WO2024110415A1

WO2024110415A1 - Method for controlling emissions of a chemical reaction

Info

Publication number: WO2024110415A1
Application number: PCT/EP2023/082454
Authority: WO
Inventors: Hai Jiang; Satya Swarup SAMAL; Marcus REBLE; Zhe Xiong YANG; Nan Zhang
Original assignee: Basf Se; Basf (China) Company Limited
Priority date: 2022-11-21
Filing date: 2023-11-21
Publication date: 2024-05-30

Abstract

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

Description

Method for Controlling Emissions of a Chemical Reaction

Description

The invention is in the field of controlling emissions of a chemical reaction. The invention relates to a method for controlling emissions of a chemical reactions, an emission control system, a non-transitory computer-readable data medium and the use of a operational instruction for controlling the emissions of a chemical reaction.

Background

Chemical reactions usually produce undesirable, but hardly avoidable by-products. Such byproducts are often harmful to the environment, for example because they are toxic, produce acidic rain or promote algae growth in waters. To avoid or at least reduce detrimental environmental impact, by-products of chemical reactions are often treated, for example in a subsequent reaction, to convert them into non-harmful substances before releasing them into the environment. A typical example is the exhaust treatment of a combustion engine. Traditionally, exhaust gases are measured to control the exhaust treatment. However, this feedback loop can be slow and thus inefficient. Prediction models have been proposed in order to react before any undesirable emission occurs.

WO 2020 / 176 914 A1 discloses a method for predicting exhaust gases using artificial neural networks. Exhaust gas stream parameters are measured and fed into the neural network to control the exhaust treatment. However, such an approach can often not react quickly enough to changes, so the throughput of the reaction is limited.

Summary

It was an object of the present invention to provide a method for emission control for a chemical reaction which can reliably control the emissions in order to maximize the reaction throughput. The method was aimed to be flexible so it can be applied to various different reactions. Furthermore, the method should be robust against sudden changes of the chemical reaction.

These objects were achieved in a first aspect by a computer-implemented method for controlling emissions of a chemical reaction comprising

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

(b) determining an operational instruction related to emission treatment in a second reactor of at least one product obtained by the chemical reaction based on the sensor data, and

(c) outputting the operational instruction.

According to a second aspect the invention relates to a non-transitory computer-readable data medium storing a computer program including instructions for executing steps of the method according to present invention. According to a third aspect the invention relates to the use of the operational instruction obtained by the method according to the present invention for controlling the emissions of a chemical reaction.

According to a fourth aspect the invention relates to an emission control system comprising

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

(b) a processor for determining an operational instruction related to emission treatment in a second reactor of at least one product obtained by the chemical reaction based on the sensor data, and

(c) an output for outputting the operational instruction.

The invention allows an improved 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 a first reactor, whereas in conventional methods, the concentration of by-products is first measured to determine any measure for the emission control. Hence, the measure cannot be determined before the by-product is actually formed. In contrast, the present invention allows an earlier determination. This is because the data related to the chemical reaction in a first reactor can be used to predict changes in of the chemical reaction in the near future. Consequently, a faster reaction to control the emissions is possible, i.e. the emission control can be adjusted before any emissions go up due to an increase of the by-product concentration. For example, the concentration of a by-product which should not be released into the environment above a certain level can be predicted, so measures, like increasing the dosing of a treatment reagent in a second reactor, can be taken before the emissions increase. In this way, it is possible to control the emissions more reliably, i.e. in fewer cases any emission levels are exceeded. Also, in some cases the chemical reaction can be operated more efficiently without an increased risk of exceeding emission levels. For example, a chemical reaction may be pushed towards the limits, i.e. with lower security margins with regard to the emission limits. In addition, the throughput of the chemical reaction may be increased due to the fact that the emission control can react faster, so any time delays in traditional methods, for example the measurement of a by-product level, can be avoided. Additionally, the treatment reaction can be operated more efficiently, for example because it needs just the right amount of treatment reagent, energy or time to treat the by-product. In this way, no treatment reagent, energy or time is wasted.

The term “emissions” as used herein generally refers to a substance or a mixture of substances which when released to the environment has a detrimental impact on the environment. Emissions include air pollutants, water pollutants or soil pollutants. Air pollutants include toxic and/or acidic gases, like CO, NO_X, SO_X, O3; volatile organic compounds including aliphatic hydrocarbons like methane or aromatic hydrocarbons like benzene; toxic metals like lead or mercury; chlorofluorocarbons; dust particles like fine particles. Water pollutants include chemicals from insecticides, herbicides fungicides; petroleum hydrocarbons including fuels; lubricants like motor oil; fuel combustion byproducts including partially oxidized hydrocarbons like alcohols, aldehydes, ketones or carboxylic acids; volatile organic compounds such as industrial solvents; persistent organic pollutants such as per- and polyfluoroalkyl substances; organochlorides such as polychlorinated biphenyl (PCBs), trichloroethylene, oxidants such as perchlorate; acids such as hydrofluoric acid, hydrochloric acid, sulfuric acid, nitric acid; nutrients like nitrates or phosphates; plastic particles such as microplastic. Soil pollutants include most of the water pollutants described above. Generally, many substances can be both air pollutants and water pollutants, or air pollutants and soil pollutants or water pollutants and soil pollutants or air pollutants, water pollutants and soil pollutants.

