Editorial
It gives me immence pleasure to publish second edition of referred journal which comprises all field of engineering Science. The contribution of Indians to the fields of Science, Mathematics, Technology and many others is well known. The ability of the Indian mind to deal with complex and abstract ideas contributes to their success. Extensive research work is carried out in all corners of the world and a Technical Journal is a medium which presents this work to the whole world. This precisely is the motive behind publishing this journal. Any innovation is a collaboration of ideas from all quarters. Knowledge sharing is the key purpose, keeping this View in mind JAES is having Partnership with PublishingIndia.com which provides the digitization of research and non-research information which serves important societal, cultural, managerial, scientific and technological development goals, including the preservation of effort made by Indian author. It stimulates new ideas and instigates a thinking mind to come up with new thoughts and approaches, which eventually contribute to the betterment of technology.
It is our aim to bring about exchange of ideas, through this journal, between the Indian researchers and researchers of other parts of the world, thus furthering the cause of research in the technical field. It is also our endeavour to inspire more number of engineering faculties to do research, by giving them an international platform for presenting their work.
JAES is a half yearly journal which publishes articles on a wide variety of subjects related to all branches of Engineering. A team of members work for successful completion of JAES, every paper get reviewed and evaluated by expert from industry or academic. For the convenience of readers and publisher entire journal is divided into four main sections A to D. Where section A is designed to have fields related to Electrical, Material Science and Engineering, Section B contains papers related to Mechanical, Production, Automobile and Industrial Engineering. Section C contains Electronics & Telecommunication, Computer Engineering, Whereas Section D deals with Civil Engineering.
I congratulate all the authors to contribute their research work to publish in this issue. I thank the entire reviewer and other members of editorial team for their tireless efforts to bring this issue in time.
Dr. D.K. Singh Chief Editor . Journal of Advances in Engineering Science A half yearly journal being published by Sinhgad Technical Education Society, Pune Pune Office: Mumbai Office: S. No. 44/1, Vadgaon (Bk) Geeta Building, 2nd Floor Off Sinhgad Road 392/6, Sion (East) Pune - 411041 (MS) India Mumbai - 400 022 (MS) India Tel: (020) 24357482 / 24355884 Tel.: (022) 24080119 / 20 Telefax: (020) 24354721 Fax : (022) 24080121 Email: [email protected] Website: www.sinhgad.edu PATRON Prof. M. N. Navale Dr. Mrs. S. M. Navale Founder President Secretary Sinhgad Technical Education Society, Pune Sinhgad Technical Education Society, Pune
ADVISORY BOARD Dr. Narendra S. Chaudhari Dr. Lech Kwiatkowski Associate Professor Corrosion Centre School of Computer Engg (SCE) Institute of Precision Mechanics Nanyang Technological University (NTU) Duchnica 3, Warsaw 01-796, Poland Singapore Dr. P. R. Kumar Dr. Pramod G. Chaphalkar Franklin W. Woeltge Professor Department of Mechanical Engineering Dept. of Electrical and Computer Engineering Grand Valley State University Research Professor Grand Rapids, MI 49504-6495, USA Coordinated Science Laboratory University of Illinois, Urbana-Champaign USA Dr Michael Blumenstein Head of School Dr. Morteza Marzjarani Professor, Computer Science and Information Systems School of Information and Communication Technology Department Gold Coast Campus, Griffith University SE179, Saginaw Valley State University Queensland 4222, Australia 7400 Bay Road, University Center , MI 48710, USA Prof. Dr. Bulent SANKUR Professor Graham Leedham Department of Electrical-Electronic Engineering Head, School of Science and Technology Bogazici University University of New England, Armidale Bebek, Istanbul, 34342 Turkey New South Wales 2351, Australia
EDITORIAL BOARD Publisher Chief Editor Dr. S. S. Inamdar Dr. D. K. Singh Sinhgad Institute of Technology Professor & Dean (R & D) Lonavala, Pune (MS), India Sinhgad Institute of Technology Lonavala, Pune (MS), India EDITORS Dr. K.C. Khare Dr. S. D. Markande Dr. C. B. Bangal Prof. A. K. Kanse SCOE SKNCOE SAE SITS Vadgaon Bk., Vadgaon Bk., Saswad Road, Narhe (Ambegaon), Sinhgad Road, Pune Sinhgad Road, Pune Kondhawa Bk., Pune Pune SUB EDITORS Dr. Prachi Mukharjee Prof. R.S. Deshpande Prof. S.C. Shilwant SKNCOE SCOE SAE Vadgaon Bk., Sinhgad Road, Pune Vadgaon Bk., Sinhgad Road, Pune Saswad Road, Kondhawa Bk., Pune Prof. Geetika Narang Prof. V. N. Deshmukh Sinhgad Institute of Technology, Lonavala, Pune SITS, Narhe (Ambegaon), Pune All correspondence including paper MISSION submission and subscription should be The Journal of Advances in Engineering addressed to: Science is devoted to high quality Dr. D. K. Singh publications in major fields of engineering (Email : [email protected]) and sciences containing experimental, Chief Editor theoretical and applied research results of Journal of Advances in Engineering outstanding significance and are not under consideration else where. Section A is Science designed to study the frontiers between Sinhgad Institute of Technology Material Science and Engineering, Chemical Gat No 309 & 310, Kusgaon (BK) and Polymer Engineering and Electrical Off Mumbai - Pune Expressway Engineering. Section B contains advances Lonavala 410 401, Pune (MS) India in the leading areas of Mechanical Telefax : (02114) 278304 Engineering, Production, Automobile and Phone : (02114) 304401, 304353 Industrial Engineering. Section C is devoted to Electronics and Telecommunication, ANNUAL SUBSCRIPTION : 2 ISSUES Computer Engineering and Information Technology. Section D is devoted to Civil Inland Rs. 400 engineering, Architecture, Structural Overseas US $ 50 Engineering, Energy and Environment. Draft or cheque should be drawn in favor of Principal, Sinhgad Institute of Technology PUBLISHED BY Lonavala (Pune) and payable at Lonavala For Dr. S. S. Inamdar outstation cheques add Rs 50/-. Principal ISSN : 0973 : 9041 July - December 2010 Sinhgad Institute of Technology Gat No. 309 & 310, Kusgaon (Bk) Front Cover Photograph Off Mumbai - Pune Expressway Sinhgad Institutes : STES Campus Lonavala 410 401, Pune (MS) India Vadgaon, Pune Telefax: (02114) 278304 Tel: (02114) 304353, 304355 PRINTED AT Copy rights are reserved with Sinhgad Technical Education Society, Pune. Views SHREE J PRINTERS PVT. LTD. expressed by the authors are their own and 1416, Sadashiv Peth, Pune 411030 (MS) India do not reflect the opinion of the editor or the Tel: (020) 24475372, Fax: (020) 24474762 editorial board. E-mail: [email protected] CONTENTS
SECTION (A) 1. Reactive Distillation Technology for Cleaner Production of Kiran D. Patil 1 Chemicals on Industrial Scale
2. Active Filter For Harmonic Reduction V.S.Baste 9
SECTION (B) 3. Performance Analysis of Diesel Engine with Mixtures of Jagannath Hirkude 15 Waste Fried Oil and Methanol Molar (Biodiesel)
4. Prediction and Optimization of Spur Gear Pair by A.G. Kamble 23 Response Surface Method
SECTION (C) 5. The Buried Gate MESFET with Turning on Characteristic Jaya T. 29
6. Discriminative Common Vectors for Face Recognition Sachin R. Jadhav 33 Using Iterative Approach
7. Recognize Human Emotions: Recognition from DCT S. B. Nimbekar 41
8. Embedded Linux Based Graphics LCD D.K.Shende 45 Application Development
9. Cost Estimation for Distributed Systems using S. V. Pingale 49 Use Case Diagram
SECTION (D) 10. Limnological Study of Pashan Lake, Pune M.S. Jadhav 57 (Maharashtra, India) Guide lines for Author
1. Papers are invited to contribute the next issue on - All four sections.
2. The manuscript should be typed in 1.5 space and printed in 10 point font on one side of Letter size papers with 1 inch margin on all four sides. Soft copy is preferred.
