Determining operating conditions in chemical production plants

Abstract

A system and method for determining operating conditions of a chemical production plant including at least one catalytic reactor is provided. Operating data and a catalyst aging indicator are received (10) via a communication interface. At least one target operating parameter for the operating conditions for a predetermined production run or a current production run is determined (14). The at least one target operating parameter for the operating conditions can be used to monitor and / or control the chemical production plant.

Description

Technical Field

[0001] The present disclosure relates to systems, methods, and computer program products for determining operating conditions in a scheduled or current production run of a chemical production plant that includes at least one catalytic reactor. Background Art

[0002] Catalytic reactors are widely used in the chemical industry to convert raw materials into valuable chemicals. The performance of the catalyst, such as conversion, selectivity and yield, is related to the operating parameters of the reactor and the aging of the catalyst. In order to achieve certain performance targets, plant operators must adjust the reactor parameters based on experience and engineering knowledge. Typically, reactor models based on kinetics, heat transfer and mass transfer phenomena are used to describe the performance of the catalyst and achieve more direct control of the reactor operation. However, such models are complex, and the experimental determination of kinetics and transport parameters is very cumbersome and expensive.

[0003] Recently, hybrid models have been published that partially rely on process knowledge but also use machine learning methods. For example, Clough and Ramirez (AIChEJ 22, 1976, p. 1097) and Gujarati and Babu (Chem. Eng. Sci. 65, 2010, p. 2009) published a knowledge-based model for a styrene reactor and used it to optimize the reactor operation that maximizes styrene yield or process economy. In both cases, the kinetic rate equation and coefficients are taken from previous literature. Considering that the kinetic coefficients should be calculated separately for each actual catalyst (different catalysts have different reaction rates for all chemical conversions occurring in the system), the practical utility of the model is limited. In addition, the catalyst aging rate is not considered.

[0004] Tamsiian et al. (Comp. Chem. 40, 2012, p. 1) developed a model based on kinetic equations published in the literature and fitted the rate coefficients using data from a styrene plant that was operated for 5 days.

[0005] H. Lim et al. (Ind. Eng. Chem. Res. 43, 2004, p. 6441) developed a hybrid model consisting of a first principle (knowledge-based) part that describes the dependence of catalyst performance (conversion and selectivity) on reactor operating parameters pressure, temperature, STO ratio and ethylbenzene flow rate as input variables, and a neural network model that predicts the deactivation factor to be used in the first principle model. The neural network predicts an updated deactivation factor based on the deactivation factors of ethylbenzene and steam, inlet temperature, total pressure and partial pressures of the most recent time step.

[0006] Shahhosseini et al. (Int. J. Chem. React. Eng. 9, 2011) developed a hybrid model to optimize the performance of an adiabatic industrial reactor system. First, they fitted a kinetic model using 7 data points measured in an experimental isothermal reactor. Using the kinetic coefficients estimated in this way, they used a taboo search algorithm or a genetic algorithm to effectively optimize the operating conditions to maximize two objective functions: ethylbenzene conversion and styrene selectivity. Based on literature values, the catalytic activity was modeled as an exponential function of time, decaying from 100% to 40% in 48 months. Modeling deactivation in this way cannot reflect the impact of different operating scenarios, which is essential for operational and plant-specific deactivation predictions.

[0007] Wu et al. (Lecture Notes in Computer Science, 10357, pp.301-312, 2017) and WO2018 / 035718 A1 disclose a data-driven model for real-time prediction of styrene production based on real operating data. 33 sensor-based predictors were identified from a combination of principal component analysis (PCA) and random forest variable importance analysis. Three models, generalized regression neural network (GRNN), GRNN after Kalman filtering, and random forest regression model were compared to predict the yield of styrene monomer. Wu et al. focused on predicting current styrene production based on available sensor data, and the motivation for the statement was that the information was usually only available after analytical testing, which may take several hours. However, the method currently proposed does not involve performance prediction or performance forecasting based on user-defined operating scenarios, including styrene monomer yield at future time points. Catalyst deactivation is not considered.

[0008] Therefore, many knowledge-based models or kinetic models have been developed to model styrene production in catalytic reactors, and recently elements of machine learning (ML) have been integrated (hybrid models), but in some cases used to optimize the performance of the kinetic model rather than modeling it itself. In all hybrid modeling and machine learning approaches, the ML methods are trained on small data sets, usually consisting of a single experiment or a single run of a production plant.

[0009] Hybrid reactor models based on knowledge or kinetics are complex and often require simplifying assumptions to reduce computational workload. In most cases, not all physicochemical processes that contribute to catalyst performance and aging are understood. For example, in the case of styrene catalysts, potassium loss is controlled by the temperature gradients occurring along the catalyst bed depth and in the catalyst particles, as well as the STO ratio, the carbon dioxide content in the gas mixture (generated by coke gasification), and the pressure gradient along the bed, making it impossible to accurately estimate the potassium evaporation rate. The potassium loss rate at each point of the reactor bed cannot be calculated a priori to estimate the aging rate of the catalyst in the catalyst bed and during the catalyst life.

[0010] An object of the present disclosure is to provide a method for determining operating conditions in a chemical production plant comprising at least one catalytic reactor, which allows for robust, stable and reliable reactor operation and enhances process control in catalyst-based production plants. Summary of the invention

[0011] According to a first aspect of the present invention, there is provided a system for determining operating conditions of a chemical production plant comprising at least one catalytic reactor. The system comprises:

[0012] A communication interface and a processing device in communication with the communication interface, the system being configured to:

[0013] - receiving, via the communication interface, operating data indicative of predefined operating conditions of a predetermined production run, or measured operating data indicative of current operating conditions of a current production run, wherein at least one operating data point comprises an expected operating value indicative of a change in the current operating condition,

[0014] - receiving via the communication interface a catalyst aging indicator associated with a period of time that the catalyst has been used in a current or scheduled production run,

[0015] - determining, via the processing equipment, at least one target operating parameter for a predetermined production run or a changed operating condition of a current production run based on the operating data and the catalyst aging indicator using a data-driven model, preferably a data-driven machine learning model, wherein the data-driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including the operating data, the catalyst aging indicator and the at least one target operating parameter,

[0016] - providing via the communication interface at least one target operating parameter for a predetermined production run or a changed operating condition of a current production run.

[0017] According to another example of the first aspect of the present invention, a system for determining an operating condition of a chemical production plant including at least one catalytic reactor is provided. The system includes a communication interface and a processing device in communication with the communication interface.

[0018] (a) For a scheduled production run, the system is configured to:

[0019] - receiving, via the communication interface, operating data indicative of predefined operating conditions for a predetermined production run,

[0020] - receiving, via the communication interface, a catalyst aging indicator associated with a time period during which the catalyst has been used in a predetermined production run,

[0021] - determining, via the processing equipment, at least one target operating parameter for operating conditions for a predetermined production run based on the operating data and the catalyst aging indicator using a data-driven model, wherein the data-driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including the operating data, the catalyst aging indicator, and the at least one target operating parameter,

[0022] - providing via the communication interface at least one target operating parameter for the operating conditions of a predetermined production run, or

[0023] (b) For changes to the current production run, the system is configured to:

[0024] - receiving, via the communication interface, measured operating data indicative of current operating conditions of a current production run, wherein at least one operating data point comprises an expected operating value indicative of a change in the current operating conditions,

[0025] - receiving, via the communication interface, a catalyst aging indicator associated with a time period during which the catalyst has been used in a current production run,

[0026] - determining, via the processing equipment, at least one target operating parameter for a changed operating condition of a current production run based on the operating data and the catalyst aging indicator using a data driven model, wherein the data driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including the operating data, the catalyst aging indicator, and the at least one target operating parameter,

[0027] - providing via the communication interface at least one target operating parameter for the changed operating conditions of the current production run.

[0028] According to a second aspect of the present invention, there is provided a system for optimizing the operating conditions of a scheduled production run or a change in the current production run of a chemical production plant. The system comprises the above system and an optimization processing device, wherein the optimization processing device is configured to:

[0029] - receiving, via the communication interface, determined target operating parameters for more than one operating condition of a predetermined production run or a variation of a current production run,

[0030] - based on the received target operating parameters for each operating condition, determining via the optimization processing device a minimum value or a maximum value of the target operating parameter, or a minimum value or a maximum value of at least one optimization parameter derived from the target operating parameter,

[0031] - providing via the communication interface a minimum or maximum value indicative of an optimal operating condition for a predetermined production run or a current production run, such as a minimum or maximum value of a target operating parameter or a minimum or maximum value of an optimized parameter derived from the target operating parameter.

[0032] According to another example of the second aspect of the present invention, a system for optimizing the operating conditions of a chemical production plant is provided. The system comprises the above system and an optimization processing device, wherein the optimization processing device is configured to:

[0033] - receiving, via the communication interface, determined target operating parameters for more than one operating condition as determined (a) for a predetermined production run or (b) for a change in a current production run,

[0034] - based on the received target operating parameters for each operating condition, determining via the optimization processing device a minimum value or a maximum value of the target operating parameters, or a minimum value or a maximum value of an optimization parameter derived from the target operating parameters,

[0035] -Providing via the communication interface a minimum or maximum value indicative of optimal operating conditions in (a) a predetermined production run or (b) a current production run, such as a minimum or maximum value of a target operating parameter, or a minimum or maximum value of an optimized parameter derived from the target operating parameter.

[0036] According to a third aspect of the present invention, a production monitoring and / or control system is provided, comprising, for example, a communication interface coupled to a system for determining operating conditions as listed above or a system for optimizing operating conditions as listed above via a wired or wireless connection. The production monitoring and / or control system may include a display device configured to receive and display the determined operating conditions. The production monitoring and / or control system may include a control unit configured to receive the determined operating conditions to control the current or scheduled production operation in a chemical production plant based on the determined operating conditions. The determined operating conditions preferably include the determined target operating parameters and further optionally include the optional operating data of the desired operating value.

[0037] According to a fourth aspect of the present invention, there is provided a computer-implemented method for determining a change in operating conditions of a scheduled production run or a current production run of a chemical production plant comprising at least one catalytic reactor. The method comprises the following steps:

[0038] - receiving, via the communication interface, operating data indicative of predefined operating conditions of a predetermined production run, or measured operating data indicative of current operating conditions, wherein at least one operating data point comprises an expected operating value indicative of a change in the current operating condition,

[0039] - receiving via the communication interface a catalyst aging indicator associated with a period of time that the catalyst has been used in a current or scheduled production run,

[0040] - determining, via the processing equipment, at least one target operating parameter for a changed operating condition for a predetermined or current production run based on the operating data and the catalyst aging indicator using a data-driven machine learning model, wherein the data-driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including the operating data, the catalyst aging indicator, and the at least one target operating parameter,

[0041] - providing via the communication interface at least one target operating parameter for a changed operating condition of a predetermined or current production run.

