JPH04236B2 - - Google Patents

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a method and a device for predicting changes in the state inside a boiling water reactor during an accident, and in particular to a method and apparatus for predicting changes in the state inside the reactor during an accident in a boiling water reactor. The present invention relates to a method and apparatus for estimating and predicting changes that will occur in a furnace in the future.

[Background of the invention]

Nuclear power plants have various safety systems in place to avoid accidents. Furthermore, even in the unlikely event that an accident occurs, operational guidance is provided to prevent the accident from spreading to surrounding areas and to safely shut down the furnace. As described above, double and triple safety measures have been taken at nuclear power plants, but the Three Mile Island accident has led to the need for equipment that can predict changes in the reactor due to an accident should an accident occur. Gender is said. this is,
In the event of an accident such as multiple pipe ruptures, the purpose is to estimate the approximate scale of the accident and determine the most appropriate operating procedure. It is necessary to predict behavior quickly.

Various dynamic simulators have been developed to simulate the dynamic behavior of conditions inside a nuclear reactor. However, many of these dynamic simulators require calculation time that is longer than the speed at which the phenomenon progresses, and even some simulators used for operator training are barely able to calculate at the same speed as the actual progress of the phenomenon. , that is, it can only perform real-time calculations. Therefore, it is impossible with this type of simulator to predict the development of phenomena caused by an accident in advance when an accident occurs.

In order to improve this problem, efforts have been made to simplify models that simulate nuclear reactor conditions and to develop simulators that are several times faster than real time. Although this method provides high-speed calculation,
This simplification comes at the expense of computational accuracy during prediction. To predict the development of nuclear reactor accident phenomena,
It can be said that a prediction method that requires prediction over a long period of 2 to 3 hours and does not guarantee sufficient accuracy will not function as a prediction. On the other hand, as a method for improving prediction accuracy, there is a method of adjusting some parameters of a prediction simulator based on actual data, but this method is also useless for the above-mentioned long-term prediction.

[Purpose of the invention]

The purpose of the present invention is to be able to predict the development of phenomena for two to three hours in about 10 to 20 minutes while maintaining sufficient accuracy when an accident occurs in a boiling water reactor. An object of the present invention is to provide a method and device for predicting the state inside a reactor.

[Summary of the invention]

The present invention estimates the scale of an accident in a short time using a sufficiently simplified model based on measurement data of the nuclear reactor plant at the time of the accident, and also estimates the scale of the accident with sufficient accuracy based on the estimated parameters. Initial settings are made with a model that has been simplified to the extent that it can be guaranteed, and based on this it is possible to predict the evolution of long-term phenomena caused by accidents.

[Embodiments of the invention]

Hereinafter, the present invention will be explained based on FIGS. 1 to 8.

First, a system configuration of an example of an in-reactor state prediction device for a boiling water reactor according to the present invention will be explained. FIG. 1 shows the system configuration. Process quantities from the nuclear reactor plant 1 are taken into the input device 2 of this system. This process quantity includes reactor water level, reactor pressure, reactor power,
Signals 8 used in the accident scale estimator, such as recirculation flow rate control signals and control rod drive signals, as well as core flow rate, main steam flow rate, feed water flow rate, containment vessel temperature, containment vessel pressure, recirculation flow rate, and various emergency cooling systems. It consists of a signal 9 such as a flow rate signal. The accident scale estimating device 3 uses the upper air process signal 8 to estimate, for example, the scale of a pipe rupture. As described later, the simulation model used in this device is highly simplified so that the scale of an accident can be estimated in a short time, and the items for estimating the scale of a pipe rupture are not limited to the rupture area and rupture quality. I don't get it. Note that here, the fracture area is the cross-sectional area where the contents leak from the fractured portion, and the fracture quality represents the content of water in the leaked material, and is a numerical value between 0 and 1. In other words, it is 0 if there is no water, and 1 if there is only water. On the other hand, in order to predict long-term phenomena after an accident, the prediction device 5 is equipped with a relatively detailed simulation model that can simulate all of the process quantities described above. This model also simulates various types of piping. Therefore, in order to start prediction with this device, it is necessary to switch from the already obtained fracture area and fracture quality to a fracture area that matches the actual fracture position, and for this purpose, the fracture position estimation device 4 and the database 7 are required.
is available. The predicted values of the state inside the reactor or the plant obtained by the prediction device 5 are displayed in chronological order on a display device 6 such as a CRT to support the judgment of the operator.