3. The cover page should contain title of paper, author’s name and affiliation and email address of the corresponding author. Not more than five key words should be indicated separately. A non mathematical abstract of about 250 words should be submitted along with the manuscript.
4. Unit of measure should be in SI.
5. References must be given in the international format.
6. Editorial decisions will be communicated within a period of two months after receipt of paper. Unaccepted papers will not be sent back. Kiran D. Patil et al. : Reactive Distillation Technology for Journal of Advances in Engineering Science1 Cleaner Production of Chemicals on Industrial Scale Section A (1), July - December 2010, PP 1-8
REACTIVE DISTILLATION TECHNOLOGY FOR CLEANER PRODUCTION OF CHEMICALS ON INDUSTRIAL SCALE
Kiran D. Patil1*, Bhaskar D. Kulkarni2
1 - Dept. of Petrochemical& Petroleum Engg., MAEER’s, M.I.T, Pune, INDIA 2 - Chemical & Process Development Division, National Chemical Laboratory, Pune, INDIA
ABSTRACT
Considerable progress was made by the chemical industry in the development and implementation of cleaner production processes via Reactive Distillation (RD) technology. The main challenge for reactor engineering is to ascertain the scientifically based cleaner technologies necessary for meeting the future energy, environmental,
and materials needs of the world. Minimisation of direct or indirect CO2 emission for energy intensive processes is precedence for reducing their environmental impact. For distillation, this issue is observed in redesigning the separation processes to improve the energy efficiency. In this paper a new methodology is presented for assessing for cleaner production of chemicals economically. We present here the potential applications of RD technology for cleaner production of industrial chemicals by illustrating case studies. This paper presents the principles of green technology, with case studies of greener technologies, reactive distillation and highlights the economic benefits of adopting environmentally friendly processes. We also describe the positive and negative aspects of typical RD technology and the potential and tradeoffs for RD technology in the context of green engineering principles.
1. INTRODUCTION reactions can be run in water rather than in traditional organic solvents by applying GCE principles to Green Chemical Engineering (GCE) involves chemical products and processes.[5,11] Elimination/ alternative more environmental - friendly synthetic minimisation of waste, improvement in safety, processes, and was another area for production of enhancement of security, and saving industry money chemicals on industrial scale with the mantra – ‘Think are the few dividends of these technologies. Green’.[1,2,17] Traditionally chemical engineers have been more concerned about selectivity than 2. NEED FOR NOVEL TECHNOLOGIES conversion. Green technologies explore alternate Rapid globalization of the industries, economic reaction conditions, alternate (solvent free) media and and political changes in the world order, even alternate energy sources. These technologies environmentally driven governmental regulations and aimed for the design of chemical products and social pressures, declining reserves of some raw processes that reduces or eliminates the use and materials, and increasing competition and generation of hazardous substances, which is the most overcapacity are the circumstances wherein fundamental approach to pollution prevention. [17] companies are fighting for their future existence. As Green technology addresses the need to produce a response, the chemical and petroleum industries are the goods and services that society depends on in a slowly realizing the significance of new technologies more environmentally benign manner. Life-saving - compared to incremental improvements in the pharmaceuticals can be produced while minimizing existing processes - as a means of producing the amount of waste generated, plastics that products more competitively while simultaneously biodegrade can be synthesized from plants, and protecting the environment. The need for more
*[email protected] 2 Journal of Advances in Engineering Science Section A (1), July - December 2010 efficient and environmentally ‘cleaner’ processes has to humans and the environment “[15, 16]. Cleaner led to a growing interest in novel processes and hybrid production aims at progressive reductions of the process systems, which combine conventional and environmental impacts of processes, products and non-conventional processes.[11] services, through preventative approaches rather The demands from the society for ‘cleaner’ than control and management of pollutants and technologies rather ‘clean-up’ technologies, the wastes once these have been created. It is an emergence of ‘performance chemicals and materials’ integrated approach, since it includes all relevant etc., is driving the profession towards achieving environmental aspects and impacts, and is not symbiotic relationship with other disciplines.[2,11] confined to one environmental impact category like The need for more efficient and environmentally most end-of-pipe technologies. Cleaner technology ‘cleaner’ processes has led to a growing interest in is a manufacturing process which by its nature or novel processes and hybrid process systems, which intrinsically: combine conventional and non-conventional ü Reduces effluent and other waste production; processes. A drive towards ‘Process Intensification’ ü Maximizes product quality; is leading to the search of novel reactor configurations, which will enhance selectivity as well ü Maximizes raw materials and energy and any as productivity. Emerging equipment, processing other input use. techniques, and operational methods promise Cleaner Production stands for a proactive and amazing improvements in process plants, noticeably preventive approach to industrial environmental shrinking their size and significantly boosting their management and aims for process- and/or product- efficiency. These developments may consequence in integrated solutions that are both environmentally and the disappearance of some traditional types of economically efficient (‘eco-efficiency’). Cleaner equipment, if not whole unit operations. [12] Production in the process industry is not a new Today, we are witnessing important new concept [15]. It has essentially been practiced since developments that go beyond “traditional” chemical the first chemical processes were utilized in our engineering. Modern industries are looking for novel industrial society. Initially, the industry focused on equipment and techniques that potentially transform issues such as yield improvement rather than concept of chemical plants and lead to compact, specifically preventing pollution from entering the safe, energy-efficient, and environment-friendly environment. However, the consequence of both sustainable processes. Process intensification activities was the same: less material constituting consists of the development of novel apparatuses waste streams entering the environment. The brings dramatic improvements in manufacturing and application of cleaner production technologies and processing, substantially decreasing equipment-size/ practices has already enabled the process industry production-capacity ratio, energy consumption, or to reduce and better manage pollution risks waste production, and ultimately resulting in cheaper, associated with wastes and other releases. Industry sustainable technologies.[2,3] is now tasked with addressing emerging environmental issues, including the emissions of 3. CLEANER TECHNOLOGY AND/OR CLEANER greenhouse gases, and providing a positive PRODUCTION CONCEPTS AND BENEFITS contribution to sustainable development in general. Cleaner Production is generally defined as “the Further and renewed application of the cleaner continuous application of an integrated preventive production approach can provide a competitive edge environmental strategy to processes, products, and for process industries while addressing those services to increase eco-efficiency and reduce risks emerging environmental concerns. Kiran D. Patil et al. : Reactive Distillation Technology for 3 Cleaner Production of Chemicals on Industrial Scale
