1. Field of the Invention
This invention relates to geological structural modeling of subsurface rock formations, particularly subsurface formation comprising caves (i.e., karst systems).
2. Background Art
While hydrocarbon reservoirs are more often found in porous rocks, karst carbonate reservoirs are important in certain regions in the world, such as China, Middle East, and Russia. For example, the Tarim basin in China contains many oil-rich carbonate fields, in which the oil is contained in caves that formed after dissolution of carbonate rocks by water. The Tarim field is one of the five giant Chinese oilfields with the significant potential for oil production growth in the coming decades.
Unlike conventional porous rock reservoirs, karst systems comprise caves connected with various types of conduits, such as channels and fractures (see
Due to open channel flow and variable geometries of the caves and conduits, general oil-water contact (OWC) in the region often does not define the boundary of the oil and water. As shown in
Therefore, dynamic simulation of fluid flows that have been developed for conventional reservoirs based on porous rock models cannot be directly applied to karst systems. In order to simulate flow flows in karst systems, one would need new types of dynamic simulators.
Karst systems have been studied for some time by geologists, speleologists, hydrologists. Many publications covering geomorphological studies, karst creation mechanisms, and bi-phasic flow considerations exist, such as P. Popov et al., “multiscale modeling and simulations of flows in naturally fractured karst reservoirs,” Communications in Computer Physics, 2009, Vol. 6, No. 1, pp. 162-184; and Xiaolong Peng et al., “A New Darcy-Stokes Model for Cavity-Fractured Reservoirs,” 2007, SPE 106751. These publications often concern single-phase flow models. In addition, many fresh water reservoirs have been exploited for potable water. Such water reservoirs are often contained in karst systems. The dynamic simulation of these water reservoirs involves two phases: water and air.
Although dynamic simulators for karst systems are known, there remains a need for better dynamic simulators for karst systems, including simulators for karst systems containing more than two phases.
One aspect of the invention relates to multi-phasic dynamic reservoir simulators.
A multi-phasic dynamic reservoir simulator in accordance with one embodiment of the invention includes a reservoir model for a karst system comprising: a plurality of caves connected via at least one conduit, wherein the plurality of caves and the at least one conduit are filled with at least two types of fluids, an exit point for a fluid to leave the karst system and at least one entry point for a fluid to enter the karst system from a surrounding rock matrix, and a set of parameters defining volumes and distributions of the at least two types of fluids in the plurality of caves and the at least one conduit; and a program having instructions for causing a processor to simulate fluid flows in the reservoir model for the karst system.
Another aspect of the invention relates to methods for simulating fluid behavior in a karst system using a multi-phasic dynamic reservoir simulator. A method in accordance with one embodiment of the invention includes the steps of: constructing a reservoir model for the karst system, wherein the reservoir model comprises: a plurality of caves connected via at least one conduit, wherein the plurality of caves and the at least one conduit are filled with at least two types of fluids, an exit point for a fluid to leave the karst system and at least one entry point for a fluid to enter the karst system from a surrounding rock matrix, and a set of parameters defining volumes and distributions of the at least two types of fluids in the plurality of caves and the at least one conduit; and simulating fluid flows in the reservoir model for the karst system.
Other aspects and advantages of the invention will be apparent from the following description and the appended claims.
Embodiments of the invention relate to multi-phasic dynamic reservoir numerical simulator systems for simulating fluid flows in karst systems. The simulator systems are capable of simulating dynamic fluid flows in the karst systems. In accordance with embodiments of the invention, the dynamic simulator systems may be designed to handle multi-phasic fluid flows, i.e., fluid flows including two, three, or more phases (e.g. water, oil, and gas). “Fluid” as used herein has its common meaning, i.e., a fluid can be a liquid or a gas. “Muti-phasic fluids” indicates fluids having at least two phases. Two non-mixable liquids would have two different phases.
In oilfield operations, it is sometimes necessary to deal with more than three phases, for example in the case of CO2 injection in a deep karst oil reservoir, such as those encountered in Northwest China. In these situations, CO2 may take two forms: gaseous or super-critical. In addition, CO2 may partially dissolve in oil and/or water. Therefore, having simulators capable of handling multi-phases can be useful.
