Simple multiphase flow production measurement method and system based on power diagram instrument and water content meter

Abstract

The present invention discloses a simple multiphase flow production metering method based on a dynamometer and a water content meter. First, a GCN model is constructed, trained and tested based on data to obtain a liquid production metering model. Then, a dynamometer and a water content meter are installed at the wellhead of the oil well that needs to be metered to obtain real-time displacement data X, real-time load data Y, real-time water content WLR and real-time gas content GVF, thereby predicting the real-time liquid production, oil production, water production, gas production, and total liquid production, oil, water, gas production, and average water content and gas content. The significant effect of the present invention is that it relies on a dynamometer and a water content meter with a single function and low cost, and proposes a metering method for the flow / production and component content of the oil well output fluid based on the GCN algorithm, which is significantly different from the traditional metering method and has the advantages of being timely, accurate, convenient and low-cost.

Description

Technical Field

[0001] The invention relates to the collection, processing and application of information in the oil field production process, and in particular to a virtual metering method for oil field output. Background Art

[0002] During oilfield exploitation, under current or expected flow conditions, in order to understand changes in formation oil and gas content, it is necessary to continuously measure the oil well output fluid flow rate (production) and the content of each component therein and provide real-time measurement data. The real-time measurement data of the oil well output fluid flow rate (production) and the content of each component are crucial for obtaining the reservoir production information required to determine production strategies and production allocation.

[0003] There are three traditional ways to measure the oil well output flow rate (production) and the content of each component:

[0004] First, the separator glass tube is used for measurement, and its main disadvantages are: the separator glass tube oil measuring device has large measurement errors for oil wells with high water content and little associated gas, as well as oil wells with low liquid volume and intermittent oil production, and cannot achieve the measurement accuracy required by the specifications; at the same time, the oil measurement operation, reading the glass tube liquid level, and the card meter of this measurement method are all performed manually, with many human factors and large measurement errors; finally, the valve needs to be opened and closed many times during the measurement process, the labor intensity of the workers is high, it is difficult to achieve long-term continuous measurement, the real-time performance of the data is poor, and there is a possibility of human false reporting.

[0005] The second method is to use a tipping bucket weighing and metering device for metering. This metering method is affected by sand, liquid and gas, which causes corrosion and stagnation of the tipping bucket bearing, inflexible rotation or failure of the tipping bucket to turn over, and increased leakage. Secondly, the gas disturbance may cause the tipping bucket to turn over by mistake. When measuring high-yield liquid wells, the impact of the liquid volume on the tipping bucket may cause the tipping bucket to turn over in advance. At the same time, factors such as incomplete oil and gas separation and foam will affect the accuracy of measurement. In addition, the cost of the tipping bucket weighing and metering device is high.

[0006] The third is to use the gamma-ray based multiphase flow meter proposed in recent years for measurement. However, due to its powerful functions and high integration, it also has problems such as high price, large size, inconvenient movement and redundant functions. Summary of the invention

[0007] In view of this, in order to meet the single demand of measuring the flow rate (production) and the content of each component of the oil well produced fluid, the present invention proposes a virtual metering method based on graph neural network (GCN) to meet the requirements of timeliness, accuracy, convenience and low cost of measurement.

[0008] Specifically, the present invention provides a simple multiphase flow production measurement method based on a work graph instrument and a water content meter, and the technical solution is as follows:

[0009] Step 1: Obtain historical liquid production data for several wellheads of oil wells The corresponding historical displacement data X0 and historical load data Y0 are collected by the dynamometer;

[0010] Step 2: Take the historical displacement data X0 and the corresponding historical load data Y0 as input feature data, and the corresponding historical liquid production data For the output target data, the node data required for GCN modeling is constructed; then based on the working condition categories corresponding to the historical wellhead displacement data X0 and the historical load data Y0, the node relationship is established, and the GCN network model is constructed for training and testing to obtain the liquid production measurement model;

[0011] Step 3: Install a work graph on the pumping unit at the wellhead of the oil well that needs to be measured to obtain real-time displacement data X and real-time load data Y;

