CA3026694A1 - Advanced control of steam injection network - Google Patents

Abstract

Systems and methods are provided for controlling and optimizing the operation of a steam-assisted gravity drainage (SAGD) production facility for recovery of hydrocarbons from a hydrocarbon-containing reservoir. One aspect provides a system including a steam header controller determining a first set of outputs to control the distribution of steam from the plant to the steam pad, and one or more steam pad controllers to determine a second set of outputs for controlling the distribution of steam to the one or more well pairs within the steam pad. Each of these controllers can be implemented as an advanced process control (APC) controller. The APC controllers receive target values and a prioritization of the steam pads and well pairs, and allocate steam to the steam pads and the well pairs based at least in part on their prioritization and relative to their respective target values while maintaining process constraints.

Description

ADVANCED CONTROL OF STEAM INJECTION NETWORK
TECHNICAL FIELD
[0001] The present disclosures relates to systems and methods for controlling and optimizing operation of a hydrocarbon production facility.
BACKGROUND

[0002] Steam-assisted gravity drainage ("SAGD") uses a pair of wells to produce hydrocarbons from a hydrocarbon-containing reservoir. The well pair typically includes two horizontal wells vertically spaced from one another, with the upper well used to inject steam into the reservoir and the lower well used to produce the hydrocarbon. The injected steam generates a steam chamber in the reservoir, and heat from the steam operates to lower the viscosity of the hydrocarbon, allowing for gravity drainage, and thereby production from the production well. The produced fluids typically include a mixture of hydrocarbons and water, including water formed from the condensing of the steam.

[0003] Some technologies that are used to manage SAGD operations include regulatory process control (RPC), which is found at many industrial sites. The RPC
solution is limited to maintaining control of individual process measurements, such as a level in a tank or flow through a pump. Furthermore, this solution lacks digital optimization of the process.

[0004] Another technology that is used to manage SAGD operations is advanced regulatory control (ARC). ARC utilizes existing control system architecture.
This requires extensive programming and is difficult to troubleshoot and maintain.
Such deployments often become quickly obsolete as the level of complexity in the deployments become difficult to manage and end up being abandoned as a solution.

[0005] There is a need for systems and methods which address the above-noted problems and disadvantages.

SUMMARY

[0006] In one aspect, there is provided a system for controlling a steam-assisted gravity drainage (SAGD) production facility, wherein the facility has a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads having one or more well pairs. The system includes a steam header controller adapted to determine a first set of outputs for controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to each one of the steam pads. The system also includes, for each one of the steam pads, a corresponding steam pad controller adapted to determine a second set of outputs for controlling the distribution of steam received at the steam pad via the steam header, to the one or more well pairs within the steam pad. Each of the steam header controller and each one of the steam pad controllers is implemented as an advanced process control (APC) controller. The APC controllers receive target values and a prioritization of the steam pads and well pairs, and allocate steam to the steam pads and the well pairs based at least in part on their prioritization and relative to their respective target values while maintaining process constraints. In particular implementations, each of the APC
controllers is a dynamic matrix control (DMC) multivariable predictive controller.

[0007] In some implementations, the steam header controller is adapted to maintain the pressure of the steam header at a target steam header pressure value. The target steam header pressure value is based at least in part on a total steam flow for the plant.
The steam header controller is adapted to adjust the allocation of steam to the steam pads in response to a trip detected in the plant or in one of the steam pads.

[0008] In some implementations, the steam header controller is configured to allocate steam to the steam pads based at least in part on a give up factor assigned to each one of the steam pads, wherein a high priority steam pad is assigned a small give up factor so that steam allocation to the high priority steam pad does not significantly deviate from its respective target value.

[0009] In some implementations, the steam pad controller for a steam pad is adapted to maintain the total steam used by the steam pad at a pad steam target value.
The steam pad controller is adapted to balance steam flows relative to their target for each of the well pairs having the same priority. The steam pad controller is configured to calculate a capacity of each well pair for both tubing and casing steam flows based on a constraint which can be one or more of: pressure high limit, valve high limit, and steam flow limit. For each one of the steam pads, the well pairs within the steam pad are ranked, and the corresponding steam pad controller is configured to give up on the well pairs in order of their ranking if a steam flow limit is reached during operation of the well pairs.

[0010] Particular implementations include an interface module for connecting a distributed control system of the SAGD production facility to one or more of the APC
controllers. The interface module receives outputs from the one or more of the connected APC controllers and performs one or more of data conditioning, data processing and further optimization of the outputs, to produce APC outputs suitable for execution by the distributed control system.

[0011] In some implementations, the system includes a steam quality controller adapted to control and optimize steam quality in each of the flow passes of a boiler of a steam generating plant while maintaining process constraints. The steam quality controller is implemented as a model predictive control (MPC) controller which is configured to predict in advance a trajectory for one or more process variables to enable proactive control of the steam quality.

[0012] In another aspect, there is provided a system for controlling the operation of SAGD production facility, the system incorporating an advanced process control (APC) controller adapted for determining one or more outputs for controlling the distribution of steam from a steam generating plant to steam pads via a steam header operatively coupling the plant to each one of the steam pads. In determining the one or more outputs the APC controller receives target values and a prioritization of the steam pads, and allocates steam to the steam pads based at least in part on their prioritization and relative to their respective target values while maintaining process constraints. The system also includes an interface module for connecting a control system of the SAGD

production facility to the APC controller, wherein the interface module delivers the one or more outputs of the APC controller to the control system for execution by the control system to control the distribution of steam. In some implementations, the APC
controller is configured to allocate steam to the steam pads based at least in part on a give up factor assigned to each one of the steam pads, wherein a high priority steam pad is assigned a small give up factor so that steam allocation to the high priority steam pad does not significantly deviate from its respective target value.

[0013] Another aspect provides a method for controlling a SAGD production facility, wherein the facility has a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads having one or more well pairs. The method includes controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to each one of the steam pads, and, for each one of the steam pads, controlling the distribution of steam received at the steam pad via the steam header, to the one or more well pairs within the steam pad.
Controlling the distribution of steam includes receiving target values and a prioritization of the steam pads and well pairs and applying APC techniques to allocate steam to the steam pads and the well pairs based at least in part on their prioritization and relative to their respective target values while maintaining process constraints.

[0014] Another aspect provides an APC server for controlling the operation of a SAGD production facility, wherein the facility has a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads having one or more well pairs. The server incorporates a model predictive control (MPC) controller configured to determine one or more outputs for controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to each one of the steam pads. In determining the one or more outputs, the MPC
controller receives target values and a prioritization of the steam pads, and allocates steam to the steam pads based at least in part on their prioritization and relative to their respective target values while maintaining process constraints.

_

[0015] The details of one or more implementations are set forth in the description below. Other features and advantages will be apparent from the specification and the claims.

[0016] Among other advantages, and as explained in further detail below, the APC
solutions described herein enable improved steam quality and improved distribution of steam, leading to reduced greenhouse gas emissions per unit of bitumen produced, reduced water and chemical consumption, and reduced manual interventions.
= BRIEF DESCRIPTION OF DRAWINGS

[0017] Features and advantages of implementations of the present application will become apparent from the following detailed description and the appended drawings, in which:

[0018] Figure 1 illustrates a well pair in an oil sands reservoir used in a steam-assisted gravity drainage (SAGD) process;

[0019] Figure 2 illustrates a steam distribution network for an SAGD
production facility;

[0020] Figure 3 illustrates a steam distribution network for an SAGD
production facility, incorporating a system according to one implementation for optimizing the distribution of steam; and

[0021] Figure 4 illustrates the flow of data through a system for optimizing the distribution of steam in a steam distribution network according to one implementation.
DETAILED DESCRIPTION

[0022] Throughout this specification, numerous terms and expressions are used in accordance with their ordinary meanings. Provided below are definitions of some additional terms and expressions that are used in the description that follows.

[0023] A "formation" or "geological formation" is a fundamental unit of lithostratigraphic classification. A formation includes rock strata that have comparable lithologies, fades, or other similar properties. Formations can be defined on the basis of the thickness of the rock strata of which they consist, and the thickness of different formations can vary widely. A given stratigraphic column can include a number of formations. In the oil sands area of Northeastern Alberta, for example, the stratigraphic column consists of the following major formations (from basement to surface):
Pre-Cambrian (basement), Devonian carbonates, McMurray oil sands, Wabiskaw sands and mudstones, Clearwater shales, Grand Rapids sandstones, and Quaternary sediments.

[0024] The "McMurray formation" or "McMurray sands" is a stratigraphic unit of Early Cretaceous age in the Western Canada Sedimentary Basin of Northeastern Alberta. It lies unconformably on Pre-Cretaceous erosion surfaces that generally comprise Devonian limestone, which is mainly carbonate rock. The McMurray sands are largely unconsolidated and the sand grains that form the formation are mostly held together by very viscous crude oil. The McMurray formation holds most of the vast hydrocarbon resources of the Athabasca bituminous sand deposit.

[0025] "Reservoir" refers to a subsurface formation containing one or more natural accumulations of hydrocarbons, which are generally confined by relatively impermeable rock or other geological layers of materials, including subsurface formations that are primarily composed of a matrix of unconsolidated sand, with hydrocarbons occurring in the porous matrix.

[0026] "Hydrocarbons" refer to a combination of different hydrocarbons or a combination of various types of molecules that contain carbon atoms and attached hydrogen atoms. Hydrocarbons include a large number of different molecules in gaseous, liquid, or solid phase having a wide range of molecular weights, and can include bitumen, heavy oil, lighter grades of oil, and natural gas. Elements (e.g., sulphur, nitrogen, oxygen), metals (e.g., iron, nickel, vanadium), and compounds (e.g., carbon dioxide, hydrogen sulphide) are sometimes present in the form of impurities in a desired hydrocarbon mixture.

[0027] The term "drilling" refers to the creation of a borehole in a formation by rotating a drill bit and simultaneously applying an axial load to the bit.

[0028] An "injection well" or "injector includes a well into which a fluid or gas/vapour is injected into a formation.

[0029] A "production well" or "producer" includes any well or wellbore from which hydrocarbons can be produced, regardless of its configuration or arrangement.
The production well can be configured vertically, horizontally, or at any angle from vertical to horizontal or beyond horizontal, in any portion thereof.

[0030] "Bitumen" and "heavy oil" are normally distinguished from other petroleums based on their relative densities and/or viscosities, which often depend on context.
Commonly-accepted definitions classify "heavy oil" as petroleum (the density of which is between 920 and 1,000 kg/m3) and "bitumen" as oil produced from bituminous sand formations (the density of which is greater than 1,000 kg/m3). For purposes of this specification, the terms "bitumen" and "heavy oil" are used interchangeably such that each one includes the other. For example, where the term "bitumen" is used alone, it includes within its scope "heavy oil".

