Automatic planning method for the transportation of petroleum products in the petroleum product transportation pipeline network

Abstract

This invention is dedicated to providing a method for automatic planning and optimization of oil product transportation in oil product pipeline networks. Using a color state space model expressed as a linear variable model with parameters and economic predictive control, this method enables real-time management of the sequence of injection of packages. The proposed structure defines an economic objective function, taking into account pumping costs, product interference, and tracking product delivery targets in specific time intervals, which is iteratively optimized over a sliding forecast horizon. The proposed model, using online data, is capable of identifying potential disturbances such as equipment failure or changes in demand, and ensures stable system performance by quickly revising the schedule. The proposed method also avoids the computational complexity of conventional methods and enables real-time implementation with optimization algorithms such as branch and bound. This invention has applications in industrial systems of petroleum product transmission lines and reduces costs, increases productivity, and improves operational flexibility.

Description

Description of the invention Title of the invention Automatic planning method for the transfer of petroleum products in the pipeline network for the transfer of petroleum products Technical background of the relevant invention In sequential injection petroleum product pipelines, the main challenge is scheduling and route selection in complex and dynamic conditions with sudden disturbances that increase costs. The proposed approach, by modeling the color state space in the LPV framework and using online data, continuously updates control decisions. The use of sliding horizon predictive control, while reducing computational complexity, makes the network resistant to disturbances and allows rapid adaptation to new conditions. In addition, its economic predictive nature reduces mixing losses, optimizes energy costs, improves productivity, and accurately tracks product delivery times. Technical problem and stating the objectives of the invention In petroleum product pipeline networks where products are transported in sequential and batched forms, one of the main challenges is to optimally schedule batch injections, determine flow routing, and plan pumping in a situation where the network has a complex structure, multiple refineries and destinations, and various operational constraints. This problem becomes more complicated when unforeseen events such as equipment failures, demand fluctuations, or sudden restrictions on routes occur continuously in the network. Conventional methods for scheduling are often based on solving a large and complex combinatorial problem, which not only requires a lot of computational time, but also is not efficient in real-world situations where real-time decision making is required. The purpose of this invention is to design an intelligent and practically applicable method for automatic scheduling and control of transmission operations. This method, using the LPV color state space model and the economic predictive control algorithm, is designed in such a way that by repeatedly solving optimization problems with limited dimensions in a short time interval (sliding horizon), it can produce updated decisions at any time and appropriate for new conditions. In this method, only the first decision among the optimization results is executed and then the problem is solved again when new information is received. This structure ensures that the system always remains flexible to environmental changes and possible disruptions, and optimizes network performance by considering economic criteria, including reducing product interference losses, reducing pumping costs, and adhering to delivery schedules. A description of the state of the prior art and the history of developments related to the claimed invention. Planning the transportation of petroleum products in pipelines is one of the complex and multifaceted problems in the field of energy systems engineering. This process faces challenges such as determining the order of product injection, delivery scheduling, selection of transportation routes, managing the capacity of lines and warehouses, and controlling product quality at delivery points. In addition, the occurrence of sudden disturbances such as equipment failures, demand changes, and blockages of routes makes traditional planning unstable, delayed, and costly. Existing models are often static, unable to adapt to instantaneous changes, and computationally heavy. In the research of Rubens Rejowski Jr. and Pinto (2003) in the article "Scheduling of a multiproduct pipeline system" published in the scientific journal Computers & Chemical Engineering, a mixed integer linear programming (MILP) model was introduced to reduce waste in simple networks with a Y-shaped structure. However, this method lacks the ability to deal with complex topologies and sudden disturbances in the network. In the research of Cafaro and Cerda (2008) in the article "Dynamic scheduling of multiproduct pipelines with multiple delivery due dates" in the journal Computers & Chemical Engineering, dynamic scheduling of oil products is presented considering product delivery dates. Despite considering delivery dates, this model has a static structure and does not have an online and disturbance-resistant decision-making mechanism. In the research of Mirhassani and Ghorbanalizadeh (2008) in the article "The multi-product pipeline scheduling system" in the journal Computers & Mathematics