CN102930345B - A kind of self-adaptation well net optimization method based on gradient algorithm - Google Patents

Abstract

The invention belongs to oil-gas field development field, particularly, relate to the well net optimization method in extensive oilfield development process, well net optimization method is as follows: (1), acquisition oil reservoir data, set up three-dimensional digital geological model of oil accumulation, carry out numerical reservoir simulation; (2), build adaptation rule element pattern, generate well pattern; (3), computing economy benefit, evaluate initial well pattern; (4), setup parameter, for well net optimization is prepared; (5), use gradient method, optimize cell formation parameter, generate new well pattern; (6), calculate and evaluate the economic benefit of new well pattern, then interative computation, until generate optimum well pattern.The present invention utilizes Optimum Theory to carry out well net optimization, can find optimum well pattern form; Well net optimization and note are adopted optimization combine, optimize the optimum note amount of adopting under different well pattern form; While optimized well pattern form, obtain its corresponding optimum note adopt total amount; Can realize optimizing to multiple rule well pattern.

Description

A kind of self-adaptation well net optimization method based on gradient algorithm

Technical field

The invention belongs to oil-gas field development field, particularly, relate to the well net optimization method in extensive oilfield development process, under the condition of given initial well network parameters, in conjunction with numerical reservoir simulation, by optimized algorithm optimization object function, while ensureing counting yield, obtain the economically viable optimum well pattern form adapted with actual oil reservoir, improve oil reservoir ultimate recovery factor.

Background technology

Current increasing late period of oil field in the world, hydrocarbon resources sharply reduces, and exploration finds that new hydrocarbon resources difficulty strengthens, and how at utmost raising oil and gas production and economic net present value become the key of oil and gas development.The exploitation of field produces New Optimizing Method is oil worker problem demanding prompt solution, and well spacing optimization relies on its advantage progressively to become independent in oil reservoir development and the problem of key.

Oilfield development program is main goal in research to reduce production cost, raising stable production period and recovery ratio, and the layout of oil field injection and extraction well is the key of oil reservoir Optimal Development.Oilfield well array optimization is to realize the maximization of productivity effect by adjustment oil-water well well location layout, because oil reservoir well net optimization problem control variable is many, as the diversity (optimization is adopted in well spacing optimization, note) of optimisation strategy, diversity (the 1. geologic parameter: the distribution of reservoir configuration, permeability distribution, degree of saturation, fluid contact level etc. of correlation parameter; 2. manufacturing parameter: well location, well number, well type, oil recovery rate etc.; 3. economic parameters: production fluid cost, drilling cost etc.), the difficulty therefore faced in research process is all well imagined from scale or complicacy.

For a long time, well net optimization design is main relies on artificial experience, and assist and carry out multi-scheme testing and debugging with numerical reservoir simulation, finally preferably scheme carries out field conduct preferably.But because Reservoir Heterogeneity is strong, physical difference is large, artificial well pattern of selecting is difficult to differentiate optimum well pattern type, yardstick and angle, can not find the optimum well pattern layout being suitable for this oil reservoir, have larger limitation.In recent years, field engineering technician proposes vector well pattern design concept, it is the distance according to the single element pattern oil-water well of simple permeability relational design, several mouthfuls of wells for negligible amounts are comparatively applicable, but for extensive well spacing, be just difficult to the indexs such as the spacing obtaining all unit corresponding optimum well pattern size, well pattern type and oil-water well.Therefore can find out, traditional well network design method efficiency is low, can not obtain the well pattern form of total optimization, need to be made improvements, meet the requirement of actual oil-field development.

Present inventor spy devises " a kind of self-adaptation well net optimization method based on gradient algorithm ", for optimized well pattern form quickly and easily, obtain being conducive to the parameter such as trend of the optimum well pattern form of engineering staff's reference, the spacing of oil-water well, angle, well pattern.

Summary of the invention

For the deficiencies in the prior art, the object of the invention is to carry out well net optimization for extensive oil reservoir development, propose the well net optimization method based on gradient algorithm.The method introduces new well pattern improved construction form, while considering element pattern form, consider that the note in oil field adopts situation, find optimum well pattern layout in conjunction with optimized algorithm, be intended to solve the problems such as the counting yield that current domestic well net optimization aspect exists is low, precision is little, difference of overall importance.

For achieving the above object, the present invention is by the following technical solutions:

Based on a self-adaptation well net optimization method for gradient algorithm, it is characterized in that, comprise the steps:

(1), obtain oil reservoir data, set up three-dimensional digital geological model of oil accumulation, carry out numerical reservoir simulation;

(2), build adaptation rule element pattern, generate well pattern;

(3), computing economy benefit, evaluate initial well pattern;

(4), setup parameter, for well net optimization is prepared;

(5), use gradient method, optimize cell formation parameter, generate new well pattern;

(6), calculate and evaluate the economic benefit of new well pattern, then interative computation, until generate optimum well pattern.

