METHOD FOR DETERMINING DRAIN CONFIGURATIONS OF

WELLS IN A FIELD

Technical Field [0001] The present disclosure relates to the blanking and splitting design of drains for injection or productions operations at wells in a field, and more specifically in a field containing a hydrocarbon reservoir.

[0002] The positioning of wells is a critical task in the production of a field containing a hydrocarbon reservoir. Indeed, the respective positions of producer wells and/or injector wells may greatly affect the productivity of the field and the volume of hydrocarbon recovered, and hence its profitability.

[0003] The reservoir generally contains at least a first fluid to be produced, and potentially other auxiliary fluids to be produced along with the first fluid. A third fluid and/or a fourth fluid are advantageously used to be injected in the reservoir to drive the production of the first and/or of the second fluid.

[0004] For example, the first fluid is oil or gas, the second fluid being the other within gas or oil. The third fluid and/or fourth fluid are generally water, gas, and/or oil. The first fluid and the second fluid are preferentially hydrocarbons.

[0005] After the choice of positioning of wells, a drill may be performed to and through the reservoir for each well. The part of a well crossing the reservoir may be used to perform the drainage of the reservoir, i.e. the extraction fluid from the reservoir and/or injection of fluid into the reservoir. Indeed, the well drain, drain or “drain portion” may constitute the location on the well crossing the reservoir where injection operations, for instance water and/or gas, or productions operations, for instance oil or gas, are performed. Thus, a drain or portion drain may be constituted only by a small interval along of the part of the well crossing the reservoir.

[0006] The reservoir may comprise several regions, for example at least an aquifer, an oil leg, and a gas cap. An aquifer is generally delimited upwards by a water oil contact or “WOC”. An oil leg is delimited between a water oil contact and a gas oil contact or “GOC”. The gas cap is located above the gas oil contact. [0007] By “production operation”, it is meant operation performed at a well called “producer well”, in which a desired fluid, i.e. the first fluid and/or the second fluid, is produced. Hence, producer wells aim at the extraction of the desired fluid by the intermediary of the drain of the well producer. [0008] By “injection operations”, it is meant operation performed at a well called injector well, in which fluids, i.e. the third fluid and/or the fourth fluid, are injected rather than produced. Injector wells, performed at the drain of the well, aim at maintaining reservoir pressure and substituting one fluid by another in the reservoir thus enhancing the production of the desired fluid at the producer wells. [0009] Usually, a numerical gridded model of the field is generated to express the properties of the reservoir contained in the field, including geology, infrastructure, and fluid properties. Each property (e.g. geology or fluid properties) of the reservoir may be associated at respective uncertainties. Thus, it is usually common that the uncertainties of the reservoir properties in a numerical gridded model of a field are expressed by a set of model realizations. Those realizations may be associated, equally likely, or characterized by a weight or a probability. For instance, the use of a variability of properties of the model across realizations and the space may capture a priori knowledge of relations between properties and across the space. [0010] Based on this model derived from raw field data, a team of scientists (e.g. composed of reservoir engineers and/or reservoir geologists) usually determines the best potential locations for wells (producers and/or injectors), as well as the design of drain of each well (i.e. location of the openings or blanking of the drain and/or the location of the packer or splitting of the drain), based on experience, and taking into account the constraints which exist in the field, such as distance to surface well head clusters or platforms, the subsoil properties, the distance of the closest water source, etc. Key design parameters include spacing between wells or drains, well drain length, well or drain configurations/designs, and the kind of performed operations at each well. [0011] However, such process is time consuming and requires significant human effort and skill. Furthermore, there is no guarantee that the proposed solution is optimal. [0012] Software products have been developed to help optimizing well configurations relative to the reservoir. However, these software products do not usually allow to carry out a full optimization design of wells (i.e. locations of the wells and designs of the associated drains), and still require the intervention of the team scientist to complete the process, especially at the drain design level.

[0013] There is thus a need for a method for determining in an efficiency way a design of well drains in a field which does not require too much human or computational resources and which give reliable results for improving productivity.

[0014] The present disclosure relates to a method implemented by computer means for determining drain configurations of wells in a field containing a hydrocarbon reservoir, said method comprising:

/a/ receiving a first set of geological gridded models,

- each gridded model among the first set of geological gridded models comprising a respective plurality of cells; - wherein each geological gridded model comprising a drain associated with a reference drain configuration (910), said reference drain configuration comprising N respective cells within the respective plurality of cells;

/b/ receiving a set of criteria; Id for each geological gridded model among the first set of geological gridded models:

- determining possible configurations for the N respective cells,

- for each possible configuration among the possible configurations and for each cell among said N respective cells, computing criterion measures, each criterion measure representing a suitability to a respective criterion among the received set of criteria; determining, based on said criterion measures, a set of non- dominated configurations among said possible configurations via a non-dominated sorting algorithm for jointly optimizing the set of criteria;

- determining a first configuration among the set of non-dominated configurations, said first configuration minimizing a distance between the reference drain configuration respectively associated to the geological gridded model and the set of non-dominated configurations;

- determining intermediate weights based on coordinates of the first configuration in a space defined by the set of criteria, each intermediate weight being associated to a respective criterion among the plurality of criteria;

161 estimating a plurality of weights, each weight being associated to a respective criterion among the plurality of criteria and estimated based on intermediate weights among the pluralities of intermediate weights associated to said respective criterion; lei receiving a second geological gridded model of the field,

- the second geological gridded model comprising a plurality of cells,

- and comprising a second drain, said second drain comprising N ₂ cells within the plurality of cells;

IV determining a configuration for said N ₂ cells based on the received set of criteria and the plurality of estimated weights.

[0015] By “reference drain configuration” it is meant a drain configuration received (i.e. not obtained by applying the method of determination of drain configuration) and used for estimating some parameters. For instance, a reference drain configuration may be determined by an expert. By estimating parameters from reference drain configurations, it may then be possible to obtain automatically new drain configurations built according to a “logic of construction” similar to that used to build the reference drain configurations.

[0016] It is noted that a “set” of elements may refer to one or more than one elements.

[0017] The “set of criteria” may correspond to a set of criterions, for which each criterion may be related to a specific constraint for determining a configuration of the cells of a drain. Indeed, the fact that a drain configuration be optimal at a location in the subsoil may depend of the physicochemical properties of the subsoil (e.g. the fluidic properties of the cells of the drain), or/and operations performed by the drain (e.g. injection or production). Therefore, the expert (geologist or reservoir engineer) may generally determine a drain configuration according to a set of criteria that must be defined in order to automate the determining method. The criterion may depend on the type of well (e.g. injector / producer), the type of operations (injection or production), the location of the others drains of wells, the physicochemical properties of sedimentary layers at the drain. [0018] Moreover, a criterion may be computed from a set of criterion specific to respective realizations by averaging, weighted averaging or rank averaging (i.e. rank of realizations or order index, see figure 5 for an example). For instance, the use of rank averaging may refer to the process by the rank of the cell relative to a criterion may be determined for every realization and the average global criterion may be computed by averaging the rank this obtained over the respective realizations.

[0019] The drain configuration may be a blanking configuration determined by a blanking design or a splitting configuration determined by a splitting design.

[0020] By “blanking configuration”, it is meant to decide to open (or not) a part of the drain of a well on the reservoir in order to perform production or injection operations. The blanking configuration may be a combination of opened or/and closed cells among the N cells of the drain.

[0021] By “splitting configuration”, it is meant to decide to split a part of the drain of a well on the reservoir in order to enable independent control of the flow over the isolated part. For instance, the isolation may be achieved by the use of a packer for instance. The splitting configuration may be a combination of insulated device between cells of the N cells of the drain.

[0022] A blanking design may correspond to determine an optimal blanking configuration aiming to the opening of the drain portion (i.e. the opening of the wall of the well in the drain portion).

[0023] A splitting design may correspond to determine an optimal splitting configuration aiming to the dividing of a portion of the drain in sub-portion, each sub-portion being isolated from each other (at least in the annular zone of the well).

