SPE-107831-PP

Streamline-Based Integral Modeling for Waterflooding Design Optimization,

Surveillance and Monitoring
Mirta Galacho1, Pablo Vázquez1, Gerardo Laiz1, Andrés Krittian2, Fabián Gutiérrez2, Victor Pacci2
1: MG&A Oil & Gas; 2: Petrobras Energía

Abstract Introduction
The Dynamic Numerical Simulation in Flowlines The classical techniques of waterflooding surveillance,
(DNSFL) is an alternative tool adapted to handle Dynamic including methods like: mapping (gas-oil ratio, water cut,
Models in Fine Scale. This feature has been particularly pressure, etc.) total liquid production vs. time, injected poral
relevant for studying the current case, a huge multilayered volumes vs. recovery factor, Hall plots, etc., jointly with the
waterflooding process developed in a giant field of great monitoring process have backed-up successful Exploitation
extension reservoirs with considerable facial and Optimization processes[1]. Most of these techniques, as those
stratigraphical variations. mentioned above, are 0D; some recent practices made 1D or
The DNSLF develops these tasks suitably because 2D spatial distribution of the relevant characteristics to the
uncouples reservoir geometry and heterogeneity from process, essentially productions and injections, but without
transport equations, solving the problems dominated by integrating the pressure fields.
convective flows in a faster and computationally more Streamline simulation goes further and incorporates
efficient way. This allows to build models of greater space pressure fields, which determine streamline as the most
discretization and, therefore, to represent better the probable way of fluid movement. In this sense, and always
heterogeneity of the reservoirs. in the pressure field, there are two well differentiated work
The analyzed field is constituted by fluvio-lacustrine levels in the waterflooding projects[2] (Fig. 1):
deposits, nine sand-clay cycles of normal grading (only 1. The first level, called “Production-injection surveillance”
eight of them were modeled), partially connected reservoirs; based on the historic analysis of the waterflooding,
with 250 actives wells in commingled production and water allows for the examination of injectors and producers,
injection; and with a long and detailed history of identifying well and poorly swept areas.
simultaneous primary and secondary events. 2. The second level, “Streamline-based simulation” based
In a previous paper (SPE 94815) a Streamline-based on a detailed geological model, fluid physics and history
Global History Matching of this field was presented. This matching, which allows, in addition to the previous one,
process enabled to achieve the Geological Modeling for an integral redesign and forecast of the waterflooding
Calibration, a clear conceptualization of the current primary towards its optimization.
and secondary production mechanisms, its productive Both levels are sustained and fed by monitoring the field
behavior, and to evaluate the geostatistical and Upscaling process; as it is clear, the development of the second level,
procedures to apply for the definition of the Simulation as in the example of Puesto Hernández Rayoso Field– Block
Model. 4, implies achieving the first level.
This paper illustrates how the Integral Model achieved, As it is known, the classical surveillance techniques have
with a detailed Streamline-based History Matching, is used been widely discussed in several well-known publications
for Waterflooding Design Optimization, Surveillance and and experts have recommended some high-value principles
Monitoring, showing that these principles are key factors to to develop them. The following principles make up the
understanding reservoir performance and identifying minimum platform required by simulation techniques,
opportunities that will improve the ultimate recovery. especially by the streamlines, here considered:
During the detailed History Matching process CPU • A key ingredient of any surveillance and simulation
runtimes around 200 minutes were achieved using a program is the planning, gathering and validation of
1225000 grid cells Model, with 190 timesteps, quarterly at “all” available information.
the starting, and monthly after, based in a Pentium 4 PC, • To implement surveillance and simulation efforts it is
3.2GHz CPU and 2GB RAM. It showed that it is possible to essential to “understand” the reservoir expressed in
work with a big Streamline-based Model in relatively short their characteristics and fluid flow, while reducing
processing time. the uncertainty of the interpretation. In the case of
simulation, this understanding is reflected in reservoir
models adjusted in the History Matching process.