[0042] According to another example of the fourth aspect of the present invention, there is provided a computer-implemented method for determining operating conditions of a chemical production plant including at least one catalytic reactor. The method comprises the following steps:

[0043] (a) For a predetermined production run, the method comprises the following steps:

[0044] - receiving, via the communication interface, operating data indicative of predefined operating conditions for a predetermined production run,

[0045] - receiving, via the communication interface, a catalyst aging indicator associated with a time period during which the catalyst has been used in a predetermined production run,

[0046] - determining, via the processing equipment, at least one target operating parameter for operating conditions for a predetermined production run based on the operating data and the catalyst aging indicator using a data-driven model, wherein the data-driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including the operating data, the catalyst aging indicator, and the at least one target operating parameter,

[0047] - providing via the communication interface at least one target operating parameter for the operating conditions of a predetermined production run, or

[0048] (b) For changes to the current production run, the method comprises the following steps:

[0049] - receiving, via the communication interface, measured operating data indicative of current operating conditions of a current production run, wherein at least one operating data point comprises an expected operating value indicative of a change in the current operating conditions,

[0050] - receiving, via the communication interface, a catalyst aging indicator associated with a time period during which the catalyst has been used in a current production run,

[0051] - determining, via the processing equipment, at least one target operating parameter for a changed operating condition of a current production run based on the operating data and the catalyst aging indicator using a data driven model, wherein the data driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including the operating data, the catalyst aging indicator, and the at least one target operating parameter,

[0052] - providing via the communication interface at least one target operating parameter for the changed operating conditions of the current production run.

[0053] According to a fifth aspect of the present invention, there is provided a method for optimizing the operating conditions of a scheduled production run or a change in the current production run of a chemical production plant. The method comprises the following steps:

[0054] - receiving, via the communication interface, determined target operating parameters for more than one operating condition of a predetermined production run or a variation of a current production run,

[0055] - based on the received target operating parameters for each operating condition, determining via the optimization processing unit a minimum value or a maximum value of the target operating parameters, or a minimum value or a maximum value of an optimization parameter derived from the target operating parameters,

[0056] - providing via the communication interface a minimum or maximum value indicative of an optimal operating condition for a predetermined production run or a current production run, such as a minimum or maximum value of a target operating parameter or a minimum or maximum value of an optimized parameter derived from the target operating parameter.

[0057] According to another example of the fifth aspect of the present invention, a method for optimizing a change in operating conditions of a scheduled production run or a current production run of a chemical production plant comprises the following steps:

[0058] - receiving, via the communication interface, determined target operating parameters for more than one operating condition as determined (a) for a predetermined production run or (b) for a change in a current production run,

[0059] - based on the received target operating parameters for each operating condition, determining via the optimization processing device a minimum value or a maximum value of the target operating parameters, or a minimum value or a maximum value of an optimization parameter derived from the target operating parameters,

[0060] -Providing via the communication interface a minimum or maximum value indicative of optimal operating conditions in (a) a predetermined production run or (b) a current production run, such as a minimum or maximum value of a target operating parameter, or a minimum or maximum value of an optimized parameter derived from the target operating parameter.

[0061] According to a sixth aspect of the invention, a method for monitoring and / or controlling a chemical production plant is provided, comprising the steps of performing a method for determining or optimizing operating conditions as listed above. The method may further comprise displaying the determined operating conditions on a display device and / or controlling a current or scheduled production run in the chemical production plant based on the determined operating conditions. The determined operating conditions preferably comprise determined target operating parameters and / or optionally comprise operating data of a desired operating value.

[0062] According to a seventh aspect of the present invention, there is provided a method for training a data driven model for determining the operating conditions of a scheduled production run or a change in the current production run of a chemical production plant comprising at least one catalytic reactor. The method comprises the following steps:

[0063] - receiving via a communication interface a training data set based on a historical data set including operating data, a catalyst aging indicator, at least one target operating parameter, and optionally plant metadata,

[0064] - training the data-driven model via the processing device by adjusting the parameterization according to the training data set,

[0065] -Providing the trained data-driven model via a communication interface.

[0066] According to an eighth aspect of the present invention, a computer program or a computer program product or a computer readable non-volatile storage medium comprising computer readable instructions, when loaded and executed by a processing device, performs the method disclosed herein.

[0067] According to a ninth aspect of the present invention, there is provided a catalyst comprising a catalyst type identifier associated with a data-driven model trained according to the methods listed herein, wherein the model is trained for a catalyst type indicated by the catalyst type identifier. In other words, there is provided a catalyst system comprising a catalyst and a catalyst type identifier associated with a data-driven model trained according to the methods listed herein, wherein the model is trained for a catalyst type indicated by the catalyst type identifier.

[0068] According to a tenth aspect of the present invention, there is provided a catalyst comprising a catalyst type identifier associated with a computer program listed herein. In other words, there is provided a catalyst system comprising a catalyst and a catalyst type identifier associated with a computer program listed herein.

[0069] According to an eleventh aspect of the invention, there is provided a chemical process using a catalyst associated with a data-driven model trained according to the methods outlined herein, wherein the data-driven model is used to design plant components and optimize the operation of the chemical process to achieve target performance, in particular the operating conditions of a chemical production plant including at least one catalytic reactor.

[0070] The following disclosure applies to the systems, methods, computer programs, computer-readable non-volatile storage media, catalysts, chemical processes, and computer program products disclosed herein, among others. Therefore, no distinction is made between systems, methods, computer programs, computer-readable non-volatile storage media, or computer program products. All features are disclosed in relation to the systems, methods, computer programs, computer-readable non-volatile storage media, catalysts, chemical processes, and computer program products disclosed herein.

[0071] The present invention provides a system or method for determining operating conditions or operating parameters based solely on a data-driven model, which allows for more robust, stable and reliable reactor operation and enhances process control in a catalyst-based production plant. In particular, the determination is more accurate than known methods based on kinetic models or hybrid models because no prior information about the reaction kinetics is required, and therefore no estimates or simplifying assumptions about the underlying process are required.

[0072] Specifically, the determination takes into account catalyst deactivation by providing a catalyst aging indicator associated with the period of time that the catalyst has been used in the current production run. Including the catalyst aging indicator as a model input parameter allows for a more accurate determination of operating conditions because the model inherently takes into account catalyst deactivation or aging. The determination of operating parameters taking into account catalyst deactivation is applicable to a wide range of chemical production plants and operating parameters. In contrast, data-driven models that do not take into account catalyst aging are only narrowly applicable. In particular, such models are limited to the chemical production plant for which the model was trained, and sometimes even only run once in the production plant, and the range of operating conditions is also narrow.

[0073] The systems and methods of the present disclosure perform short-term discrete predictions as well as time series forecasts. In the latter case, they are able to cover the complete deactivation process of the catalyst during the entire production run. The methods and systems enable plant operators to improve and optimize operating strategies on a daily basis based on energy costs, market supply of raw materials or demand for plant products, and other constraints that may occur in the plant (such as power outages in different plant parts or utilities).

[0074] Furthermore, the system and method allow for the short-term behavior of a catalyst-based production plant to be predicted and / or foretell its long-term behavior using the plant's operating data and catalyst aging indicators. This allows for enhanced process control in a catalyst-based production plant, as, for example, a plant operator can easily assess scheduled or current operations based on the operating conditions existing in the production plant. Thus, the system provides a powerful tool for planning, monitoring and controlling production processes.

[0075] In one embodiment of the invention, the chemical production plant comprises one or more catalytic reactors. In the context of the present invention, the term "catalytic reactor" refers to a chemical reactor in which a catalytic chemical reaction occurs and which generally contains a catalyst. The catalytic reactor may be a fixed bed catalytic reactor. The chemical production plant may be a styrene production plant.

[0076] In this context, a data-driven model, preferably a data-driven machine learning model or simply a data-driven model, refers to a trained mathematical model that is parameterized according to a training data set to reflect the reaction kinetics or physicochemical processes of a chemical production plant or a catalytic reactor. An untrained mathematical model refers to a model that does not reflect the reaction kinetics or physicochemical processes, such as an untrained mathematical model that does not originate from physical laws that provide scientific generalizations based on empirical observations. Therefore, kinetics or physicochemical properties may not be inherent to an untrained mathematical model. An untrained model does not reflect such properties. The parameterization of an untrained mathematical model can be achieved by feature engineering and training using a corresponding training data set. The result of this training is only a data-driven model, preferably a data-driven machine learning model, which reflects the reaction kinetics or physicochemical properties as a result of the training process, preferably only as a result of the training process.

[0077] Historical data in this context refers to a data set comprising at least operational data, catalyst aging indicators and at least one target operating parameter, wherein each data set is associated with a single production run. Thus, each data set comprises data associated with a production run of a catalyst from the start of production to the end of production. When the catalyst needs to be replaced again, such data can be measured and recorded during the production run during the life of the catalyst (e.g., from the start of production after the catalyst replacement to the end of production).

[0078] The chemical reaction catalyzed does not limit the scope of the present invention. As an example, the catalyst can be a dehydrogenation catalyst. In particular, the catalyst of the catalytic reactor can be an iron oxide-based catalyst for the dehydrogenation of aliphatic or alkyl aromatic hydrocarbons to form corresponding unsaturated hydrocarbons. Examples of such dehydrogenation processes are the dehydrogenation of ethylbenzene to styrene, the dehydrogenation of isopropylbenzene to α-methylstyrene, the dehydrogenation of butylene to butadiene or the dehydrogenation of isopentene to isoprene. The method and system are particularly useful in styrene production plants. Preferably, the chemical production plant is a styrene production plant that converts ethylbenzene into styrene using a styrene catalyst. The preparation of styrene generally includes heterogeneous catalytic dehydrogenation of ethylbenzene in the presence of steam. The catalytic dehydrogenation of aliphatic or alkyl aromatic hydrocarbons is generally performed industrially in the presence of steam at a temperature range of 500 to 700° C. In these processes, hydrocarbons and steam are generally mixed at high temperature and low pressure and passed through an iron oxide dehydrogenation catalyst.

[0079] The term operational data refers to quantities indicative of the operational state of a chemical production plant. In particular, such quantities relate to measurement data collected during production operation of the chemical production plant and may be derived directly or indirectly from such measurement data. In a preferred embodiment, the operational data comprises sensor data measured by sensors installed in the chemical production plant, quantities derived directly or indirectly from such sensor data, analytical data measured in samples taken from the chemical production plant, quantities derived directly or indirectly from such analytical data, or a combination thereof.

[0080] Sensor data may include measured quantities available in a chemical production plant by means of installed sensors (e.g., temperature sensors, pressure sensors, flow sensors, etc.). Analytical data may include quantities provided by analytical measurements of samples extracted from a chemical production plant at any point in a process or time. In particular, such analytical data may include the composition of reactants, starting materials, products, and / or byproducts as determined, for example, from samples extracted during a production process at different stages of the production process (e.g., before or after a catalytic reactor) via gas chromatography. Analytical data preferably forms the basis for determining catalyst performance characteristics.

[0081] The operational data set may include raw data, which refers to basic, unprocessed analytical and / or sensor data; or processed or derived parameters derived directly or indirectly from the raw data. In the case of a chemical production plant, the derived parameters may include the average inlet temperature of a plurality of catalytic reactors derived from the corresponding temperature sensors; the steam-to-oil ratio derived from raw data of steam flow rate and reactant flow rate; the conversion and selectivity derived from analytical data before and after the reactor; any type of normalized data, such as production values ​​normalized by catalyst volume or catalyst mass; any data derived from time series data, such as cumulative production, maximum load to date, or any combination thereof.