with Applications, the MILP model was presented for scheduling segmented pipelines. This model is suitable for long-term horizons, but it does not have the necessary dynamics against instantaneous changes and unexpected conditions. Subramanian, Maravlias, and Rawlings (2012) in their paper "A state-space model for chemical production scheduling" published in the journal Computers & Chemical Engineering, examined the use of a state-space model in the context of chemical scheduling considering disturbances. Although this work is close to the application of MPC, it lacks the economic structure and sliding horizon for optimal dynamic decision making. In the research of Bueno et al. (2020) in the article "Assigning and sequencing batches and blends of oil derivatives in a mesh-like pipeline network" in the journal Computers & Chemical Engineering, a heuristic approach for scheduling in real mesh networks is presented. Due to its heuristic nature, this method does not guarantee to reach the optimal solution and is not able to use real-time data to revise decisions. The present invention, by utilizing the LPV state space color model and the EMPC sliding horizon structure, covers these gaps and provides a scalable, real-time, and fault-tolerant solution for scheduling the transportation of petroleum products in complex pipeline networks. Providing a solution to an existing technical problem along with an accurate, sufficient, and integrated description of the invention To overcome the problems raised in the previous section, this invention presents an integrated solution based on EMPC economic predictive control and Colored LPV State Space Model. The proposed structure is designed to intelligently respond to environmental changes and disturbances while reducing computational complexity, aiming at real-time management of the petroleum product pipeline network. In this method, the current state of the network is measured using online data. Then, an updated state model with color features (to distinguish between different products) and time-varying parameters simulates the dynamic behavior of the network over a specified forecast interval. Based on this prediction, at each time step, a finite-dimensional optimization problem (including injection sequence, active paths, pumping rate, and compliance with constraints) is defined in a forecast horizon. This problem is solved with respect to economic performance indicators (including pumping cost, interproduct waste, and compliance with delivery times), and only the first decision is implemented. Then, in the next step, with new data, the process is repeated. This sliding horizon structure keeps the system in an optimal state at all times. By combining three key elements—accurate dynamic modeling, economic optimization, and real-time updating—the present invention provides an innovative framework for automated and responsive scheduling in response to disturbances and changes in complex petroleum product transmission networks. Next, we examine how to model the transmission network to solve the finite-horizon transmission scheduling problem. Common assumptions in petroleum product transmission line networks The main objective is to accurately schedule the transfer of oil product packages from refineries to terminals, so that all demands on different network routes are met, in accordance with product delivery dates and based on an economic performance index. At the same time, the actual system conditions such as production capacity, existing routes, previous schedules, and the occurrence of possible disruptions are also taken into account. The proposed structure is designed based on the following conventional features: Network topology, demands, delivery dates, production constraints, initial state of packages in the lines, current operational state of the network, and sequences sent in previous time intervals are all considered as known data at the beginning of the planning process. Each segment of the pipeline contains, at any given moment, a certain volume of an incompressible petroleum product. By injecting a new package at a constant pumping rate, the contents of the current segment are transferred to the next segment in the same path. This process provides the basis for dividing time into discrete, equal-sized steps. At each time step, only one input packet is injected into the network from one of the refineries and only one output packet is directed to one of the tanks. In other words, there is no simultaneous injection from multiple refineries or simultaneous delivery to multiple destinations, which is common in many industrial systems to simplify operations and prevent product interference. The volumetric flow density in each branch (Polyduct) of the transmission line - which contains multiple sections - is considered to be the same and uniform, so that at each time step, the packets located in the sections of each branch in the active path are continuously transferred to the next section. It is possible to use more than one route to transport product from each refinery to a tank, but creating multiple routes simultaneously is usually costly and requires more extensive infrastructure. The amount of interference losses depends on the type of consecutive petroleum products; in such a way that in some sequences, unwanted or undesirable chemical mixtures may be produced. Therefore, choosing the appropriate transmission sequence plays an important role in optimizing network performance. Transmission network predictive model using graphs and colored state space A petroleum product transportation network typically consists of several refineries that produce different types of petroleum products. These products