Reactor scale-up simulator and reactor scale-up simulation method

Abstract

To provide a reactor scale-up simulator and simulation method suitable for scaling up from a laboratory-scale reactor to a plant-scale reactor.SOLUTION: A reactor scale-up simulator 1, which calculates changes in plant characteristics due to changes in a condition α, includes: a scaler that uses first measurement values from a laboratory-scale reactor R1 and a first model that simulates characteristics of the laboratory-scale reactor to determine parameters of a plant-scale reactor when operated under a condition α1 from the parameters of the laboratory-scale reactor when operated under condition α1, the parameters of the laboratory-scale reactor when operated under a condition α0, and the parameters of the plant-scale reactor when operated under the condition α0; an updater that uses these parameters to generate a second model that simulates the characteristics of the plant-scale reactor; and a plant tester that uses the second model to calculate changes in the plant characteristics of the plant-scale reactor due to changes in the condition α.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a reactor scale-up simulator and a reactor scale-up simulation method for simulating plant-scale behavior based on laboratory-scale trials. [Background technology]

[0002] To improve productivity and product quality, plant operators may want to modify existing plant conditions (e.g., reactor jacket material, agitation speed, or masses of raw materials and additives when operating a reactor) and verify the improvement. However, testing these modifications on an actual full-scale reactor is expensive and time-consuming, so simulators are commonly used to verify the effect of the desired modifications.

[0003] Plant simulators are constructed using physics-based models that represent the behavior of the target plant. However, the accuracy of the model is highly dependent on the applied model parameters (e.g., heat transfer parameters and fluid properties). Therefore, the model parameters of the plant simulator must first be optimally set to enable accurate simulation under modified conditions. For example, increasing the amount of additive in a reactor can promote faster heat transfer rates. To reflect this phenomenon in the plant simulator, the heat transfer parameters of the existing plant model must be identified and adjusted to properly represent the changes in plant behavior due to the increased amount of additive. As a result of this parameter adjustment, high simulation accuracy is maintained.

[0004] Regarding these points, Patent Document 1 discloses a method for adjusting plant model parameters for reactor simulation. With the aim of providing a B2B control system that is more practical and versatile than conventional systems, Patent Document 1 proposes a configuration in which "the control device includes a cascade control execution unit 6 that realizes a PID controller for controlling a batch reaction process, a model storage unit 1 that stores a reaction process model, a model adjustment unit 7, and a control parameter adjustment unit 5 that adjusts PID parameters of the PID controller using a transfer function model that linearly approximates the reaction process model. The model adjustment unit 7 smooths actual data for the coolant inlet temperature Tci and provides the smoothed data as an input to the reaction process model, calculates the squared error between time-series data for the reaction temperature Trav, which is the output of the reaction process model, and the actual data for the reaction temperature Tr, determines adaptive parameters for the reaction process model that minimize the squared error by nonlinear optimization, and adjusts the model parameters of the reaction process model using the adaptive parameters."

[0005] Furthermore, Patent Document 2 aims to provide a method for simulating the dynamic temperature behavior of at least a portion of a process plant on a laboratory scale, and describes a method for simulating the dynamic temperature behavior of at least a portion of a process plant using an experimental apparatus, the process plant including an industrial reaction vessel for performing a chemical and / or physical reaction, an industrial temperature control device that interacts with the industrial reaction vessel, an industrial heat transfer unit that interacts with the industrial temperature control device, and a control unit for controlling at least the industrial temperature control device, the experimental apparatus including an experimental reaction vessel for performing a chemical and / or physical reaction, an industrial temperature control device that interacts with the experimental reaction vessel, and a control unit for controlling at least the industrial temperature control device. "A method of simulating the dynamic temperature behavior of a process plant using an experimental apparatus, comprising: an experimental temperature control device; an experimental heat transfer unit interacting with the experimental temperature control device; and a controller for operating and controlling the experimental apparatus, wherein a temperature profile and / or temperature set points are provided at predetermined time intervals to control the experimental apparatus, the temperature profile and / or temperature set points being derived from a mathematical model, the mathematical model describing the dynamic temperature behavior of at least a portion of the process plant, and the temperature profile and / or temperature set points being used by the controller to control the experimental apparatus." [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2013-69094 [Patent Document 2] Japanese Patent Application Laid-Open No. 2007-87383 Summary of the Invention [Problem to be solved by the invention]

[0007] According to Patent Document 1, the objective can be achieved. However, in order to realize a tuning method for adjusting model parameters of a reaction process model using adaptive parameters, data from the target reactor is required. This means that tests must be conducted on a plant-scale reactor to adjust the plant model. As pointed out above, conducting tests on a plant-scale reactor can be expensive and time-consuming. In this regard, it is useful to employ a scale-up method that allows for adjustment of a plant-scale model based on experimental results using a laboratory-scale reactor, where tests on a smaller scale (such as in a laboratory) are more practical.

[0008] However, laboratory-scale reactors often exhibit different thermal behavior compared to plant-scale reactors due to the difference in scale, and therefore, scaling up heat transfer parameters from laboratory-scale reactors is important for predicting the behavior of plant-scale reactors.

[0009] In this regard, Patent Document 2 provides a means to scale up the heat rate from a laboratory-scale reactor to a plant-scale reactor in order to test new operating conditions. Specifically, Patent Document 2 demonstrates that the heat rate (Q lab ) is then calculated based on the difference in the volume (or reaction mass) of the lab-scale and plant-scale reactors. lab Apply a scale factor to the plant-scale heat (Q ind ) estimates based solely on reactor scale differences. lab ) may not accurately reflect the actual conditions of a plant-scale reactor. This approach in Patent Document 2 can reduce the accuracy of calculating plant-scale heat transfer parameters (Q ind :Patent Document 2).

[0010] In view of the above, an object of the present invention is to provide a reactor scale-up simulator and a reactor scale-up simulation method suitable for scaling up a reactor from a laboratory-scale reactor to a plant-scale reactor. [Means for solving the problem]

[0011] Based on the above, in the present invention, there is provided a reactor scale-up simulator for calculating changes in plant characteristics due to changes in condition α, which uses a first measurement value in a laboratory-scale reactor and a first model simulating the characteristics of the laboratory-scale reactor to calculate parameters β of the laboratory-scale reactor when the laboratory-scale reactor is operated under condition α1. lab、α1 and the parameter β calculated by the tuner. lab、α1 and the parameter β of the laboratory-scale reactor when the laboratory-scale reactor is operated under the condition α0. lab、α0 and the plant-scale reactor parameter β when the plant-scale reactor is operated under the condition α0. plant、α0 From this, the parameter β of the plant-scale reactor when operating the plant-scale reactor under condition α1 is plant、α1 The scaler to find and the parameter β lab、α1 , β lab、α0 , β plant、α0 , β plant、α1 a reactor scale-up simulator characterized by comprising: an updater that generates a second model that simulates the characteristics of a plant-scale reactor using the above; and a plant tester that calculates changes in the plant characteristics of the plant-scale reactor due to changes in condition α using the second model that simulates the characteristics of the plant-scale reactor.

[0012] Furthermore, in the present invention, there is provided a reactor scale expansion simulation method for calculating changes in plant characteristics due to changes in condition α using a computer, wherein the computer calculates parameters β of the laboratory-scale reactor when the laboratory-scale reactor is operated under condition α1 using a first measurement value of the laboratory-scale reactor and a first model simulating the characteristics of the laboratory-scale reactor. lab、α1 Calculate the parameter β lab、α1and the parameter β of the laboratory-scale reactor when the laboratory-scale reactor is operated under the condition α0. lab、α0 and the plant-scale reactor parameter β when the plant-scale reactor is operated under the condition α0. plant、α0 From this, the parameter β of the plant-scale reactor when operating the plant-scale reactor under condition α1 is plant、α1 Calculate the parameter β lab、α1 , β lab、α0 , β plant、α0 , β plant、α1 a second model that simulates the characteristics of a plant-scale reactor using the above formula, and a second model that simulates the characteristics of a plant-scale reactor is used to calculate changes in the plant characteristics of the plant-scale reactor due to changes in condition α. [Effects of the Invention]

[0013] According to the present invention, it is possible to provide a reactor scale-up simulator and a reactor scale-up simulation method suitable for scaling up a reactor from a laboratory-scale reactor to a plant-scale reactor.

