Method for controlling emissions of a chemical reaction

Abstract

The present invention is in the field of controlling emissions of chemical reactions. The present invention relates to a computer-implemented method for controlling emissions of a chemical reaction, the method comprising: (a) receiving sensor data relating to the chemical reaction in a first reactor, (b) determining, based on the sensor data, operational instructions relating to the treatment of emissions in the second reactor of at least one product obtained by the chemical reaction, and (c) outputting the operational instructions.

Description

Technical Field

[0001] The present invention belongs to the field of controlling emissions from chemical reactions. The present invention relates to a method for controlling emissions from chemical reactions, an emissions control system, a non-transitory computer-readable data medium, and the use of operating instructions for controlling emissions from chemical reactions. Background Art

[0002] Chemical reactions typically produce unwanted but almost unavoidable by-products. Such by-products are usually harmful to the environment, for example, because they are toxic, cause acid rain, or promote the growth of algae in water bodies. To avoid or at least reduce the harmful effects on the environment, the by-products of chemical reactions are usually treated, for example, by converting them into harmless substances in subsequent reactions and then releasing them into the environment. A typical example is the exhaust gas treatment of combustion engines. Traditionally, exhaust gases are measured to control exhaust gas treatment. However, this feedback loop can be slow and thus inefficient. Prediction models have been proposed in order to react before any unwanted emissions occur.

[0003] WO 2020 / 176 914A1 discloses a method for predicting exhaust gases using an artificial neural network. Exhaust gas flow parameters are measured and fed into the neural network to control exhaust gas treatment. However, this method usually cannot react quickly enough to changes, and thus the reaction throughput is limited. Summary of the Invention

[0004] The object of the present invention is to provide a method for emissions control of chemical reactions that can reliably control emissions in order to maximize the reaction throughput. The method aims to maintain flexibility, so it can be applied to various different reactions. In addition, the method should also be able to cope with sudden changes in chemical reactions.

[0005] In a first aspect, these objects are achieved by a computer-implemented method for controlling emissions from a chemical reaction, the method comprising:

[0006] (a) receiving sensor data related to the chemical reaction in a first reactor,

[0007] (b) determining, based on the sensor data, operating instructions related to emissions treatment of at least one product obtained by the chemical reaction in a second reactor, and

[0008] (c) outputting the operating instructions.

[0009] According to a second aspect, the present invention relates to a non-transitory computer-readable data medium storing a computer program, the computer program comprising instructions for performing the steps of the method according to the present invention.

[0010] According to a third aspect, the present invention relates to the use of operating instructions obtained by the method according to the present invention for controlling the emissions of a chemical reaction.

[0011] According to a fourth aspect, the present invention relates to an emissions control system comprising:

[0012] (a) an input for receiving sensor data related to a chemical reaction in a first reactor,

[0013] (b) a processor for determining, based on the sensor data, operating instructions related to emissions treatment of at least one product obtained by the chemical reaction in a second reactor, and

[0014] (c) an output for outputting the operating instructions.

[0015] The present invention enables better control of the emissions of a chemical reaction. This is partly due to the fact that the control is based on sensor data related to the chemical reaction in the first reactor, whereas in conventional methods, the concentration of by-products has first to be measured in order to determine any measures for emissions control. Thus, the measures cannot be determined before the by-products are actually formed. In contrast, the present invention enables the measures to be determined earlier. This is because data related to the chemical reaction in the first reactor can be used to predict changes in the chemical reaction in the near future. Thus, a reaction to emissions control can be made more quickly, i.e., the emissions control can be adjusted before any increase in emissions due to an increase in the concentration of by-products. For example, the concentration of a by-product that should not be released into the environment above a certain level can be predicted, and thus measures such as increasing the dose of treatment reagent in the second reactor can be taken before the emissions increase. In this way, the emissions can be controlled more reliably, i.e., any overshoot of the emissions level can be reduced. Additionally, in some cases, the chemical reaction can be carried out more efficiently without increasing the risk of overshooting the emissions level. For example, the chemical reaction can be pushed towards its limit operation, i.e., a lower safety margin is applied with respect to emissions limits. Moreover, due to the fact that the emissions control can react more quickly, the throughput of the chemical reaction can also be increased, and thus any time delays in conventional methods, such as the measurement of the by-product level, can be avoided. Additionally, the treatment reaction can be carried out more efficiently, for example, because the treatment reaction only requires an appropriate amount of treatment reagent, energy or time to treat the by-products. In this way, treatment reagent, energy or time is not wasted.

