PLC-based chemical leakage monitoring and integrated control system for semiconductor processing

Abstract

The present invention relates to a system that monitors in real time whether leakage may occur during the operation of chemicals within a semiconductor manufacturing facility (FAB) and prevents accidents by immediately cutting off the supply in the event of an abnormality. The effects of the present invention are as follows. By monitoring the overall chemical operation status in the office, management efficiency is maximized, precise pressure is maintained through regulators, immediate response is possible when the standard pressure is exceeded through air checks, and intrinsically safe signals can be reliably detected even in organic explosion-proof areas by applying barriers.

Description

Technology Field

[0001] The present invention relates to a system that monitors in real time whether leakage may occur during the operation of chemicals within a semiconductor manufacturing facility (FAB) and prevents accidents by immediately cutting off the supply in the event of an abnormality. Background Technology

[0002] Semiconductor manufacturing processes require the essential use of various high-purity chemicals, which are supplied to process equipment via Chemical Insulated Boxes (CIBs) and Valve Manifold Boxes (VMBs) within the Fabrication Assembly (FAB). While the stable supply of chemicals is directly linked to process efficiency, leaks can lead not only to equipment damage but also to casualties and environmental pollution, making real-time monitoring systems of paramount importance.

[0003] Conventional management methods often relied on simply checking for leaks or abnormal conditions in the supply line on-site, or on individual alarm devices. However, due to the increasingly complex nature of semiconductor processes, there is a growing demand for integrated management of multifaceted process data—such as pressure at the supply end, exhaust airflow velocity, and waste liquid temperature—going beyond simple leak detection. Furthermore, there was a technical limitation in that it was difficult to prevent the spread of damage if immediate physical containment measures were not taken upon the occurrence of a leak.

[0004] Furthermore, checking various alarms and signals occurring across extensive areas within the FAB solely on-site hinders rapid initial response. Consequently, there has been a continuous need for an integrated management system that can monitor the overall chemical operation status within the semiconductor plant in real time from remote locations, such as offices, thereby ensuring ease of maintenance and minimizing accident response time.

[0005] Against this backdrop, there is a demand for a technical alternative that collects various on-site sensor data through the logic control of a PLC (Programmable Logic Controller) and, based on this, drives the pneumatic unit to physically and immediately cut off the supply. In particular, there is an urgent need to introduce an integrated control system that can maximize the overall safety of chemical operations by linking valve control via a pneumatic system with office monitoring, while ensuring the intrinsic safety of the organic explosion-proof zone. Prior art literature

[0006] Korean Registered Patent Publication No. 10-1926368 (December 3, 2018) The problem to be solved

[0007] The present invention aims to facilitate maintenance by constantly monitoring chemical leaks, wire breakage, and hazardous gases within the FAB, prevent the spread of damage by automatically shutting off valves of supply equipment (CIB, VMB) in the event of a leak, and integrate the management of chemical supply pressure, usage request signals, exhaust status, and waste liquid temperature within a single system. means of solving the problem

[0008] To solve the above-mentioned problem, the present invention is a system that monitors in real time whether a leak may occur during the operation of chemicals within a semiconductor manufacturing facility (FAB) and prevents accidents by immediately cutting off the supply in the event of an abnormality. The system comprises a leak panel control unit (100) that analyzes signals received from on-site sensors and issues a control command if they deviate from set safety standards, a leak panel pneumatic unit (200) that converts electrical control signals received from the leak panel control unit (100) into physical pneumatic signals to drive an actual valve, and an external interface (300) that collects on-site status signals, transmits them to the control unit, and executes a physical cutoff command based on the judgment of the control unit. The leak panel control unit (100) includes a PLC (110) for monitoring the internal condition of the FAB in real time using I / O contacts and issuing control commands including solenoid valve closure in the event of an abnormality, a leakage circuit breaker (120) for detecting a leakage of the main AC power supply flowing into the system and cutting off the power, and a leak sensor, a pressure sensor (PT), and a wind speed / differential pressure gauge. The leak panel pneumatic section (200) includes a circuit protector (130) for protecting the circuit by blocking overcurrent of the individual purpose-specific configuration circuit, a DC power supply (140) for supplying stable DC power to various instruments installed inside the semiconductor production facility for measuring chemical substances, a barrier (150) for safely detecting the signal of a leak sensor installed in an organic explosion-proof area and transmitting it to the PLC, and an IO terminal block (160) that performs the role of an interface for interconnecting individual contacts of the PLC and cables coming up from the field, and the leak panel pneumatic section (200) includes a ball valve (210) for manually controlling the supply and cutoff of main CDA (Clean Dry Air) supplied to the entire system, and a regulator (220) for adjusting the air pressure to a preset constant pressure for automatic valve operation.A PLC-based semiconductor process apparatus is characterized by comprising: a solenoid valve (230) for driving an automatic valve of supply equipment within a FAB by receiving an electrical signal from a PLC (110) and turning actual air on / off; an air check unit (240) for detecting when the pneumatic supply pressure exceeds a reference value and generating an alarm through the PLC (110); and a finger valve (250) for individually turning on / off the pneumatic supply to the solenoid valves installed for each manifold; wherein the external interface (300) includes input sensors that transmit physical status signals of the site, such as a leak sensor, a terminal pressure transmitter, a wind speed / differential pressure gauge, a gas detector, and a waste liquid temperature sensor; supply and shut-off equipment including a CIB (Chemical Insulated Box), VMB (Valve Manifold Box), and an automatic valve that perform physical operations by receiving a control signal from the leak panel pneumatic unit (200); and a remote monitoring center that visualizes the final information collected and determined by the leak panel and displays it to an administrator. It provides a chemical leak monitoring and integrated control system. Effects of the invention

[0009] The effects of the present invention are as follows.

[0010] By monitoring the overall chemical operation status in the office, management efficiency is maximized, precise pressure is maintained through regulators, immediate response is possible when the standard pressure is exceeded through air checks, and intrinsically safe signals can be reliably detected even in organic explosion-proof areas by applying barriers. Brief explanation of the drawing

[0011] FIG. 1 is a schematic diagram showing the overall configuration of the present invention. FIG. 2 is a photograph according to an embodiment of the leak panel control unit (100) of the present invention. FIG. 3 is a photograph according to one embodiment of the leak panel pneumatic part (200). Figure 4 is a block diagram visualizing the overall signal flow and physical control mechanism of a PLC-based integrated leak monitoring and safety control system. FIG. 5 is a diagram showing the operation sequence according to an embodiment of the present invention. Specific details for implementing the invention

[0012] Throughout the entire specification below, the same reference numerals refer to the same components unless there are special circumstances. Terms with the addition of "part" used below may be implemented in software or hardware, and depending on the embodiment, it is possible for a single "part" to be implemented as a single physical or logical component, for multiple "parts" to be implemented as a single physical or logical component, or for a single "part" to be implemented as multiple physical or logical components. Throughout the specification, when it is stated that a part is connected to another part, this may mean a physical connection between the part and the other part, or an electrical connection. Furthermore, when it is stated that a part includes another part, this does not mean that another part other than the other part is excluded unless specifically stated otherwise, but means that additional parts may be included at the designer's choice. Terms such as "first" or "second" are used to distinguish one part from another part, and unless specifically stated otherwise, they do not imply sequential expressions. Also, singular expressions may include plural expressions unless there is an obvious exception in the context.

[0014] The present invention relates to a system that monitors in real time whether leakage may occur during the operation of chemicals within a semiconductor manufacturing facility (FAB) and prevents accidents by immediately cutting off the supply in the event of an abnormality.

[0016] Referring to FIG. 1, the present invention comprises a leak panel control unit (100), a leak panel pneumatic unit (200), and an external interface (300).

[0018] Each component is explained in detail as follows.