The dynamic simulators in accordance with embodiments of the invention may be applied to simulate or predict the production behaviors of karst reservoirs. In such applications, the dynamic simulators may be used to predict the fluid movements in the reservoirs and to suggest optimal well placements for better production. In addition, these dynamic simulators may be used to suggest how to enhance the production, for example, by gas injection.
In addition, one may use such dynamic simulators to solve the inverse problem—i.e., to simulate the actual production from wells to arrive at better understanding of the underground cave systems. For example, these simulators may be used to explain the observed production rates for water, oil, and gas, as a function of time. In solving the inverse problems, the karst model parameters may be varied in order to try to reproduce a given production curve. The parameters that can be varied may include, for example, the karst system geometry, the distribution of fluids, or other parameters. Therefore, simulators in accordance with embodiments of the invention can help an operator understand and interpret well testing and production results.
Furthermore, karst system dynamic simulators of the invention are capable of handling a very detailed and complex geomorphology of caves. Such simulators can also be used to predict recovery factors for hydrocarbons in karst systems. For example, as shown in
In order to produce the trapped hydrocarbons, enhanced oil recovery (EOR), for example by gas injection, may be performed. In accordance with embodiments of the invention, EOR operations can be simulated with a multi-phasic system before the injection well is drilled. The simulation can be used to optimize the efficiency of EOR. For example, an injection well, Well C in
As illustrated in
In accordance with embodiments of the invention, the simulation may include complex well trajectories and cave systems. For example one production well may be designed to be connected with several caves that are produced commingled.
As noted above, the initial distribution of various phases in a karst system may not be entirely correlated with the depth. For example, water may be trapped in higher caves and oil may be trapped in lower caves—i.e., both oil and water may be found at different depth levels in the system, regardless of the general OWC. The reservoirs may have been charged with oil from surrounding source rock above, below and laterally. While gravity would force most hydrocarbons to move upward in the connected carves and conduits, some carves may have ceilings that can trap hydrocarbons. That situation can even be more pronounced after the reservoir has been produced for some time. The level of water in the cave system goes up with time, replacing the produced oil, but oil can be left at lower levels trapped in cave ceilings or siphons.
The simulator can have various levels of complexities depending on the karst system models, and depending on the physical phenomena. In accordance with embodiments of the invention, a simple system may simulate multi-phasic flows in a model containing a series of tanks connected with conduits. Even with such a simple model, one can reproduce very complex production behaviors. Such a simulator may be adequate for optimizing production parameters in actual karst reservoir development.
In accordance with some embodiments of the invention, more complex karst simulation systems may take into account various factors, such as fractures, pipe-type conduits, permeable layers, and the slow exchange with a tight rock matrix surrounding the caves. Such complex systems may contain a wide range of time constants. Such a simulator can help understand and interpret well testing and production results.
The examples illustrated in
Furthermore, one skilled in the art would realize that the above simplified examples may be used as building blocks to construct more elaborate models that mimic actual cave systems in the subterranean formations.
For example,
The flows of oil, water, and free gas in a cave system are controlled by the geometry of the cave system and by the pressure inside the cave system and in the production well. Therefore, a simulator of the invention would take these factors into account.
Karst volumes may have fractal-like shapes that are extremely complex, and caves that may have been free of rocks debris initially may be filled with blocks of rocks, with inter-space filled with rubble and pebbles, and cave floor may be covered by sediments, such as clay or carbonate mud. One skilled in the art would know that various techniques, such as seismic, may be used to determine subsurface cave geometries and shapes. Use of seismic data to validate cave model will be discuss in a later section with reference to
In accordance with embodiments of the invention, the models for simulation can be 2-D or 3-D. Almost all systems can be modeled with reasonable accuracy as 2-D systems. However, because most commercial reservoir simulators are designed for 3-D simulations, 3-D representations of karst systems may be more easily adapted to use such commercial systems. For clarity of illustration, the examples in this description are 2-D models, as shown in
To represent the 3D cave system of
As shown in
On the other hand, chambers C2 and C3 in this example communicate through a conduit located close to the bottom of the rock wall between the two chambers. Thus, the connection between C2 and C3 is a U-tube connection. Due to buoyancy and fluid segregation, the “U-tube” conduit would allow water to flow first between chambers C2 and C3. In this example, the production well PW is connected directly to chamber C3 or indirectly through a fracture in the rock, for example.