Preferably, above-mentioned steps 1 specifically comprises the steps:

(11), the favourable area of oil-gas accumulation in oil reservoir is found out, the oil field block that selected well spacing is favourable;

(12), reservoir static data is collected;

(13) three-dimensional digital geological model of oil accumulation, is set up;

(14), reservoir behavior data is collected;

(15), integrate static data and dynamic data, build numerical reservoir simulation model, develop analog computation.

Preferably, step 2 specifically comprises the steps:

(21), select well pattern type, estimate initial parameter, adopt 6 strat test net unit construction methods, generate initial self-adaptation element pattern.

6 parameters refer to well spacing, array pitch, rotation angle, the angle of shear, transverse direction and longitudinal translation distance, use them can obtain the different geometric deformation of unit, by optimizing this 6 parameters, indirectly achieve the optimization of self-adaptation well pattern.

(22), based on initial self-adaptation element pattern, in oil reservoir region, carry out spread, generate self-adaptation well pattern;

(23) well pattern that, will generate mates commercial reserve simulator Corner-point Grids, setting well location.

Preferably, well pattern type comprises four-spot pattern, five-spot network, seven spot pattern and nine-spot well network.

Preferably, the concrete steps of step 3 are as follows:

Foundation reservoir model and the well pattern layout obtained, total note amount of adopting that setting is suitable, utilizes commercial reserve numerical simulation software simulation oil field to develop, record its oil offtake, water production rate and water injection rate, consider the drilling cost of every mouthful of well, computing economy benefit, evaluate the quality of well pattern.

Preferably, the concrete steps of step 4 are as follows:

(41), the maximum and minimum value of setting well pattern adjustable yardstick, setup parameter comprises 6 element patterns and builds parameters and note adopts total amount;

(42) step-length of total simulated time and offset well network parameters, is set.

Preferably, the concrete steps of step 5 are as follows:

(51), distinguish disturbance 6 element patterns structure Optimal Parameters, ask production economy net present value (NPV) relative to the partial derivative of each parameter;

(52), upgrade well pattern and build parameter;

(53), generate new well pattern, search better well pattern form;

Preferably, the concrete steps of step 6 are as follows:

(61) if benefit improves, step 2-6 is repeated, until generate optimum well pattern.

(62) if benefit is deteriorated, shorten step-length, repeat step 5-6, until benefit improves; After this, then repeat step 2-6, until generate optimum well pattern.

Optimum well number, the type (water injection well or producing well) of diverse location well can be known in oil reservoir, the economic net present value under the note amount of adopting of every mouthful of well and corresponding well pattern layout by optimum well pattern.Utilize the parameter optimized and obtain, the parameter such as angle, position angle, the residing optimum planimetric position of unit of the element pattern size corresponding to optimum well pattern, element pattern can be calculated, according to oil field actual conditions, carry out correcting rear field conduct.

Compared with prior art, tool of the present invention has the following advantages in the present invention:

1, utilize Optimum Theory to carry out well net optimization, optimum well pattern form can be found.

Artificial setting is organized well pattern parameter more and is carried out the pseudoreduced method of mould and can find well pattern form relatively preferably, but this well pattern is limited to size and the array configuration of parameter, is difficult to find optimum well pattern.The maximum advantage of the present invention is reservoir simulation to combine with Optimum Theory, automatic tune ginseng coupling oil reservoir, design adapts to the optimum well pattern of this oil reservoir, obtains being conducive to the parameter such as trend of the optimum well pattern form of engineering staff's reference, the spacing of oil-water well, angle, well pattern.

2, propose the construction method of element pattern, reached the object of well pattern change by the adjustment of 6 structure parameters.

Gradient method for solving is often calculated the partial derivative of a parameter just needs a numerical reservoir simulation, and actual reservoir simulation once normal several hours consuming time even several days, therefore parameter too much will affect the speed of optimization greatly.By simplifying, the present invention proposes 6 parameter well pattern construction methods, only needing 6+1 reservoir simulation just can obtain the gradient of well pattern parameter.

3, well net optimization and note are adopted optimization to combine, optimize the optimum note amount of adopting under different well pattern form.

Different well patterns requires that it is also not identical for noting the condition of adopting, and coordinate above-mentioned 6 parameters, the present invention considers total note amount of adopting of oil reservoir, obtains its corresponding optimum note and adopt total amount while optimized well pattern form.

4, can realize optimizing to multiple rule well pattern, comprise 4 points, 5 points, 7 points, nine-spot well network.

The method that parameter builds not only is applicable to modal five-spot network, and for other 4 points, 7 and nine-spot well network, the present invention adopts the structure of 6 parametric techniques equal energy well pattern, and is optimized it.

