DESIGN OF AN ETHERNET BUS INTERFACE CONTROLLER IN A NUCLEAR POWER PLANT SIMULATOR

Size: px

Start display at page:

Download "DESIGN OF AN ETHERNET BUS INTERFACE CONTROLLER IN A NUCLEAR POWER PLANT SIMULATOR"

Jessie Aubrey Sullivan
5 years ago
Views:

1 Proceedings of the 18th International Conference on Nuclear Engineering Proceedings of the 18th International Conference on Nuclear Engineering ICONE18 ICONE18 May 17-21, 2010, Xi'an, China May 17-21, 2010, Xi'an, China ICONE18- ICONE DESIGN OF AN ETHERNET BUS INTERFACE CONTROLLER IN A NUCLEAR POWER PLANT SIMULATOR Lixuan Lu Faculty of Energy Systems and Nuclear Science, Faculty of Engineering and Applied Science, University of Ontario Institute of Technology, Oshawa, ON Canada Dong Le Pickering Learning Center Ontario Power Generation Pickering ON, Canada ABSTRACT Obsolescence presents great challenge to Nuclear Power Plants (NPP) and plant simulators around the world. Old designs will have to be either modified, or replaced by new designs, in order to simplify maintenance, increase availability and meet ever-increasing operational and training requirements. Control system upgrade and Distributed Control System (DCS) design for both old plants and new builds have become the center of interest. In a DCS, communication networks connect control systems together to allow the exchange of information and feedback. Among the many communication network protocols, Ethernet can be a promising one. This paper describes a new Ethernet Bus Interface Controller (ebic) used in the Input/Output (I/O) system of a Nuclear Power Plant (NPP) simulator in Canada. Key words: Ethernet Bus Interface Controller, Nuclear Simulator, Nuclear Power Plant, Distributed Control System 1. INTRODUCTION Nuclear Power Plant (NPP) control systems based on conventional control technology face a serious problem of obsolescence. With nuclear power plant life extension and new builds expanding globally, new control techniques are expected to be implemented. Distributed Control System (DCS) design for both old plants and new builds have become the center of interest (Lee, 2005 and Kim, 2000). Atomic Energy of Canada Limited (AECL) has been investigating the use of DCS for Advanced CANDU Reactor (ACR) (Brown, 2004). It is indicated that the use of DCS for ACR can provide redundancy, partitioning, and channelization of control function as well as separation of control and user interface function. This can produce a control system with improved maintainability and enhanced reliability. Custom designed networks have been created specifically for use in NPPs. The design of a three-level communication network for a DCS in NPPs is presented by Kim et al. [2000]. The three levels of communication networks include: information network, control network, and field network. Park et al. [2000] proposed a control network for a DCS to satisfy the requirements of NPPs. The network is referred to as Plant Instrumentation and Control NETwork+ (PICNET+). PICNET+ uses the physical layer of fast Ethernet. Its design adopts optical fibre as a transmission medium and star connection as the network topology to enhance its reliability. Choi [2002] presented a communication network called Ethernet based Real-time Control Network (ERCNet). It was also specifically developed to allow the implementation of a control network of a DCS. ERCNet adopts ring topology with token passing mechanism for ease of maintainability. It uses fibre optic physical media of fast Ethernet and also uses replicated memory. Simulators in NPPs play an important role in not only training operators, but testing of newer designs. The Darlington NPP simulator in Canada has been placed in service since The computer equipment and electronics with which the simulator was built are typical of the mid 1980 s. For instance, the Encore Seahawk 32/2040 was used as simulation computer, while Ramtek 9400 was used as the display system. The system this paper focuses on is the Bus Interface Controller (BIC) used in the Input/Output (I/O) system of the 1 Copyright 2009 by ASME 1 Copyright 2010 by ASME