[0020] The leak panel control unit (100) is intended to perform a pivotal role in integrating and collecting all alarms and signals generated during chemical operation within the semiconductor factory, making logical judgments, and controlling the system. It analyzes signals received from sensors at the site and issues control commands when they deviate from set safety standards.

[0021] The leak panel control unit (100) described above performs integrated safety management functions such as determining and monitoring for leakage, real-time monitoring, and controlling the suspension of chemical supply (closing CIB, VMB valves), and is positioned in the upper section of the leak panel structure to facilitate the manager's line of sight and access. In addition, it analyzes digital / analog signals received from various sensors at the site and simultaneously generates a pneumatic unit drive signal and an office monitoring alarm if the set safety standards are exceeded.

[0022] FIG. 2 is a photograph according to an embodiment of the leak panel control unit (100) of the present invention. Referring to the figure, the leak panel control unit (100) is configured to include a PLC (110), a leakage circuit breaker (120), a circuit protector (130), a DC power supply (140), a barrier (150), and an IO terminal block (160).

[0024] The PLC (110) is the core control device of the system and is intended to monitor the internal conditions of the FAB in real time through I / O contacts and to transmit control commands to the pneumatic unit (200). In other words, it performs control functions such as logic, sequence, and timing through a software program instead of a hardware relay circuit.

[0025] The above PLC (110) monitors the internal conditions of the FAB in real time using I / O contacts and issues control commands, such as closing solenoid valves, when an abnormality occurs. It is also mounted on a base rack in the central part of the control unit and connected to various input / output modules. It constantly scans sensor signals through output modules and activates output contacts according to logic defined in the program (e.g., Leak=ON AND Request=ON) to perform an immediate shutdown process.

[0027] The leakage circuit breaker (120) is intended to protect equipment by detecting a leakage current in the main AC power supply flowing into the system and cutting off the power. That is, it is intended to prevent fire and electric shock accidents by detecting the ground fault current when a leakage current occurs in the circuit and automatically cutting off the power, thereby monitoring and protecting the main AC power supply flowing into the entire system.

[0028] The above leakage circuit breaker (120) monitors the balance between the incoming and outgoing currents, and if an imbalance (leakage) is detected, it cuts off the magnetic circuit to immediately cut off the power flowing to the system.

[0030] The circuit protector (130) is intended to protect the circuit by blocking overcurrent in each individual purpose-specific circuit, such as a leak sensor, a pressure sensor (PT), and a wind speed / differential pressure gauge.

[0031] That is, the circuit protector (130) is a small circuit breaker installed to protect individual electrical devices and wires from overload or short-circuit accidents, and has the function of independently protecting the circuits of each measuring instrument, such as a leak sensor, PT sensor, wind speed / differential pressure gauge, to prevent an abnormality in a specific circuit from spreading to the entire system.

[0032] The above circuit protector (130) is preferably installed so as to be arranged in parallel near the power distribution block, and is intended to prevent damage to the equipment by individually turning off only the switch of the corresponding circuit when an overcurrent exceeding the rated current flows in a specific sensor circuit.

[0034] The DC power supply (140) is installed inside the FAB to supply stable DC power to various instruments for measuring chemicals.

[0035] That is, the DC power supply (140) is a device that converts AC 220V commercial power input from the outside into a stable DC power (mainly 24V) suitable for driving electronic instruments, and supplies the power required for various precision chemical instruments and sensors used inside the FAB.

[0036] The above DC power supply (140) is positioned in the power conversion section within the control unit to facilitate heat dissipation, generates clean DC power with low voltage fluctuation through rectification and smoothing circuits, and includes a protection function that limits the output in the event of a short circuit of the load.

[0038] The barrier (150) safely detects the signal from the leak sensor installed in the organic explosion-proof (intrinsically safe) zone and transmits it to the PLC.

[0039] That is, the barrier (150) is an intrinsically safe explosion-proof device that prevents explosions by limiting electrical energy transmitted to a dangerous area (explosion-proof zone) to a safe level and acts as a bridge to safely transmit the minute signal from a leak sensor installed in the organic explosion-proof zone to the PLC.

[0040] The above barrier (150) is preferably located near the input / output terminal of the control unit into which the external sensor cable is introduced, and when a field sensor signal is introduced, it suppresses the voltage and current to a certain level or lower to completely block the possibility of sparks occurring in the dangerous area, and then transmits only a safe signal to the PLC input contact.

[0042] The IO terminal block (160) serves as an interface that interconnects the individual contacts of the PLC with cables coming up from the field (Local).

[0043] The above IO terminal block (160) is a connection device designed to easily and systematically connect internal control wiring and external cables coming up from the field, and provides an interconnection interface between individual contacts of the PLC and field cables, and can serve as an inspection point during maintenance.

[0044] The above IO terminal block (160) is arranged in large quantities on the bottom or side of the control panel to support external wiring connections, and although it is a simple electrical medium, it serves as a channel to transmit sensor signals from the field to the PLC or distribute control signals from the PLC to field drivers through paths designated by number.

[0046] The leak panel pneumatic unit (200) is intended to convert an electrical control signal received from the leak panel control unit (100) into a physical pneumatic signal to drive the actual valve.

[0047] The above leak panel pneumatic unit (200) controls the clean dry air (CDA) supplied from the outside to an appropriate pressure and distributes it to each piece of equipment through individual solenoid valves to realize remote control.

[0048] Additionally, the leak panel pneumatic section (200) is located in the lower part of the leak panel structure or in a separate pneumatic control section, and it is preferable that various pneumatic pipes and manifolds are concentrated therein.

[0049] As an operating algorithm, air introduced from the main air supply is stabilized after passing through a regulator, and then transmitted to or blocked from the final drive unit (CIB, VMB, etc.) through a solenoid valve that operates according to an electrical signal from the PLC.

[0051] FIG. 3 is a photograph according to an embodiment of a leak panel pneumatic unit (200). Referring to the figure, the leak panel pneumatic unit (200) is configured to include a ball valve (210), a regulator (220), a solenoid valve (230), an air check unit (240), and a finger valve (250).

[0053] The ball valve (210) is for manually controlling the supply and cutoff of the main CDA (Clean Dry Air) supplied to the entire system.

[0054] That is, the ball valve (210) is a manual valve that opens and closes the flow of fluid using a rotating ball, and serves to manually completely block or open the supply of main CDA (Clean Dry Air) supplied to the entire system.

[0055] The ball valve (210) is preferably placed at the forefront of the pneumatic system in the main air inlet line and serves to immediately connect or disconnect the air passage by rotating the sphere inside the valve through the operation of a handle by the manager.

[0057] The regulator (220) provides a constant pressure (e.g., 6 kgf / cm²) suitable for driving the automatic valve. 2 It is intended to precisely control air pressure using ).

[0058] That is, the regulator (220) is designed to maintain a constant output pressure regardless of fluctuations in the incoming air pressure and serves to precisely reduce and regulate the air pressure to a constant pressure (e.g., 6 kgf / cm^2) suitable for automatic valve operation.

[0059] The regulator (220) is preferably installed in the pneumatic line between the ball valve and the solenoid valve, and finely adjusts the air flow rate by utilizing the balance between the set spring tension and the output pressure, thereby lowering the output pressure to a reference value of 6 kgf / cm² 2 Maintain as.

[0061] The solenoid valve (230) is intended to drive the automatic valve of the supply equipment in the FAB by receiving an electrical signal from the PLC (110) and turning the actual air on / off.

[0062] That is, the solenoid valve (230) is an electronic valve that controls the opening and closing of the valve using magnetic force generated by an electrical signal, and by receiving an electrical signal from the PLC (110) and turning the actual air on / off, it effectively drives the automatic valve of the supply device (CIB, VMB) inside the FAB.

[0063] It is preferable that a number of the above solenoid valves (230) be arranged in alignment on the manifold within the pneumatic section, and when current is applied from the PLC, magnetic force is generated in the coil to move the plunger, thereby opening or closing the blocked air passage to operate the valve of the field equipment.