1
 • In general, one classical surveillance technique is not petrophysical characteristics, low initial reservoir pressure
meaningful because different parameters and oil viscosities that vary laterally and with depth.
(characteristics of reservoirs, production The main characteristics of these reservoirs, in the
mechanisms, etc.) can produce similar answers. modeled area of Fm. Rayoso are:
Instead, a well planned simulation model integrates • Average depth: 500 to 700 mbgl
the relevant analysis to the process under evaluation • Average crude density: 25ºAPI
–waterflooding evaluation in this case- thereby • Crude viscosity: 15 to 95 cp
avoiding this problem. • Initial static pressure ≈ 26 Kgf/cm2
• Achieving a waterflooding surveillance controlled by • Bubble pressure ≈ 15 Kgf/cm2
fluid balance in the patterns requires important • Gas in oil solubility ≈ 6 m3/m3
technical efforts (from engineering and geology) • Reservoir temperature: 33ºC
during the life of the project. Simulation also requires
important technical efforts but specially concentrated Development and Exploitation History
on the initial phase –i.e. when models are developed. By mid 1976 the primary exploitation of Fm. Rayoso in
• Surveillance techniques should always induce Puesto Hernández Field started (Fig. 5). Production
thorough deeper studies which include numeric increased at the beginning of 1983 by perforating this
simulation. formation in several wells that produced before the deepest
As mentioned above, streamline simulation in Puesto levels. Water injection started at the beginning of 1994 in
Hernández Rayoso – Block 4, and the related surveillance some peripheral wells; by mid of 2000 new wells drilling
tasks, feed and depend from the monitoring process of the was intensified and changed to an irregular 9 spot injection
field to optimize the waterflooding presently in progress. pattern.
In all simulation process, including proposed By the end of 2005, over 1500 wells were drilled in the
optimizations, the methodological approach was based on Puesto Hernández area, 37 % from Fm. Rayoso.
blocks, sectors, well groups and wells resulting from the The simulation area contains 265 wells (185 producers
review of many cases of waterflooding. This type of work and 80 injectors) comprising Block 4, where the model is
avoids implementing partial action plans or fast judgment, developed, and well strips from Block 3 and 5 (South and
which is especially important in up to date situations where North from Block 4), as boundary conditions (Fig. 6).
human and capital resources are critical. Production or injection of most of these wells takes place
in over one Formation. Only 240 wells perforated in Fm.
Development Rayoso, produce this formation exclusively (38%), these are
Puesto Hernández field is at the NW border of Neuquén called turnkey wells and the remaining are non-turkey wells.
Basin of Argentina (Fig. 2), extended along 147 km2. It is Most of injectors have valves to assure the selective
developed in the Neuquén embayment, where the injection among various formations but not among the
stratigraphical column is complete although thickness is various hydraulic units of Fm. Rayoso. To the Rayoso non-
reduced due to the proximity to the basin boundary (Fig. 3.) turnkey producers the oil production was allocated
prioritizing the trends noted in Agrio and Huitrín
Description and Main Characteristics of Reservoirs formations.
The productive reservoirs of the field, Rayoso, Huitrín
and Agrio belong to the Cretacic age. This work is Approximation in different levels
developed in the Rayoso Fm, the main field producer. Model Definition and Adjustment
Fm. Rayoso belongs to a clastic-evaporitic unit These tasks were performed by a work flow based on
deposited in a predominantly continental environment of procedures issued by EIA (Estadística Integral
fluvio-lacustrine character[3]. The clastic section is Autocorrelada – Autocorrelated Integral Statistics[4]) for the
interesting from the economic perspective; there are 11 simulator input model and its subsequent adjustment in the
sand-clay cycles, of normal grading out of which the most streamline simulation process (SFL). As expressed above, in
important 8 were modeled. The cycles are truncated towards the creation of the model, history matching and prediction
the E-NE by the intercenomanian unconformity at the top of process, and in the optimization proposals, the work
Fm Rayoso, with three different CAPOs (Original water-oil methodology was based on block, sectors, well groups and
contacts); they consequently form three Hydraulic Units: well per well.
inferior (IHU), medium (MHU) and superior (SHU). Fig. 4 In the primary phase, the field produced due to
shows the truncation of the various cycles of the Hydraulic monophasic fluid and rock expansion, while during the
Units, used to define the sectors of Block 4. waterflooding convective flows are predominant. SFL is an
Structurally the reservoirs are a homoclinal of 5º average appropriate choice for this type of flow[5 a 12].
slope and dip to the SW. The 3D seismic registered in 1995, To better map the behavior of fluids in the simulator the
