throughout and heavy computational work in computer science, grid ... tool in electrical engineering field. Because .... Moreover different kinds of job: batch jobs,.
12th WSEAS International Conference on SYSTEMS, Heraklion, Greece, July 22-24, 2008
Integrating Grid Computing Technology for Developing Power Systems Reliability and Efficiency R. Al-Khannak, L. Ye Electrical and Automation Engineering Department South Westphalia University of Applied Sciences, Campus Soest Luebecker Ring 2, D-59494 Soest GERMANY
Abstract: - In modern society, energy production is one of the most critical issues. In order to generate cleaner, more reliable and ecological friendly energy, integrating of the whole power system which disperse in a wide area is very important. By integration of such geographically dispersing power system as a single substance, huge database and corresponding data/computational intensive processing is of necessary. For huge data throughout and heavy computational work in computer science, grid computing technology offers an alternative solution. Therefore corresponding research about integrating power plant control system into grid computing technology is started in resent years. In this paper, this kind of integration is described, analyzed and simulated. This paper starts with introducing state of arts of grid computing technology and explicating middleware used in this case. Application used in power system is real time critical applications. A real time data processing application is designed for power system control system. An innovative reliable backup system is also developed with grid computing technology and shows its advantages in comparison with conventional backup system.
Key-Words: - grid computing, grid computing middleware, resource broker, redundant system, execution host. computations and information to be transferred between power systems’ nodes. Within all these changes normal and conventional computing systems become unable to do achieve these huge dataset and complex computations. So, modern power system needs more and more computational power. Grid computing technology which can optimize the utilization of available resources insider grid network can therefore provide a more efficient and cheaper way for power system in comparison with single super computer. Modern electrical grids are computer enabled power networks that provide efficient and smooth management, monitoring and information exchange of distributed power networks with diverse widespread resources of power generation [5]. The electrical grid will be transformed from central generation down to collaborative networks that will incorporate customer appliances and equipment and modern information devices in the distribution. The future electrical grids will consist of large small-scale generation units of renewable energy sources and other disparate energy sources. Highly scalable and decentralized integrated communication, computing and power networks will be necessary to monitor future smart grids [3].
1 Introduction Modern power systems’ scenario shows major changes, compared with traditional and existed systems. Due to the big changes in global weather conditions and specially the increment in heat temperature (glob warming), and oil prices increase as well, power load demands increased rapidly which forced power engineers and scientists to look to new solutions to handle and balance the new situation. Green and clean power generation get the highest prior attention by integrating many small modern “clean” power generator units as renewable resources like wind turbine, photovoltaic, solar energy, hydrogen, …etc. into tradition power generation systems. This integration brings a lot of difficulties and complexity due to the amount of information to be exchanged and computed for systems controlling, monitoring and scheduling. On the other hand power provider/consumer scenario becomes different. That in the current new scenario both of them are playing dynamic, exchangeable role in the power generation environment. Power consumers are playing some times as power provider especially with renewable power sources. This scene shows a lot of
ISBN: 978-960-6766-83-1
491
ISSN: 1790-2769
12th WSEAS International Conference on SYSTEMS, Heraklion, Greece, July 22-24, 2008
2 Computational and Power Grid
4 Modern Power Systems
From the basic understanding of power electrical grid that whenever any kind of generator is connected to the grid, means this generator is ready to feed the grid with the required or demanded power. This system is working in a dynamic mode that generators are joining and leaving the grid in a specific scheduling and mechanism. In the same how, grid computing came analogy following the same scenario that in the computational grid any computing resource once connected to the grid (grid computing network) is ready to support the computational grid with demanded computing power or storage facility. It is represented as a dynamic and heterogeneous computational system which may include machines from different specification and platforms.
Application of suitable Information Technologies such as GC will enable active and collaborative participation of consumers, utilities and third parties in energy markets. This will help to stabilize the prices and to be a key for shaping future power market [13]. In modern power system, the position of energy seller and customer is changeable. Seller and customer exchange their positions with market condition changing. New power organization behaviors as a totally dynamic system. They will allow the participants to exchange their positions and activities when outside conditions are changing. This kind of power system require high performance of controlling system that processing data from each partners to ensure quality of services for all partners. From figure 1, it illustrates power grid virtual organization system including all participants in one cluster.