are transported to final storage tanks through branch pipelines. These network components work together to provide a continuous flow of transportation from refineries to consumption destinations. Figure 1, in the drawings file, shows an example of a petroleum product pipeline network with two refineries and two storage tanks. In the following, graph theory is used to mathematically model the system to simplify the physical concepts. Pipeline networks can be represented as simple directed graphs in the form of a waterfall system. A directed graph is an ordered pair consisting of a set of nodes (or vertices) and a set of edges (or arcs). In this graph, the incoming degree of each node is equal to the number of edges that end in it; that is, and the outgoing degree of each node is equal to the number of edges that leave it; that is, . The adjacency matrix is ​​a square matrix where the element indicates the existence of an edge from node to node. In this waterfall graph, four types of nodes are defined: Nodes with zero incoming degree are known as sources. Nodes with zero outgoing degree are defined as depots. Nodes with an input degree of one are known as successive nodes, and nodes with an input degree greater than one are considered receiving nodes. In modeling pipelines with simple directed graphs, each part of the network is considered as a node, while the connections between them are represented by edges.The binary weight of each edge indicates whether the pumping operation at that connection is active or inactive during transmission. Figure 2, available in the maps file, shows the directed graph of a sample transmission pipeline network with four different types of nodes to model the cascade structure. Since only one type of petroleum product is present in each section of the pipeline at any given time, the concept of a color graph is used to represent the different petroleum products. In this method, the color of each node indicates the type of petroleum product present in that section. For each color (i.e., each petroleum product), a separate subgraph is constructed and these subgraphs are analyzed in parallel. Finally, by combining all the color subgraphs, the main network graph is formed. Figure 3, available in the maps file, shows the main color graph and the subgraphs related to each petroleum product in the sample transmission line network. The LPV state space model is a linearized model whose matrices depend on a set of measurable and time-varying parameters; therefore, some nonlinear behavior of the system can be represented as a quasi-linear model with variable parameters. In this invention, an LPV model in the form of a state space is used, whose variable parameters depend on the decision variables. The developed model uses the concept of color states to specify the type of petroleum product present in each section of the network. In this model, the type of product present in the j-th section of the network is represented by a binary vector of one, with one of its elements always equal to one. Thus, at any instant of time, each section contains only one type of petroleum product. Similarly, it represents the amount of petroleum products in the n-th tank.A single binary decision vector with length equal to the total number of possible states (including all possible product-path-destination combinations), denoted by , specifies the product transfer path at each time interval t; each component of this vector indicates whether a specific product from a given refinery has been injected to a given destination. Given , the active or inactive pumping state for each path branch at time interval t is determined by a set of binary vectors that define the transfer operation at the discrete time interval t. To complete the mathematical formulation of the state-space color model, it is first necessary to define the discrete transfer between the sections. Equation (1) represents the initial injection from the refinery to the source sections. (1) where represents the presence of the product in the source section at time. represents the sum of all components of the decision variable vector that represent the transfer of petroleum product type p from a refinery to each of the reservoirs, through source section j. also represents the pumping status in the branch path that includes source section j. The set of all petroleum products is the set of all sections of the transmission line network and the set of source sections in the transmission line network. Equation (2) represents the transfer of the product from one section to the next in the intermediate sections. (2) where is the input section and the output section in the petroleum product transmission network. The pumping state in the branch containing the input section and is the set of intermediate sections in the transmission line network. Equation (3) expresses the product transmission for the receiving sections to which several branches end. (3) Here, the last segment of the branch is connected to the receiving node and its corresponding pumping status is indicated by the symbol . It also represents the set of all receiving segments. Finally, equation (4) describes the delivery of the oil product package from the last segment connected to the reservoir. (4) Given that the decision vector is responsible for directing petroleum products from a refinery to a reservoir through the active path, Figure 4, available in the Maps and Tables file, shows the one-way transfer of products through the active path for different types of topologies of sections in the pipeline network, according to relations (1) to (4), so that the active path is displayed in blue. By introducing the state matrix and output