showed a main faulting that result in E-W vertical and sub- option of Little Compressible Black Oil was chosen.
vertical faults, of few meters of throw. This paper analyzes Considering the petrophysical conditions of reservoirs, the
the hydrodynamic connectivity of these faults, especially in model chosen was of Simple Porosity and Permeability.
Block 4 (Fig. 5). Also, given the large field vertical extension, the presence of
Productive levels are relatively shallow, depth is below gravitating action was activated in the model.
600 mbgl (meters below ground level), show good The faults were represented in the model with their
throw. The best adjustment was achieved considering them
2
like communicating ones, acting as boundary conditions The conclusions for other HU are analogous, MHU
with their natural throw. records higher sweep than Inferior and Superior Hydraulic
The History Matching process included: Units (SHU).
Global Adjustments at Reservoir levels: they were There is a preferential communication trend, West-East,
limited to the surroundings of the poorly adjusted wells (in in the injector-producer pairs. Given the higher density of
the liquid production or static pressure) or to the correction lines, this distribution shows more clarity in MHU.
of poorly informed situations, as in the case of aquifers. These characteristics of sweep efficiencies are related
with the distribution of absolute reservoir permeabilities.
Local Adjustments at Well level:
The sweep areal efficiency in different cycles is
• Incorporating new well data as they were drilled.
conditioned by the variation trend of absolute permeability
• By adjustment of productivity rate to model
(E-W) and the semi - parallel fault direction (Fig. 16).
fractures.
Although in the lower cycles this trend is washed out by the
• By adjustment of transit profiles in the injectors.
limited extension of reservoirs, the productive relevance of
• By adjustment of available RFT (repeat formation
them is low.
tester).
• By adjustment of dynamic pressure of producers. Vertical sweep efficiency is ruled by the vertical
distribution of permeability, the layers of higher admission
The Goodness of the Model History Matching, in the
are those from MHU, with average K (permeability)
various levels, as usual practice, was measured by
between 200 and 250 mD (mili Darcy), followed by the
comparing the results of the simulator with the available
Superior Hydraulic Units, average K ranges between 150
historic data.
and 200 mD; finally Inferior Hydraulic Units –average K
For this process the Block 4 was divided in 4 areas, Fig.
between 100 and 150 mD.
4, limited by the “unconformities” lines in the indicated
cycles. Aquifer sweep, according to the simulation, is produced
Figs. 7, 8, 9 and 10 show the “goodness” of the fluid by some injectors located in the aquifers, but essentially by
produced matching in the Block, Sectors, Well Groups the flow orientation to the inner part of them due to the
(turnkey wells) and well per well. higher current static pressure of the oil areas, as a
Figs. 11, 12 and 13 show the “goodness” of the consequence of the water injection.
adjustment at well level as to injection profiles, static Poorly swept areas were also analyzed in well groups
pressures, and water saturation at the time of well using streamlines. For example, Fig. 17 shows one of these
completion. areas in a N-S cross-section, which includes 3 injector-
The achieve adjustments enables to consider that a producer pairs with the corresponding permeability grid.
“good” simulation streamline model was available (Block Permeability shows relatively important variations (100 to
4). 400 mD). It is also observed that injector wells, not totally
perforated in the permeable thickness, result in a re-flow
Analysis of the Actual Exploitation Scenario around the same well and even to some producers. Instead,
The analysis of this Scenario comprised the sweep injectors with perforated permeable thickness sweep better
efficiencies, the well productivity index and the evolution of to the producers.
pressures, presently and to 10 years of exploitation. That is, sweeping is controlled by permeability,
perforations and dynamic pressure of producers.
The Sweep efficiencies were calculated and graphically
conceptualized, in the different levels of the analysis, based The Time variation of Productivity Index was finally a
on the streamline from injectors, from all the history and well-based analysis. Productivity was estimated by using
future projections, in connection with current oil saturation. dynamic pressures and fluid productions measured with the
The streamline density is proportional to the water injected simulator estimation of static pressure (pseudo-stationary)
by well-layer. around the well.
In the current waterflooding recovery process, the Several wells reduced its Productivity Index; some of
Middle Hydraulic Unit (MHU) is the best swept. As an them in an important way (Fig. 18) resulting from the
example the Fig. 14 shows the streamlines. The Simulation damage in the hydraulic fracture together with a subsequent