Cleaner Production is related to several other d) Good housekeeping refers to changes in environmental management concepts [16]. operational procedures and management in a) Environmental impact categories covered, in order to eliminate waste and emission particular whether only one environmental generation. Examples are spill prevention, impact category is targeted (‘single media’) or improved instruction of workers and training. several (‘multi media’); e) On-site recycling refers to the useful b) Primary motivation driving the respective application of waste materials or pollutants at environmental management strategy: a division the company where these have been is made between environmental regulation or generated. This could take place through re- corporate self-responsibility as the key use as raw material, recovery of materials or drivers; useful application. c) Reactive versus preventive approaches; 4. GREEN CHEMICAL ENGINEERING whether the environmental strategy addresses PRINCIPLES waste and pollution once it has been generated or aims to avoid waste and The chemical process industry aims particularly emissions in the first place; at energy, capital expenditure and variable feedstock cost savings due to fierce global competition and d) Focus: whether the environmental management requirements for sustainable development. concept centres on waste streams, production Increasingly novel processes are used in the industry facilities or product life cycle. to achieve these aims.[1,6] They are used in: Cleaner Production aims at making more efficient a) Existing processes to renew parts; use of natural resources (raw materials, energy and water) and reducing the generation of wastes and b) Process re-designs based on existing feed emissions at the source. This can be achieved in stocks and catalysts; various ways. A division in five prevention practices is c) Innovative processes (new feed stocks, new most common. [e.g. 17]. catalysts, new process routes, new a) Product modifications change the product multifunctional equipment). characteristics, such as shape and material composition. The lifetime of the new product GCE is much more than a method for addressing is, for instance, expanded, the product is environmental problems. It offers a framework for easier to repair, or the manufacturing of the achieving innovation. It is a way to not only improve the product is less polluting. Changes in product environment but also positively impact the client’s packaging are generally also regarded as bottom line. Avoiding the generation of waste product modifications. (including energy) or pollutants can often be more cost-effective than controlling or disposing of b) Input substitution refers to the use of less polluting raw and adjunct materials and the pollutants once formed. use of process auxiliaries (such as lubricants 4.1 Twelve Principles of Green Engineering and coolants) with a longer service lifetime. 1. Inherent Rather Than Circumstantial c) Technology modifications include for instance Designers need to strive to ensure that all improved process automation, process optimization, equipment redesign and process materials and energy inputs and outputs are as substitution. inherently nonhazardous as possible. 4 Journal of Advances in Engineering Science Section A (1), July - December 2010
2. Prevention Instead of Treatment 12. Renewable Rather Than Depleting It is better to prevent waste than to treat or clean Material and energy inputs should be renewable up waste after it is formed. rather than depleting. 3. Designs for Separation 5. GREEN CHEMICAL ENGINEERING ASPECTS Separation and purification operations should be OF REACTIVE DISTILLATION designed to minimize energy consumption and Green engineering principles have been described materials use. recently in the context of design, both for 4. Maximize Efficiency manufacturing molecules in chemical processes and for more general products and systems. Green Products, processes, and systems should be engineering can also have a major impact on by- designed to maximize mass, energy, space, and product formation as well as other factors identified in time efficiency. green engineering principles. Reactive distillation has 5. Output-Pulled Versus Input-Pushed been used as a “clean technology” for chemical Products, processes, and systems should be synthesis in the industry because of advantages like “output pulled” rather than “input pushed” through the saving of energy cost and capital cost. Reactive use of energy and materials. distillation has been used as a “clean technology” for chemical synthesis in the industry because of 6. Conserve Complexity advantages like saving of energy cost and capital Embedded entropy and complexity must be cost.[6] viewed as an investment when making design choices 6. PRINCIPLES OF REACTIVE DISTILLATION on recycle, reuse, or beneficial disposition. TECHNOLOGY 7. Durability Rather Than Immortality 6.1 Reactive Distillation Process: Targeted durability, not immortality, should be a With ever-growing environmental concerns, design goal. petrochemical and fine chemical industries face a 8. Meet Need, Minimize Excess ubiquitous issue in recovering dilute acetic acid from Design for unnecessary capacity or capability its aqueous solutions. Reactive distillation (RD) holds (e.g., “one size fits all”) solutions should be dominance over conventional physical separation considered a design flaw. methods such as distillation and extraction. Distillation is associated with the high costs involved 9. Minimize Material Diversity in vaporizing the more volatile water that exists in Material diversity in multicomponent products high proportions and possesses a high latent heat of should be minimized to promote disassembly and vaporization. Extraction is limited in view of the value retention. distribution of the components in the reacting system. 10. Integrate Material and Energy Flows The execution of RD reduces capital and operating costs, and allows for a wider range of operating Design of products, processes, and systems conditions. RD is receiving increasing attention and must include integration and interconnectivity with holds a huge potential for the recovery of acetic acid available energy and materials flows. in many industrial processes. [9, 13] 11. Design for Commercial “Afterlife” RD is the combination of chemical reaction and Products, processes, and systems should be distillative product separation in single piece of designed for performance in a commercial “afterlife”. equipment, offers several dividends over conventional Kiran D. Patil et al. : Reactive Distillation Technology for 5 Cleaner Production of Chemicals on Industrial Scale processes in which the reaction and the product 6.2 Continuous fatty acid esterification separation are done in series, especially for reactions Processes limited by equilibrium constraints. Improved The following section describes the basic selectivity, increased conversion, better heat control, principles and suggests a possible process set-up for effective utilization of reaction heat, scope for difficult RD process. In this example, the esterification of separations and the avoidance of azeotrope are a few acetic acid with Isoamyl alcohol is described. of the advantages that RD offers. [4] As the products in RD are continuously separated from the reaction In principle, RD column consists of three different zone, no limiting chemical equilibrium can be sections; middle one reactive section, bottom one established and thus the reaction velocity is stripping section and top one rectifying section. The maintained at a high rate, resulting in greater yields. reaction and distillation (separation) takes place in middle section; i.e. reactive section. A peristaltic Other benefits of RD can include the minimization pump was being used to introduce feed to the of side reactions and the utilization of the heat of column. Two peristaltic pumps were used for two reaction for the mass transfer within the same feeds. A double-coiled condenser was used and it has column. Therefore the capital investment and been ensured that the condensation is complete. operating costs are significantly lower with RD than The experimental setup of a laboratory scale for conventional processes. Examples for successful reactive distillation consists of 3 m tall distillation applications of RD have, among others, been reported column of inside diameter 50 mm that operates at for esterifications, etherifications, alkylations and atmospheric pressure is used. The reboiler (2 lit) is isomerizations [16, 17]. Sulzer Chemtech has heated with the help of a heating mantle (2KW), developed special structured catalytic packings for provided in the reboiler. The non-reactive rectifying and Reactive distillation columns, see Figure 1. [12, 17] stripping sections are packed with wire mesh RD is a potentially important method of packing. (Evergreen Technologies, Mumbai) The separation for the recovery of dilute acetic acid from middle reactive zone is packed with structured its aqueous streams. Moreover, a value added product packing (Fenix Technologies, Pune) embedded with in the form of isoamyl acetate is produced during the TULSIONR –T- 63 MP (Courtesy Thermax India Ltd.), recovery of acetic acid by esterification with Isoamyl ion exchange resins as a catalyst. The stripping alcohol. An additional column will be required for the section is 1 m tall, reactive section 1 m, and complete separation of Isoamyl alcohol and Isoamyl rectification section 1 m tall in height respectively. A acetate. The energy costs for the additional column proper insulation (with asbestos) with external wall would be minimized as the water content of the top heating arrangement is provided to minimize the heat organic product is kept to a minimal [18]. losses to the surrounding.
