Abstract: This paper focuses on the on-site control unit upgrade project of the computer monitoring system at the Qingjiang Geheyan Hydropower Plant. It explores a novel architecture for direct-internet-connected LCUs, the first of its kind in a large-scale hydropower plant in China, utilizing Schneider Electric Ethernet products. The paper discusses redundancy strategies for inputs, outputs, power supplies, CPUs, and connections within this architecture, particularly methods for achieving redundancy in binary input scenarios. The paper analyzes the distinct characteristics of the five languages of the International Electrotechnical Commission (IEC) 1131: FBD, LD, SFC, ST, and IL, and discusses strategies and methods for writing PLC programs using Structured Text (ST). The paper also analyzes strategies for using AC sampling devices and transmitters. Finally, it discusses effective ways to shorten the construction period for on-site LCU upgrades at hydropower plants.
Keywords: Hydropower plant on-site control unit modification structure, unmanned operation
After years of effort, computer monitoring systems are being used more and more widely in hydropower plants and other fields. For hydropower plants, adopting a well-structured, fully functional, highly reliable, and user-friendly computer monitoring system is crucial for improving safety and achieving "unmanned operation (shutdown)." Encouragingly, thanks to the efforts of domestic colleagues, China's computer monitoring technology has developed rapidly, approaching or reaching the international advanced level for similar products.
With the rapid development of computer hardware and software in recent years, domestic computer monitoring technology has continued to advance. The author participated in the computer monitoring system upgrade project of the Qingjiang Geheyan Hydropower Plant. This article presents a preliminary exploration of the characteristics of the plant's LCU upgrade, the new technologies adopted in the upgrade, and the novel LCU structure, offering personal perspectives. Any inaccuracies are open to criticism and correction.
1. Objectives of the monitoring system upgrade
The Geheyan Hydropower Plant originally used a Canadian computer monitoring system, which had been operating stably for many years, making due contributions to the plant's safe production and its goal of becoming a first-class hydropower plant in China. However, with the development of the national economy, the requirements for power systems and power plants are becoming increasingly higher. It is imperative to align with the technical and management levels of world-class hydropower plants and create a world-class hydropower plant, thereby achieving the goals of high management standards, advanced technology, further refined personnel, and closed-loop operation. On the one hand, the original system's functions can no longer meet the requirements; on the other hand, ordering spare parts is becoming increasingly difficult and expensive, posing a potential threat to the plant's safe operation. Therefore, the old system must be upgraded to lay a solid foundation for creating a world-class hydropower plant. For the LCU (Limited Control Unit), the upgrade method is as follows: only the original panel cabinets, automatic synchronizing devices, 24V power supplies, lighting, and a few other accessories are retained on-site; everything else is removed and replaced with a new LCU, using Schneider Electric's Quantum series PLC as the controller. The Automation Institute of the China Institute of Water Resources and Hydropower Research provided five LCUs. The author of this article participated in the development, on-site installation, commissioning and other modification work of the LCUs. This article is a summary and reflection on the modification work.
2. Characteristics of LCU Retrofit
2.1 Different control process methods
The original monitoring system was developed by CAE in Canada, and CAE's approach differs significantly from the conventional practices in China. For example, there are nine steps involved in starting up and nine steps involved in stopping the unit. For conventional hydropower plants, our conventional approach is a five-state transition: shutdown, idling, no-load, generating, and unpredictable (the transitional state among the first four states is called the unpredictable state). (For units with phase-shifting tasks, there is also a phase-shifting state; for pumped storage units, there is also a pumping state; but these are not within the scope of the conventional units discussed.) The unit is always in one of these five states. The operations of starting up, stopping, disconnecting from the grid, and reconnecting to the grid after disconnection are simply transitions between the four states: shutdown, idling, no-load, and generating. Although the two representations are essentially the same, those accustomed to the five-state transition need time to become familiar with the nine steps involved in starting up and stopping. Considering that power plant personnel, from operators to maintenance staff, are familiar with the nine steps of starting and stopping the unit, the original nine-step start-up and shutdown procedure was adopted to facilitate operation and maintenance for power plant personnel, even though programming and debugging require considerable effort to adapt to.