matrix for each petroleum product p, equation (5) provides an accurate expression of the dynamic behavior of the petroleum product transportation network based on the colored state space model with variable parameters (Colored LPV). (5) In this equation, the state matrix depends on the decision variables. The sparse matrices are constant and binary matrices that represent the process of injecting product type p into the pipeline network. Finally, the constant matrix acts as the output matrix and specifies the final inventory of petroleum products stored in the tanks. For each petroleum product type p, equation (6) describes the prediction structure of the pipeline network behavior in terms of colored state variables and outputs, over the forecast horizon NP​. (6) This prediction method, using the initial state of the system and control inputs, allows for the prediction of the future behavior of the network, making the presented model suitable for use in predictive control (MPC). To convert this model into a matrix form, first the color state vector, outputs, and decision vectors are defined along the forecast horizon. Then, with the help of equation (6), these predicted states can be expressed in matrix form as equation (7). (7) In this regard, the expanded matrices and are defined in Equation (8) and Equation (9), respectively, and is a fixed block diagonal matrix. (8) (9) By using the matrix structure presented in equation (7) as the proposed Colored LPV-MPC model, a suitable platform can be provided for planning the transmission of petroleum products in the transmission line network, so that this planning is both economically optimal and can be stable against possible performance disruptions. Automatic planning of transmission line networks using the proposed Colored LPV-MPC model While tracking the delivery of petroleum products on specific dates in the transmission network constitutes the essential part of the objective function, other economic sectors strongly emphasize the reduction of the total material interference costs and pumping costs that extend over the forecast horizon. Equation (10) refers to the tracking objectives, material interference costs, and pumping costs, respectively. (10) where is the added reference path of the system that expresses the required quantities on the delivery dates of petroleum products. In addition, denotes the weight norm . denotes the unordered binomial sets of __, which are used to define the interference combination set between each pair of petroleum products. The transpose represents the occurrence matrix of the transmission network graph that removes the edges corresponding to the tank nodes and consists of (S-1) edges to express each consecutive pair of sections and (S+D) nodes as sections and tanks. The edges between tanks and the end sections of the corresponding transmission network are ignored because there is no interference cost during the delivery of materials to the tanks. Finally, the coefficients and constant matrices , and \[{{W}_{u}}\in {{\mathbb{Z}}^{a}}\] are used to balance between the objectives over the forecast horizon. Predictive control-based methods effectively manage constraints in industrial processes and help optimize performance due to their ability to predict and adjust system dynamics. Equation (11) as the injection constraint restricts the decision variable vector to a single binary vector that expresses the movement of a product along a specific path from a refinery to a reservoir at each time interval. (11) Equation (12) applies production constraints over the forecast horizon, taking into account previous time intervals. In fact, each constraint is defined to be considered over the entire forecast horizon, with the caveat that past sequences are considered as fixed values. (12) In this regard, the time-varying parameter expresses the production limit of product type p at refinery r, which can change due to network disruptions and changing supply and demand conditions. To reduce product interference losses, the optimization model seeks to inject several consecutive batches of the same type of petroleum product, but long repetition of this process in refineries may not be technically or economically feasible. For this reason, a limit is considered in equation (13) for the maximum number of similar batches that can be injected consecutively. (13) In this regard, the parameters express the maximum allowed number of consecutive injections of each type of petroleum product p from refinery r. Some consecutive combinations of petroleum products can cause high interference losses. To avoid this situation, the constraint defined in equation (14) specifies which pairs of products should not be placed in adjacent sections of the pipelines, as an unordered pair set. (14) Also, to maintain the model's feasibility after the end of the forecast horizon, a terminal constraint is defined. This constraint uses a fixed parameter to control the maximum allowable difference between the desired inventory in the warehouses and the planned inventory at the end of the forecast horizon. This recursive feasibility means that by creating a constraint at the end of the forecast horizon, the feasibility of the system can be ensured in the future. This method allows the system to continue to work continuously after the end of the forecast horizon without the need to review the conditions. Using the proposed Colored LPV-MPC model, an online structure can be created that calculates the transmission line network states in the forecast horizon by considering the economic performance index and considering the constraints on product injection. The proposed structure is described as equation (15). (15) In this