Figure 1 : RD column Internals 6 Journal of Advances in Engineering Science Section A (1), July - December 2010
* Blank entries indicate no known advantage or disadvantage. Table 1 : Positive and negative aspects of typical RD technology with respect to green engineering principles*[5]
The reaction mixture consisting of acetic acid, alcohol is fed continuously to the column. An electronically driven metering pump is used to transfer the liquid from the feed tank to the column. The boil- up rate is an important parameter in RD processes. As the reflux ratio is fixed, the boil-up is the only operating parameter that can be efficiently changed to obtain the better performance. Hence it is possible to operate the column over a wide range of boil-up rates to study its effect. A wattmeter is used to measure the wattage power supplied to the heating mantle. The continuous process scheme for RD process on Laboratory scale is shown in Figure 2. In the condenser, two immiscible phases are formed, an aqueous phase i.e. almost pure water and an organic phase containing water, amyl alcohol and amyl acetate. Phase separator with the condenser is used to provide reflux to the column and to continuously withdraw water formed during the reaction. Thermometer wells are provided at different Figure 2 : Continuous RD process for locations in the column to measure these production of Industrial Chemicals temperatures (Position 1- Position 8) Kiran D. Patil et al. : Reactive Distillation Technology for 7 Cleaner Production of Chemicals on Industrial Scale
The conventional separation methods such as proposed as a promising technique for separation/ distillation and extraction suffer from several recovery of acetic acids from wastewater. drawbacks and are very expensive in terms of time, energy and chemicals. RD is a method of separation Moreover, a value added product in the form that holds huge potential in the recovery of acetic acid of acetate is produced during recovery of acetic from aqueous streams. Through the application of RD acid by esterification with alcohol. Finally we via the reaction of acetic acid with an aliphatic alcohol conclude that RD process is cleaner technology (e.g. Isoamyl alcohol), a useful ester in the form of (e.g. and front-runner in recovery/separation and Iso amyl acetate) was produced. The esters of acetic production of chemicals on industrial scale. [12] acid, namely, Iso amyl acetate, n-butyl acetate, n-hexyl acetate, have a wide range of industrial 8. ACKNOWLEDGEMENTS applications. [7, 13, 14] This research work enjoyed financial support from 6.3 Advantages of the Reactive Distillation AICTE-RPS scheme, New Delhi and BCUD, Process University of Pune. The support of these organizations is gratefully acknowledged. KP is thankful to Thermax The combination of reaction and separation by India Ltd. for sponsoring the catalyst and MAEER’s distillation in one unit allows a continuous production, MIT, Pune for providing the basic infrastructure for this with reduced processing time. This leads to constant work. high product quality and at the same time simpler maintenance and process control, which is especially 9. REFERENCES worthwhile for larger production capacities. Well- 1) Anastas, P.T., and Zimmerman, J.B., Design defined and narrow residence time at gentle conditions through the Twelve Principles of Green throughout the whole plant minimizes degradation of Engineering, Env. Sci. and Tech., 37, 5, 2003, the fatty acids and fatty acid esters. No neutralization, PP 95 - 101. separation or recycling of catalyst is necessary. There 2) W. Jim Swindall, Recycling and clean is no necessity of emptying and cleaning the technology, Clean Technologies and equipment, reducing the waste streams to the Environmental Policy, Volume 5, Number 1 (2003). absolute minimum. The energy consumption of the RD process is only half of the conventional batch one. 3) Sundmacher, K., Kienle, A., Eds, Reactive Also the size of the plant could be drastically reduced Distillation-Status and Future Trends, Wiley- VCH: New York, 2003. [16]. 4) Taylor, R.; Krishna, R. Modeling reactive 7. CONCLUSION distillation. Chem. Eng. Sci., 55, (2000) It is important to have an efficient and PP 5183-5229. sustainable technology for the separation/ 5) Malone, M.F., Doherty, M.F., Reactive recovery of acetic acids from the wastewater Distillation, Ind. Eng. Chem. Res. 2000, 39, stream from the points of view of pollution control PP 3953-3957. and recovery of useful materials. Several 6) Malone, M.F., Doherty, M.F., Huss, R.S., Green conventional separation techniques discussed in Chemical Engineering Aspects of Reactive this paper have been employed to remove Distillation, Environ. Sci. & Technol, 2000, carboxylic acids from aqueous solution. Some of these techniques are not environment friendly and 7) Sharma, M. M.; Mahajani, S. M. Industrial application of reactive distillation (A review). In others are not cost effective. RD process allows Reactive Distillation-status and future trends; the operation to run continuously, leading to a Sundmacher, K., Kienle, A., Eds.; Wiley-VCH: consistently high product quality. RD has been New York, 2003; PP 1-29 (ISBN 3-527-30579-3. 8 Journal of Advances in Engineering Science Section A (1), July - December 2010
8) Hiwale, R. S.; Bhate, N. V.; Mahajani, Y, S.; 13) Teo, H. T. R., B. Saha and A. Alqahtani, iso- Mahajani, S. M. Industrial Applications of Amyl acetate synthesis by catalytic distillation, Reactive Distillation: Recent Trends. Ind. J. Ind. J. Chem. Reactor Eng, 3, 2005, Article A11, Chem. Reactor Eng. 2004, 2, R1. PP 1-14. 9) Ajay Singh, Anand Tiwari, Sanjay M. Mahajani , 14) Chiang S. F., Chien Lin Kuo, Cheng Ching Yu, Recovery of Acetic Acid from Aqueous Solutions and Wong, Design alternative for the amyl by Reactive Distillation, Ind. Eng. Chem. Res. acetate process: coupled reactor/ column and 2006, 45, PP 2017-2025 reactive distillation, Ind. Eng. Chem. Res., 41, 2002, PP 3233-3246. 10) Mahajani S. M. and Chopade S. P., Reactive Distillation: Processes of Commercial 15) Zosel, T. (1994), Pollution Prevention in the Importance, in Encyclopedia of Separation Chemical Industry, in Edgerly, D. (ed), Science, by Edlard T. R., Poole C. A. and Cooke Opportunities for Innovation: Pollution M. (Eds.), Academic Press, London, UK, (2001), Prevention, National Institute of Standards and 4075-4 Technology, Gaithersburg, USA, PP 13-25. 11) Santi Kulprathipanja, Reactive Separation 16) Van Berkel, R., E. Willems and M. Lafleur (1997 Processes, Taylor & Francis Group, 2002 a), the Relationship between Cleaner Production and Industrial Ecology, in Journal of Industrial 12) Harmsen, G. Jan, Reactive distillation: The front- Ecology, Volume 1, Number 1, PP 51-66. runner of industrial process intensification. A full review of commercial applications, research, 17) USEPA (1992) Facility Pollution Prevention scale-up, design and operation, Chemical Planning Guide, United States Environmental Engineering and Processing, 46, 2007, Protection Agency, Cincinnati, EPA 600/R92/088. PP 774–780. V.S.Baste et al. : Active Filter For Harmonic Reduction Journal of Advances in Engineering Science9 Section A (2), July - December 2010, PP 9-14
ACTIVE FILTER FOR HARMONIC REDUCTION
V.S.Baste1*, R.T.Patil2 and S.R.sonawane3
1 - Dept. of Electronics & Telecommunication, SIT, Lonavala 2 - TKOET, Warnanagar 3 - ACSC, Alandi, Pune
ABSTRACT
In the modern power distribution system, majority of loads draw reactive power and harmonic currents from ac source along with main power currents. These non-unity power factor linear and non-linear loads cause low efficiency of supply system, poor power-factor, destruction of other equipments due to excessive stresses and EM1 problems. Active filters have been considered an effective solution to reduce these problems. This paper presents an Active Power Filter (APF) based on simple control technique to provide reactive power and harmonics compensation for linear and non-linear single-phase loads. A voltage source inverter with carrier less hysteresis PWM current control is used to form an APF. A simple P-I (proportional-integral) dc bus voltage controller with reduced energy storage capacitor is employed in the APF
1. INTRODUCTION converter using the optimization technique to reduce harmonics. Power Electronics plays an important role in the control of generated electrical energy for today’s 2. OVERVIEW OF HARMONICS industrial applications in various industrial, Harmonics are undesirable components in the commercial, residential and military environments. sinusoidal waveform of the AC Power supply. The control of electrical power conversion from one Harmonics occur as integral multiples of the form to another is necessary, and is achieved through fundamental frequency. power converters. The control strategy for power converters plays an important part in the harmonic That is, the third order harmonic will have a generation and the output waveform distortion. frequency of 3 times the fundamental frequency; 150 Advancement of semiconductor technology has Hz which is 3 times the fundamental 50 Hz frequency. boosted the power electronic field due to the Harmonics affect power quality and efficiency. It is availability of power devices such as Power MOS therefore necessary that Harmonics in any power Field Effect Transistors (Power MOSFETs), Insulated system be monitored. Gate Bipolar Transistors (IGBTs) and Gate Turn Off The harmonics can arise in three ways: Thyristors (GTO’s) that have high power rating and good switching characteristics. These devices are 1) through the application of a non sinusoidal widely used in the power converter circuits. However driving voltage to a circuit containing nonlinear the output currents and voltages of static power impedance. converters are generally associated with low-order 2) through the application of a sinusoidal driving harmonics. The harmful effects of low-order harmonics voltage to a circuit containing nonlinear produced by these devices can be reduced by impedance. incorporating the use of microcontroller in the control 3) through the application of a non sinusoidal strategy of the converter circuits. This paper describes driving voltage to a circuit containing linear a computer-based control strategy for an ac to dc impedance.
