ESTIMATION OF RESERVOIR FLUID PROPERTIES USING MUD AND GAS DATA
Patent Information
- Authority / Receiving Office
- MX · MX
- Patent Type
- Patents
- Current Assignee / Owner
- EQUINOR ENERGY AS
- Filing Date
- 2023-01-05
- Publication Date
- 2026-05-19
Smart Images
Figure MX434083B0
Abstract
Description
ESTIMATION OF RESERVOIR FLUID PROPERTIES USING MUD AND GAS DATA Field of invention This description refers to a logging technique for use while drilling a borehole, and particularly a technique that uses mud and gas data to predict reservoir fluid properties. Background of the invention Drilling fluid is a fluid used to assist in drilling boreholes in the earth. The main functions of drilling fluid include providing hydrostatic pressure to prevent formation fluids from entering the wellbore, keeping the drill bit cool and clean during drilling, carrying drill cuttings, and suspending the cuttings while drilling is paused and when the drill string is run into and pulled out of the hole. Drilling fluids are broadly classified into water-based drilling fluids, non-aqueous drilling fluids (often referred to as oil-based drilling fluid), and gaseous drilling fluid. Liquid drilling fluids, that is, water-based or non-aqueous drilling fluids, are commonly referred to as drilling mud. Acquiring mud and gas logs involves collecting data from hydrocarbon gas detectors that record the levels of gases carried to the surface in the drilling mud during a well drilling operation. Mud and gas logging is conventionally used to identify the location of oil and gas zones as they are penetrated, which can be identified by the presence of hydrocarbon gas in the mud system. This can be used to provide a general indication of the reservoir type, as well as to determine whether to take fluid samples downhole for more detailed fluid composition analysis. The presence of hydrocarbon gas can be detected, for example, with a total gas detector. Once the presence of hydrocarbon gas is detected, its composition can be examined, for example, with a gas chromatograph. The most common gas component present is usually methane (C2). The presence of heavier hydrocarbons, such as C2 (ethane), C3 (propane), C4 (butane), and C5 (pentane), may indicate a wet oil or gas zone. Heavier molecules, up to around C7 (heptane), may also be detectable, but are typically present only in very low concentrations. Consequently, the concentrations of these hydrocarbons are not frequently recorded. There are two types of mud and gas data that can be collected, sometimes referred to as "standard" mud and gas logging and "advanced" mud and gas logging. The equipment for standard and advanced mud and gas logging is different. ccnnn / eznz / E / YiAi For a standard mud and gas system, the degasser typically does not have heating or uses a constant volume for gas separation. There is also only one mud sample sampling point ("outside") and therefore it is not suitable for recycle correction. The measured gas composition is usually referred to as standard mud and gas data, which is not directly comparable to the actual C5 or C5 composition of the reservoir fluid sample. For an advanced sludge and gas system, the degasser is heated and typically uses a constant volume for gas separation. There are two sludge sampling points ("outside" and "inside"), allowing for recycle correction. The measured gas composition is usually referred to as advanced sludge and gas data. When generating advanced log-gas data, in order to make the data correspond closely to the current Ci to C5 concentrations of the current reservoir fluid, two correction processes are applied to the “raw” mud and gas data from the advanced mud and gas logging system. First, a recycling correction is made to remove gas contamination from gases originating from previous drilling mud injections. This correction is applied based on a separate mud and gas measurement taken before the drilling mud is injected into the drill string. Second, an extraction efficiency correction step is applied to increase the concentration of intermediate components (C2 to C5) so that their concentrations, relative to the Ci concentration, more closely resemble the relative compositions of a corresponding reservoir fluid sample. The extraction efficiency correction is applied based on the type of drilling mud used for drilling. In the past, advanced mud gas data have been examined to estimate certain reservoir fluid properties using broad, empirical correlations between advanced mud and gas composition and certain reservoir fluid properties. For example, extremely dry gas reservoirs should comprise mostly Ci and not too much C2+, for example, with each of the Ci / C2, C1 / C3, C1 / C4, and C1 / C5 ratios (for raw mud and gas data) being greater than 50. Wet gas reservoirs will frequently have ratios between 20 and 50, and oil reservoirs will have ratios between 2 and 20. Recently, an advanced machine learning model has been developed, making it possible to predict reservoir properties much more accurately from advanced mud and gas data, even where those properties are dependent on the oil (C7+) portion of fluid that is not measured by the mud and gas data. Details of how extensively a machine learning model was trained to determine a reservoir fluid gas-oil ratio based on advanced mud and gas data can be found in the paper by Tao Yang et al. (2019), “A Machine Learning Approach to Predict Gas Oil Ratio Based on Advanced