Method for planning and optimising thermal stores in industrial installations for producing cold and / or hot water

Abstract

The invention relates to a method for planning and optimising thermal stores in industrial installations for producing cold and / or hot water, which comprises two separate steps based on respective predictive algorithms: a step of planning, which proposes use strategies to the operator by means of predictions based on the analysis of the different actors present in the installation and provides the user with a Gantt chart with thermal store usage forecasts for a time period; and a step of optimising, which uses an algorithm based on artificial intelligence in order to optimise the charge and discharge cycles of thermal accumulators.

Description

[0001] DESCRIPTION

[0002] Method for planning and optimizing thermal storage in industrial facilities producing cold and / or hot water

[0003] Technical sector

[0004] The invention falls within the field of control or regulation systems in general; functional elements of such systems; monitoring or testing arrangements for such systems or elements; and energy supply, distribution, or storage systems.

[0005] More specifically, the present invention relates to a predictive method for the operability of thermal storage in industrial facilities with high-demand cold or heat production requirements. This method is intended to improve energy efficiency and optimize the production of industrial facilities that have this type of storage.

[0006] The predictive method is based on mathematical algorithms that draw on data: on the one hand, this data is external to the industrial facility, such as the energy tariff, the nature of the energy sources, or the instructions issued by the electricity supplier; and, on the other hand, data extracted from the facility itself, collected during the commissioning and operational qualification phase, such as sensors, probes, consumption or production demand, etc.

[0007] These algorithms employ advanced mathematical models that, through data analysis and pattern identification, propose specific usage (planning) and operational improvement (optimization) strategies for more efficient use of thermal storage.

[0008] State of the art

[0009] In recent years, industry has been seeking innovative and green solutions aimed at energy savings, enabling them to reduce their electricity bills while simultaneously reducing their carbon footprint. The immediate future forecast includes rising energy supply costs and the implementation of new carbon-related taxes. TES (Thermal Energy Storage) technology stores thermal energy by heating or cooling a medium, allowing the stored energy to be subsequently used for cooling / heating applications or in industrial processes where the generation of cold or heat is essential for the industrial process itself. These TES systems are used in both service buildings and industry. Examples of this type of product include:

[0010] - Cold storage: CALMAC, FAFCO, CRISTOPIA.

[0011] - Heat storage: THERMAL BATTERY.

[0012] This thermal storage allows the owner to implement different types of strategies to reduce the energy costs derived from its productive activity, such as:

[0013] - Shift the production of cold or heat from hours of high production demand, which generally coincide with high-rate hours, to hours of low-rate hours.

[0014] - Absorb peaks in renewable energy production. Electricity production from renewable energy sources, such as photovoltaic or wind, is inconsistent and out of sync with the operational needs of industry. This results in peaks in electricity production that cannot be used for production processes and are wasted.

[0015] - Respond to demands issued by the grid distributor. In the context of a smart grid, thermal storage allows, for example, to discharge energy upon request from the grid distributor in the event of a grid overload. This way, the facility consumes less energy from the grid and contributes to grid stability, thus qualifying for rewards offered by the distributor.

[0016] Industrial automation is the control technique used in industry, primarily based on platforms with Programmable Logic Controllers (PLCs). These PLCs are characterized by implementing fixed logic, which means their sequences and operations are established and programmed in advance. This ensures both the safety and proper functioning of industrial equipment. The advantage of this integration lies in its reliability and predictability, as it allows specific scenarios to be addressed and ensures that processes are executed consistently, minimizing risks and optimizing efficiency and safety in industrial operations. Currently, the industrial sector is undergoing significant change driven by the adoption of emerging technologies such as the Internet of Things (IoT), analytics, Big Data, and Artificial Intelligence, among others.This transformation is known as the fourth industrial revolution or Industry 4.0.

[0017] The incorporation of these technologies into industrial automation is paving the way for new, more flexible and adaptable approaches that could transform the traditional industrial control paradigm into smarter, more connected solutions.

[0018] In the energy sector, there are many products that offer models for operational analysis and prediction aimed at energy efficiency and grid stability. Leading companies in the industrial sector offer applications based on predictive models for the management of smart grids, always focused on "smart grids" or "microgrids." These are essentially decentralized electrical networks composed of small and diverse energy sources that operate in parallel or autonomously with respect to the main grid. The goal is to obtain a more reliable and highly efficient electricity supply that can provide higher quality of service, while also making it safe and sustainable in both rural and urban areas.

