58 Oilfield Review Virtual Testing: The Key to a Stimulating Process Syed Ali ChevronTexaco Houston, Texas, USA Wayne W. Frenier Bruno Lecerf Murtaza Ziauddin Sugar Land, Texas Hans Kristian Kotlar Statoil Trondheim, Norway Hisham A. Nasr-El-Din Saudi Aramco Research and Development Dhahran, Saudi Arabia Olav Vikane Statoil Stavanger, Norway For help in preparation of this article, thanks to Ernie Brown, Steve Davies and Vincent Dury, Sugar Land, Texas; Keng Seng Chan and Ray Tibbles, Kuala Lumpur, Malaysia; Matt Gillard and Richard Warren, Aberdeen, Scotland; Abigail Matteson, Ridgefield, Connecticut, USA; and Carlos Torres, Maturin, Venezuela. ClayACID, ELANPlus (Elemental Log Analysis), NODAL, PLT (Production Logging Tool) and Virtual Lab are marks of Schlumberger. Matrix acid stimulation in a sandstone reservoir involves complex chemical reactions that are strongly dependent on mineralogy. A new process includes a model for simu- lating acid reactions to help operators choose an optimal treatment for a formation. Matrix acidizing reestablishes productivity in many damaged formations in a cost-effective way. The damage can be natural, an artifact of produced reservoir fluids moving through a formation, or induced by fluids used in well operations, such as drilling, completions and workovers, or stimulation. Formation damage can be caused by fines migration, scale formation, deposition of paraf- fins, asphaltenes or other organic material, or mixed organic and inorganic deposition. It can also result from plugging by foreign particles in injected fluids, wettability changes, clay swelling, emulsions, precipitates or sludges caused by acid reactions, bacterial activity and water blocks. 1 These damage mechanisms can be either natural or induced. Wellbore cleanup, matrix stimulation treatments or acid fracturing may be used to remove or bypass near-well damage. This article focuses on sandstone matrix acidizing. 2 A matrix acid treatment forces acid into a formation at a pressure below its fractur- ing pressure. The treatment often involves several stages that may be repeated. A new matrix-treatment design process includes use of the Virtual Lab software, a state- of-the-art geochemical simulator that properly accounts for secondary and tertiary reaction mechanisms. Field examples from the North Sea and from the Gulf of Thailand demonstrate the usefulness of this new process for evaluating treatment designs. Reacting to Damage Acid treatments for sandstones differ signifi- cantly from those for carbonate rocks. Carbonate rocks dissolve rapidly in hydrochloric acid [HCl], and the reaction products are solu- ble in water. In carbonate rocks, a matrix acid job is usually designed to bypass near-well damage by dissolving minerals and creating channels, or wormholes, in the rock, providing a flow path past the near-well damage. Acid- fracturing techniques in carbonates create a hydraulic fracture that has a differentially etched surface, so that the fracture maintains its conductivity during production. 3 In contrast to acidizing reactions in carbon- ate rocks, the reaction chemistry for silicate rocks is quite complex. Sandstones comprise quartz grains, clays of various types, feldspars, chert, micas, and carbonate materials as cement or overgrowths on grains, along with other minerals (next page). HCl is not effective in dis- solving most constituents of silicate rocks. Sandstone acid jobs typically use hydrofluoric acid [HF] in combination with HCl, formic or acetic acid. 4 HF dissolves silica and silicates, and HCl or organic acids are included to keep reaction products in solution. Sandstone matrix acidizing primarily targets damage from migrating fines, swelling clay, carbonate or hydroxide scales, and plugging par- ticles from drilling and completion operations. Understanding formation mineralogy and the
Transcript
58 Oilfield Review
Virtual Testing: The Key to a Stimulating Process
Syed AliChevronTexacoHouston, Texas, USA
Wayne W. FrenierBruno LecerfMurtaza ZiauddinSugar Land, Texas
Hans Kristian KotlarStatoilTrondheim, Norway
Hisham A. Nasr-El-DinSaudi Aramco Research and DevelopmentDhahran, Saudi Arabia
Olav VikaneStatoilStavanger, Norway
For help in preparation of this article, thanks to Ernie Brown,Steve Davies and Vincent Dury, Sugar Land, Texas; Keng Seng Chan and Ray Tibbles, Kuala Lumpur, Malaysia;Matt Gillard and Richard Warren, Aberdeen, Scotland; Abigail Matteson, Ridgefield, Connecticut, USA; and CarlosTorres, Maturin, Venezuela.ClayACID, ELANPlus (Elemental Log Analysis), NODAL, PLT (Production Logging Tool) and Virtual Lab are marks of Schlumberger.
Matrix acid stimulation in a sandstone reservoir involves complex chemical reactions
that are strongly dependent on mineralogy. A new process includes a model for simu-
lating acid reactions to help operators choose an optimal treatment for a formation.
Matrix acidizing reestablishes productivity inmany damaged formations in a cost-effectiveway. The damage can be natural, an artifact ofproduced reservoir fluids moving through a formation, or induced by fluids used in welloperations, such as drilling, completions andworkovers, or stimulation.