2.2 Programming using a structured text language
The original computer monitoring system's LCU program was written in a text-based language, similar in style to C. Similar to the nine-step process for unit start-up and shutdown, and given that power plant maintenance personnel are familiar with text-based languages, it was required that all LCU programs be written in a text-based programming language. When using a programmable logic controller (PLC), ladder diagrams are typically used. Their advantages include ease of learning, intuitiveness, and a strong resemblance to electrical secondary circuit diagrams, making them very suitable for power plant personnel and facilitating program maintenance by on-site personnel. In the LCU upgrade of the Geheyan computer monitoring system, Schneider Electric's Quantum series PLCs were adopted, with Concept 2.2 as the programming software. This software supports all five languages of the IEEE 1131 standard: FBD (Function Block Diagram), SFC (Sequential Function Chart), LD (Ladder Diagram), ST (Structured Text), and IL (Instruction List). The first three languages are graphical, while the latter two are text-based. Due to the characteristics of Instruction List (IL) languages, they suffer from poor readability, simple but unintuitive instructions, poor portability, unstructured text (with JUMP instructions), and weak data processing capabilities (no FOR loop statements), making them suitable only for small-scale control. ST language is a structured text language, very similar to C. It not only has rich logic processing capabilities but also includes statements such as IF, CASE, FOR, WHILE, REPEAT, EXIT, and EMPTY, resulting in very strong data processing capabilities. It lacks GOTO, JUMP, or similar instructions. Therefore, it has excellent portability, which is beneficial for program standardization. Compared to FBD, LD, and SFC, it is less intuitive and deviates significantly from the electrical secondary circuit diagram. Furthermore, its disadvantages include higher memory usage and longer scan cycles (compared to FBD, LD, and SFC). The advantages of LD language were mentioned above; FBD diagrams are closer to the electrical secondary circuit schematics. FBD, SFC, and LD lack statements such as IF, CASE, FOR, WHILE, REPEAT, EXIT, and EMPTY, resulting in insufficient data processing capabilities. Based on my personal experience, the preferred methods are: (1) using ST exclusively; (2) combining ST with FBD; and (3) combining ST with LD. Methods (2) and (3) combine the characteristics of the two languages, making them better approaches. Because using a text-based language with strong data processing capabilities to process data, and using the intuitive LD or FBD to develop sequential control processes, makes it easier for on-site technicians to accept, understand, and engage with the process. On-site technicians are most concerned about sequential control processes. I personally prefer method (3).
However, for engineers familiar with C or other text-based languages like C, or for situations where implementing particularly complex sequential control processes using LD or FBD is difficult, using the structured text-based ST language is a wise choice. This is the case with Geheyan Power Plant. Their original LCU (Local Unit) computer-monitored system, based on Canadian CAE, was entirely programmed in a text-based language similar to C. Their unit sequential control processes were also very complex, so the power plant required all processes to be programmed in ST language. This way, the modified LCU program is similar in style to the original program, making it easier for the power plant's technicians to understand and maintain. Practice has proven that choosing ST language was the correct decision.
2.3 PLC Direct Internet Connection
After years of exploration and practice, computer monitoring systems generally adopt a hierarchical, distributed system architecture, that is, divided into units according to the controlled equipment, namely LCUs. A common LCU typically consists of an industrial computer, a PLC (for data acquisition and control), an automatic synchronizing device, a speed control device, a transmitter, a power supply, and other accessories. The industrial computer acts as a node on the intranet of the computer monitoring system. Various data are sent to other nodes on the network via the industrial computer, and control commands are issued to the controllers and other devices via the industrial computer. Therefore, the reliability of the industrial computer is extremely important. Although the industrial computer is an industrial control product, its reliability is somewhat reduced due to the presence of rotating components such as fans, hard drives, and floppy drives. In response to this situation, people have turned their attention to Ethernet, considering direct internet access for PLCs. Currently, PLCs from several major international manufacturers can achieve direct internet access, such as Schneider Electric's entire Quantum and Premium series, General Electric's GE90-70, GE90-30, and VersaMAX series, Siemens' PLCs, and Rockwell's PLC controllers.
In the Geheyan computer monitoring system upgrade project, direct internet access was adopted. However, its structure still conforms to the principle of hierarchical distribution (unit). This structure meets the goal of "unmanned operation (operation with doors closed)".
2.4 Redundant Structure
Dual-machine hot standby redundancy
While PLCs are now highly reliable, to further enhance the reliability of large and extra-large power plants, especially to meet the "unmanned operation (shutdown)" requirements of large backbone power plants like Geheyan, and to facilitate maintenance (one unit running while the other can be programmed), a dual-CPU hot standby structure is adopted. There are two ways to implement dual-CPU hot standby: hardware and software. Hardware methods include Schneider Electric's Quantum series PLC dual-CPU hot standby and GE's GE90-70 series dual-CPU hot standby; software methods, such as GE's GE90-30 series dual-CPU hot standby, generally offer better performance than hardware methods. However, regardless of the method, seamless switching is crucial. This means that the switching process must ensure continuous control and no data loss. This is extremely important.