regard, , and define constraint sets for states, outputs, and decision vectors, such that represent the decision vectors allowed to define sequences in the form of batches of petroleum products injected from refineries to specified routes into reservoirs. The constraint function g applies the production constraints based on equations (11) to (13) in the forecast horizon Np. Given the initial conditions in the network, the objectives, and the constraints, the finite-horizon automatic scheduling problem of the transmission network is solved recursively using appropriate solvers such as branch-and-bound or other heuristic methods, and only the first element of the optimized sequences, i.e., is applied to the transmission network. Then, the updated states are used as new initial states to solve the problem in the next time interval. Figure 5, available in the Figures and Tables file, shows the pseudocode for the proposed algorithm for automatic transmission line network planning. Validation of the automatic transmission line network planning method on a case network To verify the validity of the proposed automatic transmission network planning method, transmission planning was performed for the sample transmission network shown in Figure 1, available in the Maps and Tables file. Table 1 shows the state of the decision vector for the transmission of petroleum products from refineries to the reservoirs of the target transmission network. Relations (16) to (20) express the state of activity (pumping of petroleum products) in the branches containing the transmission sections of the sample network based on the selected sequence _shown in Table 1_. (16) (17) (18) (19) (20) Solving the transmission scheduling problem requires having initial information. In this regard, Table 2, available in the Maps and Tables file, states the initial conditions of the network sections, including the type of material and the pumping status. Table 3, available in the Maps and Tables file, states the previous sequences sent to the network. Table 4, available in the Maps and Tables file, states the demand for various petroleum products for delivery to the tanks in the required time intervals. Table 5 shows the vector of coefficients used in the economic objective function. Meanwhile, to limit the production capacity of refineries in equation (12), the variable parameter for each petroleum product p in the refineries at time t is considered equal to the difference between the existing demand of all tanks at the end of the forecast horizon and the quantities available in the tanks for each product p. To solve this automatic scheduling problem of petroleum products transportation based on the Colored LPV_MPC model and the economic objective function in the presence of constraints, the branch and bound algorithm was used so that the online optimization process in equation (15) solves the constrained problems step by step and based on a moving forecast horizon of 10 and time intervals of 6 hours. To illustrate how this process is implemented, Figure 6 shows the calculated forecast horizon and the first optimal sequence applied to the sample transmission line network. Unlike classical MPC, where the goal is to reach a desired steady state from an initial state and the value of the objective function gradually decreases, in the economic predictive control model there may not be a steady state that the system reaches; therefore, decreasing the objective function does not necessarily mean improving economic performance. The main components of the economic objective function are shown in Figure 7, available in the Maps and Tables file. As can be seen, the part related to tracking the product delivery date targets has an oscillating behavior due to demand changes over the forecast horizon. To reduce the overall cost of product interference, the optimization model selects the injection sequence in such a way that, as much as possible, consecutive packages in the polyducts are of the same product type. The weighting vector represents the difference in the production cost of different products. For this reason, the results obtained from the optimization prevent the injection of the third product in refinery R1 and the first and second products in refinery R2 due to the higher cost. Finally, Figure 8, available in the Maps and Tables file, shows the output status in terms of tank inventory after performing the online optimization step by step. In the event of disturbances such as sudden changes in demand, equipment failure, or route blockage, the previously planned optimal sequence will no longer be valid and may severely reduce network efficiency. However, in the proposed online planning approach, the network status is updated in real time at each stage, and by re-running the optimization process based on new information, it is possible to quickly adapt to new conditions and restore system efficiency. This feature is a key advantage of the proposed method in dealing with unstable and unpredictable conditions. To examine the performance of the algorithm in the presence of disturbances, Table 6, available in the Maps and Tables file, presents different disturbance conditions in the transmission network. In this scenario, the dynamic model and the initial conditions of the existing network are calculated at each time interval according to the new conditions, and then the planning problem is solved. Figure 9, available in the Maps and Tables file, shows the output status as the final inventory of the reservoirs after running the iterative online optimization in the presence of disturbances. Explanation of shapes, maps and diagrams Figure 1 shows a sample network of pipelines for transporting