WO2013135288A1

WO2013135288A1 - System and method for controlling the processing of oil sands

Info

Publication number: WO2013135288A1
Application number: PCT/EP2012/054490
Authority: WO
Inventors: Jan Richard Sagli; Elvira Marie Bergheim ASKE; Kalpesh Patel
Original assignee: Statoil Canada Limited
Priority date: 2012-03-14
Filing date: 2012-03-14
Publication date: 2013-09-19
Also published as: CA2812394A1

Abstract

Model predictive control (MPC) is implemented on a steam-assisted gravity drainage (SAGD) injector-producer well pair. MPC uses measurements and models to predict future well and reservoir behaviour and optimize the changes on the manipulated variables to keep the controlled variables at their targets and within given constraints.

Description

System and Method for Controlling the Processing of Oil Sands Field of the invention The invention relates to a system and method for controlling the system to extract oil from oil sands.

Background to the invention Oil sands are a mix of sand, water, clay and bitumen (unrefined oil). Under normal conditions, the bitumen is too heavy and/or thick to float or be pumped, for example at 1 1 °C, bitumen is solid. Accordingly, the bitumen needs to be heated and/or thinned out so that this valuable resource can be collected. Figure 1 shows a reservoir 10 of oil sand beneath other layers of rock, soil etc. A known method of extracting the bitumen from the reservoir is to drill two well lines 12, 14 into the reservoir forming a wellpair. The first well line 12 is for inputting steam into the reservoir to heat the bitumen. The arrows along the well line indicate the direction of steam flow. The second well line 14 is for extracting the bitumen from the reservoir. As shown, the two well lines are not connected to each other and are spaced apart, perhaps generally parallel to one another into the reservoir.

Initially, the steam injected into the reservoir creates a small steam chamber. The steam expands in both the longitudinal and axial direction as indicated by arrows A and flows to the interface between the steam chamber and the reservoir. The steam expansion gradually expands the steam chamber 18. The steam heats the bitumen which flows under gravity to the lower portion of the steam chamber as indicated by arrows B. The bitumen (and any other condensate) is drawn off through the second well line using a pump. The arrows along the well line indicate the direction of output fluid flow.

Uniform steam chamber development is critical to ensure a productive well. The current challenges facing efficient and effective extraction include the variation in response from the reservoir depending on steam chamber size, reservoir heterogenity, presence of lean zones. In general, the reservoirs are not well understood. It is known to use instrumentation to measure the state of the wellpairs. An operator observes these measurements on a visualization system and takes action if the measurements are outside normal behavior or outside pre-defined limits. The operator may change the pump rate on the second well line and/or injection rates into the first well line. These changes are made based on the operator's training and/or based on directions from production engineers responsible for the wellpairs.

Typically, workflow depends very much on the experience of the operator and the production engineer working on that specific shift. Operators will normally have to take care of many wellpairs. This means that wells have to be operated with a reasonable margin of error or "comfort level" which often leads to a production that is less optimal.

The present applicant has recognised the need for a method and system to improve extraction of bitumen from oil sands.

Statements of invention

According to a first aspect of the invention, there is provided a control system for a well pair extraction system in an oil sand reservoir, the well pair extraction system comprising an injection well for injecting steam into the reservoir and a production well for outputting fluid from the reservoir, the control system comprising:

a central processor configured to:

receive measurements from at least one sensor within the well pair extraction system, the at least one sensor measuring at least one control variable of the well pair extraction system;

predict a future value of the at least one control variables by inputting the measurements into a model of the well pair extraction system, the model being dependent on the at least one control variables and at least one manipulated variable which is adjustable to control the at least one control variable;

determine, based on the predicted future value of the at least one control variable, whether the at least one manipulated variable for the well pair extraction system is to be changed, and

output any change to the at least one manipulated variable According to a second aspect of the invention, there is provided a method of controlling a well pair extraction system in an oil sand reservoir, the well pair extraction system comprising an injection well for injecting steam into the reservoir and a production well for outputting fluid from the reservoir, the method comprising:

receiving measurements from at least one sensor within the well pair extraction system, the at least one sensor measuring at least one control variable of the well pair extraction system;

predicting a future value of the at least one control variable by inputting the measurements into a model of the well pair extraction system, the model being dependent on the at least one control variable and at least one manipulated variable which is adjustable to control the control variables;

determining, based on the predicted future value of the at least one control variable, whether the at least one manipulated variable for the well pair extraction system is to be changed, and

outputting any change to the at least one manipulated variable.

The following apply to both aspects of the invention. The invention allows production from the well pair extraction system to be controlled by monitoring various parameters of the system and manipulating or changing other parameters of the system (which are termed manipulated variables) to control the monitored parameters. The control variables may thus be defined as the parameters (or a subset of the parameters) of the system which are monitored and/or measured. There are preferably a plurality of control variables which are measured and thus there may be a plurality of sensors. The manipulated variables may thus be defined as the parameters (or a subset of the parameters) of the system which are manipulated and/or changed to control the control variables. Use of the above system and method allows the pair of wells (injection and production) to be operated in a consistent way. For example, measurements and models are used to predict future well and reservoir behaviour and optimize the changes on the manipulated variables to keep the controlled variables at their targets and within given constraints. Thus, the system and/or method can be used, for example, to reduce start up time after a circulation phase and/or shut down. The operation may be independent of the operator and may thus be a closed-loop control system. In other words, said processor may be configured to iteratively repeat said receiving, predicting, determining and outputting steps. However, the operators can intervene in the more problematic cases so that the system can also operate in open-loop or manual mode under the control of the operators.