The term “controlling” as used herein generally refers to any action which may have an influence on the emissions, in particular an action which reduces the detrimental impact of the emissions on the environment. The action can be direct, for example by changing the state of a valve in order to adjust the flow of a treatment reagent, changing the temperature by additional heating or increasing the cooling. The actions can also be indirect, for example by prompting an operator to take actions, for example exchanging a filter or adjusting throughput. Controlling emissions can mean taking an action to keep the concentration of at least one pollutant below a predetermined threshold. In some cases, controlling can also mean taking an action to keep the concentration of more than one pollutant below thresholds, wherein typically different thresholds are applied for different pollutants. Usually, the pollutant is a by-product of the chemical reaction in the first reactor.

A “chemical reaction” as used herein generally refers to a process that involves the transformation of one set of chemical substances to another. Chemical reactions can be operated in a homogeneous or heterogeneous way. Homogeneous chemical reactions involve one phase, for example in a gas phase or in a liquid phase, such as a solution. Heterogeneous chemical reactions involve at least two phases. The at least two phases can be of different state of matter, for example one phase is solid and one phase is liquid, or one phase is solid and the other phase is gaseous, or one phase is liquid and the other phase is gaseous. The at least two phases can be of the same state of matter it they are immiscible, for example two immiscible liquid phases or two immiscible solid phases.

Chemical reactions can be operated in a continuous or discontinuous way, sometimes also referred to as batch chemical reactions. In a continuous chemical reaction, the reagents are continuously fed into a reactor where the reaction takes place and at the same time the products are continuously output from the reactor. In a discontinuous chemical reaction, a reactor is charged with the reagents, then the reaction takes place and after that the products are collected from the reactor. The reactor may be cleaned and is then again charged with new reagents.

The term “reactor” as used herein generally refers to a container to accommodate the chemical reaction. A reactor can be a batch reactor, continuous stirred-tank reactor (CSTR), a plug flow reactor (PFR), a semibatch reactor or a catalytic reactor.

The term "sensor data" as used herein generally refers to any data which represents the operational state of the chemical reaction or parts of it as measured by sensors of the reactor. The sensor data may be received directly from the sensors. Typically, the sensor data are collected by a digital signal controller or programmable logic controller and further transmitted from there. The sensor data may be adjusted, for example by a calibration system before being transmitted. The sensor data from the reactor may also be stored on a storage medium, for example a database on a hard drive or in a cloud system. Hence, the sensor data may be obtained from such storage medium for the purpose of the present invention.

The sensor data can comprise any measurable physical-chemical value, such as temperature, pressure, pH, concentration or partial pressure of a compound such as oxygen or water content, flow rate of reagents, of the reaction mixture in the reactor or of the products after the reactor, stirrer speed, viscosity, turbidity. Usually, the sensor data also comprises a physical-chemical value associated with an identifier identifying the sensor which has measured the value. The identification of the sensor may comprise the type of sensor, for example a thermometer, and the location of the sensor. The latter is particularly useful if more than one sensor of the same type measures at different locations of the equipment. A typical example is a pressure sensor at the inlet of a reactor and another pressure sensor at the outlet of a reactor. The sensor data can also comprise time information, i.e. the time at which the sensor has collected the physicalchemical value, sometimes referred to as time stamp. The sensor data may contain only one value from one sensor or it may contain more than one value from one sensor, for example a time series of values. Hence, the sensor data may contain multiple values measured by a sensor at different points in time. The sensor data may contain a time series of values measured by a sensor, wherein a value is measured after a predefined time period after the other, for example one value every second.

Sensor data related to the chemical reaction in a first reactor is received. The term “related” has to be understood in a broad way, namely any information of a sensor which has an influence on the chemical reaction or correlates to the state of the chemical reaction. The information of a sensor may be the value the sensor outputs, for example the temperature value of a thermometer, or it may be derived value, for example a viscosity value derived from a pressure sensor and a flow rate from a flow meter.