[0014] Furthermore, according to the embodiment of the present invention, not only the actual conditions of the laboratory-scale reactor but also the actual conditions of the plant-scale reactor are taken into consideration. As a result, the accuracy of the scaled-up model parameters is improved. Specifically, according to the embodiment of the present invention, the plant-scale heat transfer parameter β under the existing condition α is plant、α0 By additionally using the actual plant state, represented by , the plant-scale model parameters can be more accurately estimated from laboratory tests. [Brief explanation of the drawings]

[0015] [Figure 1] FIG. 1 is a diagram showing an example of the hardware configuration of a reactor scale expansion simulator according to a first embodiment of the present invention. [Figure 2] FIG. 10 is a diagram showing an example of software processing of a reactor scale up simulator according to a second embodiment of the present invention. [Figure 3]FIG. 3 is a diagram showing specific processing content of processing step S10 in FIG. 2. [Figure 4] 10A and 10B show exemplary display output examples implemented by tuner 30. [Figure 5a] FIG. 10 is a diagram showing a first example of the process of the scaler 40 applied when α does not depend on the reactor scale. [Figure 5b] FIG. 10 is a diagram showing a second example of the process of the scaler 40 applied when α depends on the reactor scale. [Figure 6] FIG. 3 is a diagram showing the specific processing content of processing step S20 in FIG. 2. [Figure 7] FIG. 10 illustrates an example of a display output generated by evaluating the generated plant-scale simulation model. [Figure 8] FIG. 10 is a diagram showing an example of a processing flow that enables processing of both Case 1 and Case 2. [Figure 9] FIG. 10 is a diagram showing an example of the hardware configuration of a reactor scale-up simulator according to a third embodiment of the present invention. [Figure 10] FIG. 10 is a diagram showing the processing contents of a reactor scale-up simulator according to a third embodiment of the present invention. [Figure 11] FIG. 10 is a diagram showing an example of the hardware configuration of a reactor scale-up simulator according to a fourth embodiment of the present invention. [Figure 12] 10 is a diagram showing an example of software processing of a reactor scale-up simulator according to Example 4 of the present invention. [Figure 13] Figure 1 shows the effect of regular adjustment of βplant, α1 every Δt on scheduling plant maintenance. DETAILED DESCRIPTION OF THE INVENTION

[0016] Hereinafter, embodiments of the present invention will be described with reference to the drawings. [Example]

[0017] Fig. 1 is a diagram showing an example of the hardware configuration of a reactor scale-up simulator according to Example 1. Fig. 1 shows an example of the hardware configuration of a scale-up simulator 1 for simulating the characteristics (e.g., temperature and concentration) of a plant-scale reactor in a target chemical plant or industrial plant under new conditions α1. Note that a user of the present invention may be an operator of the target plant (a plant-scale reactor in a target chemical plant or industrial plant).

[0018] Functionally, the scale-up simulator 1 implemented using a computer has the functions of a tuner 30, a scaler 40, an updater 50, and a plant tester 60, as well as databases DB1 and DB2, an input device 3, and a display 4. It executes simulation processing for scale-up based on input from a laboratory-scale reactor R1. The arrows in Fig. 1 indicate the direction of data flow between the components of the scale-up simulator 1 to generate simulated characteristics of the target plant.

[0019] First, the laboratory-scale reactor R1 will be described. The laboratory-scale reactor R1 is preferably a reactor of the same design as the plant-scale reactor in the target plant. For example, it is preferable that it be a similar reactor type (e.g., batch type, continuous stirred tank, fixed bed, etc.), a similar agitator type, similar dimensional ratios (e.g., height-to-diameter ratio, diameter-to-agitator ratio, etc.), and a similar cooling or heating mechanism, such as a cooling or heating jacket (hereinafter also referred to as a jacket). In this example, it is assumed that the laboratory-scale reactor R1 and the plant-scale reactor are jacketed reactors. However, it goes without saying that the present invention is not limited to this type and can also be applied to reactors with other cooling / heating mechanisms.

[0020] The data collected from the laboratory-scale reactor R1 may include, for example, a time series of temperature (T obs、lab、α1 ) and enrichment (C obs、lab、α1), as well as the reactor surface or outer shell of the laboratory-scale reactor R1 (T c、obs、lab、α1 ) time series of temperature. To collect these data, the laboratory-scale reactor R1 is also preferably equipped with a temperature sensor and a sampling port for concentration measurement.

[0021] Next, we will explain the outline of the scale-up simulator 1. The scale-up simulator 1 according to Example 1 of the present invention calculates changes in plant characteristics (concentration, temperature, etc.) due to changes in reactor conditions α (α is a factor that affects the heat transfer parameters that describe the reactor other than the scale, such as the amount of additive, stirring speed, insulation material, etc.).

[0022] Here, the reactor condition α is considered to be the condition α0 and the condition α1, where the condition α0 is the existing condition in the current operation and the condition α1 is the unknown condition in the new operation. As a result, the parameters under the condition α1, which is an unexperienced value, are estimated using the condition α0 as an empirical value.

[0023] The scale-up simulator 1 illustrated in FIG. 1 uses the above data (T obs、lab、α1 , C obs、lab、α1 T c、obs、lab、α1 etc.) are required.

[0024] The parameters for laboratory scale heat transfer under condition α1 are as follows: β lab、α1 This means that the parameter β is for example a heat transfer parameter under the condition α1 in the laboratory scale lab.

[0025] Furthermore, the scale-up simulator 1 estimates parameters for heat transfer, for example, for the plant scale under the condition α1. The heat transfer parameters for the plant scale plant under the condition α1 are as follows: β plant, which will be denoted as α1.

[0026] Some of these heat transfer parameters are known values ​​that have already been found through past demonstrations or measurements, or are values ​​that can be estimated from measurement data, or are unknown values ​​that are not yet known.

[0027] Each part of the scale-up simulator 1 will be described in detail below. First, the database DB1 and the input device 3 will be described. The scale-up simulator 1 inputs measurement data of the laboratory-scale reactor R1 via the input device 3 and stores it in the database DB1, enabling various searches. The database DB1 can be a computer storage device. The measurement data obtained from the laboratory-scale reactor R1 includes at least a time-series temperature T obs、lab、α1 , concentration C obs、lab、α1 , and the time series temperature of the reactor surface or outer shell T c、obs、lab、α1 Includes.

[0028] Next, database DB2 stores in advance data (processing data) used in various processes of scale-up simulator 1 executed by the calculation unit (not shown) of the computer, and stores intermediate products and final deliverables as the results of the processes, enabling various searches of this data. Database DB2 can be a computer storage device.

[0029] The output data (intermediate products and final products) stored in the database DB2 includes at least the β lab、α1 , β calculated with scaler 40 plant、α1 , including the plant scale model Md2 generated by Updater 50.

[0030] Furthermore, the processing data stored in the database DB2 may be user input data via the input device 3 and / or data previously stored by the vendor or user of the present invention. The processing data includes input data for evaluating at least the model Md1 (laboratory-scale simulation model) and the model Md2 (plant-scale simulation model). The input device 3 may be, for example, a computer keyboard or a touchscreen display.

[0031] The tuner 30 calculates the plant-scale heat transfer parameters β lab、α1 is an unknown value but can be estimated, this estimation process yields β lab、α1 The tuner 30 performs an estimation calculation process of the temperature data (T) measured in the laboratory-scale reactor R1 under the operating condition (hereinafter simply referred to as the condition) α1 using the model Md1 stored in the database DB2. obs、lab、α1 ) and the simulated temperature data (T sim、lab、α1 ) and calculate β lab、α1 Generate.

[0032] β lab、α1 It is important to calculate α1 when α1 results in heat transfer characteristics that differ from the current conditions α0 under which the laboratory-scale reactor R1 is currently or previously operated. For example, increasing the amount of additive in the reactor from a current amount of 5 kg (α0) to a new amount of 15 kg (α1) can promote a faster heat transfer rate. As a second example, changing the coating material from the existing material A (α0) to material B (α1), which has a greater heat transfer coefficient (h), can also result in a faster heat transfer rate. In these cases, the heat transfer parameters based on condition α0 must be adjusted to accurately reflect the changes in plant behavior under condition α1.

[0033] Specifically, the tuner 30 measures the temperature T obs、lab、α1 and the estimated temperature T sim、lab、α1and based on this comparison, the heat transfer parameter β lab、α1 More specifically, the tuner 30 calculates the heat transfer parameter β lab、α1 As an example for calculating the σ, a data assimilation method (DA) such as a Kalman filter and an ensemble Kalman filter can be used.

[0034] In general, the data assimilation method DA involves calculating the difference between measured and simulated state values ​​and calculating adjusted model parameters based on the calculated difference. In the case of the tuner 30, the measured state values ​​are T obs、lab、α1 and the simulated state value is T sim、lab、α1 and the model parameters are β lab、α1 The data assimilation method DA has been shown to be effective in accurately adjusting model parameters such as heat transfer parameters.

[0035] Details of the calculation formula are presented in various literature, for example, "Oka, Y. & Ohno, M. (2020). Parameter estimation for heat transfer analysis in casting processes based on ensemble Kalman filter. International Journal of Heat and Mass Transfer, 1-9." Therefore, it is possible to apply it.

[0036] Next, we will explain the model Md1 pre-stored in the database DB2. The model Md1 is a simulation model that represents the behavior of the laboratory-scale reactor R1 under condition α1. Specifically, the model Md1 is a physics-based simulation model that includes model equations and pre-defined model parameter values.

[0037] The pre-stored model equations for model Md1 include at least a mass balance equation, a reaction rate equation, and an energy balance equation describing the laboratory-scale reactor R1.

[0038] Generally, the model parameters relate to system properties, such as heat transfer properties, physical properties, chemical properties, fluid dynamic properties, etc. The predefined model parameter values ​​for the model Md1 include at least values ​​for the dimensions (e.g., diameter, height, and thickness) of the laboratory-scale reactor R1, kinetic parameters (e.g., activation energy and Arrhenius constant) of the reaction in the laboratory-scale reactor R1, properties of the reacting species in the laboratory-scale reactor R1, and properties of the jacket material (e.g., density and specific heat capacity). The predefined model parameters are independent of the condition α and can be easily retrieved from the design data of the laboratory-scale reactor R1.