[0016] As used herein, the term "emissions" generally refers to substances or mixtures of substances that have a harmful impact on the environment when released into the environment. Emissions include air pollutants, water pollutants, or soil pollutants. Air pollutants include: toxic gases and / or acidic gases, such as CO, NO x , SO x , O3; volatile organic compounds, including aliphatic hydrocarbons (such as methane) or aromatic hydrocarbons (such as benzene); toxic metals, such as lead or mercury; chlorofluorocarbons; dust particles, such as fine particulate matter. Water pollutants include: chemicals from insecticides, herbicides, fungicides; petroleum hydrocarbons, including fuels; lubricants, such as engine oil; fuel combustion by-products, including partially oxidized hydrocarbons, such as alcohols, aldehydes, ketones, or carboxylic acids; volatile organic compounds, such as industrial solvents; persistent organic pollutants, such as perfluoroalkyl and polyfluoroalkyl substances; organic chlorides, such as polychlorinated biphenyls (PCB), trichloroethylene, oxidants (such as perchlorate); acids, such as hydrofluoric acid, hydrochloric acid, sulfuric acid, nitric acid; nutrients, such as nitrates or phosphates; plastic particles, such as microplastics. Soil pollutants include most of the above water pollutants. Generally, many substances can be either air pollutants or water pollutants, or either air pollutants or soil pollutants, or either water pollutants or soil pollutants, or either air pollutants or water pollutants and soil pollutants.

[0017] As used herein, the term "control" generally refers to any action that may affect emissions, particularly actions that reduce the harmful impact of emissions on the environment. The action can be direct, for example, by changing the state of a valve to adjust the flow rate of a treatment reagent, or by changing the temperature by additional heating or increased cooling. These actions can also be indirect, for example, by prompting an operator to take an action - for example, replacing a filter or adjusting throughput. Controlling emissions can mean taking actions to keep the concentration of at least one pollutant below a predetermined threshold. In some cases, control can also mean taking actions to keep the concentration of more than one pollutant below a threshold, where different thresholds are typically applied to different pollutants. Generally, the pollutants are by-products of chemical reactions in a first reactor.

[0018] As used herein, "chemical reaction" generally refers to a process involving the transformation of one set of chemical substances into another. Chemical reactions can occur in homogeneous or heterogeneous ways. Homogeneous chemical reactions involve one phase, for example, the gas phase or the liquid phase, such as a solution. Heterogeneous chemical reactions involve at least two phases. The at least two phases can have different states of matter, for example, one phase is a solid phase while the other phase is a liquid phase, or one phase is a solid phase while the other phase is a gas phase, or one phase is a liquid phase while the other phase is a gas phase. If the at least two phases are immiscible, they can have the same state of matter, such as two immiscible liquid phases or two immiscible solid phases.

[0019] Chemical reactions can occur in a continuous or discontinuous manner, sometimes also referred to as batch chemical reactions. In a continuous chemical reaction, reagents are continuously fed into a reactor where the reaction occurs, while products are continuously output from the reactor. In a discontinuous chemical reaction, reagents are injected into the reactor, then the reaction occurs, and afterwards the products are collected from the reactor. The reactor can be cleaned and then new reagents can be injected into it again.

[0020] As used herein, the term "reactor" generally refers to a container that houses a chemical reaction. Reactors can be batch reactors, continuous stirred tank reactors (CSTRs), plug flow reactors (PFRs), semi-batch reactors, or catalytic reactors.

[0021] As used herein, the term "sensor data" generally refers to any data representing the operating state of a chemical reaction or its parts measured by sensors of a reactor. Sensor data can be received directly from the sensors. Usually, sensor data is collected by a digital signal controller or a programmable logic controller and further transmitted from there. Sensor data can be adjusted, for example, by a calibration system before being transmitted. Sensor data from a reactor can also be stored on a storage medium, for example, stored on a hard disk drive or in a database in a cloud system. Thus, for the purposes of the present invention, sensor data can be obtained from such a storage medium.

[0022] Sensor data can include any measurable physicochemical value, such as temperature, pressure, pH, concentration or partial pressure of a compound (such as the content of oxygen or water), flow rate of a reagent, flow rate of the reaction mixture in the reactor or flow rate of the product after the reactor, stirrer speed, viscosity, turbidity. Usually, the sensor data also includes a physicochemical value that is associated with an identifier that identifies the sensor that has measured that value. The identification of the sensor can include the type of the sensor (e.g., thermometer) and the location of the sensor. The location of the sensor is particularly useful if more than one sensor of the same type makes measurements at different locations of the device. A typical example is a pressure sensor at the reactor inlet and another pressure sensor at the reactor outlet. The sensor data can also include time information, i.e., the time at which the sensor has collected the physicochemical value, sometimes called a timestamp. The sensor data can contain only one value from one sensor or can contain more than one value from one sensor, such as a time series of values. Thus, the sensor data can contain multiple values measured by the sensor at different time points. The sensor data can contain a time series of values measured by the sensor, where one value is measured after another value after a predefined period, e.g., one value per second.

[0023] Receive sensor data related to a chemical reaction in a first reactor. The term "related" must be understood in a broad sense, i.e., any information of the sensor that affects the chemical reaction or is related to the state of the chemical reaction. The information of the sensor can be the value output by the sensor, e.g., the temperature value of a thermometer, or a derived value, e.g., the viscosity value derived from a pressure sensor and the flow rate derived from a flow meter.

[0024] The chemical reaction can contain more than one reaction step. In this case, all reaction steps can occur in one reactor, or different reaction steps can occur in different reactors. For example, sequential or reaction steps can occur in multiple reactors, where substances are transferred from one reactor to another. The first reactor as used in the context of the present invention can refer to any one of these multiple reactors, or can refer to a subgroup or a combination of all of the multiple reactors. If the chemical reaction occurs in more than one reactor, the sensor data can contain values from sensors attached to one reactor, a selected reactor, or each reactor among these multiple reactors. In many cases, the sensor data contains at least one value of a sensor attached to the reactor in which the last reaction step occurs before the emissions are processed in a second reactor.