[0031] A "well pad" or "steam pad" includes a collection of wells (which can be arranged in well pairs) drilled into a hydrocarbon formation and associated process equipment, which can include piping (e.g., steam lines, gas lines, solvent lines, and the like), instrumentation, electricity supply, testing equipment, and/or other equipment, positioned at the surface generally above the hydrocarbon bearing formation, and operatively connected, directly or indirectly, to a plurality of the wells or at least one well in a plurality of the well pairs. In some implementations, the well pairs include an injection well and a production well configured to provide a SAGD operation.

[0032] Specific examples of the present methods and systems are described below with reference to the drawings. Details are provided for the purpose of illustration, and the methods and systems can be practiced without some or all of the features discussed herein. For clarity, technical materials that are known in the fields relevant to the present methods and systems are not discussed in detail.

[0033] One aspect of the application herein provides a system for controlling the operation of a steam-assisted gravity drainage (SAGD) production facility, wherein the facility comprises a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads comprising one or more well pairs. The system incorporates one or more advanced process control (APC) controllers which are adapted for determining one or more outputs for controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to the steam pads. In determining the one or more outputs, the APC controller receives target values and a prioritization of the steam pads, and allocates steam to the steam pads based at least in part on their prioritization and relative to their respective target values while maintaining process constraints. An interface module is provided for connecting a control system of the SAGD production facility to the APC
controller, wherein the interface module delivers the one or more outputs of the APC
controller to the control system for execution by the control system to control the distribution of steam. APC controllers can further be provided at the pad level to determine outputs for controlling the distribution of steam received at the steam pad via the steam header, to the one or more well pairs within the steam pad. Additional aspects will become apparent by reference to the implementations described below.

[0034] Figure 1 illustrates a common well pair 12 in a SAGD operation on a reservoir 14. It consists of a pair of horizontal well bores 12A, 12B that are spaced 5 to 8 meters apart. The upper bore 12A injects steam into the reservoir 14 and the lower well bore 12B extracts the bitumen from the reservoir 14. Figure 2 shows a typical steam distribution network 20 for an SAGD production facility. As seen in Figure 2, there are multiple steam "pads" 16 (a pad is a group of well pairs 12) connected to a common steam pipeline manifold, commonly known as the "steam header" 18. Steam carried through the steam distribution network 20 is provided by steam generation plants 15.
The control of steam injection through the steam distribution network 20 in the SAGD
process poses a challenge. Any change in the injection rate for one pad 16 will affect , the steam flow as well as the steam header pressure to the other pads 16.
Existing control technologies focus on maintaining the steam header pressure within safe limits.
Reservoir engineers can set specific targets for each pad 16. However, using existing control technologies, the steam allocation to an individual pad 16 is often far from the optimal target as specified by reservoir engineers. A similar problem also exists for steam distribution among individual injection wells 12 within a pad 16. Each individual steam flow set point typically has to be manually adjusted by production engineers and operators. Due to the difficulties with applying manual intervention to a large number of wells 12, existing methods have not been able to achieve the steam optimization desired to improve steam utilization. Some of the pads 16, and some of the wells 12 within a pad 16, might be off from their target, as they receive an insufficient amount of steam or too much steam. As a result, steam is not optimally distributed, resulting in higher steam to oil ratio (SOR) and energy wasted. This lowers the SOR and the energy intensity per barrel of production. In addition, current approaches do not consider process constraints. This can result in inaccurate capacity estimates which affect decisions during the planning stages.

[0035] Particular implementations provide advanced process control (APC) solutions, including for instance, model predictive control (MPC) solutions, to optimize steam generation, steam distribution, and bitumen production at SAGD production facilities.
Improvements in steam generation and distribution achieved using the technology described herein have been found to improve the energy intensity per barrel as a result of the increased bitumen production for a given amount of steam supplied. This translates to reduced production costs and reduced greenhouse gas (GHG) emissions for the production of a given amount of bitumen.

[0036] Natural gas used for steam generation in some SAGD production facilities can account for over 20% of SAGD operational costs. These costs are dictated at least in part by the energy efficiency of the steam generators. The higher the energy efficiency, the less natural gas will be used and hence the lower the operational cost.
Types of steam generation processes used in SAGD bitumen production include: Once Through Heat Recovery Steam Generator (OTSG) and cogeneration units (also referred to as "cogen" units), wherein each is a type of tube boiler without a boiler drum.
For either an OTSG or cogeneration unit, boiler feed water is fed to the boiler via multiple feed tubes, often called passes, to generate steam. Steam quality is a key variable that can affect steam generator energy efficiency. Steam quality is defined as the mass fraction of dry steam in a saturated mixture of vapor, where dry steam has a steam quality of 100%, and water has a steam quality of 0%. Since the boiler feed water is recycled from underground steam condensate (coming back from the reservoir) which has a high concentration of impurities, steam quality that is too high will lead to deposition of solids in the boiler tubes, causing decreased heat transfer efficiency and local hot spots. Tube damage or rupture can result if steam quality remains high for a prolonged period. On the other hand, steam quality that is too low means lowered energy efficiency since the energy is used largely to increase water temperature, but not to convert water into steam.

[0037] As such, a target steam quality constraint needs to be observed to ensure both safe and efficient steam generator operation. However, the steam qualities among the different flow passes can be uneven due to an imbalanced heat distribution.
The pass with the highest steam quality tends to constrain the overall steam quality of the unit and restrict it from moving closer to a target steam quality.

[0038] In addition to steam quality, the distribution of steam for SAGD
production poses a challenge in a steam distribution network consisting of multiple steam pads (groups of well pairs or wells) connected to a steam header (see, for example, network 20 of Figure 2). In such a network, any change in the injection rate for one pad will affect the steam flow as well as the steam header pressure affecting the other pads.
Existing control technologies are directed primarily to maintaining the steam header pressure within safe limits without consideration of whether an allocation to an individual pad is optimal. Existing methods also typically involve manual adjustment of each individual steam flow set point to control the steam distribution among individual injection wells within a steam pad.

,

[0039] The individual control of production wells (which typically number in the several hundred) significantly affects the plant production rate. Well dynamics are very slow due to the size and complexity of the reservoir. In addition, changes in reservoir condition in one well can affect adjacent wells. Manual control or univariate controllers are not capable of managing such complex dynamics or interactions among the wells.

[0040] Methods and systems described herein are directed to the use of APC
solutions to improve energy efficiency in SAGD production facilities in two stages. The first stage (also referred to herein as "Stage 1") is to control and improve the quality of the steam produced by the steam generating plant. An objective of the APC
solution used at the first stage is to maintain steam quality as close to a predetermined steam quality constraint as possible to ensure both safe and efficient steam generator operation. For example, a steam quality constraint of 80% is observed in SAGD
production facilities in particular implementations.

[0041] Increasing steam quality improves efficiency as the energy from the combusted natural gas converts more of the water into steam which is then used to produce bitumen. Energy waste is reduced as there is less boiler blowdown (water not converted to steam is recycled in the water processing plant), the energy of which is not fully utilized for bitumen production. Thus, energy efficiency of steam generation is improved, resulting in reduced operational costs.

[0042] An APC digital optimization solution can be applied at Stage 1 to control and optimize steam quality while ensuring all process safe limits are maintained.
The APC
solution eliminates or reduces the steam quality deviation among the different passes, thus removing the constraint posed by the pass with highest steam quality. The steam quality can then be increased closer to the limit once all the pass qualities are aligned.
Pass flows and natural gas firing are controlled automatically and continuously. The solution uses empirical models configured within an MPG controller along with a linear programming algorithm to predict the trajectory of the process variables up to a certain period in advance (e.g. 30 minutes in advance) to allow for proactive control.

,

[0043] The second stage (also referred to herein as "Stage 2") is to control and improve steam distribution in the steam distribution network, which results in increased production from the wells. An objective of the APC solution at the second stage is to control multiple wells in a coordinated fashion to achieve steady production from a well pad. Difficulties in managing a large number of wells is reduced by the APC
solution as the method automatically optimizes the allocation of steam to the wells to the given targets while maintaining process constraints. "Optimize" and "optimization", as used herein, refer to the determination of a solution to a problem (e.g. allocation of steam in a steam distribution network) that strives to meet one or more targets while maintaining process constraints. An "optimal" solution to such problem includes, without limitation, any solution that is more effective than other solutions which do not satisfy process constraints and/or are further away from meeting the target.

[0044] Particular implementations of the second stage are directed to optimizing the distribution of steam from the steam supply (which can include one or more steam generation plants 15) to each pad 16 (i.e. a group of well pairs 12) and to the well pair 12 within each pad 16. Thus, there are two levels of advanced process control for steam injection and distribution within the steam distribution network 20. A first level of control operates at the steam header level for managing and controlling the distribution of steam to each pad 16 from the one or more steam generation plants 15 via the steam header 18 (also referred to as "plant level APC" or "steam header APC"). A
second level of control operates at the pad level for managing and controlling the distribution of steam to each well pair 12 within a pad 16 (also referred to as "pad level APC" or "steam pad APC"). In particular implementations of the second stage, each steam pad and well pair is prioritized according to their efficiency, and steam distribution is optimized to these requirements while maintaining process constraints, such as pressure constraints. The instances of steam short scenarios (not enough steam) and steam long scenarios (too much steam) are reduced or eliminated, while decreasing the energy wasted.

[0045] Referring to Figure 3, shown therein is a system 25 according to one implementation for managing and controlling steam distribution within a steam distribution network 20 having a supply of steam provided by a steam supply 15.
System 25 incorporates Model Predictive Control (MPC) solutions at each of the plant level and the pad level. Specifically, plant level MPC 22 is provided to manage and control the distribution of steam generated by steam supply 15 to a plurality of pads 16 which are coupled to the steam supply's steam header 18. In addition, pad level MPC
24 is provided to manage and control the distribution of steam delivered to each pad 16, to the well pairs 12 within the pad 16. Each of the well pairs 12 shown in Figure 3 can be assigned to one of the pads 16 (e.g. well pairs w1 through w4 constitute pad 1; well pairs w5 through w8 constitute pad 3; well pairs w9 through w12 constitute pad 3; and well pairs w13 through w16 constitute pad 4).

[0046] In particular implementations of system 25, one MPC controller is provided per asset and there are separate MPC components at each of the plant and pad levels. For example, to implement plant level MPC 22 shown in Figure 3, each steam generation plant is assigned an MPC controller to manage and control the distribution of steam generated by the plant, to the various pads 16 operatively connected to the plant via the steam header 18. To implement pad level MPC 24 shown in Figure 3, each pad 16 is assigned an MPC controller to manage and control the distribution of steam delivered to pad 16.

[0047] In particular implementations, each controller application (implementing an MPC component) resides in a single server. Once the application has been configured on the server, the application is deployed on a host server, referred to as an online server. The online server provides an online engine to execute the application and read/write to the production facility. Some implementations have three controller applications on average running on one online server.