[0039] On the other hand, the heat transfer parameter β of the laboratory-scale reactor R1 under the new condition α1 lab、α1 is expected to be different from the condition α0. Therefore, the new condition α1 is not known with certainty compared to other model parameters. Furthermore, this is because the pre-stored model Md1, as mentioned above, lab、α1. This means that we do not need a predefined numerical value for β lab、α1 is calculated via the tuner 30.

[0040] Furthermore, physics-based simulation models such as model Md1 are generally a function of boundary conditions. Boundary conditions relate to independent state values ​​(e.g., temperature, flow rate, pressure, and concentration) that affect the calculated state / s of the model. Model Md1 includes at least C obs、lab、α1 and T c、obs、lab、α1 Includes:

[0041] Next, the scaler 40 will be explained. The scaler 40 is a heat transfer parameter β of a plant-scale reactor under the condition α1. plant、α1 More specifically, the scaler 40 calculates β plant、α0 and β lab、α0 Using β lab、α1 By scaling up β plant、α1 Calculate β plant、α0 is the heat transfer parameter describing the plant-scale reactor in the current state α0.

[0042] In Updater 50, the β calculated with Scaler 40 lab、α1 and input data I2 previously stored in database DB2 to generate a plant-scale reactor model Md2. The input data I2 is pre-stored data stored in database DB2 and / or user-entered data. The input data I2 includes at least the equation of the reaction occurring in the plant-scale reactor and reactor specifications such as the type of reactor.

[0043] The plant tester 60 evaluates the model Md2 obtained by the updater 50. This evaluation is performed based on input data I3 previously stored in the database DB2. The input data I3 includes boundary conditions, an initial simulation time tstart (e.g., t=0), and an end trend of the simulation. The input data I3 can be user-entered data or pre-stored data. The boundary conditions are time-series values ​​that can be based on the measured jacket temperature or virtual jacket temperature of a plant-scale reactor.

[0044] On the display 4, various data in the computer are appropriately edited and visualized and displayed. [Example]

[0045] In Example 1, an example of the hardware configuration of the reactor scale expansion simulator was explained, whereas in Examples 2 and 3, examples of software processing of the reactor scale expansion simulator will be explained.

[0046] FIG. 2 is a diagram illustrating an example of software processing of a reactor scale-up simulator according to a second embodiment of the present invention. This processing flow outlines the overall processing content in the computing unit of the computer. Here, an example of processing performed by the scale-up simulator 1 to generate simulated behavior of a plant-scale reactor in state α1 is described.

[0047] First, in processing step S10, the scale-up simulator 1 performs calculations on the tuner 30, and β lab、α1 A specific example of this process will be described later with reference to Fig. 3. Then, the process proceeds to process step S20.

[0048] In processing step S20, the scale up simulator 1 performs the calculation of the scaler 40, and β plant、α1 Then, the process proceeds to step S30.

[0049] In process step S30, the scale-up simulator 1 executes the processing of the updater 50 to generate the plant-scale simulation model Md2, and then the process proceeds to process step S40.

[0050] In processing step S40, the scale-up simulator 1 evaluates the model Md2 in the plant tester 60 and presents the calculated dynamic plant characteristics on the display 4. In this way, the user can investigate how the plant-scale reactor behaves under the condition α1 without performing trials under the condition α1 in the target plant.

[0051] The detailed processing contents of processing step S10 are shown in FIG. lab、α1 A specific calculation method is shown.

[0052] 3, first, in processing step S100, the tuner 30 searches for a model Md1 stored in advance in the database DB2 and reads out input data I1 to evaluate the model Md1. Specifically, the input data I1 includes values ​​of the boundary conditions of the model Md1. The boundary conditions are defined as the C measured at least at time t. obs、lab、α1 and T c、obs、lab、α1(which are stored in database DB1). The input data I1 further includes a value for the current simulation time t (e.g., t=0) and a simulation end time tend. The values ​​of time t and trend can be predefined values ​​stored in database DB2 or user-entered values ​​via input device 3.

[0053] Then, the process proceeds to step S101. In step S101, the tuner 30 calculates β lab、α1 As mentioned above, the model Md1 generates an estimate of β lab、α1 Since there is no predefined value for β, a guess is required. Specifically, the guess generated in process step S101 is a random value, a value based on user input via input device 3, or a heat transfer parameter value (pre-stored in database DB2) describing laboratory-scale reactor R1 in the existing state α0. Note that in process step S101, β lab、α1 is still a figure with little certainty.

[0054] Then, the process proceeds to step S102. In the simulation evaluation process in step S102, the β calculated in step S101 is lab、α1 and β calculated in processing step S106. lab、α11 In either case, in processing step S102, the tuner 30 evaluates an equation including the model Md1 to obtain T sim、lab、α1 This process produces T sim、lab、α1 is generated.

[0055] In the process of process step S102 executed following process step S101, specifically, the boundary condition value read in process step S100 and the β generated in process step S101 are compared. lab、α1In the process of process step S102 executed after process step S106, specifically, the boundary condition updated in process step S106 and the estimated value of β lab、α1 The value of the model Md1 is evaluated using the following: The details of processing step S106 will be described later.

[0056] Thereafter, the process proceeds to step S103. In step S103, the tuner 30 calculates the temperature T measured at time t+1 stored in the database DB1. obs、lab、α1 Read out.

[0057] Then, the process proceeds to step S104. In step S104, the tuner 30 calculates the T sim、lab、α1 and T read out in processing step S103 obs、lab、α1 The tuner 30 then calculates and compares the difference between the calculated T sim、lab、α1 and T obs、lab、α1 Using the difference with lab、α1 Calculate the amount of tuning for T sim、lab、α1 and T obs、lab、α1 The tuned β lab、α1 Once calculated, the process of step S104 ends.

[0058] Then, processing proceeds to processing step S105. In processing step S105, the tuner 30 checks whether the end of the calculation has been reached. The tuner 30 can check whether the end of the calculation has been reached, for example, by checking whether the simulation time t+1 tends to be equal. If the end of the calculation has been reached, processing proceeds to processing step S107. If the end of the calculation has not been reached, processing proceeds to processing step S106.

[0059] In processing step S106, the tuner 30 receives the input data I1 and β lab、α1Updating the input data I1 involves updating the current simulation time t to t=t+1 and obtaining the boundary conditions at the updated current simulation time t. Meanwhile, β lab、α1 The value of is replaced with the tuning value calculated in processing step S104 and updated. Then, the processing returns to processing step S102. Thereafter, the processing is repeated.

[0060] According to the above repeated process, the T calculated in process step S102 is calculated using the updated β value from process step S106. sim、lab、α1 is the T calculated in process step S102 using the estimated β value from process step S101. sim、lab、α1 This is because the β value is adjusted from the initial guess to the tuning value generated by tuner 30 in process step S104.

[0061] In processing step S107, the tuner 30 stores the β lab、α1 This calculation is based on the calculated β from the first simulation time t. lab、α1 For example, the tuner 30 may analyze the calculated T sim、lab、α1 and T obs、lab、α1 If the difference between lab、α1 This threshold value v can be pre-stored by the user or vendor of the invention.

[0062] 4 shows an exemplary display output from implementing tuner 30. The horizontal time axis display output (plots P100, P101, P102) shown in FIG. 4 allows a user of the present invention to visualize and understand the tuning process for laboratory-scale heat transfer parameters.

[0063] Plot P100 shows the value β of the model Md1 calculated in processing step S102 during the period from the initial simulation t=0 to the end of the simulation t=tend (not shown).lab、α1 The plot P101 is used to evaluate the time series of T calculated in the processing step S102. sim、lab、α1 and T read out in processing step S103 from time t=1. obs、lab、α1 The plot 102 shows the time series of T calculated in processing step S104. sim、lab、α1 and T obs、lab、α1 The figure shows the difference between the two, and indicates that the difference decreases and converges.

[0064] According to the above display example, as already explained, β lab、α1 The value of is only an estimate. sim、lab、α1 As shown in plot P101, at t=1, T obs、lab、 Also, as shown on page 102, T at t=1 sim、lab、α1 and T obs、lab、α1 Based on the calculated difference, the tuner 30 calculates T sim、lab、α1 Tuned β is set as the updated β value to predict lab、α1 Calculate.

[0065] Prediction T at t=2 sim、lab、α1 and T obs、lab、α1 The difference between the time t_conv and the predicted temperature T sim、lab、α1 and the measured temperature T obs、lab、α1 The difference between the calculated β and the threshold v is smaller than the threshold v. lab、α1 is the final β final. It also becomes stable.