[0024] By “criterion measure”, it is meant a measure associated to a respective criterion among the set of criteria, which quantifies the suitability of given configuration to this criterion. For instance, if the criterion is related to a distance to a given fluid (e.g. oil, water, gas), the respective criterion measure may be based on a mathematical distance or any measurement relative to a distance (e.g. a Time of Flight). A criterion measure may enable to rank the suitability of a configuration (e.g. blanking or splitting configuration) versus another. Each respective criterion measure may be determined from a set of computed values associated to a respective criterion, and each computed value of a respective criterion may be determined for each cell of a respective geological gridded model.

[0025] Mi criterion measures may be defined, and each M ^th criterion measure may be associated with a respective set of computed values, i being an integer superior or equal to 1. Each computed value of an M ^th respective criterion measure may be defined according to a scale attached to the respective M ^th criterion measure, and covering the set of computed values of the M ^th criterion measure.

[0026] As mentioned above, the determining of a drain configuration is subject to a set of criteria and therefore may correspond to a joint optimization problem. However, it is generally not possible to determine a configuration that optimizes all the criteria. Therefore, an appropriate solution may consist to use a non- dominated sorting algorithm (e.g. based on the Pareto dominance principle). Such algorithm may eliminate drain configuration that are less suitable than others, the remaining configurations being called “non-dominated configurations”. [0027] The “distance between the reference drain configuration respectively associated to the geological gridded model and the set of non-dominated cells” may be any mathematical tool characterizing a proximity between the two groups of drain configuration, for instance a Hausdorff distance

[0028] Each criterion may be associated to a respective weight which may reflect the importance that the expert gave to the criterion when determining drain configuration. These weights may be estimated to be used to determine new drain configurations, according to a similar construction logic.

[0029] By “intermediate weight”, it is meant a quantity determined as a calculation intermediate to estimate a weight associated with a criterion. [0030] The above method advantageously may allow to determine drain optimal configurations (i.e. blanking or splitting configurations) of wells in a field completely automatically from reference drain configurations (or “training data”) and constraints of the field.

[0031] Thus, advantageously, the automatic determination of optimal drain configuration may reduce the time and materials resources (e.g. computing equipment) consumed while allowing to obtain an optimal configuration, that a skilled man in the domain could not have identified.

[0032] Moreover, advantageously, the automatic determination of optimal drain configuration may enable to consider model uncertainty through the joint operation of the process on ensemble of model realizations.

[0033] Furthermore, advantageously, the automatic determination of optimal drain configuration may enable to consider blanking and splitting aspects of well design jointly with other aspect thus enabling to leverage inter-relations between design aspects.

[0034] In one or several embodiments, step Id lay may comprise:

- determining a plurality of intermediate weights w corresponding to a weight w _t associated to a criterion M _t among the plurality of criteria, k being an integer between 1 and a number of reference drain configuration; wherein /d/ may comprise:

- estimating the weight w _t as a function of intermediate weights w among the plurality of intermediate weights.

[0035] For instance, the function of intermediate weights may be a mean of the intermediate weights w ^J . According to another example the weights may be considered as any function of the well rank, if there is a sequence of wells/drains with a well rank. For instance, a constant function (e.g. simple averaging) or linear function (e.g. fit of weight function of the well rank in the sequence of wells pattern).

[0036] In one or several embodiments, in step Id, the determining of the possible configurations may comprise:

- receiving a first set of constraints depending on a type of operation at the drain. [0037] In such embodiments, the criterion measures used to determine the drain configuration may be differed according to the type of operation performed at the drain (e.g. injection or production). For injection operations, as example, the criterion measure may take into account the location of the others drains used for productions operations and located around the drain of which the configuration is to be determined. For production operations, according to another example, the criterion measure may take into account the location of the unwanted liquid in the subsoil around the drain of which the configuration is to be determined. The source of the unwanted liquid may be a water leg (e.g. from aquifer) or another drain used to perform injection operation for instance.

[0038] In one or several embodiments, the determining IV of the configuration may comprise:

- determining possible configurations for said N ₂ cells;

- for each configuration among the possible configurations, computing criterion measures, each criterion measure representing a suitability to a respective criterion among the received set of criteria;

- determining, based on said criterion measures, a second set of non- dominated configurations among said possible configurations via a non- dominated sorting algorithm for jointly optimizing the set of criteria;

- determining a configuration for said N ₂ cells, based on said second set of non-dominated configurations and on the plurality of weights estimated.

[0039] The non-dominated sorting algorithm may be based on the Pareto dominance principle or any NDS algorithm.

[0040] In another possible embodiment, a plurality of configuration for said N ₂ cells may be determined.

[0041] In one or several embodiments, in /f/, the determining of the second group of configurations may comprise:

- receiving a second set of constraints depending on a type of operation at the drain.

[0042] The type of operation may be an injection operations or production operations. The injection operation performed at well injector may correspond to the injection of liquid in the reservoir through the drain. The production operation performed at well producer may correspond to the extraction of oil from the reservoir through the drain.

[0043] It is noted that the second set of constraints may be the same of the first set of constraints, or it may be different.

[0044] In one or several embodiments, in /f/, the determining of the configuration for said N ₂ cells may comprise:

- determining N _s points in the space defined by the set of criterion measures, wherein N _s is an integer, via a random sampling from a multidimensional distribution of the plurality of weights estimated;

- determining, among the second set of non-dominated configurations, a configuration which minimizes a predefined proximity criterion to the N _s points determined;

- selecting said configuration as being the configuration for said N ₂ cells. [0045] The random sampling may be performed by any adapted statistical method, for instance via Latin hypercube sampling, or Normal distribution, or Gaussian distribution.

[0046] By “proximity criterion” it is meant any mathematical tool for quantifying a proximity to a point / set of points. The proximity criterion may be a mathematical distance, but other tools may be used (e.g. a proximity in terms of angles with respect to a given axis).

[0047] In a possible embodiment, a plurality of configurations may be determined. Each configuration may minimize a respective predefined proximity criterion to the N _s points determined. [0048] In one or several embodiments, the criterion measures may comprise at least one measure among:

- an ability of the N ₂ cells to flow from or into the drain ;

- an ability of the N ₂ cells to equalize a flow of the drain;

- a likeliness of the N ₂ cells to produce at the drain an undesired fluid at a rate; - a likeliness of the N ₂ cells to determine a breakthrough of an injected fluid at distance of production wells at a reference/given rate;

- a sum of the length of all blanked cells of the drain;

- a number of splits of the drain;

- an ability of the N ₂ cells to flow from or into the drain, function of a location of a considered cell of the N ₂ cells in the drain.

[0049] The ability of cells to flow may correspond to a first criterion measure M-i. In the case of a blanking design, the first criterion measure Mi may represent the relative ability of cells of drain to flow from or into the well in function of geology parameters, as example, the Peaceman transmissibility index of the considered cell divided by the drain penetration length in the considered cell.

[0050] The Peaceman index relating well in/outflow rates to cell block pressure according to cell size (reference: https://doi.org/10.2118/6893-PA).

[0051] The ability to flow of a cell may correspond to the facility of oil extraction from each cell of the drain in the geological gridded model. This facility may be dependent of the permeability attached to the cell, and related to the Peaceman transmissibility index. The ability to flow may correspond to the ability of a liquid to pass through the cell.

[0052] The ability to equalize a flow of the drain may correspond to the first criterion measure Mi In the case of a splitting design.

[0053] The ability to equalize flow across different sub-drains may be captured by the measure of the variance across sub-drains of the sum of cell Peaceman indexes belonging to the considered sub-drain.

[0054] The likeliness of cells to produce at the drain (i.e. production operation) an undesired fluid may respectively correspond to a second criterion measure M ₂ for a reference pressure gradient, a third criterion M ₃ measure for a high pressure gradient, and a fourth criterion measure M ₄ for a low pressure gradient.

[0055] It is noted that according to the Darcy Law, the rate or the pressure gradient is equivalent in the present context.

[0056] By “pressure gradient”, it is meant well operating conditions which may be related to a flow rate of an unwanted liquid. For instance, the use of high pressure gradient (low bottom hole well pressure) determines high flow rate of oil production but may also generate a high flow rate of unwanted liquid production.