3 GC and Power Engineering Grid Computing become a powerful and important tool in electrical engineering field. Because of increasing complexity of modern power system and more geographical remote disperse power generation, these days power system need more and more computational capability to processing data and abstracting useful massage for monitoring system and optimal management. Therefore grid computing can be an alternative/better solution in the future, to help modern power system working in an efficient way. The control and operations of modern power
systems are becoming data-intensive, information-intensive, communication-intensive and computation-intensive due to much of the reliance of these power systems on computerized communications and control. [11]
Fig 1: Power grid virtual organization
5 Power System Conversion Modern power systems in general are changing into larger numbers of smaller-scale highly dispersed generation's units that use both current generation technologies and renewable energy sources to reduce carbon dioxide emissions [1,3]. These modern systems need to be connected into modern facilitated network to enable transferring, monitoring and controlling their situation within high speed techniques [1, 15]. While generators are connected into the power system, output that generated by these devices need to analyze, monitor and control with advance scheduling. Transmission level control technology at current time is not suitable to such new power system with huge numerous small scale generators. So, GC is the technology which is capable of providing a relatively inexpensive new solution, allowing the output of embedded generators to be analyzed, monitored and when necessary controlled [16].
Therefore, since Grid Computing is considered as an inexpensive means for “super-computing” for dealing with heterogeneous and distributed computing [7], grid computing is an alternative choice for constructing huge data storage and high computational processing capability. On the other side, electrical power system continue to improve its technology in order to achieve more stable, accurate power supply, power load forecasting, better scheduling. All these achievements are very important part of electrical energy and depend on real time data processing. Hence, GC technology is seen as an alternative solution for electrical energy application with secured access, high data throughout and computational workload processing. Grid computing integration mechanism will be illustrated to achieve mentioned goals.
ISBN: 978-960-6766-83-1
492
ISSN: 1790-2769
12th WSEAS International Conference on SYSTEMS, Heraklion, Greece, July 22-24, 2008
proper resource to the proper task according to its priority, consuming time and hardware/software requirement. Unexpected peak load of activity in large organization is therefore handled effectively by the grid and smooth peak load. This invaluable feature of GC is realised through the process of profiling resource by its availability and capacity and advance scheduling/management of tasks.
Due to two major reasons, either insufficient machines performance for specifically technical requirement or inefficient utilization of machines; many obstacles limit the development in engineering applications and technology in power system. And with the advance of engineering technology, which depends on the performance of computers for solve problem, this kind of obstacles will become more obvious. Using more powerful computing technology is the efficient answer to overcome these obstacles.
7 GC Challenges / Problems Although Grid services transparently provide access to the entire collection of information processing system, development and deployment of applications still face great challenges. Application development remains a great barrier due to the heterogeneous nature of the underlying resources [19]. Moreover different kinds of job: batch jobs, scripts and executables are not easily transferable from one system to another by identical manner. For example, scripts job depends on installation of its script language in the distributed heterogeneous resource, which should be considered before deployment of grid application. Also executable file need to be recompiled in order to implement on different operation system. Grid Web services are possible solutions to overcome this barrier by taking over the program installation and deployment for individual users. Further more, programs language like java, which run on the java virtual machine and not depend on the actual software/hardware architecture of under layer resource, can also partially minimize this portability problem. One more factor for development of grids need be taken into account. Not all middleware are compatible with available toolkits. Due to this fact, concerning of software compatibility is very important for design grid application. Successful grid application is realized by proper selection of toolkit and middleware according to its specific requirement.