It is possible that the chemical reaction contains more than one reaction step. In this case, all reaction steps may take place in one reactor, or different reaction steps may take place in different reactors. For example, a sequence or reaction steps may take place in multiple reactors, wherein substance is transferred from one reactor to another. The first reactor as used in the context of the present invention may refer to any of these multiple reactors, or it may refer to the combination of a subgroup or all of the multiple reactors. If the chemical reaction takes place in more than one reactor, the sensor data may contain values from sensors attached to one, a selection or each of 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 takes place before emissions are treated in the second reactor. The sensor data may be received directly from sensors of the first reactor, or it may be received from a data storage medium. The sensor data on the data storage medium may be recorded sensor data or manipulated sensor data. A reason for manipulating sensor data may be to simulate deviations and analyze the impact on the chemical reaction with the goal to control the emissions in case such situation happens in reality. Another reason may be that a change of the chemical reaction can be foreseen, for example a different grade of reagents is going to be employed which shall be taken into account as early as possible.

The chemical reaction in the first reactor usually produces one or more than one product. Often, not all products are desired. The undesired products are often referred to as by-products. However, there are occasions where the complete outcome of a chemical reaction is undesired and needs to be disposed. For example, the chemical reaction may not have yielded the product within a certain specification, so it cannot be sold or further processed. Also, it may happen that the desired product is usually further processed in a following process step, but the following process step can temporarily not be executed, for example due to an unplanned down-time for some technical difficulties. Sometimes, the product cannot be stored due to insufficient storage capacities or too high security risks associated with storing large amounts of certain substances, so it needs to be disposed. It is also possible that the desired product cannot be separated completely from the by-products and the desired product causes the major detrimental environmental impact in the emissions. Hence, at least one product of the chemical reaction in the first reactor may be transferred to the second reactor, or more than one product, for example all or essentially all, products of the chemical reaction in the first reactor may be transferred to the second reactor.

A step of separating the products obtained from the chemical reaction in the first reactor may take place. In this separation step, the desired products may be separated from the by-products. This can be very simple in case product and by-product have a different state of matter. For example, the by-product may be a gas, for example CO, and the product may be a solid or a liquid. The separation step may just be releasing the gas by a valve to the second reactor. Other separation methods are conceivable, for example distillation, crystallization, filtration, centrifugation, extraction, precipitation. The separation step can be a continuous process or a discontinuous process. Often, the separation step is continuous if the chemical reaction in the first reactor is continuous, and it is discontinuous if the chemical reaction in the first reactor is discontinuous.

The at least one product transferred to the second reactor may be a pollutant which cannot be directly released into the environment, for example through an exhaust into the air or as wastewater into a sewer or directly into a river. Instead, such product may need to be treated to convert at least parts of it into less harmful substances. Such conversion or treatment may take place in the second reactor and only the residue is released as emissions into the environment.

Treatment reactions in a second reactor include thermal treatment, plasma treatment, combustion, neutralization, catalytic conversion. A treatment reaction may involve dosing a treatment reagent into the second reactor which reacts with the at least one product of the chemical reaction in the first reactor. In case of a combustion the treatment reagent may be air or oxygen. In case of a neutralization reaction, the treatment reagent may be an acid or a base depending on the product of the chemical reaction in the first reactor. In case of a catalytic conversion, a wide range of treatment reagents exist depending on the product to be treated. For example, carbon monoxide (CO) may be treated by reacting it with oxygen on a platinum catalyst to form the non-toxic carbon dioxide. Another example is the treatment of nitric oxides (NO_X) with ammonia or urea on vanadium oxide catalyst to form nitrogen and water. In order to achieve the best conversion rate, it is important to control the reaction parameters like temperature pressure and, in case of a treatment reagent, the dosing, i.e. the amount of treatment reagent per time, of the treatment reagent.

According to the present invention, an operational instruction related to emission treatment in a second reactor based on the sensor data is determined. The term “operational instruction” as used herein generally refers to any data which can be used to control the emissions of the chemical reaction by any means. The operational instruction may include the instruction to leave all settings as they are in case the emissions are in an acceptable range, for example below a given threshold. The operational instruction may also include the instruction to adjust one or more than one setting related to the emission treatment in the second reactor. An adjustment may refer to an increase or decrease of the temperature, pressure or throughput in the second reactor. In particular, the operational instruction may include the instruction to increase or decrease the dosing of a treatment reagent in the second reactor.

It is also possible that the operational instruction includes an instruction to adjust one or more than one setting of the first reactor. In this way, the emissions are controlled by influencing the chemical reaction in the first reactor. For example, the chemical reaction in the first reactor may be adjusted by decreasing the pressure to reduce a particular product which is treated in the second reactor. It is also possible that the operational instruction contains both an instruction to adjust one or more than one setting of the first reactor and one or more than one setting of the second reactor. By adjusting the settings of both the first and the second reactor, the emissions can be controlled even more reliably and the emission treatment can be more efficient, for example because it needs less treatment reagent and/or energy.