[0057] The likeliness of cells to produce undesired fluid at reference rate, in production operation, may correspond to the diffusive time of flight in real space from the considered undesired fluid source (contact or injection well) to each cell of the portion.

[0058] The variation in likeliness of cells to produce undesired fluid at low and/or high pressure gradient, in production operation, may respectively correspond to the diffusive time of flight from the considered undesired fluid source in a space vertically exaggerated (or stretched) and/or reduced (or squeezed or shrinked).

[0059] The likeliness of cells to determine a breakthrough of an injected fluid (i.e. injection operation) at distance of production wells at a rate may correspond respectively to the second criterion measure M ₂ for a reference rate, the third criterion M ₃ measure for a high rate, and the fourth criterion measure M ₄ for a low rate.

[0060] The reference rate, in injection operation, may correspond to the diffusive time of flight in real space from producers to each cell of the drain.

[0061] The low and/or high rates, in injection operation, may respectively correspond to the diffusive time of flight in exaggerated and/or squeezed space from producers to each drain cell.

[0062] Thus, the use of exaggerated or reduced (squeezed) space may allow to take in consideration different pressure gradient (or productions rates, linearly related by Darcy’s law), and according to the subsoil condition (e.g. the balance between gravitational forces and viscous). [0063] Thus, advantageously, blanking and splitting design may help to determine an optimal drain configuration suitable to produce oil at different rates without producing any unwanted liquid (e.g. gas or water).

[0064] The sum of the length of all blanked cells of the drain may correspond to a fifth criterion measure M ₅ in the case of blanking design. The fifth criterion measure may define the ability of the total well drain to flow at high rate. [0065] The number of splits of the drain may correspond to the fifth criterion measure M ₅ in the case of splitting design. The number of splits may correspond to an investment degree, as for instance, the isolation of sub-drains may be typically operated through the use of packers bearing procurement and installation costs. Packers may be located outside casing and / or liner (“swell” or “open-hole” packers) or inside the casing or liner (“conventional” packers).

[0066] The ability of the cells to flow from or into the drain, function of the relative location of the considered cell within the drain may correspond to a sixth criterion measure M ₆ . The ability may be defined by a simple indicator bi-linear function of the length, equal to 0 at the middle of the drain and equal 1 at the extremities of the drain. Extremities of a drain may be expected to contribute more (per unit length) to the flow than the drain middle since the flow pattern in the vicinity of extremities will tend toward spherical behavior while it will tend toward radial behavior elsewhere. [0067] Thus, advantageously, it may be possible to take into account in the determination of blanking or splitting design the fact that all other factors being equal well extremities will produce more.

[0068] The present disclosure also relates to a method implemented by computer means for determining drain configurations of wells in a field containing a hydrocarbon reservoir, said method may comprise:

- receiving a geological gridded model of the field, o the geological gridded model comprising a plurality of cells, o and comprising a drain, said drain comprising N ₂ cells within the plurality of cells;

- receiving a set of criteria;

- determining possible configurations for said N ₂ cells;

- for each configuration among the possible configurations, computing criterion measures, each criterion measure representing a suitability to a respective criterion among the received set of criteria;

- determining, based on said criterion measures, a set of non- dominated configurations among said possible configurations via a non- dominated sorting algorithm for jointly optimizing the set of criteria; - determining a configuration for said N ₂ cells, based on said set of non- dominated configurations and on the plurality of weights estimated.

[0069] Another aspect of the invention relates to a non-transitory computer readable storage medium, having stored thereon a computer program comprising program instructions, the computer program being loadable into a data-processing unit and adapted to cause the data-processing unit to carry out the steps of any of claims 1 to 8 when the computer program is run by the data-processing device.

[0070] Yet another aspect of the invention relates to a device for determining drain configurations of wells in a field containing a hydrocarbon reservoir, the device may comprise:

/a/ an interface for receiving a first set of geological gridded models,

- each gridded model among the first set of geological gridded models comprising a respective plurality of cells;

- wherein each geological gridded model comprising a drain associated with a reference drain configuration (910), said reference drain configuration comprising N respective cells within the respective plurality of cells;

/b/ an interface for receiving a set of criteria;

Id for each geological gridded model among the first set of geological gridded models:

- a circuit for determining possible configurations for the N respective cells,

- for each possible configuration among the possible configurations and for each cell among said N respective cells, a circuit for computing criterion measures, each criterion measure representing a suitability to a respective criterion among the received set of criteria; a circuit for determining, based on said criterion measures, a set of non-dominated configuration among said possible configurations via a non-dominated sorting algorithm for jointly optimizing the set of criteria;

- a circuit for determining a first configuration among the set of non- dominated configurations, said first configuration minimizing a distance between the reference drain configuration respectively associated to the geological gridded model and the set of non-dominated configurations;

- a circuit for determining intermediate weights based on coordinates of the first configuration in a space defined by the set of criteria, each intermediate weight being associated to a respective criterion among the plurality of criteria;

161 a circuit for estimating a plurality of weights, each weight being associated to a respective criterion among the plurality of criteria and estimated based on intermediate weights among the pluralities of intermediate weights associated to said respective criterion; lei an interface for receiving a second geological gridded model of the field,

- the second geological gridded model comprising a plurality of cells,

- and comprising a second drain, said second drain comprising N ₂ cells within the plurality of cells;

IV a circuit for determining a configuration for said N ₂ cells based on the received set of criteria and the plurality of estimated weights.

[0071] Yet another aspect of the invention relates to a device for determining drain configurations of wells in a field containing a hydrocarbon reservoir, the device may comprise:

- an interface for receiving a geological gridded model of the field, o the geological gridded model comprising a plurality of cells, o and comprising a drain, said drain comprising N ₂ cells within the plurality of cells;

- an interface for receiving a set of criteria;

- a circuit for determining possible configurations for said N ₂ cells;

- for each configuration among the possible configurations, a circuit for computing criterion measures, each criterion measure representing a suitability to a respective criterion among the received set of criteria;

- a circuit for determining, based on said criterion measures, a set of non-dominated configurations among said possible configurations via a non-dominated sorting algorithm for jointly optimizing the set of criteria; - a circuit for determining a configuration for said N ₂ cells, based on said set of non-dominated configurations and on the plurality of weights estimated.

Brief Description of Drawings [0072] Other features, details and advantages will be shown in the following detailed description and on the figures, on which:

Fig. 1

[0073] [Fig. 1] illustrate a schematic view of subsoil comprising an oil reservoir to exploit. Fig. 2

[0074] [Fig. 2] is an example of a gridded model for performing blanking or splitting design in a possible embodiment of the disclosure.

Fig. 3a

[0075] [Fig. 3a] illustrate, in a possible embodiment of the present disclosure, a representation of computed values. .

Fig. 3b

[0076] [Fig. 3b] illustrate, in a possible embodiment of the present disclosure, a representation of computed values.

Fig. 3c [0077] [Fig. 3c] illustrate, in a possible embodiment of the present disclosure, a representation of computed values.

Fig. 3d

[0078] [Fig. 3d] illustrate, in a possible embodiment of the present disclosure, a representation of computed values. Fig. 4

[0079] [Fig. 4] presents, in a possible embodiment, sets of computed values of each criterion measure for the 2 ⁴ configurations of cells blanking in the case of productions operations. Fig. 5

[0080] [Fig. 5] is shows, in a possible embodiment, a ranking of the previous determined blanking configurations for each criterion measure.

Fig. 6 [0081] [Fig. 6] shows a non-domination analysis in one or several embodiment of the present disclosure.

Fig. 7

[0082] [Fig. 7] illustrates, in a possible embodiment, a selection process of blanking configurations among the 7 previous blanking configurations (or among all the possible configurations, if process described in Figure 6 is not performed).

Fig. 8

[0083] [Fig. 8] is a flow chart describing the determination of blanking or splitting configuration of drain cells, for injection or production operations, in a possible embodiment. Fig. 9

[0084] [Fig. 9] is a flowchart describing the estimation of parameters in a possible embodiment.

Fig. 10

[0085] [Fig. 10] represents a determination of the weights for a current well part, in a possible embodiment.