6 Why Grid Computing Development of GC is motivated by the potential benefits for the specific application. This technology has also its unique characteristics for the application development. Following entries are the major benefits and characteristics of GC technology. a) Exploiting idle Resources In most organisations, many computing resources are idle and underutilised at most of the times. For example, most desktop computers are idle more than 95% of their time [17]. This idle time is wasted and unprofitable for the owner and other organization which require high performance computer. Along with the idle time of processor, there is very often unused storage place in these computers. These two kinds of resources can be maximal used with GC technology. b) Parallel CPU Capacity Grid computing technology enables the possibility of applying massive parallel CPU activity. Massive parallel CUP is initially applied in super computer for scientific purposes, now is extended to other fields like financial modelling, oil exploration and motion picture animation. This extension of parallel computing appliance causes revolutionary working methodology in these fields. Although applying massive parallel CPU is really attractive for developer of grid computing application, there are still obstacles limiting appliance of this technology. A successful implementation of parallel computing depends on the characteristics of object application itself. If object application can be split into numerous subtasks with less interactive activities between each subtask, it will achieve better result and reduce the complexity of program design and failure possibility during implementation. c) Resource Balancing Grid computing integrates heterogeneous resources into one united virtual resource pool. In this virtual resource pool grid can assign the
ISBN: 978-960-6766-83-1
8
Proposed Model for Integrating GC with Power Applications
The architecture of integrating GC with power applications model is shown in Figure 2. This model has following features: a) The model infrastructure is integrating power systems with grid resources into a GC system. This system constructs a virtual resource pool and allows share processing power and data storage.
493
ISSN: 1790-2769
12th WSEAS International Conference on SYSTEMS, Heraklion, Greece, July 22-24, 2008
information, could assign best backup component to compensate that fault.
b) The power applications inside GC system can get the current status of all power applications within GC and interact with each other for optimal controlling. c) GC resource broker will find available resources which meet certain specifications for electrical energy application.
Fig 4: GC application with electrical power system In Figure 4 the GC-PS integrated system fault reaction mechanism is illustrated. Real time status massages are to broadcast inside whole system (GC and PS). All GC components will have real time information about system status. Resource broker is in charge of managing job assigning and decision making. For any failure grid computing system is responding to compensate the fault by send run job or activating existed schedule or most convenient power system after analysing available information.
Fig 2: Grid Computing Model Integrating with Power Applications
9
GC Implementation Mechanism in Electrical Energy Engineering
Both electrical and computational grids are located in a network and interact as one system. Name of such a system is not settled yet. It can be named as smart grids, modern power systems, micro grids, virtual power grid, and VPP virtual power plant. In figure 3, the architecture of integrating GC with power application, showing that during normal and stable power system running, GC system is sharing power system operations. [1, 2]
10 Power Application Backup System Reliable power system is the one of the most important requirement for any scientific and industrial research in power application field. Modern power systems are dealing with huge dataset due to its complexity. This feature makes reliability to risk ratio very critical. With the GC technology a innovative methodology of backup is developed in order to offer an alternative/better choice to conventional methodology. Traditional controlling - monitoring backup systems, depending on (2xN) methodology which means duplicated system, is applied in current power system. For example, if (N) operational power system components are installed, there should be additional (N) machines installed to wait for replacing any possible failure, see in figure 5. In such conventional backup systems many parameters have to be considered. system place size, cost and conditions, energy consumed by the whole system, machines cost, repairing cost, environmental issues …etc, all this aspects increasing the difficulty and complexity of design and installation such backup system.
Fig 3: Integration GC into electrical power system In this system, when any generator fails, GC system which are exchanging real time power system
ISBN: 978-960-6766-83-1
494
ISSN: 1790-2769
12th WSEAS International Conference on SYSTEMS, Heraklion, Greece, July 22-24, 2008
consists of SGE master host (SGE Administrator and information management), SGE shadow master (SGE master backup PC to increase reliability of SGE system) and various execution hosts located in different locations to process the computational jobs.
Fig 5: Conventional controlling/monitoring power system with backup system model
11 SWUGPS
Fig 7: SGE system integration into power system
In the South Westphalia University of Applied Sciences grid power system (SWUGPS) is set by integrating Power plant module and SGE grid computing system. The power plant module is a module of nuclear power plant ‘Mülheim Kärlich’ located in Germany. This nuclear power plant is a PWR (Pressured Water Reactor), which means heating water till around 315℃ by keeping water pressure of 15 MPa inside the primary loop of nuclear power plant. Power plant module simulates power generation with electrical heater and small scaled water pressurized in stead of nuclear reactor and water pressurized tower. The power system data streams to the grid computing resources via I/O device into to the server and database which are connected to the power system via Ethernet, as seen in figure6.
