
H+the sub- f (Gr )
(6) exp 1 (4)
surface. It rigorously couples fluid flow and geochemical RT equation
The governing mathematical employed in
reactions to predict the spatial and temporal distributions of TOUGHREACT to describe geochemical processes involving
fluid fluid-rock interactions can be written as3, 16:
Caand
2+ + rock
SOproperties. The change in concentration(7)
4 = CaSO4 (anhydrite)
2- over
time is a product of the sum of all transport processes —
C C C
advection, dispersion and diffusion — and the geochemical (Ci ) D i q i i
reactions. The complexity of the coupling between trans- t x x x k t k (5)
port and reaction terms, together with complex boundary
conditions, requires a numerical solution. With the increase where Ci is the concentration of a specific species in the pore
in computational power, RTMs have been developed that CaD
fluid; +the
2+ is HCO 3 = CaCO
-
diffusion (calcite)
term; q3 is the linear+fluid
H+ flow rate;
enable the simulation of complex geological systems over and ϕ is the porosity. The first two terms on the right-hand
extended time scales. side describe the transport process — diffusion and advec-
The simulations of seawater injection into a carbonate tion
Ca—2+while
+ SOthe
4 last term describes
2- = CaSO the effect of geochemi-
4 (anhydrite) (7)
reservoir, the selected parametric study, were conducted with cal reactions.
the computer code TOUGHREACT, a numerical simulation
program for chemically reactive, non-isothermal, density de- Geological Data
pendent flows of multiphase fluids in porous and fractured
media4. This was facilitated by an interactive graphical inter- As noted, for this initial tryout of the kinetic approach, we
face pre-processor and post-processor (PetraSim v2015). The considered a parametric study of seawater injection into a
carbonate reservoir in Saudi Arabia. A 19.3 m section with
good reservoir quality was selected for the model construc-
tion. Table 2 shows the simplified porosity, permeability and
lithology data for this section. Within this section, there is
a 0.3 m grainstone unit that has unusual high permeability
(overUse
Saudi Aramco: Company General one Darcy), which is typical of a super-K interval.
Fig.
Fig.4.4.A A
cross-section view of
cross-section the lithofacies
view distributiondistribution
of the lithofacies in the model forina the
water injection
model forwell in the
Middle Jurassic carbonate reservoir. This is a simplified representation of reservoir heterogeneity (Yellow
A 2D radially symmetric grid model was set up. The vertical
=a Grainstone,
water injection
Orangewell in the Packstone,
= Mud-Lean Middle Jurassic carbonate
Blue = Packstone, reservoir.
Purple This is Gray
= Wackestone, a =
assignment of cells reflects the rock properties in Table 2. In
Mudstone
simplified and Green = “super-K”).
representation of reservoir heterogeneity (Yellow = Grainstone,
Orange = Mud-Lean Packestone, pH Blue = Packestone, Purple = Wackestone, horizontal directions, 10 radial blocks with logarithmically
Gray = Mudstone and
6.08
Green
6.30 = “super-K”).
6.52 6.74 6.96 Perm (md)
0
Perm (md)
1000
increasing radii represent the first 100 m distance from the
0 1000
(a)
(a) (b)
(b)
-4.3
SAUDI ARAMCO JOURNAL OF TECHNOLOGY WINTER 2016 13
-9.3
-
= Grainstone, Orange = Mud-Lean Packstone, Blue = Packstone, Purple = Wackestone, Gray =
Mudstone and Green = “super-K”).
pH Ca (M)
Ca (M)
6.08 6.30 6.52 6.74 6.96 Perm
Perm(md)
(md)
0.018 0.44 0.862 1.28 1.71 Perm
Perm(md)
(md)
0 1000 0.018 0.44 0.862 1.28 1.71 Perm
Perm(md)
(md)
0 1000 000 1000
1000
0 1000
1000
(a)
(a) (b)
(b) (a)
(a) (b)
(b)
(a)
(a) (b)
(b)
-4.3 -4.3
-4.3
-9.3 -9.3
-9.3
-14.3 -14.3
-14.3
(m)
(m)
Depth(m)
(m)
(m)
Perm (md)
(m)
Perm (md)
Depth
-19.3 0 1000
-19.3 0 Perm1000
(md)
Depth
Depth
Depth
-19.3 0 1000
Depth
(c)
(c) (d)
(d) (c)
(c) (d)
(d)
-4.3 -4.3
(c)
(c) (d)
(d)
-4.3
-9.3 -9.3
-9.3
-14.3 -14.3
-14.3
-19.3 -19.3
0 300 600 900 0 300 600 900 -19.30 200 400 600 0 200 400 600
0 200 400 600 0 200 400 600
Distance (m)
Distance (m) Distance (m)
Distance (m)
Distance (m)
Distance (m)
Fig.5.5.Temporal
Fig. Temporal and
and spatial
spatial evolution
evolution of pHofatpH
(a)at (a) 1 (b)
1 year, year, (b) 5 years,
5 years, Fig. 6.
(c) 10 and (d) 20 years.
(c) 10 years, Fig. 6.The
Temporaland
Temporal andspatial
spatialevolution
evolutionofofCa
Caconcentrations
concentrations(Molal)
(Molal)at at
(a)(a) 1 year,
1 year, (b)(b)
5 years, (c) 10 years,
initial Fig. 6. Temporal and spatial evolution of Ca concentrations (Molal) at (a) 1 year, (b) 5 years, (c) 10 years,
years,reservoir
and (d)permeability
20 years. Theas ainitial
function of depth
reservoir is shown onas
permeability the
a right for comparison.
function of depth is The vertical
and (d) 20
5 years, (c)years. The and
10 years, initial
(d)reservoir permeability
20 years. The initial as a function
reservoir of depth is
permeability asshown on the right for
a function
exaggeration is 30x. and (d) 20 years. The initial reservoir permeability as a function of depth is shown on the right for
shown on the right for comparison. The vertical exaggeration is 30x. comparison. The vertical
of depth is shown on theexaggeration is 20x.
right for comparison. The vertical exaggeration is 20x.
comparison. The vertical exaggeration is 20x.
HCO3 (M)
HCO3 (M)
1.57e-4 5.63e-4 9.69e-4 1.37e-3 1.78e-3 Perm
Perm(md)
(md)
Perm
Perm(md)
well; another 49 radial blocks (100 m length each) represent 1.57e-4 5.63e-4 9.69e-4 1.37e-3 1.78e-3 000
0
(md)
1000
1000
1000
1000
(a)
(a) (b)
(b)
the 100 m to 5,000 m distance from the well, Fig. 3. The grid -4.3
(a)
(a) (b)
(b)
-4.3
system comprises 1,180 active cells. A denser gridding assign-
-9.3
-9.3
ment near the injection well allows
Saudi for the
Aramco: Company evaluation
General Use of well
-14.3
performance during injection. Figure 4 shows a cross-section Depth (m)
(m)
(m)
(m) -14.3
Perm (md)
-19.3 0 Perm1000
(md)
-19.3 0 1000
Depth
(c)
(c) (d)
(d)
The flow domain is initially specified as having 100% cal- -4.3 (c)
(c) (d)
(d)
-4.3
cium carbonate mineralogy. A constant pressure of 200 bars -9.3
-9.3
(or 20 MPa) and a temperature of 75 °C are assigned to the -14.3
-14.3
formation. The seawater injection is set at a constant rate of
-19.3
3 kg/s, which is equivalent to 1,630 barrels per day, for 20 -19.30
0
200
200
400
400
600 0
600 0
200
200
400
400
600
600
Distance
Distance (m)
years. The current injection rate used in our model is only for Distance
Distance (m)
the left boundary. This no-gradient boundary is implemented 6.83e-3 2.17e-2 Saudi 3.66e-2
Aramco: Company
5.15e-2 General Use
6.64e-2
Saudi Aramco: Company General Use
Perm
Perm(md)
(md)
00 1000
1000
-19.3 0 1000
-14.3
Chemical Parameters
-19.3
0 200 400 600 0 200 400 600
Distance
Distance (m)
(m)
The geochemical calculations include 11 primary aqueous
species (Na+, Ca2+, Mg2+, K+, H+, Cl-, SO42-, HCO3-, H2O, Sr2+ Fig.
Fig. 8.
8. Temporal andspatial
Temporal and spatialevolution
evolutionofofSOSO concentrations
4 concentrations
4
(Molal)
(Molal) at 1(a)
at (a) 1 year,
year, (b) 5 years, (c) 10
and Br-) and three solid phases (calcium carbonate, dolomite years,
(b) and (d)
5 years, (c)20
10years.
years,The
andinitial
(d) 20reservoir permeability
years. The as a function
initial reservoir of depthas
permeability is ashown on the right for
comparison.
function The vertical
of depth is shownexaggeration
on the rightis for
20x.comparison. The vertical exaggeration
and calcium sulfate). Br- is a conservative species used to is 20x.
Anhydrite (%)
trace the injectate migration. The concentration of aqueous 0 2.9 5.8 8.7 11.6 Perm
Perm
00
(md)
(md)
1000
1000
components of Arabian Gulf seawater — injection water used for their precipitation.(a)
A rate law from Arvidson
(a) (b) and
(b)
-4.3
— and the Middle Jurassic carbonate formation water were Mackenzie (1999) is used for dolomite precipitation, be-
5
-9.3
previously shown in Table 1. cause it is derived directly from the relevant experiments.
-14.3
A TST-type rate law is used for the calcium carbonate The kinetic parameters employed here are taken directly
(m)
Depth(m)
Perm (md)
-19.3 0 1000
and calcium sulfate dissolution, but a BCF-type rate law is from Palandri and Kharaka(c)(2004)17, except for dolomite.
Depth
(c) (d)
(d)
-4.3
-14.3
Fig. 8. Temporal and spatial evolution of SO4 concentrations (Molal) at (a) 1 year, (b) 5 years, (c) 10
years, and (d) 20 years. The initial reservoir permeability as a function of depth is shown on the right for
comparison. The vertical exaggeration is 20x.
Anhydrite (%)
Perm (md)
-19.3 0 1000
reservoir.
Depth
(c)
(c) (d)
(d)
-4.3 The formation water and the injection water contribute to
-9.3 a high Ca2+ and SO42-, respectively. When these two waters
-14.3
mix, a remarkable amount of calcium sulfate scale, up to
-19.3
11.6% at 20 years, occurs, Fig. 9. The solute mixing zones
0 100 200 300 0 100 200 300
Distance
Distance (m)(m)
where the maximum amount of calcium sulfate scale forms is
not at the place where the highest Ca2+ concentration occurs,
Fig. 9. Temporal
Fig. 9. Temporaland andspatial
spatialevolution
evolution of of calcium
calcium sulfate
sulfate (anhydrite)
(anhydrite) abundance
abundance (%) at (a) 1 year, (b) 5
years,
nor where peak SO42- is found, but where the highest activity
(%) at (c)
(a)10 years,(b)
1 year, and (d) 20 years.