2 Darlington simulator. The Bus Interface Controller (BIC) is the most integral component in the whole I/O system. It controls the passage of information between the plant simulator and the I/O devices. The original BIC designed in the late eighties cannot meet the training requirements and therefore needs to be updated. A new design, the Ethernet Bus Interface Controller (ebic) is presented in this paper. This design replaces the traditional point-to-point communication between the simulator and the I/O devices. Instead, the Ethernet bus is deployed to pass information. Ethernet uses Carrier Sense Multiple Access with Collision Detection (CSMA/CD) mechanism for resolving contention on the communication medium. The maximum number of transmission attempts is 16. The maximum number of nodes that can be connected to the network is larger than 1000 over a maximum length of 2500 m. The data transfer rate is 10 Mbps and the maximum data size is 1500 bytes [Lian, 2001]. The paper is organized as follows: firstly, the overall I/O system is introduced. Subsequently, the electronic components in the original BIC and the newly designed ebic are described in detail. Finally the conclusion is drawn. Fig. 1: Simplified Simulator Block Diagram 2. I/O SYSTEM DESCRIPTION A simplified pre-upgrade block diagram of the simulator is depicted in Fig. 1. As shown, the simulator consists of the Main Control Room (MCR), the Main Control Panel Interface (MCPI), and the Simulation Computer System (SCS). The SCS is an Alpha ES40 server running Tru64 Unix OS hosting the simulation software which runs in real time. The MCPI comprises of a set of fan-out and I/O cabinets through which the simulation software controls the panel devices. The MCP is a set of panels mimicking the real control panels at the station. In addition, the simulation software employs a display subsystem to display various kinds of graphical data, such as trend, alarms summary, etc. The display system includes a set of standard PCs running Exceed, acting as local X servers for the simulation software. Finally, the PDP 11/70 digital control computer (DCC) is emulated in the simulator using the SIMH emulator, a computer history simulation system. Of all the sub-systems of the simulator, the MCPI is the focus of this paper. A detailed block diagram of the existing MCPI is presented in Fig. 2. As shown, the MCPI consists of several I/O cabinets connected to two fan-out chassis. The SCS communicates to the MCPI fan-out chassis through the PCI- VME adapters and the DACBUS-SCS Adapter (DSA). The data transfer between the fan-out chassis and the I/O cabinets is Fig. 2: Main Control Panel Interface (MCPI) handled by a number of Bus Interface Controller (BIC) and Input/Output Buffer (IOB) cards. All communication to the MCPI system, however, relies on the first BIC in the chain. It represents the weakest link, whose failure would render the whole I/O system unavailable. The complexity of the I/O system and the parts obsolescence problem prompted the upgrade of the MCPI system. A new 2 Copyright 2009 by ASME 2 Copyright 2010 by ASME

3 board, ebic, was designed to replace the existing BIC and eliminate all intermediate hardware components between the SCS and the MCPI. In doing so, the hardware configuration is greatly simplified, resulting in higher system robustness, flexibility and hence higher reliability. The new MCPI hardware configuration is depicted in Fig. 3. Fig. 4: BIC Functional Block Diagram (CAE, 1980) Fig. 3: New MCPI Hardware Configuration (Chan, 2007) The new configuration possesses many changes; among them the most critical ones are as follows: Each I/O chassis is individually connected to the SCS via Ethernet. The whole VME chassis is eliminated. The two fan-out chassis are entirely eliminated. All intermediate circuit boards, such as the PCI-VME adapters, the DSA board, and the IOB are eliminated. 3. INTERFACE CARDS DESCRIPTION DBUS (P2), through which the BIC communicates with the host, the I/O cards and the IOB, respectively. For the DBUS, the BIC acts as a relay, passing on the data from the host to the IOB. For the CBUS, the BIC acts as a bus master, organizing the data flow between the host and the I/O function cards. A typical transfer cycle starts with an address (ADR) probe, followed by either a data request (DR) or data available (DA) probe from the host. The function card with the matching address will acknowledge the request and send or accept the data, provided that the parity check is satisfactory (CAE, 1980). The IOB performs the function of a buffer. It is equipped with a set of drivers and receivers to provide the electrical interface between the DBUS and DACBUS. It functions in two modes: relay and receive. In the former mode, the receiver is disabled, allowing data to flow directly from the DBUS to the drivers, though which the TTL data and control signals are converted to differential signals and passed on to the remote BIC. In the latter mode, the driver is disabled, permitting the data to flow in the reverse direction: from the DACBUS to the DBUS. The simplified data flow is shown in Fig. 5 (CAE, 1980). In this section, the detailed description of the following hardware is provided. For the existing architecture: The Bus Interface Controller (BIC). The Input/Output Buffer (IOB). The DSA board. For the new architecture: The Ethernet Bus Interface Controller (ebic). 3.1 The Existing Bus Interface Controller As shown in Fig. 4, the BIC is essentially a three-port device. These ports are formed by the DACBUS (P3), CBUS (P1) and Fig. 5: IOB Functional Block Diagram (CAE, 1980) 3 Copyright 2009 by ASME 3 Copyright 2010 by ASME

