method for controlling a flow network to improve the performance of the flow network, and control ap

专利摘要:
CONTROL OF FLOW NETWORKS. The present invention relates to a method for controlling a flow network in order to improve the performance of the flow network, comprising: (a) applying predetermined excitations at multiple control points within the flow network, wherein the multiple control points are on different branches of the flow network; (b) receive measurements of changes in one or more flow parameters on one or more flow paths where flows from more than one of the different branches have been combined; (c) performing an analysis of the flow parameter measurements to identify variations induced by applied excitations; (d) determine an adjustment to be made in one or more of the control points in order to improve the performance of the flow network, for example, by building and solving an optimization model; (e) make the determined adjustment at the flow network control point(s) or make an alternative adjustment decided by the flow network operator; and (f) repeating steps (a) through (e) one or more times to thereby iteratively improve the performance of the flow network.
公开号:BR112015026401B1
申请号:R112015026401-8
申请日:2014-04-17
公开日:2021-05-04
发明作者:Bjarne Foss；Vidar Gunnerud
申请人:Norwegian University Of Science And Technology (Ntnu)；
IPC主号:

专利说明:

[0001] The present invention relates to an apparatus and a method for controlling a flow network in order to improve the performance of the flow network, for example, to optimize the production of oil and gas wells.
[0002] There are many industries in which flow networks are used, for example, in the processing and manufacture of fluid and liquid products in factories and refineries. The oil and gas industry is an example of particular interest, as the flow network includes oil and gas wells that result in entries into the flow network that can be difficult to model and, in many cases, can vary unpredictably. Additionally, the availability of critical process components changes over time, and thus capabilities vary equivalently. As such, it is difficult to optimize production settings for such networks. Simulations and models can be used in an attempt to predict the response of flow networks to changes in process parameters such as flows, pressures, mixture of different constituents and so on. However, these joint models and optimization problems can become very cumbersome and require significant computational power, while still providing nothing more than a computer-assisted guess for optimal definition for the flow network.
[0003] Viewed from a first aspect, the invention provides a method for controlling a flow network in order to improve the performance of the flow network, wherein the method comprises: (a) applying predetermined excitations at multiple control points within the flow network, where multiple control points are on different branches of the flow network; (b) receive measurements of changes in one or more flow parameters on one or more flow paths where flows from more than one of the different branches have been combined; (c) performing an analysis of the flow parameter measurements to identify variations induced by applied excitations; (d) determine an adjustment to be made at one or more of the control points in order to improve the performance of the flow network; (e) make the determined adjustment to the flow network control point(s) or make an alternative adjustment decided by the flow network operator; and (f) repeating steps (a) through (e) one or more times to thereby iteratively improve the performance of the flow network.
[0004] This method provides a significant advantage compared to prior art methods for controlling flow networks as it makes it possible to do an iterative improvement of the performance of the flow network where each iterative adjustment is made, and the impact on the real world system is measured, before the next adjustment is decided. Advantageously, the proposed fit can be determined based on data that includes data obtained from real-time measurement of the system, which is an effective real-time online experiment using the applied oscillations.
[0005] Preferably, the analysis in step (c) includes creating a model, which can be a simple model such as a localized linear model and this is then optimized in the determination step (d) to identify the best fit to ( s) control point(s).
[0006] There are clear advantages for creating a model based on online experiments, that is, by applying excitations to the real flow network and receiving real world results. This allows the iterative process to consider the actual operating point of the flux network and be based on the actual reaction of the flux network to excitations.
[0007] Step (c) may include creating a local mathematical optimization problem to calculate a fit of one or more of the control points, and in this case, step (d) may include solving this optimization problem in order to determine the necessary adjustment.
[0008] With a well characterized flow network, or parts of a flow network that are well characterized, it may also be possible to create simplified models, such as localized linear models, by exciting a simulator. This will allow useful data to be obtained and a proposed fit determined without the need for online experiments. Such simulation-based excitations can provide a significant advantage in terms of speed and ease of testing, as it is appropriate to use a simulation. Thus, the analysis in step (c) can include creating models from simulator data in conjunction with models created from real world data. In this case, step (a) includes applying excitations to the simulation and step (b) includes receiving simulated response to the excitations. Compared to conventional simulation, this technique offers significant advantages as model optimization is much faster than simulator optimization. Furthermore, there is a significant advantage to including operator input in the iteration in steps (d) and (e), where the model is then preferably updated to account for changes in the flow network caused by the adjustment to the control points.
[0009] The excitations in step (a) always include online experiments with control points of the flow network as well as, optionally, simulation data where it is considered that such data is sufficiently accurate. When simulation-based models are used, these models are preferably updated when steps (a) to (e) are subsequently repeated to account for the adjustment made in previous step (e).
[00010] In prior art simulations and models, attempts are made to predict the performance of a flow network and to obtain a single "offline" solution for an optimal configuration of control points. A typical setup for such a simulation is to use an iterative solution that starts from a known point, based on measuring flow network parameters obtained using a conventional method, and then trying to converge towards optimized performance without additional input in relation to the real-world impact of the iterative adjustment on the flow network. Of course, this is not ideal as there is an inherent risk of diverging from the real path that the real world system, which can include unpredictable and non-linear elements, will react to small adjustments to the control points.
[00011] The use of models, eg simple models such as linear models, together with an iterative approach that allows the entry of a human being in step (e), allows considerable improvements compared to known simulations, especially since excitations they are also applied to the flow network to obtain real-world data about the reactions of the flow network. Instead of a complex and lengthy simulation aimed at obtaining the best “optimized” solution, it is possible to advance steps towards an improved solution and, at each stage, observe the real changes that result from the adjustments to the flow network. The model is much faster than a full simulation.
[00012] Different experimental patterns can be used for excitations, such as gradual changes, linear patterns and sinusoidal variations. Models can then be extracted from the results of these experiments using the measurements and analysis in steps (b) and (c) and these models can be used to carry out step (d). Different experimental patterns have different pros and cons. Gradual changes are, for example, easier to implement than sinusoidal patterns, although sinusoidal patterns can be analyzed more easily and more accurately than gradual changes.
[00013] In a preferred mode, the excitations are oscillations applied at known frequencies. Preferably, the oscillations applied at different control points of the multiple control points are at different test frequencies and, in step (c), a frequency analysis of the measured flow parameters is performed. Oscillations can therefore be applied in parallel to frequency analysis, allowing the identification of responses that result from the excitation of different control points. This allows a model of the flow network to be obtained for use in determining the adjustment to be made in step (d). It is particularly preferred that the applied techniques are similar to those discussed in WO 2013/072490 by Sinvent AS and Norwegian University of Science and Technology (NTNU), which proposes the use of oscillations to monitor oil and gas wells. Compared to this prior technique, the present disclosure adds the non-obvious capability of iteratively improving flow network performance.
[00014] With this frequency-based method, the properties of individual branches of the flow network can be easily determined without the need to run individual tests for each branch and without the need to stop flow to allow the individual branches to be tested . No dedicated test equipment other than a measuring device is required for the combined flow(s) as existing control points can be used to apply the required excitations. Furthermore, use of the flux network for its normal purpose can continue with minimal interference. For example, when the method is used for an oil and gas production flow network, then production can continue through the production head over the course of the test and although applied excitations will likely reduce the average flow rate, the reduction in production is low compared to the reduction in production for a conventional test such as an accumulation test. For a field with ten wells, production during a test campaign can be more than 4% higher for the method of the invention compared to an equivalent accumulation test. Different branches of the flow network (eg different wells) are tested in parallel with measurements of each individual branch and determined by observing the effects of the oscillation frequency applied across the control point for that branch. Through frequency analysis, these effects can be isolated from other variations in the output stream.
[00015] An alternative method, which can be used instead of or in addition to the frequency-based technique described above, is to use sequentially applied excitations rather than in parallel and isolate the effects of the excitations by means of bandpass filters or similar. One example is excitations applied through the use of repetitive gradual changes that approximate a sinusoidal waveform in a very rough way. Measurements in step (b) can be filtered by a band-pass filter in step (c), ie a device that passes frequencies in a certain range and attenuates frequencies outside that range. This allows the calculation of sensitivity between properties on different branches in the flow network for a control point. An example is the sensitivity between changes in a gas rise rate in a well and the pressure drop in a pipeline.
[00016] For any of the methods discussed above, the control points can be any means that has the ability to apply a controlled fit to the fluid with a known frequency of the fit. The adjustment can be on any suitable fluid parameter, such as a fluid flow and/or pressure. For example, suitable control points might include flow control valves, pumps, compressors, gas lift injectors, expansion devices, and so on. The basic principle of the above methods can be applied with any device that can apply an excitation in a flux network conduit (or in a simulation), since no matter what is used to apply the excitation, it is still possible to obtain information about the contribution of different branches of the network to flow combined through a frequency analysis performed downstream. Excitations do not need to be in flow rate or pressure, but can include other parameters such as level in a subsea separator and ESP setting and pump. The measurement for step (b) should, of course, be selected in relation to the excitation that is applied to ensure that what is being affected by the applied excitation. For example, a pressure excitation will affect pressure and flow rate, but it can also create output variations in temperature, water content, and so on.
[00017] In preferred modalities where the method is applied to an oil and gas production flow network, the control points may include one or more among: throttling control valve; gas lift valve settings or rates in riser ducts or wells; ESP (Electric Submersible Pump) settings, effect, speed, pressure rise, etc.; well branch valve definitions, subsea and deck control definitions on one or more: separators, compressors, pumps, scrubbers, condensers/coolers, heaters, annular seal columns, mixers, dividers, cooling systems, etc. (any equipment that affects production).
[00018] The measured flux parameter(s) can be any parameter that is affected by the excitation applied to the control points. Therefore, the flow parameter(s) can include one or more of pressure, flow rate (in volume or flow velocity), level or temperature, all of which are parameters that can vary across a volume of a combined flow in response to variations in individual branches of the flow network. The flow parameter(s) could alternatively or additionally include one or more parameters related to the characteristics of the fluid in the flow network, such as a gas to liquid ratio, proportions of certain components in the flow, density, pH and so on. against. In the example where the flow network is an oil and gas production flow network, then the flow parameter(s) might include, for example, water content (WC), productivity index (PI) , Oil to Gas Ratio (GOR), wellhead pressures and BHP, rates after deck separation, other rate measurements, eg water after subsea separation, other pressures, eg pipe line pressure, pressure separator, other line pressures, temperatures (many locations along the production system), flow velocities or sand production, among other things.
[00019] Flow parameters can be measured directly, eg via a pressure or temperature sensor, or alternatively can be measured indirectly, eg via calculations based on directly measured parameters.
[00020] Control points can include gas lift rates. It is preferable to use excitations at gas lift rates and also excitations applied with throttling valves.
[00021] Preferably, the excitation is applied to more than one type of control point and, in the case with maximum preference, it is applied to most or all of the control points available in the flow network or in a part of the network flow that is of interest. This allows an assessment to be made of the flow network's reaction to disturbances in any of the available control mechanisms and therefore allows the best possible fit to be identified through variation analysis to determine which control point adjustment will produce the most desirable change.
[00022] For similar reasons, it is preferable to measure a plurality of flow parameters in step (b) and, in particular, measure the response for most or all of the flow parameters that are relevant to the required improvement in network performance flow. This can be, for example, flow parameters relevant to increased production for an oil and gas production flow network.
[00023] The improvement to the performance of the flow network can be accomplished through any advantageous change in any part of the performance of the flow network. In one example, enhancement includes increasing or decreasing one or more output parameters of interest, and output parameters are therefore the focus of iterative changes in step (e) and iterations of the process. Output parameters can be related to production volume or quality, for example. Enhancement may alternatively include changing another aspect of the flow network.
[00024] Thus, the improvement may involve one or more of: increasing or decreasing the one or more output parameters of interest, increasing the accuracy of information provided by the analysis in step (c), adjusting the operational parameters of network components to extend the service life of those components or other components of the flow network, or improve another aspect of the flow network not listed above.
[00025] The output parameter(s) of interest, which the method seeks to change in some examples in order to improve performance, can be any parameter(s) of the flow network. Such a parameter can be a parameter of the type that is measured in step (b), for example, a total combined flow rate or a pressure required for a given production and so on. In the example where the flow network is an oil and gas production flow network, then the output parameter(s) of interest could be, for example, pressure drop across the production bottleneck, or total production. There can be only one output parameter of interest, or instead the enhancement to the system can be related to a combination of output parameters. If an output parameter of interest is not measured, eg flow velocity, other output parameter(s) can be used, eg pressure and temperature, to compute the interest, if an analytic expression is known, through first-order physics or through an empirical correlation.
[00026] Alternatively or additionally, adjustments can be made in order to increase the accuracy of the information provided in step (c), for example, so that better local mathematical optimization problems can be defined. When the information is improved, then better production recommendations can be computed in subsequent iterations, and therefore this can provide a way to improve the iteration process for improvements to the flow network by changing output parameters.
[00027] In an additional alternative, which can also be performed in addition to (or in parallel) to the above enhancements, the enhancement to the flow network may comprise adjusting operational parameters of flow network components in order to increase the service life of those components or other components of the flow network, preferably without compromising other aspects of the performance of the flow network. Therefore, for example, one constraint applied might be that overall production should remain at or above a given level, while another constraint might be that there is a maximum flow rate for given parts of the flow network to avoid overworking certain components. and therefore extend its service life.
[00028] In steps (d) and (e), the nature of adjustment to control points will vary with different types of control points. For example, when the control point is a valve or similar, then the adjustment will be an opening or closing movement of the valve. Often a control point will automatically adjust gradually. In this case, the adjustment can comprise one or more steps.
[00029] The analysis in step (c) preferably comprises the computation of the relationship between the excitations applied as an input to the control points in step (a) and the effect on the measured flow parameter(s) as an output in step (b). In preferred embodiments, the analysis may include a step of finding a ratio between the input amplitude of the excitations and the output amplitude of the resulting variation in the flux parameter(s). As noted above, it is preferred that the analysis includes the creation of a model.
[00030] This analysis can use any analysis method that has the ability to link input excitations to output variations. A preferred example of this is a frequency analysis that uses multiple frequencies applied in parallel, as described in more detail elsewhere in this document. This allows for a quick and effective online experiment. Minimizing the time required to perform steps (a) and (b) provides significant advantages. Alternatively, the analysis can link output variations to input excitations based on the timing or sequence of the input variations and a corresponding timing or sequence of the measured output variations, for example, as described above in relation to sequentially applied excitations and the use of bandpass filters. Once the relationships between the excitations and the flux parameter(s) are known, then it is trivial to create a linear model, for example, by dividing the excitation amplitude with the corresponding flux amplitude in order to obtain a mapping linear between all inputs and all outputs of the system. Determining a fit in step (d) may involve a simple comparison of the relationships in step (c) above to identify the fit that will generate the greatest improvement in the output parameter(s) of interest. The enhancement can be a positive or negative change in the output parameter depending on the nature of the optimization. In the total production enhancement example for the oil and gas production flow network, the required enhancement would be an increase in an output parameter that is related to total production.
[00031] In preferred modalities, step (c) includes the creation of a local optimization model, for example, by combining the linear input-output model. Step (d) may include optimizing the model to determine the necessary fit. When a suitable model is created then the user can define an objective/goal (eg a flow parameter of interest to maximize or minimize), system limit constraints and operational constraints, it is possible to have the ability to build a model of local optimization that can suggest changes to improve production.
[00032] The method may include a step of advising users of the analysis results through a control or support system. Control points can be automatically adjusted through a control system to improve performance based on the determined adjustment. Alternatively, the determined adjustment may be presented to the flow network operator as a proposed adjustment in order to allow the operator the choice of following a proposed adjustment or applying an alternative adjustment based on the operator's judgment. With either alternative, there are significant benefits to the control method as the further analysis of the flow network performance in steps (a) to (c) repeated is based on actual measured values. This leads to better knowledge of the system and ensures that the performance of the flow network can be iteratively improved by adjusting the control points towards an optimal configuration. For example, well production rates can be controlled to optimize production for the oil field or a group of wells.
[00033] In modalities where oscillations are used for excitations, the excitation application step may include sending control signals to the equipment at the control points and/or may include the flow and/or pressure control step at the points of control. Excitations can be applied through existing control devices such as existing valves or pumps and so on. Using existing valves in this way means that the method may require no modification to existing equipment to apply excitations to the flow network, other than changes to a flow network control system to implement the required control of valve opening/closing .
[00034] The oscillations are preferably approximately sinusoidal, eg waveform applied through gradual changes in valve position in order to approximate a sine wave. The use of a sine wave, or an approximation of it, provides accurate results when the output data is analyzed using conventional frequency analysis techniques such as Fourier transform-based techniques.