5 years, (c) 10 The
years,initial
andreservoir permeability
(d) 20 years. as areservoir
The initial function of depth is shown on
the right for comparison. The vertical exaggeration is 10x.
permeability as a function of depth is shown on the right for comparison. The product of a(Ca+2) and a(SO42-) occurs, if we apply the mass
vertical exaggeration is 10x. action law to Eqn. 7:
(a)
(a) (b)
(b)
permeable grainstone facies and the super-K interval, while
-4.3
the second is at a depth of 0 m to -3 m, with the relatively
-9.3
permeable mud-lean packestone facies. The pH front migrates
up to 431 m at 1 year, 562 m at 10 years and 885 m at 20 -14.3
(m)
Depth (m)
years. At 20 years, the left ~9 m near the injection well has -19.3
Perm (md)
Depth
0 1000
which reduces the pH. This explains why the highest pH (~7), -9.3
occurring near the injection well, is lower than that of sea-
-14.3
water (8). Due to the mixing effects, a transition zone occurs
between the high pH area near the wellbore and low pH (6.6) -19.3
0 10 20 0 10 20
area dominated by the formation water. Distance(m)
Distance (m)
Fig. 10. Temporal and spatial evolution of calcite abundance (%) at (a) 1 year,
Fig.
Ca2+ + HCO3- = CaCO3 + H+ (6) (b) 5 years, (c) 10 years, and (d) 20 years. The initial reservoir permeability as a
comparison.
function of The
depthvertical exaggeration
is shown is 10x.
on the right for comparison. The vertical exaggeration
Porosity (%)
is 10x.
Ca2+, HCO3- and SO42- exhibit similar time-dependent 0 4.5 9 13.5 18 Perm
Perm(md)
(md)
00 1000
1000
(a)
(a) (b)
(b)
-4.3 SAUDI ARAMCO JOURNAL OF TECHNOLOGY WINTER 2016 15
-9.3
Fig. 10. Temporal and spatial evolution of calcite abundance (%) at (a) 1 year, (b) 5 years, (c) 10 years,
and (d) 20 years. The initial reservoir permeability as a function of depth is shown on the right for
comparison. The vertical exaggeration is 10x.
Porosity (%)
-14.3 CONCLUSIONS
(m)
Depth(m)
Perm (md)
-19.3 0 1000
Depth
(c)
(c) (d)
(d) We introduced a kinetic approach for predicting scale forma-
-4.3
tion during production. The kinetic method overcomes the
-9.3
limitations in the thermodynamic approach, which provides
-14.3
no information on how fast scales precipitate and ignores the
-19.3
0 100 200 300 0 100 200 300 intermediate processes and products that are important to
Distance
Distance (m)(m)
the waterflooding. The kinetic approach is a powerful tool
Fig. 11. Temporal and spatial evolution of porosity loss (%) at (a) 1 year, (b)
Fig. 11. Temporal and spatial evolution of porosity loss (%) at (a) 1 year, (b) 5 years, (c) 10 toyears,
predict the time-dependent spatial distribution of scales
and (d)
520years,
years. (c)The
10 initial
years,reservoir
and (d) 20permeability
years. Theas a function
initial of depth
reservoir is shownason
permeability a the right for near
comparison.
the wellbore and in the formation, as well as to show
The vertical
function exaggeration
of depth is shownison 10x.
the right for comparison. The vertical exaggeration
is 10x.
the associated reservoir quality deterioration. In addition, it
facilitates the optimal design of both pre-scale treatment and
carbonate scales form mainly near the injection well. They post-scale remediation.
Saudi Aramco: Company General Use
may be pushed away from the well by employing mechanical Using a parametric study of waterflooding in a carbonate
force — agitation, hydrodynamic strategies, high pressure, reservoir in Saudi Arabia, we evaluated the impact of scale
etc. Once released, the particles are either redeposited on the formation potential on the reservoir and well properties,
matrix, transported with flow, or entrapped at pore constric- taking into account the reservoir heterogeneities, during
tions, clogging pore throats . In addition, because calcium
21 waterflooding operations. The results indicate that reservoir
sulfate mainly precipitates at the high permeability, preferen- heterogeneity has a significant impact on the preferential
tial conduits, the deposition of calcium sulfate and calcium flow paths and so on the temporal and spatial distribution
carbonate scales will reduce the permeability of the preferen- of scales, which highlights the importance of reservoir char-
tial conduits and slow down the flow. This sets up a negative acterization and prediction of changes in its quality. The
precipitation of scale may dynamically change the reservoir
feedback loop between the initial exploitation of preferential
quality and flow behavior, and thereby affect the production
flow paths and subsequent deterioration in rock properties
efficiency.
in these paths. If there were no calcium sulfate and calcium
In the future, a multidisciplinary team with experts from
carbonate scales, the injection water would be transported
engineering (reservoir, production and process engineers),
much faster through the super-K zone and get into the
from reservoir description and simulation groups, and from
production wells. It can be argued, therefore, that the flow
the geosciences (geology, geophysics, hydrogeology and geo-
retardation in the preferential conduits, which changes the
chemistry) should be formed to diagnose the scale problems
flow behavior from focused flow toward uniform flow, may
during waterflooding in a reservoir and plan a cost-effective
increase the sweep efficiency.
scale management strategy. An integrated approach, includ-
ing reservoir characterization and the prediction of changes
MODEL LIMITATIONS
in its quality, e.g., sequence stratigraphy, facies, lithology
and rock physics, as well as kinetic RTM and reservoir simu-
The models in this study used the B-dot equation22 to cal- lation, should be used for the next-generation scale modeling
culate ion activity coefficients instead of using the Pitzer to provide optimal strategies for reservoir management.
ion interaction model23, 24, although a special version of
TOUGHREACT-Pitzer is available. Errors crop up when us- ACKNOWLEDGMENTS
ing the B-dot equation, with is an ion complexation model,
on the high ionic strength calculations. We conducted a The authors would like to thank the management of Saudi
speciation solubility calculation comparing the two methods Aramco for their support and permission to publish this
for calculating concentrations in the formation and injection article. We also would like to thank Dr. Shaun Hayton and
waters using code Phreeqc V2.025 with llnl.dat and Pitzer.dat Dr. Peter Birkle for their internal review, which significantly
databases for the B-dot and Pitzer methods, respectively. The improved the quality and clarity of the manuscript.
results indicate that using the B-dot method overestimated This article was presented at the SPE Kingdom of Saudi Arabia
the concentrations of ion pairs, such as CaCl+, CaSO4o and Annual Technical Symposium and Exhibition, Dammam, Saudi
NaSO4-, and underestimated the activity of Ca2+, SO42-, etc., Arabia, April 25-28, 2016.
which in turn underestimated the supersaturation states of
calcium carbonate and calcium sulfate. We therefore spec-
Fig. 1. Main principle of a PITT (Fig. 1a) and a SWCTT (Fig. 1b), as well as a comparison of typical test zones for PITTs and SWCTTs (Fig. 1c).
For PITT operations, it is imperative to determine the The injector-producer pair of I-2 and P-2 was chosen for the
partitioning coefficient (K-value) of each of the tracers to be PITT because of the possibility of corroborating the PITT
deployed. Since the K-value will differ with different oil-wa- Tracer K-value (Tracer Added to K-value (T
results with data that will be acquired via the observation
ter systems, it is important to determine these coefficients in Oil Phase) Water P
wells, Wells O-1 and O-2, as part of the elaborate M&S
the actual reservoir fluid (both oil and water). For the exper-
WTP-2 2.2 2.0
plan for the IOR pilot project. The tracer amounts were cal-
WTP-3 2.2 1.9
iments, a dead oil sample from the actual production well, culated
WTP-4using a volumetric method, 4.3 taking into account po- 3.6
Well P-2, was used. In addition, given the assumption that rosity, reservoir height and well spacing, as well as the tracer
the partitioning tracers will need to measure the ROS in the Table 1. Results
detection limits as from theinK-value
verified Sanni et experiments
al. (2015)7. The tracer
flooded zone, injection water of about 54,000 ppm salinity, injection operation was carried out at the onset of water
instead of the original connate water with its very high sa- injection. Two unique chemical water tracers, one qualified
linity of about 200,000 ppm, was used as the water phase in tracer and one pilot tracer, were injected into each injector
the experiments. well, using the best practices that were developed from previ-
The K-values of the tracers WTP-2, WTP-3 and WTP-4 ous IWCTT operations. The water tracers were mixed in the
were determined by conducting batch-type experiments. A same container to ease logistics, Table 2.
standard solution for the injection water and the oil phase, For the field injection, connection was made through a
each containing 100 ppm concentration of the individual trac- ½’’ female national pipe thread taper socket on a side valve
ers, was prepared. Then 5 ml of the oil standard was mixed to a manometer on the flow line about 10 m upstream from
with 5 ml of unspiked injection water and 5 ml of the water the wellhead. Such an injection point is generally well suited
standard was mixed with 5 ml of unspiked oil. Mixing was to this task, with its low dead volumes and its position on
done in 10 ml vials, ensuring a low headspace volume, Fig. 3. the top of the flow line. The tracers were added “on the fly”
The vials were then rigorously shaken before they were placed during normal operation of the water injection wells and were
in a heating cabinet at the reservoir temperature. At regular thereby diluted to low ppm concentration in the injection wa-
intervals, the vials were taken out, shaken, and immediately
ter immediately, so they had no potential to influence injected
placed back into the heating cabinet. On the third day, the
water chemistry, were compatible with surfaces, etc. After the
vials were shaken one last time. Then the two phases in the
injection, the pump and the high-pressure hose were flushed
vials were allowed to separate in the heating cabinet, after
which 1 ml of each phase was transferred into a headspace K-value K-value
vial for analysis. The tracer concentration in the phases was Tracer (Tracer Added (Tracer Added K-value
to Oil Phase) to Water Phase)
measured using a headspace technique in combination with
gas chromatography/mass spectroscopy (GC/MS). WTP-2 2.2 2.0 2.1 ± 0.2
The results from the K-value experiments are summarized WTP-3 2.2 1.9 2.1 ± 0.2
in Table 1. WTP-4 4.3 3.6 4.0 ± 0.4
Partitioning
WTP-3 10 10 TRACER RESULTS AND DISCUSSION
tracer
Partitioning
WTP-4 10 10
tracer The collected samples have been consistently analyzed over
the past 30 months to define both water breakthrough times
Table 2. Summary of tracers injected
and the geometry of the elusion concentration over time.
with 20 liters of freshwater to ensure complete injection of the It is important to note that these tracers were designed to
tracers and avoid potential contamination with other tracers. be detected at very low detection limits, i.e., in the parts
per trillion range. In addition to water breakthrough times,
Tracer Sampling reservoir heterogeneity, sweep efficiency, and ROS can be
inferred from careful analyses of the tracer curves.