[0019] Currently, in the industrial sector, to install this type of technology, the interested entity must purchase the cold or hot container and contract engineering services from an automatic control systems integrator.

[0020] The control and regulation of these systems is carried out during the commissioning phase, where the adjustments to the control loops and parameterization are performed. This allows the operator to have a functional thermal storage device for the operation of the facility.

[0021] Once the facility is operational, the maintenance team routinely makes periodic adjustments to the system's regulation based on user experience and analysis of collected data to improve loading and unloading processes and test new operating strategies.

[0022] The nature of these fixed PLC logics restricts the ability to adapt to changing situations and improve performance, leading to missed opportunities for energy efficiency and continuous system optimization. Industrial thermal storage facilities lack specific prediction algorithms that propose operating strategies and optimize production processes.

[0023] GB2604953 describes a computer-implemented method for predictively controlling a water supply system installed in a building, the water supply system comprising a heat pump configured to transfer thermal energy from outside the building to a thermal energy storage medium within the building and a control module configured to control operation of the heat pump, the water supply system being configured to provide water heated by the thermal energy storage medium to an occupant of the building at one or more water outlets.The method comprises: receiving first sensor data indicating cold water usage at a first water outlet; determining whether the cold water usage at the first water outlet is correlated with a subsequent hot water demand based on the first sensor data; and upon determining that the cold water usage at the first water outlet is correlated with a subsequent hot water demand, preparing the water supply system to supply hot water.

[0024] Description of the invention

[0025] The objective of the application is to offer a predictive method for the operability of thermal storage systems that guarantees their most efficient use, ensuring the following functions:

[0026] 1. Predictive usage planning based on the different stakeholders involved, such as energy rates, availability of renewable energy sources, production demand, weather data, etc.

[0027] 2. Automatic optimization of the cold or heat production process: maximizing loading and unloading cycles and thus avoiding operational inefficiencies (underutilization or over-stressing of the system).

[0028] Through the implementation of advanced techniques, the method enables proactive management of the energy contained within the thermal storage facility, maximizing loading and unloading cycles and proposing usage strategies based on the different actors involved. This allows the operator to optimize their operations, increase productivity, and reduce costs. To achieve the proposed objectives, the invention proposes a method for planning and optimizing thermal storage facilities in industrial facilities producing cold and / or hot water, having the characteristics of claim 1.

[0029] The method object of this invention is divided into two independent phases based on predictive algorithms that perform:

[0030] 1. A planning phase: This phase proposes usage strategies to the operator through predictions based on the analysis of the different actors present in the facility.

[0031] 2. An optimization phase: This phase uses an artificial intelligence-based algorithm to optimize the charge and discharge cycles of the thermal accumulators.

[0032] Preliminary considerations: Industrial control systems are generally based on platforms composed of PLCs and SCADA. These SCADA systems are inherently data acquisition (DAQ) systems and generally have a database to store the status and measurements of variables read from equipment and field instrumentation. This history system is assumed to exist in the facility. Otherwise, a data acquisition system must be integrated beforehand.

[0033] The algorithms of the two phases that make up this method are executed on a standard processor of an industrial PC.

[0034] The planning phase provides the user with a Gantt chart with forecasted thermal storage utilization for a predefined time interval (the time interval is defined based on the nature of the process). To run this planner, the algorithm requires the following input data:

[0035] - Hourly pricing, from the network distributor(s).

[0036] - Reception of instructions issued by the electrical distributor.

[0037] - Disaggregation of the nature of available energy sources.

[0038] - Consumption prediction, based on daily history or production campaigns.

[0039] - The weather data for the location.

[0040] - Operator restrictions. The algorithm is executed at the operator's request. The steps followed in this phase for the correct execution of the algorithm are: a) Acquisition of electricity pricing data for the time interval established by the process. This information is acquired directly from the distributor's website via the Internet. b) Acquisition of the energy mix composition. If the electricity grid is the only available energy source, the data is retrieved solely from the energy distributor's website. In the case of a more complex energy management system, with different sources available, the grid data is supplemented with data from the energy management system (EMS). c) Acquisition of distributor setpoints. Depending on the technology implemented, these setpoints can be acquired from the Internet or from the electricity grid connection itself (smart meters).d) Acquisition of meteorological data. This information is acquired directly from the website through the meteorological services of the facility's location. e) Consumption prediction. Consumption prediction is based on historical data from the facility found in the operating system's database. Depending on the facility, the demand for cooling or heating may follow a daily pattern or over a period of time, depending on the production campaign. f) Evaluation of restrictions introduced by the operator. The operator has a web tool available through a scheduler that allows them to force shutdowns or use storage for maintenance or other reasons. g) Execution of the planning algorithm. h) Deployment of the planning. This display is shown to the operator through a web browser.