Mud Gas Data.” Society of Petroleum Engineers. do¡:10.2118 / 195459-MS ¡ ccnnn / eznz / E / YiAi Advantageously, this model can be used to generate a substantially continuous record of the reservoir fluid properties. This was not previously possible, and in the past, it was necessary to rely on downhole fluid sampling. Additionally, the model allows predictions of reservoir fluid properties to be made at a very early stage of the drilling process without interrupting the drilling operation, as might be required for similar downhole fluid sampling. This model has proven very useful, but it is limited in that it requires the availability of advanced mud and gas data. There is a need for a technique that can be used when advanced mud and gas data are not available. Brief description of the invention The present invention provides a method for generating a model to predict at least one property of a fluid at a sample location within a hydrocarbon reservoir, comprising: provide a training dataset comprising input data and target data, the input data comprising mud and gas data and petrophysical data for each of a plurality of sample locations, and the target data comprising at least one fluid property for each of the plurality of sample locations; and generate a model using the training dataset such that the model can be used to predict at least one fluid property at the sample location based on the measured mud and gas data and the measured petrophysical data for the sample location, wherein no drilling fluid recycling correction has been applied to the mud and gas data. It is a common belief within the oil and gas industry that petrophysical data provide only a qualitative indication of a reservoir fluid. The data usually predict lean gas with good certainty, but have reduced accuracy when used to distinguish condensate from rich gas and oil. However, it has been found that by supplementing standard mud and gas data with petrophysical data, it is possible to provide an estimate of certain reservoir fluid properties with an accuracy approaching that achievable using only advanced mud and gas data. This is particularly advantageous where generating fluid ownership records for large numbers of new and existing wells is desirable because standard mud and gas data and petrophysical data are collected for almost all wells, including both exploration and production wells. However, the additional cost of collecting advanced mud and gas data, and particularly of having the two sets of mud and gas analysis tools required to perform recycling correction, means that it is often collected only when a few exploration wells are drilled. The number of wells with advanced mud gas data represents only a small fraction of the total wells with standard mud and gas data and petrophysical data. Additionally, the above technique allows for the generation of reservoir fluid property logs for new wells at a reduced cost, as it does not require the additional costs associated with collecting advanced mud and gas data. Importantly, petrophysical data can be collected as a substantially continuous log, similar to mud and gas data. This contrasts with downhole fluid sampling, which requires interrupting the drilling process, adding significant additional costs to the drilling operation. In some modalities, the input data may not include downhole fluid sample data, and the model may not require downhole fluid sample data as an input to predict at least one fluid property at the sample location. The method is preferably a computer-implemented method, and the model generation may comprise instructing a machine learning algorithm to generate the model using the training dataset such that the model can be used to predict at least one fluid property at the sample location based on the mud and gas data measured for the sample location. At least one property is preferably one influenced by the petroleum-related components of the fluid. That is, a property that is not solely the product of the gaseous hydrocarbons within the fluid, whose composition can be predicted based on mud and gas data. At least one property may include the fluid density at the sample location. It will be appreciated that the density can be calculated under either atmospheric or reservoir conditions (for example, taking into account the oil formation volume factor). At least one property may include a gas-oil ratio. That is, a ratio between the amount of gaseous hydrocarbon and the amount of liquid hydrocarbon, typically determined under surface conditions. The gas-oil ratio is preferably a volume ratio. The gas-oil ratio may be a single-flash gas-oil measurement. However, any suitable gas-oil measurement may be used. At least one property may include a saturation pressure of the fluid at the sample location. That is, the pressure at which a secondary phase will appear upon pressure depletion. At least one property comprises a formation volume factor of the fluid at the sample location. That is, the ratio of the fluid volume under reservoir (in-situ) conditions to the fluid volume under surface conditions. At least one property may comprise the concentration of a hydrocarbon within the fluid at the sample location. The hydrocarbon may be a hydrocarbon not included in the mud and gas data. For example, the hydrocarbon may be a C7+ hydrocarbon. That is, the hydrocarbon may be a C7 hydrocarbon or a hydrocarbon heavier than C7, for example, a heavier C8 