The success of any tracer test is greatly dependent upon its
sampling protocol. It is extremely important to ensure that Residence Time Analysis for Interpreting Tracer Data
the sampling is done in a meticulous manner to avoid any
contamination. The sampling frequency should be congru- The distribution of the times it takes a population of passive
ent with anticipated possible breakthrough times. For the tracer particles to travel through a medium can be used to
current IWCTT and PITT, additional steps were taken by quantify
𝜕𝜕 the flow through 𝑞𝑞 𝑞𝑞 that medium. The tracer 𝑞𝑞 𝑞𝑞 particles
travel
𝜕𝜕𝜕𝜕
(∑ 𝑖𝑖=𝑜𝑜,𝑔𝑔,𝑤𝑤
through
𝜑𝜑 𝑆𝑆 𝑖𝑖
different
𝐾𝐾𝑖𝑖 𝐶𝐶 )
paths
+ ∇ ⋅
and
(∑ 𝑖𝑖=𝑜𝑜,𝑔𝑔,𝑤𝑤
therefore
𝒗𝒗 𝑖𝑖
use
𝐾𝐾𝑖𝑖 𝐶𝐶 ) − ∇ ∙ (∑𝑖𝑖=𝑜𝑜,𝑔𝑔,𝑤𝑤 𝜑𝜑𝑆𝑆𝑖𝑖 𝑫𝑫∗𝑖𝑖
different
the office project team to enhance the success of the project,
including the following: amounts of time to go from the inlet to the outlet. The distri-
𝜕𝜕
bution, of these times 𝑞𝑞 exit age𝑞𝑞distribution,
𝜕𝜕𝜕𝜕
(∑E(t),
𝑖𝑖=𝑜𝑜,𝑔𝑔,𝑤𝑤
𝑇𝑇
𝜑𝜑−𝑇𝑇𝑆𝑆𝑖𝑖 𝐾𝐾𝑖𝑖 𝐶𝐶 𝑞𝑞 ) is+ called
∇ ⋅ (∑the 𝑞𝑞
𝑖𝑖=𝑜𝑜,𝑔𝑔,𝑤𝑤 𝒗𝒗𝑖𝑖 𝐾𝐾𝑖𝑖 𝐶𝐶 ) − ∇ ∙ (∑𝑖𝑖=𝑜𝑜,𝑔𝑔,𝑤𝑤 𝜑𝜑𝑆𝑆𝑖𝑖 𝑫𝑫𝑖𝑖
∗
S𝑗𝑗o(𝑡𝑡)
𝐸𝐸 𝑇𝑇𝑅𝑅 −𝑇𝑇𝑊𝑊
=order
𝐶𝐶𝑗𝑗 (𝑡𝑡) ∙ 𝑄𝑄𝑗𝑗 (𝑡𝑡)/𝑀𝑀 𝑞𝑞 (4)
∑𝑗𝑗 ∫−∞ 𝐸𝐸𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡 = 1 (5)
WTP-2
(𝑡𝑡)/𝑀𝑀 The = (3)
zero moment
𝑇𝑇𝑅𝑅 +𝑇𝑇𝑊𝑊 (𝐾𝐾−1) (m0) represents =(2)
𝐾𝐾𝑤𝑤𝑞𝑞 the 𝑞𝑞
𝐶𝐶relative
𝑤𝑤 /𝐶𝐶 =amount
1 8.0E-4 WTP-3
1 𝜕𝜕of tracer produced in the production well, j; the first order mo- WTP-4
Ec(t) (1/day)
Ec(t) [1/day]
)= ∫ 𝜏𝜏 ∙ 𝐸𝐸𝑗𝑗 (𝜏𝜏)𝑑𝑑𝜏𝜏
𝑚𝑚1,𝑗𝑗 0ment (m
(7) 6.0E-4
) represents the average residence time for the tracers ∞ ∞
∞
1
(6) (3)
to extrapolations based on the data for larger times.
(10) 𝑚𝑚 𝐸𝐸(𝑡𝑡)
= ∫ =𝑡𝑡𝐶𝐶(𝑡𝑡) 𝑄𝑄 (𝑡𝑡)/𝑀𝑀
∙ 𝐸𝐸 ∙(𝑡𝑡)𝑑𝑑𝑡𝑡
∞1 𝜕𝜕̅ ) ∞ ∞ 2
and 𝑚𝑚 (𝑇𝑇
(𝑡𝑡)𝑑𝑑𝑡𝑡 𝑚𝑚 = ∫−∞ 𝑡𝑡 ∙ 𝐸𝐸𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡 (7) 𝑝𝑝
0 == ∫−∞ 𝐸𝐸𝑗𝑗∫0 𝜏𝜏 ∙ 𝐸𝐸,𝑗𝑗 (𝜏𝜏)𝑑𝑑𝜏𝜏 and
𝑤𝑤
𝜏𝜏)𝑑𝑑𝜏𝜏 Φ(𝑡𝑡) 𝜕𝜕 𝑀𝑀0 𝑗𝑗 (𝜕𝜕−𝜕𝜕0 )2
𝑉𝑉𝑤𝑤,𝑗𝑗 (𝑡𝑡) = ∫0 𝑄𝑄𝑗𝑗 (𝜏𝜏)(1 −1𝐹𝐹𝑤𝑤 (𝜏𝜏))𝑑𝑑𝜏𝜏
𝑚𝑚1,𝑗𝑗 (7)
2
(8)
𝐶𝐶(𝑡𝑡) = −∞ 𝑒𝑒𝑒𝑒𝑒𝑒 [− ]
were not measured, but can be estimated from the water cut
(9)
√𝐷𝐷0 𝜕𝜕 𝐷𝐷0 𝜕𝜕
𝐸𝐸𝑗𝑗 (𝑡𝑡) = 𝐶𝐶𝑗𝑗 (𝑡𝑡) ∙ 𝑄𝑄𝑗𝑗 (𝑡𝑡)/𝑀𝑀 (4) and the mass produced in the water phase14.
and combined in a 𝐹𝐹 − Φ diagram to quantify a measure of
1 𝜕𝜕 1 𝜕𝜕 𝐸𝐸𝑗𝑗 (𝑡𝑡) = 𝐶𝐶𝑗𝑗 (𝑡𝑡) ∙ 𝑄𝑄𝑗𝑗 (𝑡𝑡)/𝑀𝑀 (4)
the𝐹𝐹(𝑡𝑡) =
heterogeneity ∫of 𝐸𝐸 the (𝜏𝜏)𝑑𝑑𝜏𝜏
system.
(𝜕𝜕−𝜕𝜕 and
The) Φ(𝑡𝑡)
2 swept
(8) =
reservoir ∫
volume 𝜏𝜏 ∙ 𝐸𝐸 (𝜏𝜏)𝑑𝑑𝜏𝜏 (7)
𝑤𝑤 (𝜏𝜏))𝑑𝑑𝜏𝜏
𝑀𝑀 𝑗𝑗 𝑗𝑗
0
𝐶𝐶(𝑡𝑡) = 𝑚𝑚of0,𝑗𝑗time
as a function √𝐷𝐷
0𝑒𝑒𝑒𝑒𝑒𝑒 [−
can be estimated
0 𝑚𝑚1,𝑗𝑗 0
] from F(t). For a water
(9) 𝑀𝑀 = 𝑀𝑀𝑜𝑜 + 𝑀𝑀𝑤𝑤 = (1 + 𝐾𝐾 1−𝑊𝑊𝑊𝑊) 𝑀𝑀𝑤𝑤 (10)
∞ 0 𝜕𝜕 𝑞𝑞 𝑞𝑞 𝐷𝐷𝑞𝑞0 𝜕𝜕 𝑊𝑊𝑊𝑊 (10)
∑𝑗𝑗 ∫it−∞
tracer, 𝐸𝐸𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡
is given 18 𝐾𝐾𝑖𝑖 =
as: = 1 𝐶𝐶𝑖𝑖 /𝐶𝐶 (5) ∞
∑𝑗𝑗 ∫ 𝐸𝐸𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡 = 1 (5)
(𝜕𝜕−𝜕𝜕0 )2 𝜕𝜕 In the−∞ present case, water cuts were about 96%. For a
] 𝑉𝑉𝑤𝑤,𝑗𝑗 (𝑡𝑡) = ∫0 𝑄𝑄𝑗𝑗𝑞𝑞(𝜏𝜏)(1 (𝜏𝜏))𝑑𝑑𝜏𝜏
(9) − 𝐹𝐹𝑤𝑤1−𝑊𝑊𝑊𝑊 (8) (8)of 2.11 (WTP-2 and WTP-3), this yields M = 0.92
w (11)
𝑀𝑀 = 𝑀𝑀𝑜𝑜∞+ 𝑀𝑀𝑤𝑤 𝐶𝐶𝑖𝑖 = (1 + 𝐾𝐾 K-value
(10)𝐿𝐿𝑐𝑐 =∞2 ∫0 𝐹𝐹(Φ)𝑑𝑑Φ − 1 Saudi Aramco: Company General
𝐷𝐷0 𝜕𝜕 ∞ ) 𝑀𝑀𝑤𝑤 Use
𝑚𝑚0 = ∫−∞ 𝐸𝐸𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡 , 𝑚𝑚1 =𝑊𝑊𝑊𝑊 ∫−∞ 𝑡𝑡 ∙ 𝐸𝐸𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡 and 𝑚𝑚M, 2 =i.e.,∫about
−∞
𝑡𝑡
∞
28% of
∙ 𝐸𝐸𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡
the tracer is produced ∞ (6)with the oil. For
In cases where produced fluid is re-injected, contributions 𝑚𝑚0 = of
a K-value ∫−∞ 𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡 , a corresponding
4.0𝐸𝐸(WTP-4), ∙ 𝐸𝐸𝑗𝑗 (𝑡𝑡)𝑑𝑑𝑡𝑡 gives
𝑚𝑚1 = ∫−∞ 𝑡𝑡 evaluation and 𝑚𝑚2
𝑀𝑀 𝑞𝑞 (𝜕𝜕−𝜕𝜕 )2
due to re-injection can
𝐶𝐶 be [−removed 0 using de-convolution . 13
1−𝑊𝑊𝑊𝑊 𝐶𝐶(𝑡𝑡) = 1 0 𝜕𝜕 𝑒𝑒𝑒𝑒𝑒𝑒 ] (9) Mw = 0.86 M, i.e., about 14% of the tracer is produced with
)
Moment𝑀𝑀
𝑊𝑊𝑊𝑊 𝐿𝐿𝑐𝑐 = 2 ∫01𝐹𝐹(Φ)𝑑𝑑Φ
𝑤𝑤 √𝐷𝐷
analysis
0 𝜕𝜕 of −1
tracer (10)
𝐷𝐷 𝜕𝜕
curves
0 requires 1(11)𝜕𝜕
normalization the oil. By taking these corrections into account, we find
𝐹𝐹(𝑡𝑡) =
of RTDs, integrated ∫ 𝐸𝐸𝑗𝑗 (𝜏𝜏)𝑑𝑑𝜏𝜏
𝑚𝑚0,𝑗𝑗 0 to infinity. As and tracerΦ(𝑡𝑡) = 𝑚𝑚 includ-
campaigns, ∫0 𝜏𝜏 ∙ 𝐸𝐸𝑗𝑗 (𝜏𝜏)𝑑𝑑𝜏𝜏
estimated produced 1 𝜕𝜕 tracer mass (7) measurements of 0.45, 1 𝜕𝜕
0.29,
𝐾𝐾 𝑞𝑞
= 𝐶𝐶 𝑞𝑞
/𝐶𝐶 𝑞𝑞 1,𝑗𝑗 𝐹𝐹(𝑡𝑡) = ∫ 0
𝐸𝐸 𝑗𝑗 (𝜏𝜏)𝑑𝑑𝜏𝜏 and Φ(𝑡𝑡) = ∫ 𝜏𝜏 ∙ 𝐸𝐸𝑗𝑗 (𝜏𝜏)𝑑𝑑
ing the present one, are 𝑜𝑜 ended 𝑜𝑜 at a finite time after injection, 0.26 and 0.28 𝑚𝑚0,𝑗𝑗
for WT-60, WTP-2, WTP-3 and WTP-4, 𝑚𝑚1,𝑗𝑗 0re-
integration 1−𝑊𝑊𝑊𝑊
−1 𝑀𝑀 = 𝑀𝑀to 𝑜𝑜 +
infinity
𝑀𝑀𝑤𝑤 =
(11)
must
(1be+based 𝐾𝐾 𝑊𝑊𝑊𝑊 on an
) 𝑀𝑀extrapolation of
𝑤𝑤 (10)
spectively. The fact that the three partitioning tracers yield
Saudi Aramco:the tracer General
Company curves.𝜕𝜕UseVarious
𝑞𝑞 approaches
𝑞𝑞 for extrapolation have smaller values than the ideal one reflects that the secondary
been𝑉𝑉𝑤𝑤,𝑗𝑗 (𝑡𝑡) = 13,∫017,𝑄𝑄𝐾𝐾
suggested 19𝑗𝑗
= 𝐶𝐶𝑤𝑤 /𝐶𝐶−𝑞𝑞 𝐹𝐹
𝑤𝑤(𝜏𝜏)(1
. We used
= 1(𝜏𝜏))𝑑𝑑𝜏𝜏
the𝑤𝑤approach proposed by peak
(8) not=been𝜕𝜕 𝑄𝑄
has(𝑡𝑡)
𝑉𝑉𝑤𝑤,𝑗𝑗 ∫0 observed yet for the partitioning tracer
𝑗𝑗 (𝜏𝜏)(1 − 𝐹𝐹𝑤𝑤 (𝜏𝜏))𝑑𝑑𝜏𝜏
Viig et al. (2013) , based on fitting type curves of the form:
13
curves. This implies that the saturation estimated from the
1
𝐿𝐿𝑐𝑐 = 2 ∫0 𝐹𝐹(Φ)𝑑𝑑Φ 𝜑𝜑 − 1 (11) average residence time (𝑇𝑇̅) should be considered uncertain,
𝑀𝑀 (𝜕𝜕−𝜕𝜕 ) 2
𝐶𝐶(𝑡𝑡) = 0 𝜕𝜕 𝑒𝑒𝑒𝑒𝑒𝑒 [− 𝐷𝐷 0𝜕𝜕 ] (9) and that the mode 𝑀𝑀0
of the distribution
(𝜕𝜕−𝜕𝜕0 )2
(Tm) provides a more
√𝐷𝐷0 0 (9) 𝐶𝐶(𝑡𝑡)quantification
robust = 𝑒𝑒𝑒𝑒𝑒𝑒 of [− the saturation ] in thisSaudi
case.Aramco: Company (9)
∗ √𝐷𝐷0 𝜕𝜕 𝐹𝐹 − Φ 𝐷𝐷0 𝜕𝜕
