A method for predicting high-pressure physical property parameters of formation fluid in oil and gas reservoirs
Through the prediction model with the composition of fluid components as the input characteristic value, the problem of harsh experimental conditions and low efficiency in the prediction of high-pressure physical properties parameters of fluid in the oil and gas reservoir formation is solved, the accuracy and interpretability of the prediction results are improved, and reserve calculation and production are guided. practice.
Patent Information
- Application Number
- CN202111146793.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-09-28
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2041-09-28
AI Technical Summary
The existing high-pressure physical property parameter prediction methods for oil and gas reservoir formation fluids have harsh experimental conditions, long periods, low efficiency, and lack of interpretability in neural networks, resulting in deviations and irrationality of the prediction results.
Starting from a microscopic perspective, the fluid component composition is used as the input feature value of the prediction model, the prediction model is established through XGBoost, and the model is optimized using the cross-validation method until the correlation coefficient R2 is greater than 0.95, improving the correlation and interpretability of the prediction model.
A more accurate prediction of high-pressure physical properties of fluids in oil and gas reservoir formations has been achieved, experimental efficiency has been improved, and reserve calculation and production practice have been guided.
Smart Images

Figure CN114492153B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of petroleum and natural gas technology, and more specifically to a method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluid. Background Art
[0002] The high-pressure physical property parameters of oil and gas reservoir formation fluids are the basis for oil and gas reservoir reserve calculation, development plan preparation and plan adjustment.
[0003] The high-pressure physical properties of formation fluids in oil and gas reservoirs are mainly obtained through conventional PVT testing experiments, which have harsh experimental conditions, long physical simulation cycles, and low efficiency. With the rapid increase in the workload of oil and gas reservoir exploration and development experiments in my country, the problem of experimental efficiency has become increasingly prominent, and the existing PVT experimental equipment and personnel have been unable to meet the needs of oil and gas reservoir exploration and development.
[0004] In addition, the currently disclosed method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluids is to establish a prediction model through a neural network, with formation pressure, formation temperature, surface crude oil density, gas-oil ratio, and natural gas relative density as the input layer of the prediction model, and saturation pressure, crude oil dissolved gas-oil ratio, volume coefficient, formation crude oil density, and formation crude oil viscosity as the output prediction results of the prediction model. However, the high-pressure physical property parameters of oil and gas reservoir formation fluids are more closely related to the composition of fluid components, and the neural network lacks interpretability, which will lead to deviations and irrationality in the prediction results of the disclosed method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluids. Summary of the invention
[0005] The present invention overcomes the shortcomings of the prior art. Conventional PVT test experimental conditions are demanding, the physical simulation cycle is long, and the efficiency is low. A method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluids is provided. The present invention starts from a microscopic perspective and uses the fluid component composition as the input characteristic value of the prediction model, so that the correlation and interpretability between the prediction model and the high-pressure physical property parameters of the formation fluid are higher, the prediction results are more accurate, and it is more conducive to improving experimental efficiency and guiding reserve calculation and production practice.
[0006] The purpose of the present invention is achieved through the following technical solutions.
[0007] A method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluids is performed according to the following steps:
[0008] Step 1, data collection: collect historical data on the composition of formation fluids in the target oil and gas reservoir blocks;
[0009] Step 2, obtaining samples: the formation fluid components and high-pressure physical property parameters of each well in the target oil and gas reservoir block are regarded as a sample, so that the target block should have n samples;
[0010] Step 3, data division: divide the collected samples into s groups of training set m and validation set p;
[0011] Step 4, data prediction: establish a high-pressure physical property parameter prediction model for oil and gas reservoir formation fluids, use the components of the formation fluids in the training set m as the input features of the model, and obtain output results;
[0012] Step 5, result analysis and verification: Compare the output results with the data in the verification set to obtain error data;
[0013] Step 6, model optimization: select another set of training set m and validation set p, repeat steps 4-5, and continuously build new prediction models based on the new training set m and validation set p based on XGBoost until the correlation coefficient R of the prediction model is 2 If it is greater than 0.95, the error is considered to meet the requirements, and the optimization and establishment of the prediction model is completed;
[0014] Step 7, predicting the high-pressure physical property parameters of the oil and gas reservoir formation fluid: using a set of formation fluid contents as input characteristic values of the optimized model, the input of the optimized model is the predicted high-pressure physical property parameters of the oil and gas reservoir formation fluid.
[0015] In step 3, s sets of training set m and validation set p are obtained using the cross-validation method.