Formation damage can be caused by finesmigration, scale formation, deposition of paraf-fins, asphaltenes or other organic material, ormixed organic and inorganic deposition. It canalso result from plugging by foreign particles ininjected fluids, wettability changes, clay swelling,emulsions, precipitates or sludges caused by acidreactions, bacterial activity and water blocks.1
These damage mechanisms can be either naturalor induced. Wellbore cleanup, matrix stimulationtreatments or acid fracturing may be used toremove or bypass near-well damage.
This article focuses on sandstone matrixacidizing.2 A matrix acid treatment forces acidinto a formation at a pressure below its fractur-ing pressure. The treatment often involvesseveral stages that may be repeated.
A new matrix-treatment design processincludes use of the Virtual Lab software, a state-of-the-art geochemical simulator that properlyaccounts for secondary and tertiary reactionmechanisms. Field examples from the North Seaand from the Gulf of Thailand demonstrate theusefulness of this new process for evaluatingtreatment designs.
Reacting to Damage Acid treatments for sandstones differ signifi-cantly from those for carbonate rocks.Carbonate rocks dissolve rapidly in hydrochloricacid [HCl], and the reaction products are solu-ble in water. In carbonate rocks, a matrix acidjob is usually designed to bypass near-well damage by dissolving minerals and creatingchannels, or wormholes, in the rock, providing aflow path past the near-well damage. Acid-fracturing techniques in carbonates create ahydraulic fracture that has a differentiallyetched surface, so that the fracture maintainsits conductivity during production.3
In contrast to acidizing reactions in carbon-ate rocks, the reaction chemistry for silicaterocks is quite complex. Sandstones comprisequartz grains, clays of various types, feldspars,chert, micas, and carbonate materials as cementor overgrowths on grains, along with other minerals (next page). HCl is not effective in dis-solving most constituents of silicate rocks.Sandstone acid jobs typically use hydrofluoricacid [HF] in combination with HCl, formic oracetic acid.4 HF dissolves silica and silicates,and HCl or organic acids are included to keepreaction products in solution.
Sandstone matrix acidizing primarily targetsdamage from migrating fines, swelling clay, carbonate or hydroxide scales, and plugging par-ticles from drilling and completion operations.Understanding formation mineralogy and the
Spring 2004 59
nature of the damage is critical for designing aproper acid treatment. An improperly formulatedacid treatment can precipitate reaction productsin the formation, reducing rock permeability.
A primary objective of designing an acidtreatment in sandstones is optimizing damageremoval, while minimizing formation of damag-ing precipitates. The first 3 ft [0.9 m] into aformation from a wellbore experiences thegreatest pressure drop during drawdown, and iscritical for flow. This region, sometimes calledthe critical matrix, is the volume that matrixacidizing treatments target for cleanup.
If HF comes in contact with calcium carbon-ate [CaCO3] during a treatment, then it leads tocalcium fluoride [CaF2] precipitation. For thisreason, a matrix treatment usually includes apreflush stage with an acid such as HCl or anorganic acid to dissolve most of the carbonate
minerals.5 The main treatment that follows isoften either a mud acid, a combination of HFand HCl, or a retarded formulation such as theClayACID fines-control retarded acid, which is acombination of fluoboric acid [HBF4] and HCl.The HBF4 hydrolyzes slowly to form HF and also
1. A water block is a production impairment that may occurwhen the formation matrix in the near-well area becomeswater-saturated, thereby decreasing the relative perme-ability to hydrocarbons. Water block may result from theinvasion of water-base drilling or completion fluids orfrom fingering or coning of formation water.
2. For more on matrix acidizing: Crowe C, Masmonteil J,Touboul E and Thomas R: “Trends in Matrix Acidizing,”Oilfield Review 4, no. 4 (October 1992): 24–40.
3. Al-Anzi E, Al-Mutawa M, Al-Habib N, Al-Mumen A, Nasr-El-Din H, Alvarado O, Brady M, Davies S, Fredd C,Fu D, Lungwitz B, Chang F, Huidobro E, Jemmali M,Samuel M and Sandhu D “Positive Reactions in CarbonateReservoir Stimulation,” Oilfield Review 15, no. 4 (Winter 2003/2004): 28–45.
> Sandstone minerals and clays. Pore-filling and pore-lining minerals and clays in sandstones can decrease permeability. The minerals and clays havedifferent morphologies, such as pore-filling kaolinite books (A), fibrous illite (B), carbonate overgrowth (C), feldspar overgrowth (D) and quartz cement (E).
A
C
B
E
D
4. The acid formulation used in any specific instance isdependent on formation mineralogy.
5. Organic acids are blended with ammonium chloride[NH4Cl] brine to minimize clay swelling. For further infor-mation: Thomas RL, Nasr-El-Din HA, Mehta S, Hilab V andLynn JD: “The Impact of HCl to HF Ratio on Hydrated SilicaFormation During the Acidizing of a High TemperatureSandstone Reservoir in Saudi Arabia,” paper SPE 77370,presented at the SPE Annual Technical Conference and Exhibition, San Antonio, Texas, USA, September 29–October 2, 2002.
reacts with clays, leaving behind a glassy borosil-icate coating that cements and stabilizes clayparticles.6 Acid treatments are often followed byan overflush, either diluted HCl or ammoniumchloride [NH4Cl], to remove the treatment-reaction products from the near-well volume. Atreatment normally includes injection of adiverter followed by a repetition of these three stages.
