Versioning-based service-oriented model for multi ...

Computers and Electrical Engineering xxx (2012) xxx–xxx

Contents lists available at SciVerse ScienceDirect

Computers and Electrical Engineering journal homepage: www.elsevier.com/locate/compeleceng

Versioning-based service-oriented model for multi-area power system online economic load dispatch q K. Nithiyananthan a,⇑, V. Ramachandran b a b

Birla Institute of Technology and Science, Pilani Dubai Campus, Dubai International Academic City, Dubai, United Arab Emirates National Institute of Technology, Nagaland, Dimapur, India

a r t i c l e

i n f o

Article history: Received 9 June 2011 Received in revised form 24 September 2012 Accepted 24 September 2012 Available online xxxx

a b s t r a c t The main objective of this paper is to construct a distributed environment through which the economic load dispatch solutions of multi-area power systems can be monitored and controlled. A single-server/multi-client architecture which enables the neighboring power systems to access the remote economic load dispatch (ELD) server at any time with their respective data and to get the economic load dispatch solutions from the remote server has been proposed. A .NET remoting architectural model has been implemented in such a way that for every specific period of time, the remote server obtains the system data simultaneously from the neighboring power systems which are the clients registered with the remote ELD server and the optimized economic load dispatch solutions from the server have been sent back to the respective clients. The ELD server creates a new thread of control for every client’s request and hence complete distributed environment is exploited. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction The economic dispatch problem of power systems is to determine the optimal combination of power outputs for the all the generating units, which minimizes the total fuel cost while satisfying the constraints. The economic load dispatch optimizes a power system operating objective functions (such as the operating cost) while satisfying a set of system operating constraints, including constraints dictated by the power system network. In its most general formulation, the optimization of economic load dispatch is a nonlinear, non-convex, large-scale, static optimization problem with both continuous and discrete control variables. Even in the absence of non-convex unit operating cost functions, unit prohibited operating zones, and discrete control variables, the optimal power flow problem is non-convex due to the existence of the nonlinear ac power flow equality constraints. The presence of discrete control variables, such as switchable shunt devices, transformer tap positions, and phase shifters, further complicates the problem solution. The Power System economic load dispatch solution obtained through conventional client server architecture is complicated, memory management is difficult, source code is bulky, and exception handling mechanism is not so easy. In the conventional power system operation and control, it is assumed that the information required for monitoring and controlling of power systems is centrally available and all computations are to be done sequentially at a single location [1]. With respect to sequential computation, the server has to be loaded every time for each client’s request and the time taken to deliver the economic load dispatch solution is also comparatively high [2–4]. This paper outlines a new approach to develop a solution for economic load dispatch analysis by the way of distributed computing. Among distributed object-oriented software, .NET remoting provides a language layer of abstractions for performing parallel and distributed computing in .NET environments.

q

Reviews processed and approved for publication by Editor-in-Chief Dr. Manu Malek.

⇑ Corresponding author.

E-mail address: [email protected] (K. Nithiyananthan). 0045-7906/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.compeleceng.2012.09.016

Please cite this article in press as: Nithiyananthan K, Ramachandran V. Versioning-based service-oriented model for multi-area power system online economic load dispatch. Comput Electr Eng (2012), http://dx.doi.org/10.1016/j.compeleceng.2012.09.016