The volumes of oil and water in chamber C1, C2, and C3 in
In accordance with embodiments of the invention, a karst simulation model would take into account key elements, such as cave chambers C1, C2, C3, conduits U-tubes and inverted U-tubes, and aquifer entry points. These basic elements form key building blocks of the cave system model. Other hydraulic elements, or building blocks, may be designed and used in the models, if needed. Such other elements may include, for example, the “water trap” and “oil trap” elements, shown in
As noted above, the flows of fluids in the karst systems are mostly governed by gravity or buoyancy. The oil trap or water trap in the systems are typically caves where one type of fluid (water or oil) is temporarily removed from the main flow stream (or temporarily stored or trapped in the caves) due to the effect of cave geometry and gravity.
As illustrated above, all caves and traps in the models are linked by conduits, which are the simplest elements in the models. The conduits may correspond to natural fractures, vuggy layers, or karst conduits that may be the remains of ancient water flow paths.
In accordance with embodiments of the invention, other elements of a model may include one or more “flow splitter” elements, as shown in
In the example of
3-D flow splitters are common in real cave systems. However, it is not always easy to design and represent such 3-D flow splitters using 2-D models.
The above describes examples of various elements and parts, as well as their functions, that may be used in simulation models to represent a karst system. The following will show some examples of how these parts that may be used to simulate karst systems.
In accordance with embodiments of the invention, the karst system models may be presented in various graphical representations that can serve the same purpose. For example,
In an alternative representation, the same karst system shown in
In this example, the oil trap OT1 contains an initial oil volume Voi1 and an initial water volume Vwi1. It is connected to cave chamber CH3 through a conduit CO1 of length L1. The production well PW1 produces fluids from cave chamber CH3.
Another example of a cave system model representation is shown in
In the middle of
The above description shows that a model of the invention may be represented in graphical models or symbolic representations. In addition, the various parameters may be included in the models. Based on these models, the fluid movements in the system may be simulated with various approaches.
As noted earlier, when one volume of fluids is removed (produced) form the karst system, an equal volume of water form an aquifer outside the karst system being modeled is allowed to come in to replenish the removed volume. Based on this principle, simple approaches may be implemented to simulate fluid movements in the model. For example, the following will describe an example implementing a very simple Cellular Automata (CA) 2-D simulator for multi-phase flow dynamic simulation of cave systems. This implementation has been successfully implemented and tested with a prototype software using Visual BASIC on Excel. This example is limited to two-phase flows. However, this can be easily extended to three-phase flows. Similarly, the description is for a 2-D simulator. However, one skilled in the art can easily extend this to a full 3-D system.
As an example, the space (a 2-D Euclidian plane in this example) where the fluid movement is modeled may be divided into small cells, i.e., “tiled” using a square grid, as shown in
Cells may be color coded according to their contents or functions. For example, solid cells (S) representing non-permeable rock may be in grey color; oil cells (O) may be in black; water cells (W) may be in blue; and free gas cells (G) may be in white. Alternatively, these may be shown with different patterns or in different gray scales, as illustrated in
Some examples of functional elements have been described above. These include the production well (PW) shown in
These square cells may be symbolic representations of reality. For example, the actual vertical size Δz and horizontal size Δx of the cell may not be equal. However, for simplicity and clarity, these cells are shown as squares in the example.
Having defined the space as 2D cells, one would next define the rules of how fluids (cells) move in the models. By definition, solid cells and functional cells do not move because they are not fluid cells.
In CA simulators, one can consider the fluid movements in discrete time steps. At each time step Δt, the fluid cells can “move” to or exchange content (or not) with one of their neighbor cells. One time step corresponds to a discrete jump in the vertical direction (up or down), or in the horizontal direction (right or left). That implies that the maximum velocity in the vertical direction for a cell is equal to Δz/Δt and the maximum velocity in the horizontal direction is Δx/Δt.
The relative values chosen in the model for the cell size (Δx, Δz) and the time step Δt preferably are compatible with actual fluid velocities observed in the real word. By defining CA rules for cell moves according to real world expectations, the simulation outcomes may be more relevant. For example, what is the vertical velocity of a gas bubble moving in water under the sole influence of buoyancy, as compared to the velocity of an oil droplet of same diameter?