[00035] In a particularly preferred embodiment, the method is applied to an oil and gas production flow network. In this case, the control points can be control points to control flows and/or pressures from wells in the oil and gas production flow network, for example, control points at wellheads and at a base of a riser . Preferably, throttling valves and/or gas lift rates (at wellheads and a riser base) are used to apply the excitations to the flow rate of the wells. The throttling valves can be easily controlled to gradually open and close to apply a waveform of the selected frequency at the flow rate. Gas rise rates can also be easily controlled to increase or decrease the rate gradually to apply a waveform of the selected frequency to the flow rate. It is preferred to use applied excitations with throttling valves and gas lift rates.
[00036] Preferably, the method includes selecting frequencies for excitations based on characteristics of a typical frequency spectrum for the flux network. This allows frequencies to consider the underlying frequency spectrum that occurs at typical variations in pressure, flow rate and/or temperature that occur during normal operation of the flow network and thus can allow frequencies to be selected to prevent frequencies where factors such as damping or noise can interfere with the analysis results. The frequency spectrum for the flow network can be a measurement of combined pressures or flow rates over a period of time, for example over several days.
[00037] The method can therefore include selecting frequencies for the oscillations by performing the frequency analysis of the production waveform and identifying a suitable frequency range, preferably a frequency range with low damping and low noise . In the example of oil and gas production flow networks, it was found that production waveforms typically exhibit damping and therefore reduced amplitudes at higher frequencies and that, at lower frequencies, there is noise from operational processes. Thus, preferably, the frequency range is selected by removing higher frequencies that exhibit damping and/or by removing lower frequencies that are obscured by noise.
[00038] The frequency analysis used in identifying suitable oscillation frequencies is preferably based on a Fourier transform. The use of a fast Fourier transform (FFT) algorithm is preferred, as this provides considerable advantages in terms of analysis speed.
[00039] When a suitable frequency range is selected, it is necessary to determine the test frequencies to be used for the well oscillations. The step of determining the test frequencies preferably includes determining frequency ranges in the frequency range that will not interfere with each other. The step of determining frequency ranges preferably includes determining a spacing for the frequency ranges based on the number of frequencies required and/or the total available test period.
[00040] The selected frequencies should avoid interference with each other and with significant harmonic. Thus, they should be far away and should avoid the main harmonic (2nd harmonic) of other test frequencies. The main harmonic will be double the test frequency. So, for example, if a first test frequency is defined as 0.1 mHz, then this means that 0.2 mHz should not be another test frequency.
[00041] In a Fourier analysis, the total test period needed to provide the resolution for a given frequency spacing is the inverse of the frequency spacing. So, for example, a 0.5 mHz spacing requires a minimum total sampling time of about 30 minutes and a 50 μHz spacing requires a minimum total sampling time of about 6 hours. A large reduction in frequency spacing can therefore result in an excessively long test time. The frequency spacing can be selected to ensure that the total test time is limited to 60 hours or less (ie a spacing of 5 μHz or above), preferably 12 hours or less (ie a spacing of 25 μHz or above), most preferably 6 hours or less (ie, a spacing of 50 μHz or above).
[00042] The number of frequencies required will refer to the number of control points that need to be excited. In the simplest case, the method may include selecting numerous frequency ranges that will provide available test frequencies for the total number of control points to be excited. However, for large numbers of control points, it is not necessarily desirable to simply divide the available frequency range into sufficient frequency ranges to provide available frequencies for all control points. To allow testing of large flux networks that have many branches without the need to use an undesirably small frequency spacing, the method can include grouping the control points and exciting the oscillations into batches of control points. Control point groups can each include 2 to 25 control points, preferably 5 to 20 control points.
[00043] The amplitude of the oscillations should be defined to ensure that the frequency analysis provides results that can be distinguished from the baseline amplitude of frequency spectrum variations for the flow network, for example, the amplitude can be defined in an order of magnitude greater than the amplitude for the selected frequency range in a normal frequency spectrum for the flux network. The amplitude of the input oscillations can be in the range from 10 to 10000 Sm3/h, preferably 50 to 1000 Sm3/h. Production constraints or other constraints in the flow network can set a maximum for the amplitude, as an increase in amplitude can give rise to a decrease in production. The method may include determining a baseline amplitude for the selected frequency range by determining a best-fit line for the frequency/amplitude data, for example, by a least squares analysis. The amplitude for the input swings can then be set to at least five times the baseline, preferably ten times greater. All oscillations can be applied to the same amplitude, which could be, for example, a factor greater than the average baseline amplitude for all frequencies. This simplifies control of the control point mechanisms used to apply oscillations. In a preferred embodiment, the amplitudes for each test frequency are scaled to match the baseline amplitude at the test frequencies. This can improve accuracy while avoiding unnecessary loss in production. This allows precision to be set to a desired minimum based on baseline amplitude, without introducing unnecessarily large amplitudes.
[00044] Measured flow parameters, such as pressure, flow rate and/or temperature, for the example of an oil and gas production flow network, may include one or more of wellbore pressure, wellbore temperature, wellhead pressure, wellhead temperature, oil flow rate, gas flow rate and/or water flow rate. The method may include measuring this data, for example, by means of sensors placed to capture the flow in the relevant flow passages. Flow measurements for the flow rate of the total flow or separate flow(s) can be taken at any point downstream of the production head. Preferably, flow measurements are taken at a point downstream of a separator that receives flow from the production head. After the separator, further measurements are possible as they can be measurements from the separate streams.
[00045] The step of performing a frequency analysis to determine the pressure, flow rate and/or temperature variations induced by the applied oscillations may include the use of a Fourier transform presented above, preferably an algorithm of Fast Fourier transform (FFT). This produces an output frequency/amplitude graph where the effects of the ripple frequencies can be seen. The method preferably comprises determining the properties of the different branches of the flow network by determining the flow rate and/or output pressure amplitude values at the test frequencies and, using these amplitudes, to determine the properties of individual branches, or groups of branches. The baseline amplitude for the outflow pressure/rate can be determined by removing data points related to the test frequencies and their second harmonic and then determining a best fit line for the remaining results, by example, through a minimum frame analysis as above and this baseline range can be used to provide an indication of the accuracy of the results.
[00046] Additional properties of the flow network can then be calculated based on the flow and/or pressure data. For example, in the case of an oil and gas production flow network that uses data related to oil flow rate and flow rate error propagation theory can be used to determine water content (WC) and the productivity index (PI). Given Ao and Aw as amplitudes for oil and water flow rates, respectively, then WC = Aw/(Ao+Aw). Similarly, PI = Ao/Ap, where Ap is the well pressure range. In addition, GOR = Ag/Ao where Ag is the gas flow amplitude and IPR can be calculated by PI measured at two operating points or using the second harmonic if the input is large enough. Any of these parameters, or any other parameter from a different flow network, can be selected as the parameter that is enhanced by adjusting the control point(s).
[00047] The step of applying oscillations can include the application of different frequencies in different phases. If the oscillations are all applied in phase, then this creates a large peak in the cumulative effect on the total flow rate of the combined branches. This is not a problem in all flow networks, for example, in an oil and gas production flow network when production is very limited, as the effect of fluctuations on production output will be the same whatever it is the phase relationship. However, it can have an adverse effect in some scenarios, for example, in an oil and gas production flow network when production has limited process. Thus, in a preferred modality, the phases of the applied oscillations are shifted to reduce variations in the combined outflow of the flow network.
[00048] The method may include a second harmonic level measurement step for the applied test frequencies. This can be used as a test to check nonlinearity in the system and thus the validity of the local mathematical optimization problem, since if the second harmonic is low then this is a good indicator of an absence of major harmonics. In addition, the second harmonic amplitude can be used in conjunction with the test frequency amplitude to determine higher order polynomial model parameters for the flux network.
[00049] Viewed from a further aspect, the present invention provides a control apparatus for controlling a flow network to improve the performance thereof, wherein the apparatus comprises: a controller for applying excitations at multiple control points within the flow network. flow, where multiple control points are on different branches of the flow network; and a data analysis device for receiving measurements of changes in one or more flow parameters on one or more flow paths where flows from more than one of the different branches have been combined, to perform an analysis of the flow parameter measurements to identify variations induced by applied excitations and to determine a proposed adjustment to be made in one or more of the control points in order to improve the performance of the flux network; wherein the controller is arranged to make the proposed adjustment to the flow network control points or an alternative adjustment decided by the flow network operator; and wherein, after an adjustment is made, the controller is arranged to repeat the excitation of the control points and the data analysis device is arranged to repeat, thereafter, the measurement, analysis and determination steps to thus allow for an improvement iterative to the performance of the flow network.
[00050] The controller can control the control points to apply excitations by sending control signals to the control points. In some preferred embodiments, the apparatus includes control points, which may be at points distributed along the flow network. Alternatively, the control points can be a part of a different apparatus, while being directly or indirectly controllable by the controller.
[00051] The control points, the flow parameter(s) and the output parameter(s) may be as discussed above in relation to the first aspect of the invention. The applied excitations can be as discussed above in relation to the first aspect of the invention. The apparatus may optionally include the use of applied excitations in a flux network simulation and the measurement of the simulator's response. Using a simulator can increase real-world measurements as described above.
[00052] The controller and data analysis device can be separate or can be combined into a single device, for example a computer device for flow network control and flow network data analysis.
[00053] Still further noted, the present invention provides a computer program product with instructions for execution in a data processing apparatus, wherein the apparatus includes hardware or software connections to allow excitations to be applied at multiple control points within a flow network and, optionally, in a simulator of all or part of the flow network; wherein the instructions, when executed, will configure the data processing apparatus to execute a method as described in the first aspect above.
[00054] The computer program product can configure the apparatus to perform the method steps as in any or all of the preferred features presented above. The data processing apparatus may include features as discussed above for the apparatus of the second aspect. The apparatus may include an interface to provide the proposed adjustment to the user and/or to receive input from the user to adjust the control points.
[00055] It will be understood that, in the above discussion, the flow path(s) in which the flows were combined will be flow paths that are downstream of the control points, although in some situations, such as with pressure variations , upstream effects may occur, requiring upstream measurements.
[00056] Certain preferred embodiments of the invention will now be described by way of example only and with reference to the accompanying drawings, in which: Figure 1 is a flowchart showing a preferred embodiment of a method of controlling a flow network; Figure 2a is a schematic view of a typical definition for oil and gas wells; Figure 2b shows an array of field equipment in an oil field simulation used to demonstrate a well test method; Figure 3 shows oil pressure and saturation for the oil field simulation; Figure 4 illustrates the results of a frequency analysis of real world production data from an oil field; Figure 5 is a graph derived from the oil field simulation that shows variations in production flow rate during a standard build-up test campaign and variations in production flow rate during a well test using swing input in the wells; Figure 6 shows the results of a well test frequency analysis in Figure 5 including wellbore pressure, water flow rate and oil flow rate; Figure 7 is a graph derived from the oil field simulation that shows variations in production flow rate during a standard build-up well test and variations in production flow rate using swing input into the wells, with the adding noise; Figure 8 shows the results of a frequency analysis of the data in Figure 7 including wellbore pressure and flow rate; Figure 9 shows an alternative field equipment definition; Figure 10 shows a time series for simulated pressures in three wells and a pipeline when excitations are applied to throttling valves and also to gas lift valves for the wells; Figure 11 shows the results of a frequency analysis of the pressure and flow measurements of Figure 10 Figure 12 is a diagram of an exemplary flow network comprising ten production wells with a common pipeline; Figure 13 shows a time series for simulated pressures in the ten wells and in the pipeline when excitations are applied; Figure 14 shows the results of a frequency analysis of the data in Figure 13.
[00057] The preferred embodiments described in this document are for controlling a flow network to improve the performance of the flow network. This is done by analyzing the response of the flow network to incoming excitations and analyzing a small adjustment that will result in an improvement in the performance of the flow network, under given constraints. These adjustments can be done iteratively, in which each moment considers the analysis of the flow network and its behavior after the implementation of the previous adjustment.
[00058] The basic principle, as shown in Figure 1, is as follows: 1. Use the current operating point as the starting point for the search for a proper adjustment. Excite the control variables at that operating point to build a flow network model and/or derive a flow network model from a simulation to obtain a model centered on the current operating point. The model can be a simple model such as a linear localized model. 2. Since the model of the well and production system is very close to this operating point and more unsatisfactory in the opposite direction, a proposed new operating point should be in close proximity to the current operating point. 3. Search for a new operating point in the domain where the current model is valid, in order to provide improved operation for the flow network. In the preferred mode, this is done by optimizing the model. 4. Propose the new operating point to the flow network operator and adjust the flow network to move it to the new operating point or optionally to a different operating point chosen by the operator. 5. Preferably, after allowing time for the flow network to stabilize, the excitation/modeling is repeated to gather new information at this new operating point, and then the method repeats the process of finding another new operating point to move. iteratively towards an ideal solution.
[00059] To summarize, compared to known simulation techniques for optimization, the proposed method focuses on small gradual improvements rather than a large step towards the ideal solution. The result is a more robust and acceptable “online” production optimization concept than a conventional optimization approach.
[00060] This approach essentially solves a simple linear program, quadratic program, or smooth nonlinear program iteratively. However, in the preferred mode, this does not create the model only by linearizing the simulator at the work point. Instead, this is done by linearizing the real system, that is, by creating part or all of the model directly from real system measurements. This is a second core difference between preferred modalities and a regular optimization approach.
[00061] A consequence of this approach is that the operator becomes an active part of an ideal solution as the user can intervene at each new operating point. The user can deploy their preference based on intuition and experience, the recommendation of the proposed control method, or a combination of both.
[00062] The preferred modality extends to the use of techniques developed for the purpose of well testing in step 1 above, in which oscillations are applied as input excitation at control points within the flow network, for example, in throttling valves that control pressure and/or flow in wellheads. First, it is useful to understand the well testing method before considering the currently proposed method of optimizing flow networks. However, it will be understood that the current method is not limited to use with oscillations applied to wells or flow networks that form all or part of the oil and gas production system.
[00063] The well test method is described in WO 2013/072490 by Sinvent AS and Norwegian University of Science and Technology (NTNU). Parts of the disclosure of it are thus repeated below to help understand the currently proposed optimization method. The well testing method of WO 2013/072490 provides a significant advance in that field as it has allowed the properties of individual wells to be determined without the need to carry out individual tests for each well and without the need to close the wells. A dedicated test head is not required and this can reduce the complexity and cost of field equipment.
[00064] Production continues through the production head over the course of the test, and although the applied swings will likely reduce the average flow rate, the reduction in production is low compared to the reduction in production for a conventional test such as a accumulation test. For a field with ten wells, production during a test campaign can be greater than 4% for the well test method described in this document compared to an equivalent accumulation test. Wells are tested in parallel with measurements from each individual well being determined by looking at the effects of the oscillation frequency applied to that well. Through frequency analysis, these effects can be isolated from other variations in the output stream. Testing takes place with online production and with normal flow patterns during multi-well flow mixing. Thus, unlike conventional testing, due to the fact that there is no disruption of the well during testing, then there is no subsequent reconciliation of measured results to account for changes in flow patterns that arise from the testing process. This removes a source of error from the test procedure.
[00065] A typical definition for conventional well testing is shown in Figure 2a. A production head 2 connects through a production flow line 4 to a production separator 6. In this case, the production head 2 is connected to three wells 8. The wells 8 in this example are extraction oil and gas from the same oil field. Each well 8 is connected to the head by a master valve 10, side valve 12, throttling valve 14 and check valve 16. Isolation valves 18 connect the production head to the lines of the wells 8. The lines of the wells are also connected through another set of isolation valves 19 to a test head 20. Test head 20 is a dedicated head used for testing purposes only. It connects to a test flow line 22 in a test separator 24. The production separator 6 and the test separator 24 are tanks that separate oil and gas. Under the influence of gravity, the oil settles at the bottom of the tank, where the gas occupies the space at the top of the tank. Each separator 6, 24 is equipped with a pressure control line 26 which connects the gas-laden head space of the separator 6, 24 to a valve in the gas outlet line 30. For each of the production separator and the separator There is also a sampling and measuring device 28 in the oil outlet line 32 after separation, the oil and gas are piped separately for further processing through the oil outlets 32 and the gas outlets 30. The separators 6, 24 may also include a water outlet for extracting water from under the oil.
[00066] During conventional well testing, one well 8 is tested at a time using prior art methods by controlling the flows in production head 2 and dedicated test head 20. The current well test system avoids the need to test only one 8 well at a time and instead allows multiple 8 wells to be tested in parallel.
[00067] As described above, the newly developed well test system involves the use of oscillations applied to wells 8 at defined frequencies using throttling valves 14. A different frequency is used for each well 8, thus allowing them to be obtained data over multiple 8 wells simultaneously through the use of a subsequent frequency analysis. In frequency analysis, different frequencies are used to mark data related to a particular well 8. There is no need to adjust flows compared to normal production and thus this test method provides data that directly refer to properties of the 8 wells during normal production.
[00068] With an equipment definition of the type shown in Figure 2a, the inlet excitations are implemented by applying swings in pressure and flow rate from the wells 8 through the choke valves 14. A typical choke valve 14 can be opened and closed in 200 steps over a period of about five minutes. Throttling valves 14 can therefore be used with a suitably configured controller to apply fluctuations in flow rate over a wide possible frequency range.