[0065] The air check unit (240) is intended to detect when the pneumatic supply pressure exceeds a reference value and to generate an alarm through the PLC (110).

[0066] That is, the air check unit (240) is designed to detect the presence or absence of air pressure or the pressure status within the pipe and generate an electric signal. When the supply pressure is above a set reference pressure, it turns on, allowing alarms and monitoring to be performed through the PLC.

[0067] The above air check unit (240) is preferably installed on the main pneumatic supply line after the regulator, and when the internal diaphragm detects the air pressure and exceeds a set threshold, it connects the electrical contact, and this signal is displayed as an alarm on the office's integrated monitoring system via the PLC.

[0069] The finger valve (250) is for individually turning on / off the pneumatic supply to the solenoid valve installed for each manifold.

[0070] That is, the finger valve (250) is a small manual On / Off valve that can be easily operated with a finger using a small lever, and allows the pneumatic pressure supplied to the solenoid valve installed for each manifold to be individually manually cut off or supplied.

[0071] The finger valve (250) is preferably located upstream of the solenoid valve in each branch line of the manifold, and when inspection or replacement of a specific solenoid valve is required, only the finger valve of the corresponding line is manually operated to cut off the air supply, thereby enabling individual maintenance without interrupting the entire system.

[0073] The external interface (300) refers to any external environment element that the leak panel manages and communicates with, and which the leak panel system exchanges data with or performs physical control to ensure safety within the semiconductor process (FAB).

[0074] The above external interface (300) collects status signals from the site and transmits them to the control unit, executes physical blocking commands based on the judgment of the control unit, and serves as a channel for reporting the overall situation to a remote location.

[0075] Additionally, the external interface (300) is preferably distributed extensively throughout the entire chemical supply path inside the semiconductor factory (FAB) and the management office (Office), and operates by receiving sensor data from the field and, when the leak panel performs calculations, driving the valve of the supply equipment or updating information on the screen of the office monitoring system according to the result.

[0077] The external interface (300) may be configured to include input sensors, supply and cutoff equipment, and a remote monitoring center.

[0078] Input sensors are devices that transmit physical status signals from the site to the control unit (100) of the leak panel. Specifically, input sensors include a leak sensor, a terminal pressure transmitter, a wind speed / differential pressure gauge, a gas detector, and a waste liquid temperature sensor.

[0079] Leak sensors detect the occurrence of leaks during chemical operations within semiconductor factories and send a signal. Terminal pressure transmitters measure and transmit the terminal pressure of chemical supply lines in real time, while wind speed / differential pressure gauges generate signals to monitor the exhaust operation status of supply equipment (CIB, VMB).

[0080] In addition, the gas detector senses hazardous gases near the organic supply equipment, and the waste liquid temperature sensor measures the temperature of the drained waste liquid to monitor whether it exceeds the standard limit.

[0082] The supply and cutoff equipment is a target that receives a control signal from the pneumatic part (200) of the leak panel and performs a physical operation.

[0083] The above supply and shut-off equipment includes a CIB (Chemical Insulated Box), a VMB (Valve Manifold Box), and automatic valves.

[0084] The CIB (Chemical Insulated Box) and VMB (Valve Manifold Box) can be core chemical supply equipment within the FAB, and the supply can be immediately stopped via an automatic valve in the event of a leak by a pneumatic signal sent by the solenoid valve on the leak panel.

[0086] The remote monitoring center is the destination that visualizes the final information collected and assessed by the leak panel and displays it to the manager.

[0087] The aforementioned remote monitoring center monitors overall chemical-related alarms and signals occurring inside the semiconductor factory in real time from the office, and comprehensively manages equipment usage, pressure levels, exhaust conditions, etc.

[0089] The external interface (300) operates through the following organic scenario.

[0090] 1. Real-time status synchronization and monitoring

[0091] The PLC (110) collects all data from the site in real time, such as monitoring end pressure, detecting whether exhaust operation is being performed, detecting hazardous gas fumes, and monitoring waste liquid drain temperature. This data is transmitted to a remote monitoring center so that real-time monitoring is performed in the office.

[0093] 2. Anomaly detection and automatic blocking

[0094] When a chemical leak or gas leak is detected, the PLC (110) immediately controls the solenoid valve (230). Accordingly, the valve of the FAB internal supply equipment (CIB, VMB) is shut off to automatically stop the supply of chemicals on-site. In addition, if the drain temperature exceeds a threshold, the drain is shut off to prevent secondary accidents.

[0096] 3. Remote Alarm and Manual Intervention

[0097] Simultaneously with the automatic shutdown at the site, an emergency alarm is triggered at the office monitoring center. Administrators identify the source of the anomaly through the transmitted signal, and the system is provided with an integrated control environment that allows for forced shutdown via the remote monitoring center if necessary, or for safely resetting and restarting the system after maintenance is completed.

[0099] Figure 4 is a block diagram visualizing the overall signal flow and physical control mechanism of a PLC-based integrated leak monitoring and safety control system. The integrated control system according to the present invention has a double-loop structure that detects abnormal signs at the site and reports them remotely, while simultaneously performing physical shut-off control.

[0101] To explain this in detail, it is as follows.

[0103] 1. Detection and Input Stage: FAB Field Sensors → Leak Panel Control Unit (PLC)

[0104] Sensors such as leak sensors, gas detectors, and pressure transmitters placed in the chemical supply line and surrounding environment within the FAB detect physical changes. The detected information is converted into an electrical signal (Analog / Digital) and input to the leak panel control unit (100). At this time, the signal from the organic explosion-proof zone is safely transmitted through the barrier (150), and all signals are integrated into the PLC (110) via the IO terminal block (160).

[0106] 2. Judgment and Calculation Phase: Leak Panel Control Unit (PLC)

[0107] The PLC (110) compares the input sensor data with the programmed safety standard in real time.

[0108] If an abnormal situation (liquid leakage, gas leakage, pressure anomaly, etc.) is detected, the PLC simultaneously outputs a 'data signal' for monitoring and an 'electrical command' for actual control.

[0110] 3. Information Sharing Phase: Leak Panel Control Unit (PLC) → Office Monitoring

[0111] Status data processed by the PLC is transmitted to the remote monitoring system (300) in the office via network communication.

[0112] Accordingly, managers can identify the leak location, current pressure, exhaust status, etc. in real time without going to the site and direct a rapid initial response.

[0114] 4. Physical Control Step: Leak Panel Control Unit (PLC) → Leak Panel Pneumatic Unit → Equipment Shutdown

[0115] The PLC sends an electrical drive command to the solenoid valve (230) of the pneumatic unit (200). Upon receiving the command, the solenoid valve operates and sends air of a constant pressure, regulated through the regulator (220), to the supply equipment at the site.

[0116] The transmitted pneumatic signal physically closes the automatic valves of the CIB or VMB, which are supply equipment inside the FAB, to immediately stop the supply of chemicals.

[0118] The external interface (300) further includes a digital twin simulator module (310), and the digital twin simulator module (310) is built inside the remote monitoring center of the external interface (300) and is a software architecture that replicates all physical components and process states of the leak panel control unit (100), leak panel pneumatic unit (200), and external interface (300) identically in a virtual space. The digital twin simulator module (310) receives sensor data (pressure, wind speed, temperature, etc.) collected from the PLC (110) in real time and updates the virtual process model.

[0119] The digital twin simulator module (310) visualizes the automatic valve opening and closing status of the chemical supply line and CIB / VMB in 3D, allowing a manager to intuitively monitor the field situation from the office, and derives safety thresholds for the system by performing risk scenarios such as "valve shut-off delay" or "pressure surge" in a virtual environment before an actual accident occurs. In addition, it detects signs of sensor failure in advance by analyzing minute errors or noise data from the sensor, and trains an AI model through virtual scenarios even when actual accident data is insufficient.