*Vaishali.baste@gmail 10 Journal of Advances in Engineering Science Section A (2), July - December 2010
Harmonics can rectified by using suitable 3. ACTIVE FILTERS methods such as filters. Using a mathematical An Active Harmonic Filter is an electronic power technique known as Fast Fourier Transforms, the inverter using IGBT semiconductors with various distorted AC waveform can be resolved into its control loops to increase Power Factor and reduce component waveforms. Of the measured harmonics, harmonics by injecting a dynamic cancellation signal the even harmonics(harmonics whose frequency are into the power line. The Active Harmonic Filter the fundamental frequency multiplied by even solution from Power Correction Systems improves the numbers such as 100Hz(2 *50) or 200Hz(4*50) get efficiency and operation of AC Motor Systems, AC cancelled out and have no effect. For the study and Variable Frequency Drive (VFD) Systems and works management of Harmonics, only the odd harmonics well with virtually every Electrical and Electronic are considered. Device in your facility. The harmonic content must be reduced due to Active power filters were first proposed for following reasons harmonic compensation in the early 1970’s, but they could not be used in real power systems because 1) Performance of loading device fed through high-power high-speed switching devices were inverter is affected by harmonics. unavailable. Since then, and because of the high 2) Harmonics cause additional losses. development of power electronics technology, much research has been done on active filters and their High frequency harmonics can be reduced easily practical applications. The operation of an active filter by low size filter. However, to reduce lower order is based on a continuous monitoring and conditioning harmonics, big size filter are required. Therefore, lower of the distorted current created by the non-linear load. order harmonics must be reduced by some technique The same harmonic currents, but with a 180º phase other than the filtering i.e. PWM. shift are generated by the filter, so that harmonic 2.1 Types of Harmonic Sources components are cancelled and only fundamental component flows from the point of common coupling There are two types Harmonic Sources: of the load. The increasing use of power electronics- 1) Current-Source Type of Harmonic Sources based loads (adjustable speed drives, switch mode power supplies, etc.) to improve system efficiency 2) Voltage-Source Type of Harmonic Sources and controllability is increasing the concern for 2.2 Harmonic Detection Methods harmonic distortion levels in end use facilities and on the overall power system. The application of passive There are different algorithms for harmonic tuned filters creates new system resonances which detection, which lead to a large scientific debate on are dependent on specific system conditions. In which part the focus should be put on, the accuracy, addition, passive filters often need to be significantly speed, the filter stability, easy & inexpensive overrated to account for possible harmonic absorption implementation etc. from the power system. Passive filter ratings must be co-ordinated with reactive power requirements of the Frequency Domain loads and it is often difficult to design the filters to 1) Discrete Fourier Transform avoid leading power factor operation for some load conditions. 2) Fast Fourier Transform Active filters have the advantage of being able to 3) Recursive Discrete Fourier Transform compensate for harmonics without fundamental Time Domain frequency reactive power concerns. This means that the rating of the active power can be less than a 1) Synchronous fundamental “dq-frame” comparable passive filter for the same non-linear load 2) Synchronous individual harmonic ”dq-frame” and the active filter will not introduce system resonances that can move a harmonic problem from 3) Generalized integrators & variants one frequency to another. V.S.Baste et al. : Active Filter For Harmonic Reduction 11
To demonstrate reactive power compensation capability of the APF, linear loads of lagging and leading power-factors are considered with the step change. Non-linear loads consisting diode rectifier with capacitive loading and solid state ac regulator with inductive loading, arc taken APF system to illustrate its capability for harmonic and reactive. power compensation The main role of the proposed APF is to eliminate harmonics and to provide reactive power requirement of the load locally so that ac Figure 1 : Shunt Active Power Filtering source feeds only Fundamental sinusoidal active component of unity power-factor current. Since this The active filter concept uses power electronics APF system is connected in shunt with load, it to produce harmonic current components that cancel improves the system efficiency because it does not the harmonic current components from the non-linear process the active power delivered to the load loads. The active filter uses power electronic 3.1 Control Scheme switching to generate harmonic currents that cancel the harmonic currents from a non-linear load. The Fig. 3 shows the block diagram of an overall active filter configuration investigated in this lecture is control scheme for the APF system. DC bus voltage based on a pulse-width modulated (PWM) voltage and supply voltage and current are sensed to control source inverter that interfaces to the system through the APE. AC source supplies fundamental active a system interface filter as shown in Figure 1. In this power component of load current and a fundamental configuration, the filter is connected in parallel with the component of a current to maintain average dc bus load being compensated. Therefore, the configuration voltage to a constant value. The later component of is often referred to as an active parallel or shunt filter. source current is to supply losses in VSI such as switching loss, capacitor leakage current etc. in 3. RELATED WORK steady state and to recover stored energy on the dc Fig2. show the fundamental building block of the bus capacitor during dynamic conditions such as proposed parallel APF. It is comprised of a standard addition or removal of the loads single phase voltage source MOSFET based bridge inverter with dc bus capacitor and dc boost voltage for an effective current control. A hysteresis rule based carrier less PWM current control technique is used to provide fast dynamic response of the APF
Figure 3 : Control Scheme of the APF
The sensed dc bus voltage of the APF along with its reference value are processed in the P-I voltage controller. The truncated output of the P-I controller is taken as peak of source current. A unit vector in phase with the source voltage is derived using its sensed value. The peak source current is multiplied with the unit vector to generate a reference sinusoidal Figure 2 : Basic Circuit of the Active Power Filter unity power factor source current. The reference 12 Journal of Advances in Engineering Science Section A (2), July - December 2010 source current and sensed source current are reactive power components of load “current. AC source processed in hysteresis carrier less PWM current feeds only fundamental active power component of controller to derive gating signals for the MOSFETs of load current resulting in source current exactly in the APF. In response to these gating pulses, the APF phase with source voltage. Source current remains impresses a PWM voltage to flow a current through sinusoidal during transient conditions and attains new filter inductor to meet the harmonic and reactive steady state value without any overshoot and components of the load current. Since all the oscillation. A drop and a rise is observed in dc bus quantities such as dc bus voltage etc. are symmetric voltage at application and removal of additional load. and periodic corresponding to the half cycle of the ac However, during steady state the P-I voltage controller source. A corrective action is taken in each half cycle maintains a constant average voltage at the dc bus of of the ac source resulting in fast dynamic response the AF. Harmonic spectrum of load and supply of the APF . currents at 15 A and 30 A peak values of load current is shown in Fig. 7 for rectifier type non-linear load. It 4. PERFORMANCE METRICS may be observed that the harmonics are eliminated Fig. 4 shows performance of APF system with from source current. The THD in supply current is non-Linear rectifier load. Load draws discontinuous reduced to4.67 from 83.92 % of the load current (15 peaky current from ac’ bus. APF feeds harmonic and A peak)
Figure 4 : Performance of the APF System under Non-linear V.S.Baste et al. : Active Filter For Harmonic Reduction 13
electric system and can improve in the real system. A simple P-I controller based APF has been found effective to provide reactive power and harmonic compensation for the variety of loads. An excellent performance of APF system has been observed as a universal power-factor controller and an ideal reactive power compensator. APF is able to reduce the harmonics well below 5 % in all the cases of extremely reactive and harmonic polluted loads. APF has maintained sinusoidal supply current in phase with the supply voltage resulting in unity power-factor of the supply both in steady state and transient conditions This PWM inverter also can be modified to study voltage sag, voltage swell, interruption, voltage transients, voltage fluctuation, voltage (light) flicker, voltage imbalance, over voltage, under voltage and also other harmonics performance parameters