[00069] During the application of these oscillations with throttling valves 14, the well test also includes the measurement of pressures at the wellhead and wellbore, the measurement of flows for oil and gas outside the separator 6 and also measurements of water flow rate if the separator also allows for the separation of water from oil. Testing can also include gathering data related to gas to oil ratio, water content, and so on. Measured data is analyzed and effects arising from the outputs of the various wells are identified based on a frequency analysis of the type discussed below.
[00070] The frequencies that are used are determined based on the characteristics of the oil field and wells as defined in more detail below with reference to Figure 4, which is taken from WO 2013/072490. According to several different examples of a well test method, oscillations can only be applied for a short period of time as described below with reference to Figures 5 and 6. The analysis in relation to those Figures is derived from a simulation of an oil field using the Eclipse black oil simulation modeling package as fueled by Schlumberger Limited. The test model used an arrangement of 20 square blocks by 20 blocks high with a large spacing of 25 square meters by 10 meters high. Permeability was set at 300 mD and porosity at 25%. In the model oil saturation was set to 0 in layers 1 and 7 and in layers 15 to 20 and above zero in the central layers, with a maximum of layers 9 to 12. Pressure increases along the layers as is conventional. The simulation includes 10 vertical wells, drilled in layer 12. For the purposes of the Eclipse model, the field equipment is as shown in Figure 2. Figure 3 shows the pressure and oil saturation for the model.
[00071] For illustrative purposes, Figure 2 shows only two of the ten wells 8. Wells 8 connect through throttling valves 14 to a production head 2 which then feeds into a production separator 6 in a manner similar to the system described above in conjunction with Figure 1. The production separator 6 has an oil outlet 32 with an oil flow rate Fo, a gas outlet 30 with a gas flow rate Fg and also an outlet of water 34 with a water flow rate Fw. As with the system in Figure 1, there is a pressure control 26. As noted above, during operation in a real world system, it is anticipated that oscillations according to the currently proposed methods will be applied across the wells by through the wellhead choke valves 14. The choke valves 14 would be opened and closed in order to induce fluctuations in the wellhead flow rate and pressure. However, for the purposes of this model and due to restrictions in Eclipse modeling packaging, variations in wellhead pressures are applied in a simulation not by a throttling valve 14, but rather by a pressure variation of simulated wellhead created by the software. Of course, it will be understood that the end result is the same. The definition of field equipment in Figure 2 refers to platform wells without a subsea pipeline, although well testing is not limited to this definition only. An alternative definition is discussed below in conjunction with Figure 19.
[00072] In order to achieve the best results using the proposed iterative production optimization method with built-in experiments, it is important to select an appropriate set of frequencies that will allow multiple wells to be tested simultaneously, where frequencies minimize interference between each other, and where it is possible to clearly identify the induced oscillations in the oil field outputs, that is, in measurements of wellbore pressure, oil flow rate and water flow rate during the test procedure. It will be understood that in wellbore pressures and outflow rates for an oil field, there are ongoing variations in production rate. Figure 4 shows a production waveform based on real world data from a multi-well oil field as used in WO 2013/072490 by Sinvent AS and Norwegian University of Science and Technology (NTNU). The production stream oscillates considerably over its average flow rate and the production waveform also includes a degree of noise. In order to determine the frequency components of this signal, a Fourier transform is applied. Various Fourier transform variations can be used as distinct Fourier transforms and distinct time and distinct frequency transforms, and so on. A fast Fourier transform (FFT) algorithm can also be used and is preferred as FFT tends to be considerably faster and more efficient in terms of computational power. The input to frequency analysis is a production waveform for total pressure or flow rate for an oil field and the output is a complex data series whose absolute value can be presented as shown in Figure 4 as a series of points showing frequencies and amplitudes of those frequencies.
[00073] As it will be seen in Figure 4 that, at relatively large frequencies, that is, frequencies in excess of 1 mHz and approaching 10 mHz and above, there are some damping effects and, thus, increasingly high frequencies have a generally decreasing amplitude. Also, at low frequencies, below 0.1 mHz for this example, high amplitude events start to appear as a consequence of processes that occur during the oil production operation and this creates excess noise in the system. Similar phenomena will be observed in the production waveforms for other oil fields. The frequency window for oscillations to be applied to the wells should be selected to avoid these problems. Therefore, in that case, an appropriate frequency window for selecting frequencies that should not experience damping and should be easily distinguishable from other frequency components of natural variations in the oil-laden production flow would be within a frequency window between 0.1 at 1 mHz, which is almost equal to periods between 15 minutes and two and a half hours.
[00074] Frequencies in this type of range are expected to be appropriate for many oil fields. However, an analysis of production data should be performed for each oil field in order to find an appropriate set of frequencies that can be used in order to provide effective results from the well test method. Another point to note is that while frequencies within the 0.1 to 1 mHz window are generally best for that particular oil field, it may also be useful to consider higher frequencies for some types of testing, such as composition tests, since for all compositing tests, dampening effects that arise at high frequencies will not be an issue. When the optimization method described below is used in other industries, then different frequency bands may be required, but they can be designed using a similar methodology.
[00075] Once a frequency range is determined, it is also necessary to select appropriate frequencies in that range. The selected frequencies should avoid interference with each other and with other significant harmonics. The relationship between downhole pressure and wellhead pressure is non-linear and therefore a second harmonic and possibly more is expected to be produced. Ideally, the second harmonic should be checked to see if it is small. A small or negligible output at the second harmonic of the input frequency is an indicator that there are no upper harmonics and there is no problem with non-linearity that could distort the analysis results. Test frequencies should therefore be selected to avoid frequencies that will be affected by or will mask the second harmonic of other test frequencies. Therefore, for example, if a test frequency is set to 0.1 mHz, then 0.2 mHz should not be used as another frequency in a test. Similarly, if a frequency is set to 0.15 mHz, then 0.3 mHz should not be used as a frequency for another well in the test. In addition, the selected frequencies should have a spacing that is small enough to provide a large enough total number of frequencies to cover all wells, but large enough to avoid excessively long sampling times. The total sampling time required is the inverse of the minimum spacing between selected frequencies.
[00076] In the present example with a frequency window from 0.1 mHz to 1 mHz, so in order to test the simulated oil field with ten wells, naturally ten frequencies are needed. Since numerous frequencies will not be available for use, then, in order to obtain ten test frequencies, it is necessary to consider frequencies spaced close enough to produce something more than ten frequencies. This can be done by providing twenty frequency ranges, allowing up to half of the frequency ranges to be removed by conflict between harmonics and so on. For this example, the available range for test frequencies is 0.1 mHz to 1 mHz, so the potential frequency ranges should be separated from each other by 50 μHz to provide twenty possible frequencies. With a frequency spacing of 50 μHz, then the total time required to complete the test in order to provide a complete set of frequency analysis results will be six hours. This compares very favorably to the minimum total time for an equivalent accrual test campaign, which can require five days.
[00077] With the frequency range of 0.1 mHz to 1 mHz and a spacing of 50 μHz it is relatively straightforward to determine a set of frequencies are available and do not conflict with the second harmonic of other frequencies. A possible set of frequencies is 0.1 mHz, 0.15 mHz, 0.25 mHz, 0.35 mHz, 0.4 mHz, 0.45 mHz, 0.55 mHz, 0.6 mHz, 0.65 mHz , 0.75 mHz and 1 mHz. In the present exemplary simulated oil field, which has ten wells, it is possible to select ten of these eleven frequencies to be applied to the ten wells. As noted above, swings should be applied to the wells for a minimum time period of six hours.
[00078] Figure 5 shows the total outflow rate for the simulated oil field of Figure 3 when it is tested in a conventional accumulation test and also when it is tested over a six hour period using the new method with oscillation frequencies selected from those mentioned above. The two different test regimes can be easily distinguished. In the accumulation test, there are ten evident cycles, including a significant drop in overall flow when one well after another is closed and then started again. For the oscillation-based test method, a considerably shorter time period is required. As will be seen in the Figure, the accumulation test takes place over five days, whereas the swing-based test only takes six hours. The swings are applied during the final six hours of the chart. It will be understood that, as a consequence of the act of avoiding the need to stop each well, production in turn during the test campaign is enormously increased and production remains uninterrupted only with changes being applied oscillations of normal changes flow rate and wellhead pressure until the final six hours of the week period. The consequence of this is that the production using this six-hour swing-based well test method is considerably higher in total compared to the test period. With the example shown, the total production is approximately 4.3% greater than the production when the accumulation test is run. In the example shown, the flow rate for the oil field is on the order of 6,000 m3 per hour and this means that the production added over the test period shown can be around 42,000 m3. This added oil production would be worth several tens of millions of dollars at current rates, therefore providing a significant benefit.
[00079] While the test is run with oscillations being applied to the resulting changes in wellbore pressure, the water flow rate and oil flow rate are measured and then subjected to a frequency analysis of the type described above. The results are shown in Figure 6. As can be seen, there are evidently identifiable oscillations in the test results that correspond to the input oscillations at frequencies of 0.1 mHz, 0.15 mHz, 0.25 mHz, 0.35 mHz, 0.4 mHz, 0.45 mHz, 0.55 mHz, 0.6 mHz, 0.65 mHz and 0.75 mHz. This can be seen most evidently in the wellbore pressure measurement, but they are also evidently identifiable in the flow rate measurements. Since the amplitude of the input swing is known, then by measuring the amplitude of the output swing, it is possible to determine well properties.
[00080] As an example, one can consider oil flow Fo, water flow Fw, and well bore pressure p for wells 1, 5 and 10 (frequencies 0.1 mHz, 0.4 mHz and 0 .75 mHz) and the information that can be derived from the results shown in Figure 6. Oil and water production and wellbore pressure can be read from the appropriate graph in Figure 6 and the background noise, which can also be read from Figure 6, and used to estimate the uncertainty in value: For well 1, Fo = 30 ± 14 Sm3/h, Fw = 3.7 ± 1.5 Sm3/h, p = 3.3 ± 0.07 MPa (0.7 bar). For well 5, Fo = 33 ± 5 Sm3/h, Fw = 4.3 ± 0.5 Sm3/h, p = 2.81 ± 0.03 Mpa (0.3 bar). For well 10, Fo = 31.4 ± 2 Sm3/h, Fw = 4.2 ± 0.2 Sm3/h, p = 2.74 ± 1 Kpa (0.01 bar).
[00081] So, one can use error propagation theory to calculate the water content (WC) and the productivity index (PI): For well 1, WC = 0.11 ± 0.07 and PI = 9, 1 ± 4.5 Sm3/h bar For well 5, WC = 0.115 ± 0.02 and PI = 117 ± 2.2 Sm3/h bar For well 10, WC = 0.118 ± 0.01 and PI = 115 ± 0.7 Sm3/h bar
[00082] It is evident that the uncertainty is too large for well 1 due to the fact that the uncertainty is too large at low frequencies.
[00083] Uncertainties can be reduced by extending the test period. If, for example, the 6-hour test is extended to the five-day test period for the equivalent accumulation test, the uncertainties in the estimates are considerably reduced and the following is reached: For well 1, Fo = 31.3 ± 2 Sm3/h, Fw = 6.6 ± 1.5 Sm3/h, p = 3.31 ± 5 Kpa (0.05 bar)
[00084] These values are much more accurate and similar improvements can be calculated for the other estimates.
[00085] It should be noted that the oscillations could be applied over a longer period of time than that shown in Figure 5. For example, the oscillations could be applied for the entire period of the accumulation test. A longer period of time to apply the swings can improve the data that is obtained, but the downside is that production losses are increased. It is also possible to scale the oscillations, with different amplitudes for different frequencies; in particular, it may be useful to increase the amplitude at lower frequencies as larger oscillations at lower frequencies can increase accuracy.
[00086] Figure 7 shows another set of simulation data in which the frequency swing well test is applied over a period of five days. In Figure 7, random measurement noise is added to the data in order to simulate noise that can be shown in real world data. Figure 8 shows the result of a wellbore pressure and oil flow rate frequency analysis for the noise data in Figure 7. Noise often creates additional data points at higher frequencies, which are out of range. frequency selected for the applied oscillations and thus does not reduce the accuracy of the estimates.
[00087] Other alternatives and refinements are possible, for example, following the techniques proposed in WO 2013/072490 by Sinvent AS and Norwegian University of Science and Technology (NTNU). In this way, oscillations can be applied during the interruption and the start of the well without significant adverse effect. Also, a phase difference can be applied to the oscillations in order to avoid peak synchronization. A synchronization of peaks would undesirably increase production losses.
[00088] As noted above, although the simulation uses field equipment based on platform wells without a subsea pipeline, it is also possible to make use of the swing-based well test method in other equipment definitions. Figure 9 shows an arrangement with a subsea pipeline 36 that connects to a platform choke 38. As in Figure 2, only two wells 8 are shown, although naturally a larger number of wells 8 can be connected to the pipeline 36 Wells 8 connect to subsea pipeline 36 through valves 14. With this arrangement, when an oscillation is applied to valves 14, a problem can arise as pipeline 36 can oscillate at both frequencies. The reason this can occur with a subsea piping arrangement as shown in Figure 9 is that there is often no pressure control of the subsea piping 36. Piping pressure can therefore vary and will be affected by throttling valve oscillations. Thus, when a first valve 14 is excited with frequency WI and a second valve 14 is excited with frequency W2, then there is frequency leakage and the pipe 36 can oscillate with both frequencies WI and W2. As a consequence, both wells 8 will be excited at the two frequencies and the frequency marking of wells 8 is lost. This does not prevent the application of the swing-based method of well testing in settings that use a subsea pipeline, but steps must be taken to avoid frequency leakage.
[00089] One solution is to apply pressure control to pipe 36. Another solution is to use supersonic flow in check valves 14. Many wells already use supersonic flows and existing systems could be adapted to use supersonic flow rates. With supersonic flows, the pressure in the pipeline will have no effect on the flow rates through the valve and any pressure variations in the pipeline are essentially invisible to the valve and to the flow and pressure on the opposite side of the valve. The throttling flow will only be affected by the well pressure and the throttling position. As a consequence, an oscillation can be applied that will only affect the well connected to that particular choke valve and will not leak into other wells. A more comprehensive solution, which does not require changing well flow regimes, is to consider all frequencies through matrix inversion. In this more general approach, the oscillation amplitudes of all test frequencies are measured in the well pressure of each well that is related to the oscillation and related to the oscillation amplitudes in the oil, gas and/or water flow through the indices of productivity of the wells. The result is for each test frequency, an equation of the form: JI * PI,Í + J2 * p2,i + ... = Fi where Fi is the amplitude of oscillation in the flow of gas, oil or water at frequency i , and pj,i is the amplitude of oscillation in well pressure for well j at frequency i. Having measured all the oscillations Fi and pj,i, the values of the productivity index J can be found by means of matrix inversion. There are also methods available in the literature open to public inspection to calculate error propagation through a matrix inversion.
[00090] As noted above, the preferred mode is for the control of a flow network and can use excitations similar to those used in the well test method described above. It is necessary that excitations create variations in measured flow parameter(s) and that measured variations can, as an output, be analyzed to determine the relationship between input excitations and output variations. These relationships, together with the knowledge of absolute values for the magnitude of input excitations and the magnitude of output variations, allow a model to be created that reflects how changes in inputs affect the output(s). Data obtained through real-world excitation of control points within the flow network can be augmented by data obtained from models derived from simulations of the flow network. This can be useful in cases where excitations cannot be easily applied and/or where there is an adverse effect on the flow network operation if excitations are applied.
[00091] An example will now be described where excitations are applied to throttling valves and gas lift rates in an oil and gas flow network. The simulation includes a dynamic model of three vertical wells, a pipe and a pipeline. It requires the construction of a local model of the flow network over the current operating point. To be able to build this local model, the control points (production bottlenecks, gas or valve lift rates, pump effect, and so on) are excited in sinusoidal patterns at different frequencies to obtain system response information to changes in input changes. It will be understood that these excitations are analogous to those used in the well test method discussed above, but are applied not only to choke valves but also to gas lift rates. Figure 10 illustrates an example of simulated pressures in a simple system with oscillations applied to the system to be able to extract the necessary parameters. Sinusoidal excitations are applied at different frequencies to different control points. The Figure shows a time series of pressures in three wells and pressure in a pipeline with the combined flows (bottom graph) and the resulting variations in flow rates (top graph). Four different frequencies are used for four different control points which, in this example, take the form of a throttle valve and a gas rise rate for a first well, a throttle valve for a second well and a rise rate of gas to a third well.
[00092] Fast Fourier transform is used in this time series to extract frequency information similar to that described above. The excitations, which are at 0.185 mHz, 0.278 mHz, 0.463 mHz and 0.648 mHz, are easily detectable as “four lone points” on each frequency graph in Figure 11. The top three graphs show measured flow rates for oil, gas and water and the other graphs show pressures in the pipeline and three wells.
[00093] As an example, the change in total oil flow in relation to gas rise rate changes in well 2 (with frequency 0.278 mHz), is given by the amplitude at that frequency in the upper left graph of Figure 21. This amplitude can be divided by the amplitude of the gas rise rate swing to obtain a mapping between the gas rise rate input variable/control and the oil rate measurement/output.
[00094] Each control point is allocated a specific frequency. By using the Fourier transform of the relevant measurements (rates, pressures, total temperature, etc.), it is possible to map the effect of a change in a particular input variable/control points to the change in all measurements.
[00095] For a more detailed example, consider a flow network consisting of ten production wells connected to a platform through a pipeline. The simulation includes a dynamic model of 10 vertical wells, a pipeline, a duct and lift gas supply. This is shown in Figure 12. Production head 2 is connected to ten wells 8, each of which has a throttling valve 14. A lift gas system, also connected to each of the ten wells, can apply a gas lift at a rate controlled by the gas lift valve 15.
[00096] In this example, wells 81 to 84 are high-GOR wells that produce at high rates without the use of lift gas. They can be throttled again to reduce production. Wells 85 through 88 are low-GOR wells that need lift gas to produce and therefore also produce with fully open throttle valves. Wells 89 and 810 are wells that produce at intermediate rates without the use of lift gas. These wells can be throttled again to reduce production or injected with lift gas to increase production.
[00097] For this example, it is assumed that it is not allowed to adjust the operating conditions for one of the wells, 810. Due to operating conditions, adjust the gas lift rate to 81 to 84 or for the throttling valve 14 in wells 85 to 88 is not relevant. Therefore, in this scenario, one wants to find suggestions for improving production (ie, increased oil production) by adjusting one or more of the choke valves 14 in wells 81 to 84, the gas lift rates in wells 85 to 88 and choke valve 14 or the gas rise rate in well 89. According to the preferred mode, sinusoidal disturbances are added to choke valves 14, except for choke valve for well 84 and gas lift valve 15, except the gas lift valve from well 88. However, since there are accurate simulation models to control changes in wells 84 and 88, the linear models for these wells are derived directly from simulation models . Excitations are described in the Table below.