[0121] The digital twin module determines the shut-off priority by calculating the 'potential leakage spread' in the event of an accident based on real-time incoming supply end pressure and fluid characteristics.

[0123] To quantitatively determine the potential leakage spread, the system collects and defines the following five key factors in real time.

[0125] Leakage duration (t): This is the time interval from the point in time when the leak is first detected (t_0) until the point in time when the solenoid valve (230) and the automatic valve are completely closed (t_close). The closing completion point (t_close) is determined by summing the calculation speed of the PLC (110) and the mechanical response delay time of the pneumatic unit (200), and a reference value is set based on actual data during the initial operation of the system.

[0127] Differential Pressure (ΔP): This is the difference between the pressure inside the chemical supply line and the external atmospheric pressure. The internal line pressure is received in real time from a pressure transmitter at the supply end, while the external atmospheric pressure is set as a fixed constant or acquired through a separate environmental sensor. As the differential pressure increases, the displacement force of the fluid becomes stronger, and thus the leakage rate increases proportionally.

[0129] Discharge coefficient (C_d) and cross-sectional area (A): The discharge coefficient is a constant that corrects friction loss due to pipe roughness or the shape of the leak hole, and the cross-sectional area is the size of the expected leak hole. The digital twin simulator module (310) assumes a worst-case scenario that may occur by considering the pipe diameter and aging degree, sets this as a primary reference value, and then adjusts it through virtual HAZOP analysis.

[0131] Specifically, the digital twin simulator module (310) sets an initial first reference value based on the physical limit of the hardware, and the process is as follows.

[0133] Based on the diameter (D) of the main piping connected to the chemical supply equipment (CIB, VMB), the maximum value of the cross-sectional area (A) is calculated by assuming a 'Full Guillotine Break,' which is the largest physical failure type that can occur. The aging coefficient (K) is calculated by combining the installation years and the corrosiveness data of the chemicals.

[0134] The specific calculation method and judgment criteria for the aging coefficient (K) applied to predict changes in the physical state of the piping in the digital twin simulator module (310) are as follows.

[0135] The aging coefficient (K) according to the present invention is an indicator calculated by quantitatively combining the temporal deterioration state of the piping and the chemical reactivity of the internal fluid, and largely utilizes an installation years weighting factor and a chemical corrosion index as basic factors. The installation years weighting factor is a numerical representation of the elapsed time from the initial operation of the piping to the present in years, and the chemical corrosion index is a constant value pre-stored in an internal system database, which grades the rate of thickness reduction and surface roughness change of the piping material according to the type and concentration of the supplied chemical substance.

[0136] The specific procedure for calculating the above aging coefficient (K) is as follows. First, the initial state of the newly installed piping is set to a reference value of 1.0. Then, the basic deterioration value over time is derived by multiplying the above installation years weighting factor by the inherent deterioration rate of the pipe material, and the final aging coefficient (K) is determined by assigning a weighting factor by multiplying this by the corrosiveness index of the chemical substance actually flowing.

[0137] That is, it has a logical structure of [aging coefficient (K) = 1 + (years of installation × corrosiveness index)], and is designed so that the coefficient value increases linearly as time passes and as the corrosiveness of the fluid being handled increases.

[0138] The above-calculated aging coefficient (K) is reflected in the control logic to form an inverse relationship with the leakage coefficient (C_d) for predicting the final potential leakage diffusion amount. Since a higher aging coefficient (K) implies greater frictional resistance inside the pipe, the simulator incorporates physical friction losses that may occur in the actual field into the prediction model by adjusting the leakage coefficient (C_d) to be relatively lower. For example, in the case of a pipe that has been installed for 10 years and supplies highly corrosive sulfuric acid (H_2 SO_4), the aging coefficient (K) may be calculated to be approximately 1.5 or higher due to the combination of the installation years and the inherent corrosion index of sulfuric acid. In this case, the simulator adjusts the leakage coefficient (C_d) to be lower than the theoretical maximum value to accurately predict that the actual leakage amount will be somewhat less than the theoretical value due to the rough surface inside the pipe. These specific calculation and correction methods enable a three-dimensional aging assessment that considers chemical reactivity beyond simple time elapsed, thereby providing the technical basis for maximizing the precision of virtual process models within the digital twin space and preventing false positives.

[0140] Preferably, as the degree of aging increases, the internal roughness of the pipe increases and friction loss increases, so the outflow coefficient (C_d) is conservatively set high.

[0141] By combining the above data with the maximum operating pressure of the chemical supply end pressure (END POU PRESSURE), a maximum flow rate scenario is generated that can be leaked per second in the event of an accident.

[0143] Virtual HAZOP analysis implements traditional HAZOP techniques, which evaluate the safety of actual processes, as a virtual simulation in a digital twin space.

[0144] Specifically, variables (Deviations) such as 'No Flow', 'More Flow', and 'High Pressure' are sequentially input into a virtual model, and the system response when each variable occurs is calculated thousands of times for real-time pressure and gas concentration data collected by the leak panel control unit (100). Then, specific risk figures are derived, such as "when the pressure drops by 20%, how many seconds does the amount of leak diffusion exceed the safety standard set by the office?"

[0146] The step-by-step procedure for optimizing excessive threshold values ​​set in the worst-case scenario to suit actual field conditions is as follows.

[0147] The digital twin simulator module (310) sets an initial primary reference value based on the physical limit of the piping during the initial stage of accident detection. Specifically, it calculates the maximum value of the cross-sectional area (A) by assuming a full guillotine break of the chemical supply line and, based on this, generates the theoretically steepest pressure drop curve.

[0149] The above simulator (310) compares the pressure drop magnitude (ΔP) and drop slope (dP / dt) received from the actual pressure transmitter with the data of the virtual model in real time, and performs a gradient descent-based optimization algorithm to minimize the error between the two data.

[0150] The error function is defined by summing the residuals between the actual field pressure drop curve and the simulation curve at every sampling period (10ms).

[0151] If the actual rate of pressure drop is gentler than the worst-case scenario, the simulator gradually scales down the expected cross-sectional area (A) along the derivative direction (slope) of the error function. This is not a simple numerical reduction, but an iterative computational process that updates parameters to approximate the actual curve at every frame.

[0152] When the degree of agreement of the pressure curve reaches a certain level, the diffusion speed of the signal detected by the exhaust wind speed / differential pressure gauge and the gas detector (12) is combined and analyzed. If the diffusion of the gas concentration is faster than the prediction based on the pressure data, the fluid outflow coefficient (C_d) is adjusted upward to ensure physical consistency with the actual situation.

[0153] In the present invention, the criterion for converging the error range of the model to within 0.1% is based on the physical resolution of the field equipment. Considering that the measurement precision (Full Scale error) of the pressure transmitter installed in the field is typically between 0.05% and 0.1%, the virtual model is synchronized within the minimum error range that the hardware can measure. When the model error enters within 0.1%, which is the precision limit of the sensor, through iterative calculations, the simulator determines that optimization is complete and confirms the cross-sectional area (A) and outflow coefficient (C_d) at that point in time as final parameters. This is intended to ensure the highest data reliability required for real-time response while preventing unnecessary infinite loop calculations in the control system.

[0154] The solenoid valve (230) is driven only when the potential leak diffusion amount (V_leak), recalculated based on the optimized cross-sectional area (A) and the leak coefficient (C_d), exceeds a preset management threshold.

[0155] This [data comparison - gradient descent optimization - sensor-based convergence] process enables the accurate identification of the 'scale of pipe damage (pinhole or total rupture)' by backtracking during the early stages of an accident. This fundamentally prevents unnecessary process interruptions (false positives) caused by minute pressure fluctuations, while simultaneously delivering the technical effect of executing emergency shutdown sequences based on the most scientific evidence in actual crisis situations.

[0158] Fluid Density (ρ): This is an inherent physical property of the chemical being used. It exhibits an inverse relationship, meaning that the higher the fluid density, the relatively lower the outflow rate under the same pressure. This is automatically extracted from a chemical database pre-entered into the system.