The control variables may include parameters of said injection well which are to be controlled by said control system, for example pressure within said injection well and/or rate of injection of steam into said injection well. The injection well may comprise of a short string which injects steam into the heel part of the well (i.e. towards the beginning of the horizontal section of the well) and a long injection string which injects steam into the toe part of the well (i.e. at the end of the horizontal section of the well). The manipulated variables may preferably include pressure and/or rate of injection of steam within one or both strings.

The control variables may include parameters of said production well which are to be controlled by said control system. There may be many sensors on the production well. For example, there may be pressure and temperature sensors (PI, Tl) at both the toe of the well and the heel of the well. There may also be temperature and pressure sensors (PI, Tl) at various points along the length of the well. The pressure and temperature sensors may be combined sensors. The production well preferably comprises a pump or other similar lifting mechanism for drawing fluid out from said production well. At the pump there may be a variety of different sensors including for example, winding temperature sensor (Tl), current (II), voltage (El), variable frequency drive VFD speed (SI), pressure (PI). There may also be temperature and flow rate sensors (Tl and Fl) at the output from the well or downstream from the output, e.g. at the surface facilities. Accordingly, the control variables may be any one or more of (but not limited to): a) Pressure at one or more points along the production well

b) Temperature at one or more points along the production well

c) Temperature differential across one or more pairs of points along the production well

d) Flow from the production well e) Parameters of said pump for outputting fluid from said production well, wherein said pump parameters may include winding temperature sensor, current, voltage, variable frequency drive (VFD) speed and/or pressure (PI). The control variables may be individually measured and controlled by the system. Alternatively, the measurements of the control variables may be combined to allow control a key overarching control variable, for example said at least one subcool within said production well, where said subcool is defined as: Subcool = T_sat - T,

where

T_Sat is the saturation temperature at production pressure and

T, is the temperature at or near location /^' in the producer (production well). One of the main objectives is to control the subcool along the wellbore. The higher the liquid level above the producer the lower the temperature and higher is the sub-cool. Where there is no subcool, live steam may be flow directly into the production well. Where the subcool is too high, the well is functioning inefficiently. A balance needs to be obtained between these two extremes. The optimum subcool is where the reservoir has a steam zone which is large relative to an oil extraction zone but which has no overlap at an outlet to the production well.

At least one manipulated variable may comprise a parameter of said production well which is adjustable to control said control variables, for example said extraction system may comprise a pump for outputting fluid from said production well and said at least one manipulated variable may include speed of said pump. Alternatively, or additionally, at least one manipulated variable may comprise a parameter of said injection well which is adjustable to control said control variables. Each of said production well and said injection well may comprise at least one valve to control flow into or from said well, for example said injection well comprises separate valves to control flow into said short and long strings and said production well comprises a valve to control output flow. Said at least one manipulated variable may comprise a position (i.e. open/closed/partially open) of one or more of these valves. Said processor may be further configured to input a limit associated with each control variable and/or each manipulated variable. Limits may be associated with some or all of the manipulated variables. Said limit(s) may be input by an operator, e.g. as part of an initialisation phase. In the absence of definitions from an operator, preprogrammed default limits will be implemented. The limits may be associated with safety or may represent optimised limits.

The limits may also include maximum rates of change for each of the manipulated variables. These maximum rates are always respected. In other words, said limit may actually be said at least one manipulated variable. For example, said at least one manipulated variable may be said limit for rate of said steam injection and may be the set point for the injection rate into one or both of said short string and long string.

The processor may be configured to determine whether said at least one manipulated variable is to be changed by comparing said predicted values of said control variables with said at least one limit. Alternatively, or additionally, the processor may determine whether said at least one manipulated variable is to be changed by determining a state of each manipulated variable required to optimise the system, comparing said determined state with a current state of said manipulated variable and determining that a change is required if the determined state and current state do not match. The state required to optimise the system may be determined by adjusting a manipulated variable, repeating said predicting step and comparing the predicted values from said repeating predicting step with the previous predicted values to see whether or not there has been any improvement.

Said control system may further comprise a subsidiary controller for controlling said at least one manipulated variable and wherein said central processor is configured to output any change to said subsidiary controller to implement said change. Said subsidiary controller may be a flow controller (e.g. a PID) controlling a valve on an input or output to said extraction system or a product controller (e.g. a PID) controlling an input or output to said extraction system. Thus, in other words, said subsidiary controllers implement the changes recommended by the system.

The model may be further dependent on at least one disturbance variable which disables said processor when activated and wherein said processor is configured to: determine whether said at least one disturbance variable is activated and cease outputting any change to said at least one manipulated variable until said at least one disturbance variable is deactivated or a predetermined time limit has passed.