Accompanying drawing explanation

Fig. 1 is the self-adaptation well net optimization method flow schematic diagram that the present invention is based on gradient algorithm;

Fig. 2 is five-spot network schematic diagram

Fig. 3 is seven spot pattern schematic diagram

Fig. 4 is nine-spot well network schematic diagram

Fig. 5 is five-spot network shear transformation schematic diagram

Fig. 6 is seven spot pattern shear transformation schematic diagram

Fig. 7 is five-spot network rotational transform schematic diagram

Fig. 8 is seven spot pattern rotational transform schematic diagram

Fig. 9 is five-spot network scaling conversion schematic diagram

Figure 10 is seven spot pattern scaling conversion schematic diagram

Figure 11 is five-spot network translation transformation schematic diagram

Figure 12 is seven spot pattern translation transformation schematic diagram

Figure 13 is five-spot network deformation schematic diagram

Figure 14 is that multiple-unit combines five-spot network schematic diagram mutually

Figure 15 is well location and mesh fitting schematic diagram

Figure 16 is three-dimensional heterogeneous reservoir geologic model schematic diagram

Figure 17 is the five-spot network cell schematics under initial controling parameters

Figure 18 is the five-spot network three-dimensional geological model schematic diagram under initial controling parameters

Figure 19 is the five-spot network schematic diagram under initial controling parameters

Figure 20 is five-spot network unit deformation schematic diagram

Figure 21 is final five-spot network well location distribution schematic diagram

Embodiment

Below in conjunction with accompanying drawing, the invention will be further described.

As shown in Figure 1, based on the self-adaptation well net optimization method of gradient algorithm, comprise the steps:

Step 1, obtains oil reservoir data, sets up three-dimensional digital geological model of oil accumulation, carry out numerical reservoir simulation.

Numerical reservoir simulation is the overall process presenting oil-field development on computers by digitized mode, set oil reservoir, fluid, well location are in one, can simulated oil, gas, water flowing in the earth formation, the water_bearing escalating rate in the distribution of pre-detecting residual oil, crude oil yield, oil field, preferred optimum development scheme, guide field is produced.

Concrete steps are as follows:

(11), the favourable area of oil-gas accumulation in oil reservoir is found out, the oil field block that selected well spacing is favourable;

From region, carry out the entirety investigation of oil reservoir, understand geologic aspects, find out the favourable area of oil-gas accumulation, and carry out the estimation of oil-gas geology reserves, selected range of profitability carries out cloth well development.

(12), reservoir static data is collected

In controlled oil-bearing area, beat information well, by well logging, core, drill stem testing, data acquired by well logging interpretation, obtain the lithology on stratum, factor of porosity, permeability and oil-containing degree of saturation parameter, carry out detailed Strata Comparison, make character and the distribution of oil reservoir clear, collect that mouth coordinate, hole-deviation correction data, reservoir top are dark, individual-layer data, layer data.

(13) three-dimensional digital geological model of oil accumulation, is set up

Geology, well logging, geophysical information and various explanation results or conceptual model are combined and generates three-dimensional digital geological model of oil accumulation.

(14), reservoir behavior data is collected

Obtain reservoir fluid and rock core, obtain reservoir fluid (component) parameter, petrophysical parameter by experiment, determine oil reservoir starting condition (water-oil interface, oil gas interface, pressure gradient) and Production development data (perforation completion data, formation testing and pilot production data, injection-production well dynamic data: day produce oil, day production fluid, daily gas, day water filling, sand face pressure).

(15), integrate static data and dynamic data, build numerical reservoir simulation model, develop analog computation

Arrange the static data and dynamic data that obtain, provide profit production history, build numerical reservoir simulation model, carry out the flow simulating calculating of oil reservoir development oil, gas, water.Here the Reservoir simulation model built just is similar to have been moved to real oil reservoir on computer, is a kind of digitized oil reservoir, its be imbedded in the degree of depth of underground, the position of drilling well, the attribute of each layer, fluid attribute all consistent with actual oil reservoir.On this basis, at different position well spacing, can design different well pattern forms, the ratio of amendment oil-water well well number, finds out optimum well pattern form by analog development.

Step 2, builds self-adaptation element pattern, generates well pattern.

Geometric well pattern comprises various ways, the most typically comprise 4 points, 5 points, and nine-spot well network at 7, often kind of well pattern is made up of multiple identical element pattern, therefore, only need the angle between the length of side of the single element pattern of optimization, position angle and the length of side, just can determine the well pattern layout of this oil reservoir optimum.This kind of well pattern is according to the difference of oil reservoir, and the optimum element pattern size, well number, the injection-to-producing well ratio that are automatically adapted with it, so be a kind of new self-adaptation well pattern.Its concrete steps are:

(21), select well pattern type, estimate initial parameter, adopt 6 strat test net unit construction methods, generate initial self-adaptation element pattern

6 parameters are selected to build well pattern form: element pattern size changing amount (Δ a, Δ b), element pattern angle (θ), finite element rotation angle (γ), transverse translation distance (Δ x) and longitudinal translation distance (Δ y), the multiple evolving form of element pattern scaling, rotation, shearing and translation can be realized by these 6 parameters, realize the self-adaptation to oil reservoir.Because regular well pattern type is more, choose the Constructed wetlands that most typical rectangular pattern (five-spot network) and triangular pattern (seven spot pattern) describe element pattern here.