[0041] Once the planning phase is executed, it is re-executed periodically at a frequency set by the operator. This means that the steps described above are re-executed sequentially, providing the operator with an updated plan at all times.

[0042] The planning phase can also be updated autonomously if it detects a change in the input parameters. This is because it constantly monitors the input data for changes: red lines entered by the operator, distributor demands, etc. If a change in the input data is detected, the planning phase executes the algorithm to obtain the new schedule.

[0043] Formulating a model involves representing a real-world problem using mathematical expressions that accurately and precisely describe the situation to be modeled. The planning phase uses mathematical modeling focused on optimizing the use of the thermal storage.

[0044] These types of models are composed of:

[0045] Decision variables: Decision variables are identified as the elements over which control is exercised and on which a decision is made.

[0046] Objective function: The objective function (FO) is a mathematical expression that represents the objective of maximizing warehouse utilization.

[0047] Constraints: Constraints are mathematical expressions whose function is to limit the values ​​that decision variables can take.

[0048] The formulation that allows for resource planning in our case is presented below.

[0049] • Parameters: oi: ranging from 1 to m, where m is the number of existing constraints. oj: ranging from 1 to n, where n is the number of existing variables. o bi: Resource availability. This can be a limitation or requirement. oa¡,: Technical coefficients. o Q: Coefficients of the variables in the objective function.

[0050] • Decision variables: o Xf. Decision variables.

[0051] • Objective Function: o Z: Value of the objective function.

[0052] • Restrictions:

[0053] The optimization phase includes two algorithms that attempt to optimize the optimal values ​​for maximum efficiency in the unloading and loading cycles of the thermal storage unit. The following results are obtained:

[0054] The loading algorithm.

[0055] The download algorithm.

[0056] The two algorithms in the optimization phase implement a multiple linear regression model along with an ordinary squares method to reduce the cost function (error between predicted and actual values).

[0057] The multiple linear regression model is an algorithm from the family of machine learning algorithms. Machine learning is a branch of artificial intelligence that enables machines to learn autonomously, without requiring specific programming. This model is the one that best suits our need for resource optimization.

[0058] The optimization phase attempts to find the relationships between each configuration parameter and the machine's efficiency.

[0059] The operational implementation of these machine learning algorithms is designed in two phases: a. The training phase. b. The operational phase.

[0060] For the training phase, a dataset containing the values ​​of the sensors, probes, and industrial equipment involved in the processes is required. This data is obtained from the facility's operating system, using data collected during the commissioning and operational qualification phase. This data is stored in the acquisition system's database. The initial model is built from this historical data.

[0061] The advantage of modeling these charge and discharge cycles is that it allows for a relationship between the quality of the charge or discharge and each of the input variables. The following table describes the input parameters and the quality index measured for each cycle:

[0062]

[0063] * kWf / c : kilowatt of cold

[0064] Secondly, in the operational phase, the model is put into operation. From this point on, the model generates an assessment of the current loading and / or unloading cycle, offering the operator parameter variations to optimize the process.

[0065] Once the cycle has concluded, the model uses this new data to make a new contribution to the model.

[0066] The algorithm is executed at the operator's request. The steps followed in this phase for the correct execution of the algorithm are: a) Detection of the start of the charge or discharge cycle. This signal is recovered from the control and / or monitoring system through an OPC UA communication interface. The industrial OT protocol used may vary depending on the installation's connectivity. b) Acquisition of the input values ​​into the model at runtime. This is done through the same communication interface as in step 1. c) Execution of the model. The execution of the model evaluates the values ​​of the current charge or discharge cycle and provides the operator with an efficiency prediction for the process, showing the deviation of the current parameters from the most efficient model. d) Once the cycle is completed, the model integrates the cycle values ​​to complete itself automatically. The advantages of the method of the present invention include the following:

[0067] Focus on specific industrial facilities: This method is specialized for a particular type of industrial facility that currently does not have advanced optimization mechanisms.

[0068] Precision and customization: The algorithm is designed to adapt and optimize more precisely to the specific needs and characteristics of each industrial facility based on the instrumentation and number of devices involved.

[0069] Real-time efficiency: The algorithm is capable of providing real-time results and predictions, allowing operators to take immediate corrective action.

[0070] Scalability: This tool allows you to connect to a higher-level energy management system, thus increasing the number of strategies you can implement regarding the use of thermal storage.