hydrocarbon. The hydrocarbon may be a hydrocarbon that is substantially petroleum under reservoir conditions. The hydrocarbon concentration may be an absolute concentration (for example, a molar concentration), or it may be a relative concentration (for example, a ratio compared to Ci), or it may be a concentration otherwise normalized. The reservoir can be a gas reservoir, a multiphase reservoir, or an oil reservoir. At least one property for each sample location can be determined from reservoir fluid property data associated with that location. Reservoir fluid property data may include composition data measured for a fluid at the sample location. This reservoir fluid property data may contain hydrocarbon composition data from Ci to C7+ at the sample location, and preferably hydrocarbons from Ci to C20+, and more preferably hydrocarbons from Ci to C36+ at the sample location. As used herein, the notation “Cx+” is to be understood to mean heavier hydrocarbons (Cxo). The mud and gas data in the training dataset may comprise mud and gas data measured for the sample location, i.e., standard mud and gas data measured for the sample location. The measured mud and gas data may be indicative of the composition of gases released from the drilling fluid used while drilling through the sample location (i.e., passing through a drill bit performing the drilling). The measured mud and gas data may be indicative of a concentration of at least gaseous hydrocarbons from C1 to C4, and preferably at least gaseous hydrocarbons from C1 to C5, that were released from the drilling mud. As stated above, mud and gas data do not primarily include advanced mud and gas data, i.e., where mud and gas data have not been corrected to correspond to the gaseous hydrocarbon composition of the fluid at the sample location. A drilling fluid recycling correction refers to correcting mud and gas data to eliminate errors due to gases released from previous drilling operations, such as those resulting from drilling fluid recycling. Typically, this would require measuring baseline mud and gas data collected before the drilling mud was injected into the drill string. Optionally, no extraction efficiency correction has been applied to the mud and gas data. A extraction efficiency correction refers to correcting mud and gas data (O( to C5) to closely correspond with the reservoir fluid composition due to different hydrocarbon components having different capacities to evaporate from the drilling mud. Where a extraction efficiency correction has not been applied, the training data can additionally include drilling mud composition data. This can allow the machine learning model to correct the data within the model generated by the machine learning algorithm ccnnn / eznz / E / YiAi. Alternatively, a drawdown efficiency correction may have been applied to the standard mud and gas data. Often, the type of drilling mud used for a well is known, and in many cases, drawdown efficiency corrections can be estimated by Equation of State (EOS) simulation or by testing. Consequently, even when standard mud and gas data are used, it may be possible to retrospectively apply a drawdown efficiency correction to the standard mud and gas data. Optionally, mud and gas data were collected without heating. Although standard mud and gas data can use heating, it has not been frequently used when collecting mud and gas data. Consequently, where it is desirable to use the model to examine existing wells, a model trained using mud and gas data collected without heating is particularly useful. Petrophysical data may comprise any one or more of the following: bulk density, neutron porosity, resistivity data, acoustic data, natural gamma rays, nuclear magnetic resonance data, as well as deceleration time and gamma-ray spectroscopy data from pulsed neutron measurements, and the like. Optionally, petrophysical data may comprise two or more of these data types. Model generation may include: training a machine learning algorithm on a first subset of the training dataset; and testing the machine learning algorithm on a second, non-continuous subset of the training dataset. The first subset preferably comprises at least 50% of the samples in the training dataset. The second subset preferably comprises at least 10% of the samples in the training dataset. Viewed from a second aspect, the present invention provides a computer-based model for predicting at least one property of a fluid at a sample location within a hydrocarbon reservoir based on measured mud and gas data and measured petrophysical data for that sample location; the computer-based model has been generated by the above method. Viewed from the third aspect, the present invention provides a tangible, computer-readable means of storing the computer-based model. Viewed from a fourth aspect, the present invention provides a method for predicting the value of a fluid property at a sample location within a hydrocarbon reservoir. The method comprises: receiving measured mud and gas data and measured petrophysical data for the sample location; and predicting the value of the fluid property at the sample location by supplying the measured mud and gas data and measured petrophysical data to the computer-based model. The method may also include determining a quality for the measured mud and gas data and / or the measured petrophysical data. ccnnn / eznz / E / YiAi The method may also include generating a confidence indicator