[0082] Particularly for chemical production plants, conversion, selectivity and yield can be derived from analytical data. The conversion here refers to the fraction of reactants completely converted in the reactor, preferably a percentage. In the case of, for example, styrene production, this corresponds to the conversion of the starting material ethylbenzene to any product. The selectivity of the desired product refers to the number of conversion reactants converted into the desired product. In the case of, for example, styrene production, this corresponds to the selectivity of ethylbenzene to styrene reaction. The yield of the desired product refers to the mathematical product of conversion and product-specific selectivity. The yield can be represented by the percentage of the reactants entering the reactor converted into the desired product. In another embodiment, the plant metadata indicating the physical plant layout is received via a communication interface. The plant metadata can include a plant-specific amount describing the characteristics of, for example, a reactor, which is predefined by the physical plant layout and can be related to the performance of the plant or reactor. For example, the plant metadata includes multiple reactors that the reactant mixture passes through subsequently, such as 2 or 3 reactors, the total catalyst volume, the catalyst volume of the reactor, the size (length, diameter, height ...) of each reactor, the catalyst type used in the plant, or a combination thereof. In another embodiment, determining, via the processing unit, at least one target operating parameter is additionally based on plant metadata using a data driven model, wherein the training data set is based on a historical data set additionally including the plant metadata. Incorporating plant metadata into the determination of operating conditions allows for building data driven models applicable to different plants, which in turn increases the number of data points available for training the data driven model. Thus, the data driven model broadly captures the operating conditions of different plants operating under different operating conditions with different physical plant layouts, which allows for more accurate determinations.

[0083] In another embodiment, the set of historical data includes data from multiple production runs, multiple factories and / or multiple catalyst batches of the same type of catalyst. Incorporating multiple production runs into training allows different operating conditions of the same or different factories to be covered. Including data from multiple catalyst batches allows differences between catalyst batches to be taken into account. Data from multiple catalyst batches can include data from multiple production runs, wherein for at least one production run, different production batches of the same type of catalyst are used. By including multiple factory data from one or more production runs from different factories, it can be included. Therefore, operating conditions in different factories can be covered, thereby providing a wider range of model applicability. In this context, the same type of catalyst refers to the same type of catalyst formulation. Multiple catalyst batches include the same type of catalyst provided from different manufacturing batches or different delivery dates.

[0084] In another embodiment, the catalyst aging index is based on a time point, a time period, an amount derived from time-related operating data, and / or an amount derived cumulatively from time-related operating data. The catalyst aging index can be specified by the time spent in the reactor by the catalyst under reaction conditions since it first contacts the reaction mixture. Additionally or alternatively, the cumulative load or cumulative production of the catalyst can be used as a catalyst aging index, which is preferably defined by the total amount of reactants fed or converted from the start of the run until the predicted start time point. In addition to the indicators mentioned here, any other amount that can be used as a catalyst aging index can be used. The catalyst aging index can be provided via a client device, wherein the factory operator inputs the time when the production run starts or the time period from the start of the production run until the predicted start time point. Alternatively or additionally, the catalyst aging index can be determined based on the time series of operating data preferably from the start of the production run, wherein the production run starts can be determined based on the operating profile in the operating data. This operating profile can include certain temperature, pressure, flow profiles or a combination of such profiles.

[0085] The operating conditions for predetermined or current production changes refer to the operating conditions that a chemical production plant may run in the future or after the predicted starting time point. Such operating conditions may include one or more discrete time points, multiple discrete time points, one or more operating conditions of a period of time or several time intervals. In the latter case, the operating conditions may include different time intervals, for which at least one predefined or expected operating value adopts different values ​​in different intervals. Multiple discrete time points, time periods or time intervals can extend from the predicted starting time point to the remaining production operation until the production operation ends, when the catalyst needs to be replaced.

[0086] The operating conditions for a predetermined production run may refer to the operating data of the predefined operating conditions of a chemical production plant before the production run begins. The predefined operating conditions may include a set of predefined operating data specifying the operating conditions. Determining the operating conditions for a predetermined production run is particularly useful for production design and planning before the production run begins. The operating conditions for a current production run may refer to the variation of the operating conditions currently set in a chemical production plant. The operating conditions may include a set of operating data specifying the current operating conditions, wherein the operating data is located in at least one operating data point different from the operating parameters currently set in the chemical production plant. Determining the operating conditions for the current production run is particularly useful for monitoring and controlling production during the current production run.

[0087] Determining operating conditions includes predicting or forecasting the performance / behavior of catalyst or reactor, and is used to describe the application of data-driven model to a set of appropriate input parameters in a general manner. In the example of a styrene production plant, for example, this determination refers to the determination of the associated selectivity of the reactor inlet temperature and ethylbenzene to the reaction of styrene under the given specific operating conditions via operating data. This selection of the output parameters of the model may be driven by the fact that most factories operate in a manner that the operator regulates the inlet temperature to achieve the desired conversion. However, given the same data set, a model can be developed in a similar manner, as described in further detail below, with different input and output allocations, such as using the reactor temperature as input to predict the conversion rate expected under operating conditions.

[0088] In one embodiment, the determination of at least one target operating parameter for an operating condition is based on a short-term model that determines the target operating parameter at a discrete time point or a long-term model that determines the target operating parameter for a period of time (particularly the future). In the case of a short-term model, at least one target operating parameter for an operating condition is determined based on a short time frame. Here, a short time frame can refer to a discrete or single time point. The short-term model can be based on different machine learning techniques, including, for example, regression models such as linear regression models, nonlinear regression models, Bayesian linear regression, random forest models, neural networks, or combinations thereof. Other methods can also be applied. Preferably, the short-term model does not have inherent time correlation. More preferably, the short-term model predicts the performance or operating condition at time t based on the model input at the same time t.

[0089] In the case of a scheduled production run, the operating data may specify predefined operating conditions. For example, under the assumption that the predefined operating conditions are constant over time or have a predefined variation over time or based on previous production runs, the catalyst aging indicator may be estimated for the scheduled production run. The determination via the short-term model may, for example, be applied to more than one discrete time point for maintaining constant predefined operating conditions throughout the production run, and the catalyst aging indicator may be estimated for each of the discrete time points. This implementation is particularly advantageous for designing and planning future production runs.

[0090] In the case of a current production run, the operating data may specify operating conditions currently set in the chemical production, wherein at least one expected operating value indicates a change in the current operating conditions or a deviation from the current operating conditions. The determination via the short-term model may, for example, be applied to one or more discrete time points for determining the impact of at least one expected operating value indicating a deviation from the operating conditions currently set in the chemical production. Such an implementation is particularly advantageous for monitoring and controlling the current production run.

[0091] In the case of a long-term model, at least one target operating parameter for operating conditions is determined based on a long time frame. The long time frame here can refer to multiple time points in the future. The number of points and the prediction range therefore depend on the time scale of the time dynamics in the production plant. For a production process based on a catalyst, this dynamic can be determined by the catalyst aging dynamics and its time range. For a heterogeneous catalyst reaction like styrene production, this time scale may be in the range of weeks, months or years. The long-term model can be based on a time series forecasting method. This method, for example, includes such as autoregressive models, particularly autoregression (AR), moving average (MA), autoregressive moving average (ARMA), autoregressive integrated moving average (ARIMA), vector autoregression (VAR), vector autoregressive moving average (VARMA), known regression methods with exogenous regressors (VARMAX), random forest models, neural networks, convolutional neural networks, recursive neural networks or combinations thereof. Other methods can also be applied.

[0092] In one embodiment, a time series of at least one target operating parameter as measured, predicted or derived during the current or previous production run until the prediction start time point is received via a communication interface. For example. In the case of a scheduled run, a target operating parameter as measured, predicted or derived during the previous production period can be used. Preferably, a data-driven model is used to determine at least one target operating parameter at one or more time points after the prediction start time point based on operating data that optionally includes an expected operating value, a time series of at least one target operating parameter, a catalyst aging index, and optional factory metadata. More preferably, the data-driven model includes inherent time correlation. In this embodiment, at least one target operating parameter may include an uncontrolled or endogenous parameter that cannot be controlled via machine settings in a chemical production plant. On the contrary, the operating data including at least one expected operating value may include a controlled or exogenous parameter that can be controlled via machine settings in a chemical production plant. The determination of the target operating parameters at the forecast start time t may include, for example, determination for times t,…,t+N, N>0, based on uncontrolled parameters or a subset of uncontrolled parameters, up to time point t-1 or shorter, and optionally based on controlled parameters or a subset of controlled parameters, for example at time points t,…,t+N (optionally further including, for example, t-1, t-2,...) or at other time points suitable for forecasting using the specific structure of the selected model.

[0093] Preferably, the long-term model is a time series based model that includes inherent time correlations and forecasts target operating parameters for time t, ..., t+N based on model inputs at time points up to time point t-1. Here t refers to the prediction starting point in time. Therefore, the long-term model allows the determination of reactor or catalyst performance and changes in reactor or catalyst performance over a preferably extended period of time without the need for available information on uncontrolled parameters beyond the prediction starting point. These models have at least some inherent time correlations and forecast performance for time t, ..., t+N based on uncontrolled parameters up to time point t-1 and optionally based on controlled parameters (e.g., at time point t, ..., t+N or other time points suitable for prediction using the specific structure of the selected model).

[0094] In one embodiment, the processing device is further configured to pre-process the operating data prior to determination via the data driven model. Preferably, the pre-processing includes transformations of quantities that are independent of the physical plant layout. The pre-processing allows for accounting for systematic and non-systematic differences between different plants. Thus, even if data from the specific plant in question was not used to train the data driven model, it can be broadly applicable to the determination of target operating parameters. In particular, the transformation includes systematic factors as input parameters for the data driven model and / or normalization of the operating data.

[0095] In another embodiment, the data-driven model is verified before determining at least one target operating parameter. This verification enhances the interpretability and credibility of the determined target operating parameter. In order to verify, operating data and at least one target operating parameter measured or derived for one or more time points during the current production run can be received via a communication interface. The operating data can be used to determine at least one target operating parameter corresponding to the time point, wherein the measured target operating parameter is available. The results of the determined target operating parameters for a certain time point and the measured operating parameters corresponding to the same time point can be compared. If the comparison results in an effective model operation, such as if the difference is less than a threshold value or a systematic error is not identified, then the determination of at least one target operating parameter for operating conditions based on a set of operating data including at least one expected operating value can be followed. If the comparison results in an invalid model operation, an alarm may be triggered, such as via a display device or an audio device, indicating that the model operation is not suitable for monitoring and / or controlling the target chemical plant under current operating conditions.

[0096] In another embodiment, before determining at least one target operating parameter, a data driven model is selected based on a catalyst type. The catalyst type can be received via metadata, which indicates the type of catalyst used, for example, via a catalyst type identifier. The catalyst type can specify a catalyst recipe. This option allows highly flexible use of systems, computer programs, computer program products, and methods, not only covering different chemical production plants, but also covering different catalyst types. By selecting a data driven model based on input and output parameter identifiers before determining at least one target operating parameter, a further degree of flexibility can be added. The input and output parameter identifiers can specify which parameters are used as operating data and which parameters are used as target operating parameters in a system or method. In this implementation, based on the selection steps that can be performed, different data driven models can be stored in the memory of the system, including catalyst types and / or parameter identifiers for each model.

[0097] In one embodiment, training of a data-driven model is performed based on training data including a catalyst type identifier. The trained model can be stored with the catalyst type identifier and can be provided with a catalyst of that type that preferably includes the catalyst type identifier. This identification can be achieved electronically via a catalyst ID stored in a database or, for example, in a mobile storage medium attached to a catalyst delivery container in conjunction with a corresponding chemical production plant. In this case, the model is preferably trained based on a training data set measured in a production run that uses a catalyst of the same catalyst type indicated by the catalyst type identifier. Therefore, the catalyst can be bundled with the data-driven model and can help to more robustly control a production plant that operates with a catalyst of a specific catalyst type.

[0098] Provided is a method for predicting the short-term behavior of catalyst and predicting the long-term behavior of catalyst as a function of reactor operating parameters or operating data and catalyst aging or catalyst aging index in a fixed-bed catalytic reactor. The method uses a data-driven model, preferably a data-driven machine learning model, which does not involve prior information about reaction kinetics. The model can predict both the short-term and long-term behavior of the catalyst as a function of input parameters (comprising typical reactor operating parameters or operating data and parameters derived from available analytical raw data in sensors and production plants). A software product for performing the method is also provided. As an application example, the method is used to predict and predict the behavior of the catalyst and process reactor that ethylbenzene is converted into styrene.