[0016] In step 3, the training set m is composed of the formation fluid components in m samples, and the validation set p is composed of the high-pressure physical property parameters of the formation fluid in p samples.
[0017] In step 4, a high-pressure physical property parameter prediction model for oil and gas reservoir formation fluids is established based on XGBoost.
[0018] In step 4, the output results are formation crude oil viscosity, formation crude oil density, saturation pressure, and volume coefficient.
[0019] The validation set in step 5 is in the same group as the training set in step 6 to ensure that the error comparison is valid.
[0020] In step 1, the historical data of the formation fluid composition of the target oil and gas reservoir block is the formation fluid CO of each well in the block measured by PVT experiment. 2 、N 2 , C 1 , C 2 , C 3 、iC 4 、nC 4 、iC 5 、nC 5 , C 6 , C 7 , C 8 , C9 , C 10 , C 11+ The content and high-pressure physical parameters such as formation crude oil viscosity, formation crude oil density, saturation pressure, volume coefficient, etc.
[0021] In step 7, the input characteristic value is formation fluid CO 2 、N 2 , C 1 , C 2 , C 3 、iC 4 、nC 4 、iC 5 、nC 5 , C 6 , C 7 , C 8 , C 9 , C 10 , C 11+ The content.
[0022] In step 7, the input characteristic value data is the acquisition data, rather than the historical data of the formation fluid component composition of the target oil and gas reservoir block in step 1.
[0023] The beneficial effects of the present invention are as follows: the prediction model established by the present invention is reliable and reasonable, and the high-pressure physical property parameters of oil and gas reservoir formation fluids can be accurately predicted by using the model; the present invention starts from a microscopic perspective and uses the fluid component composition as the input characteristic value of the prediction model, so that the correlation and interpretability between the prediction model and the high-pressure physical property parameters of the formation fluid are higher, the prediction results are more accurate, and it is more conducive to improving experimental efficiency and guiding reserve calculation and production practice. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] Figure 1 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION
[0025] The technical solution of the present invention is further described below through specific embodiments.
[0026] like Figure 1 As shown, a method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluids comprises the following steps:
[0027] i. Data Collection
[0028] Collect historical data on the composition of formation fluids in target oil and gas reservoir blocks;
[0029] ii. Obtaining samples
[0030] The formation fluid components and high-pressure physical property parameters of each well in the target oil and gas reservoir block are regarded as a sample, so that the target block should have n samples;
[0031] ⅲ. Data Partitioning
[0032] Divide the collected samples into s groups of training set m and validation set p;
[0033] ⅳ. Data prediction
[0034] Establish a high-pressure physical property parameter prediction model for oil and gas reservoir formation fluids, use the components of the formation fluids in the training set m as the input features of the model, and obtain output results;
[0035] ⅴ. Result analysis and verification
[0036] Compare the output results with the data in the validation set to obtain error data;
[0037] ⅵ. Model Optimization
[0038] Select another set of training set m and validation set p, repeat steps iv to v, and continuously build new prediction models based on the new training set m and validation set p, until the correlation coefficient R of the prediction model is 2 If it is greater than 0.95, the error is considered to meet the requirements, and the optimization and establishment of the prediction model is completed;
[0039] ⅶ. Prediction of high-pressure physical properties of formation fluids in oil and gas reservoirs
[0040] A set of formation fluid contents is used as input characteristic values of the optimized model, and the input of the optimized model is the predicted high-pressure physical property parameters of the oil and gas reservoir formation fluid.
[0041] In step ⅲ, the training set m and validation set p of group s are obtained using the cross-validation method.
[0042] In step iii, the training set m is composed of the formation fluid components in m samples, and the validation set p is composed of the high-pressure physical property parameters of the formation fluid in p samples.
[0043] In step ⅳ, the high-pressure physical property parameter prediction model of oil and gas reservoir formation fluid is established based on XGBoost.
[0044] The output results in step iv are formation crude oil viscosity, formation crude oil density, saturation pressure, and volume coefficient.
[0045] The validation set in step ⅴ is in the same group as the training set in step ⅳ to ensure that the error comparison is effective.
[0046] The historical data of the formation fluid composition of the target oil and gas reservoir block in step i is the formation fluid CO of each well in the block measured by PVT experiment. 2 、N 2 , C 1 , C 2 , C3 、iC 4 、nC 4 、iC 5 、nC 5 , C 6 , C 7 , C 8 , C 9 , C 10 , C 11+ The content and high-pressure physical parameters such as formation crude oil viscosity, formation crude oil density, saturation pressure, volume coefficient, etc.