associated with the predicted value of the fluid property. The confidence indicator may be a numerical indication, but other indications may be used, such as color indications (e.g., red / yellow / green), or word indications (e.g., “good” / “poor”). The confidence indication can be based on the quality of the measured mud and gas data and / or the measured petrophysical data. For a single data point, the confidence indication can be reduced by one or more of a Cq C4o C5 concentration that is below a respective predetermined threshold. Where standard mud and gas data are taken at a number of locations at different depths, the confidence indication can be reduced by fluctuations in a component concentration of the mud and gas data greater than a threshold amplitude within a predetermined depth range. Where standard mud and gas data and / or petrophysical data are taken at a series of locations at different depths, the confidence indication can be reduced by the loss of a predetermined number of previous measurements or over a predetermined depth range. Viewed from a fifth aspect, the present invention provides a method for predicting the value of a fluid property of a fluid along a length of a well through a hydrocarbon reservoir, the method comprising: predicting a value of a fluid property of a fluid at a plurality of sample locations along a length of a well using the above method. The method may comprise: displaying, using an electronic display screen, a graph that plots the predictive value of the fluid property against a respective sample location for each of the plurality of sample locations along the length of the well. The method may further include: displaying, using the electronic display screen, a confidentiality indicator associated with one or more of the predicted values. For example, the confidentiality indicator may be illustrated numerically, verbally, chromatically, or iconographically. The entire method described above—that is, the methods in the first, fourth, and fifth aspects—can be carried out in any suitable and desired manner and on any suitable and desired platform. In a preferred embodiment, each of the methods is a computer-implemented method; for example, the steps of the method are carried out by a processing circuit. The methods according to the present invention can be implemented at least partially using software, for example, computer programs. Thus, it will be observed that, viewed from other perspectives, the present invention provides computer software specifically adapted to carry out the methods described herein when installed on a data processor, a computer program element comprising portions of computer software code for carrying out the methods described herein when the program element is executed on a data processor, and a computer program comprising code adapted to carry out all the steps of one or more of the methods described herein when the program is executed on a data processing system. The present invention also extends to a computer software carrier comprising this software arranged to carry out the steps of the methods of the present invention. This computer software carrier could be a physical storage medium such as a ROM chip, CD-ROM, DVD, RAM, flash memory, or disk, or it could be a signal such as an electronic signal over wires, an optical signal, or a radio signal such as one transmitted to a satellite or the like. It will be further appreciated that not all steps of the methods of the present invention need to be carried out by computer software, and thus in a further broad manner, the present invention provides computer software and this software installed on a computer software carrier to carry out at least one of the steps of the methods set forth herein. Accordingly, the present invention can be suitably incorporated as a computer program product for use with a computer system. This implementation may comprise a set of computer-readable instructions that can be stored on a tangible, non-transient medium, such as a computer-readable medium, for example, a floppy disk, CD-ROM, DVD, ROM, RAM, flash memory, or hard drive. It could also comprise a set of computer-readable instructions that can be transmitted to a computer system, via a modem or other interface device, over either a tangible medium, including but not limited to optical or analogous communication lines, or using intangible wireless techniques, including but not limited to microwave, infrared, or other transmission techniques. The set of computer-readable instructions incorporates all or part of the functionality described herein. Those skilled in the art will appreciate that these computer-readable instructions can be written in a variety of programming languages for use with many computer architectures or operating systems. Furthermore, these instructions can be stored using any memory technology, present or future, including but not limited to semiconductor, magnetic, or optical, or transmitted using any communications technology, present or future, including but not limited to optical, infrared, or microwave. It is envisaged that this software product may be distributed as removable media with accompanying printed or electronic documentation, for example, shrink-wrapped software, pre-loaded with a computer system, for example, in a ROM system or fixed disk, or distributed from a server or electronic bulletin board over a network, for example, the Internet or the