[0099] The present disclosure provides a computer-implemented method for predicting the short-term performance of a catalyst in a chemical production plant including a catalytic reactor and forecasting its long-term performance (including catalyst aging effects). The method involves using a mathematical model of the chemical production plant and in particular the catalytic reactor, the mathematical model is based on machine learning, does not involve a priori information on reaction kinetics, and uses input parameters selected from sensor raw data, derived parameters, reactor operating parameters or operating data, plant metadata, and parameters indicating catalyst aging or catalyst aging indicators.

[0100] In one embodiment, the operational data is selected from sensor data obtainable from a chemical production plant, in particular a catalytic reactor, analytical data from, for example, gas chromatography (GC) analysis, and derived parameters as listed above.

[0101] In one embodiment, the operating data includes the inlet temperature and outlet temperature of one or more catalytic reactors (preferably each of the one or more catalytic reactors), the inlet pressure and outlet pressure of one or more catalytic reactors (preferably each of the one or more catalytic reactors), and the composition of the reaction mixture at the inlet and outlet of the one or more catalytic reactors (preferably each of the one or more catalytic reactors).

[0102] In one embodiment particularly applicable to a styrene production plant, the operating data includes steam to oil ratio (STO), liquid hourly space velocity (LHSV), total styrene production normalized by catalyst volume, target ethylbenzene conversion, styrene selectivity, average inlet temperature, normalized pressure after the last catalytic reactor, normalized pressure drop across one or more catalytic reactors, temperature loss across one or more catalytic reactors, and normalized deviation of temperature loss compared to expected values ​​(calculated based on the conversion). The advantage of this parameter set is that it remains interpretable because many parameters correspond to actual operating parameters or can be easily interpreted in this context. Other methods that reduce the number of problem dimensions (e.g., PCA or RFA) can produce parameters that are useful for the modeling process and prediction accuracy, but generally lack interpretability.

[0103] The data-driven model preferably performs time series forecasting and is able to cover the complete deactivation process of the catalyst during the entire production run in the reactor system. The model enables the plant operator to improve and optimize the operating strategy on a daily basis based on energy costs, market supply of raw materials or demand for plant products, and other constraints that may arise in the plant (such as outages of different plant parts or utilities).

[0104] In one embodiment, the forecast period spans the remaining life of the catalyst determined by the limits of the operating conditions (preferably the maximum reactor temperature that can be operated). The model allows the remaining life of the catalyst in use to be determined based on the forecasted operating conditions and the limits of these operating conditions (e.g., the maximum reactor temperature or pressure that can be operated). This enables the operator to reliably plan the time remaining until the next catalyst change, and also simulate different operating strategies in order to extend the catalyst life (if necessary), or to obtain maximum productivity from the remaining time until the next planned plant shutdown.

[0105] In one embodiment, the data driven model is used to predict operating conditions of a catalyst in a production plant that does not provide historical data for training the data driven model.

[0106] In one embodiment, the output of the data-driven model is used to optimize the operating conditions of the scheduled production operation or the current production operation of the chemical production plant. In such an embodiment, the target operating parameters determined for more than one operating condition via the data-driven model are received and fed into the optimization process equipment. Optimization may include one or more optimization targets. The optimization target may be specified by, for example, an optimization parameter or a target parameter to be optimized. In this context, the optimization target may further include finding the minimum or maximum value of the specified optimization parameter or target parameter. In addition to the target parameter for more than one operating condition, the optimization target may be received and fed into the optimization process equipment. For example, the selection of possible optimization targets may be provided to the operator and the selected optimization target may be received based on the user selection.

[0107] In the case of one or more optimization objectives, an optimal solution may exist and can be provided as a result of the optimization. This optimal solution can be provided to a system for monitoring and / or controlling. The optimal solution can be displayed on a display device or used to control a chemical production process. In the case of more than one optimization objective or multi-objective optimization, one or more optimal solutions may exist and can be provided as a result of the optimization. Multiple optimal solutions can be provided to a system for monitoring and / or controlling and displayed on a display device. In this case, the operator of the plant can choose between multiple optimal solutions, which simplifies the decision-making process in the complex situation of operating a chemical production plant.

[0108] In another embodiment, the optimization processing device determines the minimum or maximum value of the target operating parameter or the minimum or maximum value of at least one optimization parameter derived from the target operating parameter based on the target operating parameter as output by the data-driven model. In an exemplary scenario, the remaining life of the catalyst can be an optimization parameter derived from the target operating parameter and finding the maximum value is the goal of the optimization process. In addition, constraints can be included in the optimization problem. Restrictions on the target operating parameters may include the highest reactor temperature, the lowest selectivity, or the lowest production per day. In another exemplary scenario, the production during the remaining life of the catalyst or until the predetermined time point for catalyst replacement can be an optimization parameter derived from the target operating parameter and finding the maximum value is the goal of the optimization process. Constraints on the target operating parameters may include the maximum reactor temperature, the minimum selectivity, or the minimum and optional maximum production per day.

[0109] In another exemplary scenario, multiple objectives can be combined. For example, during the catalyst life, the remaining life of the catalyst under the production volume can be an optimization parameter derived from the target operating parameter, and finding the maximum remaining life combined with the maximum production volume is the goal of the optimization process. Constraints on the target operating parameters can include maximum reactor temperature, minimum selectivity, or minimum production volume per day. In another example, multiple objectives further include the optimal time for catalyst replacement that further considers cost aspects (such as remaining production, energy demand, catalyst replacement costs, or a combination thereof). For multi-objective optimization, known Pareto optimization techniques can be used.

[0110] In one embodiment, a data-driven model is used to simulate expected catalyst performance, productivity (e.g., the total amount of styrene produced over a period of time), energy requirements (associated with, for example, reactor or steam reactant heating), and costs or profits (e.g., related to reactant costs, market prices of products, energy costs) for selecting operating data or operating parameter sets in a catalytic reactor. Here, output, energy consumption, CO2 emissions, costs, byproducts, catalyst replacement intervals, remaining life, or a combination thereof can be a target operating parameter or an optimization parameter derived from the target operating parameter. Different constraints can be defined in the optimization, such as constraints on the operating data on which the determination of the target parameter is based.

[0111] In one embodiment, a data-driven model is used to simulate expected catalyst performance, productivity, and energy requirements for selected operational data sets that are not practically achievable in a given plant setup (e.g., lower pressure levels (deeper vacuum), lower STO ratios, or additional reactors). This can help plant managers better evaluate the economics of potential upgrades to the plant.

[0112] In one embodiment, the data driven model is used to simulate and / or assess the expected performance and operating conditions of a new plant that has not previously used a catalyst.

[0113] The present disclosure also provides a computer program product configured to perform the method of the present disclosure. In one embodiment, the computer program product is a computer program implemented in a chemical production plant or a styrene production plant, in particular in a computing unit (computer) integrated therein and / or connected thereto. In one embodiment, the computer program product is integrated into a dashboard of the chemical production plant or the styrene production plant.

[0114] In one embodiment, the computer program product comprises an interface for inputting operating conditions of a chemical production plant. The operating conditions may be historical data of previous production runs or actual operating conditions of a chemical production plant in current operation.

[0115] In one embodiment of the computer program product, the input parameters or operating data for the forecast are manually uploaded into the computer program product and / or read out from the process control system by the computer program product. In one embodiment, the input parameters or operating data are provided in a formatted data table. In one embodiment, a graphical user interface is provided for manually uploading data into the computer program product. In another embodiment, an application programming interface is provided for uploading data into the computer program product.

[0116] In one embodiment, a user of the computer program product provides the computer program product with all of the information specified above (e.g., raw data, corresponding units, plant metadata). This information may be entered manually, uploaded via a structured data file, or provided via an application programming interface (manually or automatically). For long-term forecasts, the input time series of target operating parameters preferably cover at least the range required by the time lag structure used in the model. Control parameters may also be included if their time lags are also used in the model.

[0117] In one embodiment, exogenous operating parameters are input by a user, for example for a plurality of operating scenarios whose performance is to be predicted.All additional data processing (including formatting, aggregation and prediction) is performed by the software product.

[0118] In one embodiment, the forecast generated by the computer program product is presented to a user via a graphical user interface, as a structured data file or via an application programming interface.

[0119] According to a further embodiment, the computer program product is a computer program product that, when loaded into a memory of a computing device and executed by at least one processor of the computing device, performs the steps of the above-described computer-implemented method.

[0120] The computer program product may be used with or incorporated into a computer system, which may be a stand-alone unit or include one or more remote terminals or devices that communicate with a central computer via a network, such as, for example, the Internet or an intranet. Thus, the computers or processors and related components described herein may be part of a local computer system or a remote computer or online system, or a combination thereof. Any databases and computer program products described herein may be stored in a computer internal memory or in a non-transitory computer readable medium.

[0121] Another aspect of the present disclosure is a computer system for predicting the performance of a catalyst in a chemical production plant or for determining the operating conditions of a chemical production plant. The computer system includes: at least one interface component, which is configured to access and read operating parameters or operating data and catalyst-specific parameters, in particular catalyst aging indicators; and a processor unit, which implements a data-driven model and is configured to predict the performance of the catalyst by providing reactor operating parameters or operating data and catalyst-specific parameters provided via the interface component to the data-driven model. In one embodiment, the computer system is configured to be coupled to a chemical production plant including a catalytic reactor via a wired and / or wireless communication connection, and to automatically access and read out reactor operating parameters or operating data and / or catalyst-specific parameters from a process control system of the chemical production plant including the catalytic reactor via the interface component at least in part.

[0122] Another aspect of the present disclosure is a computer-implemented method for training a machine learning-based data-driven model for a chemical production plant including one or more catalytic reactors, the data-driven model being used to predict or forecast the performance of a catalyst in the catalytic reactor. The method includes providing a mathematical model as an initial basis; providing historical data, such as multiple production runs from the same type of catalyst and / or multiple production runs from multiple chemical production plants or catalytic reactors including the same type of catalyst, and optionally predetermined target operating parameters or operating and performance parameters of the chemical production plant (and in particular the catalytic reactor); accessing and importing the provided historical data into the mathematical model; adapting the parameterization of the data-driven mathematical model to the provided historical data; providing an update of the data-driven mathematical model based on the adapted parameterization; and iteratively repeating the method steps by setting the updated data-driven mathematical model as the initial basis.

[0123] Another aspect of the present disclosure is a computer-implemented method for determining operating and / or performance parameters or target operating parameters of a chemical production plant including one or more catalytic reactors. The method includes accessing sensor data indicating operating conditions existing in the reactor system and catalyst-specific sensor data indicating a catalyst currently used in the reactor system, determining the operating and / or performance parameters or target operating parameters of the chemical production plant using a data-driven model, wherein the data-driven model is parameterized based on a training data set, wherein the training data set includes, for example, historical data from multiple production runs of the same type of catalyst and / or from multiple production runs in multiple catalytic reactors including the same type of catalyst and previously determined operating and performance parameters or target operating parameters, and providing the determined operating and / or performance parameters or target operating parameters of the chemical production plant including one or more catalytic reactors.