A wide variety of acid formulations is available, and the best treatment for a given formation depends on the characteristics of thatformation.7 The new Virtual Lab geochemicalsimulator provides a tool that helps guide theselection based on formation parameters andtreatment chemicals. The simulator models reac-tions and indicates the amount and location ofdissolution and precipitation of mineral species.
The primary reaction between aluminosili-cates and HF from ClayACID and mud-acidtreatments yields fluosilicic acid [H2SiF6], alongwith several aluminum-fluorine complexes. Inthe presence of sodium and potassium, andunder certain conditions of temperature andacid concentration, precipitation of compoundssuch as sodium fluosilicate [Na2SiF6] and potas-sium fluosilicate [K2SiF6] can occur. In thepresence of additional aluminosilicates, H2SiF6
can react to produce amorphous silica [H4SiO4]as a secondary reaction. Amorphous silica canalso result from tertiary reactions of aluminumfluorides with aluminosilicates.8
Amorphous silica and the other compoundslisted above can block pores when they precipi-tate. A successful treatment design mustminimize the precipitation of these compounds inthe formation, particularly in the critical matrix.9
Simulated ReactionsThe reaction of HF with minerals in sandstonesis slow, and the secondary and tertiary reactionsthat generate precipitates are even slower. Theoutcome of an acid treatment depends stronglyon the amount and location of the precipitates.Therefore, predicting the results of a treatmentrequires knowledge not only of the equilibriumreaction products, but also of the reaction kinetics of the acid in the formation.
Reaction kinetics determine the rate atwhich the concentrations change as the systemapproaches equilibrium. The composition atequilibrium depends on the stability of thespecies at the given conditions and is calculatedfrom thermodynamic data. Both kinetic andthermodynamic parameters must be known forall reactive fluids and minerals to predict theamount and the location of dissolved and precipitated minerals around the wellbore.
Past practice has been to obtain specificreaction information through core-plug tests.Ideally, a core should come from the well andformation that is to be acidized, but it oftencomes from a nearby well. Outcrop samples andsamples formed of packed sand mixed with clayminerals have also been used, but matching aspecific formation mineralogy and sedimentol-ogy may be difficult.
Although a core-flow test can provide vitalinformation for designing an acid job, there aretwo inherent problems with such tests: the core plug is too short and the radial geometryaround a wellbore is not honored (left).
Most core plugs are only a few inches long.Reaction products flow out of the core before thesecondary and tertiary reactions can occur andcan generate precipitates. Use of 3-ft long coreshas been recommended to alleviate this prob-lem.10 However, obtaining sufficient formationmaterial for long cores is difficult. Analog outcropsor sandpacks can provide sufficient material, butat the expense of potentially poor matches to for-mation mineralogy and sedimentology.
Long linear cores do not address the geome-try problem. As injected fluid flows out from awellbore, the cross-sectional area that it flowsthrough increases proportionally to the radius.With the same volume flowing through a largercross section, the flow rate decreases away fromthe wellbore. For an 8-in. diameter wellbore, theflow rate 3 ft into the formation is only 10% of therate at the sandface. This slower flow ratestrongly impacts the location of precipitates fromsecondary and tertiary reactions (next page).
60 Oilfield Review
> Acid cleanup in core and near-well formation. A short core plugrepresents a small part of a treatment volume in a formation. At thelength scale of a short core, the permeability appears to improve after a 12% HCL and 3% HF (12/3) mud-acid treatment, but reprecipitationdamages formation permeability just beyond that length. The retardedacid results in better permeability over the treatment volume extending0.9 m [3 ft] or more. Permeability (k) is plotted as a ratio to the far-field,undamaged permeability (k0), which is shown for comparison with thetreated permeabilities (dashed line).
6
4
2
5
3
1
k/k 0
00 0.5 1.0
Radius, m1.5
Wel
lbor
e
Retarded acid12/3 mud acid
Model parameters:149°CKaolinite, 10%Calcite, 3%
Quartz, 87%Porosity, 20%
Spring 2004 61
The new Virtual Lab simulator overcomes theproblem of unrepresentative geometry and pro-vides guidance for successful matrix acidizing insandstone reservoirs. It is the foundation of asystem for designing acid treatments that prop-erly accounts for the cylindrical geometryaround a wellbore (see “A New Stimulation Process,” page 62).11 In addition, Schlumbergerhas created a large, proprietary database ofreaction kinetics and thermodynamics to usewith this simulator. This database saves clientstime and money because additional tests arenecessary only when a formation or a new acidformulation contains compounds that are not inthe database. The need for new tests hasbecome less common as the database has filledwith reaction parameters.
Formation mineralogy can be obtained fromeither whole core or sidewall cores. A short-coreflow test gives an estimate of the surface area ofthe reacting minerals in a formation. This testalso provides information about core permeabil-ity and the effect of an acid on permeability aspore-blocking material dissolves. Short-coretests alone do not provide sufficient informationfor determining an acid treatment, but a short-core test provides data necessary to model
reactions using the Virtual Lab simulator.12
Numerous treatment designs can be tested inthe simulator, and Virtual Lab results will indi-cate the best design for field conditions.