Fig. 11a

[0086] [Fig. 11a] presents, in a possible embodiment, sets of computed values of each criterion measure for the configurations of cells splitting in the case where all cells of the drain are opened, and for productions operations. Fig. 11b

[0087] [Fig. 11b] shows, in a possible embodiment, a ranking of the previous determined splitting configurations for each criterion measure. Fig. 12

[0088] [Fig. 12] is a possible embodiment for a device that enables the present disclosure.

Description of Embodiments [0089] Figures and the following detailed description contain, essentially, some exact elements. They can be used to enhance understanding the disclosure and, also, to define the disclosure if necessary.

[0090] Figure 1 illustrates a schematic view of subsoil comprising an oil reservoir to exploit.

[0091] Figure 1 is a representation of subsoil in two dimensions. Said subsoil may comprise a first zone 120 where no hydrocarbon and no water are present.

[0092] In addition, a zone 115, namely a reservoir, may comprise hydrocarbon fluid(s). Finally, said subsoil may also comprise a zone 106 with water, namely an aquifer.

[0093] Moreover, the zone 120 may have several sedimentary layers (no represented in figure 1).

[0094] For the sake of comprehension, figure 1 is presented in two dimensions, but 3D dimensions are possible. In addition, layers of porous rocks comprising water may surround the oil reservoir, more precisely on sides of the oil reservoir.

[0095] Layers in zone 120 may have their own physicochemical properties, (e.g. a specific porosity or permeability).

[0096] In order to exploit the oil in the reservoir 115, it is usually common to make one or several wells by drilling from the soil surface 101 to the reservoir 115. Function of the drilling conditions, i.e. for instance the layers of sedimentary rocks crossed, wells 102-103 may have different trajectory, for instance vertical or/and horizontal, or even be curved.

[0097] The area for each well 102-103 where the oil may be extracted or a liquid may be injected through the pipeline in the well is called the drain 130-131 (or drain portion). The injection/extraction may be performed through equipment such as a FCV (Flow Control Valve), an ICD (for Inflow Control Device) or an AICD (for Automated Inflow Control Device).

[0098] As described previously, an oil reservoir 115 may be composed of several sedimentary layers, and each layer may have its own physicochemical properties. [0099] In order to facilitate the oil extraction, once that the location of drain portions 130 or 131, is determined, it may be interesting to determine the blanking or splitting configuration of the drain portion, according to the physicochemical rocks properties for instance.

[0100] Indeed, depending of the location of the drain portion 130 or 131 along wells 102 or 103, several configurations of blanking or splitting may be possible. Some of them may present better configuration according to their location along their respective well relative to the subsoil geology.

[0101] These blanking or splitting configurations may be functions, for instance, of physicochemical properties of sedimentary rock layers of the reservoir 115, but also sedimentary rock layers in the neighbourhood of the reservoir 120, function of the presence of a fluid 106 close to the oil reservoir (as water or gas for instance), function of the operation performed at the well, or function of the distance between producer/injector wells. For example, the constraint of configuration design may be different if the performed operations are for an oil production or a liquid injection. [0102] Indeed, according to the physicochemical of sedimentary rock layers and/or forces exerted at the interface 115;106 (e.g. the gravity force, pressure force) an unwanted fluid from zone 106 may interfere in the production operations carried out at the drain portion 130 if said fluid is conducted to the drain portion of a production well. This phenomenon is called “coning water” or “coning gas”, and may reduce the rate of oil (or gas) production. The phenomenon of “coning” mainly happens when an unbalance appears between three forces: gravity, viscosity and capillarity between liquids of the reservoir (e.g. water and oil, or oil and gas, or any combination).

[0103] This unbalance may be caused by a high oil production rate associated to high pressure gradient, and high fluid velocity. Consequently, it may be necessary to adjust the oil production rate in order to avoid coning water/gas phenomena. The adjustment of the oil production rate may consist to be under a critical oil production rate, responsible of the coning water.

[0104] Another cause of oil rate decreasing may be the injection of fluid close to a producer well. Indeed, in the case of injection operations carried out at the portion 131 , the injection of fluid (e.g. water) in the reservoir 115 may create locally an area of the injected fluid around the drain portion 131. And, consequently, this liquid may diffuse through the sedimentary rock layers from the portion 131 to portion 130 (of producer well) and cause the decrease of the rate oil production.

[0105] Thus, to avoid unwanted fluid from aquifer or injection operations (e.g. in the case of production operations) or to control the injection effect on the oil production, it may be interesting to determine an optimal configuration of blanking or splitting for the drain portion. This optimal configuration may be a maximum of the rate of oil production in regard of the environment conditions/constraints. [0106] Figure 2 is an example of a gridded model for performing blanking or splitting design in a possible embodiment of the disclosure.

[0107] The field may be numerically modeled using a two-dimensional (2D) or three-dimensional (3D) gridded model comprising a plurality of adjacent cells 201. Each cell 201 may have a specific geographical position in the model, defined by geographical coordinates (x, z) or (x, y, z). Furthermore, each cell 201 may have a shape and a surface in the case of a 2D gridded model, or a volume in the case of a 3D gridded model.

[0108] Each cell 201 may be associated with cell infilling properties, which characterize the content of the cell 201 , as well as the properties of the fluid contained in the cell 201 when applicable.

[0109] This represented part of reservoir may also include the drain portion (or drain) of the well 102 or 103 for which the optimal configuration of the blanking or splitting design has to be determined. This drain portion, may be represented by a set of N cells 201 , N may be an integer superior or equal to 1. [0110] The represented part of reservoir may also include a part of layer 106 corresponding to the presence of fluid, for instance water from aquifer, and named “water leg”. This water leg may be considered as an unwanted liquid source for cells used for operation productions. According to another example, an unwanted liquid source may be also the location of cells used for injection operations, for instance performed at the drain portion 131. [0111] For the sake of comprehension, 4 cells defining the drain portion 130 or

131 are considered, named a, b, tt, W, and only 5 criterion measures are considered, named criterion measures M-i _, M ₂ , M ₃ , M ₄ , and M ₅ .

[0112] Therefore, the purpose of blanking design may be defined as a determination of an optimal configuration or a set of optimal configurations of opening or closing for each cell a, b, tt, W of the drain portion. An opened cell of drain portion may be understood as an opening of the wall of the well on the reservoir in order to access the reservoir, each wall be associated to one respective cell of the drain.

[0113] Likewise, the goal of splitting design may be defined as a determination of an optimal configuration or a set of optimal configurations of independent sub-portions for a drain, fully or partially opened. For instance, two sub-portions composed of a, b for the first one, and tt, W for the second one may be separated by an insulating device, as example a packer. Each sub-portion may be then considered as only one respective cell. [0114] In a possible embodiment of the present disclosure, the splitting design of a drain portion may follow a blanking design previously performed on the same drain portion.

[0115] Thus, in a purpose to determine one or several optimal configurations of blanking (or splitting) of cells, for injection or productions operations, it may be possible to compute values of each criterion measures for all possible configuration of a, b, tt, and W, i.e. 2 ^N blanking configurations for N considered drain cells or 2 ⁴ configurations for a drain with 4 cells considered in the gridded model, and then perform a ranking between all configurations on the basis of criterion measures computed values. Each configuration may correspond to a combination of opened and/or closed cells.

[0116] In the case of splitting, for N considered drain cells in the gridded model, the number of possible configurations may be equal to the binomial coefficient A = CN-1 with S corresponding to the number of splits. For 4 considered cells a, b, tt, W, and 3 splits, 8 possible splitting configurations may be obtained.

[0117] Figure 3a to figure 3d illustrate, in one possible embodiment of the present disclosure, a representation of computed values for 5 criterion measures (M _I ;M ₂ ;M ₃ ;M4;M5) for a first configuration of blanking or splitting intended to be used for production operations.

[0118] In the case of a blanking design, the current configuration corresponds to the opening of the cells a, b, tt, W on the reservoir. For this first configuration, none of cells is closed (said “blanked”).