𝐷𝐷𝑖𝑖 Several other interesting parameters are provided by(𝑇𝑇 the
̅)
with three parameters, D0, t0 and M0, to the data. For the RTD analysis, in addition to the fraction of produced mass
ideal tracer, a bimodal distribution 1−𝑊𝑊𝑊𝑊
was observed. To ex-
𝑀𝑀 = 𝑀𝑀𝑜𝑜 + 𝑀𝑀𝑤𝑤 Φ(𝑡𝑡)= (1 + 𝐾𝐾 ) 𝑀𝑀𝑤𝑤Saudi Aramco: Company(10) in the well (m0). The heterogeneity
General Use
𝑞𝑞 𝑞𝑞
𝐾𝐾𝑖𝑖 = 𝐶𝐶𝑖𝑖 /𝐶𝐶 𝑞𝑞1−𝑊𝑊𝑊𝑊
of the flow given from
trapolate the curve for this tracer, a𝑊𝑊𝑊𝑊 linear combination of the𝑀𝑀 = 𝑀𝑀coefficient,
Lorenz 𝑜𝑜 + 𝑀𝑀 𝑤𝑤 = (1
defined + as 𝐾𝐾the area) between
𝑀𝑀𝑤𝑤 the 𝐹𝐹 − Φ
𝑊𝑊𝑊𝑊
solutions given by Eqn. 9 is convenient. Tracer data, as well curve and the diagonal, normalized by one-half18,
as the analytical models used to extrapolate the curves, are 𝑞𝑞
𝐶𝐶𝑖𝑖
1 𝑞𝑞 𝑞𝑞
𝐾𝐾𝑖𝑖 = 𝐶𝐶𝑖𝑖 /𝐶𝐶 𝑞𝑞
𝐿𝐿𝑐𝑐 =Saudi
displayed 2in∫Fig.𝐹𝐹(Φ)𝑑𝑑Φ
5. For
Aramco: the−
Company 1General Usetracers, no (11)
partitioning bimodal 1
0
behavior is observed yet. An obvious explanation for this 𝐿𝐿𝑐𝑐 = 2 ∫0 𝐹𝐹(Φ)𝑑𝑑Φ − 1 (11) (11)
is that the partitioning tracers are delayed compared to the 𝑞𝑞
𝐶𝐶 𝑞𝑞
𝐶𝐶𝑖𝑖
ideal, and so the secondary peak has yet to appear. Results and the sweep volume are two such parameters. These are
Saudi Aramco: Company General Use
from the RTD analysis based on the data and the extrapola- also summarized in Table 4. It is interesting to note that the
tions are summarized in Table 4. heterogeneity of the flow 𝐾𝐾𝑜𝑜𝑞𝑞is=relatively
𝐶𝐶𝑜𝑜𝑞𝑞 /𝐶𝐶 𝑞𝑞 large, which is in line
𝐶𝐶 𝑞𝑞
The total produced mass (M) of a partitioning tracer is the with values typically found in carbonate reservoirs, such as
sum of the produced tracer mass in the water and oil phases. the Lagrave case reported𝑞𝑞 by Viig et al. (2013)13.
𝐾𝐾𝑤𝑤 = 𝐶𝐶𝑤𝑤𝑞𝑞 /𝐶𝐶 𝑞𝑞 = 1
MW and M0, respectively. The concentrations in the oil phase 𝐾𝐾𝑜𝑜𝑞𝑞 = 𝐶𝐶𝑜𝑜𝑞𝑞 /𝐶𝐶 𝑞𝑞
(c) (d)
(c) (d)
Fig. 5. Non-partitioning tracer, WT-60 (Fig. 5a) and partitioning tracers WTP-2, WTP-3 and WTP-4 (Figs. 5b, 5c and 5d, respectively). The measured data are represented by
Fig. 5. the
circles and Curves
fits used tofor the non-partitioning
extrapolate tracer,
the RTD to infinity are indicated WT-60
by dashed (Fig.
curves. The 5a)blue
combined and partitioning
curve tracers
represents the full RTD used to WTP-2, WTP-3
analyze the tracer data. and
WTP-4 (Figs. 5b, 5c and 5d, respectively). The measured data are represented by circles and the fits and
Fig. 5. Curves for the non-partitioning tracer, WT-60 (Fig. 5a) and partitioning tracers WTP-2, WTP-3
WTP-4
used
Estimation (Figs.
to extrapolate
of Oil5b,Saturation
5ctheand RTD5d, torespectively).
infinity are indicatedThe measured byfore data
dashed
not are
wellrepresented
verycurves. Theto combined
suited by circles
estimate blueand
saturations curve the Eqn.
using fits
used to extrapolate the RTD to
represents the full RTD used to analyze the tracer data.infinity are indicated by dashed curves. The combined blue
2. The best estimate for the saturations in our case is there- curve
represents
Saturations canthebe full RTD from
estimated usedEqn.to analyze
2 if the oilthe in atracer
sys- data. fore given from T of the RTD. The values for saturations
m
tem is mostly stagnant. In our case, with water cuts at about obtained from (𝑇𝑇̅) and K are 0.19, 0.22 and 0.21 from the
96%, this condition is satisfied. This estimation requires partitioning tracers WTP-2, WTP-3 and WTP-4, respectively.
knowing the times of the partitioning tracer, TR, and the These values are summarized in Table 5.
Tracer
ideal tracer, TW, in K
addition to m 0 values of m
the K.1Given that m22 ̅ 𝐹𝐹 − Φ
𝑻𝑻 Tm Lc Vp
Tracer K m0 m 1 5m ̅
𝑻𝑻 Tm Lc V5p 3
(days) (10
the bimodal shape of the ideal tracer is not yet seen for the 5 days 2
) (days)
DISCUSSION (days)
AND LESSONS LEARNED (10 m)
(days) (10 days 2
) (days) (days) (105 m3)
WT-60 tracers,—it is challenging
partitioning 0.45 to extrapolate
199 the parti- 1.2 445𝐾𝐾 𝑞𝑞 = 𝐶𝐶 𝑞𝑞 /𝐶𝐶221
𝑞𝑞 0.32 2.5
WT-60 — 0.45 ̅ 199 1.2 445𝑖𝑖 𝑖𝑖 221 0.32 2.5
WTP-2 2.1 0.27
tioning tracer curves. The values for (𝑇𝑇 ) 188
in Table 4 are there- 2.0 1. Careful execution of the tracer sampling protocol2.4
705 329 0.36 and
WTP-2 2.1 0.27 188 2.0 705 329 0.36 2.4
WTP-3 2.1 0.24 163 1.6 679 352 0.34 2.1
WTP-3 2.1 0.24 163m1 1.6
m2 679 T352 0.34 2.1
Vp
WTP-4
Tracer 4.0
K 0.24m0 219 3.0 923𝐶𝐶𝑖𝑖𝑞𝑞 448
m 0.35
Lc 2.8
WTP-4 4.0 0.24 𝐹𝐹 − Φ 219 (days) (1053.0
days2) 923
(days) 448
(days) 0.35 2.8
(m)
Table
WT-604. Summary
— of results from analysis
0.45 199 of the RTDs
1.2 of the 445
ideal tracer221 WT-60 and0.32
the partitioning
2.5
Table 4. Summary of results from analysis of the RTDs of the ideal 𝑞𝑞tracer WT-60 and the partitioning
tracers WTP-2, WTP-3 and WTP-4 𝐶𝐶
tracers
WTP-2 WTP-2, 2.1
WTP-3 and0.27WTP-4𝑞𝑞 𝑞𝑞 𝑞𝑞
𝐾𝐾𝑖𝑖 = 𝐶𝐶𝑖𝑖 /𝐶𝐶188 2.0 705 329 0.36 2.4
𝑞𝑞 𝑞𝑞𝑞𝑞
WTP-3 2.1 0.24 𝑞𝑞 163 1.6 679 𝐾𝐾𝑜𝑜 = 𝐶𝐶𝑜𝑜 /𝐶𝐶
352 0.34 2.1
𝐶𝐶𝑖𝑖
WTP-4 4.0 0.24 219 3.0 923 448 0.35 2.8
𝐾𝐾𝑤𝑤𝑞𝑞 = 𝐶𝐶𝑤𝑤𝑞𝑞 /𝐶𝐶 𝑞𝑞 = 1
𝐶𝐶 𝑞𝑞
Table 4. Summary of results from analysis of the RTD of the ideal tracer WT-60 and partitioning tracers WTP-2, WTP-3 and WTP-4
𝜑𝜑
26 WINTER 2016 𝑞𝑞 TECHNOLOGY
SAUDI ARAMCO JOURNAL𝐾𝐾OF = 𝐶𝐶 𝑞𝑞 /𝐶𝐶 𝑞𝑞
𝑜𝑜 𝑜𝑜
Tm So from Eqn. 2 So from Eqn. 2
Tracer K
(days) (days) and Tm and
Table 5. Saturation estimates from the partitioning tracers WTP-2, WTP-3 and WTP-4. Saturations based on average residence time (𝑇𝑇 ̅ ) as well as from the mode of the residence
time distribution (Tm ) are given. The latter are expected to be more reliable, as they are not based on extrapolation of the data.