[0160] The digital twin simulator module (310) performs technical judgments in the following temporal order.

[0162] 1. Data Collection and Normalization

[0163] Pressure data is received from a supply end pressure transmitter at a rate of more than 100 times per second, and normalized by applying a moving average filter to remove noise from the incoming data.

[0165] The specific design parameters and responsiveness optimization method of the Moving Average Filter applied to cancel out noise in the pressure data collected from the digital twin simulator module (310) and to normalize the signal are as follows.

[0167] The moving average filter according to the present invention targets pressure data received from a PLC (110) at a frequency of 100 times per second (100Hz) and includes a window size (N, number of weighted samples), which is a key parameter determining the precision of the filtering and the response speed of the system. The window size (N) represents the number of consecutive data samples included in the filter operation based on the current time point, and in an environment where the sampling period is 10ms (0.01 seconds), the response delay time (T_delay) due to filtering has a correlation of approximately '(N-1) × sampling period'.

[0168] In order to simultaneously secure electrical noise blocking performance at the site and immediate response capability in the event of a leakage accident, it is preferable for the above digital twin simulator module (310) to set the window size (N) to an integer between 5 and 15. If the window size (N) is less than 5, the delay time is reduced to 0.04 seconds or less, but the possibility of misjudging minute pressure vibrations at the site as abnormal signs increases; conversely, if it exceeds 15, the delay time increases to 0.14 seconds or more, causing an unnecessary increase in the potential leakage spread amount (V_leak). Accordingly, in the present invention, the window size (N) is set to 10 as the default value to obtain a meaningful data normalization effect while maintaining the response delay within approximately 0.1 seconds, and the resulting delay time of 0.09 seconds is treated as a constant within the overall safety threshold of the system and is utilized for inverse correction of the leakage time (t_0).

[0169] Looking at the specific filtering procedure, the system sequentially stores incoming pressure data in a buffer of size 10. Whenever new data is collected, the oldest sample is deleted, and the arithmetic mean of the 10 samples currently in the buffer is calculated to determine the normalized pressure data. This normalized data is then passed to the pressure drop change rate (dP / dt) judgment logic described earlier and used as basic data to determine whether an actual leak has occurred. By specifying a specific window size (N) linked to the sampling period and combining it with error correction logic based on delay time, the present invention prevents detection delays caused by indiscriminate averaging and provides a technical basis for predicting the scale of damage that may occur in the event of an actual leak accident most quickly and accurately.

[0171] 2. Filtering Anomalies

[0172] Normalized pressure data within a preset normal range (e.g., 6 kgf / cm²) 2 Perform primary filtering to see if it drops sharply beyond ±10%).

[0173] The digital twin simulator (310) does not simply look at the current pressure value being low, but constantly monitors the 'speed' at which the pressure drops. The system compares the currently measured pressure value with the pressure value measured in the previous period at intervals of a moment (e.g., 0.01 seconds) to quantify how much the pressure has changed during that short period.

[0174] To define a 'sudden drop,' the system first learns the maximum pressure fluctuations that can occur under normal operating conditions. It collects data on pressure fluctuations that occur when chemical supply pumps start or stop, or when normal process valves open and close, and sets a rate approximately 1.5 to 2 times faster than the fastest rate of pressure drop observed during these normal fluctuations as the criterion for an 'abnormal drop.' This is intended to identify only pressure drops at a level that cannot occur under normal process control.

[0175] The system completes the first filtering through the following specific drop rate criteria and verification procedures.

[0176] For example, the main supply pressure is 6 kgf / cm² 2 In a system, the pressure reaches 10% of the total (0.6 kgf / cm²) for a short period of 0.1 seconds 2 A case where it falls below ) is set as the primary condition for an 'abnormal sharp drop'.

[0177] A 'duration' condition is combined to filter out a single pressure spike caused by electrical interference around the wires or a temporary sensor error. Only when the aforementioned phenomenon of 'a 10% drop per 0.1 seconds' is observed three or more times consecutively (lasting for approximately 0.03 seconds or more) is it finally confirmed as an 'abnormal sign' caused by actual pipe damage rather than simple noise.

[0178] This provides a logical basis for the system to not react to minute vibrations at the site or normal valve operation (preventing false positives), while detecting danger within 0.1 seconds in the event of an actual accident and preparing for immediate shutdown.

[0180] 3. Calculation of instantaneous spill rate

[0181] For the data that has passed the first filtering, the cross-sectional area (A) and the outflow coefficient (C_d) are sequentially multiplied by the square root of the ratio of the differential pressure (ΔP) and the fluid density (ρ) to calculate the outflow rate per unit time.

[0183] The model used to calculate the instantaneous outflow velocity in this invention is based on the Bernoulli equation, which applies the law of conservation of energy to a fluid. That is, when high-pressure energy inside the pipe is released into the atmosphere through a leak hole, this pressure energy (ΔP) is converted into kinetic energy (1 / 2ρv 2 It is a quantification of the physical process of being converted into ).

[0185] The Bernoulli equation used in the present invention is based on the assumption of incompressible flow, which means that the fluid is not compressed. This assumption holds true under conditions where the pressure is relatively low, that is, when the difference between the internal pressure of the pipe and the atmospheric pressure is less than 30% of the absolute pressure.

[0187] In the case of sulfuric acid, a chemical commonly used in semiconductor fabs, the supply pressure is 6 kgf / cm² (gauge pressure of about 0.6 bar). When calculated based on atmospheric pressure (absolute pressure of 1.0 bar), this results in a relative pressure of about 37%, which exceeds the ideal application range of 30% for the Bernoulli equation.

[0189] However, the difference between the actual leakage rate measured in the field and the theoretical value calculated using the Bernoulli equation is only about 2 to 3 percent. This is because liquid sulfuric acid maintains a relatively constant density even under high pressure conditions, and also because friction loss due to mechanical failure is a more significant influencing factor than the high pressure effect in situations where the pipe suddenly ruptures.

[0191] Therefore, this system can predict actual leakage situations with sufficient accuracy without adding a conservative margin to the leakage rate calculated by the Bernoulli equation. Furthermore, considering the multi-condition AND combination and 120% safety margin applied by this system, a 2–3% error in the Bernoulli equation does not affect the safety assessment of the entire system.

[0193] In addition, the temperature of the semiconductor process environment is generally managed within a range of 15°C to 30°C, and within this temperature range, changes in chemical density are limited to approximately 3%. The digital twin simulator uses the temperature set during initial system startup (typically 25°C) as a reference value, and if the on-site ambient temperature deviates from this reference value, it corrects and applies the density in real time. This minimizes leakage rate prediction errors caused by seasonal changes or minute temperature fluctuations in the process environment.

[0195] The reason for taking the square root of the ratio of pressure difference to density is that, according to the kinetic energy formula, the outflow velocity (v) of the fluid is proportional to the square root of the ratio of pressure difference to density during the energy conversion process. This is to accurately reflect the physical characteristics that the outflow velocity increases as the pressure increases, and the outflow velocity is relatively lower due to inertia under the same pressure as the density of the fluid increases (heavier).

[0196] The reason for multiplying by the cross-sectional area (A) is to convert the calculated 'velocity (m / s)' into 'leakage rate per unit time (m²)' 3 This is a procedure for converting to / s). Since the absolute amount of chemical substances that spread at the same speed increases as the size of the leakage channel increases, this makes it possible to predict the actual scale of damage.

[0198] The reason for finally multiplying by the outflow coefficient (C_d) (correction for actual phenomena) is that theoretically calculated flow rates assume an ideal state free from friction or viscosity. However, in actual field conditions, factors such as pipe hole roughness, vena contraction as the fluid passes through narrow passages, and viscous friction occur, causing the actual outflow to be smaller than the theoretical value. The outflow coefficient (C_d) plays a key role in ensuring control reliability by correcting for these actual physical losses, thereby minimizing the error between simulation data and the actual leakage amount.