Next, each device of the reactor state prediction device according to this system will be explained. FIG. 2 shows the configuration of the fracture size estimating device 3. The scale estimation process signal 8 is input to the state determining section 10. Based on these process signals 8, the status determination unit 10 determines whether changes in the reactor water level, reactor pressure, etc. are due to an accident. As an example of the determination, for example, a change in the furnace output is defined as an event A, and is expressed as A={a change in the furnace output}. Similarly, B = {reactor water level has changed} C = {reactor pressure has changed} D = {recirculation flow rate control signal has changed} E = {control rod drive signal has changed}.

When A∧B∧(C∧(∧)) is true, that is, A∧B∧(C∧(∧))=1...(1), the determination unit 10 is opened and the process signal 8 is transferred to the storage device 11. is input. In the above pool algebraic expression, ∧ indicates AND, and .- indicates NOT.

The storage device 11 stores time-series data of the process signal 8. Additionally, a trigger signal 14 applied by the operator to the system may cause a process signal 8
is output to the current state estimation section 12 and comparison section 15. In the determination section 10, even if the above theoretical determination is negative, the valve is opened approximately once every hour. That is, even when no accident occurs, the determination unit 10 is opened and the process signal 8 is input to the storage device 11. In this case, the determination section 10 closes in about 2 to 3 minutes, and the signals when the logical determination is negative are averaged and stored as one value. This value is similarly input to the current state estimation unit 12 as a steady state signal 16 by a trigger signal.

In the current state estimating unit 12, the initial value of the model is automatically set based on the value of the steady state signal 16.
Along with this, the cooling system flow rate, enthalpy, and recirculation flow rate are taken in from the process signal 8, and the fracture quality and fracture area are set to assumed values, and the reactor water level and reactor pressure are calculated by the model of the estimator 12. do. This calculation is performed for several minutes, and the comparing section 15 compares the actually measured reactor water level and reactor pressure corresponding to the simulation section with the values obtained by the above calculation. Now, the actual measured values of reactor water level and reactor pressure are respectively
Calculated values of x ₁ (t), x ₂ (t), reactor water level and reactor pressure are
If y ₁ (t) and y ₂ (t), then J=∫ ^t ₀ ^+T _t . [α{t ₁ (t)−y ₁ (t)} ² +
The assumed values of the simulated fracture area and fracture quality are sequentially set to give the fracture area and fracture quality that minimize the evaluation quantity β{x ₂ (t)−y ₂ (t)} ² ]dt...(2) The optimal estimating unit 13 has the function of changing. α and β of the above evaluation quantities are appropriate non-negative values, t ₀ is the time when the calculation is started by the trigger signal 14, and T is the time to evaluate the actual measured value and the calculated value. It only takes about a few minutes.

The optimal estimator 13 can use a nonlinear programming method based on the above evaluation amount. The fracture quality and fracture area obtained by the optimal estimation unit 13 are replaced with those values previously used by the current estimation unit 12, other input data are set to the same values as last time, and the reactor water level and reactor are calculated again in the same time interval. The change in pressure is determined, and the above evaluation J is determined again. If the evaluation value J is larger than a predetermined sufficiently small value ε, the wave break quality and fracture area are estimated again by the optimum estimator 13, and the simulation is similarly repeated by the current estimator 12.

When such a procedure is repeated and the value of the evaluation quantity J becomes smaller than the predetermined value ε, the fracture quality and fracture area at this time represent the scale of the accident in this case.

Now, the types of process quantities of the simplified model used in the current state estimating section 12 will be briefly described. The model of the current state estimating unit 12 may use other process variables, such as reactor pressure, reactor water level, feed water flow rate, emergency cooling system flow rate, and enthalpy. However, here, it is modeled as shown in FIG. In other words, the inside of the reactor pressure vessel is filled with saturated water 17
The pressure is assumed to be uniform within the pressure vessel. In addition, the heating element 19 is used to simulate heat generation in the reactor core.
I'm thinking. The inlet/outlet of the cooling water to the pressure vessel is
As shown in the figure, it consists of a saturated steam outlet channel 20, a saturated water outlet channel 21, and a cooling water inlet channel 22. Modeling of such a system takes into consideration the mass balance and energy balance regarding water and steam.