proper communication between office staff and field per- a nearby well and sponge core 𝐹𝐹 −analyses
Φ performed in adja-
sonnel have been highly instrumental in eliminating sam- cent wells. Therefore, these new sets of PITT tracers are well
𝑞𝑞 𝑞𝑞
ple contamination and other headaches that could have suited as a tool to estimate oil
𝐾𝐾𝑖𝑖 saturations
= 𝐶𝐶𝑖𝑖 /𝐶𝐶 𝑞𝑞 in the interwell
jeopardized the tracer project. region and are adjudged qualified for field deployment.
2. The partitioning tracer test performance confirms that 𝑞𝑞
𝐶𝐶𝑖𝑖
the new set of partitioning tracers devised by Viig et al. ACKNOWLEDGMENTS
(2013)13 are well suited as a tool to estimate oil satura- 𝐶𝐶 𝑞𝑞
tions in the interwell region. The authors would like to thank the management of Saudi
3. The estimated saturations from the individual partitioning Aramco for their support and 𝑞𝑞permission to publish this ar-
𝐾𝐾𝑜𝑜 = 𝐶𝐶𝑜𝑜𝑞𝑞 /𝐶𝐶 𝑞𝑞
tracers are mutually consistent, with a variation of about ticle. Furthermore, this article would not have been possible
1 to 2 saturation units. The values are also consistent with without the contributions and𝑞𝑞efforts of many colleagues.
𝐾𝐾𝑤𝑤 = 𝐶𝐶𝑤𝑤𝑞𝑞 /𝐶𝐶 𝑞𝑞 = 1
estimations obtained in a SWCTT performed in an adja- In particular, we would like to acknowledge and thank
cent producer in the same field15, which reported an oil Ibrahim Zefzafy, Yousif Abdulmohsin, Fouad Sadis and
𝜑𝜑
saturation of 0.21 ± 0.02, further strengthening the conclu- Senthilmurugan Balasubramanian, our colleagues from Field
sions that these tracers are reliable for interwell estimations Services, for the periodic wellhead sampling, and many oth-
of saturation. 𝐷𝐷𝑖𝑖∗
ers for their contributions toward the project.
4. The saturation estimations based on suffer from insuffi- This article was presented at the SPE Annual Technical
cient data in the tail of the partitioning tracer curves. The Conference and Exhibition, Φ(𝑡𝑡)Dubai, UAE, September 26-28,
use of the Tm of the distribution, representing the apex of 2016.
the tracer curves, is a satisfactory replacement of the aver-
age for this project. REFERENCES
5. Two of the three partitioning tracers used in this field pi-
lot test are similar in their partitioning coefficients. That 1. Dugstad, Ø.: “Well-to-Well Tracer Tests,” Chapter 6
these two yield similar results in breakthrough times, and in Petroleum Engineering Handbook, Vol. 5, Reservoir
consequently saturation estimates, is satisfying. The third Engineering and Petrophysics, E.D. Holstein (ed.), SPE,
tracer, WTP-4, which has a larger partitioning coefficient, Richardson, Texas, 2007, pp. 651-683.
is more delayed than the two others but still yields a
2. Beier, R.A. and Sheely, C.Q.: “Tracer Surveys to Identify
similar result for saturation. Using tracers that differ in
Channels for Remedial Work Prior to CO2 Injection at
partitioning properties is an advantage, as it proves the
MCA Unit, New Mexico,” SPE paper 17371, presented
robustness of the estimate.
at the SPE Enhanced Oil Recovery Symposium, Tulsa,
Oklahoma, April 16-21, 1988.
CONCLUSIONS
3. Du, Y. and Guan, L.: “Interwell Tracer Tests: Lessons
The field pilot test for the new set of partitioning tracers has Learned from Past Field Studies,” SPE paper 93140, pre-
been a success. The estimated saturations from the individual sented at the SPE Asia Pacific Oil and Gas Conference and
tracers are mutually consistent, with a variation of about 1 Exhibition, Jakarta, Indonesia, April 5-7, 2005.
to 2 saturation units. In general, near wellbore saturations 4. Galdiga, C.U. and Greibrokk, T.: “Ultra-Trace
and interwell saturations may differ, as in the field case Determination of Fluorinated Aromatic Carboxylic Acids
presented here. The derived values are also consistent with in Aqueous Reservoir Fluids Using Solid-Phase Extraction
saturations obtained via other methods, e.g., a SWCTT from in Combination with Gas Chromatography–Mass
ABSTRACT system that can improve the speed, efficiency and ease of ESP
deployment.
Electric submersible pumps (ESPs) are a widely used artificial
lift technology. Conventional ESP systems provide power to BACKGROUND
the pumps through a cable banded to the outside of the tub-
A majority of the electric submersible pumps (ESPs) used
ing. These systems have drawbacks in terms of slow instal-
today are installed with jointed tubulars using workover rigs.
lation and limited efficiency. To overcome these obstacles,
The ESP is installed as part of the tubular string. The cable
a novel cable deployed ESP system developed for use in a
that provides power to the ESP is banded or clamped to the
high hydrogen sulfide (H2S) production environment offers a
tubular. This method has been in use since the first ESP in-
future solution. This article focuses on the material selection
stallations back in the 1930s.
process and the validation of the nontraditional National
Since the 1970s, several types of alternate deployment
Association of Corrosion Engineers (NACE) materials even-
have been developed. Systems using the alternate technolo-
tually selected.
gies include cable suspended submersible pump systems1; ca-
An evaluation process was developed and used early on to
ble internal, coiled tubing (CT) deployed submersible pump
narrow down the choices to several potential alternatives. A
systems2; and umbilical systems3. Much was learned from
metal-jacketed power cable was selected as the most viable
the use of these alternate deployment systems; however, each
solution since it provided the best protection against H2S
of these systems had drawbacks that prevented widespread
attack and it provided a smooth outside diameter that could
adoption of the technology. Major drawbacks of the current
be gripped onto and sealed. The cable criteria required that
technologies include safety concerns, problems finding ma-
it be able to withstand H2S levels up to 15% and chloride
terial able to endure corrosive environments, the uncertain
levels in excess of 150,000 ppm for a minimum of 5 years. A
reliability of downhole electrical wet connectors, and deploy-
major challenge was constructing the system so it supported
ment and retrieval concerns.
the weight of the cable inside the metal jacket. The cable was
Saudi Arabia’s experience with several of the systems led
qualified through extensive physical and chemical testing.
to a rethinking of ESP design and the development of a new
The interlocking of the metal jacket (tubing) with the in-
ESP system that could operate in a high hydrogen sulfide
ner core (cable) was qualified through extensive mechanical
(H2S) environment in the long term. A major challenge was
testing of the power cable, including its pull resistance, bend,
designing the compact, H2S resistant cable. The material
thermal growth and mechanical splice connections. Electrical
used had to be strong enough to support the weight of both
testing assessed insulation resistance. Chemical compatibility
the copper power cables and the ESP system, and also pro-
testing was performed on a number of candidate metallur-
vide resistance to high H2S levels. The cable itself had to be
gies using a modified NACE H2S stress cracking test based
a compact size to minimize pressure drop along the length of
upon ANSI/NACE Standard TM0177. Test coupons of the
cable. Finally, the technology had to be simple enough to be
various metallurgies proposed for the tubing were welded
reliably and economically manufactured in large quantities.
and formed for testing. The test coupons were bent to stress
levels in excess of the final design stresses. CABLE DESIGNS
The high chloride levels in the test conditions caused sev-
eral metallurgies to fail, even at low H2S levels. A single met- The initial plans were to brainstorm potential cable design
allurgy successfully passed all the mechanical, electrical and solutions. A review of several design possibilities identified
chemical testing. It will be used for field trials of the new some key technology gaps. As expected, the most critical as-
ESP technology. pect of the cable deployed ESP was the power/support cable.
A rigorous evaluation and test program resulted in the Finding a material for the cable able to withstand H2S con-
development of a novel high H2S rated, cable deployed ESP centrations up to 15% in the vapor phase presented a consid-
into strips. Each strip is about 300 ft in length. To achieve levels vary from 0% to 10%. Water cuts vary from 10% to
the production lengths required, they are welded together 50%.
Saudi Bottom-hole
Aramco: Company General Use
using a specialized in-line laser that produces welds in a bi- temperatures range up
ased pattern — a bias weld provides greater overall strength to 220 °F.
than a butt weld. During the process of welding, complete Separate metallurgy
full production lengths with the cable inside, the smaller coils solutions offer eco-
are welded together in an in-line welding process, as needed. nomical solutions for
After the bias welding, the weld is ground and polished to the varying H2S levels
the exact same thickness as the strip. The joined strips are in different fields. It
wound onto a large diameter spool, Fig. 1, once they are long was decided to identify
enough to produce a full production length in a single longi- three levels of H2S for
tudinal pass. To complete the manufacture of the metal-jack- metallurgy testing.
Fig. 2. Metal-jacketed cable seen in cross section.
eted cable, the flat strip and a completed ESP cable are run Fig. 2. Metal-jacketed cable seen in crossSamples section. would be
METALLURGY SELECTION
0.540”
1.750”
0.540”
1.080”
825
1.080”
Nickel Alloy
825 2205 4565
Nickel Alloy (Duplex Stainless) (Austenitic Stainless)
Fig. 4. Test coupons.
Fig. 4. Test coupons.
Fig. 4. Test coupons.
tubular product. Because the final tubular product had to be
provided in an “as welded” and “cold worked” condition,
stress corrosion cracking due to H2S attack on the unan-
nealed alloys was of concern. The corrosion testing would
evaluate the different alloy candidates to ensure proper grade
selection was made depending on the well environment.
The alloys selected for H2S testing were: UNS N08825 (825)
— nickel alloy, UNS S32205 (2205) — duplex stainless steel,
and UNS S34565 (4565) — austenitic stainless steel.