[0025] Sensor data can be received directly from the sensors of the first reactor or from a data storage medium. The sensor data on the data storage medium can be recorded sensor data or manipulated sensor data. The reason for manipulating the sensor data may be to simulate deviations and analyze the impact on the chemical reaction, with the aim of controlling emissions when such a situation occurs in reality. Another reason may be that changes in the chemical reaction can be foreseen, for example, different grades of reagents will be used, and this should be taken into account as early as possible.

[0026] The chemical reaction in the first reactor usually produces one or more products. Generally, not all products are desired. The undesired products are usually referred to as by-products. However, in some cases, the entire result of the chemical reaction is undesired and needs to be processed. For example, the chemical reaction may not yield a product within specific specifications, so the product cannot be sold or further processed. Additionally, it may happen that the desired product usually needs to be further processed in subsequent process steps, but the subsequent process steps may be temporarily unavailable, for example, due to some technical problems resulting in an unplanned shutdown. Sometimes, due to insufficient storage capacity or excessive safety risks associated with storing large quantities of a specific substance, the product cannot be stored, so the product needs to be disposed of. It is also possible that the desired product cannot be completely separated from the by-products, and the desired product has a major harmful impact on the environment in the emissions. Therefore, at least one product of the chemical reaction in the first reactor can be transferred to the second reactor, or more than one product (e.g., all or substantially all products) of the chemical reaction in the first reactor can be transferred to the second reactor.

[0027] Steps can be taken to separate the products obtained from the chemical reaction in the first reactor. In this separation step, the desired products can be separated from the by-products. This may be very simple when the physical states of the products and by-products are different. For example, the by-product can be a gas, such as CO, and the product can be a solid or a liquid. The separation step may simply involve releasing the gas into the second reactor through a valve. Other separation methods can also be envisioned, such as distillation, crystallization, filtration, centrifugation, extraction, precipitation. The separation step can be a continuous process or a discontinuous process. Generally, if the chemical reaction in the first reactor is continuous, the separation step is continuous, and if the chemical reaction in the first reactor is discontinuous, the separation step is discontinuous.

[0028] At least one product transferred to the second reactor may be a pollutant that cannot be directly released into the environment (e.g., released into the air through an exhaust device, released into the sewer as wastewater, or directly released into a river). Instead, such a product may need to be treated to convert at least a portion of it into a less harmful substance. This conversion or treatment can occur in the second reactor, and only the residue is released into the environment as an emission.

[0029] The treatment reactions in the second reactor include heat treatment, plasma treatment, combustion, neutralization, catalytic conversion. The treatment reaction can involve adding a treatment reagent to the second reactor, which reacts with at least one product of the chemical reaction in the first reactor. In the case of combustion, the treatment reagent can be air or oxygen. In the case of a neutralization reaction, the treatment reagent can be an acid or a base, depending on the product of the chemical reaction in the first reactor. In the case of catalytic conversion, there is a wide variety of treatment reagents depending on the product to be treated. For example, carbon monoxide (CO) can be treated by reacting it with oxygen over a platinum catalyst to form non-toxic carbon dioxide. Another example is the treatment of nitrogen oxides (NO x ) with ammonia or urea over a vanadium oxide catalyst to form nitrogen and water. To achieve optimal conversion, it is important to control the reaction parameters such as temperature, pressure, and, in the case of the treatment reagent, the dose of the treatment reagent (i.e., the amount of treatment reagent per application).

[0030] According to the present invention, operation instructions related to the treatment of emissions in the second reactor are determined based on sensor data. As used herein, the term "operation instructions" generally refers to any data that can be used to control the emissions of a chemical reaction by any means. The operation instructions can include instructions to keep all settings unchanged when the emissions are within an acceptable range (e.g., below a given threshold). The operation instructions can also include instructions to adjust one or more settings related to the treatment of emissions in the second reactor. The adjustment can refer to an increase or decrease in temperature, pressure, or throughput in the second reactor. In particular, the operation instructions can include instructions to increase or decrease the dose of the treatment reagent in the second reactor.

[0031] The operation instructions may further include instructions for adjusting one or more settings of the first reactor. In this way, the emissions are controlled by affecting the chemical reactions in the first reactor. For example, the chemical reactions in the first reactor can be adjusted by reducing the pressure, thereby reducing a specific product to be processed in the second reactor. The operation instructions may further include instructions for adjusting both one or more settings of the first reactor and one or more settings of the second reactor. By adjusting the settings of both the first reactor and the second reactor, the emissions can be controlled even more reliably, and the emissions treatment can be more efficient, for example, because it requires less treatment reagents and / or energy.

[0032] The operation instructions may include a time indication indicating the time at which the operation should be performed. The sensor data related to the chemical reactions in the first reactor may include sensor measurements at a certain point in time, while the operation instructions need to be executed at a later point in time (e.g., 10 seconds after the sensor measurement). The advantage of the present invention is that the sensor measurements in the first reactor may include information about the future by-product concentration. By using this information, actions can be taken at the appropriate point in time (i.e., immediately when needed), so that the control of the emissions is not affected by the delay caused by the signal waiting time or the calculation time.