[0066] T from time t_conv sim、lab、α1 and T obs、lab、α1 The small difference between lab、α1 The final β calculated in processing step S107 is lab、α1 is the average β of the trend from time t_conv lab、α1 It can be said that:

[0067] From here, the beta of Scaler 40 plant、α1 A first example of the calculation process of β will be described. The first example of the calculation process is an example of the process of the scaler 40 that is applied when the condition α does not depend on the reactor scale. lab、α1 From β plant、α1 To scale up to α, scaler 40 first calculates the slope of β (dβ / dL) versus reactor scale L at a constant α, for example, by evaluating equation (1): α Calculate. [Number 1] (dβ / dL) α =(β plant、α0 -β lab、α0 ) / (L plant -L lab ) ····(1) Here, the reactor scale L is, for example, the height, diameter, volume, or mass of the reactor. lab、α0 is a value previously stored in the database DB2, obtained from past measurements (temperature, concentration, etc.) of the laboratory-scale reactor R1 in the current state α0. Similarly, β plant、α0 are also pre-stored values ​​in database DB2 obtained from past measurements (e.g., temperature and concentration) of the plant-scale reactor at the existing state α0.

[0068] β lab、α0 and β plant、α0 is based on past measurements (temperature, concentration, etc.) of a plant-scale reactor under existing conditions α0, so (dβ / dL) calculated from Eq. (1) α represents the variation of β based on the actual conditions of both the laboratory-scale reactor R1 and the plant-scale reactor.

[0069] Then, after calculating dβ / dL, the scaler 40 calculates β by, for example, evaluating equation (2). plant、α1 Calculate. [Number 2] β plant、α1 =β lab、α1 +(dβ / dL) α [(L plant -L lab )] ····(2) Equation (2) is β plant、α1 dβ / dL [(L plant -L lab )] term, β lab、α1 (dβ / dL) α is β plant、α0 Since it is calculated using plant、α1 can be scaled up by also considering the actual conditions of the plant-scale reactor. This approach is plant and L lab This approach is more accurate than the conventional approach (e.g., Patent Document 2) that scales up based on the difference in scale using β plant、α1 β lab、α1 scales as a function of L plant , L lab This approach uses β lab、α1 , β lab、α0 and β plant、α0 β as a function of plant、α1 Expand the scale of

[0070] FIG. 5a shows an example of the display output by the scaler 40 after the above-mentioned processing. The display output shows the reactor scale L on the horizontal axis and the condition α on the vertical axis. plant、α1 are displayed so that the actual numerical values ​​can also be recognized as displayed diamonds.

[0071] On the other hand, β plant、α1 The heat transfer parameters (β lab、α1 , β lab、α0 and β plant、α0 ) are displayed as circles with specific numerical values. This display output allows the user of the present invention to easily visualize the calculated β plant、α1 It is also useful to visualize how changing α from α0 to α1 (and reactor scale) affects β.

[0072] In this way, in the present invention, β plant、α1 The heat transfer parameters (β lab、α1 , β lab、α0 and β plant、α0 ), some of which are already known values ​​(β lab、α0 and β plant、α0 ), or a value that can be estimated from measurement data (β lab、α1 ) and from these we can determine the unknown value β plant、α1 It is estimated that:

[0073] In the above example, a first example of the process of the scaler 40 in process step S20 has been described. In contrast, a second example of the process of the scaler 40 in process step S20 will be described below. This second example is applied when α depends on the reactor scale. This second example will be described with reference to FIG. 5b.

[0074] In many cases, the effect of condition α is independent of the reactor scale. For example, the type of jacket material, the type of additive, or the agitation speed has the same effect regardless of the reactor scale. In such cases, if a plant operator wants to test a new jacket material, a new type of additive, or a new agitation speed, the plant operator can test the new jacket material, the new type of additive, or the new agitation speed in a laboratory-scale reactor without any adjustments.

[0075] Therefore, as shown in Figure 5a, β lab、α1 and β plant、α1 corresponds to the same value of α equal to α1, and β lab、α0 and β plant、α0 correspond to the same value of α=α0. In this case, the method of calculation example 1 is useful.

[0076] However, in some cases, the effect of the condition α depends on the size of the reactor. More specifically, 5 kg of additive in a large reactor does not have the same effect as 5 kg of additive in a small reactor. In this case, it is common to proportionally adjust the amount of additive added to a laboratory-scale reactor to mimic the effect of adding 5 kg of additive in a plant-scale reactor. Therefore, to test a new condition α for a plant-scale reactor, the adjustment value α, lab is performed in a laboratory-scale test. This also means that the actual value of the plant-scale reactor, α0, is adjusted to α0, lab This means that the process is or has been applied to laboratory-scale reactors.

[0077] Figure 5b shows an example of the display output produced by scaler 40 when α depends on the reactor scale. lab、α0 and β´ lab、α0 In the second example in Fig. 5b, α is characterized by its reactor scale dependence, and β lab、α0 and β plant、α0 The values ​​of β' are not in a corresponding relationship, as will be described later. lab、α0 and β' lab、α0 β plant、α1 is calculated in the scaler 40 as a heat transfer parameter as an intermediate product for producing

[0078] In this example, β lab、α0 and β plant、α0 is a known value, and β lab、α1 is a value that can be estimated from the measurement data, and the unknown value β plant、α1 When estimating β' as an intermediate product, lab、α0 and β' lab、α0 Through β plant、α1 The procedure for generating

[0079] Again, as shown in Figure 5b, β lab、α1 and β plant、α1corresponds to various values ​​of α, since α depends on the reactor scale. lab、α0 and β plant、α0 corresponds to various values ​​of α. More specifically, β lab、α1 The condition α=α 1、lab (instead of condition α1), and β lab、α0 is the condition α = α 0、lab (instead of the condition α0). As explained, α 1、lab and α0, lab are the adjusted values ​​of α1 and α0, respectively. α 1、lab and α0, lab can be conventionally estimated based on the scale L of the plant-scale and laboratory-scale reactors.

[0080] In this second example, scaler 40 first calculates (dβ / dL) α Also calculate β lab、α1 and β plant、α1 refers to different values ​​of α, so when applying equation (1) as in the first example, (dβ / dL) α Instead, the scaler 40 applies equations (1a) to (1c). [Math 1a] (dβ / dL) α =(β plant、α0 -β´ lab、α0 ) / (L plant -L lab ) (1a) [Number 1b] β´ lab、α0 =β lab、α0 +(dβ / dα) L (α0-α 0、lab )···(1b) [Number 1c] (dβ / dα) L =(β lab、α1 -β lab、α0 ) / (α 1、lab -α 0、lab ) (1c) Using equation (1a), the scaler 40 calculates new parameters β'lab,α0, which refer to the laboratory-scale heat transfer parameters under the condition of α corresponding to α0. Therefore, as shown in Figure 5b, β'lab,α0 and βplant,α0 correspond to the same numerical value of α corresponding to α0.

[0081] Scaler 40 can calculate β'lab, α0 using equation (1b), where (dβ / dα)L refers to the gradient of β with respect to α at constant L. Note that the remaining terms in equation (1b) are known values.

[0082] Additionally, the scaler 40 can calculate (dβ / dα)L using equation (1c), which can be evaluated using pre-stored values ​​of βlab, α0, α1,lab, and α0,lab, as well as βlab, α1 calculated by the tuner 30.

[0083] Next, the scaler 40 calculates β using equations (2a) and (2b). plant、α1 Calculate. [Number 2a] β plant、α1 =β´ lab、α1 +(dβ / dL) α [(L plant -L lab )] (2a) [Number 2b] β´ lab、α1 =β lab、α1 +(dβ / dα) L (α1-α 1、lab ) (2b) Specifically, using equation (2a), the scaler 40 calculates the new parameter β' lab、α1 Calculate β´ lab、α1 refers to the laboratory-scale heat transfer parameter under the condition of α, which corresponds to α1. Therefore, as shown in Fig. 5b, β' lab、α1 and β plant、α1 refers to the same value for α = α1. The scaler 40 calculates β' using equation (2b). lab、α1where all necessary parameters are assumed to be known values ​​and can be used in subsequent processing.

[0084] Overall, regardless of whether α is reactor scale dependent, scaler 40 uses data on heat transfer parameters at existing conditions α for the lab-scale reactor and the plant-scale reactor, as well as data on heat transfer calculated by tuner 30 at new conditions α to calculate β. plant、α1 can be calculated.

[0085] Next, a specific example of a method in which the updater 50 generates the model Md2 in processing step S30 of FIG. 2 will be described with reference to FIG.

[0086] In processing step S300 of Fig. 6, the updater 50 reads input data I2 from the database DB2. The input data I2 is data pre-stored in the database DB2 and / or user-entered data. The input data I2 includes at least a reaction equation occurring in a plant-scale reactor and reactor specifications such as the reactor type. Equation (3) shows an example of a reaction equation included in the input data I2. [Number 3] a1A+a2B>a3C (3) (3), where A and B are reactants, C is a product, and a1, a2, and a3 are stoichiometric coefficients. The reactor type specified in input data I2 can be a batch reactor, semi-batch reactor, continuous stirred tank reactor, plug flow reactor, fixed bed reactor, etc. The remainder of this description assumes that the reactor type is a batch reactor. After reading the reactor specifications, processing proceeds to step processing step S301.

[0087] In process step S301, the updater 50 constructs a model equation describing the mass and energy balance of the plant-scale reactor by using the reactor specifications read in process step S300. Although the example of the model equation constructed in this step is described using equation (3) as the reaction equation and a batch reactor as the reactor type, the present invention may also be applied to reaction equations different from equation (3) and different types of reactors.