It can be shown that the limit velocity of a bubble moving up in water is directly proportional to the difference between water density dw and the bubble fluid density df. Therefore, the ratio of the velocities for a gas bubble and an oil droplet of same size is simply: R=(dw−dg)/(dw−do). Under atmospheric conditions, this is roughly equal to 5. Under pressure conditions of a deep oil reservoir, e.g. several hundred bars, the density of a free gas (such as methane or nitrogen) is several hundred times its density at 1 bar. For example, at 400 bars the ratio R can only be 3 instead of 5.
The CA simulator can easily account for this by giving a treatment of favor to gas cells over oil cells. One example of algorithm that achieves the objective is as follows: At each time step, all oil and gas cells are considered for a move. If the cell is a gas cell, it is always considered for a move. If the cell is an oil cell, a random number X is drawn with a uniform distribution between 0 and 1, and if X is between 0 and 1/R the cell is considered for a move. Otherwise, it is not. In other words, the probability for any oil cell to be considered for a move at each time step is R times lower than for a gas cell. Therefore, its maximum velocity is automatically R times slower than that for a gas cell.
The above is only one example. One skilled in the art would appreciate that other rules may be adopted to take care of the different probabilities of a move based on the relative move propensity factor R. This kind of rule is called a conditional rule. It is in fact a conditional or probabilistic rule. The conditions applied may be made much more complicated to account for more complex situations.
CA rules for cell moves may be defined for all situations. The following describes an example of a set of rules for a two-phase 2-D CA simulator, e.g. for oil and water. However, one skilled in the art would know that this can be easily modified to include more than two phases.
In a closed system, i.e. a cave system that has no fluid exchanges (i.e., no production and no water entry), the oil and water phases segregate quickly from their initial position to a stable equilibrium with oil at the top level and water below, with a horizontal oil/water contact (OWC) separating the two fluids.
In such a closed system, only the oil cells need to be considered for moves at each time step. In reality, the water cells are moving too, to occupy the spaces left behind after oil cell moves. However, there is no need to explicitly deal with the moves for water cells.
As an example, one set of rules for oil cell movements in a closed system may be as follows:
These two rules can be symbolized as shown in
Rule 1 is obvious, and Rule 2 is easy to understand: Cells other than water cells are oil cells, solid cells, or functional cells. Obviously, oil cannot go into a solid cell, and if oil swaps with an oil cell the result is the same as if the oil cells do not swap. Finally, in a closed system oil cannot go into any functional cell.
While the above model assumes an oil cell either moves or does not move, i.e., a 0 or 1 event, one can also expand the model to include probabilistic moves. For example, one can introduce a probability of move pm (having a value from 0 to 1) as a parameter of the model for situations where a cell can move in different directions, or stay put. The outcome of the rules is probabilistic.
For example, referring to
In the real world, oil cells have the sizes of oil molecules and the oil water contact appears perfectly still to the naked eye. However, in reality the oil molecules actually perform rapid and complex random walks, similar to what the CA simulation shows with “big” cells.
The above description assumes a closed system, in which no cells (fluid) move into or out of the model. However, in a system with a production well, cells (fluid) reaching the producing well (PW) are produced one at a time (one per time step) and exit the cave system. When fluid exits the cave volume, it should be replaced by the same amount of fluid to avoid leaving an empty space behind. Therefore, for each fluid cell exiting through PW, one fluid cell would come in from a source outside the karst system (e.g., one of the aquifer entry points).
In accordance with embodiments of the invention, a simulator may pick randomly a producing aquifer entry point at each time step. On average, each aquifer entry point may contribute equally over time. If a model requires one of the caves to produce more fluids than other caves in the system (for example to match some real production data), several aquifer entry points can be placed in that particular cave (i.e., the higher production cave). In other words, the number of aquifer entry points may be adjusted in order to reproduce the flow behavior observed in a real well. Another way to allow aquifer entry points to contribute differently is to assign a water production flow capacity (e.g., relative production capacity or probability) to each aquifer entry point.
A cell exiting the cave system at PW at time step t would be replaced by the content of a neighbor fluid cell, which itself is replaced by one of its neighbor fluid cells, etc. This process propagates a chain of moves between the exit point (PW) and one of the aquifer entry points (AQx), where finally a water cell enters the cave system to fill the gap created by the cell exiting at PW.