lets us to observe: mild damage related to the productive process.
• There are well swept oil areas as well as non-swept
areas. Simulation of the First Optimization Scenario
• Current oil saturations of swept areas do not exceed This scenario considered the improvement of the Areal
50-55% Sweep Efficiency by an arrangement, agreeing injectors and
• Aquifer areas are unwillingly swept areas. producers with the absolute permeability distribution, and
the closing of injectors in the aquifers.
In Fig. 15, with the streamlines for the Inferior
Comparing the streamlines for MHU in this scenario,
Hydraulic Unit (IHU), it is shown:
Fig. 19, with those of the current exploitation scenario, Fig.
• The Oil areas are more poorly swept than MHU.
14, it is possible to observe:
• The current oil saturations of swept areas are closer
• The Increase of swept oil areas,
to those of MHU.
• The Increase of current oil saturation in some areas
• The aquifer areas are swept more significantly than
incorporated to the waterflooding.
in MHU.
• The Reduction of swept areas in the aquifers.
3
 Analyzing the streamlines in this scenario it can be noted SPE 49000 in Proceedings of the 1998 ATCE, New
improved volume efficiency in all the field hydraulic units, Orleans, LA.
Fig. 20. 7. R. O. Baker, F. Kuppe, S. Chugh, R. Bora, S. Stojanovic
and R. P. Batycky: “Full-Field Modeling Using
Conclusions Streamline-Based Simulation: 4 Case Studies”; paper
• The analysis of the current waterflooding SPE 66405 presented at the SPE Reservoir Simulation
exploitation scenario, in different levels, using the Symposium held in Houston, Texas, USA (February
streamline-based simulation made it possible to 2001).
include the monitoring and surveillance 8. R. P. Batycky, M. R. Thiele y M. J. Blunt: “A
injection/production aspects, indicated in the Fig.1, Streamline-Based Reservoir Simulation of the House
and closely related them to the characteristics of the Mountain Waterflood” , SCRF (1997).
reservoirs. From this standpoint, the most adequate 9. G. K. Grinestaff: “Waterflood Pattern Allocations:
recommendations to optimize the process are made: Quantifying the Injection to Producer Relationship with
• A distribution of the injector-producer pairs Streamline Simulation”; paper SPE 54616 (May 1999).
that considers the permeability distribution of 10. R. Baker: “Streamline Technology: Reservoir History
the area would improve the areal sweep Matching and Forecasting = Its Success, Limitations and
efficiency. Future”, Journal of Canadian Petroleum Technology
• Vertical efficiency of current waterflooding (April 2001) Vol 40, No 4.
recovery, conditioned by K vertical 11. M. R. Thiele y R. P. Batycky: “Water Injection
distribution, would substantially increase by Optimization Using a Streamline-Based Workflow”;
selective water injection per Hydraulic Unit. paper SPE 84080 presented at SPE Annual Technical
• The inclusion of “dynamic closings” with Conference and Exhibition held in Denver, Colorado,
productive wells around the aquifers would USA (October 2003).
reduce the sweep to the inner part of them. 12. A. Chakravarty, P. Carreras, C. J. Harrison y R.
• The total or partial elimination of well Cobeñas: “Reservoir Characterization of El Trapial
damages would result in increased production. Field, Neuquén Basin, Argentina: An integrated study
• The first optimization scenario, Fig.19, made (Part 2: Reservoir Simulation)”; paper 2nd Meeting on
sweep volume efficiencies grow dramatically. Reservoir Simulation, Buenos Aires, Argentina
In the simulator it is evident that these (November 2002).
efficiency improvements determine a
substantial increase in the recovery factor of Acknowledgement
Block 4 to be reached in 10 years, Fig.20. The members of the multi-disciplinary team of MG&A
wish to thank the authorities of Petrobras Energía S.A. for
References the authorization to publish this paper.
1. M. Terrado, S. Yudono and G. Thakur: ”Waterflooding
Surveillance and Monitoring: Putting Principles Into
Practice”, paper SPE 102200, presented at the SPE
ATCE in San Antonio, Texas (September 2006).
2. R. Batycky: “Using Streamline-Based Simulation to
Proactively Manage Well-Pairs to Promote Improved
Sweep and Reduce Fluid Cycling”, presented at 2006
SPE ATW – Key Technologies & Tools to Maximize
Production
3. M. Schiuma, G. Hinterwimmer y G. Vergani:
“Reservoir rocks from productive basins in Argentina”,
Symposium of V Exploration and Development of
Hydrocarbons Congress. (November 2002), Mar del
Plata, Argentina.
4. M. Galacho, N. Galacho: Auto-correlated integral
statistic, Copyright, May Mayo 2003.
5. M. Galacho, P. Vázquez, MG&A Oil & Gas and V.
Paci, G. Fernández, C. Canosa, Petrobras Energía:
“Streamline Based Global History Matching for
Geological Modeling Calibration. Application to a Giant
Fluvio-Lacustrine Sandstones Reservoir Field”. Paper
SPE 94815 presented at SPE Latin American and
Caribbean Petroleum Engineering Conference
(LACPEC) held in Río de Janeiro, Brazil (June 2005).
6. A. S. Emanuel y W. J. Milliken: “History Matching
Finite Difference Models with 3D Streamlines”; paper