[00098] Figure 13 illustrates the resulting variations in flow rates and pressures over a 12 hour period during throttling excitation and gas lift, along with the addition of simulated noise. The top graph shows the total oil, gas and water flow rate in the pipeline. The lower graph shows pressures in the system including wellhead pressures and pipeline (duct) pressure. Again, a Fourier transform is applied and the results are shown in Figures 14 and 15. Figure 14 includes the oil, gas and water rates. Figure 15 shows the pipe line pressure and the ten well head pressures. As in the examples above, it is possible to identify the effect of changes in these eight control variables, observed as eight isolated points on the Fourier plot of each measurement/result.
[00099] In order to use this data to determine what adjustments should be made to improve the performance of the flow network, the analysis includes determining parameters for a model. In this particular example, ten control points are considered. A typical flow parameter of interest in terms of improving flow network performance is to maximize oil production, which will be used as the target for this example. The model should therefore determine the impact of the ten control point adjustments on the level of oil production.
[000100] It is necessary to know the restrictions in the system. For this example, it is assumed that all wells can possibly be pressure restricted and therefore allow the option that all choke valves are fully open. Additionally, it is assumed that the production system has a limitation on the gas available for gas lift and that there is limited handling capacity in total water and gas production. Finally, the model should be limited to suggest only changes to a maximum of four of the control points (inputs). This keeps the model simple and also ensures that the real-world results of the proposed adjustments are more likely to follow the prediction of an improvement. Therefore, it is necessary for the model to find the four best control variables to change and determine the size of the adjustment that should be made.
[000101] To formulate the optimization model, it is necessary to know how a (small) change in the throttle and gas lift rates will affect, 1) the total production of oil, gas and water and 2) the pressure of gas lines piping and wellhead pressures, that is, the effect of all control variables on all measurements. Since factors such as problems with hydrates or erosion are not considered, there is no need to consider temperature measurements or computation of velocities, although it is understood that such factors and measurements could be included in alternative scenarios, which are briefly discussed below.
[000102] Therefore, it is necessary to compute the mapping of all control variables (inputs) for all measurements (outputs):