In a simple model, the fluids (oil and water) are assumed not compressible. That implies that the propagation of the gap described above all happens within one time step no matter how far the exit point PW is from the selected aquifer entry point AQx. However, the CA rules as described above are local rules: a fluid cell moves according to a set of rules that only account for neighbor cells. Thus, quickly propagating a gap from PW to AQx requires the cells to follow a path that goes through the system avoiding solid cells (e.g., the wall of the caves). This path can be convoluted and may be a function of the geometry of the entire cave system. Thus, the knowledge in a single time step of this entire path is not local. Even though this path may be convoluted, one can still simulate this path based on the simple rules described above, assigning probabilities to different alternative cells for each move step. However, an easier alternative is described next.
To overcome this difficulty (non-local) and to keep the simulator simple and fast to run on a simple computer system, such as a regular laptop PC (e.g., using a Visual BASIC software on Excel, etc.), a special algorithm may be implemented that looks once and for all—before the simulation starts—for all the “short paths” between the functional cell PW and all aquifer entry points AQx in the cave system. There are many ways to look for short paths between two cells in a maze.
For example, an algorithm may list all shortest paths between PW and AQ1, i.e. paths with a minimum length Lmin1, as well as all paths with lengths at most Lmin1+k, wherein the integer k is a parameter of the simulator. The number of all these “short paths” between PW and AQ1 is referred to as N1. The same is done for “short paths” between PW and all other aquifer entry points, AQ2, . . . , AQn, in the cave system. The numbers of “short paths” leading to all other aquifer entry points are N2, . . . , Nn, respectively.
This algorithm gives the simulator the non-local knowledge required to propagate the gap in one time step. A probabilistic routine may be used to do so. For example, at each time step, a fluid cell exits the system through PW, and the software draws a random integer m between 1 and n (with uniform probability distribution) to decide which Aquifer entry point AQm to use. The software then picks a random integer between 1 and Nm (for the selected aquifer) where Nm is the number of “short paths” between PW and the selected aquifer, AQm. The selected short path is a chain of grid cells that all contain fluids and is used to propagate in one time step the gap between PW and AQm. The software then runs the normal moving rules as defined for a closed system by checking all non-water fluid cells one by one. This completes the cycle for one time step, and time is incremented to the next time step and a new cycle begins.
The approach described above may be summarized in the diagram shown in
If it is an open system, the program would run a “short path” subroutine first to identify the “short paths” for moving the fluid from one or more aquifers to the production well. Then, the open system routine is used.
There are many possible variations in the implementation of such software, including, for example, different ending criteria (other than a maximum time). In addition, the CA rule represents one of many approaches that can be used to model the fluid movements. While the CA rule approach is simple and easy to implement, any other suitable approaches may be used without departing form the scope of the invention.
A karst system simulator according to embodiments of the invention may be used in various situations, including forward simulation (e.g., prediction of production curves based on cave systems, which may have been determined using other techniques, such as seismic techniques) and reverse simulation (e.g., modeling to fit actual production data in order to arrive at possible cave structures). The following will describe some examples to illustrate embodiments of the invention.
A simulator described above is capable of predicting or reproducing a “water cut” curve (i.e., the water volume fraction in the fluid produced versus time) observed in a well producing oil, gas, and water from a karst system. However, in a simple implementation of the simulator, it would not include pressure information in the cave system, in the aquifer, and along the producing well. Therefore, it cannot calculate the actual volume flow rate output of the cave system.
If desired, this simple system can be improved by applying a pressure calculation routine based on the cave system parameters (fluid properties, cave volume, initial reservoir pressure, etc.) and based on the exchange parameters between the cave system and the aquifer and eventually the surrounding rock matrix. Such algorithm and software are commonly available in the oil and gas industry (e.g. ECLIPSE). Each time a fluid volume is produced through PW, the effect is a small reduction of the reservoir pressure. The magnitude of this reduction depends on the system parameters: it can be very small—hardly noticeable in fact—for large systems with compressible fluids, e.g. oil containing significant amount of dissolved gas. The simulator may also account for the pressure along the producing well that depends on the type of completion in place, e.g. artificial lift system or not, chokes and nozzles, etc. The results of these simulations may include decline curves versus time for the reservoir pressures and for the produced flow rates.