[0033] The determination of the operation instructions is generally performed using a model that receives the sensor data as input and outputs the operation instructions. As used herein, the term "model" generally refers to a mathematical description of one or more physicochemical processes in the first reactor and / or the second reactor. The model can be a mechanism model, a data-driven model, or a hybrid model that includes both a mechanism model and a hybrid model. The advantage of the hybrid model is that it can, to a certain extent, strictly follow the known physicochemical laws in part, while considering historical data in the parts where understanding is insufficient. The hybrid model requires less historical data and is less prone to overfitting.

[0034] As used herein, the term "mechanism model" generally refers to a model based on the fundamental laws of natural science (e.g., any one or more of physics, chemistry, biochemical principles, heat and mass balance). Therefore, such a model uses equations to represent these principles. The mechanism model may include linear or nonlinear ordinary differential equations, linear or nonlinear partial differential equations, linear or nonlinear algebraic equations, or linear or nonlinear differential algebraic equations. These equations are related to the physicochemical process.

[0035] A typical example of a mechanism model is a chemical kinetics model. Essentially, such a model consists of ordinary differential equations or differential-algebraic equations that describe the kinetics of chemical species consumed or produced by a set of chemical reactions. The ordinary differential equations or differential-algebraic equation systems usually consist of rate laws, which are algebraic equations that describe the rate at which chemical species are consumed or produced in a reaction. Such algebraic equations typically depend on the concentrations of the chemical species, the temperature in a given reaction, and constants that are usually temperature-dependent. In addition, certain invariances such as mass conservation can also be expressed as algebraic equations in such mechanism models.

[0036] It is possible to know which mechanism models are most suitable for a certain physicochemical process. In this case, the selection of an appropriate mechanism model is straightforward. However, if it is not known which mechanism models are very suitable for the physicochemical process, then a set of mechanism models can be selected for a similar physicochemical process. Sometimes, there may be no similar physicochemical process available, either because the underlying mechanism is not yet clear or because appropriate information cannot be obtained for other reasons. In this case, it may be sufficient to pick an arbitrary mechanism model from a model library that contains various mechanism models of known physicochemical processes. Obviously, such an arbitrary mechanism model will be less suitable for the given physicochemical process. However, the associated data-driven model can compensate for at least part of the deviation, so such a result may be sufficient for less demanding purposes. Alternatively, different mechanism models can be picked arbitrarily, tried one by one, and the mechanism model that best fits the physicochemical process can be selected. Such a selection can be automated. Thus, a mechanism model can be selected from a model library by, for example, a computer program in the following way: arbitrarily select several mechanism models, apply the mechanism models to the physicochemical process one by one, determine the degree of fit of the mechanism model to the physicochemical process, and select the best-fitting mechanism model.

[0037] A "data-driven model" refers to a mathematical model parameterized based on a historical data set and used to reflect physicochemical processes such as the reaction kinetics in a first reactor and / or a second reactor. Compared with a mechanism model derived purely using physicochemical laws, a data-driven model can allow the description of relationships that are difficult or even impossible to model by physicochemical laws. The establishment of a data-driven model does not reflect any fundamental natural physical laws. These models are only considered by using the correlations in the data.

[0038] As used herein, the term "historical data" refers to a data set that includes at least sensor data and physicochemical values, wherein each data set is associated with a single physicochemical process run. Thus, each data set includes data associated with a physicochemical process run over a predefined time period. For a batch process, such a predefined time period can be from the start to the end of a batch run. For a continuous process, a characteristic period can be selected, such as the time from injecting a catalyst into a reactor until a new catalyst needs to be replaced. Historical data can be obtained from existing plants that should control emissions. However, historical data can also be sourced from laboratories, pilot plants, or similar plants. Sometimes, historical data from more than one of these plants can be obtained.

[0039] Training a model is typically done by adjusting the parameterization according to a historical data set. In this context, adjusting the parameterization means changing the parameters in a data-driven model included in a plant model such that the output of the plant model most closely resembles the response parameters of the training set. Depending on the type of data-driven model, various methods for adjusting the parameterization are known and are described in detail in the literature.

[0040] The data-driven model is preferably a data-driven machine learning model. The data-driven model can be linear or polynomial regression, decision tree, random forest model, Bayesian network, support vector machine, or preferably an artificial neural network.

[0041] A model can be used that takes sensor data as input and outputs operation instructions. A model can also be used that includes a first sub-model that takes sensor data as input and outputs predicted physicochemical values of a chemical reaction in a first reactor at a later time point (e.g., 5 seconds or 10 seconds after the sensor data is obtained). The model can include a second sub-model that receives the predicted physicochemical values of the chemical reaction in the first reactor as input and outputs operation instructions. Alternatively, the model can include a first sub-model that takes sensor data as input and outputs a predicted by-product concentration. The model can include a second sub-model that takes the predicted by-product concentration as input and outputs operation instructions. Both the first sub-model and the second sub-model can be mechanistic models, or the first sub-model is a mechanistic model and the second sub-model is a data-driven model, or the first sub-model is a data-driven model and the second model is a mechanistic model, or both the first sub-model and the second sub-model are data-driven models. The first sub-model and / or the second sub-model can also be hybrid models.