The operational instruction may include a time indication indicating the time at which the operation shall be executed. The sensor data related to the chemical reaction in the first reactor may contain sensor measurement values at a certain point in time while the operational instruction needs to be executed at a later point in time, for example 10 seconds after the sensor measurement. The advantage of the present invention is that sensor measurements in the first reactor may contain information about a future by-product concentration. By using this information, an action can be taken at the right point in time, i.e. immediately when required, so no delays due to signal latencies or calculation time compromise the control of the emissions. The determination of the operational instruction is usually executed using a model which receives the sensor data as input and which outputs the operational instruction. The term "model" as used herein usually refers to a mathematical description a physical-chemical process or multiple physical-chemical processes in the first and/or second reactor. The model can be a mechanistic model, a data-driven model or a hybrid model containing both a mechanistic model and a hybrid model. A hybrid model has the advantage that it can in parts strictly follow physicalchemical laws to the extend known and at the same time take into account historic data for parts which are less well understood. Hybrid models require less historic data and are at the same time less susceptible to overfitting.

The term "mechanistic model" as used herein generally refers to a model which is based on the fundamental laws of natural sciences, for example any one or more of physical, chemical, biochemical principles, heat and mass balancing. Such models thus represent these principles using equations. A mechanistic model can comprise linear or nonlinear ordinary differential equations, linear or nonlinear partial differential equations, linear or nonlinear algebraic equations, or linear or non-linear differential algebraic equations. Such equations relate to a physical-chemical process.

A typical example for a mechanistic model is a chemical kinetic model. Essentially, such a model is composed of ordinary differential equations or differential algebraic equations describing the dynamics of chemical species that are being consumed or produced by a set of chemical reactions. The system of ordinary differential equations or differential algebraic equations are usually composed of rate laws that are algebraic equations describing the speed at which chemical species are consumed or produced in reactions. Such an algebraic equation typically depends on the concentrations of the chemical species, temperature in the given reaction and constants, which are usually temperature dependent. Furthermore, certain invariances, such as conservation of mass, can also be represented in such a mechanistic model as algebraic equations.

It may be known which mechanistic models fit best to a certain physical-chemical process. In this case, the selection of adequate mechanistic models is straight forward. If, however, it is not known which mechanistic models fit well to the physical-chemical process, one may select a set of mechanistic models for a similar physical-chemical process. Sometimes, there may not be a similar physical-chemical process available, maybe because the underlying mechanism is not yet known or the appropriate information is not available for a different reason. In this case, it may be sufficient to pick an arbitrary mechanistic model from a model library which contains various mechanistic models for known physical-chemical process. Obviously, such an arbitrary mechanistic model will not fit very well to a given physical-chemical process. However, an associated data-driven model may compensate at least part of the deviation, so the result may be sufficient for less demanding purposes. Alternatively, one arbitrarily picks different mechanistic models, tries one after the other and selects the mechanistic model which fits best to the physical-chemical process. Such selection can be automated. Hence, the mechanistic models may be selected from a model library, for example by a computer program, by arbitrarily selecting several mechanistic models, applying one after the other to the physical-chemical process, determining how well the mechanistic model fits to the physical-chemical process and selecting the best fitting mechanistic model.

"Data-driven model" refers to a mathematical model that is parametrized according to historical data set to reflect a physical-chemical processes such as reaction kinetics in the first and/or second reactor. In contrast to a mechanistic model that is purely derived using physical-chemical laws, a data-driven model can allow describing relations that are difficult or even impossible to be modelled by physical-chemical laws. Data-driven models are set up without reflecting any underlying physical laws of nature. These are taken into account solely by using the correlations in the data.

The term "historical data" as used herein refers to data sets including at least sensor data and physical-chemical values, wherein each data set is associated with a single physical-chemical process run. Hence each data set includes data associated with the physical-chemical process run in a predefined time period. For a batch process such predefined time period may be the beginning to the end of one batch run. For a continuous process, a characteristic period may be chosen, for example the time from charging a reactor with a catalyst until it needs to be replaced by new catalyst. Historic data can be obtained from an already existing plant in which emissions shall be controlled. However, it can also originate from a laboratory, a pilot plant or a similar plant. Sometimes historic data from more than one of these are available.

Training the model is typically done by adjusting the parameterization according to the historical dataset. Adjusting the parameterization in this context means varying the parameters in the data-driven model comprised in the plant model such that the output of the plant model most closely resembles the reaction parameters of the training set. Depending on the type of data- driven model, various methods of doing so are known and well described in the literature.