This curve is also a function of the water cut produced. If the well is producing under the sole effect of the reservoir pressure, the hydrostatic pressure at the base of the well PW is directly proportional to the average fluid density in the well over the entire well column. This density is a linear function of the water cut in the well. The higher the water cut the higher the fluid density and the lower the natural production rate. All these effects can easily be accounted for in the software to produce a decline curve for the total fluid volume flow rate produced versus time Q(t).
The dotted arrow in the “Open System Routine” in
The combination of the total fluid flow rate curve Q(t) and the water cut curve given by the CA simulator allows one to immediately calculate both the oil and water flow rates versus time. Utility of embodiments of the invention in such applications will be further illustrated in the following case studies.
In this example, a well produced a mix of oil and water with a water cut curve shown in
The production decline curve shown in
This type of water cut curve shown in
Based on this water cut curve, four main phases are considered in building a cave system for this case:
Obviously, the production well is connected to a cave chamber that contains sufficient dry oil for 120 days of oil production. The total volume of oil produced during this phase (based on the production data) is 56,540 m3. That volume gives an indication of the free volume occupied by oil in this cave chamber. The actual volume of oil initially in the chamber may be larger, but not all of it can be produced. This cannot be assessed from the production data; it depends entirely on the cave chamber geometry which is unknown.
In the model for phase 1, the oil produced at the top of the cave is replaced by water entering the cave at the bottom. Once the oil level reaches the chamber ceiling, the water cut would sharply increase to 100%. However, the water cut value observed is 80%, and it is stable for 250 days. That implies that one oil cell is produced every four water cells. The cave system model needs to be designed accordingly.
One simple way to achieve this is to consider oil arriving in CH1 from one conduit, and water arriving in CH1 from another conduit with a flow rate four times the oil flow rate. A simple way this can be achieved is with a cave system model shown in
In this model, the five aquifer entry points will produce water at the same average rate. Each time AQ1 is producing a water cell, it will push oil from chamber CH3 into chamber CH2 through the inverted U-tube at the top, and the bottom oil/water contact in chamber CH2 will gradually go down until reaching the U-tube to enter chamber CH1. The volume V′ in chamber CH2 therefore is approximately equal to one fifth of the producible oil volume in CH1, i.e. V′=Vpo1/5=56,540/5=11,300 m3. With this configuration, the water cut goes to 80% when the oil initially in CH1 has been produced.
The 80% water cut plateau lasts 250 days, and in that plateau phase, the total amount of oil produced (based on the production data) is 20,300 m3. That volume corresponds to the volume Vpo3 of producible oil in chamber CH3 minus the volume V′. Therefore, Vpo3=20,300+V′=31,600 m3.
The volume of oil in CH2 cannot be determined from the production data. However, this is not important because the volume of CH2 is irrelevant, except for V′.
One interesting question is: why would the water cut increase gradually to 100% after the volume of producible oil in CH3 is produced? One would think the increase should be sudden. This is where a simulator of the invention can provide a significant contribution. The moves of oil and water cells along the complex path CH3→inverted U-tube→CH2→U-tube→CH1 and the random mixing with the water flow coming from AQ2/3/4/5 actually translate in a gradual increase of the water cut from 80% to 100%.
In fact, the water cut is actually erratic in the production well PW. When an oil cell is produced, the water cut is 0%, and when a water cell is produced, the water cut is 100%. Therefore, the instantaneous water cut in the simulator is either 0% or 100%. In order to calculate the water cut provided in the production data the simulator may take an average over a certain number (e.g., 20) of time steps (i.e. 20 cells produced). One consequence of this averaging is that the increase of the water cut from 0% to 80% after 120 days is not as sharp as in the actual well. To avoid that problem, one can reduce the cell size in the simulator, and/or reduce the time step accordingly. However, this comes at the expense of an increased running time for the software.
Large cell sizes allow for completion of a 1000-day simulation in two minutes on a laptop computer with a 2-D model that has less than 2000 fluid cells. In this Case Study 1 example, the cell size is approximately 220 m3. That is quite large. Because the total fluid rate is on the order of 220 to 500 m3/day, the corresponding time step is half-a-day to one day.