[0042] According to the present invention, operation instructions are output. Outputting can mean writing the operation instructions to a non-transitory data storage medium, for example, writing them to a monitoring file or a control file, displaying them on a user interface such as a screen, or both. The operation instructions can also be output through an interface of the control system. Such a control system can receive the operation instructions and change the settings of the devices in the second reactor based on such operation instructions.

[0043] The operation instructions can be determined to match a preset target value of the emissions. The target value can refer to the concentration of a certain compound. The target value can also refer to a range or an upper limit, that is, a threshold value that should not be exceeded. Such a threshold value can be a value obtained according to regulations or a value required by a standard, for example, for obtaining a product certificate. In the case where more than one substance in the emissions needs to be controlled, the target value can include a vector or a matrix, where the elements refer to the values or ranges of the specific substances to be controlled. The target value can be hard-coded into the model. This reduces the need for training data but makes it more difficult to adjust the target value. Alternatively, in addition to the sensor data, the model can also use the target value as an input.

[0044] The actual emissions can be measured, for example, by sensors or laboratory analysis to obtain emissions measurement data. The emissions measurement data can include the concentrations of one or more substances to be controlled. The emissions measurement data can also include a timestamp indicating the time when the emissions leave the second reactor. The emissions measurement data can be compared with the preset target value of the emissions. This comparison can yield the accuracy of the model for determining the operation instructions. Therefore, the accuracy of the model is a measure of the degree of match between the measured emissions and the target value that the model should achieve.

[0045] The emissions measurement values can be used by a controller that controls the settings related to the treatment reaction in the second reactor. In particular, the controller can be a proportional-integral-derivative controller (PID controller). The controller receives the emissions measurement values, calculates an error based on the predetermined concentration of the product contained in the emissions, and determines a corrective action, for example, changing the settings related to the treatment reaction in the second reactor. The advantage of such a controller is that it can adjust the settings in the case of unpredictable deviations, thus making emissions control more reliable.

[0046] Sensor data and emissions measurement data can be used to retrain a model to determine operating instructions. In this way, the accuracy of the model can be improved and / or adapted to changes in chemical reactions or processing reactions. For example, the activity of a catalyst may decrease over time, or the properties of a reagent may change. Retraining of the model can be triggered periodically, e.g., after a predetermined time interval (e.g., weekly or monthly). Alternatively, once the accuracy of the model drops below a predetermined threshold, e.g., below 90% or below 80%, retraining can be triggered.

[0047] To retrain the model, sensor data and / or emissions measurement value data can be added to the historical dataset. The model can then be retrained using this expanded historical dataset. This approach has the potential to result in a well-fitting model; however, it can be computationally expensive and may require the model to be stopped for some time. To avoid this stoppage, the trained model can be retrained using only the new dataset obtained from the sensors and / or emissions measurement data. This approach requires less computational power and thus can be completed more quickly. However, this approach involves the risk of the model "forgetting" the historical dataset on which it was originally trained. This effect is sometimes also referred to as catastrophic interference or catastrophic forgetting. Depending on the required accuracy of the model, this effect may be acceptable. However, one often wants to avoid this forgetting. There are various ways to retrain while reducing the impact of forgetting the historical dataset. Retraining may only allow small changes, e.g., by penalizing large changes in the cost or loss function, which is minimized during the retraining process. Alternatively, certain parts of the model can be left unchanged, e.g., if the results produced by this part of the model have a high or sufficiently high accuracy.

[0048] In another aspect, the present invention relates to a non-transitory computer-readable data medium storing a computer program, the computer program comprising instructions for performing the steps of the method according to the present invention. A "computer-readable data medium" refers to any suitable data storage device or computer-readable memory on which a set of one or more instructions (e.g., software) are stored, which instructions embody any one or more of the methods or functions described herein. The instructions may also reside, completely or at least partially, within the main memory and / or the processor during their execution by a computer, the main memory, and the processing device that may constitute a computer-readable storage medium. These instructions may further be sent or received over a network via a network interface device. Computer-readable data media include, for example, hard disk drives on servers, USB storage devices, CDs, DVDs, or Blu-ray discs. The computer program may contain all the functions and data required to perform the method according to the present invention, or may provide an interface to enable parts of the method to be processed on a remote system (e.g., on a cloud system).

[0049] In another aspect, the present invention relates to an emissions control system. Such a system may be configured to perform the method according to the present invention. Accordingly, all definitions, examples, and preferred embodiments described for the method also apply to the system.

[0050] The emissions control system includes an input end configured to receive sensor data related to a chemical reaction in a first reactor. Such an input end may include an interface for receiving sensor data. The input end may receive sensor data locally or remotely, e.g., via an interface with a telecommunications system such as the Internet. The input end may receive sensor data directly from the sensor or via a programmable logic controller, a distributed control system, or a storage medium including cloud services. The system may even be part of a distributed control system.

[0051] The emissions control system further includes a processor configured to determine at least one physicochemical parameter. The processor may be a local processor including a central processing unit (CPU) and / or a graphics processing unit (GPU) and / or an application-specific integrated circuit (ASIC) and / or a tensor processing unit (TPU) and / or a field-programmable gate array (FPGA). The processor may also be an interface with a remote computer system such as a cloud service.