In the Geheyan computer monitoring system upgrade project, a dual-machine hot standby structure using Schneider Electric's Quantum series PLCs was adopted. When the main control CPU fails or power is lost, it automatically switches to the standby CPU, which then automatically becomes the main control CPU, achieving a seamless switchover. During maintenance, the main and standby units can be manually switched. This improves reliability.
Fiber Ethernet redundancy
For LCUs, the connection method, or networking method, with other nodes in the system now commonly uses Ethernet, with fiber optic cable as the medium. While the reliability of a single network is already high, considering other unforeseen mechanical and physical factors, a dual-fiber Ethernet configuration can be considered.
The Geheyan computer monitoring system employs dual fiber optic Ethernet. From the perspective of the LCU (PLC), its dual fiber optic Ethernet operation requires no switching and operates simultaneously (ALL IN WORKING). This eliminates the need for switching; if one network fails, the second network can be switched over instantly. This results in very high performance. This functionality is achieved through the Ethernet capabilities of the Quantum series PLC.
Redundancy of connection cables to remote chassis
Typically, an LCU unit requires several expansion chassis. For example, the Geheyan project has five expansion chassis. The Quantum series PLC offers two connection methods: Remote Input/Output (RIO) and Distributed Input/Output (DIO). The Geheyan project uses the RIO method, and its expansion chassis are called RemoteDrop.
In general, a single cable connection between the main control unit and the expansion unit is sufficient. However, using a dual cable can achieve higher reliability. For example, at the Longyangxia Hydropower Plant, a single 320 MW unit has its PLC main control unit located upstream, with one remote station downstream. At this power station, a dual cable connection is used between the main control unit and the remote station, with one cable running from the left and the other from the right. This ensures that even if unforeseen factors such as fire or mechanical damage occur on one side, the normal operation of the system will not be affected.
The dual-cable structure was also adopted in the Geheyan computer monitoring system upgrade project.
Input redundancy
Generally, input redundancy should employ a "two-out-of-three" approach, where the minority accepts the majority. General Electric's GE series PLCs have this hardware structure, with triple redundancy at the CPU, inputs, and outputs to meet extremely high reliability requirements, such as in blast furnaces in the metallurgical industry. However, this requires a large investment and is unnecessary in the hydropower industry. However, for a few critical signals, such as those used for emergency shutdowns or status signals from critical equipment like circuit breakers, which have two input signals, a problem arises regarding how to handle them. This issue was encountered in the Geheyan computer monitoring system upgrade project. The solution was to use a safety bias factor. Different safety bias factors can be used in different states and processes. Thus, by adding a set of safety bias factors to a pair of redundant signals, a set of signals can be obtained for different states and processes. A safety bias factor can be the result of logical operations on several or even more signals.
5) Output redundancy
For certain critical equipment, such as demagnetizing switches and outgoing circuit breakers, to ensure high reliability under any circumstances, each such device needs to be configured with two outgoing channels, i.e., redundant channels, to guarantee reliability. Some power plants' outgoing circuit breakers, such as those at Geheyan, have two coils for tripping: a normal operating coil and a backup coil. Excitation of either coil will cause the circuit breaker to trip. This is essentially equipment redundancy. The State Grid Corporation of China included such a requirement in its "Draft for Comments on Unmanned Operation (Door-Off Operation)".
Generally, there are two strategies for redundant channel actions:
One strategy is to use two redundant channels operating simultaneously;
Another strategy is to use a redundant second channel to re-operate after the first channel fails.
Clearly, simultaneous operation is not ideal, because the equipment would function normally under normal circumstances. In this situation, the second channel's operation is unnecessary, as it might cause the second redundant channel to malfunction if the first channel fails.
The latter method is better. Monitor the actions of the first channel; if no status feedback is received within a certain time, then the second channel will take action. This way, the second channel has almost no "opportunity to perform," and once it is given the chance, it will "perform accurately and appropriately." This approach was used in the Geheyan computer monitoring system upgrade project, achieving relatively ideal results.