[0012] A water content meter is installed on the outlet pipe of the oil pump at the wellhead of the oil well that needs to be measured to obtain the real-time water content WLR and real-time gas content GVF;

[0013] Step 4: Input the real-time displacement data X and the corresponding real-time load data Y into the liquid production measurement model to obtain the predicted real-time liquid production Q l ;

[0014] Step 5: Output simulation calculation:

[0015] The predicted real-time oil production Q is obtained as follows: o :

[0016] Q o =Q l ×(1-WLR);

[0017] The predicted real-time water production Q is obtained by the following formula w :

[0018] Q w =Q l ×WLR;

[0019] The predicted real-time gas production Q is obtained by the following formula g :

[0020] Q g =Q l ×GVF / (1-GVF);

[0021] The predicted total output Q is obtained as follows iD :

[0022]

[0023] i=l,o,w,g;

[0024] j=1,2,3,...,N;

[0025] Q ij It indicates the jth predicted real-time production array when batch statistics are calculated;

[0026] N is the total number of predicted real-time production arrays when batch statistics are calculated;

[0027] max(Q ij ) is the maximum value of the predicted real-time output array when batch statistics are calculated;

[0028] min(Q ij ) is the minimum value of the predicted real-time output array when batch statistics are calculated;

[0029] The average water content WLR is obtained by the following formula D :

[0030]

[0031] j=1,2,3,...,N;

[0032] WLR j is the jth real-time moisture content when batch statistics are calculated;

[0033] max(WLR j ) is the maximum value of the real-time moisture content when batch statistics are calculated;

[0034] min(WLR j ) is the minimum value of real-time moisture content when batch statistics are calculated;

[0035] The average gas content GVF is obtained by the following formula D :

[0036]

[0037] GVF j is the jth real-time gas content during batch statistics calculation;

[0038] max(GVF j ) is the maximum value of the real-time gas content when batch statistics are calculated;

[0039] min(GVF j ) is the minimum value of the real-time gas content when batch statistics are calculated.

[0040] Compared with the prior art, the present invention relies on a dynamometer and a water content meter and proposes a new method for measuring the flow rate / production and component content of oil well produced fluid based on GCN. DETAILED DESCRIPTION

[0041] The present invention will be further described below with reference to the embodiments.

[0042] A simple multiphase flow production measurement method based on a work graph instrument and a water content meter is carried out in the following steps:

[0043] Step 1: Obtain historical liquid production data for several wellheads of oil wells The corresponding historical displacement data X0 and historical load data Y0 are collected by the dynamometer;

[0044] Step 2: Take the historical displacement data X0 and the corresponding historical load data Y0 as input feature data, and the corresponding historical liquid production data For the output target data, the node data required for GCN modeling is constructed; specifically, a two-layer GCN network is constructed, and the two-layer network structures are {(2, 16), (16, 1)}, relu is used as the activation function, L1 is selected as the loss function, and the Adam optimization algorithm is used to prevent overfitting, and the GCN model is trained and tested;

[0045] Then, based on the working condition categories corresponding to the wellhead historical displacement data X0 and the historical load data Y0, a node relationship is established. Specifically, a historical dynamometer diagram is obtained by drawing the historical displacement data X0 and the corresponding load data Y0, and based on the working conditions of the historical dynamometer diagram, a node relationship of a normal working condition and a node relationship of an abnormal working condition are established respectively;

[0046] The node relationships of the abnormal operating conditions include: the node relationship of oil well leakage, pipeline blockage, and high back pressure; the node relationship of insufficient liquid supply; the node relationship of piston jamming; the node relationship of gas influence; the node relationship of wax jamming; the node relationship of floating valve leakage; the node relationship of combined pumping and spraying; the node relationship of fixed valve leakage; the node relationship of piston decoupling; the node relationship of floating valve and fixed valve leakage; the node relationship of heavy oil; the node relationship of wax deposition; the node relationship of sucker rod breakage; the node relationship of upper impact on the pump; the node relationship of lower impact on the pump; and the node relationship of sand production.