12 Developed reliable Backup System Methodology As discussed above, the conventional controlling/monitoring systems (as shown in figure 5) have many limitations (costs, place, energy consume, etc). So grid computing technology by SGE middleware is proposed as a new manner to achieve a more reliable, cheap and efficient power backup system. The developed methodology replaces the (2xN) traditional backup system scenario (duplicated backup system) by (N+R) GC scenario. The main goal of the new method is to reduce the limitation of conventional backup system by innovative GC method. Such limitation can be the backup system size, cost, host capacity, etc.
Fig 8: GC based controlling power system Figure 8 shows grid computing (SGE) mechanism. Within this system, SGE master is in charge of monitoring, controlling available resource inside grid network and jobs to assure system stable.
Fig 6: SWU grid power system Architecture Figure 7 shows that SGE grid computing system is constructed as computational resource pool for the power plant applications. This GC SGE system
ISBN: 978-960-6766-83-1
495
ISSN: 1790-2769
12th WSEAS International Conference on SYSTEMS, Heraklion, Greece, July 22-24, 2008
exe7. In the second screen (in the middle) shot job 835 is missed due to any problem happening with host exe7 (like network communication problem or HW or SW problem), job is pended till SGE system find one next available resource to execute interrupted job. In the bottom screen, job 835 has been rescheduled to exe3 (as available resource in virtual computational pool) to compensate exe7 breaking. And after solving the problem with previous broken host exe 7, this host exe7 can joint GC system again as a back up host to compensate next possible failure inside system. In this case no duplicated machines are reserved for backup purpose and system can still be stable even sudden failure happens.
When any failure (communication failure, host break down .etc) occurs in computational side, which is critical situation for all power systems implementations, grid computing system responds towards problem and compensates it by finding available resource inside grid and automatically reschedule jobs within seconds. In grid computing system, current status of all grid resources are monitored and gathered by grid system dynamically. And resource break finder and host configuration mechanism allows grid system find out or classify the most compatible resource for specific application. Therefore whenever failure occurs in any computational resource, grid computing system will resend missed job(s) or task(s) to the most applicable/compatible resource according its predefined hardware/software specification. From figure 9, as soon as any computational resources in virtual computational pool (white area) is broken, SGE master can automatically assess the status of grid network and is reassigning the interrupted job to the available backup system machines (dark area). Then this backup host can be seen as ordinary operational resource. With this innovative methodology, backup system can be more dynamic, reliable and efficient. And the cost, required space, consuming energy by traditional methodology can be reduced through (N+R) GC scenario.
Fig 10: SGE GUI shows GC reactive performance
13
Fig 9: GC based backup system SGE middleware is a grid computing platform which, it has an advanced graphical user interface (GUI) for dynamical controlling and monitoring the status of CG system. Figure 10 illustrates SGE GUI during the failure occurrence within CG system. Three different statuses are demonstrated in Figure 10. In the top one, two power applications are running within SGE system (with “Jobid” 834 and 835). They are being executed on client hosts exe5 and
ISBN: 978-960-6766-83-1
Conclusion
Grid Computing as a new distributing computing technology can group all the resource as a virtual computer resource pool and handle geographically distributed system or solve large scale data throughout or computational intensive work with utilizing normal/high performance pc instead of super computer or maximize idle computational resource for efficient operation to save cost/energy. Because of these characteristics of Grid Computing, it could be the best tool required along with future development in electrical power system by supporting trading activities, power marketing decision analyzing, power system controlling Future power grids will consists of massive numerous small scale and geographical disperse generators for the purpose of maximizing utilization of renewable energy. This characteristic necessarily require huge storage, information exchanging, high level automatic distribution systems and intelligent control to handle the huge dataset generated by whole power grid generator. For this challenge