[000103] The input parameters are respectively the amplitudes at which the throttling valves and the gas rise rates are excited during the experiment. In this case, the values are 0.06 for the throttling valve for wells 1, 2, 3 and 9, and 0.1 for the gas lift rate for wells 5, 6, 7 and 9. Additionally, .- y, ./•' and ./ are the amplitudes of the respective measurements at that frequency i, which can be obtained from the frequency graphs above. The sign of the parameters is determined from the size of the phase shift. .- /, // and are their relative difference. Note that the mapping between the control points for well 4 and well 8 for the outputs is obtained from simulation models rather than experiments.
[000104] As an alternative to using sine waves and the resulting amplitudes to find the coefficients Á.:, one can use steady state (median) values at different defined input points in a finite difference type computation.
[000105] Given the unit y = f(u), operating on * = f(u*). Introduce a temporary step u = u* +Δi.::, giving y*+Δy. = f(u* + Δi-::). For small Δ; ., this can be used to approximate the derivative df/du = A. *+Δy. = y* - y = A Δ;., => A = *+Δy./Δi.::. These estimates of A can then be used in the same post-processing as for the wobble case.
[000106] The steps will be subjected to the same “leakage” effect that occurs in oscillation experiments and this would be treated in a similar way.
[000107] Both the amplitude approach and the multiple steady state approach are batch processes. A more continuous approach can be thought of as the well, by applying online parameter estimation principles from a simple (linearized) model around current operating points. Given the model structure Δ;.-.= A* Δ;:.. With Δy. and Δ;:. known and A* unknown. Let A be the estimate (or initial guess) of A*. The simplest online estimation rule is then given by the instantaneous cost gradient algorithm: dA /dt = r*e*Δi.:: where r is a positive gain and e is the normalized forecast error e = (Δy - A )/nA2 where n is a design parameter. Time delays can be considered at a pre-processing stage, before signals are fed into the estimation scheme.
[000108] With redundant information through multiple measurements and/or simulators or models, it is possible to use weighted least squares methods to decide the model parameter.
[000109] Based on the established objectives of the model, the limitations in the system and the mapping/relationship between input and outputs, the model described below is proposed. The first Table lists all sets and their indices, while the second Table provides model variables. The parameters can be found in the third Table, followed by the objective function and all model constraints.