2

K. Nithiyananthan, V. Ramachandran / Computers and Electrical Engineering xxx (2012) xxx–xxx

A .NET remoting architecture provides a framework that allows objects to interact with each other across the boundaries [5]. The high adaptability of the communication layer, the different object activation patterns, garbage collection without additional communication, and the advanced security features are advantages of .NET remoting [6]. .NET remoting client–server architecture overcomes the difficulties associated with sequential computation and it can be easily implemented. Remoting provides a powerful and easy way to the power system clients to communicate with remote ELD server at different domains. Even though good number of research papers available on distributed applications on .NET remoting on other engineering applications [7] only very limited research work has been carried out on distributed environment modeling on online economic dispatch for multi –area power systems. The proposed model supports many different communications protocols, including the Simple Object Access Protocol (SOAP)/Hypertext transfer Protocol (HTTP) used by Web based power system applications. Since the .NET framework provides a number of services, including activation and lifetime support, as well as communication channels responsible for transporting messages to and from remote applications. Remoting uses built in security by providing a number of hooks, which allow channel sinks to gain access to the power system data in the form of serialized stream before it is transported over the channel and hence the distributed economic load dispatch through this proposed architecture will itself a major contribution to power system online applications. In addition the proposed model secures the inbuilt safety and security of the ELD server as well as power system data. In this paper the proposed model has been compared with Remote Method Invocation (RMI) based economic load dispatch model [8] and performances with respect to round trip time & aggregate throughput have been analyzed in detail. The rest of this paper is organized as follows. Section 2 presents the frameworks for automated economic load dispatch in .NET remoting environment. Section 3 presents automated economic load dispatch algorithm. Results and analysis are presented in Section 4 followed by the conclusion of the paper in Section 5.

2. .NET remoting architecture for automated economic load dispatch solutions [9] In this present work, a distributed environment has been set up using .NET remoting to estimate and to monitor economic load dispatch solutions for different subsystems of an integrated power system. Each sub system has been considered as a power system client and hence multi power system clients – single ELD server model is implemented. In this model the tie line power flow for each area is assumed to be constant and tie line power flow is treated as load for each sub system. These

Power Systems Economic Dispatch Request

Proxy

CLIENT CHANNEL Formatter Sink

Client 1

Proxy Client 2

Proxy Economic Dispatch Result

Transport Sink

Client n

Power system data in XML form

SERVER CHANNEL Formatter Sink Economic Load Dispatch Server Economic Dispatch Request

Economic Dispatch Result

Transport Sink

Fig. 1. .NET remoting architecture for economic load dispatch.



3

power system clients are interconnected with an ELD server as shown in Fig. 1. When there is a call for a method located in a remote object, Fig. 1 shows the flow of .NET REMOTING. A client computer basically does the distributed power system monitoring through a Graphical User Interface (GUI) for every specific period of time and frequently exchanges data with the server. The server does the economic load dispatch computation and then distributes the results. Chronologically the server process should be started first, so that it can take the initiative to set up a connection link. It then starts waiting till it receives a connection request from the client. A client can register itself with the remote object (server object), just by invoking the registration procedure on the server object, when it needs a service from it. The remote object obtains the necessary data from the registered client objects and responds back to them respectively with the results. This total process can be automated by making the server get the input data for every specific period of time. Transaction of data among clients and server takes place several times and so the possibilities of the occurrence of errors may be high. Hence it must be handled properly. 2.1. .NET remoting data flow model In the proposed model, each neighboring power system is considered as a client remote object and it creates a new instance of the economic load dispatch server. The proposed remoting system creates a proxy object that represents the class and returns to the client object a reference to the proxy. When a client calls a method on an object that represents the remote object, the remoting infrastructure fields the call and it checks the type information and sends the call over the channel to the server process. Channel is a type that packages the power system data into a block of bytes that can be communicated through the as HTTP protocol. The HTTP channel transports messages to and from remote objects using the SOAP protocol. All messages are passed through the SOAP formatter [10], where the message is changed into Extensible Markup Language (XML) and serialized, and the required SOAP headers are added to the stream. This process of marshalling presents the entire economic load dispatch data in a suitable format for transporting one system to another. When the channel sends the information to the economic load dispatch server a listening channel in the server side picks up the request and unmarshals the parameters. This receiver locates the object to be called and then calls the desired method with those parameters. When the method returns, server remoting system captures, marshals the return value and sends the marshalled economic load dispatch results as packets onto a marshal stream and thus sends the results to the client channel. The client remoting system returns the results of the call to the client object. This data flow model is shown in Fig. 2. 2.2. Economic load dispatch server’s side-by-side versioning service [11] Maintaining large applications like economic load dispatch across different versions is a complex problem. This is especially true when the ELD server component under goes significant changes between versions. .NET remoting solves these problems by versioning assemblies, thereby allowing applications to run only with the versions they were built and tested. In this proposed method, the economic load dispatch server made to run economic dispatch service side-by-side using different versions of the Common Language Runtime (CLR) [12]. Since proposed model is online analysis for economic load dispatch optimization problem is formulated through classical method. Classical method is simpler and faster compare to other optimization techniques such as genetic algorithm and ant colony optimization. 2.2.1. Implementation procedure of the proposed model Many of today’s power system applications employ distributed object technology for scalability and reliability. Remoting is a new technology supporting this goal, allowing objects to communicate across application domains with minimal overhead. The remoting architecture abstracts as much of the underlying communications plumbing associated with distributed