[0200] The digital twin simulator (310) first derives the 'theoretical maximum injection velocity' that can occur under the density (ρ) of the chemical substance based on the differential pressure (ΔP) collected from the pressure sensor, calculates the total flow volume by substituting the leakage cross-sectional area (A) optimized by virtual HAZOP analysis into the derived velocity, and finally calculates the final outflow velocity that actually diffuses into the FAB floor or the atmosphere by applying an outflow coefficient (C_d) that reflects the aging and friction conditions of the pipe.

[0202] 4. Calculation of Diffusion Amount

[0203] The calculated leakage rate is integrated over the time domain (t_0 to t_close) from the time of leakage detection to the time of expected closure completion to determine the final potential leakage diffusion amount.

[0205] The reason for performing integral calculations in the cumulative diffusion calculation is that the leakage rate continuously changes over time as the leakage progresses. This is intended to reduce errors that may occur when calculating simply as 'velocity × time' and to accurately reflect actual physical phenomena.

[0207] The final blocking priority and whether emergency control is required are determined through a combination of the following multiple conditions.

[0209] Evaluate whether the calculated potential leakage spread exceeds the risk tolerance limit (e.g., 500 ml) set by the manager. The threshold value is initially set by the manager based on the risk level of the process and can be variably adjusted based on operational data statistics.

[0211] In addition, when all three conditions are satisfied (AND combined) when ① the potential leakage diffusion amount exceeds 120% of the limit value, ② the hazardous gas concentration is detected above the upper limit by the gas detector, and ③ the differential pressure of the exhaust system is below the standard value, the highest level emergency shut-off sequence is activated.

[0213] Here, the potential leakage spread of 120% of the threshold value is an empirically calculated 'safety margin' to prevent malfunctions caused by sensor errors and transient pressure fluctuations. The reason for using 120% instead of 100% as the standard is to guarantee that it is a definite risk situation even when considering the uncertainty of the prediction algorithm.

[0214] Furthermore, the reason for setting the threshold at which hazardous gas concentrations are detected above the upper limit is to simultaneously check for the dangers of hazardous gases (fumes) dispersed into the atmosphere, as well as liquid leaks. This serves as a direct indicator of the potential for casualties.

[0215] The reason for setting the condition that the differential pressure of the exhaust system is below the threshold is that if the exhaust system does not operate normally (below the differential pressure), the risk of leaked gas spreading into the FAB increases exponentially. In other words, it is to identify the worst-case scenario where 'accident (leak) + collapse of the defense line (exhaust failure)' overlaps.

[0217] If an abnormality is detected in multiple lines, the chemical toxicity index and the calculated leakage spread amount are added together with weights of 0.4 and 0.6, respectively, to calculate a comprehensive risk score, and the solenoid valves (230) are controlled sequentially starting from the line with the highest score.

[0219] For example, a density of 1,840 kg / m³ 3 Assume a situation where an accident occurs in the phosphorus sulfuric acid supply line. The pressure at the supply end is 6 kgf / cm² 2 At 3kgf / cm² 2 If it drops sharply and it is predicted that it will take 2 seconds (t) until the valve closes, the system will have a dropped pressure value (3 kgf / cm²). 2 The outflow velocity is calculated by determining the difference between the pressure and atmospheric pressure. Since the density is high, the outflow velocity is calculated to be relatively slow compared to water, but if the outflow coefficient and cross-sectional area conditions are large, the amount of diffusion increases rapidly.

[0220] Since the calculated potential leakage spread is 800ml, which exceeds the preset risk limit of 500ml, the system immediately triggers an alarm at the office monitoring center (300) and simultaneously drives the solenoid valve (230) to shut off the valve.

[0221] Unlike conventional methods that simply sound an alarm after a leak occurs, this step-by-step computation and judgment configuration enables the quantitative prediction of the scale of damage that may occur until the shutdown is completed. In particular, by precisely considering multiple independent variables such as differential pressure, density, and mechanical delay time, it prevents excessive false positives and performs immediate shutdown only for actual hazardous situations that pass multiple conditions, thereby resulting in simultaneous improvements in the reliability and safety of semiconductor process operations.

[0223] Inside the remote monitoring center (300), a predictive maintenance analysis module (320) is installed along with a digital twin simulator module (310).

[0225] The predictive maintenance analysis module (320) according to the present invention is built inside a remote monitoring center (300) and integrates and analyzes accumulated operation data transmitted from a PLC (110) and virtual process data generated from a digital twin simulator (310). The module aims to move away from a simple threshold alarm method and to derive the optimal maintenance time before failure occurs by analyzing the physical deterioration pattern of a part.

[0227] This invention defines a quantitative indicator called the Health Score to calculate the remaining lifespan of a component and calculates the future failure time based on this. The key factors included in the calculation logic and their correlations are as follows.

[0229] - Cumulative operating cycle weight: Refers to the number of times a driving component, such as a solenoid valve (230), is physically opened and closed. Since mechanical wear is most directly proportional to the number of operations, it is set as the primary deterioration factor.

[0230] To quantify the physical wear of a part, the system collects and quantifies operating cycle data through the following steps for the above cumulative operating cycle weight.

[0231] The rising edge (the point at which the signal changes from 0 to 1) of the control signal (Digital Output) transmitted from the PLC (110) to the solenoid valve (230) is detected and counted in real time. The counted data is updated and accumulated in real time in the PLC's non-volatile memory or DB server so that it is not lost even in the event of a power outage or system reboot. The ratio of the 'current accumulated count' to the 'design life (e.g., 1 million times)' of the corresponding part is calculated. For example, if a valve with a life of 1 million times is operated 500,000 times, a basic weighting value of 0.5 is derived, which serves as the basic basis for deducting the total health score.

[0233] - Environmental Stress Weight: This quantifies the average pressure, temperature, and toxicity intensity of chemicals to which the component is exposed. It is used as a correction factor because the rate of physical degradation accelerates when operating in high-pressure or high-temperature environments outside the normal range.

[0234] Environmental stress weighting internally performs the following formula to quantify how harsh the environment was in addition to the number of simple operations.

[0235] To this end, the data collected from pressure transmitters and temperature sensors is not viewed simply as instantaneous values; instead, the average exposure value over the operating time is calculated, and the difference between the 'standard operating pressure' and the 'actual operating pressure' of the line is analyzed. For example, if the standard pressure is 6 kgf / cm² but the actual operation is 7.2 kgf / cm², it is determined that there is a 20% overload relative to the standard.

[0236] The final stress index is derived by multiplying by the corrosiveness grade of the chemical being handled (e.g., 1.5 for strong acids, 1.0 for general ultrapure water). If the pressure is 20% higher and a strong acid is being handled, the deterioration rate is defined as being 1.2 x 1.5 = 1.8 times faster than in a standard environment, thereby expanding the range of score deduction.

[0238] - Signal Drift Variable: Refers to the degree to which the zero-point of sensors, such as pressure transmitters (9), deviates from the initial set value over time. This serves as an objective indicator for determining the electrical aging of the sensor's internal components.

[0239] To derive signal drift variables, a zero-calibration data comparison method is used to detect the degradation of precision caused by sensor aging.

[0240] To do this, the sensor output value is checked when the equipment is stopped or the fluid in the piping is completely emptied to a 'zero pressure' state, and the minute voltage or digital value (e.g., 0.05 kgf / cm²) actually output by the sensor is measured at the point where, theoretically, 0 (zero) should be output. The difference between the zero point data at the time of initial installation and the current zero point data is the 'signal drift'.

[0241] Calculate the ratio of drift to the full scale of the sensor. For example, if the scale is 10 and a drift of 0.1 occurs, it is considered a 1% performance degradation, and as this ratio increases toward a preset tolerance (e.g., 2%), it acts as a variable that rapidly lowers the soundness score.