Figure 5 : Harmonic Spectrum with Nonlinear Rectifier Fed Load 7. REFERENCES 5. RESULTS AND DISCUSSIONS 1. Tain-Syh Luor; “Influence of load characteristics on the applications of passive and active It is also noted from these waveforms that the harmonic filters” Harmonics and Quality of power factor Decreases with decrease in load under Power, 2000. Proceedings. Ninth International duty ratio control. This is due to the fact that the power Conference on, Volume: 1, 1-4 Oct. 2000 factor depends on ratio of switching frequency to PP. 128 -133 vol.1. resonant frequency. The APF eliminates harmonic components effectively and is able to compensate the 2. P. Enjeti, P. D. Ziogas, and J. F. Lindsay, reactive power “Programmed PWM techniques to eliminate harmonics: A critical evaluation,” IEEE Trans. 6. CONCLUSIONS & FUTURE WORK Ind.Applicat., vol. 26, PP. 302–316, Mar. 1990. In this paper we present the harmonic 3. Dariusz Czarkowski, Member; David V. reduction system using active filter to reduce total Chudnovsky; Gregory V. Chudnovsky; and Ivan harmonic distortion value by the computer which W. Selesnick “Solving the Optimal PWM work with microcontroller and active harmonic filter. Problem for Single-Phase Inverters” IEEE The harmonic reduction system program PWM TRANSACTIONS ON CIRCUITS AND and control active harmonic filter by analyze signal SYSTEMS-I: FUNDAMENTAL THEORY AND in system .The computer shows electric signal APPLICATIONS, VOL. 49, NO. 4, APRIL 2002. ,harmonic analysis ,function direct and automatic 4. Chen, Z.; Blaabjerg, F.; Pedersen, J.K.; “A study filter control. The results present ability of the of parallel operations of active and passive filters” harmonics reduction that increase the Efficiency of Power Electronics Specialists Conference, 14 Journal of Advances in Engineering Science Section A (2), July - December 2010
2002. pesc 02. 2002 IEEE 33rd Annual , Volume: 2 , 23-27 June 2002 PP.1021 -1026 vol.2. 5. Wilson E. Kazibwe, Musoke H. Sendaula. Electric Power Quality Control Techniques. New York :Van Nosteand Reinhold,1993. 6. E.O. Brigham, “The Fast Fourier transform”, Prentice- Hall,1974. 7. L.H. Thomas, “Using A Computer to Solve Problems in Physics, Application of digital Computer ,” Boston. Mass. :Ginn,1963. 8. Acha, Enrique “Power system harmonics: Computer modeling and analysis” Chichester : John Wieley & Sons, c2001. 9. B. Sing, K. A. Haddad, A Chandra, “Universal Active Power Filter for single-phase reactive power and harmonic compensation,” International conference on IEEE Power Quality, PP. 81 -87, 1998. Jagannath Hirkude et al. : Performance Analysis of Diesel Engine Journal of Advances in Engineering Science15 with Mixtures of Waste Fried Oil and Methanol Molar (Biodiesel) Section B (2), July - December 2010, PP 15-22
PERFORMANCE ANALYSIS OF DIESEL ENGINE WITH MIXTURES OF WASTE FRIED OIL AND METHANOL MOLAR (BIODIESEL)
Jagannath Hirkude1*, A.S. Padalkar2 and Jisa Randeer3
1 - Dept. of Mech. Engg., Padre Canceicao College of Engg., Verna (Goa), INDIA 2 - Dept. of Mech. Engg., Sinhgad College of Engg., Pune (MS), INDIA 3 - Dept. of Mech. Engg., Padre Canceicao College of Engg., Verna (Goa), INDIA
ABSTRACT
This experimental investigation focuses on production of biodiesel from Waste Fried Oil (WFO) by changing oil to methanol molar ratios and performance analysis of it in diesel engine. High viscosity and poor volatility are the major limitations of waste fried oil for utilization as a fuel in diesel engine. The oil to methanol molar ratios considered for transesterification were 3:1, 6:1 and 9:1. The cost of biodiesel production is presented in this paper and found more economical than mineral diesel. It satisfies the important fuel properties as per ASTM specification of biodiesel. This paper discusses the performance of biodiesel in a single-cylinder, four-stroke, direct-injection, diesel engine. The performance of engine with mineral diesel has been considered as base line. Biodiesel B50 with molar ratio of 6:1 yielded the highest thermal efficiency 30.2% closer to mineral diesel (31.6%) at rated load. The brake specific fuel consumption for B50 was 0.31 kg/kWh as against 0.29 kg/kWh of diesel. The highest exhaust gas temperature was observed for molar ratio of 3:1 and it was 31oC higher than that of diesel. For daily 6 hours operation for 300 days, it is possible to save Rs 21606 by running the engine on B50(6:1) mode.
1. INTRODUCTION total biodiesel production cost arises from the cost of raw material [7]. Everywhere in the world, there is an Biodiesel derived from vegetable oil by enormous amount of waste lipids generated from transesterification with alcohol like methanol and restaurants, food processing industries and fast food ethanol is recommended for use as a substitute for shops everyday. Reusing of these waste greases not petroleum-based diesel mainly because biodiesel is only reduces the burden of the government in an oxygenated, renewable, biodegradable and disposing the waste, maintaining public sewers and environmentally friendly bio-fuel with similar flow and treating the oily wastewater, but also lowers the combustion properties including low emission profile production cost of biodiesel significantly. Furthermore, [1-2]. The used vegetable oil classified as waste is biodiesel fuel has been demonstrated to be increasingly attracting much interest because of its successfully produced from waste edible oils by an great potential to be used as diesel substitutes known alkali-catalyzed transesterification process [8-11]. as bio-diesel [3]. Study has shown that vegetable oil Literature review indicates that there is need of based fuels can significantly reduce exhaust gas conversion of WFO from kitchen waste into biodiesel emissions, including carbon monoxide, carbon using transesterfication process. The economics of dioxide, particulate matter and sulfur dioxide [4-6]. production of biodiesel from waste fried oil waste fried As compared to petroleum-based diesel, the high oil methyl esters (WFOME) with mineral diesel in cost of biodiesel is a major barrier for its different compositions. commercialization. Approximately 70%–85% of the
* [email protected] 16 Journal of Advances in Engineering Science Section B (2), July - December 2010
2. METHODOLOGY Table 1 : Molar ratio for biodiesel preparation 2.1 Preparation of biodiesel for different oil Batch Molar Total volume ratio to methanol molar ratios ratio Methanol:WFO (ml/ml) The raw material (WFO) was collected from different hotels in Goa, a tourist destination in India. Batch 1 3:1 96:950 The used fried oil was filtered to remove food residues Batch 2 6:1 192:950 and solid precipitate by using double layer of Batch 3 9:1 288:950 cheesecloth in a funnel. In order to avoid soap formation due to water, the filtered fried oil was dried at 60oC for 10 minutes using microwave oven. In preheated mixture of waste fried oil and methanol, NaOH was added. The amount of potassium hydroxide needed was 7.7 grams per liter by titration with waste fried oil. This solution was stirred at 600 rpm for 15 minutes and glycerin was allowed to settle for 24 hours. The ester layer was separated from the glycerol layer in a separating funnel. Crude ester layer consisted of methyl ester, unreacted oil and methanol of glycerol, catalyst residue, and small amount of produced soap. In the separating funnel, this layer was washed with hot water, until the washings were neutral. This ester was dried and filtered. The transesterification process followed for biodiesel Figure 2 : Biodiesel with different molar ratios production is shown in Figure 1
2.2 Economic Analysis of Conversion For the conversion of waste fried oil, methanol and potassium hydroxide are available at a rate of Rs 50/ litre and Rs 250/litre respectively. The cost of waste fried oil considered was almost zero because it’s treated as discarded waste, harmful to the environment. Biodiesel cost will depend greatly on methanol prices and economy can be achieved by varying the grade of methanol used. By-product of transesterification is industrial grade glycerin which Figure 1 : Transesterification process has industrial use and can be sold with or without processing, as it is an important constituent in chemical, pharmaceutical and cosmetic industry. The The molar ratios used to find out amount of cost of production of 1 liter biodiesel for molar ratio of methanol used are tabulated in Table 1. Figure 2 1:6 is presented in Table 2. It can be seen from Table shows three batches of biodiesel with different waste 2 that the production cost of biodiesel is substantially fried oil to methanol molar ratios. lower than market price of mineral diesel. Jagannath Hirkude et al. : Performance Analysis of Diesel Engine 17 with Mixtures of Waste Fried Oil and Methanol Molar (Biodiesel)
Table 2 : Cost of biodiesel production from 2.3 Fuel Properties waste fried oil Viscosity of WFO and biodiesel was determined Biodiesel from WFO Cost Cost with help of redwood viscometer. The (Rs/lit.) (USD/lit.) transesterfication of WFO provided a significant reduction in viscosity. At 40oC, the reduction was from Waste fried oil Nil - 64.60 cSt to 10.75 cSt. The addition of WFOME Methanol (192 ml) 10.00 - increased the viscosities of blends by 10.75 cSt and Reagents 1.00 - 7.535 cSt for B100 and B50 respectively. Calorific value (38900 kJ/kg) for biodiesel for molar ratio 6:1 was KOH (7.7 gms) 2.00 - estimated with help of bomb calorimeter and found Electricity 0.40 - lower than that of mineral diesel ( 43000 kJ/kg). The Purification 0.50 - flash point temperature was found out by flash point apparatus and it is more than 93oC which is minimum Labour 1.40 - requirement for biodiesel based on ASTM D 6751- 09. Collection and 4.00 - The properties of different fuels with Indian Standards transportation used are given in Table 3. It can be seen from Table 3 Sub total 19.30 - that the properties of biodiesel produced from different molar ratios are in the acceptable ranges. Revenue from by 3.00 - product sales It is observed that with increase in molar ratio there was increase in calorific value and flash point Total (cost less revenue) 16.30 0.35 temperature. Viscosity of biodiesel from molar ratio Cost of Diesel 40.00 0.86 6:1 was found closer to mineral diesel.