From Laboratory to FieldCentral to any successful acidizing treatment isaccurate information about the reaction chemistry relating to formation minerals. Theliterature contains much of the relevant thermo-dynamic-equilibrium data. However, mostpublicly available reaction-kinetics data arefrom tests obtained at temperatures below fieldmatrix acidizing conditions. Schlumberger labo-ratories performed batch-reactor tests at a widerange of temperatures to create an extensiveproprietary database.13
The database of reaction-kinetics datareduces the number of fluid formulations that it is necessary to test. However, usually at leastone core-flow test is recommended to determinethe reactive surface area of minerals in the formation represented by the core. More than50 core-flow tests have been performed to validate the Virtual Lab software. This databasealso provides analogs for future cases in whichcore material is not available.
The flow test on a core sample from the Heidrun field was typical of the procedure.14 Asmall formation-core plug, 3.73 by 6.4 cm [1.47by 2.5 in.], obtained from a well near the one tobe treated, was saturated with simulated forma-tion brine and flushed alternately withlaboratory oil and brine until the effluent wasclear. A laboratory engineer heated the core toreservoir temperature and flowed prefilteredtest fluids through the core with a 1,000-psi [6.9-MPa] backpressure. This pressure kept anygenerated carbon dioxide [CO2] in solution.
The Heidrun field study used a 9/1 mud-acid—9% HCl and 1% HF—and a ClayACIDtreatment. Flow rate and differential pressuredata recorded every 30 s allowed calculation ofpermeability throughout the test. The engineercollected effluent in 10-mL plastic tubes on aregular schedule and noted any fines in the sam-ple. After filtering and diluting with nitric acid
6. Thomas RL and Crowe CW: “Matrix Treatment EmploysNew Acid System for Stimulation and Control of FinesMigration in Sandstone Formations,” paper SPE 7566,presented at the 53rd SPE Annual Technical Conferenceand Exhibition, Houston, Texas, October 1–3, 1978; also inJournal of Petroleum Technology 33, no. 8 (August 1981):1491–1500.
7. Al-Dahlan MN, Nasr-El-Din HA and Al-Qahtani AA: “Evaluation of Retarded HF Acid Systems,” paper SPE65032, presented at the SPE International Symposium on Oilfield Chemistry, Houston, Texas, USA, February13–16, 2001.
8. Nasr-El-Din HA, Hopkins JA, Shuchart CE and Wilkinson T:“Aluminum Scaling and Formation Damage Due to Regu-lar Mud Acid Treatment,” paper SPE 39483, presented atthe SPE International Symposium on Formation DamageControl, Lafayette, Louisiana, USA, February 18–19, 1998.
9. Thomas et al, reference 5.10. Gdanski R: “Fractional Pore Volume Acidizing Flow
Experiments,” paper SPE 30100, presented at the SPEEuropean Formation Damage Conference, The Hague,The Netherlands, May 15–16, 1995.
11. Ziauddin M and Robert J: “Method of Optimizing the Design, Stimulation and Evaluation of Matrix Treatment in a Reservoir,” U.S. Patent No. 6,668,992 B2(December 30, 2003).
12. Ziauddin M, Gillard M, Lecerf B, Frenier W, Archibald Iand Healey D: “Method for Characterizing Secondaryand Tertiary Reactions Using Short Reservoir Cores,”paper SPE 86520, presented at the SPE InternationalSymposium and Exhibition on Formation Damage Control,Lafayette, Louisiana, USA, February 18–20, 2004.
13. Ziauddin M, Frenier W and Lecerf B: “Evaluation of Kaolinite Clay Dissolution by Various Mud Acid Systems(Regular, Organic and Retarded),” presented at the 5thInternational Conference and Exhibition on Chemistry inIndustry, Manama, Bahrain, October 14–16, 2002.Hartman RL, Lecerf B, Frenier W and Ziauddin M: “AcidSensitive Aluminosilicates: Dissolution Kinetics and Fluid Selection for Matrix Stimulation Treatments,” paperSPE 82267, presented at the SPE European FormationDamage Conference, The Hague, The Netherlands, May 13–14, 2003.
14. Ziauddin M, Kotlar HK, Vikane O, Frenier W and Poitrenaud H: “The Use of a Virtual Chemistry Laboratoryfor the Design of Matrix Stimulation Treatments in theHeidrun Field,” paper SPE 78314, presented at the SPE 13th European Petroleum Conference, Aberdeen,Scotland, October 29–31, 2002.
> Reaction time. Longer reaction time increases aluminum [Al] concentrationin the effluent, but silicon [Si] concentration first increases from zero, thendecreases, for both mud-acid (top) and ClayACID treatments (bottom). Themodel curves show that excluding secondary and tertiary reactions, as couldhappen in a short core test, could lead to incorrect predictions.
0.6
0.4
0.2
00 100 200
Time, min
Al and Si without secondaryand tertiary reactions
Al
Si
300
Conc
entra
tion,
mol
/kg
0.6
0.4
0.2
00 100 200
Time, min
Al and Si without secondaryand tertiary reactions
Al
Si
300
Conc
entra
tion,
mol
/kg
(continued on page 64)
62 Oilfield Review
A new process for matrix acid stimulationrelies heavily on the Virtual Lab software.Mineral-fluid reactions are simulated quicklyand efficiently, so the best treatment optioncan be selected. Schlumberger has developedseveral databases in a proprietary dataarchive to use with the simulator.
The design process starts with a collectionof well data (next page). Mineralogy, which isan important parameter for proper stimula-tion design, can be obtained from X-raydiffraction of core material. The other datainclude well completion, formation tempera-ture, porosity, permeability, evidence relatingto formation damage, and well history.