[0124] Another aspect of the present disclosure is a computer-implemented method for determining operating and / or performance parameters of a chemical production plant including one or more catalytic reactors. The method includes accessing sensor data indicative of operating conditions present in the reactor system and catalyst-specific sensor data indicative of a catalyst currently used in the reactor system, determining operating and / or performance parameters or target operating parameters of the chemical production plant using a data-driven model, wherein the data-driven model is parameterized based on a training data set, wherein the training data set includes, for example, historical data from multiple production runs of the same type of catalyst and / or multiple production runs in multiple reactor systems including the same type of catalyst and previously determined operating and performance parameters or target operating parameters of the chemical production plant, and providing the determined operating and / or performance parameters or target operating parameters of the chemical production plant.

[0125] Another aspect of the present disclosure is a control system for controlling a chemical production plant including one or more catalytic reactors. The control system includes a computer system as described above and a control unit configured to control actual and / or scheduled production running in the chemical production plant based on provided operating and / or performance parameters or target operating parameters of the chemical production plant.

[0126] The computer program may be stored and / or distributed on a suitable medium, such as an optical storage medium or solid-state medium provided together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems.

[0127] However, the computer program may also be presented over a network like the World Wide Web and can be downloaded into the working memory of a data processor from such a network.

[0128] According to a further exemplary embodiment of the present invention, a data carrier or a data storage medium for making a computer program available for downloading is provided, the computer program being arranged to perform a method according to one of the aforementioned embodiments of the present invention.

[0129] It should be understood that the embodiments described herein are not mutually exclusive of each other and that one or more of the described embodiments may be combined in various ways as would be appreciated by one of ordinary skill in the art.

[0130] The computer program for executing any method of the present invention can be stored on a computer-readable storage medium (e.g., a non-transitory computer-readable storage medium). The computer-readable storage medium can be a floppy disk, a hard disk, a CD (optical disc), a DVD (digital versatile disk), a USB (universal serial bus) drive, a RAM (random access memory), a ROM (read-only memory), and an EPROM (erasable programmable read-only memory). The computer-readable medium can also be a data communication network that allows downloading program codes, such as the Internet. The methods, systems, and devices described herein can be implemented as software in a digital signal processor DSP, a microcontroller, or any other side processor, or as a hardware circuit in an application-specific integrated circuit ASIC, a CPLD, an FPGA, or other suitable device. The present invention can be implemented in a digital electronic circuit or in computer hardware, firmware, software, or a combination thereof, such as in available hardware of a conventional mobile device or in new hardware dedicated to processing the methods described herein, as described in more detail below. BRIEF DESCRIPTION OF THE DRAWINGS

[0131] Exemplary embodiments of the present invention are shown in the accompanying drawings. However, it should be noted that the accompanying drawings only show specific embodiments of the present invention and are therefore not to be considered limiting of its scope. The present invention may include other equally effective embodiments.

[0132] Figure 1 An exemplary embodiment of a method for determining operating conditions of a scheduled production run or a change in a current production run of a chemical production plant comprising at least one catalytic reactor is shown;

[0133] Figure 2 shows a high-level workflow for preprocessing raw data from a single plant into a dataset ready for model training or prediction;

[0134] Figure 3 shows an example workflow for identifying a selection of raw data sets upon which to base a model;

[0135] Figure 4 An exemplary implementation of a system for determining changing operating conditions for a scheduled production run or a current production run of a chemical production plant is shown. DETAILED DESCRIPTION

[0136] The present disclosure provides a computer-implemented method for predicting the short-term performance of catalyst and / or predicting the long-term performance of catalyst (including catalyst aging effect) in a chemical production plant including at least one catalytic reactor. The method relates to a mathematical model using, in particular, a catalytic reactor, which is based on machine learning, does not involve prior information of reaction kinetics and uses input parameters selected from sensor raw data, derived parameters, reactor operating parameters, plant metadata and parameters indicating catalyst aging. Methods, systems, computer programs and computer program products disclosed herein are further described for a styrene production plant used as an example. Methods, systems, computer programs and computer program products disclosed herein are applicable to other production plants with at least one catalytic reactor, in particular with a fixed-bed reactor.

[0137] In case of a styrene production plant, a feed stream comprising ethylbenzene is mixed with steam in a mixer. The mixed stream is fed into a catalytic reactor comprising a potassium promoted iron oxide based catalyst for dehydrogenation to styrene monomer. The styrene production plant further comprises temperature sensors, pressure sensors, flow sensors etc. located at various locations to monitor the plant operation.

[0138] In the production of styrene monomer, ethylbenzene can be dehydrogenated in an adiabatic radial flow reactor. Ethylbenzene is mixed with steam in a specific ratio called steam-to-oil (STO) ratio to provide heat for the endothermic dehydrogenation process and prevent the reduction and coking of the potassium-promoted iron oxide-based catalyst used in the process. The reaction is operated at high temperature and subatmospheric pressure in a setup that usually includes at least two consecutive reactors. Intermediate reheating can compensate for the energy consumed by the reaction. Low pressure, steam dilution and high temperature favor the dehydrogenation of ethylbenzene, resulting in higher equilibrium conversion.

[0139] In the service life of catalyst, potassium as coke gasification promoter evaporates from catalyst and is carried to the downstream of catalyst bed to its cooler outlet, causing the activity loss caused by catalyst coking. In order to compensate for the aging of catalyst, plant operators increase inlet temperature in the catalyst life period, so as to keep ethylbenzene conversion constant. Higher temperature negatively affects the selectivity of styrene by strengthening the formation of cracking and byproduct (such as benzene and toluene). In addition, the potassium deposit in the catalyst bed and the fine powder produced by the catalyst cause the reactor inlet pressure to increase, which is thermodynamically disadvantageous. When inlet temperature or pressure have been increased to exceed the operating limit of factory, must stop running and need to replace catalyst.

[0140] Catalyst performance and its aging rate depend on the operating parameters of the reactor, such as inlet temperature, STO ratio, inlet and outlet pressures, ethylbenzene flow rate, number of reactors used, etc. Process economics can be significantly improved by optimization of reactor operation. Attempts have been made in the past by using different types of reactor models that have been fitted to catalyst operating data derived from experimental or industrial reactors. The models are based on knowledge or assumptions about the reaction kinetics of primary and secondary reactions, mass, heat and pulse transport phenomena, adsorption / desorption of different chemicals present in the system, coke gasification and potassium loss kinetics, etc. The methods, systems, computer programs and computer program products described herein allow for more robust and reliable process control.

[0141] Figure 1 An exemplary embodiment of a method for determining changing operating conditions of a scheduled production run or a current production run of a chemical production plant comprising at least one catalytic reactor is shown.

[0142] If the determination relates to a predetermined production run, operating data indicating predefined operating conditions for the predetermined production run are received via the communication interface in a first step 10. The predefined operating data may be generated by a previous production run. If the determination relates to a change in the current production run, measured operating data indicating current operating conditions for the current production run are received via the communication interface in a first step 10. In addition, at least one operating data point may be adjusted so that it includes a desired operating value indicating a change in the current operating conditions. The operating data may include sensor data measured by sensors installed in the chemical production plant, quantities derived directly or indirectly from such sensor data, analytical data measured in samples taken from the chemical production plant, and / or quantities derived directly or indirectly from such analytical data.

[0143] In addition, a catalyst aging indicator associated with a time period that the catalyst has been used in a current or scheduled production run is received via the communication interface. The catalyst aging indicator can be based on a point in time, a time period, a quantity derived from the time-dependent operational data, and / or a quantity cumulatively derived from the time-dependent operational data. In addition, plant metadata indicative of a physical plant layout can be received via the communication interface.

[0144] In a second step 12, the operational data and the plant metadata may be pre-processed via a processing device before determining at least one target operational parameter.Preferably, the pre-processing comprises a conversion into quantities that are independent of the physical plant layout.

[0145] In a third step 14, at least one target operating parameter for a predetermined production run or a change in operating conditions for a current production run is determined via a processing device using a data driven model based on operating data and a catalyst aging index. The determination of the at least one target operating parameter may be additionally based on plant metadata. The data driven model is parameterized according to a training data set. The training data set may be based on a historical data set including operating data, a catalyst aging index, at least one target operating parameter, and optional plant metadata. The historical data set may include data from multiple runs, multiple plants, and / or multiple catalyst batches. The determination of the at least one target operating parameter may be based on a short-term model that determines a target operating parameter at a discrete time point or a long-term model that determines a target operating parameter over a period of time.

[0146] In the case of a long-term model, a time series of at least one target operating parameter measured, predicted or derived during a current or previous production run up to a prediction start time point may be received via a communication interface in step 10. Determining (14) via a processing unit at least one target operating parameter at one or more time points after the prediction start time point may be based on operating data including expected operating values, the time series of at least one target operating parameter, a catalyst aging indicator and optional plant metadata using a data-driven model. Preferably, the data-driven model includes inherent time correlation.

[0147] In a fourth step 16 , at least one target operating parameter for a predetermined production run or a changed operating condition of a current production run may be provided via the communication interface.

[0148] In a fifth step 18, the determined target operating conditions may be provided to an optimization process device for optimizing the operating conditions of a predetermined production run or a change of the current production run of the chemical production plant. Here, for more than one operating condition of the predetermined production run or the change of the current production run, the determined target operating parameters may be received via a communication interface between the optimization process device and the processing device. Based on the received target operating parameters, a minimum or maximum value of the target operating parameter or a minimum or maximum value of an optimization parameter derived from the target operating parameter may be determined via the optimization process device, and a minimum or maximum value indicating the optimal operating condition of the change of the predetermined production run or the current production run may be provided via the communication interface.

[0149] Preprocessing of operational data

[0150] Figure 2 An exemplary workflow is shown for preprocessing data into a format suitable for model training, prediction via a short-term model, or forecasting via a long-term model.

[0151] In a first step, measured operational data are optionally received, followed by a pre-processing method. If data are prepared for training, such operational data may include historical data sets from multiple runs, multiple factories and / or multiple catalyst batches. If data are prepared for prediction or forecasting, such operational data may include measured operational data indicating the current operating conditions of the current production run. The operational data preferably include sensor data measured by sensors installed in the styrene production plant, and / or analytical data measured in samples taken from the styrene production plant.

[0152] For every production plant, there are many sensors available, often hundreds or even thousands, which provide raw data at their respective sampling rates. In addition, analytical data (e.g. the results of a gas chromatograph) are available for the specific times that samples were taken from the plant. The frequency with which this data is available varies from plant to plant, but a typical frequency is once a day to once a week at best.

[0153] In a first preprocessing step, operational parameters may be selected from operational data that form input parameters for a data-driven model. These input parameters may be derived from raw parameters such as sensor data or analytical data. An exemplary process of how these raw parameters may be selected from all available parameters when setting up a data-driven model is outlined in more detail below.

[0154] In a second preprocessing step, the data from the analytics and selected sensors can be combined based on their timestamps and, in particular, preprocessed onto a common time scale. For many plants, only a daily summary of the raw sensor data may be available instead of high-frequency raw data, and this is also the typical frequency at which the analytics data is available, and a daily average can be used to obtain all data based on the same time. Other merging techniques can also be applied and are well known, such as interpolation of daily data (both analytics and sensors) and sampling at higher or lower frequencies of interest, for example to create hourly data.

[0155] Additionally, plant metadata may be received. Plant metadata (e.g., catalyst active volume and number of reactors) may be added to the dataset as numerical or categorical variables to complete a set of input parameters for the model, including derived parameters.