[0052] The emissions control system further includes an output end for outputting operation instructions. Such an output end may include an interface for outputting operation instructions. The output end may send operation instructions locally or remotely, e.g., via an interface with a telecommunications system such as the Internet. The output end may send operation instructions to a programmable logic controller, a distributed control system, or a storage medium including cloud services. The system may even be part of a distributed control system. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 depicts a general setting in which an emissions control system operates.

[0054] Figure 2 depicts the method of the present invention.

[0055] Figure 3 depicts a specific embodiment of the method of the present invention.

[0056] Figure 4 depicts the emissions control system of the present invention.

[0057] Figure 5 depicts a decision tree for determining whether to retrain a model involved in an emissions control system.

[0058] Figure 6 depicts an example of a plant employing an emissions control system. DETAILED DESCRIPTION

[0059] Figure 1 shows an overview of the relevant components of the method (including the emissions control system). One or more reagents (101) are fed into a first reactor (102), where the reagent reacts with one or more products (105). The products include any substances made by chemical reactions and any unwanted but almost unavoidable by-products. At least one product (105) (e.g., a by-product) is fed into a second reactor (106), where the product is converted into a compound with less environmental impact, and then the reaction product from the second reactor (106) is released into the environment as emissions (107), e.g., released into the air or wastewater through an exhaust device. The first reactor (102) is equipped with at least one sensor that generates sensor data (103). The sensor data (103) may include one or more sensor values measured at one or more time points. The sensor data (103) may also include an indicator of the sensor type and / or location for the sensor value and / or a timestamp indicating the time when the measurement was performed.

[0060] The sensor data (103) is transmitted to the emissions control system (104). The emissions control system (104) receives the sensor data (103) and uses the sensor data to determine operation instructions related to emissions treatment in the second reactor (106). The output operation instructions are such that they can be used to act on the second reactor. This can be indirectly performed via a human operator taking the required actions or directly by a controller (e.g., a valve for adjusting the amount of treatment reagent or a heater for adjusting the temperature in the second reactor).

[0061] Figure 2 Describes details of a method for controlling emissions from a chemical reaction. Sensor data (201) related to a chemical reaction in a first reactor is received. Operating instructions (202) for emissions treatment in a second reactor are determined based on the sensor data (201) from the first reactor. The determination is typically performed by a parameterized model such that it receives at least a portion of the sensor data as input and outputs the operating instructions. Optionally, the sensor data (201) from the first reactor can be used to determine the concentration or level (204) of one or more by-products that need to be treated before being released into the environment. The determination (204) of the one or more by-products can be used to confirm the accuracy of the determination, for example, by comparing the determined by-product concentration with data (205) including the measured by-product levels. This comparison can be carried out continuously or only from time to time. If time-consuming laboratory analysis is required to determine the by-product concentration, it is advantageous to carry out the comparison from time to time. Optionally, data (205) including the measured by-product levels can be used as an additional input for determining the operating instructions. This additional input can improve the accuracy of determining the operating instructions.

[0062] Figure 3 Describes a specific embodiment of the method. The sensor data (310) from the first reactor contains a time series of values obtained from sensors. A set of sensor values is associated with time t-2 (311), a set of sensor values is associated with time t-1 (312), and a set of sensor values is associated with time t (313). The times t-2, t-1, and t can be separated by a certain time interval, for example, one second. For the purpose of illustration, only three sets of sensor values are selected. The sensor data can contain more or fewer sets. The sensor data is used to predict the sensor data (320), thereby obtaining the predicted sensor data (330). To this end, a first sub-model can receive one or more than one set of sensor values (for example, the entire time series of sensor values) and output a set of sensor values at a future time point. The predicted sensor data (330) can contain sensor values related to a future time t+1 (331) (for example, the upcoming one or more seconds). Optionally, the predicted sensor data (330) can contain sensor values related to more than one time (for example, t+1 (332) and t+2 (330)), that is, predict a series of sensor values in the future.

[0063] The predicted sensor data (330) can be used to determine an operation instruction (340). To achieve this, a second sub-model can be used, which uses the predicted sensor data (330) as input and the operation instruction as output. The second sub-model can use a set of sensor values related to one time, such as a set of sensor values at t+1 (331), or can use multiple sets of sensor values related to different times, such as a set of sensor values at t+1 (331) and a set of sensor values at t+2 (332). In the latter case, the time series can be extrapolated. The time intervals between t, t+1, and t+2 can be the same or different from each other. These time intervals can be the same or different from the time intervals between t, t-1, t-2, etc. Once the operation instruction is determined, it is output (350).