Sample coupons were prepared that simulated the actual
unannealed conditions in both longitudinal and transverse
welds. The samples were then bent in a way to induce stress
levels greater than actual field tube conditions. Setup fol-
lowed best practices, including special precautions to prevent Fig. 5. Micro-hardness test location.
Fig. 5. Micro-hardness test location.
Fig.
any 5. Micro-hardness
possible test location.
galvanic reaction. Figures 3a and 3b show the
with both longitudinal and bias welds. Three samples of
test coupon cut and configuration.
each alloy were placed into each autoclave test vessel to pro-
The test coupons were welded and bent around a man-
vide multiple test point confirmation.
drel to the first position with holes drilled in each leg for
After the full duration of the test, the samples were
inserting a bolt to provide additional stress, Fig. 4. Titanium
removed and a post-test evaluation was performed. Post-
bolts added to the test coupons were then tightened to a
testing consisted of a visual examination and a dye penetrant
calculated stress more severe than in situ stress levels. The
test, after which the samples were mounted for microscopic
titanium bolts were wrapped with Teflon tape at all contact
work. Hardness values before and after testing were com-
points with the test coupons to prevent galvanic corrosion
pared along with the microstructures of the base and weld.
between the bolts and coupons. Pretest micro-hardness
Saudi Aramco:val-
Company General Use Saudi Aramco: Company General Use
ues were taken in the base metal and in the weld section, H2S Test Results
and this was repeated at three elevations across the coupon
thickness, Fig. 5. The lab report noted, “Cracking has been observed for all
Four different autoclaves — test vessels — were set up for 2205 U-bend specimens by the visual examination.” All nine
testing the coupons. Three autoclaves, each with a different samples of that alloy (three samples each from the 1%, 5%
level of H2S at 1%, 5% and 15%, were used for 30-day tests and 15% H2S tests) were returned for further examination.
of the 2205 and 4565 alloys. The fourth autoclave had 15% Evaluations indicated that the cracks were in seam weld
H2S and was planned for a 90-day test of the 825 alloy. areas; however, there was no sign of defective welds upon
The remaining test parameters were the same for all four micro-examination. There were no significant differences in
autoclaves: 150,000 mg/l chlorine, 5% CO2, a total pressure crack frequency and crack depth among the autoclave sam-
of 300 psi and a temperature of 275 °F (135 °C). Because ple lots, Figs. 6a to 6d.
the 825 alloy had the highest confidence of passing all the The lab report noted similar results for the other alloy:
testing, it was put in a 90-day test. The 825 test samples “Cracking has been observed for all 4565 U-bend specimens
consisted of U-bend specimens and the actual welded tubing by the visual examination.” All nine samples of that alloy
Unbent
Bent
Fig.
Fig. 6a.Sample
6a. Sample of
of2205
2205 alloy,
alloy, as
as prepared
preparedfor
fortesting.
testing. Fig.6b.
Fig. 6b.Sample
Sample of
of 2205
2205alloy
alloyafter
after1%
1%HH2S, 30-day test.
2S, 30-day test.
LOAD TESTING
Purpose
Fig.7a.
Fig. 7a.Sample
Sample of
of 4565
4565 alloy,
alloy, as
as prepared
preparedfor
fortesting.
testing. Fig.7a.
Fig. 7b.Sample
Sample of
of 4565
4565 alloy,
alloy after 1% H2S,for
as prepared 30-day test.
testing.
Saudi
Fig.
Fig. 8a.8a. Sampleof
Sample of825
825 alloy,
alloy, as
as prepared
preparedfor
fortesting. Fig. 8b. Sample of 825 alloy after 15% H2S, 90-day
testing.Fig. Fig. 8c. Bias weld sample of 825 alloy after 15% H2S,
8b. Sample of 825 alloy after 15% H2S, 90-day test.Bias weld sample of 825 alloy after 15%
Fig. 8c. H2
test. 90-day test.
Saudi Aramco:
No metallurgical abnormalities were associated with any
of the pinholes. Additional external and internal fatigue
cracks were observed, which is normal from a fatigue test,
e.g., once the maximum fatigue life is reached, fatigue cracks
start at different locations.
Fig. 9b. Forming rolls shape the jacket strip around the cable.
Fig. 9b. Forming rolls shape the jacket strip around the cable.
pinholes started at the seam weld.
The outside diameter measured during the fatigue test was
close to the nominal value. The ovality and ballooning de-
Fig. 9b. Forming rolls shape the jacket strip around the cable.
tected were very low — ovality of 0.1% to 0.4%; ballooning
of 0.001” to 0.009”. It is noteworthy that these results indi-
cate the dimensional changes due to low-cycle fatigue for the
Incoloy 825 tube, under the testing conditions, were below
the maximum acceptable for normal CT operations.
Some displacement of the solid conductors and their in-
sulation was observed at the sample’s ends after the bending
fatigue test. This could be the result of the movement of the
internals while the cable was being bent during the test. Fig. 10. Load testing of cable.
Fig. 10. Load testing of cable.
(a)
Fig. 10. Load testing of cable.
(b)
Saudi Aramco: Company General Use
Fig. 10. Load testing of cable.
Fig. 11. Cable samples testing using the bending fatigue machine. Both samples were fatigue tested on a 36” radius bending form and with 500 psi internal pressure. (a) Test
1: seam weld on the intrados; (b) Test 2: seam weld on neutrados.
using
Saudi Aramco: Company General Use
SAUDI ARAMCO JOURNAL OF TECHNOLOGY WINTER 2016 37 ;
Fig. 12. Transverse crack.
Fig. 12. Transverse crack.
Fig. 12. Transverse crack.
SHEAR TESTING
Purpose
ACKNOWLEDGMENTS
REFERENCES
Bander F. Al-Daajani, Dr. Nayef M. Al-Anazi, Dr. Abdelmounam M. El-Sherik and Abdullatif I. AbdulHadi
Fielda target
water
Table 1. +Dataactual
Visualfield
conditions forsludge,
each
showedspool
2
of 112 pins was then installed, covering location inspection
3.5
minor metal loss, Fig. 3. It was
stagnant water
from the 4 o’clock to 8 o’clock positions of each spool. All noticed in the photo that corrosion products were deposited
test spools were connected in a series with one inlet and onto the pipe wall where they formed a hard film. The EFM
10% HCl acid in distilled water, water
3
outlet for gas mixture purging purposes.
1
Figure 1 shows the experimental setup
for the test spools.
circulated at 15 l/min
The spools were carefully cleaned by
Spool-3
Fig. 3. showed
results Photo showing a cross
no detectable metalsection for4.spool-1.
loss, Fig. In contrast,
the UT technique detected a metal loss of 0.06 mm, which is
equivalent to a corrosion rate of 0.122 mm per year (mmpy).
Fig. 5. Average metal loss in spool-1 detected by UT and EFM.
For comparison purposes between the EFM and UT
readings, six locations were selected on the 6 o’clock Fig.5.5.Average
Fig. Averagemetal
metalloss
lossinin spool-1
spool-1 detected
detected byby
UTUT
andand EFM.
EFM.
Spool-2
The corrosivity of condition 2 was more aggressive than that Fig. 6. Photo showing a cross-section for spool-2.
Fig. 4.
Fig. 4. Remaining
Remaining wall
wall thickness
thicknessfor
forspool-1
spool-1asas
detected by EFM.
detected by EFM. Fig. 7.
Fig. 7. EFM
EFM results
results indicating
indicatingno
nometal
metalloss forfor
loss spool-2.
spool-2.
for spool-3 after the experiment. In this part of the test, the
Saudi Aramco: Company General Use
EFM demonstrated the ability to pick up the changes in metal
thickness, confirming that the EFM system does respond to
the reduction in metal thickness, but only when the reduction
in pipe metal thickness is higher than 0.2 mm.
Figure 11 exhibits the results of metal loss detection in
six locations as measured by EFM and UT. The UT readings
are higher than the EFM readings in locations 2, 3 and 4 by
30% to 50%, while in locations 1, 5 and 6, EFM readings
Fig. 9. Spool-3 photo after the experiment with UT locations marked.
are a little bit higher than UT readings by almost 7% to
Fig. 9.
Fig. 9. Spool-3
Spool-3 photo after
photo the experiment
after with UT
the experiment locations
with marked. marked.
UT locations 20%.
CONCLUSIONS
ACKNOWLEDGMENTS
Dr. Tao Chen, Dr. Qiwei Wang, Dr. Fauken F. Chang and Dr. Yahya T. Al-Janabi
Fig. 1. Scale formed in a Khuff sour gas well, including (left) scale deposited in a downhole tubular, and (right) scale deposited in a topside flow line.
Fig. 1. Scale formed in a Khuff sour gas well, including (left) scale deposited in a downhole tubular, and
(right) scale deposited in a topside flow line. SAUDI ARAMCO JOURNAL OF TECHNOLOGY WINTER 2016 47
(H2S), with an average of ~5% H2S, and 0.5% to 4.0% and casing, along with the generation of H2S during descal-
carbon dioxide (CO2). The produced water is a mixture of ing jobs, raised major concerns and barriers to their further
vapor condensate and formation water, with total dissolved use. Alternative dissolvers, however, usually have limited
solids up to 300,000 mg/l. The average water production dissolving powers.
rate is very low, often in the range of 2 bbl per million This article presents the latest laboratory results from an
standard cubic feet per day (MMscfd) of gas. The produced evaluation of new iron sulfide dissolvers. The recently devel-
hydrocarbon condensate rate is between 20 bbl per MMscfd oped dissolvers were studied for their dissolving effectiveness
and 300 bbl per MMscfd. Low alloy carbon steel T-95, against iron sulfide minerals, and for the rate of their corro-
which contains 0.5% to 1.0% Cr, 0.2% Ni and 0.2% Mo, sion of mild steel at elevated temperatures and high pressure.
has been used for well completions in the Khuff as a cost-ef- Future developments for iron sulfide dissolvers are discussed.
fective option3. No subsurface chemical treatment program
has been implemented to date for controlling corrosion and EXPERIMENTAL PROCEDURE
scale.