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

It is possible to use one model which uses the sensor data as input and outputs the operational instruction. It is also possible to use a model comprising a first sub model which uses the sensor data as input and outputs predicted physical-chemical values of the chemical reaction in the first reactor at a later point in time, for example 5 or 10 seconds after the sensor data has been obtained. The model may comprise a second sub model which receives the predicted physicalchemical values of the chemical reaction in the first reactor as input and outputs the operational instruction. Alternatively, the model may contain a first sub model which uses the sensor data as input and outputs a predicted by-product concentration. The model may contain a second sub model which uses the predicted by-product concentration as input and outputs the operational instruction. It is possible that both the first and the second sub models are mechanistic models or that the first sub model is a mechanistic model and the second sub model is a data- driven model or that the first sub model is a data-driven model and the second model is a mechanistic model or that bot the first and the second sub model are data-driven models. It is also possible that the first sub model and/or the second sub model are hybrid models.

According to the present invention, the operational instruction is output. Outputting can mean writing the operational instruction on a non-transitory data storage medium, for example into a monitoring file or a control file, display it on a user interface, for example a screen, or both. It is also possible to output the operational instruction through an interface to a control system. Such control system may receive the operational instruction and based on such operational instruction change settings of equipment in the second reactor.

The operational instruction may be determined to match a preset target value for the emissions. The target value may refer to a concentration of a certain compound. The target value may also refer to a range or an upper limit, i.e. a threshold which should not be exceeded. Such a threshold may be a value derived from legislation or a value required by a standard, for example to obtain a certificate for a product. In case more than one substance in the emissions need to be controlled, the target value may comprise a vector or a matrix, wherein an element refers to a value or a range for a particular substance which needs to be controlled. It is possible that the target value is hard coded into the model. This reduces the need for training data but makes adjustments to the target value more difficult. Alternatively, the model may use the target value as input in addition to the sensor data.

The actual emissions may be measured, for example by sensors or by laboratory analytics, to obtain emission measurement data. Emission measurement data may include the concentration of one or more substances to be controlled. Emission measurement data may also include a timestamp indicating the time when the emissions left the second reactor. The emission measurement data may be compared to the preset target value for the emissions. The comparison may yield an accuracy of the model used to determine the operational instruction. Hence, the accuracy of the model is a measure of how close the measured emissions match the target value which the model is supposed to achieve.

The emission measurement may be used by a controller which controls the settings related to the treatment reaction in the second reactor. In particular, the controller may be a proportional- integral-derivative controller (PID controller). The controller receives the emission measurement, calculates an error based on the predetermined concentration of a product contained in the emissions and determines a correction action, for example the change of settings related to the treatment reaction in the second reactor. Such controller may have the advantage to adjust settings in case of unpredictable deviations, so controlling the emissions becomes even more reliable.

The sensor data and the emission measurement data may be used to retrain the model for determining the operational instruction. In this way, the accuracy of the model may be increased and/or adapted to changes in the chemical reaction or the treatment reaction. For example, the activity of a catalyst may degrade over time, or the properties of reagents may vary. Retraining the model may be triggered regularly, for example after a predetermined time interval, for example every week or every month. Alternatively, retraining may be triggered once the accuracy of the model falls below a predetermined threshold, for example below 90 % or below 80 %.

For retraining the model, the sensor data and/or emission measurement data may be added to the historical dataset. The model may be trained from new with the thus enlarged historical dataset. This approach is likely to yield a well fit model, however, it may be computationally expensive and may require stopping the model execution for a while. To avoid such stops, the trained model may be retrained with only new datasets obtained from the sensors and/or emission measurement data. This approach needs less computational power and hence finishes faster. However, it involves the risk that the model “forgets” the historic dataset it was originally trained for. 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, often one wants to avoid such forgetting. There are various ways to execute retraining while lowering the impact of forgetting the historic datasets: The retraining may only allow small changes, for example by punishing large changes in the cost or loss function which is minimized during the retraining process. Alternatively, certain parts of the model may not be subject to any changes, for example if it has turned out that this part of the model yields results with high or sufficiently high accuracy.

In another aspect the present invention relates to a non-transitory computer-readable data medium storing a computer program including instructions for executing steps of the method according to the present invention. "Computer-readable data medium" refers to any suitable data storage device or computer readable memory on which is stored one or more sets of instructions (for example software) embodying any one or more of the methodologies or functions described herein. The instructions may also reside, completely or at least partially, within the main memory and/or within the processor during execution thereof by the computer, main memory, and processing device, which may constitute computer-readable storage media. The instructions may further be transmitted or received over a network via a network interface device. Computer-readable data medium include hard drives, for example on a server, USB storage device, CD, DVD or Blue-ray discs. The computer program may contain all functionalities and data required for execution of the method according to the present invention or it may provide interfaces to have parts of the method processed on remote systems, for example on a cloud system.