(211), well pattern construction method

According to reservoir characteristics, according to the experience of site operation personnel, 6 basic parameters of setup unit center point coordinate (x, y) and element pattern, determine the citation form of well pattern, as in Figure 2-4.Because the target of the present invention optimizes to obtain optimum well pattern, so require that this well pattern form can regulate, can convert.According to the geometry of area regular well pattern, if 6 parameters change, can derive four kinds of mapping modes: rotate, shear, scaling and translation, can be optimized these parameters like this, change, so that obtain better well pattern form.Mapping mode is defined as follows:

1. shear

The lateral deformation of element pattern of the present invention is realized by shearing inclination, after generation lateral shear angle γ, and the A of five-spot network ^*, B ^*point there occurs sidesway, seven spot pattern by 2 ^*, 3 ^*, 5 ^*with 6 ^*also there occurs sidesway, achieve the structure of different shearing inclination well pattern like this, after detrusion, the geometric configuration of well group changes, as seen in figs. 5-6.

2. rotate

Orientation and the trend of well pattern can not be constructed completely by detrusion, also need the rotation in conjunction with well pattern, as: after well pattern is rotated in a clockwise direction θ, original well pattern can develop becomes new well pattern form, but basic geometric configuration can not change, as Figure 7-8.

3. scaling

By increasing or reduce element pattern size a and b, well group just freely can convert on yardstick, and five-spot network becomes A from initial well net unit A-D ^*-D ^*, seven spot pattern becomes 1 from 1-6 ^*-6 ^*, achieve the structure of different scale well pattern like this, as shown in figs. 9-10.

4. translation

After experiencing above-mentioned 3 kinds of conversion, element pattern is still and is in original place generation deformation, is not subjected to displacement, and in order to all features making the well pattern built meet differentiation, the variable building well pattern adds and moves in parallel distance, delta x and vertically move distance, delta y.Well group can do the movement on planimetric position, and its geometric configuration can not change, as depicted in figs. 11-12.

(212), estimate initial value, generate initial well pattern

Describe based on above-mentioned well pattern conversion, according to reservoir characteristics, design initial well pattern, estimate 6 suitable initial value u of parameter ₀={ a ₀, b ₀, Δ x ₀, Δ y ₀, θ ₀, γ ₀, complete the structure of initial well pattern, describe building process for five-spot network.As shown in Figure 6, round dot is producing well to five-spot network, and triangle form point is water injection well.Wherein, the well pattern AB length of side is 2a, the AC length of side be 2b, γ is the angle of shear, and θ is rotation angle.

According to the basic understanding of field engineering personnel to underground situation, select the minimum direction of permeability as reference orientation dir-a, based on reservoir physics experiment and result of log interpretation, obtain the seepage parameters k on direction dir-a and direction dir-b _a, φ _a, k _rwfaand k _b, φ _b, k _rwfb, roughly estimate direction dir-a according to on-site experience, the incorgruous injection production pressure difference Δ p in dir-b both direction _a, Δ p _a, due in oil reservoir development, the incorgruous injection production pressure difference under regular situation in well group in all directions all relatively, also can be similar to and think that injection production pressure difference is identical in both direction, be i.e. Δ p _a=Δ p _b.According to Single Well Economic limit control reserve, computing economy critical spacing, and determine the well spacing L on reference orientation dir-a according to this _a(L _abe more than or equal to economic limit well space), then according to formula determine the well spacing L on dir-b _b, thus obtain the approximate value a of five-spot network unit size ₀≈ L _a, b ₀≈ L _b.Other 4 initial parameter Δ x ₀, Δ y ₀, θ ₀, γ ₀obtain according to actual field situation setting.

Pass through shearing and rotational transform by basic well pattern according to following formula, five-spot network is transformed to A by ABDCO ^*b ^*d ^*c ^*o ^*, as shown in figure 13.

Known C point well location coordinate is: $\left\{\begin{array}{c}{x}_c=x_0+{\mathrm{\Δx}}_0\\ y_c=y_0+{\mathrm{\Δy}}_0\end{array}\right.---\left(1\right)$

A ^*point coordinate is: $\left\{\begin{array}{c}{x}_{A*}=x_c+\frac{2(b+b_0)}{\cos \mathrm{\γ}}\sin (\mathrm{\θ}+{\mathrm{\θ}}_0+\mathrm{\γ}+{\mathrm{\γ}}_0)\\ y_{A*}=y_c+\frac{2(b+b_0)}{\cos \mathrm{\γ}}\cos (\mathrm{\θ}+{\mathrm{\θ}}_0+\mathrm{\γ}+{\mathrm{\γ}}_0)\end{array}\right.---\left(2\right)$