Despite the averaging over 40 time steps, the water cut curve obtained with the simulator remains quite erratic due to the random nature of the simulation of cell moves. To overcome that problem, one can do many simulation runs and take the average of all the water cut curves obtained. The actual curves presented at the end of this section are the average of 25 simulation runs.
In phase 4, new oil reaches CH1 around 580 days after production started. This oil comes from another cave located far away from the main cave CH1/CH2/CH3. In a 2-D model, one can choose to design this arrival from the left or from the right; it does not matter. This reflects the symbolic nature of the graphical representation. One way to delay the arrival of oil from this second cave to CH1 is to add an oil trap on the way. The oil trap serves to delay the arrival of oil. One example of the model is a cave system shown in
In
It is interesting to note that even in cave systems where free flow dominates, recovery factors can be quite poor. The recovery factor depends dramatically on the geometry of the caves. Oil is trapped in pockets located at the ceiling of the structures. Therefore, it is possible to have cave systems where 90% or more of the oil is non-producible under natural flow.
In these situations, enhanced recovery can be achieved, for example, by injecting gas (such as nitrogen or CO2) into the cave systems. Due to buoyancy, the injected gas will replace the oil trapped in ceiling pockets and push it towards the producing well. Gas injectors with optimized locations may allow for recovery of most oil in a cave system.
In the example of Case Study 1, CH4 is an obvious location for a gas injector. All the oil remaining in the cave CH4 and all the oil in the oil trap OT1 may be recovered by gas injection into CH4. However, the oil remaining in the cave chambers CH2/CH3 may not be recovered with the same gas injector. In deep karst reservoirs, nitrogen may be injected in the form of water/nitrogen foam in order to have a sufficient density of the injected fluid to limit the surface pressure to safe values.
A dynamic simulation of nitrogen foam injection can be done with the CA simulator described here to predict the enhanced recovery.
Case study 1 had a well starting with dry oil production and increasing water cuts, followed by decreased water cut. Case study 2 is the opposite; production starts with a high water content, and then water cut is reduced. The water cut increases again a couple years after the production start time. A cave system model for Case Study 2 is shown in
This model is very similar to the model shown in
The Case Study 2 is based on the production data shown in
The decline curve shown in
This type of production behavior is almost never seen in conventional (i.e., non-cave system) reservoirs, where water production generally tends to increase over time and would not decrease, as shown in
Phase 1 is a clear indication that the production well is connected to a cave chamber CH1 that is full of water. This cave is likely to be at a lower level (perhaps lower than the general local oil-water-contact interface) than other nearby cave chambers containing oil.
Phase 3 has the lowest water cut production of all phases: 20%. Assuming the oil comes from a second oil cave that is connected to the water cave CH1, there is a ratio of 4:1 between the number of aquifer entry points in the oil cave and the other aquifer entry points in CH1 bringing water. This ratio is required to reach 20% water cut. Therefore, an aquifer entry point AQ1 is placed in CH1 and four aquifer entry points AQ2,3,4,5 are placed in the oil cave.
Although the water cuts in phases 1 and 3 are easier to understand, it is a bit more complicated in phase 2. Specifically, the question is why the water cut seems to have a small plateau around 50% in phase 2. This can be accounted for by including an oil trap and a flow splitter explained below.
When oil from the oil caves starts to reach CH1, it comes at a flow rate that is four times the water flow rate entering CH1 through AQ1. In order to have a plateau around 50%, it is necessary to reduce the oil flow rate reaching the production well PW. This can be achieved by splitting the oil flow with a flow splitter element OS1 and trapping 75% of the oil flow into an oil trap TR1.
When the oil trap is full, all the oil flow can then reach the production well PW, and the water cut stabilizes at a 20% water cut plateau, as observed in phase 3.
Similarly, it is necessary to include a water trap in the second cave, or on the way between the second cave CH2 and the first cave CH1, in order to explain the 40% water cut plateau in phase 4. When the water produced from the four aquifer entry points AQ2,3,4,5 reaches the top level of the cave chamber CH3, it goes to a water flow splitter WS1 and about 75% of the water goes into the trap TR2. The other 25% of the water, as well as the oil released from the trap TR2, flow to the production well PW. This is equivalent to having two aquifer entry points bring water to PW, and the other three aquifer entry points pushing oil out of trap TR2 to the well PW, hence the 40% water cut plateau.