4 A simplified block diagram of the DSA is presented in Fig. 6. The DSA board is essentially a bus controller that interfaces two different buses, VME and DACBUS. The latter is a proprietary bus system developed by CAE. The DSA is responsible for the data transfer between the host computer and the MCPI, utilizing the Direct Memory Address (DMA) method. Its main component is a dual port RAM, which is mapped into host computer s VME address space. At start-up, as part of the initialization, the host downloads all valid DACBUS addresses to the address table of DSA residing in its RAM. During normal operation, the host initiates a transfer cycle, which includes a read and a write operation. During the write operation, the host sends a block of I/O data (including host memory address, DASBUS address and word count) to the DSA by writing directly to its data table (which also resides in the RAM). The DSA then writes the data out to the MCPI. During a read operation, the host initiates a data request to the DSA, which then scans the MCPI and stores the data in its data table. The host then reads the data directly from there. The DSA can operate in two modes: free run, where it continually cycles through each point in its address table, writing to or reading from the MCPI, and trigger, where it scans the MCPI whenever triggered by the host. The timing requirement for a transfer cycle is 50 milliseconds. Any errors encountered during the transfer cycles (e.g. address time out) are logged into a serial First In First Out (FIFO) memory, where they can be read back to the host for diagnostic. Fig. 7: ebic Functional Block Diagram (Chan, 2007) The microcontroller is a 16-bit, 40 MIPS digital signal controller, with up to 85 programmable digital I/O pins. The controller operates at 3.3V and also provides a built-in A/D converter. The C-Bus drivers and receivers provide the necessary electrical interface between the controller and the function cards. The scanning of the function card is done via the C-bus. The Ethernet controller is a stand-alone IEEE compatible controller, with Serial Peripheral Interface (SPI) via which it communicates with microcontroller. The EPROM is the storage area for an HTTP server though which on-line diagnostic is provided. The new Ethernet based interface controller has been successfully deployed in the simulator. 4. CONCLUSION In conclusion, the design of a new ebic board in the I/O system of a nuclear power plant simulator is presented. The new design possesses many outstanding features. It eliminates all intermediate hardware required by the old system, such as the PCI-VME, the DSA, and the IOB circuit boards. Therefore, it greatly simplifies the hardware architecture. The end results are the reduced cost and the higher maintainability. The new Ethernet based interface controller has been successfully deployed in the simulator. Fig. 6: DSA Functional Block Diagram 3.2 The New Bus Interface Controller As described above, for the old system, the data flow from the host to the function card goes through several bus systems, namely the PCI, VME, DACBUS, DBUS, and CBUS. This makes the control and interfacing rather complicated. The ebic, by contrast, implemented using the ModBus TCP/IP protocol, is far simpler. Its key components are, as illustrated in Fig. 7, the microcontroller, the Ethernet controller, the data and control bi-directional drivers, plus the address drivers (Chan, 2007). REFERENCES [1] Brown, R. and Basso, R., Advanced CANDU Reactor Distributed Control System Design, 25th Annual Conference of the Canadian Nuclear Society, [2] CAE,1980. The Bus Interface Controller Specifications. [3] Chan, Y. W., Ethernet Bus Interface Controller Design Specifications, SSD, OPG. [4] Choi, J.Y., Evaluation of Ethernet Based Control Network Used in the Distributed Control System, 15th Control Information Systems Laboratory (CISL) Winter Workshop, Copyright 2009 by ASME 4 Copyright 2010 by ASME

5 [5] Hespanha, J.P., Naghshtabrizi, P. and Xu, Y., A Survey of Recent Results in Networked Control Systems, Proceedings of the IEEE, Vol. 95, Issue 1, 2007, pp [6] Kim, H.S. etc., Design of Networks for Distributed Digital Control Systems in Nuclear Power Plants. International Topical Meeting on Nuclear Plant Instrumentation, Controls, and Human-Machine Interface Technologies [7] Lee, S. and Gwak, K., Analysis of a Communication Network for Control Systems in Nuclear Power Plants and a Case Study. ICCAS. [8] Lian, F., Moyne, J.R. and Tilbury, D.M., 2001, Performance Evaluation of Control Networks: Ethernet, ControlNet, and DeviceNet, IEEE Control Systems Magazine, Vol. 21, Issue 1, 2001, pp [9] Park, T.R. etc., Implementation of PICNET+ as the Control Network of the Distributed Control System for the Nuclear Power Plant, International Topical Meeting on Nuclear Plant Instrumentation, Controls and Human- Machine Interface Technologies. 5 Copyright 2009 by ASME 5 Copyright 2010 by ASME

RELIABILITY EVALUATION OF THE TWO BUS INTERFACE CONTROLLERS IN THE DARLINGTON SIMULATOR

THE UNIVERSITY OF ONTARIO INSTITUTE OF TECHNOLOGY FACULTY OF ENGINEERING AND APPLIED SCIENCE RELIABILITY EVALUATION OF THE TWO BUS INTERFACE CONTROLLERS IN THE DARLINGTON SIMULATOR Dong K. Le ENGR 5002G