Scale deposition has been one of the major issues for the Scale Solids
Khuff sour gas wells in the last few decades. Scale that is
deposited in downhole tubulars and topside pipelines causes Most iron sulfide scales deposited in Khuff wells are in
significant production and operational problems, such as the form of iron mono-sulfide phases, such as pyrrhotite
restriction of downhole surveillance and intervention, and (Fe1-xS), troilite (FeS) and mackinawite (Fe1+xS), Fig. 35. The
leads to significant losses of gas production4. Figure 1 shows iron-deficient Fe1-xS is the most abundant phase. On average,
the typical scale deposited in the downhole tubulars and top- it accounts for more than half of iron sulfide scales (58%),
side pipelines of Khuff wells1, 5. followed by FeS (13%) and Fe1+xS (11%). Pyrite and marca-
Scale formation differs significantly from well to well in site, which have the same chemical composition (FeS2), have
terms of severity and chemical composition; it also varies a combined average scale contribution of 15%. They occur
with depth in the same well . Most deposits are a mixture of
5 in less than half of the analyzed samples. Greigite is the least
different types of minerals. Figure 2 is a graph of the typical common iron sulfide phase. It is found only in a few wells
scale compositions in the Khuff sour wells. As iron sulfide and accounts for only about 3% of the iron sulfide scale on
scale is the predominant mineral found in most scale samples average. As Fe1-xS is the predominant iron sulfide form found
a Khuff
analyzed, sour
it is the maingas well,
focus including
in this study. (left) scale deposited inscale,
at field a downhole tubular,
it was used andthe dissolving perfor-
to measure
deposited
Bothinchemical
a topside andflow line. methods have been used to
mechanical mance of the iron sulfide scale dissolvers in this screening
remove downhole iron sulfide scale in these wells6-8. Hydro- study.
chloric (HCl) acid based scale dissolvers were used in several The Fe1-xS samples used in this study were prepared from
descaling jobs. Although they were often effective in remov- commercial Fe1-xS supplied by VWR. Fe1-xS chips were pul-
ing deposits9, their heavy corrosion of the production string verized, and powders from 0.85 mm to 2.0 mm in size were
Fig. 4. Microscopic image of Fe1-xS particles used in the study. Table 1. Basic chemistry and pH of candidate scale dissolvers
Fig. 4. Microscopic image of Fe1-xS particles used in the study.
collected, Fig. 4. The samples were washed repeatedly by For the dissolving tests, 30 ml of dissolver and 3 grams of
de-ionized water to remove fine particles, and then dried in pulverized Fe1-xS were used. The tests were carried out at
open air before use. 125 °C in an autoclave. The total test duration was 6 hours,
including 1 hour of heating in the autoclave from room
Dissolver Solutions temperature to 125 °C, 4 hours at the test temperature, and
Saudi Aramco: Company General Use
1 hour of cooling, depressurizing and disassembling. Tests
The ideal iron sulfide scale dissolver should not only re- were also performed using test tubes without external ap-
Code pH Chemistry
move scale deposits in downhole tubulars to restore well plied pressure. At the SD-A
end of the tests conducted at 1,000 psi,
<0 15% HCl acid
accessibility and productivity, but also cause no damage to the spent scale dissolvers
SD-Bwere collected
<0 for further
Strongcorrosion
acid
well integrity and pose no safety concerns. Considering the tests. SD-C 2-5 THPS
aggressive nature of HCl acid, non-HCl acid based dissolv- SD-D performance,
To calculate dissolving 2-5 the solidsTHPSwere
ers are desired to alleviate damage and safety concerns. The filtrated out and rinsed with de-ionized water Acidic
SD-E 2-5 chelate
to remove re-
SD-F 2-5 THPS
ideal iron sulfide scale dissolver should have the following sidual dissolver. The collected solids were then dried at 80
SD-G 2-5 THPS
characteristics: °C overnight. The amount
SD-H of dissolved
5-7 Fe1-xpH
S neutralized
was calculated
chelate
from the weight difference
SD-I between5-7the initial and finalchelate
pH neutralized sol-
• A dissolving capacity comparable to that of concen- ids, which were measuredSD-J with an>12
analytical Alkaline
balance.dissolver
trated HCl acid. SD-K >12 Alkaline dissolver
Corrosion Tests
Fig.7.7.Effect
Effectofofpressure
pressure on
on Fe
Fe1-x S dissolution at 125 °C.
1-x S dissolution at 125 °C.
Fig. 8. Corrosion
Fig. 8. Corrosion rate
rateatat125
125°C
°Cafter
after44hours
hoursininlive
liveand
and spent
spent dissolver
dissolver solutions.
solutions.
Fig.
50 7. Effect
Fig. WINTER 2016 SAUDI ARAMCO JOURNAL OF
of pressure on Fe1-xS dissolution atTECHNOLOGY
125 °C.
Fig. 10. pH of live dissolvers vs. corrosion rate at 125 °C over 4 hours.
solvers showed less corrosion, with metal loss of < 0.05 lb/
ft2, especially SD-I, SD-J and SD-K, which had a corrosion
rate less than 0.01 lb/ft2.
Compared to the corrosion caused by the live dissolv-
ers, the solutions of spent dissolvers in most cases showed
Blank SD-A SD-B
a reduced corrosivity against the test carbon steel coupon.
Their ability to corrode carbon steel, therefore, is partially
consumed by the dissolution reaction with Fe1-xS. For some
high pH dissolvers — SD-E, SD-H and SD-I — no noticeable
change in corrosivity was observed. SD-C SD-D SD-E
Fig. 9.
Fig. 9. pH
pH of
of live
livedissolvers
dissolversand spent
and dissolvers.
spent dissolvers. Saudi Aramco: Company General Use
Fig. 12.
Fig. 12. Corrosion
Corrosion weight
weight loss
loss at
at 125
125 °C
°C and
and150
150°C
°Cafter
after4-hour
4-hourimmersion
immersionin
Fig. 10. pH of live dissolvers vs. corrosion rate at 125 °C over 4 hours. live dissolvers.
Fig. 10. pH of live dissolvers vs. corrosion rate at 125 °C over 4 hours.
late in the system. That meant the D50 was decreasing from DIF. In addition, a reserve volume of 600 bbl to 800 bbl
the initial 35 µm to 27 µm, and the mud weight was creep- invert emulsion with a density of 75 lbf/ft3 was available
ing toward 68 lbf/ft3, affecting equivalent circulating density throughout the entire well drilling and completion process.
(ECD). To control this situation, 25 µm and 50 µm ground This was prepared as requested in the drilling program and
marble was added to the system and centrifuges were run. As maintained in good condition for use as a mud cap in case
Fig. 2. PSA used. of lost circulation and/or to control gas kicks. Based on the
soon as the mud weight reached 68 lbf/ft3, mud weight and
properties were held as desired. Another action included dis- information provided, the drilling fluid team established the
continuing use of ground marble with a D50 of 5 microns. following objectives as focal points for the DIF preparation
and maintenance:
• When the mud plant arrived at the location, the tanks FIELD CHEMICAL CONCENTRATION
were spotted as per the planned area, and the mud COMPARISON FOR DIF
plant was rigged up and hydrotested for leaks.
Tables 3 and 4 show the field mud formulation and field
• The mud plant was connected to the operator diesel
fluid properties, respectively, for the DIF.
tank, filled up with diesel and started to mix the DIF.
As shown in Table 3, extra lime was used because of a
• The 1,400 bbl of DIF was mixed in the mobile mud depletion of the alkalinity while drilling through the gas cap.
tank while drilling the 12.25” hole; it was sheared Actual and planned concentrations were nearly matched
daily in the mud plant to optimize the DIF properties. with the proposed formulation. All the properties shown in
• The first action taken before transferring the DIF to Table 3 were in range and constant, which indicated there
the rig tanks was to clean and flush all the pits and
Planned Actual
lines, and then measure all the tanks. Product Name
Concentration Concentration
• The initial rheology was lower than specification to Primary Emulsifier
12 14
prevent surface losses on the solids control equipment; (ppb)
then the YP was increased to 24 lb/100 ft2. Secondary Emulsifier
4 6
(ppb)
Maintenance Treatments Lime — Alkalinity
4 8
Source (ppb)
• An 8½” horizontal hole was drilled from 9,689 ft MD Fluid Loss Additives
to 12,749 ft MD with 67 lbf/ft3 invert emulsion DIF 6 5
(ppb)
without problems. The active system was treated with Viscosifier (ppb) 4 2.3
lime and invert emulsifier. Bridging Agent —
• On an hourly basis, 25 µm ground marble and 50 µm 5 µm Ground Marble 8 5.8
(ppb)
ground marble were added. The rate of addition de-
pended on the porosity trend. Bridging Agent —
25 µm Ground Marble 24 25.1
• Shale shaker screen sizes were increased from 140 (ppb)
mesh to 180 mesh to discard fine sands and solids. Bridging Agent —
50 µm Ground Marble 8 5.33
• A centrifuge was run 24 hours/day to prevent buildup
(ppb)
of low gravity content.
CaCl2 — 77% (ppb) 66 55
• Seepage losses of 3 bbl/hr to 5 bbl/hr were observed
Table 3. Field mud formulation
while drilling in porous formations; therefore, the 50
were no mud invasions, emulsion blockage or hole cleaning lateral section. The hydraulics software simulation always
problems. showed good hole cleaning with the rheology and ROP as
The PSD and PPA were run at different depths to deter- used on the drilling operation. Also, no fill and/or excessive
mine particle size so as to evaluate the condition of the DIF drag were reported on the final trip going out and in, which
and the efficiency of the solids control equipment. Figure 3 indicated the hole was clean.
shows PSD and porosity vs. depth for the DIF throughout It is recommended to not stop the pumps while a sweep is
the drilling operation. in the hole and to make short trips every 500 ft to 800 ft, or
less if needed. Pipe rotation is very important in hole clean-
HOLE CLEANING ing, and backreaming is not recommended. The mud engi-
neer should track each sweep and report on its effectiveness.
Hole cleaning is a major concern in any well. Good hole As shown in Fig. 4, at 12,489 ft, the maximum cuttings
cleaning will reduce problems, such as torque and drag on load in the annular is 0.52%, and the maximum value stan-
short trips and hole pack off. The rate of penetration (ROP) dard is less than 3%. This value represents a good cleaning
should be monitored and controlled to optimize the cuttings practice for the hole. The maximum overbalance in 02 sand
load percentages, which in turn will alleviate any issues con- was 700 psi in dynamic condition.
cerning surface equipment, annular loading and excessive
whole mud losses. Hole cleaning indicators, such as torque SOLIDS CONTROL EVALUATION
and drag and pressure-while-drilling data, should be moni-
tored. It is recommended to pump high density sweeps — 40 Before displacing, shakers were dressed with 140 mesh to
bbl to 50 bbl — for every other stand, 180 ft, or as needed 180 mesh screens to avoid blinding from unsheared mud ag-
to ensure good hole cleaning. These sweeps should be pre- glomerated with ground marble. After two full cycles and no
pared in the slug pits using mud from the active system. In losses on the surface, shakers were progressively screened up
this case, there were no hole cleaning issues through the to 200 mesh and remained as such throughout the section.
The absence of fish eyes indicated cor-
rect mud mixing. The flow was evenly
distributed although most of the time
shakers were running wet. It was
preferred to lose a bit with the finest
screens than incorporate solids with
the coarse screens.
After displacing the hole to the sol-
ids-free OBM, the shale shaker screens
were changed to 325 mesh to avoid
plugging of the 6⅝” premium mesh
sand screens used with mechanical
open hole packers and a dual perma-
nent downhole monitoring system.
The PSD analyses conducted on
Fig. 3. PSD, PPT and porosity vs. depth for the DIF.
cuttings and shaker underflow showed
Fig. 3. PSD, PPT and porosity vs. depth for the DIF.
CONCLUSIONS AND RECOMMENDATIONS 8. This article recommends building a fixed mud plant in a
remote area to assist in supplying, storing and recycling
1. A reservoir section of Well N-1 was drilled from 9,689 OBM from and to rigs. This will be reflected in the over-
ft to 12,749 ft using 70/30 invert emulsion 67 lbf/ft3 all cost and performance of the field project.
