Central unit PM 640 and PM 644: (supports SC 610)

 Performance, in this thesis, refers to the capability of a communication component in means of speed and throughput. • Availability is the term for the probability that a system will perform its specified functions when used under stated conditions. A common mathematical definition of operational availability is Ao = MT BF/(MT BF + MDT), whereas MTBF is the “mean time between failure” and MDT the “mean down time” [2]. However, in this thesis, availability is used in a more general manner, since the basis for mathematical operations is not available.

Abstract In order to modernize their infrastructure and keep up with the state of the art, ABB Power Systems decided to replace the older controller AC450 with a new generation of controllers called AC800M. Just like its predecessor, its main task is to work as a sequencer in an otherwise mostly unchanging topology. Although the new controller AC800M provides modern communication features and a sophisticated application development system, it lacks of a communication interface compatible with the residing controllers AC160. A hardware approach addressing this problem is in development, but not available at this point of time. Thus the decision was made to realize the connection using OPC, a widely spread and open software communication interface standard with a high potential of reusability. In addition, it was aimed at gaining additional knowledge about the OPC interface, which is commonly used in industry. In this thesis, we evaluate adequate hardware and software to realize this connection and we have programmed the controllers with applications to evaluate its performance and integrity. In addition, we are making considerations about redundancy that is vital in automation business in order to increase reliability and availability. We have shown that it is possible to interconnect controllers using OPC with satisfactory average performance results. Due to high maximum round trip times and high complexity when realizing redundancy, it is recommended to use such a system for testing purposes or non-critical operational applications, but not for critical systems. In this thesis we also identify and judge several alternative ways of connection.

Acknowledgements First of all, I would like to thank Prof. Dr. Bernhard Plattner of the Computer Engineering and Networks Laboratory TIK at the Swiss Federal Institute of Technology ETH Zurich for supporting and supervising this Master’s Thesis. Special thanks go to my advisors Rainer Baumann and Bernhard Tellenbach of TIK for their straightforward and helpful support during my work. Secondly I would like to thank Dr. Esther Gelle and Pascal Aebli of ABB for enabling this Master’s Thesis in the first place as well as providing aid throughout this thesis. Special thanks also to Stephan Egli for supporting me with the first steps, Swen Woldau for numerous hints concerning AC160 and AF100 as well as Juerg Beck for AC800M tips and tricks. Finally I would like to thank everyone else at PSPD for the provided aid and making my work so convenient not only in a technical but also in a human manner. Baden, June 2007 Martin Pfister

Chapter 1 Introduction This chapter will provide a rough overview of the problem treated by this Master’s Thesis. All technical devices and expressions will be explained more precisely in the next chapter. Please note that since this is a public thesis, it does not contain sensitive company-internal data. 1.1 ABB Power Systems ABB Power Systems is one of the world’s leading providers of infrastructure for controlling combined cycle power stations and waste-to-energy plants. Such a plant control infrastructure includes several hardware parts consisting of controllers, input/output-boards and communication devices as well as many software components to engineer, run, observe and analyze the power plant. A power plant control system has to satisfy a broad variety of different needs, from the efficient and reliable control of the turbines and associated supporting functions (such as lube oil) to easy configuration and operation as well as to sophisticated analysis functions addressing technical and economical aspects. 1.2 Problem Statement Due to high investment costs, the technical management of power plants is a slowgoing business with long life-cycles. Thus, a considerable amount of hardware devices currently in use are tens of years old. For future applications within ABB Power Systems it will be necessary to connect a controller of the newest series used within ABB, Control IT AC800M, with an older controller of the type Advant Controller 160 (AC160). The problem is that these two controllers do not share a fast communication interface of similar type and therefore cannot communicate directly. The standard communication intended for AC160 is Advant Fieldbus 100 (AF100). However, AC800M can support a whole range of buses except for AF100. As a consequence, the communication must be implemented using some relaying technique.

1.2.1 The Use of OPC It was decided in advance to realize the relaying connection using OPC. This solution was chosen because OPC is an open standard and very common in process and automation industry. Furthermore, this solution offers a high potential to be used for similar problems, since a lot of devices support this specification. However, OPC is normally not used for fast controller-to-controller communication but for slower visualization and logging purposes and there is no performance data available for this kind of connection. The use of OPC is therefore both challenging as well as interesting to gain more insights and know-how. It is also to mention that a hardware solution addressing our problem is not available yet. It is therefore necessary to have an alternative way using already available parts, also for testing purposes. 1.3 Goals The goals of this Master’s Thesis are stated as follows: • Setup and evaluation of a test environment • Setup of test systems • Theoretical and practical evaluation of the test systems concerning performance, availability and reliability. • Identification of improvements and different approaches • Comparison with alternatives As a starting point for the performance requirements, the current implementation was taken. The corresponding quantity and type of variables are displayed in Table 1.1 with 32-bit floating point values (floats) as analog in- and outputs and 1-bit boolean values as so-called status and command bits. In the current configuration with AC450 and AC160, all variables are written to the AF100 fieldbus with a cycle time of 256 milliseconds. Therefore we determined the minimum requirement for round-trip times from one controller to the other to exactly this time. In agreement with the advisors, instead of elaborating the optional extension stated in the task description (Appendix C), we spent more time on trying out a