The disturbance variables may include one or more of surface production pressure, casing gas pressure and/or pump speed process value. The disturbance variables may also include a purge indicator, a critical subcool, active subcool and/or stable steam.

The model may comprise a plurality of finite impulse response models and/or at least one logic relationship.

According to another aspect of the invention, there is provided a well pair extraction system for an oil sand reservoir comprising:

an injection well for injecting steam into the reservoir;

a production well for outputting fluid from the reservoir, and

a control system as defined above. Each aspect of the invention preferably improves the safety of the well operation. It may allow wells to be located more closely together, e.g. at a distance of approximately 3m apart rather than the standard minimum 5m gap, or may allow thinner reservoirs to be exploited, e.g. reservoirs having a height of 10m or less rather than the standard minimum of 15m. Thus, the system and/or method may be used to reduce the distance between the well pairs.

The invention may also improve the efficiency of the well, accelerate start-up after e.g. the circulation phase or shut-down and generally reduce the energy consumption. The control may optimise production to ensure that maximum extraction is achieved for minimum energy input.

Said system may be expanded from the above system to cover multiple well pairs. Thus according to another aspect of the invention, there is provided an extraction system for an oil sand reservoir comprising a plurality of well pair extraction systems as defined above; a source of steam connected to each of said plurality of well pair extraction systems; and an output connected to each of said plurality of well pair extraction systems.

Although there is a common steam input and a common output, each wellpair has its own control system. Each control system is individually connected to the common output to manipulate fluid flow from the associated production well and to the common steam input to manipulate steam flow into the associated injection well. Accordingly, the separate wellpairs may be individually optimised whilst drawing on the same resources. The optimisation may be done individually at each control system or by a central control. The extraction system may thus further comprise a main controller connected to the control system for each well pair extraction system, the main controller being configured to control, e.g. optimise, production and steam utilisation across the extraction system. The main controller may thus act as co-ordination system to ensure that steam is given to the well where additional steam gives the highest increase in oil production or where needed due to given constraints. Such constraints may include a minimum spacing between individual well pairs. With better co-ordination of the overall system, the spacing between individual well pairs may be reduced. Thus according to another aspect of the invention, there is provided use of said system to reduce the distance between the well pairs.

The invention further provides processor control code to implement the above- described systems and methods, for example on a general purpose computer system or on a digital signal processor (DSP). The code is provided on a physical data carrier such as a disk, CD- or DVD-ROM, programmed memory such as non-volatile memory (eg Flash) or read-only memory (Firmware). Code (and/or data) to implement embodiments of the invention may comprise source, object or executable code in a conventional programming language (interpreted or compiled) such as C, or assembly code. As the skilled person will appreciate such code and/or data may be distributed between a plurality of coupled components in communication with one another.

Brief description of drawings The invention is diagrammatically illustrated, by way of example, in the accompanying drawings, in which:

Fig. 1 is a schematic illustration of known extraction of bitumen from an oil sands reservoir;

Fig. 2 is a block diagram of the components of one arrangement of the system comprising an injection well and a production well;

Figs. 3a and 3b are schematic diagrams of an injection well and an production well which may be used of the system of Fig 2;

Figs. 4a to 4c illustrate a steam chamber having zero subcool, preferred subcool and high subcool, respectively;

Fig. 5a is a flowchart of the steps for controlling a MPC controller within the system of Fig. 2;

Fig. 5b is a graph showing example upper and lower limits defined in the method of Fig. 5a;

Fig. 5c is a flowchart of the steps for controlling a PID controller within the system of Fig. 2;

Fig. 5d is a graph showing an example set point for the PID controller;

Figs. 6a and 6b are graphs showing steam flow over time within the short injection and long injection strings respectively for a model assuming a homogeneous reservoir;

Fig 6c is a graph showing output fluid flow over time for the model of Fig. 6a;

Figs. 6d and 6e are graphs showing variation in pressure over time within the short injection and long injection strings respectively for the model of Fig. 6a;

Fig. 6f is a graph showing the level of one subcool over time for the model of Fig. 6a; Figs. 6g and 6h are graphs showing the variation of total steam injection flow and the percentage steam flow in the short string with time respectively for the model of Fig. 6a;

Figs. 7a and 7b are graphs showing steam flow over time within the short injection and long injection strings respectively for a model assuming a heterogeneous reservoir;

Fig 7c is a graph showing output fluid flow over time for the model of Fig. 7a;

Figs. 7d and 7e are graphs showing variation in pressure over time within the short injection and long injection strings respectively for the model of Fig. 7a;

Fig. 7f is a graph showing the level of one subcool over time for the model of Fig. 7a;

Figs. 7g and 7h are graphs showing the variation of total steam injection flow and the percentage steam flow in the short string with time respectively for the model of Fig. 7a; Fig. 8 is a schematic illustration of a system comprising multiple wellpairs, and Fig. 9 is a schematic illustration of one system architecture.

Detailed description of drawings Fig. 2 shows a system for extracting oil from oil sands. The system comprises an extraction system comprising a steam generator 30 for injecting steam into an injection well 41 and a production well 40 for extracting output fluid, including oil. As shown in Fig. 2, the injection well 41 and production well 40 are joined by dotted lines to illustrate that whilst the injection and production wells may be physically close to each other there is no direct connection between these two wells. The wells are connected by the creation of a steam chamber as described with reference to Fig. 1 above. The extraction system may thus be termed a closely spaced injector-producer well pair.