[000110] It is observed that the wells are represented by the subscript j in the model, including well 10, even if their control variables are not for the optimization in the test case. Control/input variables are represented by index i and given by Δ; :, for changing the throttling openings and gas lift rates. The subscript p represents the phase, that is, oil, gas and water, or a subset thereof. All variables are delta (Δ) variables, meaning that changes from current operating conditions are modeled.
[000111] Objective function: The objective of this example is to increase the oil production, Δ-s..., as much as possible. Δ-;.;, (1)
[000112] System restrictions: total gas and water production (;.,) is restricted on the platform. However, the corresponding variable in the model is not the total production rate, but the change in the total production rate of gas and water. Therefore, equation (2) (2) below constrains the changes in gas and water production rates Δs, given by their respective platform handling capabilities (c,) and total production rates (;.,). Also, there is only a limited amount of gas lift (∑^^ΔL:.) available for distribution between gas lift wells. Similar to the previous equation, Equation (3) below constrains the change in the amount of gas used for gas lift.
[000113] Additionally, the optimization model needs to suggest production strategies that are possible under the pressure point of view. This is ensured by Equation (4) below, stating that the change in wellhead pressure for each well must be greater than or equal to the change in pressure drop across the bottlenecks plus the change in piping pressure. This is derived from the pressure upstream of the choke (t<j which must be greater than the pipeline pressure (t-'-r).