[0119] In the case of splitting design, the current configuration corresponds to the use of no insulation device. Therefore, all of the cells a, b, tt, W are merged and represent an opened portion drain with only one flow control point. The opening of the drain portion may be full or partial corresponding to the determination of blanking configuration by a previous blanking design of this same drain portion.

[0120] In reference of the figure 3a, each cell of the gridded model is associated with one computed value of the first criterion measure M-i, and according to a scale of computed values 320. This scale may be any scale. Each computed value corresponds to the ability to flow of the considered cell for a blanking design, or the ability to equalize flow of the considered cell for a splitting design.

[0121] For instance, the Peaceman transmissibility index (or Peaceman well index related to the permeability) of the cell a is lower than the Peaceman transmissibility index of the cell b, tt, or W. Thus, in reference to figure 3a, the oil extraction or fluid injection at cells b, tt, or W may be easier than the cell a since the associated value index is higher.

[0122] From figure 3a, several observations may be recited and several scenarios (or configurations) may be proposed:

- a risk of vertical water coning (e.g. during oil production operations) may exist into cell a due to a maximal ability to flow around the cell a and a shortest path to water leg 106 in condition in which there is little gravity stabilization of water 106 / oil front 115. Such risk may legitimate the hypothesis that it may be optimal to not open to flow cell a while leaving all other cells open to flow. In the case of blanking, it corresponds to keep the cell a closed (blanked) on the reservoir.

- a scenario in which the cell p be closed to flow while all others be open to flow is unlikely to be optimal because the deliverability of cell b and p are same and cell b present higher water breakthrough risk than cell p being closer to the water in same permeability/Peaceman Index of rock.

[0123] In the case of a splitting design, the configuration of figure 3a using none split may be not optimal since the obtained flow (e.g. oil production flow) is function only of the subsoil properties, and may be superior to a critical oil flow. If the flow may be possible only below a certain fraction of water vs. total flow rate (e.g. due to vertical lifting conditions) such fraction might be reached for an oil cumulative volume lower than the corresponding cumulative volume achieved with a split drain (that would correspond to two distinct flow rate conditions). The difference of volume might warrant the costs associated with a split design. Consequently, it might be interesting to split the cells of the drain portion in order to allow a balance of the flow for instance.

[0124] By using additional criterion measures, it may be possible to evaluate the capacity of a configuration of blanking (or splitting) taking into account different constraints, as for instance the kind of operations to perform or the presence of an unwanted liquid (gas or water for instance), etc.

[0125] Figure 3b to figure 3d illustrate the use of 3 additional criterion measures with their associated computed values 330, corresponding to the diffusive time of flight in the gridded model for three conditions of exploitation: High rate (Figure 3b), Reference rate (figure 3c), and Reduced rate (figure 3d).

[0126] The diffusive time of flight may represent a precise and inexpensive measure of the time of arrival of a fluid from the source to the well if the source pressure is uniform at the front and at the well and the fluid from source and in reservoir share same density and viscosity. The diffusive time of flight rank may represent a low cost measure of the relative risks of source fluid arrival time and rate at various locations in the reservoir if the source pressure is uniform at the front and at the well. More generally, the computed ranks of diffusive time of flight with various vertical vs. horizontal stretch may represent a mean to capture relative risks of source fluid entry in a well under ranges of pressure conditions at source and well. Accuracy depends upon the distance from homogeneous conditions at source and well and variations of viscosity and density in the reservoir.

[0127]

[0128] Thus, in the gridded model context, the diffusive time of flight may correspond to the ability for a pressure wave to go by diffusion from a cell to another cell. For instance, the diffusive time of flight (DTOF) may be determined according to a fast marching method. In addition, the computed values of the diffusive time of flight may be associated to a scale 330 or index values. For instance, the DTOF may be computed using a Fast Marching Method (FMM) for structured grids, as described in the paper of J. Sethian, “A Fast Marching Level Set Method for Monotonically Advancing Fronts”, Proc. Natl. Acad. Sci., pp. 1591- 1595, 1996. Alternatively, the method of J. N. Tsitsiklis, “Efficient algorithms for globally optimal trajectories”, Automatic Control, IEEE Transactions, pp. 1528- 1538, 1995, may be applied using anisotropic slowness on any grid (including Corner Point Grids involving Non Neighbor Connection, NNC). However this method has a higher computational cost. [0129] Of course, other distance measures may be used.

[0130] Nevertheless, to ease the determination of the diffusive time of flight for various rates, it is also possible to “shrink” or “stretch” the dimension of the gridded model in the z direction (i.e. the vertical direction). Indeed, this “shrinking” or “stretching” is representing the modification of the pressure due to the rate of the production compared to the gravity force. Thanks to the Darcy equation, it is possible to demonstrate that a change in the pressure field may be represented by (i.e. is equivalent to) a change in the dimensions of the model.

[0131] Therefore, the model of figure 3b may be a compressed model of the subsoil in the z direction. The model of figure 3c may be the model of the subsoil without dimensions modification. The model of figure 3d may be an extended model of the model in the z direction. [0132] As for figure 3b and 3d, dimensions may have been changed, the diffusive time of flight will be modified. Indeed, the diffusive time of flight takes into account the size of the cells (e.g. the vertical size of cells - which have changed - and the horizontal size of cells - which are unmodified).

[0133] Figure 3b illustrates the case where the space of the gridded model is shrinked (e.g. compressed) in the vertical direction, which may simulate the presence of high flow rates of unwanted liquid (or the use of high flow rate of oil production) in the subsoil during an production operations performed at cells a, b, TT, and W.

[0134] According to the computed values (of diffusive time of flight) of each cell, the cell a with the value of 5 may present, compared to the cells b, tt, or W, the most important risk of producing unwanted liquid, as example water from the water leg. This result may seem consistent with the fact that the cell a is surrounded by cells with high Peaceman transmissibility index, and so, representing the shortest path between water leg and cell a.

[0135] Figure 3c illustrates the case where the space of the gridded model is considered as normal, which may correspond to a reference rate (or medium rate or normal rate or real rate).

[0136] For said figure, as the dimensions in the vertical direction are different from Figure 3b (nota: it is noted that scales of Figures 3a, 3b, 3c, and 3d may not be identical in the z direction even if it is not apparent), the computed values for the diffusive time of flight are different.

[0137] Figure 3d illustrates the case where the space of the gridded model is considered as stretched (or extended), which may correspond to a low rate (i.e. below the reference rate).

[0138] For said figure, as the dimensions in the vertical direction are different from Figure 3b and Figure 3c, the computed values for the diffusive time of flight are different. [0139] From figure 3a to 3d, it may be possible to determine the “relevancy” of a configuration of blanking or splitting based on sets of computed values for each criterion measure, according to the constraints, and the scale of computed values specific to each criterion measure. For instance, for the previous case of configuration, the sets of computed value may be (5;9;9;9) for the first criterion measure M-i, (5;3;3;3) for the third criterion measure M ₃ , (5;3;3;4) for the second criterion measure M ₂ , and (5;3;4;5) for the fourth criterion measure M ₄ .

[0140] The capacity of the above configuration of blanking or splitting may represent the ability of the configuration to produce oil compared to a risk of liquid breakthrough of this configuration.

[0141] For injection operations, similar previous figures (Fig 3a-3d) for the same blanking configuration may be obtained and presented computed values of each criterion measure. The criterion measures may be, in this case, relative to injection operation in order to evaluate the breakthrough risk of a configuration on a distant producer well or anyone of the distance producer wells in the field. Nevertheless, the above description applies likewise to the injector wells or injections operations.

[0142] Both methods are quite similar apart from criterion measures which may differ from one to another.

[0143] Figure 4 presents sets of computed values of each criterion measure for the 2 ⁴ configurations (400) of cells blanking (i.e. for cells a, b, tt, and W) in the case of productions operations.

[0144] The fifth criterion measure corresponding to the sum of blanked cell (closed). This fifth criterion measure may be used in order to normalize the other criterion measures if needed.

[0145] Each configuration 400 is defined by a set of opened and closed cells, where a closed cell is defined by a black square, and an opened cell is defined by a white square.

[0146] For each criterion measure, the sets of computed values (e.g. 410) may be associated in the form of a summed value (e.g. 411) corresponding to the sum of the computed values of the considered set. This summed value may be representative of a given configuration for a given criterion measure.