Finally, when the trap TR2 is full of water, all the water is directed to PW, and the water cut gradually increases to 100% over time. It can take a long time, e.g. several years, before the mixed oil and water flow becomes a pure water flow.
As noted above, seismic techniques may be used to map cave structures underground. These techniques can be very useful in helping to build cave systems.
Note that the model in Case Study 2 (shown in
However, seismic survey reveals an additional cave, Cave 4, which is not present in the model derived from the production data. Most likely, this is due to the fact that Cave 4 is not connected to the production well 1, and, therefore, does not contribute to the production data. If Cave 4 does not contribute to the production data, then it would not appear in the model that is based on the production data.
While seismic techniques are useful in mapping cave structures, it does not have sufficient resolution to visualize details, such as smaller caves or conduits connecting the caves. Such smaller structures can be inferred from production data or determined using other well logging techniques, such as well testing or interference test data. The integration of seismic data, as well as well testing and interference test data, is a very useful step to validate the cave system model.
A comparison between the two water cut curves obtained with the CA simulator for the model shown in
While the examples above illustrate simulation of two-phase systems (water and oil), embodiments of the invention can be used to simulate multiphase karst systems (i.e., including 2 or more phases, such as water, oil, and gas). The following example illustrates the rules that can be used to simulate 3-phase flow with a CA simulator.
As noted above in the section entitled, “CA Rules for Cell Moves,” free gas cells move up faster than oil cells due to higher buoyancy. In accordance with embodiments of the invention, the relative moving velocities of gas and fluid can be adjusted. A typical velocity ratio in high pressure reservoirs is about 3:1 (gas:liquid). In such a three-phase CA simulator (water, oil, and free gas), one should replace the box “Apply CA move rules for each cell” in
Embodiments of the invention for the simulation of karst systems can be implemented with any suitable software, including but not limited to ECLIPE (from Schlumberger Corporation, Houston, Tex.), which contains a suite of software packages for various reservoir simulations. The following example illustrates a simulation using ECLIPSE.
As noted above, fluid flows in cave systems (karst systems) are more like free flows. Free flow can be simulated using standard ECLIPSE software (using the VE module). The example shown in
As noted above, methods of the invention may be used to model a karst system based on production data.
The method next analyzes the water cut curve versus time and other production data to build a preliminary cave system model (step 344). Then, the method runs simulations with a karst system simulator of the invention and modify cave system model until reaching a good match between simulated and measured production data (step 346).
If seismic data is available and can resolve caves, compare the cave system model obtained in step 346 with seismic data and make additional adjustments if necessary (step 348).
As illustrated above, a multi-phasic dynamic simulator for a karst system of the invention may be implemented with any suitable computation systems, including a personal computer (see e.g.,
As shown in
Alternatively, the multi-phasic dynamic simulator for kart systems may be implemented as part of a software or systems that are designed for oil and gas industry, such as the ECLIPSE system from Schlumberger Technology Corporation (Houston, Tex., USA).
In addition, some embodiments of the invention may relate to a computer readable media that stores a program having instructions to cause a processor to execute steps for implementing one or more methods of the invention. Such computer readable devices, for example, may include a hard drive, a floppy disk, a CD, a DVD, a tape, etc.
Advantages of embodiments of the invention may include one or more of the following. A multi-phasic dynamic simulator of the invention can be used to model and predict flow behaviors and production data in cave systems. A model of a karst system in accordance with embodiments of the invention can be represented with simple chambers (caves) with conduits (channels or fractures) and various traps and/or flow splitters. These simple models are easier to visualize the cave systems and yet they can predict flow behaviors in the cave systems with accuracy. A multi-phasic dynamic simulator of the invention can also be used in a reverse manner to aid the understanding of a cave system based on actual production date. Embodiments of the invention are simple to implement, and yet they can produce accuracy simulation results.
While the invention has been described with respect to a limited number of embodiments, those skilled in the art, having the benefit of this disclosure, will appreciate that other embodiments can be devised which do not depart from the scope of the invention as disclosed herein. Accordingly, the scope of the invention should be limited only by the attached claims.
This claims benefit of provisional application Ser. No. 61/348,014, filed on May 25, 2010, which is incorporated by reference in its entirety.
Number | Date | Country | |
---|---|---|---|
61348014 | May 2010 | US |