[0088] First, the updater 50 generates a mass balance equation that follows the law of conservation of mass. Equation (4) shows an example of a mass balance equation generated based on reaction equation (3) and a batch reactor. [Number 4] d[A] / dt=-a1r d[B] / dt=-a2r d[C] / dt=a3r (4) In equation (4), d[A] / dt, d[B] / dt, and d[C] / dt are the differential changes in the concentrations of A, B, and C with respect to time t, and r is the reaction rate. Furthermore, the updater 50 generates an equation for r according to the law of mass action and the Arrhenius law. Equation (5) shows an example equation generated based on reaction equation (3). [Number 5] r = exp(Ar-Ea / RT)[A] a1 [B] a2 ···(5) In equation (5), Ar (Arrhenius constant) and Ea (activation energy) are kinetic parameters, R is the universal gas constant, [A] and [B] are the concentrations of reactants A and B, and T is the reactor temperature. Additionally, the updater 50 generates an energy balance equation for at least the plant-scale reactor temperature T by applying the law of conservation of energy. Equation (6) shows an example of an energy balance equation generated based on reaction equation (3) and a batch reactor. [Number 6] m r Cp r (dT / dt)=-rH+UA(T j -T) ···(6) (6) where dT / dt is the difference in reactor temperature with respect to time t, and T j is the jacket temperature, and m r is the mass inside the plant-scale reactor, and Cp r is the heat capacity associated with the reactor species, H is the enthalpy of reaction associated with reaction (3), U is the total heat transfer coefficient, A is the heat transfer area between the reactor contents and the jacket material, and T j is the jacket temperature. Updater 50 uses T j It is possible to further generate an energy balance equation for: Since this equation is well known, the details of the equation will not be explained here.

[0089] Process step S301 is completed when at least the energy balance, mass balance, and reaction rate equations have been generated, and processing then proceeds to process step S302.

[0090] In process step S302, the updater 50 calculates at least the β calculated by the scaler 40 in process step S20. plant、α1 is used to define the model parameters of the equation generated in processing step S301. Since β is a heat transfer parameter, in the energy balance equation (6), Cp r , U, UA, etc. For example, if β specifically refers to the overall heat transfer coefficient U, the updater 50 may set U to β plant、α1 The updater 50 further updates the remaining model parameters, such as the motion parameters (Ar and Ea), m, and m from the pre-stored data in the database DB2 or from user-entered data. r , Cp r , H, and A values ​​can be defined.

[0091] Instead of using pre-stored data, H may also be defined by the updater 50 by evaluating the following expression: [Number 7] H=Σai H f、product、i -Σa j H f、rectant , j (7) (7) In the formula, H f is the enthalpy of formation, and a is the stoichiometric coefficient according to the reaction formula (e.g., reaction formula (3)). Specifically, H for reaction formula (3) can be defined as the following formula (8). Note that H f are readily available in the literature and can be pre-stored in the database DB2. [Number 8] H=H f、C -(H f、A +H f、B ) ···(8) Processing step S302 in Figure 6 is completed when the model parameters have been assigned values, and processing then proceeds to processing step S303.

[0092] In process step S303, the formula generated in process step S301 is used as a model formula including the values ​​defined in process step S302 as specific model parameter values ​​to generate a plant scale model Md2. The generated model Md2 is stored in database DB2. Then, the process ends.

[0093] As will be described later, the scale-up simulator 1 enables simulation of a plant-scale reactor by evaluating the generated model Md2. More specifically, the model Md2 uses β as a model parameter value. plant、α1 Applying [mathematical formula - see original document], evaluating the model Md2 can simulate the plant-scale reactor at the new state α1.

[0094] Hereinafter, an example of evaluation of the model Md2 by executing the plant tester 60 in step S40 of Fig. 2 will be described. This example will be described taking into consideration that the model Md2 generated by the updater 50 is specifically composed of equations (4) to (6) as model equations.

[0095] First, the plant tester 60 searches for the model Md2 generated by the updater 50 in step processing step S30.

[0096] Second, the plant tester 60 reads input data I3 from the database DB2 to evaluate the model Md2. The input data I3 includes boundary conditions, an initial simulation time tstart (e.g., t=0), and an end time of the simulation. The input data I3 can be user-entered data or pre-stored data. The boundary conditions are a time series T from the initial simulation time tstart onward, which can be based on the measured or virtual jacket temperature of the plant-scale reactor. j value.

[0097] Third, the plant tester 60 reads the boundary conditions of the model Md2, and then evaluates the model Md2 to calculate simulated concentration and temperature values. In this example, the plant tester 60 generates predicted values ​​of the concentrations [A], [B], and [C] obtained by the model Md2 based on equation (4), and a change trend of the predicted value of the reactor temperature from the initial simulation time tstart based on equation (5).

[0098] Figure 7 shows an example of the output generated by evaluating the generated plant-scale simulation model Md2. The displayed output shows, from the top to the bottom, time series predictions of the concentrations of various elements A, B, and C, a time series prediction of the reactor temperature, and the value of the jacket temperature used as a boundary condition from the initial simulation time tstart to the following time points.

[0099] By viewing this display, users can easily visualize the temperature and concentration behavior of a plant-scale reactor under the specified boundary condition, condition α1. That is, they can know the boundary condition values ​​that can give the desired reactor temperature and concentration behavior under condition α1. This knowledge can then be used to set boundary conditions such as jacket temperature when implementing condition α1 in an actual factory. Overall, users can save the time and resources required to conduct tests on a plant-scale reactor under the conditions of α1.

[0100] Considering that there exist two cases, Case 1 shown in Fig. 5a and Case 2 shown in Fig. 5b, the reactor scale up simulator 1 of the present invention should be able to handle both of them. For this reason, it is preferable to perform the process of Fig. 8 so that the separation process can be performed.

[0101] In the first processing step S200 of FIG. 8, data β lab、α0 , β plant、α0 and β lab、α0 , β plant、α0 The condition α corresponding to the reactor scale L lab , L plant Enter.

[0102] In the next processing step S201, the gradient of β (dβ / dL) with respect to the reactor scale L at a constant α is calculated using equation (1). α Calculate.

[0103] In processing step S202, β lab、α0 and β plant、α0 It is confirmed that there is a relationship that corresponds to the same value of equal α, and if there is an equal relationship, the method of Processing Example 1 is executed, and if there is not an equal relationship, the method of Processing Example 2 is executed. This distinguishes between the case where α depends on the reactor scale and the case where it does not.

[0104] If α depends on the reactor scale, β lab、α1 β proportionally to plant、α1 Calculate the following.

[0105] When α does not depend on the reactor scale, the intermediate product β' is lab、α0 is calculated, and the slope (dβ / dα)L is found in processing step S205, and this is used to calculate β plant、α1 Calculate the following.

[0106] By using the procedure of Example 2, not only the actual conditions of a laboratory-scale reactor but also the actual conditions of a plant-scale reactor are taken into consideration. As a result, the accuracy of the scaled-up model parameters is improved. Specifically, according to the embodiment of the present invention, the plant-scale heat transfer parameter β under the existing condition α is calculated. plant、α0 By additionally using the actual plant state, represented by , the plant-scale model parameters can be more accurately estimated from laboratory tests. [Example]

[0107] In Example 2, β plant、α0 and β lab、α0 Based on β lab、α1 By scaling up the plant β in the new state α1, plant、α1 This explains how to calculate the heat transfer parameters of β plant、α0 and β lab、α0 is the reactor is operating or has been operating 、 Since it is based on past measurements, β plant、α1 The calculation of reflects the realities of both laboratory-scale and plant-scale reactors. While this approach provides a more accurate estimate than traditional approaches, it does not provide a means to incorporate the actual conditions of a plant-scale reactor.

[0108] In the previous example, β plant、α0 and β lab、α0 is already available as a user input or pre-stored information. However, in some cases, only measurements at state α are readily available, and β plant、α0 and β lab、α0is still unavailable. Therefore, in this example, we use the measured value of α0 to calculate β plant、α0 and β lab、α0 We provide a method to calculate β using the measured values. plant、α0 and β lab、α0 In the following, in order to avoid duplication, only the differences from the previous embodiment will be explained.

[0109] First, an example of the hardware configuration of a reactor scale-up simulator according to Example 3 of the present invention is shown in Fig. 9. The scale-up simulator 1a shown in Fig. 9 is configured to be able to input from a plant-scale reactor R2 in addition to the laboratory-scale reactor R1. The scale-up simulator 1a shown in Fig. 9 further includes a parameter tuner 31, a tuner 32, and a database DB3 in addition to the configuration of Fig. 1. The arrows in Fig. 9 indicate the β of the scale-up simulator 1a. plant、α0 and β lab、α0 The flow of information in each part for calculating β plant、α0 and β lab、α0 The calculation method will be described later with reference to FIG.

[0110] The scale-up simulator 1a can further store and retrieve measurement data of the laboratory-scale reactor R1 under the condition α0 in a database DB1. The measurement data includes at least the time series of temperature (T obs、lab、α0 ) and enrichment (C obs、lab、α0 ) and the reactor surface or outer shell (T c、obs、lab、α0 ) and time series temperature.