The creation of the steam chamber and the process of extraction is controlled by a control system which comprises a central controller or processor 34 whose operation is described in more detail below. The processor 34 is connected to a steam controller 32 which controls the injection of steam into the injection well and a production controller 42 which controls the output from the production well. The processor and controllers are shown as separate components but it will be appreciated that they may incorporated into a single component. The steam and production controllers may be connected to valves (not shown) to open and close the valves to control flow within the system. The controllers may also be connected to pump(s) within the system to control flow within the system. The production controller 42 controls the flow of output from the production well to downstream polishing separators. The separators may include an initial oil/water separator 44 to separate the output flow into an oil flow which is predominantly oil and a water flow which is predominantly oil. The water flow is fed to a water treatment apparatus 46 for further polishing and the final water flow may be fed back into the steam generator 30. The oil flow is fed to an oil treatment apparatus 48 for further treatment.

The processor 34 receives data from a variety of sensors 36 which are positioned throughout the system, e.g. in the production well and the injection well. The processor 34 is also connected to an operator terminal 35 for an operator to input data into the system and/or for the processor 34 to output data for an operator to view.

Use of the above system allows the pair of wells (injection and production) to be operated in a consistent way. The operation may be independent of the operator and may thus be a closed-loop control system. Operators can let the automated system take care of most of the control. However, the operators can intervene in the more problematic cases so that the system can also operate in open-loop mode under the control of the operators. One of the aims as explained below is that the steam assisted gravity draining subcool is reduced, leading to higher production.

Fig. 3a shows the components of one suitable configuration for the injection well 41. The injection well 41 comprises an injector 50 coupled to a short injection string 52 and a long injection string 54. The short injection string 52 injects steam into the heel part of the well and the long injection string 54 injects steam into the toe part of the well. For ease of illustration, the strings are not shown as continuous lines in Fig 3a but clearly the two parts of each string are connected. As an alternative, a single injection string with a plurality of valves along the length of the string could be used whereby the string can inject into different areas of the production by appropriate opening and closing of one or more valves. Also the production well may be equipped with a similar type of valves along the wellbore, making it possible to manipulate the steam chamber also from the producer side.

Fuel gas 56 may also be input to the injector to purge the well to prevent sand etc. clogging the pressure instrument. Each input to the injector has an associated valve 58 controlling flows into the injector. Each valve is controlled by a flow controller FC which may be incorporated in or may be controlled by the steam controller 32 of Fig. 2. Alternatively, another controller such as a pressure controller could be used in the place of the FC. There is a pressure sensor PI measuring the pressure on the fuel gas input.

Fig. 3b shows the components of the production well 40. The production well 40 comprises a producer 60 which is coupled to a pump or other lifting mechanism to draw output fluid flow from the reservoir. The fluid is output to a production header 62 and gas is output to a casing gas 64. Each output from the producer 60 has an associated valve 68 controlling flows from the producer. On the output to the production header 62 there is an additional valve 70 to allow some of the fluid output to be drawn off to be tested at a test header 66. Each valve is controlled by a product controller PC which may be incorporated in or be controlled by the production controller 42 of Fig. 2. The pump is shown as an electric submersible pump (ESP) 72 is fitted within the production well to pump fluid to the producer 60. Alternative pumps may be used as appropriate and thus the term pump and ESP may be interchanged in the following description. There are many sensors in the production well. There are pressure and temperature sensors (PI, Tl) at both the toe of the well and the heel of the well. There are also pressure and temperature sensors (ΡΙ,ΤΙ) at various points along the length of the well. These sensors may be combined sensors measuring both pressure and temperature. Temperature differential (TDI) may also be calculated at various points along the well. At the ESP 72 there are a variety of different sensors including, for example, winding temperature sensor (Tl), current (II), voltage (El), variable frequency drive VFD speed (SI), pressure (PI). There are also temperature and flow rate sensors (Tl and Fl) on each output. One of the main control objectives is to control the subcool along the wellbore. Figs 4a to 4c illustrate this objective. In each of Figs 4a to 4c, there is a steam zone 80 having a steam input 84 and an outer oil extraction zone 82 having an oil output 86. As explained in relation to Fig 1 , the oil is extracted by steam forcing through the steam input 84 to the interface of the steam chamber. Heated oil drains under gravity along the interface to the base of the chamber and is extracted through oil output 86.

The subcool is related to the fluid level and is defined as:

Subcool = T_sat - T_x

where

T_Sat is the saturation temperature at production pressure and

T, is the temperature at location /^' in the producer (production well)

The higher the liquid level above the producer the lower the temperature and higher is the sub-cool. Typically, the lowest subcool is used as the constraining or limiting subcool. In Fig. 4a, there is no subcool, in other words the steam zone 80 extends below the steam input and covers the oil output 86 so that there is no separation of the steam zone and oil extraction zone in the region of the oil output. Accordingly, live steam will be output which is dangerous. By contrast, in Fig. 4c, the steam zone is small relative to the oil extraction zone because of a relatively high subcool. Accordingly, the well is likely to function inefficiently with loss of potential head and steam distribution. The optimum subcool is shown in Fig. 4b where the steam zone is large relative to the oil extraction zone but there is no overlap at the outlet.