Remoting System

Remoting System

Proxy

CHANNEL

Power System Client Object

ELD Server Object

Fig. 2. Remoting data flow model.


4


Fig. 3. Basic remoting architecture for economic load dispatch.

Fig. 4. Implementation of ELD server object.

computing. It is easy to setup communication between remote power system distributed objects and communicate as if they were in the same process space. Remoting can be approached at many different levels. One of the significant differences between remoting and other distributed object technologies is its architectural extensibility. The remoting architecture is very flexible, enabling the capability to extend power system application by adding custom components that participate in the communication process. An integral feature of remoting is lifetime management of remote economic load dispatch server components through leases. A power system client communicates with a lease manager to control the life time of these components. At its most basic level, remoting is the capability to communicate with components in separate AppDomains. Fig. 3 shows a simplified view of two objects communicating via remoting a power system client component in AppDomain A communicates with an economic load dispatch server component in AppDomain B. The application domain is an execution environment within the remoting architecture. It provides major benefits of reliability and security of power system data communication. Since remoting supports multitier architecture as well as distributed architectures, the server component in AppDomain B could easily be extended to any power system client to service the economic load dispatch. Furthermore, a remote component could be located in another process on either the same or a different machine. Once an ELD component is set up, the underlying plumbing required to maintain remote communication is hidden by the remoting architecture.

2.2.2. Remoting economic load dispatch server A remoting ELD server object is an instance of a class that inherits from MarshalByRefObject. Fig. 4 shows the implementation of the remote ELD server object. The EldServer class in Fig. 4 inherits MarshallByRefObject, which supports the basic functionality of a callable ELD component over the proposed remoting architecture. This class is implemented as a DynamicLink Library (DLL), negating the need for a Main() method in the economic load dispatch server. The only method in this class is the calculateEconomicLoadDispatch() which receives the economic load dispatch data as a string argument and calculates the economic load dispatch data. The calculateEconomicLoadDispatch() method sends the economic load dispatch results to the power system client as a single string. The remote ELD component requires a configuration file named web.config to specify necessary operating parameters. The web.config file for the remote ELD server component is shown in Fig. 5. The web configuration file shown above is located in the same directory in which the ELD server implementation is located. It contains a special section for remoting, marked by a tag name after the remoting namespace, . Further down the hierarchy is an tag, containing a tag. The section has a tag with three attributes such as mode, type and objectUri that assist in making the remote ELD object to run properly (see Fig. 6). The attribute ‘‘mode’’ has two possible values: SingleCall and Singleton. These values identify basic life time issues associated with a remote economic load dispatch server object. SingleCall components are activated and live for the duration of a single call from the remote client. When a remote power system client calls the calculateEconomicLoadDispatch() method of Please cite this article in press as: Nithiyananthan K, Ramachandran V. Versioning-based service-oriented model for multi-area power system online economic load dispatch. Comput Electr Eng (2012), http://dx.doi.org/10.1016/j.compeleceng.2012.09.016


5

Fig. 5. web.config file for ELD server object.