Schlumberger has created an extensivedatabase of reaction kinetics and thermo-dynamics, but occasionally some specifickinetics parameters are not available. In thatcase, reactions monitored in a controlled envi-ronment, a batch-reactor, provide necessarydata. The new results are added to the database.
As the next step, experts recommend per-forming at least one flow test using corematerial relevant for each formation to bestimulated. These core tests are also stored in the database, so a new test is not necessaryif results are already available. If they are notavailable, and suitable core material can beobtained, then a flow test should be per-formed to provide data for the Virtual Labsimulator to match mineral surface area andthe permeability-porosity relationship for thespecific formation. Only for cases in whichcore tests or core material are not availableshould an analog to the formation be used.The core-flow database is the first place tolook for such an analog.
With all this information collected, a VirtualLab model can be built for the formation. Itincludes the effect of radial flow from thewellbore. The model can perform sensitivitystudies when the well-log data indicate heterogeneity in the formation-mineral compo-sition. Data selected from the reaction andcore databases feed into the model.
A stimulation expert selects a few treatmentfluids based on the information obtained forconstructing the model. Each treatmentoption is simulated. Various injection volumes,rates and shut-in periods can also be evaluated.Uncertainties in the data can be checked byrunning a sensitivity analysis, which VirtualLab software can do automatically.
With an optimal treatment schedule deter-mined, an operator can now perform therecommended treatment.
If real-time bottomhole pressure data areavailable during the operation, the treatmentdesign can be adjusted while in progress(below). If operational constraints preventthe treatment from proceeding as planned,the constraints can be put into the model and
the treatment redesigned. Once the designand the operational parameters agree, thereal-time data of bottomhole pressure, injec-tion rates and fluids injected can be comparedwith model expectations. If there is a significantdiscrepancy, the model assumptions are reex-amined. For example, the real-time data mayprovide a new insight into the type, quantityor location of damage, or may suggest that thepermeability-porosity relationship in the for-mation differs from that measured in the core.After the model is adjusted, the redesignedtreatment can continue. This ability to adjustthe model in real time provides a great benefitin helping operators optimize stimulation jobs.
After the treatment, flowback and produc-tion data can be used to adjust the model onelast time. The updated model for that fieldand reservoir is then available to optimizefuture treatment jobs.
A New Stimulation Process
> Real-time feedback loop from process to model. When real-time bottomholepressure data are available, the model can be adjusted to update the processwhile it is under way. This feedback loop continues until the job is complete.
Begin treatment
Yes
Yes
No
No
Do operationalconstraints prevent thetreatment from being
executed asplanned?
Do the real-time datamatch expectations from
the model?
Check model assumptions• Formation-damage type, quantity and location• Permeability-porosity relationship
Adjust model
Redesign treatment
Read real-time data• Bottomhole pressure• Injection rate• Fluid type and volume injected
Spring 2004 63
> The stimulation process using Virtual Lab simulation and the proprietary data archive. The process begins on the left and proceeds clockwise.Solid lines are the process steps and dashed lines are data transfers into, out of, or within the data archive. A real-time feedback loop can updatethe model while the crew performs the treatment.
Calibrate parameters fromcore-flow test• Mineral surface areas• Permeability-porosity relationship
Reservoir modelsFluid database
Reaction kineticsand thermodynamics
Proprietary data archive
Are reaction kineticsof minerals available
in database?
Yes
Yes
No
No
YesNo
Is reservoir coretest available in
database?
Is it feasibleto obtain and test
reservoir core?
Collect well data• Mineralogy• Temperature• Porosity and permeability• Formation damage• Well history• Well completion
Estimate mineral surfaceareas and permeability-porosity relationship
Build model for reservoirtreatment, accounting for• Radial flow• Heterogeneities in permeability and mineralogy from well-log data
Determine treatment using model• Optimize for fluid type, volume and injection rate• Examine sensitivity to data uncertainty• Examine treatment scenarios
Examine treatment from postjob data• Flowback• Production
Update model assumptions• Formation damage type, quantity and location• Permeability-porosity relationship
Conduct batch-reactor tests to
measure kinetics
Core-flow tests
Select reaction data
Select formation data
Client report Perform treatment
Begin
End
to prevent further precipitation, the fluid samples were analyzed to determine the concentration of aluminum and silicon (below).Changes in effluent composition provided infor-mation about the type and morphology ofreactive minerals in the core. The Virtual Labsimulator matched the flow-test results, providingthe mineral surface area and permeability-porosity relationship.
The acid treatment did not deconsolidate theHeidrun field core and did not form precipitates,indicating that this treatment fluid was compati-ble with the native mineralogy.15 It also providedthe desired permeability improvement.
First Use of Simulator for Stimulation Statoil operates the Heidrun field, located in theHaltenbanken area of the Norwegian Sea,120 km [75 miles] south of the Arctic Circle. Thetarget well, A-48, had a deviation angle of 48°across the producing interval in the Tilje forma-tion and was completed with an openhole gravelpack.16 Productivity in this zone declined afterformation-water breakthrough, and worsenedafter a scale-inhibitor squeeze treatment.Design of a matrix-stimulation job was difficultbecause this was the first well in the Tilje forma-tion to be acidized. The formation washeterogeneous, with high clay content and largeclay clasts (bottom).