Table 3 : Comparison of fuel properties
Properties WFO Diesel Biodiesel Standard No2
3:1 6:1 9:1
Viscosity at 400C (cSt) 64.14 4.320 12.20 10.75 11.51 IS: 1448 (P:25)-1976
Specific gravity 0.897 0.830 0.858 0.843 0.845 IS:2720 (Part III/Sec I)-1980
Calorific Value (kJ/kg) 31000 43000 34700 38900 42300 IS:1350 Part II – 1970
Flash point 180 70 110 125 136 IP-34 Temperature oC ASTM-D93-58 18 Journal of Advances in Engineering Science Section B (2), July - December 2010
3. EXPRIMENTATION All tests with different fuels were conducted for constant engine speed 1500 rpm and varying load on The experimental set up consisted of a single engine. Tests were carried out for 190 bar original fuel cylinder diesel engine, engine test bed, fuel and air injection pressure and injection timing 27oC before top metering equipment, exhaust gas analyzer and digital dead centre. The engine was coupled with a single temperature indicator. The specifications of engine phase, 220 V AC alternator. The alternator was used are given in Table 4. The schematic diagram of test for loading the engine through a resistive load bank. setup is shown in Figure 3. The load bank consisted eight heaters of 500 W each. The load was varied from 0.5 kW to 4 kW in step of Table 4 : Engine specifications 0.5 kW. The engine was first tested with diesel for no load and 20 minutes at rated speed of 1500 rpm until Make Laxmi Industries lubricating oil temperature reached to 80oC. Same Kolhapur (India) conditions were maintained throughout the Rated Power 3.78 kW experiment for different fuels. After the baseline test with diesel, no load test was conducted for three Rated Speed 1500 rpm batches of biodiesel prepared with different molar Number of cylinders 1 ratios 3:1, 6:1 and 9:1. For testing of each batch fuel the blends were prepared on volume basis just before Bore X Stroke 80 mm X 110 mm the experimentation. The fuel prepared from each Combustion Chamber Direct injection batch for testing purpose were B50 (50% biodiesel + with bowl in piston 50% mineral diesel), B70 (70% biodiesel + 30% Standard injection timing 270 BTDC mineral diesel) and B100 (100% biodiesel). Standard injection pressure 190 bar The specific fuel consumption was calculated by measuring the time required for a fixed volume of fuel to flow into the engine. The torque was measured using swinging field electrical dynamometer. The engine speed (rpm) was measured by electronic digital counter. The performance parameters break thermal efficiency and brake specific fuel consumption were calculated from measured data. The exhaust gas temperature was measured by using an electronic digital indicator with iron-constantan thermocouple. The exhaust gas temperature was measured by using an iron-constantan thermocouple with accuracy of ±1oC. The results from the engine with biodiesel from different oil to methanol ratios were compared with the baseline test. 4. RESULTS AND DISSCUSSION 4.1 Brake thermal efficiency (BTE)
Figure 3 : Schematic diagram of experimental The variation of brake thermal efficiency with molar setup (1) Engine (2) Alternator ratios 3:1, 6:1 and 9:1 for different blends B50, B70 and (3) Electrical Load Bank (4) Fuel tank B100 for varying load conditions is shown in Figure 4 (5) Burette (6) Two way control valve to 6. In all cases the thermal efficiency of biodiesel was (7) Air box (8) Orifice plate less than that of diesel at all power outputs. The (9) U tube manometer (10) Exhaust baseline brake thermal efficiency for diesel was 31.6%. Gas Analysis (11) Exhaust gas The maximum brake thermal efficiency for B50 with thermocouple molar ratio 6:1 was 30.2% and for other two molar Jagannath Hirkude et al. : Performance Analysis of Diesel Engine 19 with Mixtures of Waste Fried Oil and Methanol Molar (Biodiesel) ratios, it varied in the range 28.7% to 28.5%. The maximum brake thermal efficiency for B70 varied in the range 25.8% to 29.8%. For B100 the maximum thermal efficiency was in the range 23.8% to 25.7%. Lower thermal efficiencies may be due to lower heat content, higher density, higher viscosity and poor volatility of biodiesel compared to diesel. Significant improvement in thermal efficiency was observed with increase in diesel percentage in the blends. For higher molar ratio, higher the conversion of methanol was observed. However using too low molar ratio results in less conversion of methyl esters resulting in presence of un-removed glycerin and higher molar ratio with excess methanol can obstruct glycerin formation, all these can lead to higher viscosity and poor volatility of the fuel. Figure 6 : Brake thermal efficiency at brake load (B100)
4.2 Brake specific fuel consumption (BSFC) Figure 7 to 9 show the variation brake specific fuel consumption with load on engine for diesel and different biodiesel blends.
Figure 4 : Brake thermal efficiency at break load (B50)
Figure 7 : Brake specific fuel consumption at brake load (B50)
Figure 5 : Brake thermal efficiency at brake load (B70) 20 Journal of Advances in Engineering Science Section B (2), July - December 2010
reasons for increase in BSFC with increase in percentage of biodiesel in mixture may be increase in viscosity and poor spray atomization. For daily 6 hours running for 300 days a year, engine running with 50% biodiesel will consume 972 kg (1153 lit) of biodiesel and 972 kg (1168 lit) of diesel compared to 1813 kg (2178 litre) of diesel alone. Running cost of engine with 50% biodiesel will be Rs 65514 per year compared to Rs 87120 with diesel. Annual saving of Rs 21606 is possible by running the engine with B50. 4.3 Exhaust gas temperature Figure 10 to 12 show the variation exhaust gas Figure 8 : Brake specific fuel consumption at temperature with load on engine for diesel and brake load (B70) different biodiesel blends. It was observed that the exhaust gas temperature increases with load because more fuel is burnt at higher loads to meet the power requirements. For B50 (6:1), it increased from 127oC to 300oC. It was also observed that the exhaust gas temperature increases with percentage of biodiesel in the test fuel for all the loads. For B50 (6:1), the maximum exhaust gas temperature was 300oC and same was 320oC for B100 (6:1). This is because biodiesel constitutes of poor volatility, which burns during the late combustion phase. For molar ratio 3:1, exhaust gas temperature found was the highest 341oC followed by 325oC for 9:1 and 320oC for 6:1.
Figure 9 : Brake specific fuel consumption at brake load (B100)
The minimum brake specific fuel consumption was observed at rated power output for all fuels. In all cases the brake specific fuel consumption of biodiesel is higher than that of diesel. The minimum BSFC for B50 with molar ratio 9:1 was 0.284 kg/kWh as against 0.27 for diesel. The minimum BSFC for B70 with different molar ratios varied in range 0.375 kg/kWh to 0.301 kg/kWh. For B100, the minimum BSFCs were Figure 10 : Exhaust gas temperature at brake in the range 0.435 kg/kWh to 0.340 kg/kWh. The load (B50) Jagannath Hirkude et al. : Performance Analysis of Diesel Engine 21 with Mixtures of Waste Fried Oil and Methanol Molar (Biodiesel)
performance of biodiesel in a diesel engine, the following conclusions could be drawn. 1. Biodiesel B50 with molar ratio of 6:1 yielded the highest thermal efficiency 30.2% closer to mineral diesel (31.6%) at rated load. 2. For biodiesel B50 with molar ratio of 6:1, the lowest brake specific fuel consumption was 0.284 kg/kWh compared with 0.27 kg/ kWh of diesel. 3. The highest exhaust gas temperature was observed for molar ratio of 3:1(341oC) followed by 9:1(325oC) and 6:1(320oC) and Figure 11 : Exhaust gas temperature at brake mineral diesel (290oC). load (B70) 4. For daily 6 hours operation for 300 days, it is possible to save Rs 21606 by running the engine on B50(6:1) mode. For biodiesel B50 with molar ratio of 6:1, the lowest brake specific fuel consumption was 0.284 kg/kWh compared with 0.27 kg/kWh of diesel.