[0047] The input characteristic value in step ⅶ is the formation fluid CO 2 、N 2 , C 1 , C 2 , C 3 、iC 4 、nC 4 、iC 5 、nC 5 , C 6 , C 7 , C 8 , C 9 , C 10 , C 11+ The content.
[0048] The input characteristic value data in step ⅶ is the acquisition data, not the historical data of the formation fluid components of the target oil and gas reservoir block in step i.
[0049] Example
[0050] First, 220 PVT experimental sample data were collected, each of which contained formation fluid CO 2 、N 2 , C 1 , C 2 , C 3 、iC 4 、nC 4 、iC 5 、nC 5 , C 6 , C 7 , C 8 , C 9 , C 10 , C 11+ The content and high-pressure physical parameters such as formation crude oil viscosity, formation crude oil density, saturation pressure, volume coefficient, etc.
[0051] Then, the collected samples were divided into 50 groups using the cross-validation method, each of which consisted of 200 random training sets and 20 validation sets; each training set consisted of the formation fluid CO in the corresponding sample data. 2、N 2 , C 1 , C 2 , C 3 、iC 4 、nC 4 、iC 5 、nC 5 , C 6 , C 7 , C 8 , C 9 , C 10 , C 11+ Each validation set consists of high-pressure physical parameters such as formation crude oil viscosity, formation crude oil density, saturation pressure, and volume coefficient in the corresponding sample data.
[0052] Then, a high-pressure physical property parameter prediction model for formation fluids in oil and gas reservoirs was established based on XGBoost. A set of training sets and validation sets were selected, and the formation fluid component composition in the training set was used as the input features of the model, and the formation crude oil viscosity, formation crude oil density, saturation pressure, and volume coefficient were used as the output results of the prediction model.
[0053] The output results of the prediction model are compared with the historical experimental data of formation crude oil viscosity, formation crude oil density, saturation pressure and volume coefficient in the same group of validation sets for error analysis.
[0054] Then, select another set of training sets and validation sets and repeat the above steps until the error meets the requirements, completing the optimization and establishment of the prediction model.
[0055] The formation fluid CO of 10 wells randomly selected 2 、N 2 , C 1 , C 2 , C 3 、iC 4 、nC 4 、iC 5 、nC 5 , C 6 , C 7 , C 8 , C 9 , C 10 , C 11+ The content of is used as the model input characteristic value, and the established prediction model is used to predict the formation crude oil viscosity, formation crude oil density, saturation pressure, and volume coefficient high-pressure physical property parameters. The prediction results are compared with the experimental data of formation crude oil viscosity, formation crude oil density, saturation pressure, and volume coefficient high-pressure physical property parameters of 10 wells, as shown in Tables 1 to 4.
[0056] Table 1 Comparison between experimental and predicted values of formation crude oil viscosity
[0057] sample Actual viscosity value Predicted viscosity value error,% Well No. 1 22.5 23.5 4.44 Well No. 2 126.2 125.6 0.48 Well No. 3 1.1 1.3 2.73 Well No. 4 0.3 0.4 3.33 Well No. 5 0.3 0.2 6.67 Well No. 6 0.3 0.4 3.33 Well No. 7 0.3 0.4 6.67 Well No. 8 13.8 15.1 5.07 Well No. 9 67.9 68.2 1.62 Well No. 10 2.2 2.5 6.82
[0058] Table 2 Comparison between experimental and predicted values of formation crude oil density
[0059] sample Actual density value Predicted density value error,% Well No. 1 0.8934 0.8859 0.84 Well No. 2 0.9365 0.9158 2.21 Well No. 3 0.7356 0.7158 2.69 Well No. 4 0.6158 0.6254 1.56 Well No. 5 0.6159 0.5987 2.79 Well No. 6 0.6289 0.6157 2.10 Well No. 7 0.6338 0.6259 1.25 Well No. 8 0.8894 0.8566 3.69 Well No. 9 0.8659 0.8235 4.90 Well No. 10 0.8155 0.8054 1.24
[0060] Table 3 Comparison between experimental and predicted values of formation crude oil saturation pressure
[0061] sample Actual saturation pressure value Predicted saturation pressure value error,% Well No. 1 4.60 4.52 1.74 Well No. 2 8.23 8.13 1.22 Well No. 3 14.18 13.59 4.16 Well No. 4 27.44 25.91 5.58 Well No. 5 28.32 27.36 3.39 Well No. 6 21.56 20.68 4.08 Well No. 7 36.06 34.21 5.13 Well No. 8 5.59 5.64 0.89 Well No. 9 6.57 6.28 4.41 Well No. 10 14.80 14.2 4.05
[0062] Table 4 Comparison of experimental and predicted values of formation crude oil volume coefficient
[0063] sample Actual volume coefficient value Predicted volume factor value error,% Well No. 1 1.06 1.04 1.89 Well No. 2 1.05 1.03 1.90 Well No. 3 1.27 1.21 4.72 Well No. 4 1.76 1.79 1.70 Well No. 5 1.76 1.72 2.27 Well No. 6 1.52 1.45 4.61 Well No. 7 1.63 1.55 4.91 Well No. 8 1.07 1.12 4.67 Well No. 9 1.09 1.03 5.50 Well No. 10 1.21 1.25 3.31
[0064] The prediction model established by the present invention is reliable and reasonable, and the high-pressure physical property parameters of oil and gas reservoir formation fluid can be accurately predicted by using the model.