[0047] Finally, the GCN network model was constructed for training and testing to obtain the liquid production measurement model;

[0048] Step 3: Install a work graph on the pumping unit at the wellhead of the oil well that needs to be measured to obtain real-time displacement data X and real-time load data Y;

[0049] A water content meter is installed on the outlet pipe of the oil pump at the wellhead of the oil well that needs to be measured to obtain the real-time water content WLR and real-time gas content GVF;

[0050] Step 4: Input the real-time displacement data X and the corresponding real-time load data Y into the liquid production measurement model to obtain the predicted real-time liquid production Q l ;

[0051] Step 5: Output simulation calculation:

[0052] The predicted real-time oil production Q is obtained as follows: o :

[0053] Q o =Q l ×(1-WLR);

[0054] The predicted real-time water production Q is obtained by the following formula w :

[0055] Q w =Q l ×WLR;

[0056] The predicted real-time gas production Q is obtained by the following formula g :

[0057] Q g =Q l ×GVF / (1-GVF);

[0058] The predicted total output Q is obtained as follows iD :

[0059]

[0060] i=l,o,w,g;

[0061] j=1,2,3,...,N;

[0062] Q ij It indicates the jth predicted real-time production array when batch statistics are calculated;

[0063] N is the total number of predicted real-time production arrays when batch statistics are calculated;

[0064] max(Q ij ) is the maximum value of the predicted real-time output array when batch statistics are calculated;

[0065] min(Q ij ) is the minimum value of the predicted real-time output array when batch statistics are calculated;

[0066] The predicted total production Q iD include:

[0067] Predicted total liquid production Q lD ,i=l;

[0068] Forecast total oil production QoD ,i=o;

[0069] Predicted total water production Q wD ,i=w;

[0070] Predicted total gas production Q gD ,i=g;

[0071] Therefore, the calculation formula It can be written as:

[0072]

[0073] Q lj It indicates the jth predicted real-time liquid production when batch statistics are calculated;

[0074] max(Q lj ) is the maximum value of the predicted real-time liquid production when batch statistics are calculated;

[0075] min(Q lj ) is the minimum value of the predicted real-time liquid production when batch statistics are calculated;

[0076] Q oj It indicates the jth predicted real-time oil production when batch statistics are calculated;

[0077] max(Q oj ) is the maximum value of the predicted real-time oil production when batch statistics are calculated;

[0078] min(Q oj ) is the minimum value of the predicted real-time oil production when batch statistics are calculated;

[0079] Q wj It indicates the jth predicted real-time water production when batch statistics are calculated;

[0080] max(Q wj ) is the maximum value of the predicted real-time water production when batch statistics are calculated;

[0081] min(Q wj ) is the minimum value of the predicted real-time water production when batch statistics are calculated;

[0082] Q gj It indicates the jth predicted real-time water production when batch statistics are calculated;

[0083] max(Q gj ) is the maximum value of the predicted real-time water production when batch statistics are calculated;

[0084] min(Q gj ) is the minimum value of the predicted real-time water production when batch statistics are calculated;

[0085] The Q lj , Q oj , Q wj , Q gj Collectively referred to as the jth predicted real-time production array Q ij ;

[0086] The max(Q lj )、max(Q oj )、max(Q wj )、max(Q gj ) are collectively referred to as the maximum value of the predicted real-time production array;

[0087] The min(Q lj )、min(Q oj )、min(Q wj )、min(Q gj ) are collectively referred to as the minimum value of the predicted real-time production array;

[0088] The average water content WLR is obtained by the following formula D :

[0089]

[0090] j=1,2,3,...,N;

[0091] WLR j is the jth real-time moisture content when batch statistics are calculated;

[0092] max(WLR j ) is the maximum value of the real-time moisture content when batch statistics are calculated;

[0093] min(WLR j ) is the minimum value of real-time moisture content when batch statistics are calculated;

[0094] The average gas content GVF is obtained by the following formula D :

[0095]

[0096] GVF j is the jth real-time gas content during batch statistics calculation;

[0097] max(GVF j ) is the maximum value of the real-time gas content when batch statistics are calculated;

[0098] min(GVF j ) is the minimum value of the real-time gas content when batch statistics are calculated.

[0099] According to the above method, the oil well production within a certain period of time can be measured.