[0244] Here, as the cumulative operating cycles or environmental stress levels increase, the overall health score is reduced proportionally. This indicates that the physical life of the corresponding component is approaching the failure threshold.

[0245] In addition, when the health score starts at 100 points and decreases until it reaches a preset 'maintenance requirement threshold,' the system analyzes the slope of the score decline up to that point and determines the point at which the score becomes 0 points (point of failure) as the remaining life.

[0247] The predictive maintenance analysis module (320) performs the analysis in the following time sequence.

[0249] 1. Data preprocessing and feature extraction step: Load the operation history data for each part for the last 30 days collected from the PLC (110), pass it through a moving average filter to remove temporary noise, and then extract only the pure degradation trend.

[0251] 2. Individual degradation calculation step: Among the extracted data, a weight of 0.4 is assigned to the cumulative operating cycle, a weight of 0.3 to the stress index combining average operating pressure and temperature, and finally, a weight of 0.3 to the sensor signal drift.

[0253] In other words, 'mechanical wear' is the most fundamental and dominant cause of failure in drive components such as solenoid valves (230). Since the physical life of a component responds most honestly to the number of operations, the highest weight (0.4), which serves as the central axis for calculating the total degradation, is assigned as the weight of the cumulative operation cycle. That is, even if environmental conditions are favorable, the component must be replaced once the number of operations reaches its design life. Therefore, this is set as the 'basic degradation amount' to ensure the stability of the analysis.

[0255] Semiconductor processes handle high-pressure chemicals and highly corrosive chemicals. Even with the same number of operations, the rate of degradation accelerates exponentially in high-pressure or corrosive environments. Therefore, a stress index weight of 0.3 is the second highest after mechanical wear (0.4) and reflects the 'life shortening effect depending on the operating environment.' If this value is too low, it cannot predict early failure in harsh environments, and if it is too high, the system overreacts (false positives) to temporary fluctuations within the normal range.

[0257] Even if a component appears physically sound, the safety logic of the entire system collapses if the sensor's measurements drift. In particular, the "electronic aging" of gas detectors or pressure sensors is invisible and requires independent monitoring. Therefore, by assigning a weight of 0.3 to signal drift—equivalent to that of environmental stress—a "decline in measurement data precision" is regarded as a danger signal as serious as physical component failure. This serves as a strong line of defense to prevent secondary accidents caused by sensor malfunctions.

[0259] 3. Calculation of overall health score: The deterioration value to which the three weights above are applied is sequentially subtracted from the initial state score (100 points) to determine the overall health score of the current part.

[0260] In other words, this step is the process of integrating individually calculated degradation factors into a single quantitative indicator to visualize the current state of the component on a '100-point scale'.

[0261] The score at the time when the component is first installed and starts operation is defined as 100 points (best). Subsequently, the system sequentially subtracts from 100 points the value obtained by multiplying the three degradation factors (operation cycle, environmental stress, and signal drift) derived earlier by their respective weights.

[0262] As an example, each factor is first converted into an 'individual degradation percentage' ranging from 0 to 100. If 50% of the design life has been used, 50 points are derived, and 20 points are determined to be deducted by multiplying this by a weight of 0.4. If the severity is 20% higher than the standard, 20 points are derived, and 6 points are deducted by multiplying by a weight of 0.3. If it deviates by 10% of the tolerance, 3 points are deducted by multiplying 10 points by a weight of 0.3. The final score resulting from this is 100 - (20 + 6 + 3) = 71 points. This score indicates the current physical and functional health of the corresponding part and is updated in real-time at every data reception cycle and recorded in the DB.

[0264] 4. Remaining Life (RUL) Estimation Step: Calculate the slope of the health score decline over the past 7 days, and calculate the remaining time in days until the score reaches the 'failure judgment criterion (e.g., 10 points)' assuming this slope is maintained.

[0265] This step is the process of calculating the 'D-day' on which the part is expected to actually lose its function, under the assumption that the current downward trend in scores will continue into the future.

[0266] The system monitors the change in the overall health score recorded over the past 7 days. Rather than simply the difference between yesterday and today, it derives the slope (rate of decline) of the average daily score drop by performing a linear regression analysis on the 7-day data points. The reason for setting a 7-day period is that it is the optimal window for quickly reflecting recent changes in the operating environment while offsetting short-term operational fluctuations.

[0267] Since it is dangerous to wait until the component's score reaches 0, a 'failure judgment criterion (e.g., 10 points)' is established with a safety margin in mind. The point at which the score reaches 10 points is considered the actual end of life.

[0268] Divide the value obtained by subtracting the failure judgment criterion score from the current score (remaining margin score) by the previously calculated daily average drop score.

[0270] For example, if the current score is 50 points and the standard is 10 points, the margin is 40 points, and if the analysis over the past 7 days shows a decrease of 2 points per day, then 40 / 2 = 20 is calculated.

[0271] Based on this, the system outputs information to the administrator interface stating, "The remaining life of the part is 20 days." If the operating environment deteriorates rapidly and the daily drop increases to 4 points, the remaining life updated the next day is immediately shortened to 10 days and a warning is sent.

[0273] 5. Maintenance Decision-Making Step: Determine whether the remaining life calculated through the Remaining Ultimate Life (RUL) estimation step has entered within the pre-set 'safety maintenance period (e.g., 14 days)' and propose an optimal parts replacement schedule to the manager.

[0275] This module combines multiple conditions to generate a final maintenance alarm to prevent misjudgment.

[0277] The condition combination method is as follows.

[0278] Output 'Highest Level Maintenance Request' when all of the following conditions are satisfied: ① the overall health score is 20 points or less (OR), ② the slope of the health score decline within the last 24 hours has increased rapidly by more than 200% compared to the past average (AND), and ③ the remaining lifespan prediction value is within 7 days.

[0280] Here, Condition ① and [Condition ②+③] are combined via OR because there are two types of risks that the system needs to monitor.

[0282] Condition ① (Overall health score of 20 points or less) signifies 'gradual aging,' where the part wears out slowly according to its designed lifespan. It is the final safety line indicating that maintenance is required if the part itself has reached its limit, even if the slope is not steep. Conditions ②+③ (surge in slope and RUL within 7 days) detect cases where the part has entered a 'sudden failure' path due to a sudden external impact or change in environment, even if the score is still high (e.g., 80 points).

[0283] Through this OR combination, you can monitor both "old and dying parts" and "parts that suddenly started to break down" without missing any.

[0285] In addition, conditions ② and ③ are combined with AND to increase the 'certainty' of the prediction.

[0286] In other words, if only the slope is considered (② alone), the slope can temporarily spike by more than 200% during instantaneous voltage fluctuations or temporary high-voltage operation. If an alarm sounds every time this occurs, field workers will lose trust in the system (false positive).

[0287] By combining with RUL (②+③), a second verification is performed to determine whether the steep slope is not merely noise, but a critical trend where failure is predicted within 7 days based on actual calculation results.

[0288] This AND combination acts as a filter to distinguish between "brief shaking" and "real breakdown."

[0290] The reason a specific figure of 200% relative to the historical average was set here is to identify the 'accelerated deterioration zone.' When any machine part reaches the end of its lifespan, the rate of wear accelerates exponentially; the fact that the score drops at twice the normal rate (200%) indicates that the part's physical structure has already entered a breakdown phase and has begun to trace an 'exponential failure curve.' This is a clear signal that it is time for 'replacement' rather than simple repair.

[0292] In addition, 7 days (RUL) is the minimum physical preparation time required for the emergency supply of parts and the deployment of maintenance personnel in a semiconductor FAB. Even considering prediction errors, 7 days is the final golden time for responding before an accident occurs.

[0293] A Health Score of 20 indicates a state where the component's performance has been lost by 80% compared to its initial level. In places handling high-risk chemicals, such as semiconductor processes, the standard practice is not to wait until the component's performance reaches zero, but to perform maintenance while maintaining a Safety Factor of 20%.