[0064] Figure 4 Details of the emissions control system (410) are depicted. The emissions control system (410) includes an input end (411) that is used to receive sensor data (402) related to a chemical reaction in the first reactor (401). The input end (411) can be an interface of a storage medium on which sensor measurements are recorded, or an interface of a communication medium (such as a cable or wireless communication connection) for receiving the sensor data (402). A processor (412) is configured to determine an operation instruction based on the sensor data (402). The processor (412) may be able to execute a parameterized model such that it requires at least a part of the sensor data (402) as input and outputs the operation instruction. The processor can be a local processor, such as part of a computer, or a remote computer center, such as a cloud service. The operation instruction is output by an output end (413). The output end (413) can be an interface of a storage medium on which the operation instruction is written, or an interface of a communication medium (such as a cable or wireless communication connection) for transmitting the operation instruction. The operation instruction can be used to generate an operation signal (421) so as to adjust the second reactor (422) to control emissions. Alternatively, the operation instruction can be displayed on a display, so that an operator can perceive the operation instruction and take any action based on the operation instruction.

[0065] Figure 5An example of monitoring the accuracy of a model for determining operating instructions based on sensor data from a first reactor is shown. Sensor data (501) from the first reactor is received and used as input to the model. The model can output a predicted emissions level (502) and an operating instruction to control the emissions level. The emissions level can be measured (503) after the operating instruction is executed. The predicted emissions level (502) can be compared with the measured emissions level (503). If the difference is within an acceptable range, the model can be further used without modification (506). If the difference is outside the acceptable range (e.g., the difference exceeds a predetermined threshold), the model can be retrained (507). For this purpose, the sensor data (501) from the first reactor and the measured emissions level (503) can be used as training data. If only new data is used for training and only minor changes to the model are allowed, the retraining can be performed immediately, i.e., without stopping the model. In this way, the model can be quickly adjusted according to changing conditions. This method is particularly useful if the model is a hybrid model containing known physicochemical relationships that is immutable during retraining. Overfitting can be effectively avoided.

[0066] Figure 6 An example of how the method and system of the present invention can be applied is shown. The reagents (601) cumene and oxygen are injected into a first reactor (602). The first reactor (602) can be a solid bed column reactor with an immobilized oxidation catalyst. In the first reactor (602), cumene reacts with oxygen to yield the products (605) phenol and acetone. The reaction is catalyzed by the oxidation catalyst. Carbon monoxide (CO) is formed as a by-product (604) of the chemical reaction in the first reactor (602). Since CO is a gas, it can be easily separated from the liquid or solid products. CO is a toxic gas and thus cannot be released into the environment via the exhaust device. Therefore, CO is transferred to a second reactor (609). Oxygen is added to the second reactor (609) via a valve (607). The second reactor (609) can also be a solid bed column reactor with an immobilized oxidation catalyst. CO is converted to carbon dioxide (CO2) in the second reactor together with oxygen and the oxidation catalyst, and the carbon dioxide can be released into the atmosphere as an emission (610). However, the reaction is not complete, and its turnover rate depends on many parameters, particularly the ratio between CO and O2, and also on temperature, pressure, catalytic activity, and the average contact time between CO and O2 and the catalyst, which is affected by the gas flow rate through the reactor. Therefore, it is desirable to control these parameters such that the concentration of CO in the emission (610) is minimized.

[0067] To achieve this, the emissions control system (606) receives sensor data from sensors (603) attached to the first reactor (602). Typically, there are multiple sensors (603) attached to the first reactor (602). These sensors can include sensors for determining flow rates, for example, one flow rate sensor that measures the amount of reagent flowing into the first reactor (602) at the inlet of the first reactor (602), and one flow rate sensor that measures the amount of product and by-products flowing out of the first reactor (602) at the outlet of the first reactor (602). The sensors (603) can also include thermometers that measure the temperature inside the reactor. Several thermometers can be placed at different parts of the first reactor (602), for example, at the inlet of the reactor, where the reagent first contacts the oxidation catalyst, in the central part of the reactor (where the chemical reaction has partially occurred), or at the end of the reactor (where the product leaves the first reactor (602)). Thus, the sensor data can include the temperature distribution along the flow direction of the reagent and product through the first reactor (602). Such sensor data can contain information about the reaction rate of the chemical reaction in the first reactor (602). The sensors (603) can also include pressure sensors that measure the pressure inside the first reactor (602). Several pressure sensors can be placed at different parts of the first reactor (602), for example, at the inlet of the reactor, where the reagent first contacts the oxidation catalyst, in the central part of the reactor (where the chemical reaction has partially occurred), or at the end of the reactor (where the product leaves the first reactor (602)). Thus, the sensor data can include the pressure distribution along the flow direction of the reagent and product through the first reactor (602). Such sensor data can contain information about the catalyst state and / or any blockages in the first reactor (602).

[0068] Sensor data typically includes a time series of sensor measurements. This means that some or all of the sensors repeat their measurements and provide measurement values, for example, after a predetermined time interval (e.g., once per second) or whenever the measurement value changes by a predetermined amount. Sensor data can contain multiple entries, each entry including an identifier for identifying the sensor, a timestamp indicating the time when the measurement value was measured, and the measured value.

[0069] Optionally, the sensor data can include the physicochemical values of a chemical reaction prior to the chemical reaction in the first reactor. In this example, cumene (601), the reagent, can be produced by reacting benzene with propylene in another reactor. This reactor can be equipped with sensors that measure the flow rate of substances flowing into the reactor, the flow rate of substances flowing out of the reactor, the temperature and pressure at different locations. These sensor values can be included in the sensor data together with identifiers indicating that these values are related to the chemical reaction prior to the chemical reaction in the first reactor.