In another aspect the present invention relates to an emission control system. Such system may be configured to execute the method according to the present invention. Hence, all definition, examples and preferred embodiments described for the method also apply to the system.

The emission control system comprises an input configured to receive sensor data related to related to the chemical reaction in a first reactor. Such input may comprise an interface for receiving the sensor data. The input may receive the sensor data locally or remotely, for example via an interface to a telecommunication system, such as the internet. The input may receive the sensor data directly from the sensors, or via a programmable logic controller, a distributed control system, or a storage medium including a cloud service. It is even possible that the system is part of a distributed control system.

The emission control system further comprises a processor configured to determine at least one physical-chemical parameter. The processor may be a local processor comprising 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 to a remote computer system such as a cloud service.

The emission control system further comprises an output for outputting the operational instruction. Such output may comprise an interface for outputting the operational instruction. The output may send the operational instruction locally or remotely, for example via an interface to a telecommunication system, such as the internet. The output may send the operational instruction to a programmable logic controller, a distributed control system, or a storage medium including a cloud service. It is even possible that the system is part of a distributed control system.

Brief Description of the Figures

Figure 1 depicts the general setup in which an emission control system acts.

Figure 2 depicts the method of the present invention.

Figure 3 depicts a particular embodiment of the method of the present invention

Figure 4 depicts the emission control system of the present invention.

Figure 5 depicts decision tree for deciding on retraining of the model involved in the emission control system.

Figure 6 depicts an example for a plant employing the emission control system.

Description of Embodiments

Figure 1 shows an overview of the relevant components for the method including the emission control system. One or more than one reagent (101) is fed into a first reactor (102) in which it reacts to one or more than one product (105). The product includes any substance to be made by chemical reaction as well as any undesired but hardly avoidable by-product. At least one of the products (105), for example a by-product, is fed into a second reactor (106) in which it is converted into a compound of less environmental impact before the reaction products of from the second reactor (106) are released into the environment as emissions (107), for example through an exhaust to the air or into wastewater. The first reactor (102) is equipped with at least one sensor generating sensor data (103). The sensor data (103) may contain one or more than one sensor value measured at one or more than one points in time. The sensor data (103) may also contain an indicator of the sensor type and/or position for a sensor value and/or a timestamp indicating the time when the measurement was executed.

The sensor data (103) is transferred to the emission control system (104). The emission control system (104) receives the sensor data (103) and uses it to determine an operational instruction related to emission treatment in a second reactor (106). The operational instruction is output so it can be used to act on the second reactor. This can be indirect via a human operator taking the required action or directly by a controller, for example a valve to adjust the amount of a treatment reagent or a heater to adjust the temperature in the second reactor.

Figure 2 depicts the details of the method for controlling emissions of a chemical reaction. Sensor data related to the chemical reaction in a first reactor (201) are received. An operational instruction for emission treatment in second reactor (202) is determine based on the sensor data from the first reactor (201). The determination is usually executed by a model parametrized such that it receives at least part of the sensor data as input and outputs the operational instruction. Optionally, the sensor data from the first reactor (201) can be used to determine the concentration or level of one or more by-product (204) which needs to be treated before it can be released into the environment. The determination of one or more by-product (204) can be used to confirm the accuracy of the determination, for example by comparing the determined by-product concentration to data comprising measured by-product level (205). This comparison can be done continuously or only from time to time. The latter is advantageous if time-consuming laboratory analysis is required to determine the by-product concentration. Optionally, it is possible to use data comprising measured by-product level (205) as additional input for the determination of the operational instruction. This additional input may increase the accuracy of the determination of the operational instruction.

Figure 3 depicts a particular embodiment of the method. The sensor data from the first reactor (310) contains a time series of values obtained from sensors. A set of sensor values relates to the time t-2 (311), a set of sensor values relates to time t-1 (312) and a set of sensor values relates to time t (313). The times t-2, t-1 and t may be separated by a certain time interval, for example one second. Three sets were only chosen for the sake of illustrating this embodiment. The sensor data may contain more or fewer sets. The sensor data is used to predict sensor data (320) yielding predicted sensor data (330). For doing so, a first sub model can be employed receiving one or more than one sets of sensor values, for example a whole time series of sensor values, and outputs a set of sensor values for a future point in time. The predicted sensor data (330) may contain sensor values related to one time t+1 (331) in the future, for example for one or a few seconds to come. Optionally, the predicted sensor data (330) may contain sensor values related to more than one times, for example t+1 (332) and t+2 (330), i.e. a series of sensor values for the future are predicted. The predicted sensor data (330) may be used to determine an operational instruction (340). To achieve this, a second sub model may be used which uses the predicted sensor data (330) as input and the operational instruction as output. The second sub model may use a set of sensor values related to one time, e.g. set at t+1 (331), or it may use sets of sensor values related to different times, e.g. the set at t+1 (331) and the set at t+2 (332). In the latter case, the time series may be extrapolated. The time intervals between t, t+1 and t+2 may be the same or different to each other. These time intervals may be the same to the time interval between t, t-1 , t-2 and so forth, or different to these. Once the operational instruction is determined, it is output (350).