[0111] Furthermore, the scale-up simulator 1a can additionally store and search the measured values ​​of the plant-scale reactor R2 under the condition α0 in the database DB3. The measured values ​​include at least the time-series temperature (T obs、plant、α0 ) and enrichment (C obs、plant、α0 ) and the reactor surface or outer shell (T c、obs、plant、α0) and a time series of temperatures. The database DB3 may be, for example, a computer storage device.

[0112] Hereinafter, the processing contents of the scale-up simulator 1a according to the third embodiment of FIG. 10 will be described. plant、α0 and β lab、α0 This processing can be performed by the scaling simulator 1a as a pre-processing step of the processing described in FIG.

[0113] In processing step S10a of FIG. 10, the scale-up simulator 1a executes the parameter tuner 31 to calculate β lab、α0 Then, the process proceeds to step S10b. A specific implementation example of the parameter tuner 31 will be described later.

[0114] In processing step S10b, the scale-up simulator 1a executes the tuner 32 to obtain β plant、α0 , and the process ends. Exemplary methods for implementing tuner 32 are also described below.

[0115] Hereinafter, the details of the parameter tuner 31 will be described, focusing on the differences from the tuner 30. In the parameter tuner 31, α0(T obs、lab、α0 ) and the measured temperature data of the laboratory-scale reactor R1 at α0(T sim、lab、α0 ) and the simulated temperature data of the laboratory-scale reactor R1 at β lab、α0 is generated using the laboratory-scale model Md1a. Specifically, the parameter tuner 31 generates T obs、lab、α0 and T sim、lab、α0 Based on this comparison, β lab、α0 Calculate.

[0116] Next, we will explain model Md, but as a premise, we have denoted the model representing the behavior of laboratory-scale reactor R1 used in post-processing in Figure 1 as Md1, and the model representing the behavior of the plant-scale reactor used in post-processing in Figure 1 as Md2. In contrast, we will denote the model representing the behavior of laboratory-scale reactor R1 used in pre-processing in Figure 9 as Md1a, and the model representing the behavior of plant-scale reactor R2 used in post-processing in Figure 9 as Md2a.

[0117] The details of the pre-processing model Md1a will now be described, focusing on important similarities and differences with the post-processing model Md1.

[0118] The pretreatment model Md1a is a simulation model that represents the behavior of the laboratory-scale reactor R1 under condition α0. Like the posttreatment model Md1, the pretreatment model Md1a is also a physics-based model that includes model equations and predefined model parameter values.

[0119] Similar to the post-treatment model Md1, the pre-stored model equations for the pre-treatment model Md1a also include at least a mass balance equation, a reaction rate equation, and an energy balance equation describing the laboratory-scale reactor R1.

[0120] The predefined model parameter values ​​for the pretreatment model Md1a also include at least values ​​for the dimensions of the laboratory-scale reactor R1, kinetic parameters of the reaction in the laboratory-scale reactor R1, properties of the reactive species in the laboratory-scale reactor R1, and properties of its jacket material.

[0121] Heat transfer parameter β for two laboratory-scale reactors under the condition α0 lab、α0 teeth . The parameter tuner 31 is calculated from the pre-processing model Md1a, and β lab、α0. In this case, the boundary state of the preprocessing model Md1a is at least C obs、lab、α0 and T c、obs、lab、α0 Includes.

[0122] Next, the β lab、α0 Here, a specific example of the processing of the parameter tuner 31 in processing step S10a will be described with reference to FIG.

[0123] First, in processing step S100, the parameter tuner 31 searches for the preprocessing model Md1a, reads out the input data I4, and evaluates the preprocessing model Md1a. Specifically, the input data I4 read out in processing step S100 includes the values ​​of the boundary conditions of the preprocessing model Md1a. The boundary conditions include at least the C measured at time t obs、lab、α0 and T c、obs、lab、α0 (stored from database DB1). Input data I4 further includes values ​​for the current simulation time t (e.g., t=0) and the simulation end time tend. The values ​​for time t and trend can be predefined values ​​stored in database DB2 or user-entered values ​​via input device 3.

[0124] In processing step S101, the parameter tuner 31 calculates β lab、α0 Generate an estimate of

[0125] As mentioned above, the preprocessing model Md1a is lab、α0 Since there is no predefined value for β, it is important to determine the validity of the guessed value. Specifically, the guessed value generated in processing step S101 may be a random value or a guessed value based on a user input value via the input device 3. Note that in processing step S101, β lab、α0 is most likely not a certainty.

[0126] In processing step S102, the parameter tuner 31 evaluates an equation including the preprocessing model Md1a to obtain T sim、lab、α0This process generates a predicted temperature of T sim、lab、α0 is generated.

[0127] In processing steps S101 and S102, the boundary condition value read in processing step S100 and the β generated in processing step S101 are compared. lab、α0 In process step S102 from process step S106, the pre-processing model Md1a is evaluated using the boundary condition updated in process step S106 and the estimated value of β lab、α0 The output of the preprocessing model Md1a is evaluated using

[0128] In processing step S103, the parameter tuner 31 calculates the temperature T measured at time t+1 stored in the database DB1. obs、lab、α0 Read out.

[0129] In process step S104, the parameter tuner 31 calculates T sim、lab、α0 and T read out in processing step S103 obs、lab、α0 The parameter tuner 31 then calculates and compares the difference between the calculated T sim、lab、α0 and T obs、lab、α0 Using the difference with lab、α1 Calculate the amount of tuning for T sim、lab、α0 and T obs、lab、α0 The tuned β lab、α0 Once calculated, the process of step S104 ends.

[0130] In processing step S106, the parameter tuner 31 calculates the input data I4 and β lab、α0 Updating the input data I4 includes updating the current simulation time t, e.g., t=t+1, to obtain the boundary conditions at the updated current simulation time t. Meanwhile, β lab、α0 The value of is replaced with the tuning value calculated in processing step S104 and updated, and the processing then returns to processing step S102.

[0131] The T calculated in process step S102 using the updated β value from process step S106 sim、lab、α0 is the T calculated in process step S102 using the estimated β value from process step S101. sim、lab、α1 is expected to be more accurate than (1) because the β value has been adjusted from the initial guess for the tuning value generated by parameter tuner 31 in process step S104.

[0132] In processing step S107, the parameter tuner 31 is stored in the database D2 and the β lab、α0 This calculation is based on the calculated β from the first simulation time t. lab、α0 For example, the parameter tuner 31 may analyze the calculated T sim、lab、α0 and T obs、lab、α0 If the difference between is less than the threshold value v1, lab、α0 This threshold value v1 can be pre-stored by the user or vendor of the present invention.

[0133] The final β is generated through the iterative adjustment in processing step S104 performed by the parameter tuner 31 using the measurements under the condition α. lab、α0 is expected to accurately represent the behavior of the laboratory-scale reactor R1 under conditions α. This treatment also allows for the use of β, which is not available in the previous embodiment. lab、α0 Automate the calculation of

[0134] From here, the details of the tuner 32 in FIG. 9 will be described, focusing on only the differences from the tuner 30 in FIG.

[0135] The tuner 32 in FIG. 9 uses the plant-scale simulation model Md2a to calculate α0(T obs、plant、α0 ) and the measured temperature data of the plant-scale reactor R2 at α0(T sim、plant、α0) and the calculated β plant、α1 Specifically, the tuner 32 generates T obs、plant、α0 and T sim、plant、α0 Based on this comparison, β plant、α0 Calculate.

[0136] The model Md2a is a simulation model that represents the behavior of the plant-scale reactor R2 under the condition α. ​​Specifically, the model Md2a is a physics-based model that includes a model equation and predefined model parameter values.

[0137] The pre-stored model equations for model Md2a include at least the mass balance equation, the reaction rate equation, and the energy balance equation that describe the behavior of plant-scale reactor R2.

[0138] The predefined model parameter values ​​for model Md2a include at least values ​​for the dimensions of plant-scale reactor R2, kinetic parameters of the reaction in plant-scale reactor R2, properties of the reactive species in plant-scale reactor R2, and properties of the jacket material.

[0139] As mentioned above, the heat transfer parameter α0 for the plan-scale reactor R2 condition is β plant、α0. and β plant、α0 should be calculated via the tuner 32 and therefore does not require predefined model parameters.

[0140] The boundary state of model Md2a is at least C obs、plant、α0 and T c、obs、plant、α0 Includes.

[0141] Next, β in processing step S10b of FIG. plant、α0 Specific processing of the tuner 32 in processing step S10b will be described with reference to FIG. 3, focusing on only the differences from the embodiment described for the tuner 30.

[0142] First, in processing step S100 of FIG. 3, the tuner 32 searches for the model Md2a (pre-stored in the database DB2) and reads the input data I5 for evaluating the model Md2a.

[0143] Specifically, the input data I5 includes the values ​​of the boundary conditions of the model Md2a. The boundary conditions include at least the C measured at time t obs、plant、α0 and T c、obs、plant、α0 (stored in database DB3). The input data I5 further includes a value for the current simulation time t (e.g., t=0) and a simulation end time tend. The values ​​of time t and trend can be predefined values ​​in database DB2 or user-entered values ​​via input device 3.