As explained in more detail below, the objective of creating an ideal subcool can be achieved by monitoring various parameters (termed control variables) and manipulating other parameters (termed manipulated variables) within the system to control the monitored parameters. For example, the ESP speed, the injector bottomhole pressure and injection split, the short- and long steam injection rates are examples of variables which may be adjusted to achieve the overall aims. As an example, the ESP speed can be raised or decreased by a fixed amount, e.g. 5%.

Fig. 5a illustrates the various steps of the method which are implemented within the processor of Fig 2. The first initialisation step S100 is to define any limits (or set points) for the system. As shown in Fig. 5b, there may be an upper and a lower limit 90,92. Each limit may be constant with time or may change with time, e.g. have step changes in value. A step change is unusual but could be included by an operator, perhaps when overriding in open-loop mode. A separate limit (or pair of limits) is preferably defined for each control variable and/or for each manipulated variable. In the absence of definitions from an operator, preprogrammed default limits will be implemented. Any changes to the limits are communicated to the individual controllers, e.g. the steam controller or production controller of Fig. 2 and/or the flow controllers of Fig 3a and the product controllers of Fig 3b.

The limits are set to avoid the process limitations such as:

• staying above bottom water zone pressure,

• avoid live steam production,

• staying within pump operation envelope etc. Thus, the limits may be associated with safety, e.g. max pressure or may represent optimised limits, e.g. desired output flow. The limits may also include maximum rates of change for each of the manipulated variables. These maximum rates are always respected.

After initialisation, the method automatically loops through steps S102 to S1 10 repeatedly, perhaps every minute or at other regular intervals. The prediction may be for the next 12 to 18 hours. At step S102, sensor data is received from the system. As indicated above, there are a plurality of sensors throughout the system, including instrumentation measuring the state of the wellpairs, pumps etc. The measurements are primarily for the controlled variables (CVs), i.e. the system variables which the system is attempting to control. Other variables can also be measured.

The measurements are fed into the processor which operates a so-called Model Based Predictive Control System (MPC) so the processor may be termed an MPC controller. At step S104, the future states of the pair of wells are predicted based on a pre-defined model. At step S106, the processor calculates what will happen if actions to various system variables known as the manipulated variables (MVs) are taken. In particular, the processor determines the state of each manipulated variable required to optimise the system (e.g. to provide a constant subcool).

The model used in the MPC may comprise a plurality of different models for the different sections/elements of the system. The models may be of finite-impulse- response (FIR). The models are identified through iterative testing to minimise the difference between the model and real behaviour. Some of the models, e.g. the model from injection steam flow MV to subcool CV are difficult to identify due to a lack of repetitive behavior. However, these models are still important. Accordingly, these models are implemented as logic relationships, for example inject most steam (within injection split constraints) where the subcool are highest (heel, middle, toe) along the wellbore. However, other alternative implementations of the models may be used.

The processor then determines whether any of the manipulated variables require altering (step S108) as a result of the predictions and outputs any changes to be implemented (step S110). The changes are then made to the physical system itself. The changes may be implemented by subsidiary controllers, e.g. the steam controller or the production controller, the flow controller or the product controller. The subsidiary controllers may themselves be PID controllers. Fig. 5c shows the operation of such a controller. As with Fig. 5a, the first step S200 is an initialisation phase which includes setting the set point of the process variable in question. The controller receives sensor data for the process variable at step S202, compares this measurement to the set point at step S204 and outputs any resulting change to the controlled variable at step S206. The set point of the process variable may be altered, for example by the MPC controller as described in more detail below.

A summary of the differences between a PID controller and the overall system controller are set out below:

A PID controller is a Proportional-lntegral-Derivative controller which is a generic control loop feedback mechanism (controller) widely used in industrial control systems.

As examples, the manipulated variables (MVs) include some or all of: 1. Pump (ESP) speed.

2. Short steam injection rate set point.

3. Long steam injection rate set point.

As set out above, limits may be associated with some or all of the MVs. In practice, the most common limit is selecting which of the subcools determines the ESP speed; this subcool is termed the critical subcool and may be the lowest value of subcool. The limits on the steam injection rates are decided by the bottomhole injection pressure and the injection split. However, after a shut-down of the wells, the total steam injection rate is then a common limit while the injection pressure builds up.

The ESP speed is selected as an MV. As an alternative (or additionally), the production flow rate set point from the production well could be an MV. However, the ESP speed is a variable that the operators are used to manipulating using the manual operating system. Furthermore, the production flow rate is a more noisy measurement so this may be controlled with a PID controller. The ESP speed is operated with a dead-band of 0.5 Hz to avoid small incremental moves in the pump speed.