An example of the results obtained based on this model is shown in FIG.

Suppose that a recirculation system pipe rupture occurs with a rupture area of 0.1 ft ² (square feet) and a rupture quality of 0.0. Although such an accident is a hypothetical accident and there is no actual measurement data, an analysis example using a detailed accident analysis program is shown by the solid line in Fig. 4. The two solid lines indicate the reactor water level and reactor pressure, respectively. Now t=
The results simulated according to the above method using data for about 2 minutes (T=125 seconds) from time t ₀ are shown by broken lines. This result agrees well with the analysis example. In this case, the fracture area is 0.107ft ² and the fracture quality is 0.079, respectively.
It can be seen that 0.1ft ² and fracture quality of 0.0 are well estimated.

As described above, it is clear that a nuclear reactor model simplified by the model shown in FIG. 3 can function satisfactorily as long as the fracture area and fracture quality are estimated.

However, the model shown in Figure 3 is extremely simplified compared to an actual nuclear reactor, and it is impossible to predict subsequent long-term phenomena based on such a model. Therefore, the present invention includes a dynamic model that can simulate the dynamic behavior of the internal state of the nuclear reactor with sufficient accuracy.

The above-mentioned fracture size estimating device estimates the fracture area and fracture quality of a pipe rupture accident based on actual measurement data. From this value, it is necessary to determine the position of the pipe in order to set input values for a dynamic model that can simulate various pipe breaks. The various types of piping include main steam flow path piping, recirculation flow path piping, water supply flow path piping, and water level instrumentation piping. The relationship between the fracture quality and these pipes can be uniquely determined by the reactor water level and reactor pressure.

For example, when the reactor pressure and reactor water level are maintained sufficiently, the fluid in the recirculation channel piping is a single-phase flow, and in this case, the fracture quality is zero. However, the breakage quality of water supply piping and instrumentation piping may also be zero. In this case, the furnace pressure decreases from 50 to
By waiting for about 100 seconds, the quality at the upper position of the water supply piping and instrumentation piping in the furnace will have a certain value (0<x1). In this case too, the flow in the recirculation reflux line is single-phase. In this way, the fracture position can be uniquely determined from the reactor water level, reactor pressure, and fracture quality.

The database 7 shown in FIG. 1 is a file containing data on the reactor water level, reactor pressure, fracture quality, and corresponding fracture position.
Based on this information, the fracture position estimating device 4 determines the fracture position.

The trigger signal 14 for estimating the rupture size described in FIG. 2 is used to set the estimation start time as described above, and this can be input from the keyboard provided with the CRT. This trigger signal 14 can, for example, take in an alarm signal of the occurrence of an accident, operate the fracture position estimation device at the same time, and operate the same estimation device again after a certain period of time, for example 50 seconds, to automate the setting. be.

Next, a prediction device 5 for predicting the development of phenomena over a long period of time in the event of such an accident will be explained with reference to FIG.

Process signals 8 and 9 are input to the prediction device 5 at regular intervals regardless of whether an accident occurs or not. The input section 23 is for this purpose, and in this case each signal is subjected to an averaging operation. At the same time, an alarm signal indicating the occurrence of an accident is superimposed on the process signals 8 and 9, and when this signal is activated, the averaging operation is stopped. After several tens of seconds at the latest, the values of the rupture position and the rupture area are input via signal 26. In the model initial value setting section 24, initial values are set based on the steady state process signal and the values of the fracture position and area. As will be described later with reference to FIG.
The model No. 5 is more complex than the dynamic model for estimating the rupture size described above, and requires a system of algebraic equations to perform initialization settings for unmeasured physical quantities. .

Once the initial values are set, the predictive simulation unit 25 sequentially predicts changes in phenomena after the accident. At this time, the prediction start command and prediction time can be entered by the operator using the keyboard, but normally the operation is performed using standard settings.