Figure 4 depicts the details of an emission control system (410). The emission control system (410) comprises an input (411) for receiving sensor data (402) which relate to a chemical reaction in a first reactor (401). The input (411) can be an interface to a storage medium onto which the sensor measurement values have been recorded or an interface to a communication medium for receiving the sensor data (402), for example a cable or a wireless communication connection. The processor (412) is configured to determine an operational instruction based on the sensor data (402). The processor (412) may be able to execute a model which is parametrized such that it requires at least part of the sensor data (402) as input and has the operational instruction as output. The processor may be a local processor, for example forming part of a computer, or it may be a remote computer center, for example a cloud service. The operational instruction is output by an output (413). The output (413) can be an interface to a storage medium onto which the operational instruction is written or an interface to a communication medium for transferring the operational instruction, for example a cable or a wireless communication connection. The operational instruction may be used to generate an operational signal (421) which causes adjustments for the second reactor (422) in order to control the emissions. Alternatively, the operational instruction may be displayed on a display, so an operator can perceive it and take any action based on the operational instruction.

Figure 5 shows an example for monitoring the accuracy of a model for determining an operational instruction based on the sensor data from the first reactor. The sensor data from the first reactor are received (501) and used as input for the model. The model may output a predicted emission level (502) together with an operational instruction to control the emission level. The emission level may be measured after execution of the operational instruction (503). The predicted emission level (502) and the measured emission level (503) may be compared. If the difference is within an acceptable range, the model may be further used without change (506). If the difference is outside an acceptable range, for example it exceeds a predetermined threshold, the model may be retrained (507). For this purpose, the sensor data from the first reactor (501) and the measured emission level (503) may be used as training data. Retraining may be executed instantaneously, i.e. without stopping the model, if only the new data is used for training and only small changes of the model are permitted. In this way, the model can be adjusted very fast to changed conditions. This approach is particularly useful if the model is a hybrid model containing known physical-chemical relationships which are immutable during the retraining. Overfitting can be effectively avoided.

Figure 6 shows an example of how the method and system of the present invention may be applied. The reagents (601) cumene and oxygen are charged into a first reactor (602). The first reactor (602) may be a solid bed column reactor with an immobilized oxidation catalyst. In the first reactor (602) cumene reacts with oxygen to 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). As CO is a gas, it can be easily separated from the liquid or solid products. CO is a toxic gas and cannot be released into the environment via an exhaust. Therefore, the CO is transferred to a second reactor (609). Oxygen is added to the second reactor (609) via a valve (607). The second reactor (609) may also be a solid bed column reactor with an immobilized oxidation catalyst. The CO is converted in the second reactor with oxygen and the oxidation catalyst to carbon dioxide (CO2), which can be released into the atmosphere as emissions (610). However, this reaction is not complete, its turnover depends on many parameters, in particular the ratio between CO and O2, but also on the temperature, the pressure, the catalytic activity and the average contact time between the CO and the O2 with the catalyst, which is influenced by the gas flow rate through the reactor. It is hence desirable to control these parameters such that the concentration of the CO in the emissions (610) is minimized.

In order to achieve this, the emission 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 may include sensors to determine the flow rate, for example one flow rate sensor at the inlet of the fist reactor (602) measuring the amount of reagent which flow into the first reactor (602) and one flow rate sensor at the outlet of the first reactor (602) measuring the amount of product and by-product flowing out of the first reactor (602). The sensors (603) may also include thermometers which measure the temperature inside the reactor. There may be several thermometers placed at different parts of the first reactor (602), for example at the inlet of the reactor, at the place where the reagents first come in contact with the oxidation catalyst, in a central part of the reactor, where the chemical reaction has partially progressed or at the end of the reactor where the products leave the first reactor (602). The sensor data may thus comprise a temperature profile along the flow direction of the reagents and products across the first reactor (602). Such sensor data may contain information about the reaction rate of the chemical reaction in the first reactor (602). The sensors (603) may also contain pressure sensors which measure the pressure inside the first reactor (602). There may be several pressure sensors placed at different parts of the first reactor (602), for example at the inlet of the reactor, at the place where the reagents first come in contact with the oxidation catalyst, in a central part of the reactor, where the chemical reaction has partially progressed or at the end of the reactor where the products leave the first reactor (602). The sensor data may thus comprise a pressure profile along the flow direction of the reagents and products across the first reactor (602). Such sensor data may contain information about the state of the catalyst and/or any plugging in the first reactor (602). The sensor data typically comprises a time series of sensor measurement values. This means that some or all sensors repeatedly do measurements and provide their values, for example after a predetermined time interval, for example once a second, or every time the measurement value changes by a predetermined amount. The sensor data may contain multiple entries, each entry comprising an identifier for identifying the sensor, a time stamp indicative for the time the measurement value was measured and the measured value.