The first use of the Virtual Lab geochemicalsimulator was for a treatment in the Heidrun A-48 well. The software simulated both batch-reactor and core-flow tests specific to the Tiljeformation and provided the parameters neededfor a stimulation model. The team simulated sev-eral treatment scenarios and several acidformulations to optimize the fluid types,sequences, volumes and injection rates.17
The core test described earlier in “FromLaboratory to Field” showed that permeabilityincreased during the ammonium chloride flushthat followed injection of the 9/1 mud acid. Thisindicated continuing movement of fines out ofthe core. However, in the field, continued flush-ing would move those fines deeper into theformation, causing damage when the flow slowedor stopped and the fines settled. A flowbackstage was included after the mud-acid stage toclear the mobile fines out of the formation.
The treatment design was based on the coreand reactor tests.18 During the treatment, Statoilcaptured samples from all fluid returns anddetermined the profile of ions in these fluids ateach stage. With this information, Virtual Labsoftware confirmed that fines migration was themost likely primary damage mechanism andallowed the operator to examine the possibilitiesof combined damage mechanisms. This simula-tion showed that the final design improvedpermeability while limiting mineral precipitation(next page, top). The model recommended injec-tion rates that could not be maintained duringexecution because of operational difficulties. Asecond run of the model using actual flow ratesand fluid volumes indicated that the differencein fluid placement between the recommendedand executed procedures was minor.
Before the stimulation treatment, the wellproductivity index was 20 m3/bar-d [9 bbl/psi-D]and reached 55 m3/bar-d [24 bbl/psi-D] immedi-ately after the treatment. The productivity indexover the next seven-month period averaged42 m3/bar-d [18 bbl/psi-D]. The acid treatmentsuccessfully removed the near-well damage andcontrolled fines migration (next page, bottom).The Virtual Lab model optimized after treating
64 Oilfield Review
15. An acid treatment normally dissolves some cement;deconsolidation indicates that so much cement dissolvedthat the core matrix was no longer competent.
16. Ziauddin et al, reference 14.17. Ziauddin et al, reference 14.18. Ziauddin et al, reference 14.
> Core-flow test. The permeability response to treatment acids is measured during a Heidrun fieldcore test. The increasing permeability during the NH4Cl brine flush following the 9/1 mud-acidtreatment indicates movement of fines out of the core (bottom). The upper plot shows elementalconcentrations in the effluent. After changing injection fluids, the permeability change is seen beforean effluent effect because the new fluid has to pass through the core. All the solid lines are best-fitresults from the Virtual Lab model, providing essential parameters for modeling the treatment.
> Clay clasts. The Tilje formation in the Heidrun field contains large clay clasts, apparent in thecomputed tomographic image (left). The section AA’ includes large, dark, clay clasts (center). Thelower section BB’ shows clay laminae (right).
< Matrix acid-treatment model results. Thetwo-phase Heidrun field treatment startedwith a mud-acid treatment, then a flow-back stage, followed by a ClayACIDfluoboric phase (table). The geochemicalmodel predicted that the treatment wouldimprove near-well permeability (top). Thetotal silica mineral precipitation was low,less than 2.5% of the formation volume(middle). Borosilicate precipitation, usefulfor stabilizing clays, peaks near the well-bore, while amorphous silica peaks fartheraway (bottom).
> Production data for Heidrun field Well A-48. Productivity declined when water broke through, andfurther productivity was lost after a scale-inhibitor squeeze treatment. The acid-stimulation treatmentin September 2001 restored productivity without significantly increasing the amount of produced water.
5,000
6,000
3,000
Oil r
ate,
m3 /d
Prod
uctiv
ity in
dex
(PI),
m3 /b
ar-d
Wat
er c
ut, %
4,000
2,000
0
1,000
1/17/007/1/99 8/4/00 2/20/01
Date
9/8/01 3/27/02 10/13/02
125
150Injection-waterbreakthrough
Scale-inhibitorsqueeze
Acid-stimulationtreatment
75
100
50
0
25
Oil ratePIWater cut
the Heidrun A-48 well provided vital informationto shorten the learning curve for treatment ofother wells in this complex, clay-rich formation.
Damage Mechanisms in the Galley FieldOperator ChevronTexaco used the new acid-stimulation process in the Galley field on the UKcontinental shelf. The G5 well was completedhorizontally with a 650-ft [200-m] openhole section in the late Paleocene-age Cromarty formation, which comprises fine to very fine-grained, poorly consolidated, turbiditicsandstone (above). Most of the productive sec-tion has a 100-mm mesh screen in place thatwas originally intended for a gravel pack, but ashale section about a quarter of the way alongthe horizontal section collapsed. Although pro-ductive sand channels beyond the collapsedshale are accessible for flow into the wellbore,those sections could not be gravel packed.19
Oil production declined steadily from an ini-tial 7,000 B/D [1,100 m3/d], but the oil declinerate accelerated when water productionincreased in April 2002. Before the stimulationtreatment, the well produced about 1.1 millionbbl [175,000 m3] of oil and 979 MMcf [28 millionm3] of gas, along with about 31,000 bbl [4,900 m3]
of water. Significant recoverable reservesremained within the well’s drainage area.
The combined ChevronTexaco and Schlumberger stimulation team examined several possible damage mechanisms to explainthe loss of oil production.