[0065] The present invention starts from a microscopic perspective and uses fluid component composition as the input characteristic value of the prediction model, so that the correlation and interpretability between the prediction model and the high-pressure physical property parameters of the formation fluid are higher, the prediction results are more accurate, and it is more conducive to improving experimental efficiency and guiding reserve calculation and production practice.
[0066] The present invention is described above by way of example. It should be noted that, without departing from the core of the present invention, any simple deformation, modification or other equivalent replacement that can be made by those skilled in the art without inventive effort falls within the protection scope of the present invention.
Claims
1. A method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluids, characterized in that: Follow the steps below: Step 1, data collection: collect historical data on the composition of formation fluids in the target oil and gas reservoir blocks; Step 2, obtaining samples: the formation fluid components and high-pressure physical property parameters of each well in the target oil and gas reservoir block are regarded as a sample, so that the target block should have n samples; Step 3, data division: divide the collected samples into s groups of training set m and validation set p; Step 4, data prediction: establish a high-pressure physical property parameter prediction model for oil and gas reservoir formation fluids, use the components of the formation fluids in the training set m as the input features of the model, and obtain output results; Step 5, result analysis and verification: Compare the output results with the data in the verification set to obtain error data; Step 6, model optimization: select another set of training set m and validation set p, repeat steps 4-5, and continuously establish new prediction models based on XGBoost according to the new training set m and validation set p until the correlation coefficient R2 of the prediction model is greater than 0.95, which is considered that the error meets the requirements, and the optimization and establishment of the prediction model is completed; Step 7, predicting high-pressure physical property parameters of oil and gas reservoir formation fluid: using a set of formation fluid contents as input characteristic values of the optimized model, the input of the optimized model is the predicted high-pressure physical property parameters of the oil and gas reservoir formation fluid; In step 1, the historical data of the composition of the formation fluid in the target oil and gas reservoir block are the contents of CO2, N2, C1, C2, C3, iC4, nC4, iC5, nC5, C6, C7, C8, C9, C10, and C11+ of the formation fluid of each well in the block measured by PVT experiments, and the high-pressure physical property parameters of the formation crude oil viscosity, formation crude oil density, saturation pressure, and volume coefficient; In step 7, the input characteristic values are the contents of formation fluid CO2, N2, C1, C2, C3, iC4, nC4, iC5, nC5, C6, C7, C8, C9, C10, and C11+; In step 7, the input characteristic value data is the acquisition data, rather than the historical data of the formation fluid component composition of the target oil and gas reservoir block in step 1.
2. A method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluid according to claim 1, characterized in that: In step 3, s sets of training set m and validation set p are obtained using the cross-validation method.
3. The method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluid according to claim 1, characterized in that: In step 3, the training set m is composed of the formation fluid components in m samples, and the validation set p is composed of the high-pressure physical property parameters of the formation fluid in p samples.
4. The method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluid according to claim 1, characterized in that: In step 4, a high-pressure physical property parameter prediction model for oil and gas reservoir formation fluids is established based on XGBoost.
5. The method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluid according to claim 1, characterized in that: In step 4, the output results are formation crude oil viscosity, formation crude oil density, saturation pressure, and volume coefficient.
6. The method for predicting high-pressure physical property parameters of oil and gas reservoir formation fluid according to claim 1, characterized in that: The validation set in step 5 is in the same group as the training set in step 6 to ensure that the error comparison is valid.
7. Application of the method for predicting high-pressure physical property parameters of formation fluid in oil and gas reservoirs as described in any one of claims 1 to 6 in predicting high-pressure physical property parameters of formation fluid in oil and gas reservoirs.