[0147] The computed value of a closed cell of a drain may be considered equal to zero. [0148] For instance, in the case of the A configuration, a first summed value 411 may be determined by the sum (i.e. 5+9+9+9=32) of computed values of the set 410 and associated to the M-i. A second summed value 421 may be determined by the sum of computed values of the set 420 and associated to M ₃ , etc. A third summed value 431 may be determined by the sum of computed values of the set 430 and associated to M ₄ , etc. In the present sum, all values to be summed are different to 0 as no cells are blanked. If a cell is blanked, the associated value in the sum may have been set to 0. The previous sum may be performed for any criterion measures.

[0149] According to another possible embodiment, the determined value (e.g. 411) for a configuration of cells for a considered criterion measure may be based by an average of computed values (e.g. 410) of cells of the drain portion.

[0150] Figure 5 shows a ranking of the previous determined blanking configurations for each criterion measure. The rankings may be determined according to the following process:

- for each criterion measure, and referring to Figure 4, it is possible to order the summed values for each configuration;

- Therefore, for each criterion measure and for each configuration, it is possible to determine a rank (or an order index) on the basis of the determined ordering. [0151] Said ranks or order indexes are shown in Figure 5.

[0152] Thus, according to the rank of each configuration for a considered criterion measure, it may be possible to determine one or several configurations as the most relevant for a considered criterion measure.

[0153] For instance, configurations C and D, with a rank of 2 on 16 (and for criterion measure M ₃ ) may be less relevant to decrease the breakthrough water risk at high flow rate than the configuration B, with a rank of 5 on 16 of a breakthrough risk. The configuration B has also a better ability to flow than configurations C and D.

[0154] Referring Figure 5, it may be noticed that several configurations may present the same rank for a considered criterion measure, as for example the configurations H, G and F for the criterion measure M ₃ .

[0155] Thus, the determination of one or several blanking configurations may be seen as a “multicriterion measures” optimization problem, or Pareto optimization problem, for which it may not be possible to isolate a unique solution that is better than the others for all criterion measures. [0156] Flowever, it is possible to keep some configurations which are preferred to

(which “dominate”) other configurations, by using a Non-dominated sorting (NDS) algorithm.

[0157] The use of a non-domination analysis may help to select the best configurations among configurations having the same rank for one or several criterion measures.

[0158] Figure 6 shows a non-domination analysis in one or several embodiment of the present disclosure. The non-domination analysis may be performed by a comparison between configurations (e.g. by the use of a matrix shape). [0159] In another possible embodiment, the non-domination analysis may be performed by the use of any non-dominated algorithm.

[0160] Thus, it may be possible to (optionally) identify configurations that are (without doubt) dominated by other configurations.

[0161] A configuration is dominated without doubt by another configuration if every criterion measure for said configuration is below (or greater depending of the criterion - e.g. “below” for ability to flow and “greater” for water breakthrough risk (because it is a risk / drawback and not a protection / advantage)) the respective criterion measure for a second configuration.

[0162] For instance, between the similar configurations C, E, it is possible to note that configuration C is clearly dominated by configuration E (element 611 is set to 1 to show said domination) as configuration C has (as rank for each criterion measure) 3, 2, 2, 2, 12 and configuration E (as rank for each criterion measure) 3, 2, 4, 4, 12.

[0163] The results of the domination analysis lead to a number of 7 configurations remaining compared to 16 initial configurations. Said 7 configurations are not clearly dominated by another configurations (i.e. configuration having only 0 in their line - lines with an arrow at the end of the lines). It is noted that configuration Q is excluded as all cells are blanked (closed).

[0164] Figure 7 illustrates, in a possible embodiment, a selection process of blanking configurations among the 7 previous blanking configurations (or among all the possible configurations, if process described in Figure 6 is not performed).

[0165] Figure 7 is describing a case where only two criterion measures Mi and M ₂ are used, corresponding respectively to the first and second criterion measure. Of course, the process may be used for more than two criterion measures. [0166] From the summed value (or rank - in the following the rank is only mentioned but may be replaced by the summed value) of each criterion measure Mi or M ₂ , each blanking configuration may be represented in the figure 7 by a couple of coordinates, each coordinate representing the value of the respective criterion measure. [0167] In one possible embodiment, the coordinates of each configuration may be defined by their rank for considered criterion measure minus one, and divided by the number of configurations. The coordinate for a considered criterion measure of each configuration may thus be comprised between 0 and 1.

[0168] In another possible embodiment, the coordinates may be defined by the summed value of a configuration divided by the maximal summed values for all configurations.

[0169] These 7 blanking configurations, called non-dominated configurations, may define the Pareto Frontier.

[0170] The left figure of Fig. 7 represents the Pareto frontier 701 (in grey), and the configurations 702, according to their coordinates, on the Pareto frontier 701 (represented by circles with solid lines). It is noted (in case where the coordinates are based on the summed values) that the axis of the graph may be defined such that the origin is defined by the two minimum criterion measures for configurations located on the Pareto Frontier, even if this is not mandatory. The configurations 703 under the Pareto frontier 701 (represented by dashed circles) are the dominated configurations, determined by the non-domination analysis, and which are not Pareto optimal, and which are eliminated as possible blanking configurations.

[0171] In order to further reduce the configurations domain, a second selection within the group of non-dominated configuration 702 may be performed. For instance, it may be decided to keep only one or a number N _s of non-dominated configurations 702 considered as the optimal blanking configuration according to the different constraints cited above.

[0172] Each criterion measure M, is associated to a respective weight w _t comprised between 0 and 1, such that the sum of the weights is equal to 1 : w-L + ··· + w _n = 1. Therefore, it may be possible to assign more importance to certain criterion measures than to others.

[0173] Thus, the second selection may then be performed by sampling N _s points in the hyper-quadrant 704 containing the non-dominated configurations 702 according to the weights {w ₁ ,w ₂ , - , w _n } associated to the corresponding criterion measures M-i, M ₂ , ..., M _n and to select the configuration(s) among the non- dominated configurations 702 which are the “closest” to the sampled points, according to a predefined criterion measure. The sampling may be performed for instance by a Latin Flypercube sampling.

[0174] An example of this second selection scheme is represented in the center figure of Fig. 7, where N _s =2 points A and B, are randomly sampled according to the weights , w ₂ of the criterion measures M ₁ and M ₂ . For instance, the sampling of points may comprise the determination of coordinates of the points where distribution for determining said coordinates are any distribution, for instance centered around respective weights w ₁ , w ₂ (for instance, normal or gaussian distribution).

[0175] The “closest” non-dominated configuration 702 may be the one which minimizes a mathematical distance (e.g. Euclidian distance) to the sampled point. Alternatively, the “closest” non-dominated configuration 702 may be defined as follows:

- consider the lines joining the origin of the graph and the sampled points A and B;

- determine the intersection between these lines and the Pareto frontier;

- for each sampled point A and B, select the non-dominated configuration a and b which minimize a mathematical distance to the intersection of the respective line and the Pareto frontier.

[0176] This last scheme is illustrated in the right figure of Fig. 7, where the black circles 705 represent the non-selected non-dominated configurations, and the white circles 706 represent the selected non-dominated configurations, i.e. the configurations corresponding to the optimal possible blanking configurations.

[0177] Other schemes may be used for the second selection. For instance, it is possible to select, for each sampled point A or B the non-dominated configuration a or b such as the difference between the angle of the non-dominated configuration a or b and the angle of the corresponding sampled point A or B relative to an axis of the graph, is minimized. In other words, the selected non- dominated configurations a and b may be the closest configurations on the Pareto frontier in the direction of the sampled point A and B.

[0178] In another example, N _s may be equal to 1 allowing to obtain only one optimal possible blanking configuration.

[0179] Figure 8 is a flow chart describing the determination of blanking or splitting configuration of drain cells, for injection or production operations, in a possible embodiment.

[0180] At the first step, input data for the method of blanking or splitting design may be provided with the gridded model 810. The input data may be:

- the type of operations to perform, production or injection, or the type of well for which a blanking or splitting configuration has to be determined.