[0048] An additional server can be provided to enable engineers to make online changes to the controller applications. Another server can be provided to log inputs and outputs to the APC system. This server allows the engineers to analyze the black box (empirical) models used in the APC server to evaluate their effectiveness as well as troubleshoot any irregularities.

[0049] MPC is an automated process plant control strategy that involves the development of a process model that predicts and optimizes future process behaviour.
This technology can be implemented without altering existing facility control system architecture and can therefore be readily deployed to brownfield and greenfield projects.
MPC is a proactive tool which predicts future process behavior using empirical models built from extensive historical data analysis. This allows for a higher level of productivity as the process can be efficiently operated closer to the design limits, leading to improved process production with lower GHG emission intensity per barrel of oil produced.

[0050] The MPC solutions described herein are multi variant: the MPC
controller can control multiple process inputs and outputs simultaneously and optimize interacting processes, such as multiple OTSG tube flows in a process unit. This reduces or minimizes manual intervention and effectively optimizes the process despite process complexities. MPC, in addition to controlling the process, digitally optimizes the process to the best possible solution using objective functions, such as linear or quadratic programming algorithms among the control variables (e.g. balanced steam production).

[0051] MPC is a tool that utilizes advanced algorithms to predict process dynamics.
MPC uses data analytics via a system identification algorithm to produce models between interconnected variables. MPC can additionally optimize the process via constraints using economic or environmental drivers, such as minimization of natural gas consumption. MPC controls multiple parameters simultaneously and drives the whole process toward an optimal point. At the same time, MPC predicts process measurements, allowing the forecast of any upsets or deviations, thus enabling corrections in the process.

[0052] Linear or quadratic optimization functions can be utilized to drive a higher value of the process depending on the objective. Linear programming (LP) can be used to solve for the best outcome with a mathematical model whose requirements are represented by linear relationships. Quadratic programming (QP) can be used to solve a quadratic optimization problem with a quadratic function of several variables subject to linear constraints. LP will sacrifice one variable at a time within the multi-variable control to achieve the desired control objective, while QP will spread the error among all the control variables.

[0053] The development of an APC solution begins by compiling a list of available process variables and measurements. Historical process data is then collected to identify the process model. Empirical models are cross validated using a separate historical data set in order to ensure model consistency and accuracy. Then, the APC
application is configured and customized according to the control objectives and process variable priorities. Simulations are conducted to validate that the objectives and desired process outcomes are achieved. Additional advanced calculations can be incorporated due to the complexity of the SAGD process. In particular implementations, the models that are developed using APC software (i.e., the APC application) can reside on an APC server. In alternate implementations, the models reside on an embedded controller.

[0054] One or more algorithms, also referred to as "logic", are configured in the existing plant control system (such as the plant's existing distributed control system (DCS)) to interface and communicate with the APC application that resides on the APC
server. The interface logic includes various condition-based algorithms to accommodate for changing reservoir and process changes. Interface logic (live process data from the control system passed to the APC server and control commands sent from the APC
server to the control system) are connected and referenced to existing process control programming, thus allowing the APC application to assume control of the process. The combination of the APC configuration on the APC server and the interface configuration on the existing plant control system provide the APC solution for management and control of the injection and distribution of steam through the steam distribution network of an SAGD production facility.

[0055] The configuration can be customized and improved as more deployments are completed and as the solution matures. This iteration of deployments enables incremental improvements of the components of the APC solution. Over time, the r configuration difference among various deployments can grow wider, rendering the solution on various assets as non-standard.

[0056] The solutions at the plant level APC and pad level APC can be templated for more efficient deployment to different steam distribution networks, each of which can have varying numbers of pads and well pairs per pad, with various characteristics.
Templating is the process of standardizing the various components of the solution to a level that would allow any user meeting the minimum input and output criteria to take advantage of this digital optimization solution across the oil and gas industry and beyond. Furthermore, once the components are templated, the deployment time frame of the APC solution would decrease, thus increasing the rate of return and increasing the attractiveness of the solution as a whole.

[0057] The end goal of templating is to produce design files and engineering manuals containing instructions of the steps required for deployment of the APC
solutions regardless of the automation platform. The design and manuals provide the end user with the engineering knowledge to implement the solution.

[0058] Referring to Figure 4, there is shown a data flow diagram illustrating the flow of data and connectivity of an APC solution according to one implementation. This implementation incorporates a modular integration of the APC solution with an existing distributed control system (DCS) 40 of the plant to allow for safe and low risk integration within an operating SAGD production facility. An APC system 30 accepts inputs (which include process controls, measurements, and other inputs from the SAGD
process) and generates APC outputs 55 for controlling the SAGD process 42. In particular embodiments described herein, the APC outputs 55 are used for controlling the distribution of steam in an SAGD steam distribution network. In some implementations, at least some of these generated outputs 55 are used as inputs 52 in another APC system (e.g. plant level APC system or pad level APC system), or can be fed back as inputs 52 into the same APC system 30 along with new inputs 52' (which include process controls, measurements and other inputs from the SAGD process) to generate new or refined outputs 55. For example, an APC output can be the prediction -horizon which can be used as an input to a calculation in the same system to adjust a subsequent APC output of the same system. In addition, the APC output from one system can be used as the APC input (e.g. target) for another system. For example, the plant level APC outputs a value that is used as a target for the pad level APC
(e.g. pad steam rate).

[0059] Referring in more detail to the components of APC system 30, the inputs 52 to the APC solution are provided to an MPG component 32 which is implemented as an APC application residing on an APC server of the APC system 30. The MPG
component 32 residing on an APC server is also referred to as an "APC
controller" or an "MPG controller". The MPG component 32 incorporates empirical (black box) models 31, proprietary calculations 33, and advanced optimization algorithms 35.
These are used to generate intermediate outputs 54 based on the received inputs 52. An interface module 34 connects the MPG component 32 with an existing distributed control system 40 of the SAGD production facilities. The interface module 34 receives the intermediate outputs 54 from the MPG component 32 and performs data conditioning 36, data processing 37 and further optimization 38 to produce APC outputs 55. APC
outputs 55 are then sent to the plant process 42 for final control and execution by the plant process, including the steam generating plant (e.g. OTSG or cogeneration unit) and the pads or groups of well pairs.

[0060] A steam header APC controller is deployed in order to allocate the optimal amount of steam to the pads (plant level APC). This provides the optimal amount of steam required by a pad, which is at least partly dictated by requirements of the individual wells within the pad. An additional steam pad APC controller is deployed in conjunction with the steam header APC controller to enable individual wells to receive the optimal amount of steam (pad level APC). The steam header APC determines the amount of steam to be distributed to the pad while the steam pad APC
determines the optimal steam flow to be distributed to the individual wells to ensure optimal reservoir growth. Each of steam header APC controller and the steam pad APC controller can be implemented as an APC application residing on APC server infrastructure (as seen, for example, in Figure 4) or on an embedded controller. The steam header APC
controller and steam pad APC controller are communicatively coupled, such that some of the outputs or predictions of the steam pad APC controller be used as inputs or fed back into the steam header APC controller. For example, in some implementations the steam pad APC controller flags when a pad trips, and in response to such flag the steam header APC controller proactively takes action to reallocate the steam before any impact on the header pressure results from the pad trip. As a further example, if a steam pad APC controller determines that its pad is limited, it can provide a flag to the steam header APC controller not to increase the target for this pad further.
[00611 A list of APC solution inputs 52 for Stages 1 and 2 is shown in Table 1-below. As noted previously, Stage 2 includes both plant level APC (at the steam header, to allocate the optimal amount of steam to the pads) and pad level APC (at the pad, to determine optimal distribution of steam to the individual wells within each pad).
Table 1-1 APC Solution Inputs Solution Inputs Category Stage Note APC server Hardware 1, 2 Hardware requirement to host APC software APC software and Software 1, 2 APC software licensing Communication server Hardware 1, 2 Available for all automation systems Good quality historical Information 1, 2 Available as historical data process measurement data Firing rate Process 1 Available with any measurement OTSG/cogen Boiler feed water Process 1 Available with any measurement OTSG/cogen Steam tube flow rate Process 1 Available with any measurement OTSG/cogen Steam tube steam Process/soft 1 Soft sensor to estimate quality sensor steam quality Tubing steam rate Process 2 Available on a well head measurement Tubing pressure Process 2 Available on a well head measurement Casing steam rate Process 2 Available on a well head measurement Casing pressure Process 2 Available on a well head measurement Electric submersible Process 2 Available on artificial life pump (ESP) speed wells measurement Tubing discharge Process 2 Available on a well head pressure measurement Reservoir sub cool Process 2 Estimated using temperature measurements Bore level Process 2 Estimated using tubing pressure and suction pressure of ESP
Pump sub cool Process 2 Available on most well heads [0062] A list of APC solution outputs 55 for Stages 1 and 2 is shown in Table below.
Table 1-2 APC Solution Outputs Solution Outputs Category Stage Note Firing rate of steam Process 1 Available with any asset set point OTSG/cogen Boiler feed water to Process 1 Available with any steam asset set point OTSG/cogen Steam quality lab Lab 1 Available at most sites sample Steam tube flow rate of Process 1 Available at most sites steam asset set point Tubing steam rate set Process 2 Available on a well head point-injection well Casing steam rate set Process 2 Available on a well head point-injection well ESP speed set point- Process 2 Available on artificial life subsurface pump wells Steam header pressure Process 2 Available on steam measurement distribution networks Tubing discharge Process 2 Mostly available on pressure measurement wells/not critical (Optional) Downhole temperature Process 2 Available on most well measurement- heads Subsurface Suction pressure of Process 2 Available on most well ESP measurement heads [0063] In particular implementations, some objectives of the steam header APC
controller for a steam generating plant (e.g. OTSG or cogeneration unit) are as follows:
1. To maintain the pressure of the plant steam header at target.
2. To control the header pressure by optimally distributing the steam to the pads (wherein parameters concerning optimal distribution are defined by reservoir and production engineering groups).
3. To adjust the steam rates to the pads in response to OTSG or cogeneration unit trips (an unplanned event occurring in the boiler).
4. To adjust the steam rates to the pads in response to pad trips (an unplanned event occurring in the steam pad).
[0064] Steam generation rates for an OTSG and cogeneration unit are set by the control room operators. The steam sent to the pads will generally be off target by a certain amount (Le. the amount of steam produced and the sum of the steam targets at the well pairs are not always matched). An objective of the advanced control strategy according to a particular implementation described herein is to control the steam header pressure by manipulating the steam to the pads in such a way as to distribute the error from steam targets across the pads based on their prioritization.
[0065] The steam header APC controller in the implementations described below is a .. dynamic matrix control (DMC) multivariable predictive controller. For example, such DMC controller can be an Aspen DMCplusTM controller. A DMC controller determines optimal input and output targets by solving a sequence of quadratic programming equations. Control variables (CVs) and manipulated variables (MVs) are ranked in priority. The controller does not sacrifice steady state control performance for a CV to improve the performance of lower priority CVs. The priority of MVs determines how to use extra degrees of freedom.