[0156] In the third pre-processing step, the received and selected operating data can be filtered and smoothed. Here, for example, maintenance intervals, startup phases, irregular and outlier time points and durations can be identified and optionally filtered. To achieve this, many options are known to those skilled in the art, and any combination of this method can be used. For example, the program includes applying a feasible absolute threshold based on catalyst field knowledge, such as minimum reactor temperature, maximum steam / oil ratio and the maximum pressure after the reactor; For example, based on distribution indicators, such as 6 times of the interquartile range (conservative threshold used in the example application), or alternatively using an absolute threshold based on the estimated possibility of the data points derived from the overall distribution, by comparing each value or a set of values ​​with the distribution of all other values ​​available from the same parameter or a set of parameters of the corresponding production plant to identify outliers; and / or identifying irregularities based on a parameter jump compared to the monthly coefficient of variation of the parameter.

[0157] In the fourth pre-processing step, missing data points on the common time scale can be detected and replaced with statistically determined values. This possible missing parameter can be estimated. Specifically, if these data have been sampled at a frequency lower than the selected time basis, the analytical data can be interpolated. Interpolation can be determined by different methods, such as simple mean interpolation, forward or backward filling, weighted mean or estimated values ​​from Kalman filter, or comparable estimation methods. The same method can be applied to replace abnormal values. The start of operation can be identified based on the standard defined by the expert, such as the first data point of space velocity per hour>0.2 / h in the styrene catalyst example. All derived parameters can be calculated then, including, for example, cumulative plant production.

[0158] Depending on the nature of the derived parameters, in particular the cumulative parameters, identified startup phases, downtimes, etc. may be removed from the data set before or after the calculation of the corresponding derived parameters. In an exemplary embodiment, the cumulative production since the start of operation (on which the achieved aging indicator is based) is calculated before the startup phases are removed from the data set, since they contribute to the aging of the catalyst, even though these phases are not part of the operating conditions covered by the model.

[0159] Figure 2 Another preprocessing step not depicted in the figure may be to convert the data from different plants into common units. Figure 2 You do this before the workflow in the , although it can be done before or after any step in the process.

[0160] At this point, the data is ready for training, forecasting with a long-term model, or prediction with a short-term model. However, since the filtering procedure may cause some gaps in the data, and the time scale used for forecasting is usually much longer (e.g., months) compared to the available time basis (e.g., days), additional aggregation steps (e.g., weekly average or median aggregation) can be performed on the data before it is used in the long-term model.

[0161] Parameter selection process

[0162] The operational data or raw parameters of interest can be accessed via Figure 3 The parameters available to catalyst experts are used as a basis, which consist of a combination of raw sensor data, analytical data and some typical operating parameters derived from them (e.g. conversion, space velocity, steam / oil ratio, selectivity). In a first step, if data from multiple plants are used, only parameters available for all plants of interest can be selected (this criterion is outdated if only one plant is modeled), and redundant parameters (e.g. rescaled parameters) and zero-variance parameters may be removed.

[0163] Next, a correlation matrix of the remaining parameters can be calculated, and clustering of the parameters (e.g., hierarchical clustering) can be performed to identify pairs or clusters of parameters that carry similar information. A simple filter for high (anti) correlation values ​​(e.g., (>0.90 or <-0.90) or (>0.95 or <-0.95)) can be applied instead of a clustering algorithm. However, clustering of correlation values ​​with all other parameters further allows identification of parameters that may have a lower direct correlation but have very similar correlation values ​​with all other parameters. From each cluster of two or more parameters, only a single parameter can be retained based on certain selection criteria, such as: the parameter needs to be available for all plants; typical parameters used by operators on a daily basis need to be retained (this is an exception that may result in retaining more than a single parameter), and the parameter should represent an interpretable quantity, which is also the reason for not reducing the dimensionality via principal component transformation or similar methods that produce features that cannot be directly interpreted by the operator.

[0164] Based on these criteria, the set of parameters can be iteratively reduced by adjusting the clustering threshold when necessary. The number of parameters can be iteratively reduced until a set of parameters (1, ..., M) remains, where only low correlations remain between the parameters. Some of the remaining relatively high correlation values ​​arise from parameters that may be important to the operator and therefore may not be removed.

[0165] Once the parameter set is reduced, the raw sensor and analytical data are identified, which is necessary to obtain all of these remaining parameters.

[0166] Normalization of parameters and factory metadata

[0167] Different plants vary in their production levels, their typical operating conditions, and often exhibit systematic differences. To account for such differences between different plants, one or a combination of the two strategies described below can be applied.

[0168] In an additional pre-processing step, normalization may be performed. In one embodiment, plant metadata indicative of a physical plant layout may be received. Such plant metadata may include reactor layouts, such as the number of reactors, active catalyst volume, reactor type, size, or a combination thereof.

[0169] Normalization of operational data can be performed to make the model input parameters (except plant metadata) as independent of the specific plant layout as possible. For many parameters such as steam / oil ratio, conversion, selectivity, no normalization is required. In addition, parameters such as liquid hourly space velocity (LHSV) are essentially normalized by catalyst volume. Additionally or alternatively, the total product amount or cumulative total production of catalyst production runs for each catalyst can be normalized by active catalyst volume, because this is a more comparable measure of "aging" of catalyst per unit volume than time on stream (which does not capture changes in production levels) or unnormalized cumulative production (which will have different meanings for reactors of different sizes).

[0170] Additionally or alternatively, the pressure may be normalized to its initial value during the start of the run, e.g., a median value for the first 90 days on stream, in order to focus on aging effects rather than (in some cases more significant) plant-to-plant variations. Additionally or alternatively, the pressure drop across each unit (e.g., reactor or heat exchanger) may be normalized by the space velocity or total flow rate, since it is well known that it varies with the superficial linear velocity of the gas mixture. Additionally or alternatively, an average inlet temperature of the reactor greater than one (e.g., 2 or 3) may be determined.

[0171] Ultimately, there are other reasonable ways, beyond the examples mentioned in this section, to make operational data, analytical data, and any quantities derived therefrom, comparable between plants; and there may be other useful reactor / plant metadata, particularly metadata about reactor geometry, that could be used in a similar manner.

[0172] The reason for finding such normalization parameters is that using data from multiple different plants provides a number of significant advantages: 1) The parameter space covered by the aggregated dataset is much larger than that of any single plant, which typically revolves around a relatively narrow set of operating parameters that the operators are familiar with, as erroneous deviations from this can result in serious monetary losses. Therefore, a model trained on such an aggregated dataset can provide predictions outside the operating range of a specific plant, because it includes information that cannot be obtained from their own historical data. 2) Due to the long service life of the exemplary styrene catalysts (2-3 years), each plant can only run 1-4 times a certain type of catalyst, severely limiting the number of deactivation processes that can be observed at each plant (each run only provides a single independent observation of catalyst deactivation). The aggregated dataset allows many more deactivations to be included in the training data. 3) Combined with a parameter selection process that emphasizes interpretability and availability of common operating parameters, this allows the trained model to be applied to plants for which no a priori real data is available - a situation often encountered when technical recommendations need to be made for a new plant, which would not be possible if the model was strictly linked to a specific set of sensors available at a specific plant.

[0173] Short-term model

[0174] For short-term predictions of catalyst behavior, any regression model can be used and a variety of typical candidate models are known to those skilled in the art. Depending on the nature of the data in a chemical production plant, a small number of independent runs can usually be selected, and even after normalization of many parameters, some potential plant-specific biases, relatively simple in nature, and less flexible models may be selected. In such scenarios with a small number of independent runs, highly flexible models (such as random forest regression) will fit the training data set better, but may not extrapolate well to new data. Therefore, depending on the number of runs available, an appropriate type or combination of regression-based models can be selected.

[0175] One possible model could be an ensemble of linear models trained on a subset of the training data, for example to predict the average reactor inlet temperature and the selectivity of the reaction. Using such an ensemble has two advantages: first, using an ensemble of models for prediction can lead to more accurate predictions, for example by taking the median prediction of all models [e.g., Ensemble Methods—Foundations and Algorithms, Zhou Zhihua; CRC Press 2012]; second, training the ensemble in this way can estimate the uncertainty of the model by using a prediction range (or 10% and 90% percentiles, or any other prediction quantile range...).

[0176] When using training data from multiple catalyst batches, multiple runs, and / or multiple plants, the training and test data sets may be split by individual runs. Additionally, or alternatively, the training and validation data sets may be split by individual plants or catalyst batches. For example, a random set of approximately 75% of the training set may be selected, and the parameters may optionally be normalized (to zero mean and unit variance) during preprocessing. The remaining data may be used as a validation data set to test the trained model.

[0177] Long-term model

[0178] For time series forecasting, a large number of mathematical models can be used, ranging from autoregressive models to recurrent neural networks. Model requirements applicable to the exemplary problem at hand include: 1) Applicability to multivariate time series, i.e., prediction of multiple endogenous (uncontrolled) parameters with long-term trends to be forecasted, such as pressure, inlet temperature, or selectivity; 2) Integration of exogenous (controlled) parameters, i.e., highly influential parameters that are known or will be controlled externally and therefore do not require model forecasting, such as steam / oil ratio, LHSV, target conversion.

[0179] A preferred embodiment of the method relates to a mathematical model that allows regularization to avoid overfitting, which may easily occur when multiple time lags are included in the mathematical model. In one embodiment of the mathematical model used in the method for generating this disclosure, a collection of mathematical models is implemented, which is mainly based on a combination of a penalized linear model and a penalized vector autoregression model (VARX) with exogenous variables. For an overview of penalized VAR(X) models, including different structured regularization methods, see, for example, [arXiv:1508.07497v1 (Nicholson et al., 2018, VARX-L:StructuredRegularization for Large Vector Autoregressions with Exogenous Variables)] and references therein.

[0180] A training / test split is performed for short-term models between runs. All candidate models can be trained to predict the endogenous variable at time t based on the history of the exogenous variable at time t and the endogenous variable up to a maximum time lag m (tm,…,t-1). Applying the model iteratively step by step and using the forecast of the endogenous variable in each new step as input for the next step allows predictions to be made for any number of steps ahead.

[0181] The training procedure can be performed using leave-one-out cross-validation at the plant level. Here for the N plants in the dataset, N training datasets (consisting of data from all other plants) can be generated, their parameters can optionally be normalized to have mean 0 and comparable value ranges, and the trained model can be evaluated on the omitted plants (validation set). Finally, model hyperparameters (e.g., regularization parameters) can be selected that provide the most robust performance on the N validation sets, measured, for example, by the average root mean square error of the one-step-ahead forecasts, and the model can be trained on the full training set.

[0182] Since the mathematical model used in the method of the present invention is based on machine learning, it is necessary to use historical data from at least one production run in a chemical production plant including at least one catalytic reactor for training before the predicted short-term performance or predicted long-term performance of the aging of the catalyst used for such a reactor system. In one embodiment of the method of the present invention, historical data from multiple production runs of the same type of catalyst have been used to train the model. In another embodiment of the method, historical data from production runs in multiple reactor systems including the same type of catalyst have been used to train the model. In both embodiments, historical data can be provided from different manufacturing batches of the same type of catalyst. In one embodiment of the method, before being used to train the model, the operating data and the catalyst aging index of possible multiple production runs have been normalized, as described in the previous section.

[0183] It has been found that using operating data from more than one production run, whether from the same plant or from different plants, can improve the quality of predictions and broaden the range of operating parameters covered by the prediction or forecast model. Including data from multiple runs and plants further improves the generalizability of the prediction or forecast to be applied to production plants for which no data was available during model training. However, all data used to train the model are preferably provided by plants using the same catalyst formulation, as different catalysts vary significantly in their catalytic properties (reaction rates) and morphological properties (transport properties). This includes the same type of catalyst provided using different manufacturing batches or different delivery dates.

[0184] In one embodiment, input parameters for the model are selected from sensor data available from the reactor system, such as analytical data from gas chromatography (GC) analysis, and the derived parameters listed in the corresponding section above.