[0144] In processing step S101, the tuner 32 detects β plant、α0 Specifically, the estimated value generated in processing step S101 may be a random value or an estimated value based on a user input value via the input device 3. Note that in processing step S101, β plant、α0 is still a figure with little certainty.

[0145] In process step S102, the tuner 32 evaluates an equation including the model Md2a to obtain T sim、plant、α0 This process produces T sim、plant、α0 is generated.

[0146] In processing steps S101 and S102, the boundary condition value read in processing step S100 and the β generated in processing step S101 are compared. lab、α1 In process step S102 from process step S106, the boundary condition updated in process step S106 and the estimated value of β plant、α0The value Md2a is evaluated using the following: Processing step S106 will be described in detail later.

[0147] In processing step S103, the tuner 32 detects the temperature T measured at time t+1 stored in the memory unit D3. obs、plant、α0 Read out.

[0148] In process step S104, the tuner 32 calculates the T sim、plant、α0 and T read out in processing step S103 obs、plant、α0 The tuner 32 then calculates and compares the difference between the calculated T sim、plant、α0 and T obs、plant、α0 Using the difference with plant、α0 Calculate the amount of tuning for T sim、plant、α0 and T obs、plant、α0 The tuned β plant、α0 Once calculated, the process of step S104 ends.

[0149] In processing step S106, the tuner 32 receives the input data I5 and β plant、α0 The update of the input data I5 includes updating the current simulation time t, e.g., t=t+1, and obtaining the boundary conditions at the newly set simulation time t. On the other hand, β plant、α0 is replaced and updated with the tuning value calculated in processing step S104.

[0150] The T calculated in process step S102 using the updated β value from process step S106 sim、plant、α0 is the T calculated in step S102 using the estimated β value from step S101. sim、plant、α0 is expected to be more accurate than β, because the β value was adjusted from the initial guess to the tuning value generated by tuner 30 in process step S104.

[0151] In processing step S107, the tuner 32 calculates the β plant、α0 (processing step S20). This calculation is performed by subtracting the calculated β from the initial simulation time t. plant、α0 For example, the tuner 32 may analyze the calculated T sim、plant、α0 and T obs、plant、α0 If the difference between is less than the threshold v2, plant、α0 This threshold value v2 can be pre-stored by the user or vendor of the present invention.

[0152] The final β generated by the repeated adjustment in the processing step S104 performed by the tuner 32 using the measured values ​​(temperature and concentration) under the condition α0 plant、α0 is expected to accurately represent the behavior of the plant-scale reactor R2 under conditions α. This treatment also allows for the use of β, which is not available in the previous examples. lab、α0 Automate the calculation of [Example]

[0153] The simulated temperature and concentration data generated by scale-up simulator 1 of Figure 1 or scale-up simulator 1a of Figure 9 can be used to replace trials of new state α1 in plant-scale reactor R2. After studying the simulated trials, the plant operator can then proceed with the actual implementation of state α1 in the target plant.

[0154] As the plant continues to operate, material degradation may also occur. For example, thinning, scaling, or rusting of the reactor walls may reduce heat transfer efficiency. Therefore, the actual β plant、α1 is expected to decrease. In such cases, β plant、α1 Monitoring can help indicate the rate of deterioration and help determine future maintenance plans.

[0155] In this Example 4, to assist the plant operator with plant maintenance, β is used when condition α is already implemented in the plant-scale reactor. plant、α1 The description will be provided focusing only on the differences from the previous embodiment.

[0156] FIG. 11 is a diagram showing an example of the hardware configuration of a reactor scale expansion simulator according to a fourth embodiment of the present invention. In this example, when condition α1 is implemented in a plant-scale reactor, β plant、α1 1 shows an exemplary configuration of a scale-up simulator 1b that can additionally adjust the parameter tuner 31 and the parameter tuner 32. In this example, the scale-up simulator 1b will be described, focusing on the differences from the previous examples 1 and 2. Nevertheless, the scale-up simulator 1b can also be applied to example 3 (which has a parameter tuner 31 and a tuner 32).

[0157] A scale expansion simulator 1b according to the fourth embodiment of FIG. 11 further includes a tuner 33, an updater 51, and a database DB3 in addition to the components of the scale expansion simulator 1 of FIG.

[0158] As will be described later in detail, tuner 33 is a β plant、α1 On the other hand, the up-values ​​given by the updater 51 in FIG. 11 are used to generate tuning values ​​of the model Md2 generated by the up-values ​​given by the updater 50 in FIG. 1, adjusted by β plant、α1 Update to model Md2b using

[0159] The scale-up simulator 1b of FIG. 11 can additionally store and retrieve measurement data of the plant-scale reactor R2 via a database DB3. The measurement data includes, under condition α1, at least the time series temperature (T obs、plant、α1 ) and enrichment (C obs、plant、α1 ) and the reactor surface or outer shell (T c、obs、plant、α1 ) and time series temperature.

[0160] 12 is a diagram showing an example of software processing of the reactor scale up simulator according to Example 4 of the present invention. From here, according to FIG. 12, in state α1, β is calculated using the measured values ​​of the plant-scale reactor R2. plant、α1 An example of an additional process implemented by the scale-up simulator 1b to adjust the process conditions is described below. In this example, the process conditions α1 are implemented at a certain time Δt (e.g., one month later) after the process conditions α1 are implemented in the plant-scale reactor R2. Hereinafter, the time when the conditions α1 are implemented in the plant-scale reactor R2 is referred to as t_imp.

[0161] In process step S700 of FIG. 12, the scale-up simulator 1b starts the implementation of the tuner 33. Specifically, the tuner 33 searches the database DB2 for a current simulation model of the plant-scale reactor R2. More specifically, the tuner 33 searches for either the model Md2b generated by the updater 51 or the model Md2 generated by the updater 50, and gives a higher priority to the model Md2b. This means that the model Md2 or the model Md2b is selected only when it is not already available in the database DB2. Process step S700 is completed when the tuner 33 completes the search for mode 2. Then, the process proceeds to S701.

[0162] In processing step S701, the scale-up simulator 1b continues to execute the tuner 33. Specifically, the tuner 33 reads the input I6 from the database DB3. More specifically, the input data I6 includes values ​​of the boundary conditions of the selected model Md2. The boundary conditions include at least the C obs、lab、α1 and T c、obs、lab、α1 When the tuner 33 has finished reading the input data I6, processing step S701 is completed, and processing then proceeds to processing step S702.

[0163] In processing step S702, the tuner 33 sets initial calculation conditions and continues processing. Then, processing proceeds to processing step S703. Specifically, to set the initial calculation conditions, the current simulation time t (ex.t=t_imp) and the simulation end time tend (ex.tend=t_imp+Δt) are set.

[0164] In processing step S703, the tuner 33 calculates T sim、plant、α1 Processing continues by evaluating the equation containing the model Md2 retrieved in process step S700 to generate a predicted value of T sim、plant、α1 Processing is complete when the .times. ...

[0165] In processing step S704, the tuner 33 calculates the temperature T measured at time t+1 stored in the database DB3. obs、plant、α1 Then, the process proceeds to processing step S705.

[0166] In processing step S705, the tuner 33 uses the T calculated in processing step S703. sim、plant、α1 and T read out in processing step S704 obs、plant、α1 Specifically, the tuner 33 continues the process by comparing T sim、plant、α1 and T obs、plant、α1 Then, the tuner 33 uses the calculated difference to calculate β plant、α1 Calculate the amount of tuning. sim、plant、α1 and T obs、plant、α1 The tuned β plant、α1 Once calculated, the process of S703 ends, and the process proceeds to processing step S706.

[0167] In process step S706, the tuner 33 checks whether the end of the calculation has been reached. The tuner 33 can confirm this by, for example, checking whether the simulation time t+1 tends to be equal. If so, processing proceeds to process step S708. If not, processing proceeds to process step S707.

[0168] In processing step S707, the tuner 33 updates the value of the input data I6 and the β plant、α1 The update of input I6 involves updating the current simulation time t, for example t=t+1, and obtaining the boundary conditions at the newly set simulation time t. plant、α1 Temporarily updating the value of the current β lab、α1 with the tuning value calculated in process step S705, and processing then returns to process step S703.

[0169] In processing step S708, the tuner 33 outputs the tuned β plant、α1 This calculation is based on the calculated β from the first simulation time t. plant、α1 For example, the tuner 33 may analyze the calculated T sim、lab、α1 and T obs、lab、α1 If the difference between lab、α1 This threshold value v can be pre-stored by the user or vendor of the invention.

[0170] In process step S709, the scale-up simulator 1b ends the process of the tuner 33 and proceeds to the process of the updater 51. Specifically, the updater 51 retrieves the model Md2 generated by the updater 50 from the database DB2. The updater 51 generates another plant-scale simulation model Md2b using as model equations and specific model parameter values ​​copies of the equations and model parameters from the model Md2. Finally, in process step S708, the updater 51 updates β plant、α1 β was calculated plant、α1 In processing step S708, β plant、α1 was calculated using measurements of the plant-scale reactor R2 under condition α1, so model Md2b is a more updated simulation model than model Md2. This explains the importance of selecting model Md2b with a higher priority in process step S700. The generated model Md2b is also stored in database DB2 (replacing the existing model Md2b).