FIG. 6 shows the scale of the nuclear reactor model used in the prediction simulation section 25. The inside of the reactor is divided into five regions: A unboiled region in the shroud B saturated water mixing region in the shroud C unboiled region outside the shroud D saturated water region outside the shroud E steam region in the dome Define dynamic equations for energy and mass balance. Regarding the heat generating part within the core 27, heat transfer from the fuel rods 27A is taken into consideration. Various piping, namely recirculation flow path piping 28, water supply piping 29, instrumentation piping 30, main steam flow path piping 3
1 are simulated, respectively, as well as isolation valve 32 and relief valve 33. In addition, the cooling system was also simulated.

Figures 7 and 8 show the results of simulating 1000 seconds after a recirculation channel rupture accident with a rupture area of 0.1 ft ² using such a model. The results of this prediction simulation are shown as broken lines, and the results of the detailed calculations described above are shown as solid lines. Predicted using this method, 1000
Even after about a second has elapsed, there is no difference between the reactor water level and reactor pressure compared to the detailed calculations (actually measured values), indicating that accuracy is fully guaranteed. Note that the calculation time required for this calculation is approximately 1/10 of the real time, indicating that it is sufficiently fast.

〔Effect of the invention〕

According to the present invention, when predicting changes in the state inside a nuclear reactor after an accident, the scale of the accident is estimated in a short time using a simple dynamic model, and this estimation result and a relatively large-scale model are used to estimate the scale of the accident in a short time. In order to predict what will happen after that, it is possible to make highly accurate predictions over a long period of 2 to 3 hours, and it is also possible to make predictions about 10 times faster than in real time.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing the configuration of the prediction device of the present invention for predicting the internal state of a nuclear reactor at the time of an accident, Fig. 2 is a block diagram of a rupture scale estimating device, and Fig. 3 is a simplified diagram for estimating the rupture scale. Fig. 4 is a diagram showing the rupture scale estimation results, Fig. 5 is a block diagram of the prediction device, and Fig. 6 is a more detailed reactor model diagram used for relatively long-term prediction after estimation. , 7th
The figure shows the predicted results of the reactor water level, and FIG. 8 shows the predicted results of the reactor pressure. 1... Nuclear reactor, 2... Input device, 3... Accident scale estimation device, 4... Fracture position estimation device, 5... Prediction device, 6... Display device, 7... Database,
8, 9...process signal, 10...state determination section,
11...Storage device, 12...Current status estimation unit, 13...
...Optimum estimation section, 14...Trigger signal, 15...Comparison section, 16...Steady state signal, 17...Saturated water,
18... Saturated steam, 19... Heating element, 20... Saturated steam outlet channel, 21... Saturated water outlet channel, 22
... Cooling water inlet channel, 23 ... Input section, 24 ...
Model initial value setting section, 25... Prediction simulation section, 26... Fracture position and area signal, 27...
...Reactor core, 27A...Fuel rod, 28...Recirculation flow path piping, 29...Water supply piping, 30...Instrumentation piping, 3
1... Main steam flow path piping, 32... Isolation valve, 33...
...Relief valve.

Claims (1)

[Claims] 1. A method for predicting the internal state of a nuclear reactor at the time of an accident over a long period of time, which uses a simple dynamic model based on a relatively small number of parameters among the process signals of the reactor to predict the scale of the accident and the accident. A high-speed estimation stage in which the estimated position is estimated and compared with actual measured data in a short period of time, and the estimated value is updated so that the two agree to match the actual measured data, and the estimated value and estimated position that match the actual measured data are A method for predicting the internal state of a nuclear reactor at the time of an accident, the method comprising the step of predicting the internal state of the reactor over a long period of time by applying the process signal to a dynamic model of the nuclear reactor that is relatively less simplified. 2. In a device that predicts the internal state of a nuclear reactor over a long period of time during an accident, a simple dynamic model is used to predict the occurrence of an accident based on trends in a relatively small number of parameters such as reactor pressure and water level among the reactor process signals. We have an accident scale estimating device that estimates the scale and compares it with actual measured data in a short period of time, updating the estimated value so that both agree and match the actual measured data, and the estimated value of the accident scale estimating device. An accident location estimating device that estimates the location where an accident occurred from predicted data, an accident scale estimating device, an estimated value of the accident scale and accident location estimated by the accident location estimating device, and the process signal are combined into an atom with relatively low simplification. A system for predicting the internal state of a nuclear reactor at the time of an accident, comprising a predictive system that simulates the internal state of the reactor over a long period of time by applying a dynamic model of the reactor, and a system that displays the prediction results.