4
 Fig.1 Two-Step Solution with Streamlines
Taken from: Using Streamline-Based Simulation to Proactively Manage Well-Pairs to Promote Improved Sweep
and Reduce Fluid Cycling. Presented @ 2006 SPE ATW.

Fig. 2 Puesto Hernández Field Location, Argentina, Neuquén Basin

Taken from: Rocas Reservorio: Los Reservorios de la Formación Rayoso, IAPG Mar del Plata 2002

5
 Fig. 3 Stratigraphic Column, Rincón de los Sauces Area, Neuquén Basin, Argentina.
Taken from: Rocas Reservorio: Los Reservorios de la Formación Rayoso, IAPG Mar del Plata 2002

Fig. 4 Sectors of Block 4

6
Fig. 5 Production and Injection History, Fm. Rayoso

Fig. 6 Puesto Hernández Field and Simulation Area

7
Fig. 7 Matching of Produced Fluids, Block 4 Fig. 8 Matching of Produced Fluids,
Sector 2 to 4

Fig. 9 Matching of Produced Fluids, Fig. 10 Matching of Produced Fluids, by Well

Key Wells of Sector 2 to 4

8
 Fig. 11 Matching of Injectivity Logs

Fig. 12 Matching of Static Pressure

Fig. 13 Matching of Water Saturation to the Well Termination

9
 Fig. 14 Actual Scenario, IHU
Flowlines from Injectors, 10 years, with Actual Oil Saturation

Fig. 15 Actual Scenario, MHU

Flowlines from Injectors, 10 years, with Actual Oil Saturation

10
Fig. 16 Absolute Permeability Distribution per Cycle

Fig. 17 Actual Scenario – Poor Swept Zones

K Variations and Perforation Amplitude

11
 Fig. 18 Time Variation of Productivity Indexes

Fig. 19 First Optimization Scenario, MHU

Flowlines from Injectors, 10 years, with Actual Oil Saturation

12
Fig. 20 Volumetric Efficiencies from Flowlines

13