As shown in Fig 3a, flow controllers control the valves on the short and long steam injection strings and these controllers may be PID controllers controlling the rates of injection. If this is the case, the MPC controller is connected (directly or via the steam controller) to the flow controllers to vary the set point to the flow controller as necessary. Both the pressure conditions of the injection well and the valve opening on the long and short strings determine the rate of steam injected. Accordingly, it is advantageous to include control of the steam injection rate within the MPC controller in closed-loop because the MPC controller can input and output several variables whereas a PID controller only has one degree of freedom. Furthermore, some operators, for example those in the PETECH (Pulse Enhancement technology) group are used to deciding the steam flow rates rather than whether to open or close the steam valves.

As examples, the controlled variables (CVs) include some or all of (but not limited to):

1. Subcools along the wellbore and at the pump, 9 in total.

2. Injector bottomhole pressure.

3. Pump motor winding temperature

4. Pump motor current

5. Pump inflow temperatures, 2 in total.

6. Pump inlet pressure

7. Production flow from well (maps the upthrust and downthrust constraints for the Pump)

8. Total steam rate 9. Injection split between short and long steam injection

10. Steam pressures in short and long string, 2 in total

1 1. Steam injection valves, 2 in total

12. Pressure measurements along the producer well

As explained above, subcools are important when optimising performance and it is preferred to consider all the subcools along the wellbore. Additionally, injector bottomhole pressure, pump constraints and steam constraints are considered. The ESP constraints protect the pump from operating outside its defined operating window and temperature ratings. The steam constraints ensure that the steam supply is sufficient and the pressure in the injection string will not go too high.

The production flow is included as a constraint because the pump curve is mapped directly as capacity as a function of speed. Thus, the upthrust and downthrust constraints for the pump speed can be expressed as high and low limit on the production flow, where the limits on the production flow changes depend on the pump speed itself.

In addition, there are some disturbance variables (DVs) that affect the process which may also be included within the model used by the MPC controller. Some examples of DVs are (but not limited to):

1. Surface production pressure

2. Casing gas pressure

3. ESP speed process value

The surface production pressure affects the ESP deliverance pressure and thus affects the required head for the ESP pump. The ESP speed process value may be added as a DV because we want to add the model from the ESP speed towards the injector bottomhole or producer pressure without using the ESP speed as an MV for this controlled variable. In addition there are logic DVs which override the MPC controller in whole or in part. These logic DVs are added to avoid unwanted moves in special cases. As examples these may include some or all of: 4. Purge indicator. MPC controller disabled if purging of the downhole bubble tubes is detected.

5. Critical subcool. MPC controller disabled if a critical subcool turns bad.

6. Subcool active. Ensures that at least one subcool CV is active before the ESP speed can be used as an MV in the MPC controller. 7. Stable steam. Ensures that the steam supply is stable before the steam injection rates can be used as an MV in the MPC controller.

Purging of the bubble tubes is done to prevent sand etc. clogging the pressure instrument. The purging occurs regularly, e.g. every three hours, and fuel gas is purged down to the instrument at a significantly higher pressure. Hence, the pressure readings go high for a short while, less than 2 minutes. However, the temperature readings are affected for longer, and for some subcool values the disturbances last for about 30 minutes. Typically, the disturbances are only short-term and the process returns to approximately the same levels as before the purge. Accordingly, the MPC controller is disabled, i.e. turned off or at least prevented from outputting any changes to the system for approximately 20-30 minutes following a purge, depending on which subcool CVs are actively being used in the modelling.

As set out above, subcools are considered along the length of the production well. The critical subcool is the subcool having the lowest value. This is because this is the subcool that determines the ESP speed. If this critcial subcool fails or goes bad, the MPC controller is disabled. If this logic DV is not used, the controller may continue to control the second lowest subcool, and may increase the ESP speed. Before reinstating the MPC controller, the limits for the CVs may need to be revised and an investigation carried out as to why this measurement did turn bad.

The subcool active and/or stable steam DVs avoid unwanted operation of the SAGD wellpair. Subcool active ensures that at least one subcool CV is active in the controller before the ESP speed can be controlled by the MPC. This is because ESP speed is the manipulated variable having the greatest effect on the subcool controlled variable. Stable steam ensures that the steam supply is working well and the MPC has the ability to decide the steam rates.

Figs 6a to 7h illustrate the results of a simulation using MPC control of the above system. The process was simulated in the thermal reservoir modelling tool STARS. In Figs 6a to 6h, a homogeneous model assuming that the reservoir had a uniform permeability was used. In Figs 7a to 7h, a heterogeneous model with different permeability in the reservoir was used, leading to different steam distribution compared to the homogenous reservoir. In summary, the simulation results showed that the short and long string bottomhole pressures were controlled with the steam injections. The producer flow controlled the lowest subcool in the wellbore. Figs. 6a to 6c and Figs. 7a to 7c are the variation in time for each of the manipulated variables of the simulated system, namely:

• Short injection string steam flow;

• Long injection string steam flow, and

· Produced fluid flow

These are the variables which are automatically adjusted by the system under the control of the MPC controller to ensure the desired or optimal operation of the system. For each variable, the upper and lower limits are set by the operator, e.g. during an initialisation phase as described in Fig 5.