Fig. 6. GUI with economic load dispatch solution.

the EldServer class, a new server object is instantiated when the call begins. When the call ends, the server object is destroyed. Furthermore, each client receives a reference to a unique server object whenever it makes a call. The type attribute of the tag has a quoted pair of values that identify the remote ELD server object. The first value is the fully qualified name of the server class. The second value is the executable filename of the assembly to which the server class belongs to. The last attribute of the tag is the objectUri, which holds the Universal Resource Identifier (URI) of the server component. 2.2.3. Step by step procedure for versioning service The steps involved in side by side versioning of economic load dispatch server as follows: i. ELD server component has given a version number as part of its identity. The version number is stored in the assembly manifest along with other identity information like relationships and identities of other assemblies connected with the application. ii. In the absence of a policy file in the assembly, the economic dispatch server should continue to work with the CLR version they were built with. If a policy file is present specifying that two different versions of an assembly like System.Runtime.Remoting are compatible, fusion will automatically load the latest compatible version. It is therefore important to ensure that only strong assembly names are used in configuration files when referring to CLR assemblies.


6


iii. Economic load dispatch application made to run side-by-side by using strong names when calling Assembly.Load or when specifying power system clients transport sink providers. Any changes or upgradation of the economic load dispatch solutions of the server side results in the side-by-side versioning service and upgraded version is made available for clients. 3. .NET remoting based economic load dispatch monitoring algorithm Each client object registers with the ELD server component. The server uses the remote client reference to invoke its method to obtain the economic load dispatch data and then provides the service through its methods. Both client and server objects are considered as remote objects and this is how inter remote object communication is achieved. The server object uses a single thread of control to distribute the economic load dispatch solution simultaneously to the clients registered with it. The proposed model is dynamic which allows a new power system client to register with the ELD server object at runtime and to get serviced. A kind of multicasting has been achieved through this multi client-single server model. Economic load dispatch server and clients have to store in them, the necessary object codes required for economic load dispatch calculations. Subsequently, the following steps are to be carried out. i. ii. iii. iv. v. vi. vii.

Start economic load dispatch server. Economic load dispatch server should creates its own proxy object and run the proxy object at specific port number. Start power system client by loading the server side DLL in the assembly. Client registers with the server by invoking the appropriate method at the remote object. Server uses the client’s reference to receive the power system data from the client. Server computes the economic load dispatch result and returns it to the client. Client obtains the result from the server through server side HTTP channel and provides a view of the result through a specified object. viii. For every specific period of time, server automatically receives system data from the client, thereby providing an automatic economic load dispatch monitoring. 4. Results The above distributed algorithm has been implemented in Windows NT based Hewlett–Packard workstations connected in an Ethernet Local Area Network. The results are shown in a GUI as given in Fig. 3. The above GUI shows the economic load dispatch solution for a specific three generator bus power system client. When each power system client is loaded, it registers with the economic load dispatch server due that the intercommunication between server and client’s remotable object is achieved and the client sends the request and receives the output. Using this approach, different power system clients can monitor continuous updated optimized economic load dispatch solutions at regular time intervals [13]. The major factor that influences the performance of the proposed models is the round trip time (RTT) that includes the convergence time. The round trip time measures the time needed from the point when the power system client initiates a method invocation to the point when the client receives the results. The round trip time is measured for all the power system clients that invoked the load flow method simultaneously without any delay. The performance analysis of the proposed distributed models has been carried out with respect to economic load dispatch and the variations of round trip time with respect to the number of clients are shown in Fig. 7. From the above graph, it is very much evident that .NET based model running in CLR out performs the conventional client/ server architecture based on socket programming. RMI model [7] performs slightly better than .NET model with less number of power system clients connected with ELD server. As the number of power system clients increasing RTT difference

Fig. 7. RTT vs. no of clients (Economic Load Dispatch).



7

Fig. 8. Aggregate throughput vs. no. of power system clients.