petroleum products. Figure 2 shows a directed graph of a sample transmission pipeline network with four different types of nodes to model the cascade structure. Figure 3 shows the main colored graph and subgraphs for each petroleum product in the sample transmission line network. Figure 4 shows the one-way transfer of products through the active path for different types of topologies of sections in the transmission pipeline network: (a) source sections, (b) intermediate sections, (c) receiver sections, (d) reservoir sections. Figure 5 shows the pseudocode of the proposed automatic scheduling algorithm for the transmission line network. Figure 6 shows the calculated forecast horizon and the first optimal sequence applied to the sample transmission line network. Figure 7 shows the main components of the economic objective function after solving the automatic scheduling problem for the sample transmission line network. Figure 8 shows the state of the outputs in the form of reservoir inventory after solving the automatic scheduling problem for the sample transmission line network. Figure 9 shows the state of the outputs in terms of reservoir inventory after solving the automatic scheduling problem for the sample transmission line network in the presence of disturbances. Table 1 shows the decision vector state for transporting materials from refineries to the tanks of the desired transmission network. Table 2 presents the initial conditions of the network sections, including the type of material and pumping status. Table 3 presents the previous sequences sent to the sample transmission line network. Table 4 shows the demand for various petroleum products for delivery to reservoirs in the required time intervals. Table 5 presents the coefficient vector used in the economic objective function to solve the sample automatic transmission line network planning problem. Table 6 describes the different disruption conditions in the transmission network. A clear and precise statement of the advantages of the claimed invention over prior inventions. The present invention, by utilizing the combined structure of the color state space model with variable parameters and the economic predictive control strategy, provides an innovative approach for the optimal scheduling of petroleum product transportation in pipeline networks with complex and multi-source topologies. Compared with previous studies and methods, this invention has the following key advantages: The use of a color state space model with location-variable parameters provides accurate modeling of the behavior of transmission networks with diverse and multi-source-multi-storage structures, which cannot be easily achieved in classical MILP or MINLP methods. In the objective function, instead of focusing solely on tracking costs, a balanced combination of two other economic indicators, including product interference costs and pumping costs, is considered, which provides a more realistic coverage of network operation objectives. Considering product delivery dates in different time periods provides better accuracy and coordination between injection, storage, and demand over the time horizon, whereas many previous models only consider a single final due date. The automatic scheduling process, based on the proposed model, is executed iteratively and in real time, and at each time step, it generates a new injection sequence by solving a constrained optimization problem. This leads to increased accuracy and flexibility in responding to changes in demand and network status. Continuous updating of the network status using online measurements allows for rapid detection of disturbances such as equipment failure or changes in the reference path, enabling the system to respond dynamically without the need for human intervention. Instead of solving a complex macro-programming problem, this method reduces the computational burden and enables practical implementation in real time by employing a sliding forecast horizon and MPC structure. The presented structure is designed to be solvable both with exact algorithms such as branch and bound and with heuristic methods; a matter that increases the efficiency and flexibility of the model in different operational conditions. Description of at least one implementation method for implementing the invention The proposed implementation method for implementing this invention is based on an automatic and real-time programming algorithm that is designed using a time-varying color state space model and an economic predictive control strategy and is executed step by step in each time interval. The general process is as follows: Initially, a color state space model is defined according to the forecast structure for a network with multiple refineries and multiple destinations. Then, the forecast horizon is specified and the reference trajectory is set based on the recorded demands on the delivery dates during the forecast period. At each time step of the system execution, the network data including previously executed sequences and the current state of the pipelines are first updated. At this stage, the system also checks the network status so that if disturbances such as equipment failure or changes in demand patterns have occurred, the reference path is modified and the decision-making allowed range is adjusted according to the new conditions. Then, the proposed model solves a combinatorial optimization problem by considering all defined constraints (such as production constraints, sequential injection, interference considerations, and capacities). Finally, only the first optimized decision from this sequence is applied to the network, and the state of the inventory and model variables are