[0100] As an oil field manager, what you are most concerned about is often the daily production of the oil wells. The daily production data of the oil wells is more instructive for oil well management and decision-making.

[0101] The daily production of oil wells that are continuously in production on that day is measured as follows:

[0102] In step 3 and step 4, the dynamometer obtains several real-time dynamometer diagrams, selects one real-time dynamometer diagram per minute, and inputs the real-time displacement data X and the real-time load data Y corresponding to one pumping cycle into the liquid production metering model to obtain the predicted real-time liquid production Q per minute. l ;

[0103] The water content meter measures a number of water content instrument values ​​and a number of gas content instrument values ​​every minute, and takes the average water content as the real-time water content WLR, and the average gas content as the real-time gas content GVF;

[0104] Then, the calculation is performed in step 5:

[0105] Predicted real-time oil production Q per minute o ;

[0106] Predicted real-time water production per minute Q w ;

[0107] Predicted real-time gas production per minute Q g ;

[0108] For the calculation of daily output, N is 1440, and the calculation is:

[0109] The total daily output forecast Q iD ;

[0110] Daily average water content WLR D ;

[0111] Daily average gas content GVF j .

[0112] The present invention will be further described below in conjunction with examples:

[0113] Select five oil wells a, b, c, d, and e that are in continuous production. Install a dynamometer and a water content meter at their wellheads. Use the above method to predict the total daily liquid production. The results are shown in the following table:

[0114]

[0115] in:

[0116] Prediction deviation = | actual liquid production - predicted total liquid production |;

[0117] Prediction error = prediction deviation / actual liquid production*100%.

[0118] It can be seen from the above table that the maximum prediction error of the liquid production of the five oil wells a, b, c, d, and e is 4.6%, the average prediction error of the five oil wells is 2.38%, and the daily liquid production prediction error is within 5%, which proves that the metering method of the present invention has good accuracy and can be applied to actual production.

[0119] The structures and installation methods of the dynamometer and the water content meter are well documented in the prior art and will not be described in detail here.

[0120] The corresponding system includes oil pump, oil pumping unit, dynamometer, water content meter and server;

[0121] The dynamometer is installed on the oil pump, and the dynamometer obtains the historical displacement data X0, the historical load data Y0, the real-time displacement data X and the real-time load data Y, and transmits them to the server; the water content meter is installed on the oil pump, and the water content meter obtains the real-time water content WLR and the real-time gas content GVF and transmits them to the server; the server is used to build a GCN network model for training and testing.

[0122] Beneficial effects: The present invention relies on a single-function, low-cost dynamometer and water content meter, and proposes a method for measuring the flow / production and component content of oil well output fluid based on the GCN algorithm. It is significantly different from traditional metering methods and has the advantages of being timely, accurate, convenient and inexpensive.

[0123] Finally, it should be noted that the above description is only a preferred embodiment of the present invention. Under the guidance of the present invention, ordinary technicians in this field can make various similar expressions without violating the purpose and claims of the present invention, and such changes all fall within the scope of protection of the present invention.