between RMI and .NET models are very less. Due do inherent advantages of versioning, life time activation and inbuilt security features of the proposed model scores high against the RMI based distributed model. Along with the performance analysis aggregate throughput of the proposed distributed models has been carried out with respect to economic load dispatch and the variations of round trip time with respect to the number of clients are shown in Fig. 8. From the above graph it is evident that as the No. of power system clients increases, overall throughput is increasing for all distributed models. However increase in throughput is not even. If no of clients increases which in turn increases the large objects occupies the network and slows down the performance. But both RMI and Remoting architecture are automatically allocating some extra memory to accommodate large objects. As the result if the number of power system clients increases client/server model shows considerable reduction in throughput. In the proposed model throughput increases and almost reaching more than hundred Mbps. It is evident that from Fig. 8 .NET remoting occupies less bandwidth and better than RMI based model. 5. Conclusions An effective .NET based distributed model has been developed to monitor the economic load dispatch of multiple power systems. It has been tried out in overcoming the overheads associated with sequential power system economic load dispatch computation through this model. Although, client–server architecture for economic load dispatch solution is very well established, the value of this study lies in that it emphasizes a unique methodology based on .NET based model to serve a large number of clients in a distributed power system environment, across various platforms based on communication between Common Language Runtime. A practical implementation of this approach suggested in this paper was assessed based on 6, 9, 10 and 13 bus sample systems and performance analysis has been discussed. Accordingly the proposed model can be implemented for large power systems network spread over geographically apart. References [1] Wood AJ, Woollenberg BF. Power generation, operation and control. NewYork: John Wiley & Sons; 1996. [2] Bandyopandhyay G, Senguptha I, Saha TN. Use of client-server model in Power System load flow computation. IE(I) J Electr 1999;79:199–203. [3] Hart David G, Uy David, Northcote-Green James, Laplace Carl, Novosel Damir. Automated solutions for distribution feeders. IEEE Comput Appl Power 2000;13(4):25–30. [4] Qiu B, Gooi HB. Web based SCADA Display Systems (WSDS) for access via internet. IEEE Trans Power Syst 2000;15(2):681–6. [5] Chen Chung-Kai, Chang Yu-Hao, Chen Cheng-Wei, Chen Yu-Tin, Yang Chih-Chieh, Lee Jenq-Kuen. Efficient switching supports of distributed .NET remoting with network processors. In: Proceedings of international conference on parallel processing; May 2005. p 190–95. [6] Fengqing Zhang, Quancun Kong. Distributed remote control system of UAV based on man-in-loop real-time operation. In: Proceedings of the 26th chinese control conference. Hunan, China; July 2007. p 119–22. [7] Nithiyananthan K, Ramachandran V. RMI based distributed model for multi-area power system on-line economic load dispatch. J Electr Eng Romania, Europe 2005;56(1–2):41–4. [8] Nithiyananthan K, Ramachandran V, Dhamodharan D. RMI based multi – area power system load flow monitoring. In: IASTED international conference, power and energy systems in Rhodes, Greece; July 3–6 2001. p. 249–52. [9] Jianxing Zhou, Xinguo⁄ Qiu, Wei nANG, Xu Lu. Research of detecting system of Tower Crane based on.NET Environment. In: Proceedings of international conference on electrical and control engineering. IEEE Computer Society; June 2010. p. 1598–601. [10] Yorino Naoto, Hafiz Habibuddin M, Yutaka Sasak M, Zoka Yoshifumi. High-speed real-time dynamic economic load dispatch. IEEE Trans Power Syst 2012;27(2):621–30. [11] http://msdn.microsoft.com/en-us/library/2e7z38xb%28v=VS.71%29.aspx.


8


[12] http://msdn.microsoft.com/library/vstudio/z1zx9t92. [13] Aloisio G, Bochicchio MA, La Scala M, Sbrizzai R. A distributed computing approach for real-time transient stability analysis. IEEE Trans Power Syst 1997;12(2):981–7. K. Nithiyananthan received his Ph.D. in Faculty of Electrical Engineering from College of Engineering, Guindy, Anna University, Chennai, India in 2004. He is currently working as Associate Professor in the Department of Electrical and Electronics Engineering, BITS, Pilani Dubai campus, United Arab Emirates. His research interests include computer applications to power systems engineering, Internet technologies and embedded systems. V. Ramachandran received his M.E. and Ph.D. in Faculty of Electrical Engineering from College of Engineering, Guindy, Anna University, Chennai, India. He is currently working as Director, National Institute of Technology, Nagaland, India. His research interests include power systems reliability engineering, network security, component technologies and soft computing.