OBM. The previous casing of 9⅝” was set on top of the 9. A total of 26 wells were drilled successfully based on a
reservoir sandstone; a horizontal section was drilled; and similar DIF design.
the well was completed with 6⅝” premium mesh sand
screens. Close monitoring of mud properties and PSD
ACKNOWLEDGMENTS
while drilling was conducted, which led to drilling the
The authors would like to thank the management of Saudi
hole without any hole problems.
Aramco and Halliburton for their support and permission to
2. Hole cleaning was simulated using the hydraulics soft- publish this article.
ware program, which showed good hole cleaning with This article was presented at the SPE Kingdom of Saudi
minimum cuttings load in the annulus. Arabia Annual Technical Symposium and Exhibition,
3. PSD was conducted periodically every 2 to 3 hours, which Dammam, Saudi Arabia, April 25-28, 2016.
Dr. Abderrazak Traidia, Kaamil M. Shibly, Abdulrahman S. Al-Qahtani and Mohammed A. Abufour
ABSTRACT HIC has been a major concern for the oil, gas and pet-
rochemical industry since the 1960s, in particular for oil
This article discusses the development and capabilities of an and gas companies operating in sour hydrocarbon regions.
engineering software created by Saudi Aramco for the pre- Following a series of pipeline ruptures at Aramco in 1974
diction of linear hydrogen-induced cracking (HIC) growth (three ruptures occurred on a spirally welded API-5L-X42
rate in pipelines and plant vessel steels. Using a mechanistic sour gas transmission pipeline within 7 weeks of its com-
HIC model, a software prototype, named ProHICTM, was missioning), the company developed stringent material
developed and laboratory validated. This first prototype is specifications1 for sour service assets based on the work
being deployed for testing in practical field cases with a view developed earlier by British Petroleum and Sumitomo in
to improving the prediction accuracy. 1972. This work contributed to the development of the
The completion of this first version demonstrates that a NACE-TM-0284 standard test method2 (final version pub-
mechanistic modeling approach for HIC prediction is feasi- lished in 2011). This standard greatly influenced the steel
ble and can provide conservative crack growth rate estimates manufacturing industry and led to the development of HIC
as long as a strategy is in place for collecting critical model resistant steels, with greater steel cleanliness and more tightly
input parameters. This work also emphasizes the need to controlled microstructure. Today, this HIC qualification
include in future developments a means to determine the standard, in conjunction with the test acceptance criteria
coalescence of in-plane fracture toughness in a hydrogen given in NACE-MR-01753, is widely accepted and used for
environment for common steel grades used for pipeline and all new sour service steel procurements. To the authors’
plant vessels. best knowledge, there has been no documented case of HIC
caused failures of equipment (pipeline or plant vessel) manu-
INTRODUCTION factured from HIC resistant steel.
For non-HIC resistant steel equipment (also called dirty
The internal corrosion of carbon steels used in the oil and steels), most oil companies decided to keep operating this
gas and petrochemical industry produces atomic hydrogen as equipment while inspecting it more frequently to check the
a result of the cathodic corrosion reaction. The presence of progress of HIC damage. A Fitness-for-Service (FFS) assess-
hydrogen promoters, such as hydrogen sulfide (H2S), causes ment is also carried out on a regular basis, and mitigation
a significant part of the produced hydrogen to penetrate the and repair measures, such as internal coating (for vessels) or
bulk steel through an adsorption-absorption-diffusion mech- sleeving (for pipelines), are used as needed. In addition to
anism, while the balance combines into molecular hydrogen dealing with aging legacy equipment, companies are faced
and is released in the form of bubbles in the transported hy- with the increasing sourness of produced oil and gas world-
drocarbons. At the interface of existing macro-defects in the wide, which is raising a real concern about the ability of
steel, such as laminations or elongated nonmetallic inclusions non-sour service equipment — i.e., equipment originally de-
— typically manganese sulfides — the diffusing hydrogen signed to operate in non-H2S containing environments — to
is trapped and recombines to form high-pressure hydrogen handle a mildly sour environment without developing severe
gas. With time, the pressure buildup within these defects — HIC damage. For these reasons, HIC is reemerging as a topic
maintained by continuous charging flux from the corroding of interest to several oil and gas companies4-10.
wall — will trigger small cracks parallel to the pipe wall, Currently, one of the major challenges is the prediction
commonly called hydrogen-induced cracks (HICs). These of the remaining lifetime of HIC affected equipment, which
will grow, coalesce, and depending on the steel quality, may is critical for integrity management. The international FFS
lead to an advanced form of HIC called step-wise cracking standard API-57911, which is often used as a guideline for
(SWC). SWC, which progresses through the pipe thickness, FFS assessment of HIC affected equipment in the refinery
will eventually result in equipment failure. and petrochemical industries, clearly states that the remain-
zminimize
𝑛𝑛𝐻𝐻22 3
First16(1−𝜈𝜈
compressibility 0, if 𝜑𝜑1.54×10
< 0 -6 K.Pa-1 and D7, were selected from the described tests to validate
1 computational complexity.
)𝑎𝑎 A time-dependent
𝑉𝑉(𝑎𝑎, 𝑝𝑝𝐻𝐻2 ) ≈ constant 𝑝𝑝𝐻𝐻2 (5) the current model accuracy. Table 1 shows the value of the
analysis with an implicit3𝐸𝐸 time integration scheme was used
z2 Second compressibility 4.69×10-11 Pa-1 model parameters used for the simulation. Due to the uncer-
in coupling the structural mechanics module with both the Saudi Aramco: Company General Use
𝑛𝑛𝐻𝐻2 constant
diffusion-reaction module modulus
and the distributed ODE module. tainty and absence of published data on fracture toughness
E Young’s Ω∖Γ 200 GPa
νCoupling variables Poisson’s
were usedratio to evaluate numerically the 0.3 in the segregation zone of CMn steels — preferable initiation
THIC cavity volume, Temperature
which wasΩrevealed ⊂ 𝑅𝑅 𝑛𝑛 to be very 293
close sites for HIC — KIH was used as a fitting parameter.
toK General
Saudi Aramco: Company Use
the analytical approximation given in Eqn. 5, as well as the Figure 3 compares the experimental and calculated crack
Table 1. Material parameters of X60 pipeline steel in model
flux of hydrogen through the Γ . growth curves for both defects, D5 and D7. This figure
A review of the published literature showed a relevant shows that the predicted crack growth rates are in quite
𝜎𝜎
case study, which has been usedℎ to test the capabilities of good agreement with the experimental measurements. The
the proposed HIC model. Brouwer et al. (1995)19 conducted comparison also shows that the model tends to overestimate
𝜎𝜎ℎ = 𝜎𝜎𝑖𝑖𝑖𝑖 /3
Experiment19 𝜀𝜀 Experiment19
Simulation: KIH = 345 N/mm 3/2
Simulation: KIH = 345 N/mm3/2
𝜆𝜆, 𝜇𝜇
Cracked Area (mm2)
Crack Area (mm2)
𝑅𝑅(𝑝𝑝𝐻𝐻2 )
Defect D5
𝑅𝑅(𝑝𝑝𝐻𝐻2 ) =Saudi (𝑧𝑧1 /𝑅𝑅 Company
1 +Aramco: + 𝑧𝑧2 )𝑝𝑝General Use Defect D7
𝐻𝐻2
Saudi Aramco: Company General Use
16(1−𝜈𝜈 2 )𝑎𝑎3
𝑉𝑉(𝑎𝑎, 𝑝𝑝𝐻𝐻Saudi
2
) ≈Aramco:
Saudi Aramco: Company
3𝐸𝐸Company
𝑝𝑝 2GeneralUse
(a)𝐻𝐻General
Use (5) (b)
𝑛𝑛𝐻𝐻2
Time (hours) Time (hours)
Fig. 3. Comparison of calculated crack growth curves for defect D5 (a) and defect D7 (b) withSaudi
experimental
Aramco: crack growth
Company curvesUse
General
19
.
the HIC growth rate at longer exposure times. This may be even with low computing resources.
Abdulrahman S. Al-Qahtani is an
Internal Corrosion Engineer in the
Pipelines Technical Support Division
of Saudi Aramco’s Pipelines,
Distribution and Terminals Projects
and Technical Support Department.
Prior to joining Saudi Aramco in
2009, he worked as a Process Engineer in one of SABIC’s
petrochemical affiliates. After joining Saudi Aramco,
Abdulrahman worked in the field of in-line inspection
technologies, pipelines maintenance and pipelines operation
engineering. Currently, he is working as the Team Leader
of the Pipelines Internal Corrosion Group under the
Pipelines Specialists Unit, providing technical support to
the Pipelines Department on new projects and routine
maintenance and operations.
Abdulrahman had an internship assignment working
for 1 year with DNV GL, in Ohio, in the field of pipelines
internal corrosion.
He received his B.S. degree in Applied Chemical
Engineering from King Fahd University of Petroleum and
Minerals (KFUPM), Dhahran, Saudi Arabia.
To begin receiving the Saudi Aramco Journal of Technology at no charge, please complete this form.
Name _____________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
Title __________________________________________________________________________________________________________________________________________
Organization _______________________________________________________________________________________________________________________________
Address _____________________________________________________________________________________________________________________________________
City __________________________________________________________________________________________________________________________________________
State/Province ______________________________________________________________________________________________________________________________
Country _____________________________________________________________________________________________________________________________________
TO ORDER
By phone/email:
By mail:
Current issues, select back issues and multiple copies of some issues are available upon request.
The Saudi Aramco Journal of Technology is published by the Saudi Aramco Public Relations Department,
Saudi Arabian Oil Company, Dhahran, Saudi Arabia.
A New Approach of Pressure Profile and Oil Recovery during Dual and Single Coreflooding of Seawater and CO2
Injection Process for a Carbonate Reservoir
Xianmin Zhou, Fawaz M. AlOtaibi, Dr. Sunil L. Kokal and AlMohannad A. Al-Hashboul
ABSTRACT
Sweep efficiency during waterflooding and carbon dioxide (CO2) miscible injection can be challenging because of channeling
and bypassing of injected fluids. Some of the factors that contribute to this include reservoir heterogeneity, permeability contrast
and gravity override. All can lead to reduced volumetric sweep efficiency in both sandstone and carbonate reservoirs. To study
the effect of reservoir heterogeneity on pressure profile and oil recovery, and accurately describe the displacement mechanisms
during seawater and CO2 flooding, an effective experimental methodology — including the laboratory setup and procedures —
is proposed in this article.
Acoustics Deep Shear Wave Imaging (DSWI) Analysis in Fractured Carbonate Reservoirs
Mohamed L. Zeghlache, Andre S. Silva, Mohamed Gouda and Ahmed H. Abouzaid
ABSTRACT
The evaluation of carbonate reservoirs is challenging for different reasons related to stratigraphic and structural features’ hetero-
geneities. One of the major challenges is the evaluation of a fracture’s nature, intensity, aperture and extension. Near and far
field evaluation techniques have been utilized to achieve advanced fracture evaluation and modeling, the latter being a key factor
in successful carbonate reservoir development.