[0066] An initial controller design for the steam header APC controlled the plant steam header pressure as a controlled variable by breaking the proportional-integral-derivative (PID) loop of the plant PID controller, and using steam to each pad to maintain the pressure at its target. With this strategy, the APC controller had to coordinate moves among the steam pads to maintain the pressure at its target under stable conditions and catch the pressure during disturbances such as OTSG/cogen unit trips or steam pad trips. Possible causes for OTSG/cogen unit trips include poor combustion conditions, high tube temperature, high tube pressure, or a fuel rich scenario (e.g. in such case, the air louver freezes and reduces the amount of 02, leading to excess 02 measurement that is below the required ratio to natural gas, which is around 3% in particular implementations).0ther causes of OTSG/cogen unit trips include instrumentation failures or power outages to the unit. Possible causes of pad trips include instrumentation failures on the wells, high or low steam header pressure, high pressure in multiple reservoirs, or a power outage to the steam injection system.
Indications of a trip include lack of steam flow from the generator, or lack of steam flow to the pad.
[0067] The controller design was subsequently changed to instead use the output from the plant steam PID controller (i.e. the plant total steam flow, as provided by the existing header pressure control PID loop) as the target for a controlled variable. The plant PID controller already maintains the steam header pressure at its setpoint and catches it during disturbances. This strategy matches the performance of the existing PID loop (which needed to be met or exceeded) by using the PID loop and optimally distributing the total flow among well pads, satisfying the objective criteria above.
[0068] In a particular implementation, the plant total steam flow is used to set the target for the steam header APC controller. The existing distributed control system (DCS) steam distribution logic, the pad splitter, can be used as a backup control strategy for times when the advanced control strategy is not in service.
[0069] The following table lists the independent variables for a steam header APC
controller, delivering steam to 9 well pads, according to one implementation:

Table 2-1 Manipulated Variables Index Tag Name Description Units 1 1 FC001 SP Total Steam to Pad 1 m3/h 2 2FC80100SP Total Steam to Pad 2 m3/h 3 3FC80100SP Total Steam to Pad 3 m3/h 4 4FC80100SP Total Steam to Pad 4 m3/h 5FC80100SP Total Steam to Pad 5 m3/h 6 6FC80100SP Total Steam to Pad 6 m3/h 7 7FC801 00SP Total Steam to Pad 7 m3/h 8 8FC80100SP Total Steam to Pad 8 m3/h 9 9FC80100SP Total Steam to Pad 9 m3/h [0070] The following table lists the dependent variables for the same steam header APC controller:
Table 2-2 Controlled Variables Index Tag Name Description Units 1 TOTFLOW Plant Total Steam Flow m3/h 2 1 FDT80100 Pad 1 Distributable Target m3/h Value 3 2FDT80100 Pad 2 Distributable Target m3/h Value 4 3FDT80100 Pad 3 Distributable Target m3/h Value 5 4FDT80100 Pad 4 Distributable Target m3/h Value 6 5FDT80100 Pad 5 Distributable Target m3/h Value 7 6FDT80100 Pad 6 Distributable Target m3/h Value 8 7FDT80100 Pad 7 Distributable Target m3/h Value 9 8FDT80100 Pad 8 Distributable Target m3/h Value 9FDT80100 Pad 9 Distributable Target m3/h Value [0071] Calculated values for a steam pad's distributable steam measured value, MV
low limit, and MV high limit depend on whether the steam pad has a DMC
controller as a steam pad APC controller, or whether the steam pad is using existing logic.
In 5 particular implementations, existing logic is provided in the distributed control system (DCS) of the production facility. The pad splitter is logic in the DCS that distributes the steam; however, the pad splitter does so differently from the steam pad APC
controller as the pad splitter is not predictive, does not take into account the pressure in the tubing, and does not optimize for a target against a set of constraints. For a steam pad 10 with a DMC controller, the distributable steam measured value is the sum of the steam flow setpoints for the tubing and casing flows that are controlled by existing logic; the steam rate MV low limit is calculated as the sum of the absolute low steam rate limits for the tubing and casing flow that are controlled by existing logic plus the flows rates of any steam flows not controlled by existing logic; and the steam rate MV high limit is calculated as the sum of the absolute high steam rate limits for the tubing and casing flow that are controlled by existing logic plus the flows rates of any steam flows not controlled by existing logic. The manipulated variables in Table 2-1 are manipulated by the APC controller between the MV low and high limits so that the distributable steam measured value can be controlled to the targets in Table 2-2.
[0072] For pads that are using the existing logic, the MV low limit is calculated as the sum of the minimum steam flows plus the flows rates of any steam flows not controlled by existing logic, and the steam rate MV high limit is calculated as the sum of the maximum steam flows plus the flows rates of any steam flows not controlled by existing logic. The existing logic is not capable of delivering more steam than is defined by the target calculation. This is due to the fact that the existing logic uses the maximum value as the target. The consequence is that the steam header APC controller will not be able to send more steam than is defined by the target to the pads that use the existing logic.

In a steam long scenario the pads that use the existing logic will be at their target while the pads which have DMC controllers will be taking in all the extra steam.
[0073] In a particular implementation, a steam flow controller in a well can operate in one of the following modes:
1) Manual: The operator manually controls the valve opening, e.g. if a value is set at 50%, the flow of steam is not considered under control. Manual mode can be used when the operator needs to take control due to an upset or a steam flow measurement issue, for example.
2) Auto: In auto mode, the operator sets a set point for the flow that he or she desires.
3) Cascade: This is the mode where the existing logic is active. This indicates that the wells in cascade can be manipulated and have the steam adjusted. For example, take the case where there are two wells in cascade mode and two wells in auto mode.
If the wells in cascade mode have their absolute high steam rate limits set to 10, this informs the MPC controller that the wells can have their flow increased to 10 once the MPC
controller takes over these wells. If the two auto wells stay in auto mode, the MPC
controller cannot take control of these wells.
4) R-Cascade: This mode is used to enable the MPC controller to take control over the steam flow controllers. Each well in a well-pair has settings for the absolute low steam rate limit and absolute high steam rate limit, as input by the reservoir engineer. These limits are used to inform the MPC controller of the constraints. In addition, each well will have a target for the MPC controller, and an intermediate limit. For example, if the absolute low steam rate limit is 5 and absolute high steam rate limit is 20 and the target is 10, the intermediate low would be 7.5 and the intermediate high will be 15.
Depending on the backdown order and priority of the wells, steam can be decreased to the intermediate target for a set of wells with priority 2 before the priority 1 wells are decreased in a steam short scenario.
[0074] The distributable target is measured by summing the target of the wells that are not in auto or manual mode. This implies that MPC has taken control of these wells and the corresponding targets set by the reservoir engineer can be used in the distribution and optimization of steam in these wells.
[0075] The distributable measure value and the steam rate MV low limit and high limit can be related as follows: since the distributable measure references all the wells that .. are in the mode that the MPG can take control of and manipulate, it does not mean that is the total amount of steam the pad is getting. For example, take the case where there are four wells, two of which are controlled in MPG with a target of 10, an absolute low steam rate limit of 5 and an absolute high steam rate limit of 20. If the other two wells are in auto mode and have a flow rate set point entered of 5 each (i.e.
distribution to these wells is not optimized, as it is desired that their steam rates are kept constant), the distributable target would be 20 and MV high limit would be the sum of the absolute high steam rate limit of the first two wells as each well can receive up to 20 plus the sum of the setpoint of the last two wells as these wells are already getting 5 m3/h. Thus, the steam rate MV high limit would be 50, and respectively summing the absolute low steam rate limit of the first two wells plus the set points of the last two wells yields an absolute low steam rate MV rate of 20.The pad can obtain up to 50 m3/h as a maximum, but 10 are already flowing from the wells in auto mode and the MPG can only increase 10 on each of the first two wells.
[0076] To help elicit a proper controller response to OTSG or cogen trips, the boiler feed water (BFW) rates to these operations are included in the calculations to detect a trip. However, there are times when the BFW rates do not go all the way to zero after a trip despite the fact that the steam to the header is completely gone. To remedy this, the BFW flow rate can be considered along with the statuses that represent the state of the steam generator's operation. In some implementations, these include the steam to .. header status and the fuel gas status. A status value of 0 indicates that either the steam is shut off from the header or the fuel gas is shut off from the burners. Both statuses must have values of 1 for the steam to be flowing to the header. The steam generator feedforward values can be calculated as follows:
BFW Flow = BFW_Setpoint * Steam_to_Header Status * Fuel_Gas_Status __¨

Using this calculation, the BFW value is immediately set to zero upon an OTSG
or cogen trip. This activates logic that sets the maximum move appropriately for the MVs.
[0077] For the DMC controller of the steam header APC to provide a proper response to a pad trip, distributed control system (DCS) logic can be used to detect the trip and place the corresponding steam pad MV in feed forward. For example, let down valve output can be used to check if a pad trip has occurred. If a pad trip is identified, the controller sets the steam flow to the corresponding pad manipulated variable measurement to zero, sets the maximum move to the remaining MVs appropriately, and starts redistributing the extra steam to the other pads.
[0078] As noted above, the steam flow targets to the pads are generally not met exactly. By default, the desired distribution is that which will give up equally on the distributable steam (DT) values for each pad on a percentage basis. However, particular embodiments enable the prioritization of steam rates to each pad relative to the targets.
This can be implemented through a "give up" factor that is determined by reservoir and production engineers. Table 3-1 below presents the default solution for a sample scenario where a total of 3223.5 m3/hr of steam must be sent to the pads, but the pads are only requesting a total of 3050 m3/hr of steam.
Table 3-1 ¨ Pad steam distribution for default distribution on equal percentage Total Steam Percent of Distributable Final Value Percent (TT) TT Flow in Steam (DT) Error from RCAS DT
Pad 1 400 m3/hr 45% 180 m3/hr 412.0 m3/hr 6.69%
Pad 2 450 m3/hr 100% 450 m3/hr 480.1 m3/hr 6.69%
Pad 3 375 m3/hr 90% 337.5 m3/hr 397.6 m3/hr 6.69%
Pad 4 500 m3/hr 60% 300 m3/hr 520.1 m3/hr 6.69%
Pad 5 450 m3/hr 100% 450 m3/hr 480.1 m3/hr 6.69%
Pad 6 275 m3/hr 100% 275 m3/hr 293.4 m3/hr 6.69%
Pad 7 325 m3/hr 100% 325 m3/hr 346.8 m3/hr 6.69%
Pad 8 275 m3/hr 100% 275 m3/hr 293.4 m3/hr 6.69%