- The type of configuration wanted (e.g. blanking or splitting configuration) for the well drain portion, or one after the other on the same drain portion.

- One or several model realization, (e.g. based on seismic measurements) may provide geologic information about the subsoil and reservoir to exploit. - One or several criterion measures mentioned above and allowing to measure the suitability of configurations. Each criterion measure may be associated to a respective weight w, to give more or less importance to certain constraints compared to others (e.g. the presence of a unwanted liquid close to the drain portion). The weights wi,...,w _n may be also an input of the method of blanking or splitting design.

[0181] In one possible embodiment, when a splitting design follows a blanking design on the same drain portion, the same weight for a given criterion measure for both splitting and blanking may be used. In another example, different weight may be used for the splitting design and blanking design on the same drain portion.

[0182] In addition, in some possible embodiments, it may be possible to provide in input data the location of other well drains close to the current well, or the location of injector or producer wells. For instance, the location and configuration of the drain 131 may be an input when the purpose is to determine a blanking or splitting configuration at the drain 130.

[0183] In some possible embodiments, it may be possible to provide constraints related to the blanking or splitting design. For instance, it may be possible to exclude some configurations such as:

- configurations for which the proportion of cells in drain opened to flow exceed a given threshold,

- configurations for which specific cells in drain are opened or closed (blanked). For instance, according to the subsoil geology where the drain is located, it may be interesting that, for all configurations selected, a specific cell of the drain has to be opened, because the oil extraction or injection will be better, - configurations for which splits may be located at specific intervals.

[0184] Thus, at step 810, a gridded model is received with one or several previously cited input data. The gridded model represents the properties of the reservoir contained in the field, the cells corresponding to the portion drain of the well, for instance the drain 130, and comprising also a plurality of adjacent cells. [0185] At step 820, the type of operations may be determined in order to determine the criterion measures to be used at the step 840. The type of operation depends of the input data indicating if the drain portion own to a producer or injector well.

[0186] Optionally, a step 830 may be performed in order to determine the location on the neighboring drains or wells. Indeed, the design of blanking or splitting configuration may depend of the different sources of unwanted liquid in the case of production operation. The sources may be from the water leg, or neighboring drains (e.g. used for injection operation). In the case of injection operation, the knowledge of the neighboring drains used for production operations may help to determine the optimal configuration of cells blanking or splitting. [0187] At step 840, one or several criterion measures to use may be determined according to the input data, and the previous step 820 or 830.

[0188] At step 850, on the basis of the previous determined criterion measures and input data, a set of all possible configurations of blanking or splitting may be defined, and depending on the number of cells constituting the drain (or portion drain).

[0189] At step 860, summed values computed of each criterion measure for all configurations may be determined (see description related to Figure 4). Thus, each cell of a configuration is associated to a computed value for a considered criterion measure. [0190] At step 865 (optional), a ranking of all configurations for each criterion measure may be defined (see description related to Figure 5) on the basis computed values.

[0191] In order to reduce the number of possible blanking or splitting configuration, a non-domination analysis may be performed 870 (see description related to Figure 6). The optimal possible configurations may correspond to the non- dominated configurations, and representat the Pareto Frontier (see description related to Figure 7).

[0192] At step 880, a sampling of N _s points in the hyper-quadrant containing the non-dominated configuration may be performed, as described with reference to the right figure of Fig. 7. [0193] At step 890, one or several optimal blanking configuration may be determined. This/these configurations may be the configuration(s) which is/are the closest (according to a proximity criterion) to its respective line among the selected configuration (in the example of Fig. 7, the proximity criterion being a Euclidean distance).

[0194] The process may be repeated until all drain portions are processed or/and to perform a splitting design on the obtained drain portion(s).

[0195] The method described above with reference to Fig. 7 makes it possible to automatically determine a blanking configuration of drain cells, depending on the geology, according to an artificial intelligence approach based on optimality rules that are, in practice, the rules used by the reservoir geologist / engineer who establishes such patterns “by hand”. It allows a very strong reduction of the search space.

[0196] This method may receive, as input, a set of parameters, including the weights associated with the constraints used for determining the well pattern. It models the knowledge of the operator (reservoir engineer / geologist) for determining configuration of blanking, but this user still has to set the values of these parameters in relation with the specific knowledge of the problem (local knowledge). [0197] These parameters may define the axis of the space to be explored for determining the drain configuration, as represented for instance in Fig. 7. Flowever, even if the operator may have an intuitive understanding of the axes, it is complicated for him to quantify these axes. Therefore, the effectiveness of the method described above is reduced, because in practice the operator has to test different values of the weights and judge their relevance as a function of the result returned by the method.

[0198] There is thus a need for a method that does not require the knowledge of the weights associated with the constraints for determining blanking configuration. The method described below proposes for this purpose to automatically determine optimal blanking configuration(s) from:

- one or several reference blanking configuration(s), generated by an expert (reservoir engineer or geologist); and - a description of the constraints and the criterion measures used by the expert for generating said reference blanking configuration(s). The weights associated with the constraints are not a part of this description and are not known.

[0199] The purpose of the method is to estimate the weights underlying the generation of the reference blanking configuration(s) in order to provide configuration(s) “close” to the reference configuration(s), i.e. built according to a similar logic.

[0200] By “underlying”, it is meant that the weights are not positively set by the expert to generate the reference model, but they mathematically translate an intuitive knowledge of the expert on the studied geological structure.

[0201] Once these weights are estimated from reference configurations, it is then possible to apply the configuration determination method described above to similar fields in terms of geology, and to obtain new configuration(s) completely automatically.

[0202] Figure 9 is a flowchart describing the estimation of parameters in a possible embodiment.

[0203] At step 910, a set of one or several reference blanking configuration(s) may be received (associated with a respective drain portion), and each blanking configuration may be defined by one or several criterion measures associated at respective weights.

[0204] Additional input data may be provided at step 910, and which may be similar to the input data of the figure 8, except the input data related to weights since they have to be determined. [0205] At step 920, the coordinates of each reference blanking configuration may be determined on the basis of each criterion measure (as for any configuration and as described in the previous Figures). In other word, each reference blanking configuration may be located in the domain of the criterion measures (M _1; M ₂ , M ₃ , M ₄ , etc.) in order to place them in the domain of Figure 10 (see below). Therefore, the values for each criterion measures may be determined for each reference blanking configuration. [0206] At step 930, and for each drain associated with a respective reference blanking configuration, all possible configurations (for blanking or splitting) may be determined, and non-dominated blanking configurations may be selected similarly the previous disclosure. A Pareto frontier from non-dominated blanking configurations may be determined similarly to the step 820 to 870 of the figure 8.

[0207] At the step 940, reference blanking configurations considered as the closest to the Pareto frontier may be determined.

[0208] For instance, this determination 940 may be performed by minimizing a Flausdorff distance between the reference blanking configurations and the group of determined blanking configurations belonging to the Pareto Frontier and selecting the reference blanking configuration that minimize this distance.

[0209] At step 950, the weights w _t of the criterion measures M, may be determined: these weights may correspond to the coordinates of the blanking configuration belonging to the Pareto Frontier determined at step 930 on the axis corresponding to respective criterion measures.

[0210] Once steps 920-950 have been performed for each drain associated to each respective reference blanking configuration, we have a set of values

\wi , k E {l, for each weight w _£ (K the number of reference well patterns). Each weight w _t may then be estimated as a function of the corresponding set of values For instance, the estimate of the weight w _t , for a given type of well and a given type of part, may be a mean of the determined values w for said type of well and for said type of part.

[0211] In another embodiment, it is possible to sort, for each reference blanking configuration, the weights identified by increasing distance to the Pareto Frontier, and to average only the Y first weights, Y being an integer (input of the method).

[0212] Other functions may be used.

[0213] Figure 10 represents a determination of the weights for a current well part, in a possible embodiment. [0214] Figure 10 represents a determination of the weights as performed in steps 920 to 950 of Fig. 9.