496
ISSN: 1790-2769
12th WSEAS International Conference on SYSTEMS, Heraklion, Greece, July 22-24, 2008
electric power system applications. Power Engineering Society General Meeting, 2006. IEEE [10] “Smart Electric Grid of the Future: A National “Distributed Store-Gen.” Test Bed.” Retrieved October 04, 2006, from http://www.ldeo.columbia.edu/4d4/testbeds/Sm art-Grid-White-Paper.pdf [11] “Grid Computing.” Retrieved September 05, 2006, from http://en.wikipedia.org/wiki/Grid_computing [12] Global Environment Fund. (2005). “The Emerging Smart Grid.” Retrieved August 29, 2006, from http://www.globalenvironmentfund.com/GEF whitepaper_ElectricPowerGrid.pdf [13] Yini Chen, Chen Shen, Wei Zhang and Yonghua Song. Grid computing infrastructure for power systems. [14] Malcolm Irving and Gareth Taylor' Prospects for Grid-Computing in Future Power Networks, September 2003. [15] Features of the Java Commodity Grid Kit. Available from: http://aspen.ucs.indiana.edu/gce/C536javacog/c 536featuresOfCoG.pdf [16] Parvin Asadzadeh, Rajkumar Buyya, Chun Ling Kei, Deepa Nayar and Srikumar Venugopal, Global Grids and Software Toolkits: A Study of Four Grid Middleware Technologies. Available from: http://www.gridbus.org/papers/gmchapter.pdf [17] Gannon, D., Ananthakrishnan, R., Krishnan, S., Govindaraju, M., Ramakrishnan, L. and Slominski, A. “Grid Web Services and Application Factories.” Retrieved October 03, 2006, from http://www.extreme.indiana.edu/xcat/publicatio ns/AppFactory.pdf [18] D.Bevc, S.Zarantoneolla, N.Kaushik, I.Musat, “Grid Computing for Energy Exploration,” Minisymposium on Grid Computing for Oil and Gas Industry, SIAM Conference on Parallel Processing for Scientific Computing, San Francisco, February 2004. [19] S. Sheng, K.K. Li, Z. Xiangjun, “Grid Computing for Load Modeling,” IEEE International Conference on Electric Utility Deregulation, Restructuring, and Power Technologies, April 2004. http://ieeexplore.ieee.org/Xplore/login.jsp?url=/i el5/9285/29506/01338053.pdf?tp=&isnumber= &arnumber=1338053
many advanced new electrical technologies and information communications infrastructure will be developed to overcome obstacles on the way to future smart green power system. Grid computing is an innovative tool to help scientist and engineer in this field to achieve this goal. Grid computing facilitates the integration of current and future technologies in order to construct stable electrical grids consisting large number of heterogeneous generator and consumers. And it also can build the reliable power control system by its innovative backup methodology. Applying grid computing technology into power system will necessarily to reduce the difficulty of controlling and monitoring coming with integrating millions of small disperse generators into current power grid system.
References: [1] R. Al-Khannak, B. Bitzer. “Load Balancing for Distributed and Integrated Power Systems using Grid Computing”. ICCEP07, Capri, Italy May 2007. IEEE http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arn umber=4272368 [2] Taylor, G.A., Hobson, P.R., Huang, C., Kyberd, P. and Taylor, R.J. “Distributed monitoring & ontrol of future power systems via Grid computing.” Retrieved August 20, 2006, from http://people.brunel.ac.uk/~eesrgat/research/pdf _pubs/IEEEPES2006.pdf [3] Deak.O. (2005). “GRID SERVICE EXECUTION FOR JOPERA.” Retrieved October 05, 2006, from http://www.iks.ethz.ch/publications/files/Oliver _Deak_MT.pdf [4] Cascio, J. (2005). “Smart Grids, Grid Computing, and the New World of Energy.” Retrieved October 4, 2006, from http://www.worldchanging.com/archives/00215 2.html [5] Rajkumar Buyya and Srikumar Venugopal, CSIC communications, computer society of India' July 2005 [6] Foster, Ian and Carl Kesselman ed.: The Grid:Blueprint for a New Computing Infrastructure; Morgan Kaufmann; San Francisco; 1999. [7] Grid Computing Info Centre (GRID Infoware), http://www.gridcomputing.com [8] Joshy Joseph, Craig Fellenstein. Grid Computing, 2004. [9] Qi Huang, Kaiyu Qin and Wenyong Wang, Development of a Grid Computing platform for
ISBN: 978-960-6766-83-1
497
ISSN: 1790-2769