[0079] The example shown in Table 3-1 above does not include the addition biasing that can be used to prioritize the steam give up on each pad. Table 3-2, below, shows a solution for the same scenario where the pads are prioritized using a 'give up' factor.
Table 3-2 ¨ Pad steam distribution for distribution with give-up factor Total Percent Distributable Give Final Value Percent Steam of TT Target (DT) Up Error (TT) Flow in Factor from DT
RCAS
Pad 1 400 m3/hr 45% 180 m3/hr 100 417.30 9.61%
m3/hr Pad 2 450 m3/hr 100% 450 m3/hr 75 482.44 7.21%
m3/hr Pad 3 375 m3/hr 90% 337.5 m3/hr 100 407.44 9.61%
m3/hr Pad 4 500 m3/hr 60% 300 m3/hr 50 514.42 4.81%
m3/hr Pad 5 450 ms/hr 100% 450 m3/hr 75 482.44m3/hr 7.21%
Pad 6 275 m3/hr 100% 275 m3/hr 0.1 275.03 0.01%
m3/hr Pad 7 325 m3/hr 100% 325 m3/hr 100 356.23 9.61%
m3/hr Pad 8 275 ms/hr 100% 275 m3/hr 50 288.21 4.81%
m3/hr [0080] An interface module 34 connects the MPG component 32 with an existing distributed control system 40 of the SAGD production facilities. Give up factors are between 0.1 and 100, where pads with a give up factor of 100 will have the highest percent error. High priority pads can be defined by using a very small give up factor such as 0.1. In Table 3-2, Pad 6 has been specified as high priority and the steam is not permitted to significantly deviate from the target value. For the percent error on DT to be exactly proportional to the give up factors, at least one pad must have a give up factor of 100.
[0081] To provide the distribution described above, the CV Target ECE (Equal Concern Error, used as the weight-assigned steady state control error for the process variables used to distribute the steam on an equal percentage of target) can be calculated as follows:

_icouo Pad_1_ECE j equation (1) where GU is the give up factor for Pad 1, and is a value between 0.1 and 100.
[0082] For the remaining pads Pad 2 through Pad 8 (represented as Pad_x), the ECEs are calculated as follows:
Pad_x_DT Pad_x_GU
Pad_x_ECE = Pad_105_ECE x x equation (2) Pad_l_DT Pad_l_GU
where:
Pad 1 ECE is the Pac 1 CV Target ECE as calculated in equation (1) above;
Pad _I DT is the Distributable Target value for Pad 1 (m3/hr);
Pad_x_DT is the Distributable Target value for Pad x (m3/hr);
Pad 1 GU is the Give Up factor value for Pad 1; and Pad_x_GU is the Give Up factor value for Pad x.
[0083] The DT value used above is the steam target in terms of absolute flow rate. If the DT value is expressed in percent of total flow to the pad, the above equation can be modified to achieve the same behavior.
[0084] The DMC controller (used as the steam header APC controller) is configured to calculate the steady state move limit for each pad steam flow MV based on the "distance" between the sum of all of the pad flows and the target, output from the steam header pressure PID loop. The formula is given by:
ITRGT ¨ FLOWI
ITRGT ¨ FLOWI + E
SSML = (SSMLMAX)(OPPTF) ITRGT ¨ FLOW' + A + (1 OPTTF) N + E
(SSMLMAX)(OPTTF) A = A Aõ 11 equations (3) [ssmimm- (1-0 PTTF) fv-j where:
SSML = Steady State Move Limit;

SSMLMAX = Ma)'( Steady State Move Limit between the Tubing and Casing Max Steady State Move Limits;
SSMLMIN = Min Steady State Move Limit;
OPTTF = Optimization Tuning Factor;
TRGT = Pad External Target;
FLOW = Sum of Well Pair Steam Flows;
k = SS_TARGET_RATIO_FACTOR;
N = Sum of Tubing and Casing Well Pair;
E = Epsilon (to avoid diving by zero); and A = DELTA_TRGT_FLOW MIN.
[0085] The DMC controller (used as the steam header APC controller) is configured to calculate the maximum move ("max move") limit for each pad steam flow MV.
In some implementations, there are four different max move values that each of the MVs can have depending on process conditions:
1. MaxMoveJustOn: If the controller is turned on and an MV is outside of limits it will ramp to its limit at a rate of MaxMoveJustOn.
2. MaxMoveWoundUp: If the controller is turned on and the well pad total steam flow target is far away from the sum of steam flows to the pads then max move is MaxMoveWoundUp. If an MV is outside its limits moving at a rate of MaxMoveJustOn, its max move will remain at MaxMoveJustOn until it enters its own limits 3. MaxMoveTripped: If there is a trip the max move for all MVs go to MaxMoveTripped, regardless of status. A counter begins counting until it reaches its max counter value, then the max move goes to its appropriate value. If another trip happens while the counter is counting, the counter is reset.
4. MaxMoveNormal: For all other scenarios not listed above, the max move is MaxMoveNormal.

[0086] The magnitude of each of the max moves is shown below:
MaxMoveTripped > MaxMoveNormal > MaxMoveWoundUp > MaxMoveJustOn For example, in one implementation, the MaxMoveTripped = 10, MaxMoveNormal =
6, MaxMoveWoundUp = 0.2, and MaxMoveJustOn = 0.1.
[0087] The priority of the max moves is: MaxMoveTripped > MaxMoveJustOn >
MaxMoveWoundUp > MaxMoveNormal [0088] In certain implementations, an independent variable measurement is turned off immediately if it is outside of validity limits or if the measurement stays bad for a certain number of consecutive controller cycles. A dependent variable measurement is turned off if the measurement stays bad for a certain number of consecutive controller cycles.
[0089] For example, in one implementation, a "PADINFO" calculation maps the watchdog counter and why off message to output parameters, a "TRIP COUNTER
PREVIOUS" stores the previous value of the number of units tripped and the number of cycles ago since a trip recently happened; a "BAD COUNTER IND PREV" stores the number of consecutive cycles that an independent variable has a bad status;
and a "BAD COUNTER DEP PREV" stores the number of consecutive cycles that a dependent measurement has a bad status.
[0090] A steady state program, which is a combination linear program (LP) and quadratic program (QP) built into the DMC controller, can be used to "optimize" the pad steam system within constraints. For most DMC controllers, LP costs are assigned to the MVs and/or CVs in order to push the process toward the most economical set of constraints. However, for the steam pad and steam header APC controllers of certain embodiments, the LP/QP is driven by the set of external or operator-entered targets and limits.
.. [0091] The steam header APC controller can be set up with different CV
ranks to handle situations where it is unable to find a feasible solution that satisfies all the CV
limits and targets. In this case, the controller uses ranks to drop out the lower priority variables and recalculate a new solution for the remaining CVs. Different ranks can be configured for the external targets, lower and upper limits.
[0092] Table 4-1 below lists exemplary controlled variables that have external targets defined, for one implementation. The first two columns show the ranks and concerns associated with the lower and upper limits. The next columns are the concerns and ranks for the external targets. Note that the lower the rank, the higher the priority.
Table 4-1 ¨ Steady State (SS) Configuration Parameters for CVs with External Targets Variable SS SS SS Low SS High Lower Lower Upper Low High Concern Concern Target Upper Target Target Rank Rank Rank Target Concern Concern Rank TOTFLOW 20 20 1 1 30 30 0.01 0.01 All Steam Flow 32 32 1 1 35 35 Calculated Calculated CVs (with External Targets) [0093] Tables 5-1 and 5-2 (below) summarize exemplary tuning constraints implemented for the manipulated and controlled variables in one implementation. In Table 5-1 below, the upper, middle and lower dynamic equal concerns for the control calculation determine how aggressively the CVs are driven to their steady-state targets, when near or beyond the upper limit, in between limits or near or beyond the lower limit.
The transition zones are the size, in engineering units, of the region in which the control calculation equal concern increases from its middle value to its value at either limit.
Table 5-1 CV Tuning Parameters CVs Dyn Lower Dyn Middle Dyn Upper Lower Upper Equal (Target) Equal Trans. Zone Trans.
Zone Concern Concern Concern Plant Total 3 0.01 3 1 1 Flow All Steam 500 500 500 30 30 Flow CVs (with External Targets) [0094] In Table 5-2 below, the maximum move parameter defines the largest change (in engineering units) the controller is allowed to make in a manipulated variable setpoint in one cycle. This parameter is normally set as a protection for abnormal situations, i.e. it is set a bit higher than the typical moves that are made in one cycle.
The max move calculation allows the MVs to move further when a trip occurs.
Maximum move can vary for steam flows based on the Max Move Calculation.
[0095] The controller will not calculate a steady-state target that is more than the maximum steady state (SS) step away from the current value. Maximum SS Step below can change between the values shown in Table 5-2 down to 0.1 based on the SS
Move Limit Calculation. This parameter is set to "slow down" the controller when process conditions are changing and a very different steady state solution is required. The steady state solution calculation limits how far MVs can move during a trip and forces them to all move by the same direction.
[0096] The move suppression factor is the primary tuning parameter which affects how aggressively the controller will move a manipulated variable to achieve control objectives. A larger value means more suppression, i.e., less movement.
Table 5-2 MV Tuning Parameters MVs Maximum Move Maximum SS Move Move Step Suppression Suppression Increase Factor 1FC001SP Calculated Calculated 1.2 1000 2FC001SP Calculated Calculated 1.2 1000 3FC001SP Calculated Calculated 1.2 1000 4FC001SP Calculated Calculated 1.2 1000 5FC001SP Calculated Calculated 1.2 1000 6FC001SP Calculated Calculated 1.2 1000 7FC001SP Calculated Calculated 3 1000 8FC001SP Calculated Calculated 1.2 1000 9FC001SP Calculated Calculated 1.2 1000 [0097] To improve the steam header APC controller response time (e.g.
responding to a pad trip), the algorithm that performs the optimization, referred to as Target Algorithm Type (EPSMVPMX), can be adjusted to provide a more robust calculation. In addition, the Move Plan Options parameter (DYNOPT) can be changed (e.g. from 0 to 3), to allow the controller to use time-varying ECEs that vary along the horizon based on the value of the CVs and an advanced move plan that uses quadratic programming to solve the constrained dynamic move plan rather than a least squares algorithm with clipping to meet MV constraints.
[0098] Turning now to second level control (at the pad level), in particular implementations a steam pad APC controller is deployed for each pad, to allocate the optimal amount of steam to the wells within the pad for optimal production. In the implementations described below, the steam pad APC controller is a DMC
controller, such as an Aspen DMCplus TM controller. One steam pad APC controller is implemented for each well pad that is using the APC solution in an SAGD production facility.
[0099] In particular implementations, some objectives of the steam pad APC
controller are as follows:
1. To maintain the total steam used by the pad at a target value. Prior to implementation of the APC controller, the total pad steam target value was determined by the existing distributed control system (DCS) steam allocation code, which has some deficiencies, as previously noted. However, in implementations with the APC solution, the total pad steam target value is instead provided by a supervisory dynamic matrix control (DMC) controller application (the steam header APC controller) responsible for controlling steam header pressures and allocating steam to all of the pads.