[0070] The emissions control system (606) receives sensor data. The emissions control system (606) can receive the measured value of each sensor immediately after each sensor measurement is completed, or receive a set of sensor measurements at a predetermined time interval (e.g., once per second). In the latter case, the sensor data needs to be collected, for example, by a sensor controller, which in turn sends the collected sensor data to the emissions control system (606). The controller can be part of a distributed control system for the first reactor (602). Alternatively, such a controller can be part of the emissions control system (606).

[0071] The emissions control system (606) uses some or all of the sensor data as input to a model, which is parameterized based on the sensor data input. The model can be a hybrid model that includes kinetic equations for the reactions in the first reactor (602) and kinetic equations for the reactions in the second reactor (609). Additionally, the hybrid model can also include a data-driven part, e.g., one or more neural networks. These neural networks can receive some or all of the sensor data as input and output parameters that can be added to or multiplied by the kinetic equations. The output of the hybrid model is an operating instruction. The hybrid model has been trained using historical data that includes, for example, sensor data, the amount of processing reagent, and emissions components. The operating instruction output by the model can be the amount of oxygen required by the second reactor at a certain time. Thus, the emissions control system (606) determines the operating instruction based on the sensor data. The emissions control system can send the operating instruction to a valve (607), which is adjusted according to the operating instruction to adjust the amount of processing reagent (608) (i.e., oxygen) flowing into the second reactor (609) as determined by the model.

[0072] Additionally, the model can also output the byproduct concentration at a certain point in time. This output can be compared with the measured value of the byproduct concentration at that specific point in time. This comparison can indicate the accuracy of the model. Furthermore, the emissions can be measured and the measured value compared with the expected value. For example, the model is adjusted to keep the CO concentration in the emissions (610) below a predetermined value. The measured CO concentration in the emissions (610) can be compared with the predetermined value, and this comparison can yield the accuracy of the model. For example, the degree and duration by which the CO concentration exceeds the predetermined value can be used to measure the accuracy of the model. If the accuracy is below a certain value, the model can be triggered to be retrained. This situation may be due to differences in reagent quality, changes in catalyst activity, or blockages that slowly accumulate in the reactor.

[0073] From the study of the drawings, the present disclosure, and the appended claims, those skilled in the art can understand and implement other variations of the disclosed embodiments when practicing the claimed invention.

[0074] For the processes and methods disclosed herein, the operations performed in the processes and methods can be implemented in a different order. In addition, the operations outlined are provided only as examples, and some of them can be optional, can be combined into fewer steps and operations, can be supplemented with more operations, or can be extended into more operations without departing from the essence of the disclosed embodiments.

[0075] In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single unit or device can perform the functions of several items recited in the claims. The fact that certain measures are recited in mutually different dependent claims does not mean that a combination of these measures cannot be used advantageously.

Claims

1. A computer-implemented method for controlling emissions of a chemical reaction, the method comprising: (a) receiving sensor data related to the chemical reaction in a first reactor, (b) determining, based on the sensor data, operation instructions related to emissions treatment of at least one product obtained by the chemical reaction in a second reactor, and (c) outputting the operation instructions.

2. The computer-implemented method according to claim 1, wherein, The operation instructions are determined to match a preset target value of the emissions.

3. The computer-implemented method according to any one of the preceding claims, wherein, Determining the operation instructions involves determining the by-product level of the chemical reaction in the first reactor.

4. The computer-implemented method according to any one of the preceding claims, wherein, A second chemical reaction occurs in the second reactor to reduce emissions.

5. The computer-implemented method according to the previous claim, wherein, The operation instructions include instructions for adjusting reaction parameters of the second chemical reaction.

6. The computer-implemented method according to the previous claim, wherein, The reaction parameter is the dosage of a treatment reagent for reacting with the by-products of the chemical reaction in the first reactor.

7. The computer-implemented method according to any one of the preceding claims, wherein, The chemical reaction is a continuous chemical reaction.

8. A computer-implemented method according to any one of the preceding claims, wherein, The sensor data includes a time series of physicochemical values measured by at least one sensor.

9. The computer-implemented method according to any one of the preceding claims, wherein, Determining the operation instructions employs a hybrid model using at least a portion of the sensor data as input values.

10. The computer-implemented method according to any one of the preceding claims, wherein, Measuring at least one concentration of the by-products of the chemical reaction in the first reactor, and wherein determining the operation parameters is further based on the measured concentration of the by-products.

11. The computer-implemented method according to any one of the preceding claims, wherein, Determining the operation instructions is based on a model that has been trained using a set of historical training data.

12. The computer-implemented method according to the preceding claim, wherein, Continuously retraining the model using the sensor data.

13. A non-transitory computer-readable data medium storing a computer program, the computer program comprising instructions for performing the steps of the method according to any one of the preceding claims.

14. Use of the operation instructions obtained by the method according to any one of the preceding claims for controlling emissions of a chemical reaction.

15. An emissions control system, comprising: (a) an input end for receiving sensor data related to a chemical reaction in a first reactor, (b) a processor for determining, based on the sensor data, operation instructions related to emissions treatment of at least one product obtained by the chemical reaction in a second reactor, and (c) an output end for outputting the operation instructions.

Images