Figs. 6d to 6h and Figs. 7d to 7h are the variation in time for each of the controlled variables of the simulated system, namely: · Bottomhole pressure for the short string;

• Bottomhole pressure for the long string;

• Subcool level (400mts from the heel for the homogeneous model and at the toe for the hetergeneous model)

• Total steam injection flow, and

· Percentage steam flow for the short string These are the variables which are measured by the system and which the MPC controller is automatically controlling by manipulation of the manipulated variables above. For each variable, the upper and lower limits or set points are set by the operator or production engineer. Figs 6a to 7h illustrate the impact on all variables of changes to the limits/set points by the operator. The results are shown over a period of several days with the numbers 1 to 5 illustrating the points at which limits/set points are changed.

For Figs 6a to 6h, these changes were:

1) The upper limit for the short string bottomhole pressure was increased by 50Kpa and held at the increased limit for three days

2) The upper limit for the long string bottomhole pressure was increased by 50Kpa and held at the increased limit for three days

3) The lower limit for the subcool was increased by 3° and held at the increased limit for three days

4) The upper limit for the short string bottomhole pressure was decreased by 50Kpa and held at the decreased limit for three days

5) The set point for the total steam injection flow was decreased by 25m³/hr and held at the decreased limit for three days.

Each of these changes had an immediate impact on the corresponding controlled variable, e.g. Fig 6d shows that the pressure measured in the short string spiked when the upper limit was increased. Some of the changes also effected related controlled variables, e.g. Figs 6d and 6e show that the decrease in the total steam flow at (5) also results in a drop off in pressure for both strings. Figs 6d to 6h also show that most of the controlled variables were stabilised by the system shortly after the change albeit that the stabilisation may be a different level.

The stabilisation of the controlled variables is achieved by changing the manipulated variables. For example, following the increase to the upper limit for the short string bottomhole pressure, the system determined that the short injection string steam flow needed to be increased but that overall produced fluid flow needed to be decreased. In this way, the short string bottomhole pressure could be stabilised at the new higher level. The long string bottomhole pressure and subcool were unaffected.

As shown in Figs 6a and 6b, the manipulated variables were not always changed in response to the changes to the limits to the controlled variables. In Fig 6c, each change of the limit to the controlled variable resulted in a change to the manipulated variable, namely the produced fluid flow.

For Figs 7a to 7h, these changes were:

1) The lower limit for the subcool was decreased by 3° and held at the decreased limit for three days

2) The upper limit for the long string bottomhole pressure was increased by 40Kpa and held at the increased limit for three days

3) The upper limit for the short string bottomhole pressure was increased by 40Kpa and held at the increased limit for three days

4) The set point for the total steam injection flow was decreased by 25m³/hr and held at the decreased limit for three days.

5) The percentage of the total steam flow from the short string was increased by 5%.

As with Figs 6a to 6h, Figs 7a to 7h shows that each of these changes had an immediate impact on the corresponding controlled variable and also effected related controlled variable. Stabilisation of the controlled variables was also achieved by changing the manipulated variables. As with Figs 6a to 6h, Figs 7a to 7h show that the manipulated variables were not always changed in response to the changes to the limits to the controlled variables. As before, in Fig 7c, each change of the limit to the controlled variable resulted in a change to the manipulated variable, namely the produced fluid flow.

Fig 8 shows a schematic implementation expanding the above system to cover multiple wellpairs. A plurality of wellpairs are grouped into a pad (L1 , ... , L4). There a plurality of pads, each served by a common steam header 102 and a common production header 100. As with Fig. 5, the production header is connected to an oil/water separator 104 which separates the fluid into a predominantly oil flow to the oil treatment apparatus 106 and a predominantly water flow to the water treatment apparatus 108. The steam header 100 is connected to a common steam generator 1 10 which may receive polished water from the water treatment apparatus.

Although there is a common steam input and a common output, each wellpair has its own MPC which is individually connected to the production header 100 to control fluid flow from the associated production well and to the steam header 102 to control steam flow into the associated injection well. Accordingly, the separate wellpairs may be individually optimised whilst drawing on the same resources. As an optional extra, a site wide controller 1 12 could be connected to each MPC controller to optimise production and steam utilisation across the whole site.

Fig 9 illustrates the system architecture for the MPC system. There are a plurality of servers for different parts of the system, namely a CPF server, a wellpad server, a historien server, active directory domain controller server, a terminal server and a gateway server. There are also a plurality of controllers, e.g. PID controllers, for different parts of the system, namely CPF controllers, CCS controllers and wellpad controllers. Each of the servers and controllers, including the gateway server and the wellpad controllers are connected to the control system supervisory network to enable communication between the various components. There are also a plurality of workstations connected to the network, for example operator workstations and engineer workstations.

The MPC controller is an open connectivity (OPC) client that is connected to the OPC server running on the Gateway server. In this way, the MPC system is installed in separate computer on top of the digital process control system (DCS) and implemented by operators via the operator/engineer workstations. It will be appreciated that the different servers could be geographically and physically separated. Alternatively, the functionality could be combined into one or more combined servers.