[000114] Operational restrictions: it may be desirable to limit the number of changes that the model proposes. These constraints and the binary variables .•••. ensure that the optimization only allows changes to four control variables. Also, the amount by which each control variable is limited, given by ~:.

[000115] Linear system model: by excitation of the system or using simulation models, it is possible to map all the control/input variables of interest (for example, the gas rise rate in well 1 is not considered in this example) to all relevant measurements/outputs. In this example, changes in throttling settings and gas lift injection rates (Δi-:;), are mapped to changes in total oil, gas and water rates, pipeline pressure, and wellhead pressures . Each of the control variables has its different frequency

[000116] Output parameters of interest may not always be measured if, for example, the fluid velocity for a pipeline is restricted to a maximum velocity due to erosion. Although velocity is not usually measured, it could be estimated based on first order physical effects and calculated as a function of flow rates, pressure and temperature. Equation (13) illustrates a velocity constraint and Equation (14) the velocity model.

[000117] It is also possible to correct exchange or defective equipment if information is automatically provided from a condition-based monitoring system. With reference to this example, if a gas lift compressor needs to reduce speed due to high frequency vibrations, the parameter c-’- in Equation (3) is similarly reduced iteratively until the vibrations disappear. The end result is the possibility to predict, with a high degree of security, the results of (small) changes in the control points in the outflow parameters of interest. Thus, in this example, it is possible to identify what effect settings on throttling valves and gas lift rates will have on total production. Since only a small change in operating point is proposed and since this is based on real world information, then the linear model will be accurate. When the necessary changes are implemented (or, optionally, if the operator decides to make alternative changes based on his experience and judgment) then the process can be repeated. In this way, it is possible to iterate towards an optimal solution while ensuring that the proposed fit never significantly diverges from a route that will provide an improvement. This is not possible with conventional simulation.
[000118] Solving this model for the exemplifying case explained above, the model provides suggested changes to the four control points in the first iteration in order to maximize the oil production target. The Tables below contain information regarding the recommended steps obtained by running the method. In the first Table, the four suggested control point changes are listed and note that these are, in their entirety, throttling definitions.

[000119] The next Table provides the resulting change in production rates, where oil production is the target of the model, while gas and water production constitutes the capacity constraints. This clearly demonstrates that the recommended steps are in the right direction, the oil production rate increases each iteration. Note that, in Step 4, the water exceeded the maximum allowed and therefore, although the oil production rate is higher, the definition used for Step 4 is not allowed and needs to be changed to avoid the restriction on the content of Water.

[000120] In a more general context, the preferred implementation for the control of the flow network is set out below. First, the control points that undergo optimization are excited in sinusoidal patterns with different frequencies or, if there is an accurate simulation model for parts of the system, the simulator can be used instead. Frequencies can be selected to avoid interference. A Fourier transform of all relevant measurements for a suitable period of time (typically the last 6 to 18 hours of production for oil and gas networks) is conducted to find/compute the parameters of the linearized model. A suitable time period can be determined based on the frequency spectrum and/or based on a moving horizon principle.
[000121] The model is resolved to suggest changes in the production strategy. This procedure could be done every 5 minutes. The computational cost is low since the model is simple.
[000122] When suitable for the production engineer, he can look at the model suggestion for production changes. It may decide not to make any changes, implement suggested changes, or something else. However, when the process is stabilized at a new operating point and the process is excited around that operating point for 6 to 18 hours, a new (simple) model is developed and the system will provide a new suggestion. Note that the preferred system provides recommendations for production changes before the system is excited for 6 to 18 hours, however, this recommendation would then be based on the old operating point and the new operating point or with a lower quality estimate of parameter. The process can continue in a cycle to provide continuous research for an improvement in production strategy.
[000123] In addition to operational changes, the model could also be changed/changed to account for new information based on: high frequency monitoring data based on condition of equipment within the flow network, eg when compressor speed needs be reduced due to vibrations and the gas handling capability needs to be realized; planned maintenance where parts of the network are out of order, and capacities and performance are reduced.

权利要求:
Claims (22)
[0001]
1. Method for controlling a flow network in order to improve the performance of the flow network characterized by the fact that it comprises the steps of: (a) applying predetermined excitations at multiple control points (10, 12, 14, 16) within the flow network, where the multiple control points (10, 12, 14, 16) are on different branches of the flow network; (b) receive measurements of changes in one or more flow parameters on one or more flow paths where flows from more than one of the different branches have been combined; (c) performing an analysis of the flow parameter measurements to identify variations induced by applied excitations; (d) determine an adjustment to be made in one or more of the control points (10, 12, 14, 16) in order to improve the performance of the flow network; (e) make the determined adjustment to the flow network control point(s) or make an alternative adjustment decided by the flow network operator; and (f) repeating steps (a) through (e) one or more times to thereby iteratively improve the performance of the flow network.
[0002]
2. Method according to claim 1, characterized in that the measurements are used in the analysis of step (c) to create a model of relevant parts of the flow network, and step (d) includes the optimization of the model .
[0003]
3. Method according to claim 1 or 2, characterized in that step (c) includes creating a local mathematical optimization problem to calculate a fit of one or more of the control points (10, 12, 14 , 16), and step (d) includes solving this optimization problem in order to determine the necessary adjustment.
[0004]
4. Method according to any one of claims 1 to 3, characterized in that step (a) includes applying excitations both to the flow network and to a simulation of the flow network or parts thereof, and step ( b) includes taking measurements from the flow network and the simulation.
[0005]
5. Method according to claim 4, characterized in that the model obtained from the simulation is updated to consider the adjustment made in step (e) when steps (a) to (e) are subsequently repeated.
[0006]
6. Method according to any one of claims 1 to 5, characterized in that excitations are oscillations applied at known frequencies, in which oscillations are applied at different control points of multiple control points (10, 12, 14 , 16) are at different test frequencies and where, in step (c), a frequency analysis is performed.
[0007]
7. Method according to any one of claims 1 to 6, characterized in that the excitations include excitations with known characteristics applied sequentially and step (c) includes the identification of the excitation effects based on timing and/or frequency of the excitements.
[0008]
8. Method according to any one of claims 1 to 7, characterized in that the excitations are applied to more than one type of control point (10, 12, 14, 16).
[0009]
9. Method according to any one of claims 1 to 8, characterized in that a plurality of flow parameters is measured in step (b).
[0010]
10. Method according to any one of claims 1 to 9, characterized in that the analysis in step (c) comprises the computation of the relationship between the applied excitations as an input to the control points in step (a) and the effect on the measured flow parameter(s) as an output in step (b).
[0011]
11. Method according to claim 10, characterized in that the analysis in step (c) comprises finding a ratio of the excitation input amplitude to the output amplitude of the resulting variation of the flow parameter (or parameters).
[0012]
12. Method according to claim 10 or 11, characterized in that the determination of an adjustment in step (d) involves comparing the relationships determined in step (c) to identify the adjustment that will generate the greatest improvement in performance of the flow network.
[0013]
13. Method according to any one of claims 1 to 12, characterized in that it includes a step of reporting the results of the analysis through a control or support system.
[0014]
14. Method according to any one of claims 1 to 13, characterized in that the determined adjustment is presented to the flow network operator as a proposed adjustment in order to allow the operator to have the choice to follow the proposal or apply an alternative adjustment based on operator judgment.
[0015]
15. Method according to any one of claims 1 to 14, characterized in that it is used for an oil and gas production flow network, in which the control points (10, 12, 14, 16) include control points (10, 12, 14, 16) to control flows and/or pressures from wells (8) within the oil and gas production flow network, eg control points (10, 12, 14, 16 ) at the wellheads and at the base of a riser.
[0016]
16. Method according to claim 15, characterized in that the control points (10, 12, 14, 16) include throttling valves and gas lift rates.
[0017]
17. Method according to any one of claims 1 to 16, characterized in that the improvement in the performance of the flow network includes increasing or decreasing one or more output parameters of interest and the parameter(s) of output are therefore the focus of iterative changes in step (e) and iterations of the process.
[0018]
18. Method according to any one of claims 1 to 16, characterized in that the improvement in the performance of the flow network involves one or more of: increasing or decreasing one or more output parameters of interest; increase the accuracy of information provided by the analysis in step (c); or adjust operating parameters of flow network components in order to increase the service life of those components or other flow network components.
[0019]
19. Control apparatus for controlling a flow network to improve its performance, characterized in that it comprises: a controller for applying excitations at multiple control points (10, 12, 14, 16) within the flow network, wherein the multiple control points (10, 12, 14, 16) are on different branches of the flow network; and a data analysis device for receiving measurements of changes in one or more flow parameters on one or more flow paths where flows from more than one of the different branches have been combined, performing an analysis of the flow parameter measurements for identifying variations induced by applied excitations, and determining a proposed adjustment to be made in one or more of the control points (10, 12, 14, 16) in order to improve the performance of the flux network; wherein the controller is arranged to make the proposed adjustment at the control points (10, 12, 14, 16) of the flow network or an alternative adjustment decided by the flow network operator; and wherein, after an adjustment is made, the controller is arranged to repeat the excitation of the control points (10, 12, 14, 16), and the data analysis device is arranged to thereafter repeat the steps of measurement, analysis and determination to thus allow an iterative improvement to the performance of the flow network.
[0020]
20. Apparatus according to claim 19, characterized in that the controller is arranged to perform steps (a) and/or (e), as defined in any one of claims 1 to 18.
[0021]
21. Apparatus according to claim 19 or 20, characterized in that the data analysis device is arranged to perform steps (b) and/or (c), as defined in any one of claims 1 to 18 .
[0022]
22. A computer program product characterized in that it comprises instructions for executing in a data processing apparatus, wherein the apparatus includes hardware or software connections to allow excitations to be applied at multiple control points (10, 12 , 14, 16) within a flow network and/or within a simulator of all or a part of the flow network; wherein the instructions, when executed, will configure the data processing apparatus to execute the method as defined in any one of claims 1 to 18.