[0185] In one embodiment, the operating data or reactor operating parameters include the inlet and outlet temperatures of each reactor, the inlet and outlet pressures of each reactor, and the composition of the reaction mixture at the inlet and outlet of each reactor.

[0186] In one embodiment, the operating data or input parameters of the model include steam to oil (STO) ratio, liquid hourly space velocity (LHSV), total styrene production normalized by catalyst volume, target ethylbenzene conversion, styrene selectivity, average inlet temperature, normalized pressure after the last reactor, normalized pressure drop across the reactor, temperature loss across the reactor, and normalized deviation of temperature loss from expected (based on conversion calculation) value.

[0187] An advantage of the parameter set used in the method or system of the present disclosure is that it remains interpretable because many parameters correspond to actual operational parameters or can be easily interpreted in that context. Other methods of reducing the dimensionality of the problem (e.g., PCA or RFA) can produce parameters that are useful for the modeling process and prediction accuracy, but generally lack interpretability.

[0188] The mathematical model used in the method of the present disclosure performs time series forecasting and is able to cover the complete deactivation process of the catalyst during the entire production run in the reactor system. The model enables the plant operator to improve and optimize the operating strategy on a daily basis based on energy costs, market supply of raw materials or demand for plant products, and other constraints that may occur in the plant (such as outages of different plant parts or utilities).

[0189] The subject matter of the present disclosure is further described and explained in the following working examples.

[0190] Example

[0191] Generation of datasets for training mathematical models

[0192] To develop the mathematical model, data from multiple production runs of the same catalyst type (BASF S6-42) were used. The data set covers 11 industrial plants with 2 or 3 reactors and information on about 1-4 production runs per plant. For each individual plant, a complete set of parameters (including analytical and sensor data) was collected. Sensor data are usually available at daily resolution, while analytical data are available at daily to weekly resolution.

[0193] The parameter selection process of the model is as described above ( Figure 3 )

[0194] Table 1 lists the operational data from the raw sensors and the analysis parameters that were chosen at each time point in order to derive all relevant parameters for the model. These unit and format selections are just one example of what could be used; temperature could, for example, also be specified in degrees Fahrenheit, pressure could be specified in mmHg, and another date and time format could be used, etc.

[0195]

[0196]

[0197] Table 1

[0198] The number of reactors and the total catalyst volume are additionally used as metadata. Table 2 below lists the parameter sets (derived parameters and plant metadata) used to train the different models.

[0199]

[0200]

[0201] Table 2

[0202] Example application: Short-term model

[0203] Model development

[0204] As described above, an ensemble of 50 linear regression models has been trained on subsets of the training dataset, each of which is split between runs. Each of the 50 reduced training sets contains about 74% of the available runs selected at random to both improve predictions and provide an estimate of the local uncertainty in the model predictions (assuming that each model is trained on, for example, a different subset of operating conditions). Importantly, the data subsets need to be split between runs (or alternatively between plants), rather than a random sampling of training data points. Otherwise, there will be little variation between models, since all models are trained on nearly the same distribution of operating conditions.

[0205] In the present example, in particular the parameters "temperature" (average reactor inlet temperature) and "selectivity" (reaction towards the desired product styrene) have been predicted using all the other parameters in Table 2.

[0206] In this specific implementation, the reaction temperature itself is one of the main factors affecting the observed selectivity. Therefore, the prediction of the two parameters is performed in two steps. First, a set of 50 models is trained to predict temperature based on all parameters in Table 2 (excluding selectivity and temperature). Using all parameters in Table 2 as input (excluding selectivity only), a second set of 50 models is trained to predict selectivity. Then, the predicted temperature from the first set of models is used as one of the input parameters to predict selectivity.

[0207] In actual use of this developed model, the workflow for predicting both parameters has subsequently been implemented in a single prediction function that receives the input parameters and predicts both temperature and selectivity. Thus, for all uses of the model, from an external point of view, the combination of the two model ensembles can be viewed as a single entity of a "short-term" model.

[0208] In addition to directly using the parameters in Table 2, interaction terms are also considered and finally, for example, the "steam / oil ratio-conversion rate" interaction is integrated into the short-term model for predicting temperature. Such interactions or higher-order polynomial terms (such as the quadratic term for temperature for selectivity prediction) can be easily implemented into the prediction model using statistical programming languages ​​without expanding the basic set of parameters provided to the model.

[0209] Choosing which type of model to use, whether to use ensemble techniques, and which higher-order terms, transformations, or interactions of input parameters to use depends on the specific problem and dataset; and is a typical model development procedure for a data scientist.

[0210] Short-term model use case 1

[0211] In one exemplary use case, a plant operator may wish to use a short-term model to estimate the required reactor temperature to achieve a desired target conversion at specific operating conditions that may not have been used before in their plant. In the case of a styrene production plant, such operating conditions may be an increased steam / oil ratio or a decreased LHSV or a change in feed composition. Ideally, the starting point for this prediction should be the current plant state, including the normalized aging parameter "totalProduction".

[0212] Model Implementation

[0213] Figure 4 A client-server arrangement of a production monitoring and / or control system is shown, including a client having a processing device including a user application, a server having a processing device for a service provider. The client and server can be coupled in communication, for example, wired or wirelessly, via a communication interface. The client can include a display device. Preferably, the client user application is configured to receive and display operating data, expected values, catalyst aging indicators, plant metadata, target operating parameters, or determined operating conditions. More preferably, the client user application is configured to receive target operating parameters or determined operating conditions to control current or scheduled production operations in a chemical production plant based on the target operating parameters or determined operating conditions. The client user application can be an embedded part of a chemical production plant process monitoring and / or control system.

[0214] Raw sensor and analysis data are recorded at the production plant together with the necessary plant metadata since the start of the production run. These data are preprocessed according to a data preprocessing workflow, e.g. Figure 2As shown, in particular, the same workflow (filtering steps, interpolation, thresholds, aggregation, etc.) as used to prepare the training data set for the corresponding model is used. Preferably, this pre-processing is implemented by the same party that has developed the model and can be provided to the user application, for example via direct integration of the pre-processing function or via an application programming interface (API).

[0215] After the data has been preprocessed, the user can adjust the operating parameters of interest to provide expected operating values ​​that indicate changes in current operating conditions. The set of adjusted input parameters (also referred to as operating scenarios or prediction scenarios) is then transmitted to the prediction function. The prediction function may, for example, be implemented locally in the user application or addressable via an API and perform all of the above operations. The results will be reported back to the user application, for example, in order to compare and select between different scenarios. The user can further associate the input parameters and / or prediction parameters with, for example, costs or other quantities that may affect the decision-making process regarding which operating parameters to use in the plant.

[0216] Figure 4 Visualization of an exemplary implementation concept. Raw data is automatically or manually collected at the production plant and transmitted to a service provider via an API, which processes the raw data into the correct format according to a workflow, such as described above. The transformed data set can be provided to a user (e.g., a plant operator), and different scenarios can be defined based on current values. These scenarios can be transmitted to the same or a second API that runs the model, which provides the corresponding predictions or forecasts to the user.

[0217] Short-term model use case 2

[0218] In a second use case, an expert may wish to provide an estimate of the reactor temperature and the corresponding selectivity development before a catalyst is installed in a plant. This scenario often occurs, for example, before a decision is made to purchase a catalyst, during the preparation of a technical proposal that provides one or more hypothetical operating scenarios and their impact on the customer. Therefore, accurate predictions are desired for plants that have never provided data for model training.

[0219] Making some simplifying assumptions, such as estimating the “aging” parameter totalProduction based on assumed operating data, typically involving constant operating parameters for the complete run, replacing reactor temperature losses with estimates based on operating conditions, etc., the above short-term model can be used for this purpose given a typical set of operating parameters provided by the customer. The fact that this service can be provided by the model is a direct result of the selection criteria used in the parameter selection workflow ( Figure 3 ), in particular keeping as input to the model all the parameters that operators typically use to monitor and control their plants.

[0220] Model presentation to users

[0221] The interface of the user application may include a block of input parameters to the model, which may be automatically or manually populated and may be used to interactively specify operating scenarios; a block of model outputs for specifying scenarios (e.g., average reactor inlet temperature and selectivity with prediction ranges); and optionally a further output as a model prediction of the local response to the parameter of interest (e.g., predicted reactor inlet temperature for a range of target conversions).

[0222] The input parameter blocks can be automatically populated if current plant operating data is available (e.g., use case 1), or can be completely manually defined if no actual plant data is available (e.g., use case 2). Based on the user-adjusted input parameters, the parameters predicted by the model, in this example, reactor inlet temperature and styrene selectivity, can be displayed to the user as, for example, text (for prediction of a single point), or, for example, in graphical form (local model response to a single parameter change). These are used only as examples, as there are many more ways in which the model can be used using use cases, implementations, and presentation to the model user.

[0223] Example application: Long-term model

[0224] Model development

[0225] Starting from the processed dataset for developing the short-term model (Table 2), some additional steps were taken to prepare the dataset for training the long-term model. First, a smaller number of parameters were selected from the list, and second, each parameter was aggregated weekly.

[0226] The input parameters of the model can be constant (reactors, CatVolume), can be controlled (conversion, SOR, LHSV), can be calculated based on these constant or controlled values ​​(totalProduction), or can be uncontrolled in the operating scenario of interest (temperature, selectivity, pressureOut, deltaP, deltaT, dTdev_norm).

[0227] The latter set of parameters are labeled as endogenous or uncontrolled parameters, while the former (which are known over the entire forecast horizon, assuming the operational scenario executes as planned) are labeled as exogenous or controlled parameters.

[0228] Table 3 provides the parameters used to develop an exemplary forecast model and their assignment to the two types of parameters.

[0229] Parameter name Exogenous / Endogenous type reactors External Absolute temperature Endogenous Numeric pressureOut Endogenous Numeric deltaT Endogenous Numeric SOR External Numeric LHSV External Numeric conversion External Numeric selectivity Endogenous Numeric CatVolume External Numeric deltaP Endogenous Numeric totalProduction External Numeric dTdev_norm Endogenous Numeric

[0230] Table 4

[0231] In developing the exemplary models, a number of different VARX-type candidate models as described above were trained, where the candidate models differed, for example, in their regularization methods (elasticnet, ridge) or their maximum number of time lags (4-10 weeks).

[0232] For all these different model combinations, the training procedure is performed in the same way at the plant level using leave-one-out cross-validation. All trained models are candidate models for the ensemble that ultimately comprises the long-term model.

[0233] While all training is performed for one-step-ahead forecasts (which can be applied iteratively, forecasting any number of steps ahead), the model ultimately chosen should also perform well for longer-term forecasts. To evaluate this, forecasts are performed starting at multiple points for each available run in the training and test datasets using the actual exogenous variables of these runs as operational scenarios, and the error distribution of the forecasts is determined for each step ahead.

[0234] For reactor inlet temperature and selectivity on both the training and test data sets, models were selected that provided no significant long-term bias and narrow error distributions over the entire range (even though a modest widening of the error distribution for longer forecasts must be expected). Model selection was based on the error distributions of predictions performed on the test and training sets from the 30 candidate models.

[0235] In this exemplary development of a long-term model for a styrene catalyst, the models ultimately chosen were 3 two-stage models with a maximum time lag of 10 weeks trained with different types of regularization on local ramps. For the ensemble forecast, all models were iteratively run independently of each other and only the complete individual forecasts were averaged to provide the ensemble forecast. In the example presented, this is a more efficient implementation than aggregating the forecasts after each individual step, which would be a possible alternative implementation of the ensemble forecast.