2. To allocate steam to the various wells, based on operator entered limits and priorities for each of the wells, regardless of whether there is a need to shed or "dump" steam.
3. To balance individual steam flows with respect to their targets for all wells which have the same priority.
4. To react quickly to maintain the medium pressure steam header (pad level steam header pressure) when the pad splitter or supervisory DMC controller cuts steam to the pad in a OSTG or cogen trip. By way of explanation, the steam distribution process includes the steam header that collects all the steam from the steam generators and is maintained at around 10 MPa in particular implementations.
For the steam that is directed to each pad, the pressure is dropped to approximately 4 MPa by a pressure controller in particular implementations.
This pressure controller is referred to as the medium pressure header.
[00100] In some implementations, each pad steam controller has a subcontroller for each well pair in the pad. Each subcontroller has 2 MVs and 8 CVs. There is a total steam flow CV for the pad. Thus, the model matrix size is 2*(Number of Well Pairs) x 8*(Number of Well Pairs) + 1.
[00101] Table 6-1 below lists the independent variables in a steam pad APC
controller, delivering steam to 18 well pairs, according to one implementation:
Table 6-1 Manipulated Variables Index Tag Name Description Units 1 FC011SP Well 1 Tubing Steam Flow 5m3/h 2 FC012SP Well 1 Casing Steam Flow Sm3/h FC##1SP Well 144 Tubing Steam Flow Sm3/h n+1 FC14#2SP Well #71# Casing Steam Flow Sm3/h 34 FC181SP Well 18 Tubing Steam Flow Sm3/h 35 FC182SP Well 18 Casing Steam Flow Sm3/h ,--[00102] To create intermediate limits and absolute limits, duplicate CVs are created for each well steam injection flow. The operator limits for the steam flow process variable CV are the intermediate limits, while the operator limits for the steam flow duplicate variable CV are the absolute limits. The process variables for the well pair steam injection flows are read from the steam flow measurements on the DCS, while the measurements for the duplicate CVs are set equal to the measurements of the process variable steam flows internally within the steam pad APC controller.
[00103] Table 6-2 below lists the dependent variables in a steam pad APC
controller, delivering steam to 18 well pairs, according to one implementation:
Table 6-2 Controlled Variables Index Tag Name Description Units 1 FC001PV Total Pad Steam Sm3/h 2 FC011PV Well 1 Tubing Steam Flow Sm3/h 3 FC011PD Well 1 Tubing Steam Flow Duplicate Sm3/h 4 FC011OUT Well 1 Tubing Steam Flow Valve Position 5 PC011PV Well 1 Tubing Steam Injection kPag Pressure 6 FC012PV Well 1 Casing Steam Flow Sm3/h 7 FC012PD Well 1 Casing Steam Flow Duplicate Sm3/h 8 FC012OUT Well 1 Casing Steam Flow Valve Position 9 PC012PV Well 1 Casing Steam Injection kPag Pressure FC11-#1PV Well #4 Tubing Steam Flow Sm3/h m+2 FC##1PD Well #4 Tubing Steam Flow Duplicate Se/h m+3 FC#.#10UT Well #41 Tubing Steam Flow Valve Position m+4 PC#41PV Well t144 Tubing Steam Injection kPag Pressure m+5 FC##2PV Well ## Casing Steam Flow Sm3/h m+6 FC#4t2PD Well ## Casing Steam Flow Duplicate Sm3/h m+7 FC##20UT Well ## Casing Steam Flow Valve Position m+8 PC##2PV Well i#t Casing Steam Injection kPag Pressure 138 FC111PV Well 1 Tubing Steam Flow Sm3/h 139 FC111PD Well 1 Tubing Steam Flow Duplicate Sm3/h 140 FC111OUT Well 1 Tubing Steam Flow Valve ok Position 141 FC111PV Well 1 Tubing Steam Injection kPag Pressure 142 FC112PV Well 1 Casing Steam Flow 5m3/h 143 FC112PD Well 1 Casing Steam Flow Duplicate Sm3/h 144 FC112OUT Well 1 Casing Steam Flow Valve Position 145 FC112PV Well 1 Casing Steam Injection kPag Pressure [00104] In the implementation described below, the steam pad APC controller is configured to calculate the capacity of each well pair for both tubing and casing steam flows. The capacity is defined as the amount of injection steam flow that a well pair can take if the well pair is to ride the first constraint which will be encountered. Constraints include, for example, pressure high limit, valve high limit, or steam flow limits. The capacity is calculated by comparing the change from a process variable's current position to its high operator limit, and using the gains and gain multipliers, determining the minimum amount of steam necessary to reach one of these high operator limits. In particular implementations, process variables include, for example, steam flow setpoint, pressure PV (process variable), and valve opening or position of the valve (OP) (for example, a pressure controller can maintain a certain pressure by maintaining the position of the valve at a certain position). If a combined status indicator for steam flow setpoint, pressure PV, or valve OP is flagged to indicate that the controller will reach the high operator limit for any of these variables, the steady state target for the steam flow MV is used as the capacity instead.
[00105] The steam pad APC controller is configured to calculate the total steam flow capacity for the well pad. The total steam flow capacity is the sum of the steam flow capacities for each well pair, both tubing and casing.
[00106] The steam pad APC controller can be configured with a pressure riding or growth mode setting for the individual well steam flows, including both tubing and casing. When the growth mode is turned on (growth = 1) the controller sets the external target (ET) for the steam flow PV equal to the steam capacity plus a capacity bias. The capacity bias, a positive number, is added to ensure steam is pushed all the way to capacity in case the calculated capacity is lower than the amount of steam the controller can actually inject. The high intermediate limit for the steam flow PV is set equal to the ET. If the high absolute limit is lower than the ET, the high absolute limit is also set equal to the ET, otherwise, it keeps its original value. The pressure mode therefore removes the steam flow PV high limits as constraints and forces steam up to open the valve to its high limit or take the pressure to its high limit, whichever constraint is hit first.
[00107] The steam pad APC controller can be configured to accept a high priority or "auto mode" for an individual well steam flow tubing and casing. The high priority flag is turned on (e.g. high_priority = 1) for sensitive wells. The high priority flag takes priority over the backdown order number and means that the high priority well steam will not be given up on. If the external target for total pad steam is greater than or less than the sum of the external targets for tubing and casing steam from individual well pairs, the controller distributes the steam to backdown orders 1 and 2 first. The controller will only move wells marked as high priority after the backdown order 1 and 2 well pairs have reached either their high or low absolute limits. The controller uses ranks for its external targets and operator limits that are lower (higher in priority) than the ranks for back down orders 1 and 2. So the controller will give up first on the amount of steam being requested by backdown order 1 and 2 wells before beginning to give up on high priority wells.
[00108] Wells are assigned to either of the backdown order 1 or backdown order groups. The behavior of the backdown orders is configured by assigning appropriate values to the upper and lower external target ranks as well as the high and low limit .. ranks. When the backdown order flag is set to either 1 or 2 the steam pad APC
controller will set the ranks to achieve the desired behavior.
[00109] The backdown orders can be based on the stability of the reservoir, or, in other embodiments, on the steam to oil ratio (SOR) of each well pair. Backdown order 2 well pairs need to maintain their flow more consistently than backdown order 1 well pairs, which are allowed to vary.
[00110] The steam pad APC controller is configured to give up first on the backdown order 1 well pairs while maintaining backdown order 2 well pairs at their external target.
If the backdown order 1 group reaches either of its intermediate limits then backdown order 2 will give up on its external target. If backdown order 2 reaches either of its intermediate limits it will maintain the value of the limit and backdown order 1 will begin giving up on its intermediate limits until it hits its absolute limits. If backdown order 1 reaches its absolute limits, backdown order 2 will give up until it reaches its absolute limits.
[00111] The steam pad APC controller is configured to give up on the external targets and intermediate limits as an equal percentage of the targets or intermediate limits, respectively, within the same backdown order group. The ECEs for the tubing and casing steam flows for individual well pairs are calculated to result in the desired behavior.
[00112] Since ECEs determine how much they give up based on their ratio, in order to calculate ECEs a reference ECE is needed. In some implementations, the reference ECE is arbitrarily selected to be well pair 1 tubing steam. The ECEs for well pair 1 tubing steam are set equal to 1. All other ECEs are then based off this reference, and have different calculations.
[00113] In particular implementations, the target ECEs for all other tubing and casing steam flows for all other well pairs (including well pair 1 casing steam), are calculated by taking the square root of the target ratio. The ECEs for upper and lower intermediate limits are calculated by taking the square root of the upper intermediate limit ratio (the upper intermediate limit of a well pair steam flow divided by the reference upper intermediate limit of well pair 1 tubing steam) and the square root of the lower intermediate limit ratio (the lower intermediate limit of a well pair steam flow divided by the reference lower intermediate limit of well pair 1 tubing steam), respectively.

r [00114] The steam pad APC controller is configured to set the service request for the duplicate DP steam flow variable equal to the service request for the steam flow process variable PV.
[00115] The steam pad APC controller is configured to give up either on the external target or the intermediate limits depending on where the measurement of the steam flow process variable is located. The percent give up calculation calculates the ratio between the steady state target and the reference location (lower intermediate limit, target, or upper intermediate limit) within the range of measurements. If the steady state target is equal to any of the reference locations, the percent giveup is equal to 1. If the process variable measurement is located between the intermediate limits (but not equal to them), then the reference is the external target. If the measurement is equal to or below the intermediate low limit, the reference is the intermediate low limit. If the measurement is equal to or above the intermediate high limit, the reference is the intermediate high limit.
[00116] A reference flag is also listed to let the user on the DCS know which reference the controller is using to give up from. Within a given backdown order group, all steam injection flows should have the same percent give up value. If a constraint is being hit, the percent give up will be different.
[00117] In some implementations, the steam pad APC controller is configured to calculate a steady state (SS) move limit for each steam flow MV based on the "distance"
between the sum of the tubing and casing flows and the target ("Steady State Move Limit Calculation"). The formula is given by equations (3) set out above. In some implementations, the steam pad APC controller is configured to calculate the move suppression dynamically. Move suppression is provided through a tuning variable that can be modified through the advanced calculations such as equation (3) above.
The challenge in steam distribution is that the sensitivity of the entire system (i.e. quick moves) can lead to oscillation and overpressure situations and asset trips. By modifying the tuning variable, move suppression can dynamically adjust for the complexity and sensitivity of a steam distribution network to reduce or avoid these problems.