类似技术:

---

公开号 | 公开日 | 专利标题

---

BR112015026401B1|2021-05-04|method for controlling a flow network to improve the performance of the flow network, and control apparatus for controlling a flow network to improve its performance

---

CA2657996C|2014-03-25|Method for comparing and back allocating production

---

CN108779667B|2021-09-28|Evaluation of a flow network

---

BRPI1010364A2|2013-04-09|Method of modeling a computer-readable production system having a resident computer-readable instruction set and computing device

---

Karimnezhad et al.2014|A new empirical correlation for estimating bubble point oil formation volume factor

---

Mansoori et al.2015|Pressure-transient analysis of bottomhole pressure and rate measurements by use of system-identification techniques

---

Gryzlov et al.2013|Soft-sensors: Model-based estimation of inflow in horizontal wells using the extended Kalman filter

---

EA036893B1|2021-01-12|Production of hydrocarbons with test separator

---

US9612360B2|2017-04-04|Well testing apparatus and methods for measuring the properties and performance of oil and gas wells

---

US10489523B2|2019-11-26|Apparatuses, systems and methods for performing remote real-time experiments

---

Vefring et al.2006|Reservoir characterization during underbalanced drilling |: Methodology and active tests

---

Varyan et al.2015|Investigation on the suitability of virtual flow metering system as an alternative to the conventional physical flow meter

---

CA3003510C|2020-10-13|Emulsion composition sensor

---

Zenith et al.2015|Well testing by sinusoidal stimulation

---

BR112019022849A2|2020-05-26|FLOW NETWORK DATA RECORDING

---

Mursaliyev2018|Implementation of Virtual Flow Metering Concept in Kashagan Field

---

Dunca et al.2015|Discharge evaluation from pressure measurements by a genetic algorithm based method

---

Apio et al.2019|Comparison of Kalman filter-based approaches for permanent downhole gauge pressure estimation in offshore oil production

---

Monteiro et al.2018|Model Adjusting and Parameters Estimation under Uncertainties for Multiphase Flow

---

Góes et al.2021|Virtual flow metering of oil wells for a pre-salt field

---

Mwalyepelo2015|Analysis of selected Multiphase metering concepts

---

Pappu et al.2022|ACF/PACF-Based Distance Measurement Techniques for Detection of Blockages in Impulse Lines of a Pressure Measurement Circuit for Nuclear Reactors

---

GB2506818A|2014-04-09|Monitoring production of a fluid through a production separator train

---

Holmås et al.2013|Prediction of liquid surge waves at Ormen Lange

---

Cheremisin et al.2009|ADVANCED HORIZONTAL WELL MODEL

同族专利:

---

公开号 | 公开日

---

DK2986815T3|2017-09-11|

---

RU2015146599A|2017-05-22|

---

WO2014170425A3|2015-04-02|

---

MY180874A|2020-12-10|

---

WO2014170425A2|2014-10-23|

---

CA2908802A1|2014-10-23|

---

RU2664284C2|2018-08-16|

---

BR112015026401A2|2017-07-25|

---

US9946234B2|2018-04-17|

---

US20160054713A1|2016-02-25|

---

EP2986815B1|2017-05-31|

---

RU2015146599A3|2018-03-14|

---

EP2986815A2|2016-02-24|

---

GB201306967D0|2013-05-29|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

---

US3559476A|1969-04-28|1971-02-02|Shell Oil Co|Method for testing a well|

---

US3908454A|1972-10-12|1975-09-30|Mobil Oil Corp|Method and apparatus for logging flow characteristics of a well|

---

FR2467414B1|1979-10-11|1984-05-18|Anvar|

---

SU1506458A2|1987-12-14|1989-09-07|Научно-экспериментальный центр автоматизации управления воздушным движением|Device for modeling transport network|

---

CA2019343C|1989-08-31|1994-11-01|Gary R. Holzhausen|Evaluating properties of porous formations|

---

US5144590A|1991-08-08|1992-09-01|B P America, Inc.|Bed continuity detection and analysis using crosswell seismic data|

---

NO325157B1|1995-02-09|2008-02-11|Baker Hughes Inc|Device for downhole control of well tools in a production well|

---

US6615917B2|1997-07-09|2003-09-09|Baker Hughes Incorporated|Computer controlled injection wells|

---

US6333700B1|2000-03-28|2001-12-25|Schlumberger Technology Corporation|Apparatus and method for downhole well equipment and process management, identification, and actuation|

---

US6724687B1|2000-10-26|2004-04-20|Halliburton Energy Services, Inc.|Characterizing oil, gasor geothermal wells, including fractures thereof|

---

US6650280B2|2000-12-08|2003-11-18|The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration|Measurement system and method|

---

FR2818742B1|2000-12-22|2003-02-14|Inst Francais Du Petrole|METHOD FOR FORMING AN OPTIMIZED NEURAL NETWORK MODULE FOR SIMULATING THE FLOW MODE OF A POLYPHASIC FLUID Vein|

---

US6920085B2|2001-02-14|2005-07-19|Halliburton Energy Services, Inc.|Downlink telemetry system|

---

US6675892B2|2002-05-20|2004-01-13|Schlumberger Technology Corporation|Well testing using multiple pressure measurements|

---

NO325614B1|2004-10-12|2008-06-30|Well Tech As|System and method for wireless fluid pressure pulse-based communication in a producing well system|

---

US7565931B2|2004-11-22|2009-07-28|Energy Equipment Corporation|Dual bore well jumper|

---

NZ571239A|2006-04-07|2011-06-30|Shell Int Research|Method for determining the contributions of individual wells to the production of a cluster of wells|

---

NO325342B1|2006-06-19|2008-03-31|Alf Egil Stensen|Control device and flow templates and methods of using the control device.|

---

US7953584B2|2006-12-07|2011-05-31|Schlumberger Technology Corp|Method for optimal lift gas allocation|

---

US8131470B2|2007-02-26|2012-03-06|Bp Exploration Operating Company Limited|Managing flow testing and the results thereof for hydrocarbon wells|

---

US20090308601A1|2008-06-12|2009-12-17|Schlumberger Technology Corporation|Evaluating multiphase fluid flow in a wellbore using temperature and pressure measurements|

---

US8600717B2|2009-05-14|2013-12-03|Schlumberger Technology Corporation|Production optimization for oilfields using a mixed-integer nonlinear programming model|

---

US8902078B2|2010-12-08|2014-12-02|Halliburton Energy Services, Inc.|Systems and methods for well monitoring|

---

US8146657B1|2011-02-24|2012-04-03|Sam Gavin Gibbs|Systems and methods for inferring free gas production in oil and gas wells|

---

US9140108B2|2011-11-03|2015-09-22|Bp Corporation North America Inc.|Statistical reservoir model based on detected flow events|

---

US9816353B2|2013-03-14|2017-11-14|Schlumberger Technology Corporation|Method of optimization of flow control valves and inflow control devices in a single well or a group of wells|BR112015024828A2|2013-03-29|2017-07-18|Schlumberger Technology Bv|method for controlling the flow of a well, method, and system for controlling flow|

---

DE102014004336A1|2014-03-26|2015-10-01|Wilo Se|Method for determining the hydraulic operating point of a pump unit|

---

US9864353B2|2015-06-18|2018-01-09|Schlumberger Technology Corporation|Flow balancing for a well|

---

US10614378B2|2016-09-26|2020-04-07|International Business Machines Corporation|Cross-well allocation optimization in steam assisted gravity drainage wells|

---

US10267130B2|2016-09-26|2019-04-23|International Business Machines Corporation|Controlling operation of a steam-assisted gravity drainage oil well system by adjusting controls to reduce model uncertainty|

---

US10352142B2|2016-09-26|2019-07-16|International Business Machines Corporation|Controlling operation of a stem-assisted gravity drainage oil well system by adjusting multiple time step controls|

---

US10378324B2|2016-09-26|2019-08-13|International Business Machines Corporation|Controlling operation of a steam-assisted gravity drainage oil well system by adjusting controls based on forecast emulsion production|

---

US10570717B2|2016-09-26|2020-02-25|International Business Machines Corporation|Controlling operation of a steam-assisted gravity drainage oil well system utilizing continuous and discrete control parameters|

---

US10577907B2|2016-09-26|2020-03-03|International Business Machines Corporation|Multi-level modeling of steam assisted gravity drainage wells|

---

GB2562465A|2017-05-04|2018-11-21|Solution Seeker As|Recording data from flow networks|

---

CA3084875A1|2017-12-08|2019-06-13|Solution Seeker As|Modelling of oil and gas networks|

---

US10982516B2|2019-09-03|2021-04-20|Saudi Arabian Oil Company|Systems and methods for operating downhole inflow control valves to provide sufficient pump intake pressure|

---

US20210062681A1|2019-09-04|2021-03-04|Saudi Arabian Oil Company|Systems and Methods for Proactive Operation of Process Facilities Based on Historical Operations Data|

---

GB202005239D0|2020-04-08|2020-05-20|Solution Seeker As|A method of modelling a production well|

法律状态:
2018-11-13| B06F| Objections, documents and/or translations needed after an examination request according [chapter 6.6 patent gazette]|

---

2020-04-14| B06U| Preliminary requirement: requests with searches performed by other patent offices: procedure suspended [chapter 6.21 patent gazette]|

---

2020-11-03| B06A| Patent application procedure suspended [chapter 6.1 patent gazette]|

---

2021-02-23| B09A| Decision: intention to grant [chapter 9.1 patent gazette]|

---

2021-05-04| B16A| Patent or certificate of addition of invention granted [chapter 16.1 patent gazette]|Free format text: PRAZO DE VALIDADE: 20 (VINTE) ANOS CONTADOS A PARTIR DE 17/04/2014, OBSERVADAS AS CONDICOES LEGAIS. |

---

2022-02-08| B21F| Lapse acc. art. 78, item iv - on non-payment of the annual fees in time|Free format text: REFERENTE A 8A ANUIDADE. |

优先权:

---

申请号 | 申请日 | 专利标题

---

GBGB1306967.9A|GB201306967D0|2013-04-17|2013-04-17|Control of flow networks|

---

GB1306967.9|2013-04-17|

---

PCT/EP2014/057881|WO2014170425A2|2013-04-17|2014-04-17|Control of flow networks|

---