Claims

1. A simple multiphase flow production measurement method based on a power meter and a water content meter, characterized in that Follow these steps: Step 1: Obtain historical liquid production data for several wellheads of oil wells The corresponding historical displacement data X0 and historical load data Y0 are collected by the dynamometer; Step 2: Take the historical displacement data X0 and the corresponding historical load data Y0 as input feature data, and the corresponding historical liquid production data For the output target data, the node data required for GCN modeling is constructed; then based on the working condition categories corresponding to the wellhead historical displacement data X0 and the historical load data Y0, the node relationship is established, and the GCN network model is constructed for training and testing to obtain the liquid production measurement model; Step 3: Install a work graph on the pumping unit at the wellhead of the oil well that needs to be measured to obtain real-time displacement data X and real-time load data Y; A water content meter is installed on the outlet pipe of the oil pump at the wellhead of the oil well that needs to be measured to obtain the real-time water content WLR and real-time gas content GVF; Step 4: Input the real-time displacement data X and the corresponding real-time load data Y into the liquid production measurement model to obtain the predicted real-time liquid production Q l ; Step 5: Output simulation calculation: The predicted real-time oil production Q is obtained as follows: o : Q o =Q l ×(1-WLR); The predicted real-time water production Q is obtained by the following formula w : Q w =Q l ×WLR; The predicted real-time gas production Q is obtained by the following formula g : Q g =Q l ×GVF / (1-GVF); The predicted total output Q is obtained as follows iD : i=l,o,w,g; j=1,2,3,...,N; Q ij It indicates the jth predicted real-time production array when batch statistics are calculated; N is the total number of predicted real-time production arrays when batch statistics are calculated; max(Q ij ) is the maximum value of the predicted real-time output array when batch statistics are calculated; min(Q ij ) is the minimum value of the predicted real-time output array when batch statistics are calculated; The average water content WLR is obtained by the following formula D : j=1,2,3,...,N; WLR j is the jth real-time moisture content when batch statistics are calculated; max(WLR j ) is the maximum value of the real-time moisture content when batch statistics are calculated; min(WLR j ) is the minimum value of real-time moisture content when batch statistics are calculated; The average gas content GVF is obtained by the following formula D : GVF j is the jth real-time gas content during batch statistics calculation; max(GVF j ) is the maximum value of the real-time gas content when batch statistics are calculated; min(GVF j ) is the minimum value of the real-time gas content when batch statistics are calculated.

2. The simple multiphase flow production measurement method based on the power diagram instrument and the water content meter according to claim 1 is characterized in that: In the step 2, a historical dynamometer diagram is obtained by drawing the historical displacement data X0 and the corresponding load data Y0, and based on the working conditions of the historical dynamometer diagram, node relationships of normal working conditions and node relationships of abnormal working conditions are respectively established; The node relationships of the abnormal operating conditions include: the node relationship of oil well leakage, pipeline blockage, and high back pressure; the node relationship of insufficient liquid supply; the node relationship of piston jamming; the node relationship of gas influence; the node relationship of wax jamming; the node relationship of floating valve leakage; the node relationship of combined pumping and spraying; the node relationship of fixed valve leakage; the node relationship of piston decoupling; the node relationship of floating valve and fixed valve leakage; the node relationship of heavy oil; the node relationship of wax deposition; the node relationship of sucker rod breakage; the node relationship of upper impact on the pump; the node relationship of lower impact on the pump; and the node relationship of sand production.

3. The simple multiphase flow production measurement method based on the power diagram instrument and the water content meter according to claim 1 is characterized in that: In the step 2, a two-layer GCN network is constructed, and the two-layer network structures are {(2, 16), (16, 1)}, relu is used as the activation function, L1 is selected as the loss function, and the Adam optimization algorithm is used to prevent overfitting, and the GCN model is trained and tested.

4. The simple multiphase flow production measurement method based on the power diagram instrument and the water content meter according to claim 1 is characterized in that: In step 3 and step 4, the dynamometer obtains several real-time dynamometer diagrams, selects one real-time dynamometer diagram per minute, and inputs the real-time displacement data X and the real-time load data Y corresponding to one pumping cycle into the liquid production metering model to obtain the predicted real-time liquid production Q per minute. l ; The average water content measured by the water content meter every minute is used as the real-time water content WLR, and the average gas content is used as the real-time gas content GVF; Then, the calculation is performed in step 5: Predicted real-time oil production per minute Q o ; Predicted real-time water production per minute Q w ; Predicted real-time gas production per minute Q g ; N is 1440, and the calculation is: The daily predicted total production Q iD ; Daily average water content WLR D ; Daily average gas content GVF j .

5. The simple multiphase flow production metering system based on the dynamometer and the water content meter according to claim 1 is characterized in that: Including oil pump, oil pumping unit, dynamometer, water content meter and server; The dynamometer is installed on the oil pump, and the dynamometer obtains the historical displacement data X0, the historical load data Y0, the real-time displacement data X and the real-time load data Y, and transmits them to the server; the water content meter is installed on the oil pump, and the water content meter obtains the real-time water content WLR and the real-time gas content GVF and transmits them to the server; The server is used to build a GCN network model for training and testing.

Images