D ^*point coordinate is: $\left\{\begin{array}{c}{x}_{D*}=x_c+2(a+a_0)\×\cos (\mathrm{\θ}+{\mathrm{\θ}}_0)\\ y_{D*}=y_c-2(a+a_0)\×\sin (\mathrm{\θ}+{\mathrm{\θ}}_0)\end{array}\right.---\left(3\right)$

Summit well C ^*with centerwell O ^*coordinate close be: $\left\{\begin{array}{c}{x}_{o*}=x_c+\sqrt{{(b+b_0)}^2+{(a+a_0+(b+b_0)\tan r)}^2}\sin (\mathrm{\θ}+{\mathrm{\θ}}_0)\\ y_{o*}=y_c+\sqrt{{(b+b_0)}^2+{(a+a_0+(b+b_0)\tan r)}^2}\cos (\mathrm{\θ}+{\mathrm{\θ}}_0)\end{array}\right.---\left(4\right)$

In formula, ${\mathrm{\θ}}_0=\arctan \left(\frac{a+a_0+(b+b_0)\tan \mathrm{\γ}}{b+b_0}\right)$

After A, C and D point determines, B point also just can directly obtain, and the relative coordinate relational expression of the five-spot network of whole oil reservoir, obtains initial well pattern by that analogy.

(22), based on initial well net unit, in oil reservoir region, carry out spread, generate self-adaptation well pattern

In order to meet the demand of all kinds of oil reservoir shape size, with the top peak of oil reservoir for reference position, according to the element pattern structure generated, to surrounding spread, calculate the coordinate of each injection-production well, construct the well pattern form that multiple unit combines mutually, as shown in figure 14.

(23) well pattern that, will generate mates commercial reserve simulator Corner-point Grids, setting well location

The regular well pattern of above-mentioned generation, is only the geometry obtaining well pattern, is namely aware of the terrestrial coordinate of every mouthful of well.Want to evaluate it, must exploit again according to the deployment drilling well of well location, its quality can be evaluated.But can only bore a well in reality, scheme can not arbitrarily be revised, so be arranged into by well location in the digitized numerical reservoir simulation model of the 1st step generation here, analog development, evaluates its quality.

During numerical reservoir simulation, each mouth well well location is described by grid, that is: the terrestrial coordinate of A1 well is in (1,1,1) individual grid, and that just describes the position in well place with this grid.So before numerical simulation, all wells are needed to mate with numerical simulator according to terrestrial coordinate, calculate the terrestrial coordinate scope corresponding to each grid, if the terrestrial coordinate of well location falls among the scope of certain grid, this grid is exactly that this mouthful of well bores the position of meeting.The grid of all wells is confirmed, finally obtains the position that each well is corresponding in simulator, as shown in figure 15.

Complete the generation of whole well pattern form like this, as long as arrange 6 different parameters, just can generate a set of well pattern form matched with it in numerical simulation.So in step described later, by optimizing this 6 parameters, generating different well patterns, then evaluating, just can obtain optimum well location layout.

Step 3, computing economy benefit, evaluates initial well pattern.

After the initial well pattern of generation, foundation reservoir model and the well pattern layout obtained, total note amount of adopting that setting is suitable, commercial reserve numerical simulation software (as Eclipse or CMG) simulation oil field is utilized to develop, record its oil offtake, water production rate and water injection rate, consider the drilling cost of every mouthful of well, computing economy benefit (economic net present value NPV), evaluate the quality of well pattern.Economic net present value expression formula is as follows: $\mathrm{NPV}=\mathrm{\Σ}\frac{r_o{q}_o-r_{\mathrm{pw}}{q}_{\mathrm{pw}}-r_{\mathrm{iw}}{q}_{\mathrm{iw}}}{{(1+b)}^t}-(N_o{C}_o+N_i{C}_w)---\left(5\right)$

In formula: NPV is production net present value (NPV), unit;

R _ofor crude oil price, yuan/ton;

R _pwfor producing water cost price, unit/cubic meter;

R _iwfor water filling cost price, unit/cubic meter;

Q ₀for tired oil offtake, ton/sky;

Q _pwfor tired water production rate, cubic meter/sky;

Q _iwfor tired water injection rate, cubic meter/sky;

T is the production time, year;

B is annual rate, decimal;

N _ofor producing well number, mouth;

N _ifor water injection well number, mouth;

C _ofor boring producing well cost, unit/mouth;

C _wfor boring water injection well cost, unit/mouth.

Step 4, setup parameter, for well net optimization is prepared.

(41) the maximum and minimum value of well pattern adjustable yardstick, is set.According to oil field actual conditions, need the scope setting well pattern adjustable yardstick, parameter comprises 6 element patterns structure parameters and note adopts total amount.

(42) step-length of total simulated time and offset well network parameters, is set.

Step 5, uses gradient method, optimizes cell formation parameter, generates new well pattern.