[0236] As with the individual models, the final ensemble is evaluated on the test data to obtain the error distribution for the next N days, which is an exemplary implementation provided as a forecast error estimate.

[0237] As with the short-term example above, the ensemble forecast and the expected error distribution together can be treated by the user as a single entity for the “long-term” model for model application, regardless of the details of the underlying procedure (ensemble mean aging, multi-step models, …).

[0238] Long-term model use cases

[0239] In an exemplary use case, a plant operator may wish to estimate the remaining catalyst life depending on different scenarios for operating the plant and plant-specific constraints on operating parameters. Such scenarios may include changes in LHSV, steam / oil ratio, or target conversion level; there are many motivations for considering different scenarios, but an exemplary question would be whether it is ultimately worthwhile to run the plant at a lower production level during a period of low styrene prices to extend catalyst life.

[0240] The end of catalyst life can depend on many conditions defined locally in each production plant, but one limitation of catalyst life is reactor temperature or pressure that may not exceed a plant-specific threshold in all cases. Based on the predictions of the long-term model, the end of catalyst life can be estimated based on this threshold and compared for different user-defined scenarios.

[0241] Model Implementation

[0242] This model can be conceptually similar to Figure 4 , but with some minor changes. For example, the processing functions implemented locally or via the API need to be adjusted to provide the data format for long-term model training, in this specific example a weekly aggregation step after executing the program on the data for the short-term model. For example, the prediction function implemented locally or via the API now receives not only a set of operating conditions to predict a single point, but also the lagged endogenous data required by the model, as well as the operating scenario values ​​of the exogenous parameters. Similarly, the model output is a complete forecast of all or only some of the endogenous variables of the model.

[0243] User model presentation

[0244] In the long-term model implementation, the endogenous parameters used in the model need to be available for at least the last L weeks, where L is the maximum time lag used in any model element included in the "long-term" model. This data can be automatically obtained from the plant raw data as described in the Data Preprocessing and Preparation section, or data in an appropriately formatted format can be manually entered or uploaded.

[0245] The operator can use the control parameter input block to create an extended period of operation scenario, for example, in the example application, the steam / oil ratio, target conversion and LHSV can be planned in advance in up to three independent parts to simulate future changes. These scenarios may also be much more complicated than what is described here. The planned operation scenario can be displayed visually, preferably together with the history of these parameters to more easily maintain the continuity of some controls. The control parameters selected for user manipulation can be composed of any subset of all exogenous parameters used in the model. In this specific implementation, the other 3 exogenous parameters are constant, or can be directly derived from other inputs and / or time.

[0246] The forecasts provided by the long-term model may be displayed in a tabular format, ready for export of data and further analysis based on forecast trends, for example, or the results may be displayed graphically to the user for visual inspection of one or more different planning scenarios.

[0247] The interface may include a section where selected control (exogenous) parameters may be defined for user-defined operating scenarios (e.g., target conversion, LHSV, and steam / oil ratio; graphs showing actual data up to the start of the prediction and for future operating scenarios). In another section of the interface, selected endogenous parameters may be presented to the user (e.g., reactor inlet temperature and styrene selectivity; the start of the prediction may be marked by a horizontal dashed line, and everything beyond that point is predicted by the model and includes an estimate of the prediction error). In addition, the predictions for all or some of the endogenous parameters may be presented as a data box that may be exported by the user.

Claims

1. A system for determining operating conditions of a chemical production plant including at least one catalytic reactor, the system comprising a communication interface and a processing device in communication with the communication interface, (a) For a predetermined production run, the system is configured to: - receiving (10) via said communication interface operating data indicative of predefined operating conditions for said predetermined production run, - receiving (10) via said communication interface a catalyst aging indicator associated with a time period during which the catalyst has been used in said predetermined production run, - determining (14) via the process equipment at least one target operating parameter of the operating conditions for the predetermined production run based on the operating data and the catalyst aging indicator using a data driven model, wherein The data driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including operating data, catalyst aging indicators and the at least one target operating parameter, - providing (16) via said communication interface said at least one target operating parameter of said operating conditions for said predetermined production run, or (b) For changes to the current production run, the system is configured to: - receiving (10) via the communication interface measured operating data indicative of current operating conditions for the current production run, wherein at least one operating data point comprises an expected operating value indicative of the change in the current operating conditions, - receiving (10) via the communication interface a catalyst aging indicator associated with a time period during which the catalyst has been used in the current production run, - determining (14) via the process equipment at least one target operating parameter for the changed operating condition of the current production run based on the operating data and the catalyst aging indicator using a data driven model, wherein the data driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including operating data, the catalyst aging indicator and the at least one target operating parameter, - providing (16) via said communication interface said at least one target operating parameter for said changed operating condition of said current production run, wherein the system is configured to - receive (10) via the communication interface plant metadata indicative of a physical plant layout, and - determining (14) via the processing unit at least one target operating parameter additionally based on the plant metadata using a data driven model, wherein the training data set is based on a historical data set additionally including the plant metadata, wherein the system is further configured to pre-process (12) the operational data and the plant metadata via the processing equipment prior to determining the at least one target operational parameter, wherein the pre-processing includes a conversion of quantities that are independent of the physical plant layout, The pre-processing includes: pre-processing selected data from the received operational data onto a common time scale; and detecting missing data points on the common time scale and replacing the missing data points with statistically determined values.

2. The system according to claim 1, wherein: The operational data include sensor data measured by sensors installed in the chemical production plant, quantities derived directly or indirectly from the sensor data, analytical data measured in samples taken from the chemical production plant, and / or quantities derived directly or indirectly from the analytical data.

3. The system according to claim 1 or 2, wherein: The historical data set includes data from multiple runs, multiple plants, and / or multiple catalyst batches.

4. The system according to claim 1 or 2, wherein: The catalyst aging indicator is based on a point in time, a time period, a quantity derived from time-dependent operating data, and / or a quantity cumulatively derived from time-dependent operating data.

5. The system according to claim 1 or 2, wherein: The system is configured to determine at least one target operating parameter via the processing unit based on a short-term model that determines the target operating parameter at a discrete time point or a long-term model that determines the target operating parameter over a period of time.

6. The system according to claim 1 or 2, wherein: The system is configured to receive (10) via the communication interface a time series of at least one target operating parameter as measured, predicted or derived during the current or previous production run up to a predicted time point, wherein the system is further configured to determine (14) via the processing unit at least one target operating parameter for one or more time points after the predicted time point based on the operating data, the time series of the at least one target operating parameter, the catalyst aging indicator and optionally the plant metadata using the data-driven model, wherein the data-driven model includes inherent time correlation.

7. A system for optimizing operating conditions of a chemical production plant, the system comprising: The system according to any one of claims 1 to 6 and the optimization processing device communicating with the communication interface, the optimization processing device being configured to: - receiving (18) via the communication interface determined target operating parameters for more than one operating condition as determined (a) for a predetermined production run or (b) for a change in a current production run, - based on the received target operating parameters for each operating condition, determining (18) via the optimization processing device a minimum or maximum value of the target operating parameters, or a minimum or maximum value of an optimization parameter derived from the target operating parameters, - providing (18) via the communication interface the minimum value or the maximum value indicative of an optimized operating condition in (a) the predetermined production run or (b) the current production run.

8. A production monitoring and / or control system comprising a communication interface communicatively coupled to the system for determining the operating conditions according to any one of claims 1 to 6 or the system for optimizing the operating conditions according to claim 7.

9. The production monitoring and / or control system according to claim 8 comprises a display device configured to receive and display the determined operating conditions; or comprises a control unit configured to receive the determined operating conditions and control the current production operation or the scheduled production operation in the chemical production plant based on the determined operating conditions.

10. A computer-implemented method for determining operating conditions of a chemical production plant comprising at least one catalytic reactor, the method comprising the steps of: (a) For a predetermined production run, the method comprises the following steps: - receiving (10) via a communication interface operating data indicative of predefined operating conditions for said predetermined production run, - receiving (10) via said communication interface a catalyst aging indicator associated with a time period during which the catalyst has been used in said predetermined production run, - determining (14) via a processing device at least one target operating parameter of the operating conditions for the predetermined production run based on the operating data and the catalyst aging indicator using a data driven model, wherein the data driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including the operating data, the catalyst aging indicator and the at least one target operating parameter, - providing (16) via said communication interface said at least one target operating parameter of said operating conditions for said predetermined production run, or (b) For changes to the current production run, the method comprises the following steps: - receiving (10) via the communication interface measured operating data indicative of current operating conditions for the current production run, wherein at least one operating data point comprises an expected operating value indicative of the change in the current operating conditions, - receiving (10) via the communication interface a catalyst aging indicator associated with a time period during which the catalyst has been used in the current production run, - determining (14) via the process equipment at least one target operating parameter for the changed operating condition of the current production run based on the operating data and the catalyst aging indicator using a data driven model, wherein the data driven model is parameterized according to a training data set, wherein the training data set is based on a historical data set including operating data, the catalyst aging indicator and the at least one target operating parameter, - providing (16) via said communication interface said at least one target operating parameter for said changed operating condition of said current production run, Wherein, the method comprises the following steps: - receiving (10) via said communication interface plant metadata indicative of a physical plant layout, and - determining (14) via the processing unit at least one target operating parameter additionally based on the plant metadata using a data driven model, wherein the training data set is based on a historical data set additionally including the plant metadata, wherein the method further comprises the step of pre-processing (12) the operating data and the plant metadata via the processing equipment prior to determining the at least one target operating parameter, wherein the pre-processing comprises a transformation of quantities that are independent of the physical plant layout, The pre-processing includes: pre-processing selected data from the received operational data onto a common time scale; and detecting missing data points on the common time scale and replacing the missing data points with statistically determined values.

11. The method according to claim 10, wherein: The historical data set includes data from multiple runs, multiple plants, and / or multiple catalyst batches.

12. The method according to claim 10 or 11, wherein: The determination of at least one target operating parameter is based on a short-term model that determines the target operating parameter at discrete points in time or a long-term model that determines the target operating parameter over a period of time.

13. The method according to claim 10 or 11, wherein: A time series of at least one target operating parameter measured, predicted or derived during a current or previous production run up to a predicted time point is received via the communication interface (10), wherein at least one target operating parameter for one or more time points after the predicted time point is determined via the processing unit (14) based on the operating data, the time series of at least one target operating parameter, the catalyst aging indicator and optionally the plant metadata using the data-driven model, wherein the data-driven model includes inherent time correlation.

14. A method for optimizing the operating conditions of a scheduled production run or a change in a current production run of a chemical production plant, the method comprising the steps of: - receiving (18) via the communication interface target operating parameters determined by the method of claim 10 for more than one operating condition as determined (a) for a predetermined production run or (b) for a change in a current production run, - based on the received target operating parameters for each operating condition, determining (18) via the optimization processing device a minimum or maximum value of the target operating parameters, or a minimum or maximum value of an optimization parameter derived from the target operating parameters, - providing (18) via the communication interface the minimum value or the maximum value indicative of an optimized operating condition in (a) the predetermined production run or (b) the current production run.

15. A method for monitoring and / or controlling a chemical production plant, comprising the steps of performing the method for determining operating conditions according to any one of claims 10 to 13 or the method for optimizing operating conditions according to claim 14.

16. A computer-readable non-volatile storage medium comprising computer-readable instructions, which, when loaded and executed by a processing device, perform the method according to any one of claims 10 to 15.

17. A catalyst system comprising - a catalyst, and - a catalyst type identifier associated with the computer program or the computer readable non-volatile storage medium according to claim 16.

Images