In summary, model predictive control (MPC) has been implemented on a steam- assisted gravity drainage (SAGD) injector-producer well pair. Although MPC have been widely used in industry (e.g. (Qin & Badgwell, 2003)), there is no suggestion of how the technology can be implemented in a SAGD process. As described above, in the invention, MPC is implemented for stable closed-loop control of subcool in the producer and the injector bottomhole pressure. MPC uses measurements and models to predict future well and reservoir behaviour and optimize the changes on the manipulated variables (e.g steam injection rates and electrical submersible pump (ESP) speed) to keep the controlled variables (subcools, pressures) at their targets and within given constraints. Constraints on the system and ESP are also respected. With MPC the control targets can be specified and the MPC will move the SAGD process to the desired operation point.

No doubt many other effective alternatives will occur to the skilled person. It will be understood that the invention is not limited to the described embodiments and encompasses modifications apparent to those skilled in the art lying within the spirit and scope of the claims appended hereto.

Claims

CLAIMS:

1. A control system for a well pair extraction system in an oil sand reservoir, the well pair extraction system comprising an injection well for injecting steam into the reservoir and a production well for outputting fluid from the reservoir, the control system comprising:

a central processor configured to:

output any change to the at least one manipulated variable.

2. A control system according to claim 1 , wherein the at least one control variable includes parameters of the injection well which are to be controlled by the control system.

3. A control system according to claim 2, wherein the at least one control variable comprises pressure within the injection well and/or rate of injection of steam into the injection well.

4. A control system according to any one of the preceding claims, wherein the at least one control variable includes parameters of the production well which are to be controlled by the control system.

5. A control system according to claim 4, wherein the at least one control variable comprise at least one subcool within the production well, where the subcool is defined as: Subcoo^ T^ - T,

where

T_Sat is the saturation temperature at production pressure and

Tj is the temperature at location /^' in the producer.

6. A control system according to claim 4 or claim 5, wherein the at least one control variable includes parameters of a pump for outputting fluid from the production well.

7. A control system according to any one of the preceding claims, wherein the at least one manipulated variable comprises a parameter of the production well which is adjustable to control the control variables.

8. A control system according to claim 7, wherein the extraction system comprises a pump for outputting fluid from the production well and the at least one manipulated variable includes speed of the pump.

9. A control system according to any one of the preceding claims, wherein the at least one manipulated variable comprises a parameter of the injection well which is adjustable to control the control variables.

10. A control system according to any one of the preceding claims, wherein the processor is further configured to input at least one limit associated with the at least one control variable.

1 1. A control system according to claim 10, wherein the at least one limit is input by an operator.

12. A control system according to claim 1 1 , wherein the at least one limit is the at least one manipulated variable.

13. A control system according to claim 12, wherein the at least one manipulated variable is the limit for rate of the steam injection.

14. A control system according to claim 13, wherein the injection well comprises a short string and a long string and wherein the at least one manipulated variable is the limit for rate of the steam injection into one or both of the short and long strings.

15. A control system according to any one of claims 10 to 14, wherein the processor is configured to determine whether the at least one manipulated variable is to be changed by comparing the predicted values of the control variables with the at least one limit.

16. A control system according to any one of the preceding claims, further comprising a subsidiary controller for controlling the at least one manipulated variable and wherein the central processor is configured to output any change to the subsidiary controller to implement the change.

17. A control system according to claim 16, wherein the subsidiary controller is a flow controller controlling a valve on an input to the extraction system.

18. A control system according to claim 16 or claim 17, wherein the subsidiary controller is a product controller controlling an output from the extraction system.

19. A control system according to any one of the preceding claims, wherein the processor is configured to iteratively repeat the receiving, predicting, determining and outputting steps.

20. A control system according to any of the preceding claims, wherein the model is further dependent on at least one disturbance variable which disables the processor when activated and wherein the processor is configured to:

determine whether the at least one disturbance variable is activated and cease outputting any change to the at least one manipulated variable until the at least one disturbance variable is deactivated or a predetermined time limit has passed.

21. A control system according to any one of the preceding claims, wherein the model comprises a plurality of finite impulse response models.

22. A control system according to any one of the preceding claims, wherein the model comprises at least one logic relationship.

23. Use of a control system according to any one of the preceding claims to reduce startup time after a circulation phase or a shut-down.

24. A well pair extraction system for an oil sand reservoir comprising:

an injection well for injecting steam into the reservoir;

a production well for outputting fluid from the reservoir, and

a control system as defined in any one of claims 1 to 22.

25. An extraction system for an oil sand reservoir comprising:

a plurality of well pair extraction systems as defined in claim 24;

a source of steam connected to each of the plurality of well pair extraction systems; and

an output connected to each of the plurality of well pair extraction systems.

26. An extraction system according to claim 25 further comprising a main controller connected to the control system for each well pair extraction system, the main controller being configured to control production and steam utilisation across the extraction system.

27. A method of controlling a well pair extraction system in an oil sand reservoir, the well pair extraction system comprising an injection well for injecting steam into the reservoir and a production well for outputting fluid from the reservoir, the method comprising:

predicting a future value of the at least one control variable by inputting the measurements into a model of the well pair extraction system, the model being dependent on the at least one control variable and at least one manipulated variable which is adjustable to control the control variables; determining, based on the predicted future value of the at least one control variable, whether the at least one manipulated variable for the well pair extraction system is to be changed, and

outputting any change to the at least one manipulated variable.

28. A carrier carrying computer program code to, when running, implement the method of claim 27.