There are initial value and the boundary condition of 7 parameters, disturbance gradient method can have been adopted to obtain new parameter value, when its principle is exactly partial derivative=0, obtain maximal value.Process is: partial derivative NPV being asked to each parameter, this partial derivative is exactly the Grad often mentioned, new well pattern variable is generated after obtaining gradient, analog computation obtains new NPV value again, ask partial derivative again, back and forth calculate with this, when final partial derivative is 0, obtain maximum NPV value, well pattern parameter is now exactly optimum parameter.Reciprocal iterations can set 5 times, even if partial derivative is not 0, also can ensures that the well pattern now obtained is better than initial well pattern, can obtain better economic benefit.Concrete methods of realizing is:

(51), distinguish disturbance 6 element patterns structure Optimal Parameters, ask production economy net present value (NPV) relative to the partial derivative of each parameter

Can find according to partial derivative computing formula, when Δ u → 0, namely well pattern parameter u ₁with u ₂time closely, utilize the production net present value (NPV) NPV calculated ₁and NPV ₂just can calculate the partial derivative of relevant parameter, because there are 7 parameters, so need 8 numerical simulations, respectively with the NPV of rear 7 analog computations ₂deduct the NPV of last analog computation ₁value, obtains the partial derivative of 7 parameters: $\frac{\∂\mathrm{NPV}}{\∂u}=\underset{\mathrm{\Δu}\→0}{\lim }\frac{\mathrm{\ΔNPV}}{\mathrm{\Δu}}=\underset{\mathrm{\Δx}\→0}{\lim }\frac{{\mathrm{NPV}}_2-{\mathrm{NPV}}_1}{u_2-u_1}\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\left(6\right)$

(52), upgrade well pattern and build parameter

According to method of steepest descent, after obtaining gradient, only need given certain step-length (as α=10) just can generate new parametric variable: $u^{n+1}=u^n+\mathrm{\α}{\left(\frac{\∂\mathrm{NPV}}{\∂u}\right)}^n\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\·\left(7\right)$

(53), generate new well pattern, search better well pattern form

By variable u ⁿ⁺¹={ Δ a ⁿ⁺¹, Δ b ⁿ⁺¹, Δ x ⁿ⁺¹, Δ y ⁿ⁺¹, θ ⁿ⁺¹, γ ⁿ⁺¹, WIPR ⁿ⁺¹generate new well pattern form and Reservoir simulation model, utilize business simulator analog development, calculate new NPV value, calculating step as more last in NPV change is large, and then well pattern is better, as diminished, go to (2) step, shorten step-length (as α=2), then build new well pattern calculating NPV value, until NPV increases.

Step 6, calculates and evaluates the economic benefit of new well pattern, then interative computation, until generate optimum well pattern (61) if benefit improves, repeats step 2-6, until be tending towards 0, till economic benefit no longer increases, well pattern now regards as optimum well pattern.

(62) if benefit is deteriorated, shorten step-length, repeat step 5-6, until benefit improves; After this, then repeat step 2-6, until generate optimum well pattern.

The numerical reservoir simulation data file of final generation just demonstrates the well pattern layout after optimization, comprise the type (water injection well, producing well) of the number of well in oil reservoir, well, the physical location of well in oil reservoir, the economic net present value under the note amount of adopting of every mouthful of well and corresponding well pattern layout.Further, utilize 7 parameters optimized and obtain, the element pattern size corresponding to optimum well pattern, the angle of element pattern, position angle, the residing optimum planimetric position of unit can be calculated, according to oil field actual conditions, can carry out correcting rear field conduct.

Specific embodiment:

According to said method, it is as follows that certain oil field comprises step based on the self-adaptation well net optimization method of gradient algorithm:

Step 1, obtains oil reservoir data, sets up three-dimensional digital geological model of oil accumulation, carry out numerical reservoir simulation.

Selected representational theoretical model example builds geological model of oil accumulation, whole oil reservoir divides a set of straticulate exploration, collect dynamic static data data such as () oil reservoir, fluid, oil-water wells, set up three-dimensional heterogeneous reservoir numerical simulator, comprise 50 × 50 × 1 grid altogether, sizing grid is 7.5 × 7.5 × 6 meters, as shown in figure 16.

Step 2, builds adaptation rule element pattern, generates well pattern.

Five-spot network: setting element pattern size (a=75 rice, b=52.5 rice), element pattern size changing amount (Δ a=0, Δ b=0), element pattern angle (θ=0), finite element rotation angle (γ=0), transverse translation distance (Δ x=0) and longitudinal translation distance (Δ y=0), namely the five-spot network unit under initial parameter control, as shown in figure 17.