[0171] Processing step S709 is completed when model Md2b is generated, and the process then ends.

[0172] By performing the process described in FIG. 12 for a certain period Δt, β plant、 α1 The trend can capture the degradation associated with heat transfer efficiency.

[0173] The effect of β per Δt on scheduling plant maintenance. plant、α1 The effect of regular adjustment of β is explained along with Fig. 13. Fig. 13 shows the effect of regular adjustment of β by tuner 33. plant、α1 Using the above, we show an example of the display output generated by the scale-up simulator 1b, with time on the horizontal axis and parameter β on the vertical axis. plant、α1 Specifically, the display output is shown as a function of the condition α1 and the tuned β by the tuner 33 at various points after t_imp. plant、α1β at time t_imp of the first implementation of plant、α1 is shown in the interval Δt.

[0174] Note that β at time t_imp plant、α1 is a value calculated by the scaler 40. Due to degradation such as scaling, β plant、α1 begins to decrease. Finally, at some time t_imp+nΔt, β plant、α1 falls below the alarm. The alarm value can be defined by the user of the invention or the vendor. In this case, the user of the invention can schedule a maintenance period, such as cleaning the reactor at time t_imp+nΔt. Through the maintenance period, the reactor will regain its heat transfer performance similar to that at time t_imp. Without a means to measure the degradation, maintenance periods may be scheduled too frequently or too late. With the invention, the actual degradation rate can be known and the maintenance period can be scheduled just in time.

[0175] In addition, β using a small Δt of Δt=1 second plant、α1 It can also be applied to real-time tuning of [Explanation of symbols]

[0176] 1, 1a, 1b: Scale-up simulator R1: Laboratory-scale reactor 3: Input device 4: Display R2: Plant-scale reactor 30, 31, 32, 33: Tuner 40: Scaler 50, 51: Updater 60: Plant tester DB1, DB2, DB3: Databases

Claims

1. A reactor scale-up simulator that calculates changes in plant characteristics due to changes in condition α, Using first measurements in a laboratory-scale reactor and a first model that simulates the characteristics of the laboratory-scale reactor, the laboratory-scale reactor is subjected to a condition α １ The parameter β of the laboratory-scale reactor when operated at ｌａｂ、α１ a tuner for calculating The parameter β obtained by the tuner ｌａｂ、α１ and the laboratory-scale reactor is subjected to conditions α ０ The parameter β of the laboratory-scale reactor when operated at ｌａｂ、α０ and the plant-scale reactor is under the condition α ０ The parameter β of the plant-scale reactor when operated at plant、α０ Therefore, the plant-scale reactor is set under the condition α １ The parameter β of the plant-scale reactor when operated at plant、α１ A scaler for The aforementioned parameter β ｌａｂ、α１ , β ｌａｂ、α０ , β plant、α０ , β plant、α１ an updater that generates a second model that simulates the characteristics of the plant-scale reactor using a plant tester that calculates a change in plant characteristics of the plant-scale reactor due to a change in condition α using a second model that simulates characteristics of the plant-scale reactor.

2. 10. The reactor scale-up simulator of claim 1, The tuner tunes the characteristics of the first model according to a difference between the first measurement value and an output given by the first model, and calculates a parameter β of the laboratory-scale reactor. ｌａｂ、α１ A reactor scale-up simulator that calculates

3. 10. The reactor scale-up simulator of claim 1, The condition α ０ is the existing condition in the current operation, and the condition α １ is a reactor scale-up simulator characterized by unknown conditions in new operations.

4. 10. The reactor scale-up simulator of claim 1, The scaler satisfies the condition α for the scale ratio between the laboratory-scale reactor and the plant-scale reactor. ０ The gradient of the parameter ratio in the condition α １ Parameter β in ｌａｂ、α１ and the gradient to obtain the parameter β plant、α１ A reactor scale-up simulator that implements a first method for calculating

5. 5. The reactor scale-up simulator of claim 4, The scaler adjusts the parameter difference to the condition α in accordance with the ratio of the parameter difference to the condition α. ０ The parameters in １ Parameter β in plant、α１ The reactor scale-up simulator is characterized by implementing a second method for calculating

6. 6. The reactor scale-up simulator of claim 5, The reactor scale-up simulator is characterized in that the scaler can switch between the first method and the second method.

7. 10. The reactor scale-up simulator of claim 1, The reactor scale-up simulator also Using first measurements in a laboratory-scale reactor and a first model for pretreatment that simulates the characteristics of the laboratory-scale reactor, the laboratory-scale reactor is subjected to a condition α ０ The parameter β of the laboratory-scale reactor when operated at ｌａｂ、α０ a parameter tuner that calculates the time series Using second measurements in the plant-scale reactor and a second pre-treatment model that simulates the characteristics of the plant-scale reactor, the plant-scale reactor is subjected to conditions α ０ The parameter β of the plant-scale reactor when operated at ｌａｂ、α０ and a second tuner that calculates the time series of the reactor scale-up simulator.

8. 10. The reactor scale-up simulator of claim 1, The reactor scale-up simulator also Condition α in a plant-scale reactor １ and a second model simulating the characteristics of the plant-scale reactor, and １ The parameter β of the plant-scale reactor when operated at ｌａｂ、α１ and a third tuner that calculates the time series of The third tuner tunes the characteristics of the second model according to a difference between the second measurement value and an output given by the second model, and calculates a parameter β of the laboratory-scale reactor. ｌａｂ、α１ A reactor scale-up simulator that calculates

9. A reactor scale expansion simulation method for calculating changes in plant characteristics due to changes in condition α using a computer, comprising: The computer uses first measurements of the laboratory-scale reactor and a first model that simulates the characteristics of the laboratory-scale reactor to calculate the laboratory-scale reactor under conditions α １ The parameter β of the laboratory-scale reactor when operated at ｌａｂ、α１ Calculate The parameter β ｌａｂ、α１ and the laboratory-scale reactor is subjected to conditions α ０ The parameter β of the laboratory-scale reactor when operated at ｌａｂ、α０ and the plant-scale reactor is under the condition α ０ The parameter β of the plant-scale reactor when operated at plant、α０ Therefore, the plant-scale reactor is set under the condition α １ The parameter β of the plant-scale reactor when operated at plant、α１ Seeking The aforementioned parameter β ｌａｂ、α１ , β ｌａｂ、α０ , β plant、α０ , β plant、α１ generating a second model simulating the characteristics of the plant-scale reactor using a second model that simulates the characteristics of the plant-scale reactor, and calculates changes in plant characteristics of the plant-scale reactor due to changes in condition α.

10. 10. The reactor scale-up simulation method according to claim 9, comprising: The computer tunes the characteristics of the first model according to the difference between the first measurement value and the output given by the first model, and calculates a parameter β of the laboratory-scale reactor. ｌａｂ、α１ A reactor scale-up simulation method comprising:

11. 10. The reactor scale-up simulation method according to claim 9, comprising: The condition α ０ is the existing condition in the current operation, and the condition α １ is an unknown condition in a new operation.

12. 10. The reactor scale-up simulation method according to claim 9, comprising: The computer calculates the condition α for the scale ratio between the laboratory-scale reactor and the plant-scale reactor. ０ The gradient of the parameter ratio in the condition α １ Parameter β in ｌａｂ、α１ and the gradient to obtain the parameter β plant、α１ A reactor scale-up simulation method, comprising:

13. 13. The reactor scale-up simulation method of claim 12, comprising: The computer calculates the condition α in accordance with the ratio of the difference in the parameter to the difference in the condition α. ０ The parameters in １ Parameter β in plant、α１ The reactor scale-up simulation method is characterized by carrying out a second technique for calculating

14. 14. The reactor scale-up simulation method of claim 13, comprising: A reactor scale expansion simulation method, characterized in that the first method and the second method can be switched and executed.

15. 10. The reactor scale-up simulation method according to claim 0, The computer further Using first measurements in a laboratory-scale reactor and a first model for pretreatment that simulates the characteristics of the laboratory-scale reactor, the laboratory-scale reactor is subjected to a condition α ０ The parameter β of the laboratory-scale reactor when operated at ｌａｂ、α０ is calculated in time series, Using second measurements in the plant-scale reactor and a second pre-treatment model that simulates the characteristics of the plant-scale reactor, the plant-scale reactor is subjected to conditions α ０ The parameter β of the plant-scale reactor when operated at ｌａｂ、α０ A reactor scale expansion simulation method characterized by calculating the above in a time series manner.

16. 10. The reactor scale-up simulation method according to claim 9, comprising: The computer further Condition α in a plant-scale reactor １ and a second model simulating the characteristics of the plant-scale reactor, and １ The parameter β of the plant-scale reactor when operated at ｌａｂ、α１ is calculated in time series, The characteristics of the second model are tuned according to the difference between the second measurement value and the output given by the second model, and a parameter β of the laboratory-scale reactor is obtained. ｌａｂ、α１ A reactor scale-up simulation method comprising:

Images