World Wide Web. Brief description of the drawings Certain preferred embodiments of the present description will now be described in greater detail, by way of example only and with reference to the accompanying figures, in which: FIG. 1 is a schematic illustration of a mud and gas analysis tool. FIG. 2 illustrates a workflow for a machine learning algorithm to generate a first model to predict a gas-oil ratio using a training dataset; An exemplary standard gas-mode analysis tool 20 is shown schematically in FIG. 1. Detailed description of the invention Tool 20 is attached to a flow line 10 containing the drilling mud returned from a well drilling operation. As stated above, the drilling mud can be either water-based or oil-based. Tool 20 comprises a sampling probe 22 positioned with respect to flow line 10 to collect a sample 24 of the drilling mud from flow line 10. The drilling mud sample 24 is preferably a continuous sample, i.e., such that a portion of the drilling mud flow within flow line 10 is diverted through the mud and gas analysis tool 20. The drilling mud sample 24 is supplied to a gas separation chamber 26 where at least a portion of the gas carried by the drilling mud is released. The drilling mud sample can optionally be heated by a heater 28 upstream of the gas separation chamber 26. Heating the drilling mud sample 24 helps release the gas from the drilling mud sample 24. Typically, the mud sample 24 is not heated, and the temperature typically ranges from 10°C to 60°C. However, in some implementations, heating to 80°C to 90°C is used. The released gas 30 is directed from the separation chamber 26 to a gas analysis unit (not shown), while the degassed sludge 32 is returned to the flow line 10 or to another location for reuse. The gas analyzer may comprise a total gas detector, which can provide a basic quantitative indication as to how much gas is being extracted from the drilling mud by the tool. Total gas detection typically incorporates either a catalytic filament detector, also called a thermal wire detector, or a hydrogen flame ionization detector. A catalytic filament detector operates on the catalytic combustion of hydrocarbons in the presence of a heated platinum wire at a gas concentration below the lower explosive limit. The gradual heating due to combustion causes a corresponding increase in the resistance of the platinum wire filament. This increase in resistance can be measured using a Wheatstone bridge or an equivalent detection circuit. A hydrogen flame ionization detector operates on the principle of ionizing hydrocarbon molecules in the presence of a very hot hydrogen flame. These ions are subjected to a powerful electric field, resulting in a measurable current flow. The gas analysis device may additionally or alternatively include an apparatus for the detailed analysis of the hydrocarbon mixture. This analysis is usually performed using a gas chromatograph. However, several other detection devices may also be used, including a mass spectrometer, an infrared analyzer, or a thermal conductivity analyzer. / crnnn / Q7n7 / B / YiAi A gas chromatograph is a batch-processing, rapid-sampling instrument that provides proportional analysis of a series of hydrocarbons. Gas chromatographs can be configured to separate almost any quantity of gases, but typically oilfield chromatographs are designed to separate the paraffin hydrocarbon series from methane (C0) to pentane (C5) at room temperature, using air as a carrier. The chromatograph will report (in units or as a mole percent) the amount of each component of the detected gas. A carrier gas stream 34, commonly comprising air, can be supplied to the separation chamber 26 and mixed with the released gas 30 to form a gas mixture 36 that is supplied to the gas analysis unit. The carrier gas stream 34 provides a continuous flow of carrier gas in order to provide a substantially continuous flow rate of the gas mixture 36 from the separation chamber 26 to the gas analysis unit. Additionally, in the case of a gas analyzer comprising a gas chamber, the use of air as the carrier gas can provide the oxygen required for combustion. In some provisions, tool 20 can be configured to detect and / or remove H2S from the gas to prevent adverse effects that could influence hydrocarbon detection. In some models, non-combustible gases, such as helium, carbon dioxide, or nitrogen, can be detected by the gas analyzer in conjunction with the hydrocarbon record acquisition. The following technique seeks to use a machine learning algorithm to produce a model that accurately estimates certain reservoir fluid-related properties, particularly the gas-oil ratio and reservoir fluid density, based on standard mud and gas data and other petrophysical data. FIG. 2 illustrates a workflow for training the machine learning algorithm to generate a predictive model of a reservoir's gas-oil ratio based on measured standard mud and gas data. In the following example, an input dataset of 102 is used as a training dataset and comprises data relating to a plurality of reservoir samples. The input dataset comprises reservoir fluid property data from a large number of reservoir fluid samples. These reservoir samples can be obtained, for example, by downhole fluid sampling. However, other techniques could also be used, such as taking a fluid sample from the well after it has been completed. Reservoir fluid property data include at least hydrocarbon composition