Drilling-induced damage—Filtrate inva-sion; invasion of a calcium carbonate bridgingagent, polymer, starch and drilled solids; and fil-tercake plugging of the screen and sandfacecould go unnoticed initially in a horizontal well.However, such damage can create localized pro-duction areas, which can eventually lead to earlywater breakthrough, loss of screens and acceler-ated fines production.
Completion damage—The collapse of theshale section prevented a complete gravel pack,so the filtercake and mud removal in the sectionbeyond the damage was probably extremely poor.
Swelling clays—X-ray diffraction mineral-ogy from a core sample showed that the volumeof swelling clays, such as smectite, was too lowto be a damage mechanism.
Inorganic scale—Damage from barium sulfate [BaSO4] was expected to be small, butCaCO3 scale could be a major source of damage.Limited data were available to quantify the volumes of scale.
Water-holdup problems—NODAL productionsystem analysis results showed that water cut inthis field must exceed 50% to create a significantimpediment to production. The measured value of20% shows this is an unlikely damage mechanism.
Fines migration—X-ray diffraction resultsindicated the presence of migratory clays suchas chlorite and illite along with mobile quartziteparticles. A pump-in test supported fines as adamage source. Permeability increased duringthe pump-in—that is, reverse-flow—period, ascompared with the permeability during produc-tion. Further evidence of fines migration wasfound in the decreasing oil production withincreasing water production, since water candestabilize fines and cause them to migrate.20
Finally, the formation is unconsolidated, and other wells in the area had experiencedfines migration.
This analysis indicated that the treatmenthad to remove damage possibly caused bydrilling, inorganic scale and migration of claysand quartzite particles. The proposed treatmentstarted with jetting a chelating agent using acoiled tubing string with a high-pressure noz-zle.21 This treatment, which stabilized iron andalso removed CaCO3 scale, was followed by
66 Oilfield Review
> Galley field well G5 log and production history.The horizontal section of this well passes througha channel, then a shale section and threeproductive zones (top). The shale section from7,440- to 7,570-ft [2,267- to 2,307-m] measureddepth (MD) collapsed during completion operations.Oil production declined steadily starting inJanuary 2002, with water production increasingbeginning in April 2002 (left). This well later had amatrix acidizing treatment.
acetic acid to help remove additional CaCO3 andto provide a preflush for the final treatment,which was a 9/1 organic mud acid.22 The VirtualLab process provided a means to test the effec-tiveness of this treatment schedule.
Reaction kinetic parameters were availablein the database. A core-flow test on a small plugfrom the Cromarty formation provided an esti-mation of mineral surface areas and parametersfor the permeability and porosity correlation.The test showed that treatment fluids were com-patible with the native mineralogy and that theyincreased permeability within the core sample.
The next step was to simulate the reservoirgeometry using the Virtual Lab software. In thissimulation, damage was assumed to be due onlyto fines migration. The model showed that well-bore skin factor declined steadily with thetreatment, and a small quantity of amorphoussilica reprecipitated near the wellbore (below).
A PLT Production Logging Tool run justbefore the main stimulation treatment was ana-lyzed in real time and indicated no productionfrom the gravel-packed channel sand. The firsthalf of the productive interval beyond the shalesection produced oil with a 50% water cut, andthe second half produced dry oil at a low rate.Since water was not coming from an isolatedzone, it was not possible to stimulate oil produc-tion alone.
The first treatment stage was jetting achelating agent along the entire wellbore. Thisstage mechanically cleaned the wellbore andincreased the oil production rate to 1,000 B/D[160 m3/d] with a water cut of 40%. Flowbackafter the treatment was slower than plannedbecause of operational problems. A postjob Virtual Lab simulation showed that the effect ofthis additional fluid-residence time was a smallincrease in silica precipitation that would haveminimal effect on productivity.
The complete treatment increased oil pro-duction to 3,000 B/D [480 m3/d], 15 times thepretreatment production rate. The water cutincreased slightly to 45%. After three months ofproduction, the well produced oil steadily at1,500 B/D [240 m3/d].
The productivity increase was better thanthat predicted by the geochemical simulation.The model had assumed that the main cause ofdamage was fines migration, but it is possiblethat the dominant damage came instead fromCaCO3 scale or residual drilling and completionfluids. Real-time bottomhole-pressure readingsand an analysis of the flowback fluids were notavailable. Had they been, the Virtual Lab simula-tor could have estimated the contributions ofthe various damage mechanisms, furtherimproving future jobs in the field.
Sensitive Clays in the Gulf of Thailand Several fields operated by ChevronTexaco in theGulf of Thailand have similar lithologies. Theproductive sandstone formations have HCl-sensi-tive clays in proportions greater than 15%, andthe reservoir temperature exceeds 250°F[120°C]. The formation also contains carbonateminerals.23 The primary damage mechanisms areswelling of smectite and other clays and migra-tion of clays such as kaolinite-illite and
illite-smectite mixtures. These clays can eitherline or fill pore spaces.
Conventional matrix acidizing—using mud-acid and ClayACID treatments—was ineffectivein restoring well productivity in this area.24 InApril 2002, Schlumberger used a new clay-stabilizing acid in this field, a ClayACID formula-tion using an organic acid in place of the HCl. Theclay-stabilizing acid is designed to permanentlystabilize a formation containing high percentagesof silt and clay, while minimizing secondary andtertiary reactions. The treatment deposits a layerof borosilicate glass that immobilizes the clays.The formulation was successful in four of sixtreatments, and the production increase was stable for at least six months after treatment.Nevertheless, a posttreatment analysis indicatedthat a better methodology for selecting candidatewells could yield improved results.