r-- CA 3026694 2018-12-05 [00118] In some implementations, the steam pad APC controller is configured to move very slowly (with a small value for the Maximum Move parameter for each steam flow) if the absolute value of difference between the well pad target and the sum of the well pairs is greater than some preselected value, when it is first turned on (the "Just Turned On Status Holder Calculation"). This ensures that if the well pair is wound up it moves slowly as it removes the wind up. When the difference between the well pad target and the sum comes within the preselected value, or the sum switches from being lower than the target to being higher (or vice versa) then the steam pad APC controller will revert to using normal values for the Maximum Move.
[00119] The steam pad APC controller can be set up with different CV ranks to handle situations where it is unable to find a feasible solution that satisfies all the CV limits and targets. In this case, the controller uses ranks to drop out the lower priority variables and recalculate a new solution for the remaining CVs. Different ranks have been configured for the external targets, lower and upper limits.
[00120] Table 7-1 below lists the controlled variables that have external targets defined. The first two columns show the ranks and concerns associated with the lower and upper limits. The next columns are the concerns and ranks for the external targets.
Note that the lower the rank, the higher the priority.
Table 7-1 ¨ Steady State Configuration Parameters for CVs with External Targets Variable SS SS SS Low SS High Lower Upper Lower Upper Low High Concern Concern Target Target Target Target Rank Rank Rank Rank Concern Concern FC001PV 15 15 0.1 1.0 40 40 0.01 0.01 All Steam Flow Calcu Calc Calcu- Calcu- Calcu- Calcu- Calcu-Calcu-CVs (with -lated u- lated lated lated lated lated lated External lated Targets) [00121] The calculated ranks listed in Table 7-2 below are based on backdown order and high priority status.

Table 7-2 ¨ Calculated ranks Variable/rank High Priority Backdown Order 2 Backdown Order Duplicate (DP) SS 20 25 30 High Rank SS High Rank 45 50 55 Upper Target Rank 45 60 65 Lower Target Rank 45 60 65 SS Low Rank 45 50 55 Duplicate (DP) SS 20 25 30 Low Rank [00122] The steady state configuration parameters for the CVs that do not have external targets are listed in Table 7-3 below. Note that these limits tend to become active constraints only occasionally. However, when they do, they can take priority over the external targets summarized in the previous section (since the ranks are generally lower).
Table 7-3 ¨ Steady State Configuration Parameters for CVs without External Targets Variable SS Low Rank SS High Rank SS Low Concern SS High Concern FC1:401OUT 35 35 0.01 0.01 PC19#01PV 9999 10 1000000 5 FC#402OUT 35 35 0.01 0..01 PC1t1/02PV 9999 10 1000000 5 [00123] Tables 8-1 and 8-2 below summarize some of the more important tuning constants which can be implemented for the manipulated and controlled variables.
[00124] For the controlled variables (CV), the upper, middle and lower dynamic equal concerns for the control calculation determine how aggressively the CVs are driven to their steady-state targets, when near or beyond the upper limit, in between limits or near or beyond the lower limit. The transition zones are the size, in engineering units, of the , region in which the control calculation equal concern increases from its middle value to its value at either limit. The tuning parameters for CVs of a steam pad APC
controller are shown in Table 8-1 below for one implementation.
Table 8-1 CV Tuning Parameters CVs Dyn Lower Dyn Dyn Upper Lower Upper Equal Middle (Equal Trans. Trans.
Concern (Target) concern) Zone Zone Concern FC001PV 3 0.025 1 10 FC011PV 0.15 0.2 0.15 0.5 0.5 FC011PD 0.15 0.2 0.15 0.5 0.5 FC011OUT 0.5 0.5 0.1 0 3 FC012PV 0.15 0.2 0.15 0.5 0.5 FC012PD 0.15 0.2 0.15 0.5 0.5 FC012OUT 1 1 0.1 0 3 FC##1PV 0.15 0.2 0.15 0.5 0.5 FC1/#1PD 0.15 0.2 0.15 0.5 0.5 FC##10UT 0.5 0.5 0.1 0 3 PC##1PV 1000000 1000000 20 0 FC4E2PV 0.15 0.2 0.15 0.5 0.5 FC#42PD 0.15 0.2 0.15 0.5 0.5 FC##20UT 1 1 0.1 0 3 PC#42PV 1000000 1000000 10 0 FC111PV 0.15 0.2 0.15 0.5 0.5 FC111PD 0.15 0.2 0.15 0.5 0.5 FC111OUT 0.5 0.5 0.1 0 3 FC112PV 0.15 0.2 0.15 0.5 0.5 FC112PD 0.15 0.2 0.15 0.5 0.5 FC112OUT 1 1 0.1 0 3 [00125] Manipulated variables (MV) tuning parameters are shown in Table 8-2 below for one implementation. The maximum move parameter defines the largest change (in engineering units) the controller is allowed to make in a manipulated variable setpoint in one cycle. This parameter is normally set as a protection for abnormal situations (i.e. it is set a bit higher than the typical moves that are made in one cycle).
Maximum Move can go to 0.1 for both tubing and casing steam flows based on the Just Turned On Status Holder Calculation. Maximum SS Step below can change between the values shown in Table 8-2 down to 0.1 based on the Steady State Move Limit Calculation.
[00126] The controller will not calculate a steady-state target that is more than the maximum SS step away from the current value. This parameter is set to "slow down"
the controller when process conditions are changing and a very different steady state solution is required.
[00127] The move suppression factor is the primary tuning parameter which affects how aggressively the controller will move manipulated variables to achieve control objectives. A larger value means more suppression, i.e., less movement. In some implementations, the steam pad APC controller is configured to calculate the move suppression dynamically.
Table 8-2 MV Tuning Parameters MVs Maximum* Maximum SS Move Move Move Step Suppression Suppression Increase Factor FC011SP 1 2 0.03 2000 FC012SP 2 10 0.03 2000 FCtictt1SP 1 2 0.03 2000 FC##2SP 2 10 0.03 2000 FC111SP 1 2 0.03 2000 FC112SP 2 10 0.03 2000 [00128] For certain implementations, targets can be automatically adjusted based at least in part on total production and steam ratio. Some of such implementations use Total Fluid to Steam Ratio (TFSR) logic, for example. TFSR is a metric used by reservoir engineering to determine the performance of the reservoir for example, and connects production with steam injection. Automatic adjustment of targets allows the targets to be applied to full field (thereby automating the producer and injector full field).
One objective is to automate as many wells as possible to achieve prompt change in steam supply according to production change. This will have the benefit of allocating steam for better use (e.g. when the Electric Submersible Pump (ESP) is down, steam can be allocated quickly to other wells for better use).
[00129] The steam targets for wells can be determined based on various optimization purposes, including, without limitation, target steam chamber pressure, target performance indices (e.g., steam-oil-ratio for energy efficiency, total fluid to steam ratio for balanced fluid optimization, etc.), priority to younger and ramp-up wells to promote steam chamber development, and edge well management to confine heat injected and target the steam injection to pads with infills to support base well pairs and infills. Tools used to support the steam target determination can include performance analysis tools (e.g., trend analysis based on production history) and reservoir simulation.
Performance analysis tools can be used to set steam targets based on data recorded during a period in which performance was good. Reservoir simulations can be used when operating strategy has changed and there is no prior history to support prediction.
Sensitivity runs are generally required to find the optimal solution at the pad or field level.
After the steam targets are determined, an automated steam process controller (APC) is put in place to ensure steam supply trends toward the set steam targets despite the continuous steam supply swings from the steam generators.
[00130] Some implementations allow certain wells to be turned ON or OFF. The reservoir production and engineering personnel can determine if a problematic well should be turned on or off or is too sensitive to the change. If a well is targeting steam chamber pressure, the well can have the TFSR option turned off and have the pressure riding function turned on to target a certain injection pressure.
[00131] The APC solution described herein can reduce GHG emissions in terms of both absolute value (for the first stage of the APC solution) and intensity per unit of production (for the second stage of the APC solution). For the first stage, the APC
solution is applied to reduce the variance in steam quality produced among different tubes within an OTSG or cogeneration unit. This removes the constraint imposed by the tube with the highest steam quality. The overall steam quality can therefore be increased to transfer more energy into the steam. As a result, less natural gas is required to produce the same amount of steam, and steam generation costs and GHG
emissions are reduced.
[00132] For the second stage, the plant level and pad level APC solutions optimize the delivery of steam to the wells depending on the process plant criteria, and predict and avoid constraints for each steam pad or well. In particular implementations, the APC
solution predicts the reservoir condition up to 48 hours in advance through continuous data analysis with a black-box (empirical) modelling technique. The APC
solutions prioritize the steam pads and wells based on the reservoir characteristics and engineer input. By controlling and optimizing the reservoir condition (e.g. sub-cool, bore level) for each well while maintaining process safety limits, bitumen extraction is increased.
[00133] The plant level and pad level APC solutions do not reduce absolute GHG
emissions directly per se, as the same amount of steam is generally consumed during unit operations and thus generates the same amount of GHG emissions. However, for a given amount of steam, more bitumen production can be achieved with the assistance of plant level and pad level APC solutions that distribute the steam more efficiently and manage reservoir conditions. As such, less GHG is emitted per barrel of bitumen produced.
[00134] In a particular trial of the APC solution for controlling the distribution of steam, analysis of operational data of 60 days prior to and following the implementation of the solution showed that the steam injection APC solution was able to distribute steam to higher priority pads (i.e. new pads and pads with lower steam to oil ratio (SOR)). On average, 3150 m3/d of steam was determined to be redistributed to higher producing pads. Using the pad level SOR to convert the steam into bitumen production, an estimated 1500 bbl/d production increase could be achieved with a more optimal steam allocation. This resulted in a reduction of GHG emission intensity per barrel produced.