[0215] In the left figure, the non-dominated blanking configurations of the Pareto Frontier 1001 are represented, together with the provided blanking reference configuration 1003. The pair 1004 of blanking configurations comprising the configuration among the blanking configuration 1003 and the configuration among the blanking configurations 1002 of the Pareto Frontier having the lowest distance is selected. The coordinates of the blanking configuration of the Pareto Frontier belonging to the selected pair 1004 correspond to the values of the weights w , w ₂ ⁵ of the respective criterion measures M _lt M ₂ .

[0216] The estimated weights may then be used as input of the method for determining blanking configuration, overcoming the above-mentioned problems related to the definition of these parameters by the operator.

[0217] In other words, a possible method for automatically determining blanking configuration may comprise the following steps:

- Receive a set of blanking configuration corresponding to respective gridded models (e.g. constructed by hand by an operator);

- Determine, with the estimation method described in Fig. 9, a set of parameters; and - Determine a determining blanking configuration by using the construction method described in Fig. 8 with the set of parameters previously estimated, as for instance the weight for each criterion measure.

[0218] As explained previously, the presented method for a blanking design at a drain portion may be used for a splitting design according to a similar methodology, except for some criterion measures which may be different, or the same criterion measure may have a different interpretation. For instance, in the case of the criterion measure M-i, from the same computed values, in the case of blanking design, the interpretation of the computed value may correspond to the ability to flow, and may correspond to the ability to equalize to flow in the case of splitting design. [0219] For instance, a cell of a drain with high ability to flow, and isolated from other cells of the same drain may be more suitable to equalize a total from cells of the drain than a cell with a weak ability to flow. Indeed, the range of available rates (e.g. performed by a valve), at cell with high ability to flow may be more important. [0220] The criterion measure M ₅ for splitting device may be related to investment degree, For instance, the use of 3 splits corresponding to 4 sub-portion of the drain portion 130 requires more investment that the use of 1 split corresponding to 2 sub-portion of the drain portion 130.

[0221] The previous figure 3a to figure 3d in the case of splitting design may correspond to a splitting configuration with none insulation between cells. Furthermore, the splitting configuration performed may be performed on a drain with opened or/and closed cells, determined according to the previous blanking design for instance. In the case of splitting configuration presented in figure 3a to figure 3d, all cell of the drain are opened. [0222] As explained previously, this splitting configuration of drain portion may be not optimal, and according to a similar methodology used for the blanking design, it may be possible to determine a ranking for each criterion measure of all splitting configurations for a given configuration (e.g. a combination of opened and/or closed cells of a drain).

[0223] Figure 11a presents, in a possible embodiment, sets of computed values of each criterion measure for the configurations (8 possible splitting configurations) of cells splitting (a, b, tt, and W) in the case where all cells of the drain are opened, and for productions operations. [0224] Each configuration 1100 may be defined by a set of insulated devices

(packers) corresponding to the vertical black lines. For the cells which are directly neighbours (i.e. without insulating device between them), only one cell may be defined corresponding to the merging of the directly cells. For instance, the configuration A may be defined by one only cell corresponding to the merging of the four opened cells (a, b, tt, and W) of the drain portion. The configuration B using only one insulating device may be defined by 2 cells. The first cell (1) may comprise one cell (a), and the second cell (2) may correspond to the three merged cells (b, TT, and W).

[0225] For each criterion measure M-i, M ₂ , M ₃ , and M ₄ , and for each splitting configurations, a unique value 1111 for each cell of a splitting configuration defined by merged cells may be determined.

[0226] Each unique value may correspond to a mean of the computed values of the considered criterion measure, and associated to the merged cells. For instance, in the case of the criterion measure M-i, the configuration A with only one cell (4 merged cells) may be defined by a unique value equal at 8, and corresponding to the sum of computed values (5+9+9+9=32) divided by 4 (the number of merged cells). For the configuration B defined by two cells (1 , 2), the unique value for the first cell is equal at 5 (5 divided by 1), and the second unique value of the second cell (3 merged cells) is equal at 9 (9+9+9=27 divided by 3).

[0227] Thus, for each criterion measure, a set of unique values of a considered splitting configuration may be determined as previously explained. According to another example, in the case of configuration D and criterion measure M ₂ , unique values for each cell of the configuration D are 3.7 (5+3+3=11 divided by 3), and 4 (4 divided by 1).

[0228] For criterion measures M1 , M2, M3 and M4, the sets of unique values (e.g. 1111) may be associated in the form of a standard deviation (e.g. 1112) determined on the basis of the set of unique values of a considered configuration. For instance for configuration E and criteria M3 the set of computed values include three values 5, 3, 3 the standard deviation of which is 0.9.

[0229] Computed values of the criterion measure M ₅ may be defined by the opposite of the number of insulated device used for each splitting configuration. For instance, the computed value for the configuration A is equal at 0 (0 insulated devices are used), and for the configuration FI is equal at 3 (3 insulated devices are used). [0230] Figure 11b shows, in a possible embodiment, a ranking of the previous determined splitting configurations for each criterion measure. [0231] The rankings may be determined (according to the step 810 to 865 of the figure 8 for instance) according to the following process:

- for each criterion measure, and referring to Figure 11a, it is possible to order the values of variance for each splitting configuration;

- Therefore, for each criterion measure and for each configuration, it is possible to determine a rank (or an order index) on the basis of the determined ordering.

[0232] Said ranks or order indexes are shown in Figure 11a.

[0233] Thus, according to the rank of each configuration for a considered criterion measure, it may be possible to determine one or several configurations as the most relevant for a considered criterion measure.

[0234] As presented in figure 11b, the splitting configuration A corresponding to the use of none insulating and the merging of cells a, b, tt, and W may constitute an optimal configuration since it is the unique configuration bearing the minimum cost possible. .

[0235] On the contrary, the maximum cost solution H, may be excluded from the possibly optimal domain because it is dominated in term of ability of reducing disequilibrium per split device for all rate criterion while presenting higher cost than any other configuration.

[0236] In the similar way that the determination of a set of optimal blanking configurations, a set of optimal splitting configurations may be performed by the use of the non-domination analysis. For instance, in the case of 8 possible splitting configurations, it may be possible to select 3 splitting configurations which are non-dominated (A, B, E), and then to perform the methodology according to the step 870 to 890 of figure 8 in order to select one or several optimal splitting configuration(s).

[0237] In another possible embodiment, similarly to the blanking design, it may be possible to use the methodology of the figure 9 for determining weights in the case of splitting design, and then use it according to the method described in figure 8.

[0238] Figure 12 is a possible embodiment for a device that enables the present disclosure. [0239] In this embodiment, the device 1200 comprise a computer, this computer comprising a memory 1205 to store program instructions loadable into a circuit and adapted to cause circuit 1204 to carry out the steps of the present invention when the program instructions are run by the circuit 1204. [0240] The memory 1205 may also store data and useful information for carrying the steps of the present invention as described above.

[0241] The circuit 1204 may be for instance:

- a processor or a processing unit adapted to interpret instructions in a computer language, the processor or the processing unit may comprise, may be associated with or be attached to a memory comprising the instructions, or

- the association of a processor / processing unit and a memory, the processor or the processing unit adapted to interpret instructions in a computer language, the memory comprising said instructions, or

- an electronic card wherein the steps of the invention are described within silicon, or

- a programmable electronic chip such as a FPGA chip (for « Field-Programmable Gate Array »).

[0242] This computer comprises an input interface 1203 for the reception of input data used for the estimation method according to the invention and an output interface 1206 for providing a set of estimated parameters 1207. These parameters may then be used as input data of the method for determining blanking or splitting configurations of drain portion detailed with reference to figure 8.

[0243] To ease the interaction with the computer, a screen 1201 and a keyboard 1202 may be provided and connected to the computer circuit 1204.

[0244] Expressions such as "comprise", "include", "incorporate", "contain", "is" and "have" are to be construed in a non-exclusive manner when interpreting the description and its associated claims, namely construed to allow for other items or components which are not explicitly defined also to be present. Reference to the singular is also to be construed in be a reference to the plural and vice versa. [0245] A person skilled in the art will readily appreciate that various parameters disclosed in the description may be modified and that various embodiments disclosed may be combined without departing from the scope of the invention.