Five-spot network unit grid well location coordinate is:

C point coordinate is: (x _c, y _c)=(21,21)

A point coordinate is: $\left\{\begin{array}{c}{x}_A=21+\frac{2(5+0)}{\cos 0}\sin (0+0)=21\\ y_A=21+\frac{2(5+0)}{\cos 0}\cos (0+0)=31\end{array}\right.$

D point coordinate is: $\left\{\begin{array}{c}{x}_D=21+2(5+0)\×\cos 0=31\\ y_D=21-2(5+0)\×\sin 0=21\end{array}\right.$

The coordinate of O is:

B point coordinate is: (x _b, y _b)=(31,31)

In formula,

The relative coordinate relational expression of the five-spot network of whole oil reservoir, obtains initial well pattern by that analogy.

Calculate the well location coordinate of five-spot network each mouth well according to aforementioned five dot element construction methods and actual well location coordinate and numerical reservoir simulation mesh coordinate are matched, obtaining whole initial five-spot network, as depicted in figs. 18-19.

Step 3, computing economy benefit, evaluates initial well pattern.

The price of setting crude oil is 4500 yuan/ton, and producing water cost price is 400 yuan/cubic metre, and the cost boring producing well and brill water injection well is 6.5 × 10 ⁵unit/mouth, does not consider water filling cost.Total production time is 10 years, and annual rate is 0, and producing well number is 24 mouthfuls; N _ifor water injection well number is 15 mouthfuls, tired oil offtake is 1.054 × 10 ⁵ton, tired water production rate is 6.286 × 10 ⁴cubic meter, tired water injection rate is 1.890 × 10 ⁵cubic meter, the economic net present value being calculated initial well pattern by formula (5) is NPV ₁=2.4 × 10 ⁸unit.

Step 4, setup parameter, for well net optimization is prepared.

According to on-the-spot actual, set maximum well pattern size (a _max=190 meters, b _max=190 meters), element pattern angle (θ=[-90 °, 90 °]) and the anglec of rotation (γ=[-60 °, 60 °]), transverse translation distance (Δ x _max=187.5 meters) and longitudinal translation distance (Δ y _max=187.5 meters), maximum note adopts total amount (WIPR _max=190 cubic metres/day).

According to initial value and the boundary condition of 7 parameters, five-spot network is according to shearing, rotation, translation, convergent-divergent principle generation deformation, five-spot network unit presents new form (Figure 20), meanwhile, expands to whole oil reservoir and obtains new five-spot network layout (Figure 21).

Step 5, uses gradient method, optimizes cell formation parameter, generates new well pattern.

Calculate the economic net present value NPV of new five-spot network ₂=5.2 × 10 ⁸unit, is asked the partial derivative of each parameter to NPV by formula (6): $\frac{\∂\mathrm{NPV}}{\∂\mathrm{\Δa}}=\frac{5.2\×{10}^8-2.4\×{10}^8}{11.3-0}=2.48\×{10}^7,$ $\frac{\∂\mathrm{NPV}}{\∂\mathrm{\Δb}}=\frac{5.2\×{10}^8-2.4\×{10}^8}{12.5-0}=2.24\×{10}^7$ $\frac{\∂\mathrm{NPV}}{\∂\mathrm{\Δx}}=\frac{5.2\×{10}^8-2.4\×{10}^8}{1200-0}=2.33\×{10}^5$ $\frac{\∂\mathrm{NPV}}{\∂\mathrm{\Δy}}=\frac{5.2\×{10}^8-2.4\×{10}^8}{-803-(-241)}=-4.98\×{10}^5$ $\frac{\∂\mathrm{NPV}}{\∂\mathrm{\θ}}=\frac{5.2\×{10}^8-2.4\×{10}^8}{90-38}=5.38\×{10}^6$ $\frac{\∂\mathrm{NPV}}{\∂\mathrm{\γ}}=\frac{5.2\×{10}^8-2.4\×{10}^8}{60-34}=1.08\×{10}^7$ $\frac{\∂\mathrm{NPV}}{\∂\mathrm{WIPR}}==\frac{5.2\×{10}^8-2.4\×{10}^8}{933-673}=1.08\×{10}^6$

After obtaining this gradient, according to the step-length (α=2.5) of setting, utilize method of steepest descent calculate new well pattern parameter: element pattern size changing amount (Δ a=20 rice, Δ b=24 rice), element pattern angle (θ=90 °) and the anglec of rotation (γ=60 °), transverse translation distance (Δ x=1200 rice) and longitudinal translation distance (Δ y=-803 rice), note adopt total amount (WIPR=147.7 cubic meter/sky).

Step 6, calculates and evaluates the economic benefit of new well pattern, then iterative computation, until generate optimum well pattern.

This give result of calculation after repetition step 2-6, carry out 4 iterative computation altogether, table 1 gives well pattern controling parameters data during the 1-4 time well pattern distortion, and the net present value (NPV) that table 2 gives 5 points, 7 and nine-spot well network are optimized increases data.