data, which may be in the form of direct measurements of the concentration of each hydrocarbon component within the sample, typically covering hydrocarbons from Ci to C36+. In some forms, concentration data may be in the form of relative data (e.g., as a ratio of compositions of different hydrocarbons) or may be normalized in some other way. Reservoir fluid property data may also optionally include concentrations of one or more other constituents within the wellbore. Reservoir fluid property data may include one or more properties derived from the reservoir fluid sample. These derived properties may include the target property determined by the machine learning algorithm, for example, a gas-oil ratio in this case. Other derived properties may include fluid density. Reservoir fluid property data are sometimes referred to as PVT data, as they are commonly obtained in a pressure-volume-temperature (PVT) laboratory, where researchers will employ various instruments to determine reservoir fluid behavior and reservoir sample properties. Input dataset 102 also includes standard mud and gas data measured for each PVT sample at the same depth in the reservoir. The measured standard mud and gas data includes hydrocarbon composition data measured for the gas released from the drilling fluid at the sample location. It will be noted that there is a time delay between the drill bit passing through the sample location and when the mud reaches the surface and is analyzed. However, workers in this field will be familiar with the procedures for calculating the time delay to determine the depth to which the mud and gas sample corresponds. Therefore, this is not discussed in detail. The composition data for mud and gas preferably includes data for at least hydrocarbons C0 to C4, and preferably at least hydrocarbons Ci to C5 (as is the case in the present example). In some cases, concentrations for hydrocarbons up to C7 or higher may be included. Composition data can be stored either as a direct concentration measurement (e.g., measured in ppm or similar units), or alternatively as a first relative concentration (e.g., as a ratio of another hydrocarbon, usually CQ). In some modes, composition data can be normalized. The standard measured sludge and gas data are “raw” sludge and gas data, meaning they have not been corrected for recycling or extraction efficiencies. This is important because using “raw” sludge and gas data will allow the subsequent model to be used more widely in areas where advanced sludge and gas data are unavailable. Input dataset 102 further comprises petrophysical data measured for each PVT sample at the same reservoir depth. The petrophysical data may include any one or more of the following: bulk density, neutron porosity, resistivity data, acoustic data, natural gamma ray, nuclear magnetic resonance data, as well as deceleration time and gamma ray spectroscopy data from pulsed neutron measurements, and the like. The input dataset 102 comprises target data and input data for each sample that passes the classification. The target data corresponds to the desired output of the model. The input data corresponds to the data that has been entered into the eventual model. ccnnn / eznz / E / YiAi The target data in this example is a gas-oil ratio, and in this example, it is the gas-oil ratio measurement from a single flash of the sample. As stated above, this data is stored as part of the reservoir properties data within the initial datasets. Alternatively, other gas-oil ratio measurements can be used, or the gas-oil ratio can be derived from reservoir composition data, i.e., based on the concentrations of the various hydrocarbons. The input data are standard mud and gas data, i.e., data indicative of the composition of gases released from the drilling fluid at the sample location, and at least one type of petrophysical data, for example bulk density, and neutron porosity. As mentioned above, the measured sludge and gas data comprise "raw" sludge and gas data, i.e., they have not been corrected for recycling or extraction efficiency. Although it is not possible to apply a recycling correction after data collection, and it is not possible to account for the lack of heating (if heating was not used), it may be possible to apply a retrospective extraction efficiency correction. This is because the drilling mud composition for a particular well is usually known, and the extraction efficiency correction factors for that particular drilling fluid can be estimated either from the EOS simulation or approximated through experimentation. The results show that the temperature-dependent extraction efficiency correction far outperforms the recycling corrections. Accordingly, the mud and gas data used for input dataset 102 preferably comprise standard mud and gas data where a extraction efficiency correction has been applied. Next, a model generation is carried out, in which a model is generated and validated based on the input data set 102. The input dataset 102 is first divided into a training dataset 104 and a test dataset 106. The input dataset 102 is preferably curated such that at least the test dataset 106 contains data spanning the various classes of the input dataset 102 into a whole (e.g., dry gas reservoirs, wet gas reservoirs, oil reservoirs). Typically, at least 50% of the input dataset 102 should be used for training, and at least 10% of the input dataset 102 should be used for practice. Common ratios