updated to repeat the process at the next time step. This structure allows for rapid response to changes and automatic execution of decisions in real time by continuously moving the forecast horizon forward. This implementation method can be implemented in industrial systems and, using online measurement equipment and SCADA systems, can be used on a real scale for dynamic and optimal management of petroleum product transmission lines. Explicit mention of the industrial application of the invention The present invention is directly applicable to the oil, gas and petrochemical industry, and in particular in the field of management and planning of the transfer of petroleum products through transmission lines. This system can be used in the transmission infrastructure of refineries, multi-branch networks, and distribution and storage centers. The industrial application of this invention includes the following: Real-time planning of product transportation in complex transmission line networks with multiple refineries and multiple reservoirs; Reduce operating costs by reducing interference costs, optimizing injection sequences, and controlling pumping costs; Increase operational reliability by taking into account possible disruptions such as equipment failures, demand changes or maintenance operations; Increasing efficiency in the distribution of petroleum products using advanced decision-making algorithms based on economic predictive control and state space color models;  Ability to be implemented in industrial control systems such as SCADA, DCS, and refinery automation systems, with the ability to connect to real-time measurement and monitoring equipment. This invention is a completely practical and industry-wide solution for intelligently optimizing the timing of the transfer of petroleum products through transmission pipelines, creating a significant competitive advantage for operating companies and energy supply chain management.

Claims

Claim What is claimed: Claim No. 1) A method for automatically scheduling the transfer of petroleum products in petroleum product transmission lines, including the following steps: a) Modeling the network structure using a colored state space model with linearly variable parameters (Colored LPV State Space) in order to describe the dynamics of the transmission network, b) Defining the forecast horizon and demand values ​​in the time intervals required for delivering products to the tanks, c) Updating the network operational data in each time interval, including existing constraints, tank inventory status, type of products available in different parts of the network and identifying possible disruptions, d) Solving a constrained hybrid optimization problem to determine the optimal sequence of the type and route of products sent to the transmission network during the forecast horizon, e) Applying the first decision resulting from the optimized sequence to the network and re-executing the process in the next time interval with a new forecast horizon, such that this method aims to reduce the economic objective function including reducing the cost of tracking the delivery of products Oil, reducing waste interference between products and reducing network pumping costs are designed and implemented. Claim No. 2) According to claim No. 1, the color state space model for each type of petroleum product is defined as an independent vector, such that at any time, each part of the network contains only one type of product. Claim No. 3) According to claim No. 1, the economic objective function includes the weighted sum of the cost of tracking the delivery of petroleum products, the cost of interfering with petroleum products in network segments, and the pumping costs over the forecast horizon. Claim No. 4) According to claim No. 1, constraints are considered, including refinery production capacity, the maximum number of consecutive refinery injections allowed, and avoiding mixing incompatible products in close proximity to each other. Claim No. 5) According to claim No. 1, the optimization problem is solved in each time interval for a prediction horizon and only the first optimized decision from the calculated sequence is applied to the network, such that in each subsequent time interval, the network data is updated and the optimization problem is solved again. Claim No. 6) An automatic planning and control system for transporting petroleum products in petroleum product transmission network lines, comprising: a) a modeling unit for representing the network structure using a colored state space model with linearly variable parameters (Colored LPV State Space), b) a control unit for estimating and defining demand values ​​in the time intervals required for delivering products to tanks and determining the forecast horizon based on user needs, c) an optimization unit for solving a combined problem with the aim of reducing the costs of the economic objective function including product interference, pumping costs, and tracking the delivery of petroleum products, d) an execution unit for applying the optimized decision to the injection or pumping system, e) an interface for receiving real-time data from the network including tank inventory, line status, and detecting possible disturbances, so that the system can receive new information in each time interval, calculate the optimal injection sequence, and execute the first decision. Claim No. 7) According to claim No. 6, the planning system is connected to an industrial supervisory control and automation (SCADA) system and receives information on the current status of the network, including pressure, flow, product type, and inventory at various points in the transmission line network.

Images