Table 1

Table 2

The numerical reservoir simulation data file of final generation just demonstrates the well pattern layout after optimization, comprise the type (water injection well, producing well) of the number of well in oil reservoir, well, the physical location of well in oil reservoir, the economic net present value under the note amount of adopting of every mouthful of well and corresponding well pattern layout.And, utilize 7 parameters optimized and obtain, the parameter such as angle, position angle, the residing optimum planimetric position of unit of the element pattern size corresponding to optimum well pattern, element pattern can be calculated, according to oil field actual conditions, can carry out correcting rear field conduct.As seen from Figure 21, the final information of five-spot network of this example is: producing well 2 mouthfuls, mesh coordinate in oil reservoir is (8,7) and (40,47), water injection well 2 mouthfuls, mesh coordinate in oil reservoir is (2,36) and (47,17), corresponding Liquid output is 147.7 cubic metres/day, and economic net present value is NPV ₁=8.06 × 10 ⁸unit.Based on the above results: less injection-production well number effectively reduces drilling cost, with degree of saturation profiles versus before and after Figure 19, mother oil displacement is respond well, this makes it possible to ensure higher oil recovery factor, the more crude oil of extraction, thus obtain good economic benefit, therefore, namely above data message can be used as the theoretical foundation of this oil reservoir development design proposal.

Claims (7)

1., based on a self-adaptation well net optimization method for gradient algorithm, it is characterized in that, comprise the steps:

(1), obtain oil reservoir data, set up three-dimensional digital geological model of oil accumulation, carry out numerical reservoir simulation;

(2), build adaptation rule element pattern, generate well pattern; Specifically comprise the steps:

(21), select well pattern type, estimate initial parameter, adopt 6 strat test net unit construction methods, generate initial self-adaptation element pattern;

6 parameters refer to well spacing, array pitch, rotation angle, the angle of shear, transverse direction and longitudinal translation distance, use them can obtain the different geometric deformation of unit, by optimizing this 6 parameters, indirectly achieve the optimization of self-adaptation well pattern;

(22), based on initial self-adaptation element pattern, in oil reservoir region, carry out spread, generate self-adaptation well pattern;

(23) well pattern that, will generate mates commercial reserve simulator Corner-point Grids, setting well location;

(3), computing economy benefit, evaluate initial well pattern;

(4), setup parameter, for well net optimization is prepared;

(5), use gradient method, optimize cell formation parameter, generate new well pattern; Concrete steps are as follows:

(51), distinguish disturbance 6 element patterns structure Optimal Parameters, ask production economy net present value (NPV) relative to the partial derivative of each parameter;

(52), upgrade well pattern and build parameter;

(53), generate new well pattern, search better well pattern form;

(6), calculate and evaluate the economic benefit of new well pattern, then interative computation, until generate optimum well pattern.

2. the self-adaptation well net optimization method based on gradient algorithm according to claim 1, it is characterized in that, described step 1 specifically comprises the steps:

(11), the favourable area of oil-gas accumulation in oil reservoir is found out, the oil field block that selected well spacing is favourable;

(12), reservoir static data is collected;

(13) three-dimensional digital geological model of oil accumulation, is set up;

(14), reservoir behavior data is collected;

(15), integrate static data and dynamic data, build numerical reservoir simulation model, develop analog computation.

3. the self-adaptation well net optimization method based on gradient algorithm according to claim 2, it is characterized in that, well pattern type comprises four-spot pattern, five-spot network, seven spot pattern and nine-spot well network.

4. the self-adaptation well net optimization method based on gradient algorithm according to claim 3, it is characterized in that, the concrete steps of step 3 are as follows: foundation reservoir model and the well pattern layout obtained, total note amount of adopting that setting is suitable, utilize commercial reserve numerical simulation software simulation oil field to develop, record its oil offtake, water production rate and water injection rate, consider the drilling cost of every mouthful of well, computing economy benefit, evaluates the quality of well pattern.

5. the self-adaptation well net optimization method based on gradient algorithm according to claim 4, it is characterized in that, the concrete steps of step 4 are as follows:

(41), the maximum and minimum value of setting well pattern adjustable yardstick, setup parameter comprises 6 element patterns and builds parameters and note adopts total amount;

(42) step-length of total simulated time and offset well network parameters, is set.

6. the self-adaptation well net optimization method based on gradient algorithm according to claim 5, it is characterized in that, the concrete steps of step 6 are as follows:

(61) if benefit improves, step (2)-(6) are repeated, until generate optimum well pattern;

(62) if benefit is deteriorated, shorten step-length, repeat step (5)-(6), until benefit improves; After this, then repeat step (2)-(6), until generate optimum well pattern.

7. the self-adaptation well net optimization method based on gradient algorithm according to claim 6, it is characterized in that: in described 6 strat test net unit construction methods, 6 parameters refer to well spacing, array pitch, rotation angle, the angle of shear, transverse direction and longitudinal translation distance, use them can obtain the different geometric deformation of unit, by optimizing this 6 parameters, indirectly achieve the optimization of self-adaptation well pattern.