Optionally, the sensor data may comprise physical-chemical values of a chemical reaction which precedes the chemical reaction in the first reactor. In this example, the reagent (601) cumene may be produced in another reactor by reacting benzene with propene. This reactor may be equipped with sensors measuring the substance flow into the reactor, the substance flow out of the reactor, the temperature and pressure at different locations. These sensor values may be comprised in the sensor data together with an identifier indicating that these values relate to the chemical reaction preceding the chemical reaction in the first reactor.

The emission control system (606) receives the sensor data. It is possible that the emission control system (606) instantaneously receives the measured value of each sensor once it is measured or it receives a set of sensor measurements at a predetermined time interval, for example every 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 emission control system (606). The controller may be part of a distributed control system for the first reactor (602). Alternatively, such controller may be part of the emission control system (606).

The emission control system (606) uses parts or all of the sensor data as input for a model which is parametrized according to the sensor data input. The model may be a hybrid model containing a kinetic equation for the reaction in the first reactor (602) as well as a kinetic equation for the reaction in the second reactor (609). Additionally, the hybrid model may contain a data-driven part, for example one or more than one neural networks. These neural networks may receive parts or all of the sensor data as input and output a parameter which may be added or multiplied to the kinetic equations. The output of the hybrid model is an operational instruction. The hybrid model has been trained with historic data, including, for example, sensor data, amount of treatment reagent and emissions composition. The operational instruction output by the model may be the amount of oxygen required at a certain time in the second reactor. The emission control system (606) thus determines an operational instruction based on the sensor data. It may send this operational instruction to the valve (607) which is adjusted according to the operational instruction, whereby the amount of treatment reagent (608), i.e. oxygen, flowing into the second reactor (609) is adjusted as determined by the model.

In addition, the model may output the concentration of by-product at a certain point in time. Such output may be compared to a measurement of the concentration of by-product at this particular point in time. The comparison may indicate the accuracy of the model. Also, the emissions may be measured and compared to 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) may be compared to this predetermined value and the comparison may yield the accuracy of the model. For example, the extent and the time the CO concentration exceeds the predetermined value may be a measure for the accuracy of the model. If the accuracy is below a certain value, a model retraining may be triggered. Such situations may occur due to differences in the quality of the reagents, changes of the catalyst activity or slow buildup of plugging in the reactors.

Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims.

For the processes and methods disclosed herein, the operations performed in the processes and methods may be implemented in differing order. Furthermore, the outlined operations are only provided as examples, and some of the operations may be optional, combined into fewer steps and operations, supplemented with further operations, or expanded into additional operations without detracting from the essence of the disclosed embodiments.

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 may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

Claims

1. A computer-implemented method for controlling emissions of a chemical reaction comprising

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

(c) outputting the operational instruction.

2. The computer-implemented method according to claim 1, wherein the operational instruction is determined to match a preset target value for the emissions.

3. The computer-implemented method according to any of the preceding claims, wherein determining an operational instruction involves determining a by-product level of the chemical reaction in the first reactor.

4. The computer-implemented method according to any of the preceding claims, wherein a second chemical reaction takes place in the second reactor to reduce emissions.

5. The computer-implemented method according to the preceding claim, wherein the operational instruction comprises the instruction to adjust a reaction parameter of the second chemical reaction.

6. The computer-implemented method according to the preceding claim, wherein the reaction parameter is the dosing of a treatment reagent used to react with the by-product of the chemical reaction in the first reactor.

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

8. The computer-implemented method according to any of the preceding claims, wherein the sensor data contains a time series of a physical-chemical value measured by at least one sensor.

9. The computer-implemented method according to any of the preceding claims, wherein determining the operational instruction employs a hybrid model which uses at least part of the sensor data as input value.

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

11. The computer-implemented method according to any of the preceding claims, wherein determining the operational instruction is based on a model which has been trained with a set of historic training data.

12. The computer-implemented method according to the preceding claim, wherein the model is continuously retrained with sensor data.

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

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

15. An emission control system comprising

(c) an output for outputting the operational instruction.