Power redundancy
The importance of power supply is self-evident. No matter how good the equipment, without power, nothing is possible. In the Geheyan computer monitoring system upgrade project, power redundancy technology was adopted. The main power supply of the Geheyan power plant is DC110V, and the I/O power supply is DC24V. In the remote station, two power modules are used for redundancy, while a single power supply is used in the CPU chassis (each CPU chassis occupies an independent baseboard). The power plant provides two DC110V power supplies: one supplying one CPU chassis power module and one power module in the remote station, and the other supplying another CPU chassis power module and another power module in the remote station. Both DC110V supplies generate DC24V through DC/DC conversion, creating redundancy between the two DC24V supplies. Under normal circumstances, both DC110V supplies are powered, the two CPU chassis operate normally (one primary, one backup), the DC24V supplies operate normally, and the two power supplies in the remote station each bear half the load of their respective chassis. When one DC 110V circuit fails, one of the CPU chassis can continue to operate normally (when the power supply to the backup CPU chassis is lost, there is no operational intervention or impact; when the primary CPU loses power, the backup CPU seamlessly switches to become the primary CPU), while the other DC 24V circuit continues to operate normally, with one of the power supplies at the remote station handling the entire load of that chassis. This achieves power redundancy.
2.5 AC Sampling and Transmitter
The use of AC sampling is increasing, showing a strong trend of replacing transmitters. However, the understanding of AC sampling is not yet clear. The computer monitoring upgrade project at Geheyan Power Plant demonstrated a suitable and appropriate approach to AC sampling and transmitter processing, which is worthy of emulation by other power plants. The suitability and appropriateness here refers to the processing of AC quantities, as transmitters are currently the only option for acquiring and processing DC and non-electrical quantities. Specifically, transmitters are used for control components with high real-time and reliability requirements, such as generator synchronization, active power, voltage (reactive power), and guide vane opening limits. AC sampling devices are used to collect electrical parameters of the generator, such as three-phase current, three-phase voltage (phase and line), active power, reactive power, and power factor.
2.6 The CableFast high-speed wiring system is adopted.
Following conventional wiring methods, various I/O modules require wiring to the control panel terminals. The Geheyan Hydropower Plant units have large single-unit capacities and are designed with meticulous attention to detail, resulting in a large number of I/O points. There are 320 digital inputs, 128 digital outputs, 64 temperature points, 32 non-electrical analog inputs, and 8 analog outputs. There are a total of 25 I/O modules, each with 40 terminals requiring connection to existing terminals, totaling at least 1000 wires. However, on-site installation and modification time is extremely limited, necessitating a more efficient method. Using Schneider Electric's CableFast quick-connect system is a good solution. CableFast is a standard product from Schneider Electric; it pre-connects Quantum terminal blocks and terminal blocks with cables. The terminal blocks can be directly mounted on DIN rails, and external wiring can be directly connected to the terminal blocks. This reduces the amount of wiring work and saves a significant amount of time, making it a superior approach.
2.7 Carefully designed and meticulously prepared
To minimize the workload of on-site wiring and modifications, and to facilitate future maintenance, careful design is essential. A reasonable layout and configuration are crucial. Generally, external terminal wiring should not be altered to reduce construction time. Furthermore, thorough preparation is necessary, including all necessary accessories and tools; otherwise, the project schedule will be affected. During the on-site modification at Geheyan, we effectively addressed this issue, ensuring the smooth progress of the modification and assembly work. While maintaining the project schedule, we achieved the same level of process operation as the imported equipment, earning high praise from the power plant.
3. Conclusion
The transformation of the local control unit (LCU) at the Qingjiang Geheyan Hydropower Plant involved innovations in structure, technical approach, and implementation methods. These innovations are primarily reflected in the first-ever use of a QUANTUM PLC for direct internet connection in the hydropower industry (eliminating the industrial control computer), and the adoption of dual-machine hot standby redundancy, dual networks, partial I/O redundancy, and power supply redundancy.
The Geheyan Hydropower Plant's LCU, especially the unit LCU, has a large number of I/O points, three to four times that of typical units of the same size. Furthermore, the upgrade requirements for the monitoring system were high, particularly given the tight timeframe. On-site installation, wiring, and commissioning had to be completed in less than three weeks, a very tight schedule. Through active cooperation with the power plant and other relevant parties, the computer upgrade project at the Geheyan Hydropower Plant has been essentially completed. Currently, the equipment is operating well, the expected goals have been largely achieved, and the results are positive.
About the author:
Liu Xiaobo, male, is the manager of the Monitoring Engineering Department I of the Automation Institute of the China Institute of Water Resources and Hydropower Research, and a senior engineer. He is engaged in the research, development, and design of computer monitoring systems.
Wang Dekuan, male, is the director of the Automation Institute of the China Institute of Water Resources and Hydropower Research, a professor-level senior engineer, and is engaged in the research, development, and design of computer monitoring systems.
Liu Xiaopeng, male, is an engineer at the Monitoring Engineering Department I of the Automation Institute of the China Institute of Water Resources and Hydropower Research. He is engaged in the research, development, and design of computer monitoring systems.