, [00135] The reduction in GHG emission intensity can be converted into GHG
emission avoidance. The GHG emission avoidance benefit is based on a decrease of energy intensity if there are no constraints preventing the plant from processing all the extra bitumen produced as a result of the application of the APC solution.
Increasing production will not be possible if an active constraint already exists in the plant (for example, the plant is at full capacity). Alternatively, as a result of the APC
solution, the steam production rate can be lowered (less steam used) to achieve the same bitumen production rates if the plant is constrained. In this scenario the GHG
emission can be avoided.
[00136] For the purpose of calculating GHG emission avoidance, an assumption can be made that there are no constraints limiting the increase of bitumen production. The avoidance of GHG emissions is then calculated as the amount of extra GHG
emissions required to achieve the same amount of bitumen production increase without APC

solutions. The same percentage of GHG increase is needed to obtain the percentage of bitumen production increase if the GHG intensity remains the same had APC
solutions not been implemented. In summary, total GHG emissions can be reduced and avoided by applying the APC solutions (first stage and second stage solutions).
[00137] Other benefits of the APC solutions include the long-term reduction in water and chemical consumption. Using the APC solutions described herein, more bitumen can be produced with the same amount of steam, therefore reducing the water consumption per barrel of bitumen production. The reduction in water consumption leads to a reduction in chemical usage, as chemicals are needed to treat and remove impurities from water recycled from reservoirs before being used again for steam generation. Reduction in chemical usage in turn reduces the risk of chemical exposure and improves safety.
[00138] In general, some other advantages of the APC solutions include:
= avoidance or reduction of steam long and steam short scenarios by facilitating optimized steam distribution to the wells;

= more accurate steam injection rates in relation to the engineers' analysis, resulting in improved SOR;
= improved decision-making capability due to predictions of the steam system capacity; and = reduced manual intervention, allowing workers to redirect their efforts to more value-added work (e.g. the control room operator is able to focus on more urgent plant upsets rather than managing the large number of wells).
[00139] The methods and systems described above can be applied to improve the operation of a steam distribution network, such as the steam distribution network 20 that is shown in Figure 2, which delivers steam generated by an OTSG or cogeneration unit to a plurality of hydrocarbon-containing reservoirs such as oil sand reservoirs. However, in other implementations, these methods and systems can be applied to improve operation in other types of production facilities, such as for other supply/demand balancing networks or networks of distribution or production (e.g. steam turbine network for electricity generation, gas distribution network, gas production network, or water supply network).
[00140] In the above description, for purposes of explanation, numerous details are set forth in order to provide a thorough understanding of the present disclosure.
However, it will be apparent to one skilled in the art that these specific details are not required in order to practice the present disclosure. Although certain dimensions and materials are described for implementing the disclosed example implementations, other suitable dimensions and/or materials can be used within the scope of this disclosure.
All such modifications and variations, including all suitable current and future changes in technology, are believed to be within the sphere and scope of the present disclosure. All references mentioned are hereby incorporated by reference in their entirety.

Claims (32)

1. A system for controlling a steam-assisted gravity drainage (SAGD) production facility, wherein the facility comprises a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads comprising one or more well pairs, the system comprising:
a steam header controller adapted to determine a first set of outputs for controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to each one of the steam pads; and for each one of the steam pads, a corresponding steam pad controller adapted to determine a second set of outputs for controlling the distribution of steam received at the steam pad via the steam header, to the one or more well pairs within the steam pad, wherein each of the steam header controller and each one of the steam pad controllers is implemented as an advanced process control (APC) controller, wherein the APC controllers receive target values and a prioritization of the steam pads and well pairs, and allocate steam to the steam pads and the well pairs based at least in part on their prioritization and relative to their respective target values while maintaining process constraints.

2. The system of claim 1 wherein the steam header controller is adapted to maintain the pressure of the steam header at a target steam header pressure value.

3. The system of claim 2 wherein the target steam header pressure value is based at least in part on a total steam flow for the plant.

4. The system of any one of claims 1 to 3 wherein the steam header controller is adapted to adjust the allocation of steam to the steam pads in response to a trip detected in the plant or in one of the steam pads.

5. The system of any one of claims 1 to 4 wherein for each one of the steam pads, the corresponding steam pad controller is adapted to maintain the total steam used by the steam pad at a pad steam target value.

6. The system of any one of claims 1 to 5 wherein for each one of the steam pads, the corresponding steam pad controller is adapted to balance steam flows relative to their target for each of the well pairs having the same priority.

7. The system of any one of claims 1 to 6 comprising an interface module for connecting a distributed control system of the SAGD production facility to one or more of the APC controllers.

8. The system of claim 7 wherein the interface module receives outputs from the one or more of the connected APC controllers and performs one or more of data conditioning, data processing and further optimization of the outputs, to produce APC
outputs suitable for execution by the distributed control system.

9. The system of any one of claims 1 to 8 wherein the steam header controller is configured to allocate steam to the steam pads based at least in part on a give up factor assigned to each one of the steam pads, wherein a high priority steam pad is assigned a small give up factor so that steam allocation to the high priority steam pad does not significantly deviate from its respective target value.

10. The system of any one of claims 1 to 9 wherein for each one of the steam pads, the corresponding steam pad controller is configured to calculate a capacity of each well pair for both tubing and casing steam flows based on a constraint comprising one or more of: pressure high limit, valve high limit, and steam flow limit.

11. The system of any one of claims 1 to 10 wherein for each one of the steam pads, the well pairs within the steam pad are ranked, and the corresponding steam pad controller is configured to give up on the well pairs in order of their ranking if a steam flow limit is reached during operation of the well pairs.

12. The system of any one of claims to 1 to 11 wherein each one of the APC
controllers comprises a dynamic matrix control (DMC) multivariable predictive controller.

13. The system of any one of claims 1 to 12 wherein the plant comprises a plurality of flow passes and the system further comprises a steam quality controller adapted to control and optimize steam quality in each of the flow passes while maintaining process constraints.

14. The system of claim 13 wherein the steam quality controller is implemented as a model predictive control (MPC) controller which is configured to predict in advance a trajectory for one or more process variables to enable proactive control of the steam quality.

15. A system for controlling the operation of a steam-assisted gravity drainage (SAGD) production facility, wherein the facility comprises a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads comprising one or more well pairs, the system comprising:
an advanced process control (APC) controller adapted for determining one or more outputs for controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to each one of the steam pads, wherein in determining the one or more outputs the APC controller receives target values and a prioritization of the steam pads, and allocates steam to the steam pads based at least in part on their prioritization and relative to their respective target values while maintaining process constraints; and an interface module for connecting a control system of the SAGD production facility to the APC controller, wherein the interface module delivers the one or more outputs of the APC controller to the control system for execution by the control system to control the distribution of steam.

16. The system of claim 15 wherein the APC controller is configured to allocate steam to the steam pads based at least in part on a give up factor assigned to each one of the steam pads, wherein a high priority steam pad is assigned a small give up factor so that steam allocation to the high priority steam pad does not significantly deviate from its respective target value.

17. A method for controlling a steam-assisted gravity drainage (SAGD) production facility, wherein the facility comprises a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads comprising one or more well pairs, the method comprising:
controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to each one of the steam pads; and for each one of the steam pads, controlling the distribution of steam received at the steam pad via the steam header, to the one or more well pairs within the steam pad;
wherein controlling the distribution of steam comprises receiving target values and a prioritization of the steam pads and well pairs and applying advanced processing control (APC) techniques to allocate steam to the steam pads and the well pairs based at least in part on their prioritization and relative to their respective target values while maintaining process constraints.

18. The method of claim 17 comprising maintaining the pressure of the steam header at a target steam header pressure value.

19. The method of claim 18 comprising determining the target steam header pressure value based at least in part on a total steam flow for the plant.

20. The method of any one of claims 17 to 19 comprising adjusting the allocation of steam to the steam pads in response to a trip detected in the plant or in one of the steam pads.

21. The method of any one of claims 17 to 20 comprising, for each one of the steam pads, maintaining the total steam used by the steam pad at a pad steam target value.

22. The method of any one of claims 17 to 21 comprising, for each one of the steam pads, balancing the steam flows relative to their target for each of the well pairs having the same priority.

23. The method of any one of claims 17 to 22 comprising providing a server for implementing the APC techniques, and integrating the server with the SAGD
production facility by connecting the server with a distributed control system of the SAGD
production facility.

24. The method of claim 23 comprising receiving outputs from the server, and performing one or more of data conditioning, data processing and further optimization of the outputs, to produce APC outputs suitable for execution by the distributed control system.

25. The method of any one of claims 17 to 24 comprising allocating steam to the steam pads based at least in part on a give up factor assigned to each one of the steam pads, wherein a high priority steam pad is assigned a small give up factor so that steam allocation to the high priority steam pad does not significantly deviate from its respective target value.

26. The method of any one of claims 17 to 25 comprising, for each one of the steam pads, calculating a capacity of each well pair for both tubing and casing steam flows based on a constraint comprising one or more of: pressure high limit, valve high limit, and steam flow limit.

27. The method of any one of claims 17 to 26 comprising, for each one of the steam pads, ranking the well pairs within the pad and giving up on the well pairs in order of their ranking if a steam flow limit is reached during operation of the well pairs.

28. The method of any one of claims to 17 to 27 comprising providing a dynamic matrix control (DMC) multivariable predictive controller to implement the APC
techniques.

29. The method of any one of claims 17 to 28 wherein the plant comprises a plurality of flow passes and the method further comprises controlling and optimizing steam quality in each of the flow passes while maintaining process constraints.

30. The method of claim 29 comprising applying model predictive control (MPC) techniques to predict in advance a trajectory for one or more process variables to enable proactive control of the steam quality.

31. An advanced process control (APC) server for controlling the operation of a steam-assisted gravity drainage (SAGD) production facility, wherein the facility comprises a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads comprising one or more well pairs, the server comprising a model predictive control (MPC) controller configured to determine one or more outputs for controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to each one of the steam pads, wherein in determining the one or more outputs the MPC controller receives target values and a prioritization of the steam pads, and allocates steam to the steam pads based at least in part on their prioritization and relative to their respective target values while maintaining process constraints.

32. An advanced process control (APC) server for controlling a steam-assisted gravity drainage (SAGD) production facility, wherein the facility comprises a plant operable to produce steam for delivery to a plurality of steam pads, each one of the steam pads comprising one or more well pairs, the server comprising:
a steam header controller configured to determine a first set of outputs for controlling the distribution of steam from the plant to the steam pads via a steam header operatively coupling the plant to each one of the steam pads, wherein the steam header controller receives a target steam header pressure value and a prioritization of the steam pads, and determines a pad steam target value for each one of the steam pads for an optimal allocation of steam to the steam pads based at least in part on their prioritization while maintaining the steam header pressure at the target steam header pressure value;
for each one of the steam pads, a corresponding steam pad controller communicatively coupled to the steam header controller and configured to receive the corresponding pad steam target value from the steam header controller and a ranking of the well pairs, and, based at least in part on the corresponding pad steam target value and the ranking of the well pairs, determine a second set of outputs for controlling the distribution of steam received at the steam pad via the steam header, to the one or more well pairs within the steam pad, while maintaining the steam pad pressure at the pad steam target value; and an interface module adapted to communicatively connect a distributed control system of the SAGD production facility with the steam header controller and steam pad controllers, wherein the interface module is configured to receive outputs from the steam header controller and the steam pad controllers and perform one or more of data conditioning, data processing and further optimization of the outputs, to produce advanced process control outputs suitable for execution by the distributed control system to control the distribution of steam throughout the SAGD production facility, wherein each of the steam header controller and steam pad controllers is implemented as a dynamic matrix control (DMC) multivariable predictive controller.