[0295] The maintenance requirement threshold (20 points) and the safety maintenance period (14 days) are initially set by summing the average component supply cycle of the semiconductor line and the standard time required for replacement work, and subsequently, the system automatically learns and optimizes them as actual failure case data accumulates.

[0297] As an example, when applying actual values ​​to a solenoid valve (230) with a lifespan of 1,000,000 cycles, the following applies.

[0298] Situation: Currently, the cumulative operating cycle is 900,000 (90%), and the average operating pressure is 100% higher than the standard (harsh conditions)**, and the sensor drift has progressed to 80% of the allowable error.

[0300] Calculation process: The system applies the operating cycle score (90 points * weight 0.4 = 36 points deducted), stress weight (100 points * weight 0.3 = 30 points deducted), and drift variable (80 points * weight 0.3 = 24 points deducted).

[0302] Final judgment: The overall health score is calculated as 100 - (36 + 30 + 24) = 10 points. Since this falls below the maintenance threshold (20 points) and has reached the failure judgment criterion (10 points), an immediate maintenance recommendation is issued.

[0304] Unlike conventional methods that simply calculate the lifespan of parts in years, this configuration allows for quantitative evaluation by combining the actual operating environment (stress) and the part's response (signal drift). This prevents the premature replacement of unnecessary parts, thereby reducing maintenance costs, while simultaneously preventing FAB shutdowns caused by unexpected part failures, resulting in maintaining overall process availability and production reliability at over 99.9%.

[0306] FIG. 5 is a diagram showing the logical flow of a chemical leak detection and automatic shut-off process according to the present invention, which is described in detail as follows.

[0308] 1. Startup and Monitoring Phase

[0309] When the system is activated, it determines, monitors, and assesses the operational status of chemicals within the FAB in real time. It collects on-site data from leak sensors, pressure transmitters (PT), anemometers / differential pressure gauges, gas detectors, and other sources.

[0310] Signals from the organic explosion-proof zone are safely transmitted to the PLC through a barrier.

[0312] 2. Abnormality determination step

[0313] The PLC (110) compares and analyzes the data collected through the I / O contacts with the preset safety thresholds.

[0314] It determines whether chemical leakage, wire breakage, excessive gas concentration, or abnormal waste liquid temperature (set criteria such as 130°C or higher) is detected.

[0316] 3. Emergency control output stage

[0317] If an abnormal situation is determined, the PLC (110) immediately transmits an electrical driving signal to the solenoid valve of the pneumatic unit (200).

[0318] The solenoid valve turns the air on / off based on the received electrical signal and converts it into a physical pneumatic signal.

[0320] 4. Physical Blocking and Safety Assurance Steps

[0321] The converted pneumatic signal is transmitted to the automatic valves of the CIB or VMB, which are internal supply equipment within the FAB, to immediately close the valves. This immediately cuts off the supply of chemicals, preventing further leakage and the spread of accidents, and, if necessary, blocks the drainage of waste liquid to prevent environmental contamination.

[0323] 5. Alarm Generation and Remote Reporting Steps

[0324] Simultaneously with on-site containment measures, overall chemical-related alarms and status signals are transmitted to the office.

[0325] Managers can immediately check the location of the accident and the status of the equipment through the remote monitoring system and perform a rapid maintenance response.

[0327] Although various embodiments of a PLC-based chemical leak monitoring and integrated control system for semiconductor processes have been described above, the PLC-based chemical leak monitoring and integrated control system for semiconductor processes is not limited to the embodiments described above. Various devices or methods that can be implemented by a person skilled in the art by modifying and varying the embodiments described above may also be examples of the PLC-based chemical leak monitoring and integrated control system for semiconductor processes described above. For example, even if the described techniques are performed in a different order than described, and / or the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from the described method, or are replaced or substituted by other components or equivalents, it may still be an embodiment of the PLC-based chemical leak monitoring and integrated control system for semiconductor processes described above. Explanation of the symbols

[0328] Leak panel control unit (100) PLC(110) Leakage circuit breaker (120) Circuit protector (130) DC power supply (140) Barrier (150) IO terminal block (160) Leak panel pneumatic part (200) ball valve (210) Regulator (220) solenoid valve (230) Air check section (240) Finger valve (250) External interface (300)

Claims

Claim 1 A system for preventing accidents by monitoring in real time whether a chemical leak may occur during operation within a semiconductor manufacturing facility (FAB) and immediately cutting off the supply in the event of an abnormality, comprising: a leak panel control unit (100) that analyzes signals received from sensors installed at the semiconductor factory site and issues a control command if the signal deviates from a set safety standard; a leak panel pneumatic unit (200) that converts an electrical control signal received from the leak panel control unit (100) into a physical pneumatic signal to drive an actual valve; and an external interface (300) that collects status signals from sensors installed at the semiconductor factory site, transmits them to the leak panel control unit (100), and executes a physical cutoff command based on the judgment of the leak panel control unit (100). The leak panel control unit (100) includes a PLC (110) for monitoring the internal condition of the FAB in real time using I / O contacts and issuing control commands including solenoid valve closure in the event of an abnormality, and a leakage current for detecting a leakage current of the main AC power supply flowing into the system and cutting off the power. The leak panel pneumatic section (200) includes a circuit breaker (120), a circuit protector (130) for protecting the circuit by blocking overcurrent of a circuit configured for individual purposes including a leak sensor, a pressure sensor (PT), and a wind speed / differential pressure gauge, a DC power supply (140) for supplying stable DC power to various instruments installed inside semiconductor production facilities for measuring chemical substances, a barrier (150) for safely detecting the signal of a leak sensor installed in an organic explosion-proof zone and transmitting it to a PLC, and an IO terminal block (160) that performs the role of an interface for interconnecting individual contacts of the PLC and cables coming up from the field, and the leak panel pneumatic section (200) includes a ball valve (210) for manually controlling the supply and cutoff of main CDA (Clean Dry Air) supplied to the entire system, and a regulator (220) for adjusting the air pressure to a preset constant pressure for automatic valve operation, andIt includes a solenoid valve (230) for driving an automatic valve of supply equipment within the FAB by receiving an electrical signal from the PLC (110) and turning actual air on / off, an air check unit (240) for detecting when the pneumatic supply pressure exceeds a reference value and generating an alarm through the PLC (110), and a finger valve (250) for individually turning on / off the pneumatic supply to the solenoid valves installed for each manifold, and the external interface (300) includes input sensors that transmit physical status signals of the site, including a leak sensor, a terminal pressure transmitter, a wind speed / differential pressure gauge, a gas detector, and a waste liquid temperature sensor, supply and shut-off equipment including a CIB (Chemical Insulated Box), VMB (Valve Manifold Box), and an automatic valve that perform physical operations by receiving a control signal from the leak panel pneumatic unit (200), and a remote monitoring center that visualizes the final information collected and determined by the leak panel and displays it to the manager, and external A remote monitoring center of the interface (300) includes a digital twin simulator module (310) that replicates and simulates the physical components and process status of the leak panel control unit (100), leak panel pneumatic unit (200), and external interface (300) in a virtual space based on real-time sensor data, wherein the digital twin simulator module (310) collects and normalizes pressure data to filter out abnormal signs, calculates an instantaneous leakage velocity by multiplying the cross-sectional area and leakage coefficient by the square root of the ratio of differential pressure and fluid density, determines the potential leakage diffusion amount by performing an integral operation from the time of leak detection until the completion of valve closure, and activates an emergency shut-off sequence when all three conditions are satisfied: the potential leakage diffusion amount exceeds a preset threshold, the hazardous gas concentration in the gas detector is above the upper limit, and the differential pressure of the exhaust system is below the reference value. Claim 2 delete Claim 3 delete Claim 4 delete Claim 5 delete Claim 6 delete

Images