include 50:50, 70:30, 75:25, 80:20, and 90:10. However, other ratios may be used instead. In general, the larger the training dataset, the more accurate the model will be. However, if a very small test dataset is used (or if no test dataset is used at all), then it is not possible to reliably verify the model's accuracy, for example, making it difficult to detect an overfitted model (accurate only for the specific training data). To generate a model, a machine learning algorithm is provided with the training data set ccnnn / eznz / E / YiAi 104 and a set of training parameters to control the machine learning algorithm. In one example, Gaussian process regression and random forest were found to be the best-performing models. However, it should be noted that any suitable algorithm can be used, such as Universal Krlging, KMean, or Elastic Net. Those working in this field will be familiar with the procedures for selecting and using a machine learning algorithm. Therefore, this will not be discussed in detail. Model 108 variation, for example, cross-validation, can then be performed. During model 108 validation, the model is tested to determine how well it predicts new data that were not used in the model estimation, in order to identify problems such as overfitting or selection bias. Model 108 validation is an optional step. Cross-validation involves partitioning the training dataset 104 into complementary subsets, performing model fitting using one subset of the training dataset 104, and validating the analysis on the other subset of the training dataset 104. To reduce variability, most methods use multiple rounds of cross-validation, performed using different partitions, and the validation results are combined (e.g., averaged) over the rounds to give an estimate of the model's predictive performance (e.g., a mean average prediction error, MAPE). In this example, K-times cross-validation is used, specifically 4-times cross-validation. In K-times cross-validation, the 104 training datasets are divided into K discontinuous subsets (in this case, four), known as folds. Cross-validation is then performed by training the model on all datasets except one fold and validating the trained model using the fold that was not used for training. The best model is then selected as the one with the best predictive performance, for example, the lowest MAPE. A first test step 110 is then carried out, in which the model is tested using the training dataset 104 as a whole. A second test step 112 is then carried out, in which the model is tested using test dataset 106. As previously stated, this is a curated dataset that is broadly representative of the data as a whole, and was not used during the generation of the model. The model has been found to predict a reservoir fluid gas-oil ratio based on standard C5 mud and gas data and MAPE petrophysical data that is close to what has been achieved using a model based on advanced Ci to C5 mud and gas data. Understanding the quality of measured mud and gas data is important before making a prediction of fluid properties (e.g., gas-oil ratio) because the quality of the mud and gas data will significantly impact the accuracy of the prediction. The following characteristics of mud and gas data values have been identified as indicating low quality or unreliable data: ccnnn / eznz / E / YiAi • Large fluctuations of a component within a small depth interval. • First observations after the omitted measurements. ♦ Content of q below a given threshold. • C4 or C5 content below a given threshold. To quantify the quality of mud and gas data, the inventors derived a quality control measure (QC measure) ranging from 0 to 1. High-quality mud and gas data would have a QC measure value close to 1. If one or more of the factors mentioned above were present, the QC measure would decrease. Low-quality mud and gas data were indicated by a QC measure close to 0. An individual numerical quality measure between 0 and 1 could be plotted side-by-side with a predicted fluid property record (as will be discussed below) to visualize the level of confidence associated with each prediction, based on the quality of the mud and gas data. Samples with a higher QC score show a closer match, while samples with a lower QC score show a poor match. Thus, these factors provide a useful indication of the accuracy of a gas-oil ratio prediction. Mud and gas data and petrophysical data are both generated continuously during the drilling process. Therefore, by applying a machine learning model to mud and gas data and petrophysical data, it is possible to provide, at any early stage of the well installation procedure, a continuous well log of predicted reservoir properties, such as gas-oil ratio or fluid density. This has sometimes not been possible previously until much later in the process. Although previous examples have been described in the context of a gas-oil ratio as the target reservoir fluid property, the same technique can also be used to create a model for estimating other reservoir fluid properties at a sample location, based on measured mud and gas data. Exemplary reservoir fluid properties include reservoir fluid density, either reservoir tank oil density or live reservoir density, reservoir fluid saturation pressure, and a formation volume factor of the reservoir volume. Additionally, a similar technique can be used to train a model to estimate the reservoir fluid composition and the corresponding