The second stimulation campaign, carriedout in 2003, used the Virtual Lab software forprestimulation analysis to improve results. Thegeochemical model inputs included a mineralcomposition of 9% carbonate minerals, 18%clay—illite, mixed illite and smectite, kaoliniteand chlorite—and 6% feldspar. The large propor-tions of these minerals, in conjunction with thehigh reservoir temperature, make treatmentdesign difficult.
19. Ziauddin et al, reference 12.20. Relative permeability effects also can cause decreased
oil production with increased water production. It is thecombination of this effect and the pump-in tests that support a conclusion of fines migration.
21. A chelating agent stabilizes metal compounds, preventingthem from precipitating.
22. An organic mud acid uses formic acid in place of HCl, so a 9/1 organic mud acid is 9% formic acid and 1% HF.
23. Torres C, Ziauddin M, Suntonbura N, Xiao J and Tibbles R:“Application of a Unique Clay Stabilizing Acid in the Gulfof Thailand,” presented at the PetroMin DeepwaterTechnology Conference, Kuala Lumpur, Malaysia, July 14–17, 2003.
24. Torres et al, reference 23.
> Model results for amorphous silica. The originally planned treatment usingorganic mud acid produced less than 0.5% of amorphous silica, as apercentage of the formation volume, in the critical volume of the matrix nearthe wellbore (green). Operational difficulties forced a delay in the flowbackstage, so the model was rerun using the actual times. The silica depositedwith this additional soak time was still small (blue).
1.0
0.5
Silic
a vo
lum
e, %
Wel
lbor
e
00 0.5 1.0
Radius, m1.5 2.0
Treatment with shut-inTreatment without shut-in
Critical matrix
The geochemical simulation tested pre-flushes of both 10% acetic and 5% formic acid toremove near-well carbonate minerals from theformation. The two formulations provided simi-lar skin reduction, so acetic acid was usedbecause it was more readily available at thetime. The model indicated the optimal preflushvolume and the optimal clay-stabilizing acid vol-ume (above). The simulation showed that theborosilicate coating that stabilizes claysextended about 1 ft [0.3 m] into the formationwith the optimal clay-stabilizing acid treatment,but that additional clay-stabilizing acid did notextend the protected zone significantly.
ChevronTexaco planned the second phase of clay-stabilizing acid stimulations based onoptimized acid volumes from the Virtual Labsimulator. Stimulation jobs on one oil-producingwell and three gas producers were successfuland showed significant production increases(right). This use of the new stimulation designprocess increased profitability from the stimu-lated wells. The fluid system, customized for thespecific lithology in the Gulf of Thailand wells,provided a lasting solution.
Reacting to the Future The new stimulation process, including the Virtual Lab simulator, provides a tool to improvewell performance in sandstone formations.Sandstone matrix acid treatments are complex,and the success rates are historically low. Thenew process with the software and proprietarydatabases as its basis assures a much higherratio of successful matrix acid treatments.
Determining formation mineralogy is animportant first step in the process. If data suchas the ELANPlus Elemental Log Analysis areavailable, they can be used with the Virtual Labsoftware. In addition, the growing databases forgeochemistry and flow properties will providemore analogs for locations lacking core material.
The Virtual Lab software is a general-purpose geochemical simulator and is not
restricted to solving for matrix acidizing in sandstones. The tool could be used for carbonateacidizing, carbon dioxide sequestration andwater-compatibility testing. Schlumberger continues to expand the reaction database,increasing the variety of problems that this geochemical software can solve for the industry. —MAA
68 Oilfield Review
> Optimizing treatment volumes. The geochemical model accounts for acetic acid spending, or weakening, as it interacts with formation carbonateminerals. Far from the wellbore, the carbonate is 7% of the formation volume. The radius of formation that is cleaned of carbonate material is much smallerthan the invaded radius. Injecting 100 gal/ft [1.2 m3/m] of perforated height cleared carbonate to a greater radius than did a volume of 75 gal/ft [0.9 m3/m](left). However, additional injection did not significantly increase the cleared radius. Using a 100 gal/ft preflush, the model indicated an optimal treatmentusing clay-stabilizing acid of 75 gal/ft (right). Beyond that quantity of injected clay-stabilizing acid, skin increased because permeability was destroyed.
10
8
6
4
2
Carb
onat
e vo
lum
e, %
00 1 2
Radius, ft3 4
Damaged zoneW
ellb
ore
400
300
200
100
Skin
00 100 200
Volume, gal/ft300 35050 150 250 400
10% acetic-acid preflush
Clay-stabilizing acid NH4CI brine
75 gal/ft100 gal/ft125 gal/ft150 gal/ft
3.6 ft4.2 ft4.7 ft5.2 ft
Fluidinvasion
> Production improvement in Gulf of Thailand wells. Matrix acidizing increasedoil production in Well B-1 from 0 to 442 B/D [70 m3/d] (green). A comparison ofpretreatment (pink) and posttreatment (red) production from three gas wellsalso shows significant improvement.