[0071] Security and confidentiality: This tool does not depend on cloud services, ensuring the confidentiality and protection of customer information. Cost management: The tool allows you to manage the financial savings achieved, as well as quantify the CO2 not consumed and therefore determine the return on investment.

Claims

CLAIMS 1.- A method for planning and optimizing thermal storage in industrial facilities for the production of cold and / or hot water, comprising two independent phases based on predictive algorithms that perform: a) a first resource planning phase that uses mathematical modeling to propose usage strategies to the operator through predictions based on the analysis of the different actors present in the facility and offers the user a graph with the forecasts of use of the thermal storage for a period of time; in which it receives at least the following input data: o hourly pricing, from the network distributors, o reception of instructions issued by the electricity distributor, o disaggregation of the nature of the available energy sources, o consumption prediction, based on daily history or based on production campaigns, o meteorological data, o operator restrictions;and b) a second optimization phase that uses two independent algorithms to optimize the values in the discharge and charge cycles of the thermal store, based on artificial intelligence, which are executed at the operator's request, in which said algorithms implement a multiple linear regression model together with an ordinary squares method, trying to find the existing relationships between each configuration parameter and the system efficiency; in which the planning phase is executed according to the following sequence of steps: a) acquisition of the electricity pricing data for the predefined time interval, acquired from the distributor's website through the Internet; b) acquisition of the composition of the energy mix, from the data obtained from the energy distributor's own website, or with the network data completed with those of the energy management system;c) acquisition of distributor instructions, acquired online or from the electricity grid connection itself (smart meters); d) acquisition of meteorological data, acquired online through meteorological services; e) consumption prediction based on historical data from the facility stored in the operating system's database, according to a daily pattern or a time period depending on the production campaign; f) evaluation of operator-introduced constraints, which may force the storage to be shut down or used for maintenance or other reasons; g) execution of the scheduling algorithm; and h) deployment of the schedule and displaying it to the operator via a web browser. 2.- Method according to claim 1, in which the planning phase is executed periodically at a frequency set by the operator, or autonomously if a change in the input parameters is detected, for which it is permanently monitored that there are no changes in the input data that exceed certain red lines entered by the operator or demands of the distributor. 3.- Method, according to any of the preceding claims, in which the planning phase uses a mathematical modelling focused on the optimization of the use of the thermal warehouse, which is composed of: decision variables that are identified as the elements over which control is had and on which a decision is made, objective function, or mathematical expression that represents the objective of maximizing the use of the warehouse, and the restrictions (mathematical expressions) that limit the values that the decision variables can take. 4.- Method, according to claim 1, in which the operational implementation of the machine learning algorithms used in the optimization phase are designed in two phases, training and operational, where the training phase is carried out with a set of data of values of the sensors, probes and industrial equipment that intervene in the processes, obtained from the operating system available in the facility with the data collected during the commissioning and operational qualification phase. 5.- Method, according to claims 1 and / or 4, in which the input parameters to determine a quality index or ratio between the electrical consumption (kW) per kilowatt of cold / heat (kW f / c ) stored, for each charging cycle, are the following: charging time, temperature of entry to the warehouse of the cold / heat producing equipment, return temperature to the cold / heat producing equipment, Inlet flow rate from the cold / heat generating equipment, outlet temperature of the thermal storage unit, thermal storage load level. Pumping speed, number of cold / heat producers, thermal storage load setpoint. 6.- Method, according to claims 1 and / or 4-5, in which the input parameters to determine quality indices or ratios between the electrical consumption (kW) per kilowatt of cold / heat (kWf / c ) released, and between the kWf / c released per kWf / c required for each discharge cycle, are the following: discharge time, inlet temperature of the cold / hot industrial hydraulic circuit, return temperature of the cold / hot industrial hydraulic circuit, outlet flow rate of the thermal store, outlet temperature of the cold store, load level of the thermal store, pumping speed, percentage of opening of the gate valve, power setpoint to be delivered from the thermal store.

7. A method according to claims 1 and / or 4-6, wherein the optimization algorithm is executed upon operator request, in the following steps: a) detection of the start of a loading or unloading cycle, the signal of which is retrieved from the control and / or monitoring system via a communication interface; b) acquisition of the input values into the model at runtime via said communication interface; c) execution of the model to evaluate the values of the current loading or unloading cycle and provide the operator with an efficiency prediction for the process, showing the deviation of the current parameters from the most efficient model; and once the cycle is completed, d) integration of the cycle values for self-completion.

Images