C7+ fraction properties. This is advantageous because this information can be used for an equation of state (EOS) model calculation. The EOS model for a particular fluid is an expression that describes the relationship between the fluid's pressure, temperature, and volume and can be used to predict the fluid's phase behavior in order to derive additional fluid properties. It is generally considered necessary to know at least the following fluid properties in order to determine the equations of state: 1) The absolute composition of each of the hydrocarbons from C1 to C6 and the absolute composition of the combined C7+ hydrocarbons; ccnnn / eznz / E / YiAi 2) The average hydrocarbon density of C7+ hydrocarbons; and 3) The average molecular weight of the hydrocarbon of C7+ hydrocarbons. When determining the equations of state for a fluid, C7+ hydrocarbons are usually grouped together because these hydrocarbons remain in the liquid / oil phase. A standard C7+ characterization method can divide the C7+ into multiple pseudo-components for EOS calculation. Although models of individual fluid properties (such as density and GOR) were developed in the first examples, it will be appreciated that a physical model could be generated that would calculate all fluid properties. The EOS model procedure in the second example 10 shows a good solution for predicting all reservoir fluid properties. Although preferred embodiments have been described above, it will be appreciated that these have been provided by way of examples only, and the scope of the invention should be limited only by the following claims.
Claims
1. A method for generating a model to predict at least one fluid property at a sample location within a hydrocarbon reservoir, comprising: providing a training dataset comprising input data and target data, the input data comprising mud and gas data and petrophysical data for each of a plurality of sample locations, and the target data comprising at least one fluid property for each of the plurality of sample locations; and generating a model using the training dataset such that the model can be used to predict at least one fluid property at the sample location based on the measured mud and gas data and the measured petrophysical data for the sample location, wherein no drilling fluid recycling correction has been applied to the mud and gas data.
2. The method according to claim 1, wherein the model generation comprises instructing a machine learning algorithm to generate the model using the training dataset.
3. The method according to claim 1 or 2, wherein the at least one property comprises a property influenced by the petroleum-related components of the fluid.
4. The method according to any of the preceding claims, wherein the at least one property comprises one or more of: a fluid density at the sample location; a gas-oil ratio of the fluid at the sample location; a saturation pressure of the fluid at the sample location; a formation volume factor of the fluid at the sample location; a concentration of C7+ hydrocarbons within the fluid at the sample location.
5. The method in accordance with any of the preceding claims, wherein the mud and gas data of the training data set comprises standard mud and gas data measured for the sample location.
6. The method according to claim 5, wherein an extraction efficiency correction has been applied to the mud and gas data of the training dataset.
7. The method according to claim 5, wherein an extraction efficiency correction has not been applied to the mud and gas data of the training dataset, and wherein the training data comprises drilling mud composition data.
8. The method according to claim 5, 6 or 7, wherein the measured mud and gas data were collected without the use of heating.
9. The method according to any of the preceding claims, wherein the petrophysical data comprise one or more of: bulk density; neutron porosity; resistivity data; acoustic data; natural gamma rays; nuclear magnetic resonance data; and gamma-ray spectroscopy data.
10. A computer-based model for predicting at least one property of a fluid at a sample location within a hydrocarbon reservoir based on measured mud and gas data and measured petrophysical data for the sample location, the computer-based model being generated by the method in accordance with any of the preceding claims.
11. A tangible computer-readable medium that stores a computer-based model in accordance with claim 10.
12. A method for predicting a fluid property value at a sample location within a hydrocarbon reservoir, the method comprising: receiving measured mud and gas data and measured petrophysical data for the sample location; and predicting the fluid property value at the sample location by supplying the measured mud and gas data and measured petrophysical data to a computer-based model according to claim 10.
13. A method for predicting a fluid property value of a fluid along a length of a well through a hydrocarbon reservoir, the method comprising: predicting a fluid property value of a fluid at a plurality of sample locations along a length of a well using a method according to claim 12 for each sample location.
14. The method according to claim 13, further comprising: displaying, using an electronic display screen, a graph plotting the predicted values of the fluid property against a location of the respective sample location for each of the plurality of sample locations along the length of the well.