6. REFERENCES 1. R. Altin, C. Selim. The potential of using vegetable oil fuels as diesel engines. Energy Convers. Manag. 2001; 42(5): PP 529–538 2. H. Fukuda, A. Kondo, H. Noda. Biodiesel fuel production by transesterification of oils. Journal of Bioscience and Bioengineering 2001; 92(5): Figure 12 : Exhaust gas temperature at brake PP 405–416. load (B100) 3. W. Charusiri, W. Yongchareon, T. Vitidsant. 5. COLCLUSIONS Conversion of used vegetable oils to liquid fuels and chemicals over HZSM-5, sulfated zirconia The prospect of waste fried oil based fuel and hybrid catalysts. Korean Journal of Chemical production looks very attractive for energy engineering 2006; 23: PP 349-355. conversion in a developing country like India. Cost of conversion of biodiesel from waste fried oil was 4. M. S. Graboski, R. L. McCormick. Combustion Rs. 16.3 per liter compared with subsidized market of fat and vegetable oil derived fuels in diesel price of diesel approx. Rs. 40 per liter. Waste fried engines. Prog. Energy Comb. Sci. 1998; 24: oil to methanol molar ratio of 6:1 for PP 125-164. transestrification satisfies the important fuel 5. M. E. Gonzalez Gomez, R. Howard-Hildige, J. properties as per ASTM specifications of biodiesel. J. Leahy, T. O’reilly, B. Supple, M. Malone. Based on the experimental investigation on the Emission and performance characteristics of a 22 Journal of Advances in Engineering Science Section B (2), July - December 2010
2 litre toyota diesel van operating on esterified waste cooking oil and mineral diesel fuel. Environmental Monitoring and Assessment 2000; 65: PP 13–20. 6. T. Murayama. Evaluating vegetable oils as a diesel fuel. INFORM 1994; 5(10): PP 1138 - 1145. 7. Xiangmei Meng, Guanyi Chen, Yonghong Wang, Biodiesel production from waste cooking oil via alkali catalyst and its engine test, Fuel Process. Technol. 2008, 89 (9): PP 851-857. 8. M. Mittelbach, S. Gangl. Long storage stability of biodiesel made from rapeseed and used frying oil. J. Am. Oil Chem. Soc. 2001; 78: PP 573-577. 9. K.T. Lee, T.A. Foglia, K.S. Chang. Production of alkyl ester as biodiesel from fractionated lard and restaurant grease. J. Am. Oil Chem. Soc. 2002; 79: PP 191–195. 10. M. Mittelbach, H. Enzelsberger. Transesterification of heated rapeseed oil for extending diesel fuel. J. Am. Oil Chem. Soc. 1999; 76: PP 545–550. 11. M.I. Al-Widyan, A.O. Al-Shyoukh. Experimental evaluation of the transesterification of waste palm oil into biodiesel. Bioresour. Technol. 2002; 85: PP 253–256. A.G. Kamble et al. : Prediction and Optimization Journal of Advances in Engineering Science23 of Spur Gear Pair by Response Surface Method Section B (1), July - December 2010, PP 23-28
PREDICTION AND OPTIMIZATION OF SPUR GEAR PAIR BY RESPONSE SURFACE METHOD
A.G.Kamble1*, R.Venkata Rao2, A.S. Potdar3 and A.D. Lokhande4
1 - Dept. of Mech. Engg., SIT, Lonavala, Maha. INDIA 2 - Dept. of Mech. Engg., SVNIT, Surat, Guj. INDIA 3 - Dept. of Mech. Engg., SIT, Lonavala, Maha. INDIA 4 - Dept. of Mech. Engg., Maha. INDIA
ABSTRACT
Spur gear design is one of complex and time consuming design procedure. It will be great if it is automated by computer application. We can so reduce human error and make process faster. This paper presents development of mathematical models to predict module, number of teeth of pinion and gear of medium size external spur gear pair of 20 degree full depth teeth. This paper presents unique method to investigate engineering problem, its analysis, mathematical modelling and optimization with the help of RSM-response surface methodology and design of experiments (DOE). In the first part of our project we developed software which will assist in designing spur gear pair. We are taking dimensions as observations generated at the end. The second part constitutes study of relations between variables and development of mathematical models to predict dimensions directly without following design procedure. Finally we worked on optimization of design. Observations obtained from models are in compromising accuracy with actual design observations.
I. INTRODUCTION objective functions to optimize design variables were used. Spur gear design takes lot of factors into consideration. Lot of research work has been done on 2. LITERATURE REVIEW profile behavior, analysis of composite profiles, wear G. Cockerham [1] developed a program for reduction, reduction in transmission error, vibration designing spur gear of 20 degree pressure angle. The analysis, compact gearbox designs and root-stress program was fully integrated, requiring only design analysis etc. to understand different phenomenon of specifications and material properties to be supplied. spur gears. Our objective is to study and observe B.S.Tong et.al.[2] also presented an interactive different interaction effects of variables of spur gear program to design internal spur gear pairs. From a design by RSM- response surface method and specification, the program first performs a kinematic optimize it. Here we are finding center distance at analysis to determine teeth numbers and to satisfy which we can get minimum number of teeth, centre distance requirements. maximum module and at the same time maximum power transmission capability. Initially we designed a B.S. Tong and D. Walton [3] worked on set of computer program which can design a pair of external design variables are defined in terms of the number of spur gears of medium size to observe variations pinion and annulus teeth and the module. The between different variables. objective functions of minimum centre distance and volume were expressed. Some special search DOE-design of experiments method has been strategies were presented in order to solve the used to take observations. Observations are then problems of a discrete number of variables and to analyzed and different graphs have been plotted for reduce the calculation time. Hunglin Wanga et.al. [4] mathematical model. Single as well as global described in his paper, the mathematical formulation
*avi_nash1000@ yahoo.co.in 24 Journal of Advances in Engineering Science Section B (1), July - December 2010 and an algorithmic procedure to solve this multiple module, no. of teeth on pinion and no. of teeth on ear objective gear design problem. are response factors and shaft distance, rpm of To design and optimize multi-spindle gear trains pinion, rpm of gear, power output are input factors. of the non speed change type gear, S. Prayoonrat and 3.2 Development of Spur Gear Design D. Walton [5] proposed another algorithm. The Program designer may choose to optimize gear trains on the A spur gear design program is developed in C++ basis of minimum overall centre distance, minimum coding language. Spur gear beam strength, wear overall size, minimum gear volume, or other desirable strength, surface hardness etc. is calculated by the criteria, such as maximum contact or overlap ratios. program and finally it gives dimensions of gear pair. The method is based on a two-stage optimization Inputs given to the program are center distance, rpm procedure. A direct search method was used. of pinion (smaller) and gear (bigger) and power. Other Takao Yokota [6] optimized weight of gear by inputs are material elasticity, factor of safety, pressure genetic algorithm. Similarly Cevdet Gologlua et.al.[7] angle etc. Program consists of inbuilt design data automated preliminary design by genetic algorithm. tables. Required values are chosen by program itself Relation between wear and tooth width modification wherever necessary during execution. Inbuilt design is spur gear was well presented by Huseyin et.al [8]. data tables are user material selection, form factor 3. PLAN OF INVESTIGATION selection, grade of gears, deformation constant and pitch error calculation [9]. Program is developed to 3.1 Identification of Process Parameter design spur gears according to ISO 6336 (B) and 3.2 Development of Spur Gear Design Program on AGMA standards [10]. Program shows failure of C++. design when dynamic loads are equal or more than the 3.3 Finding the limits of the Process Parameters. beam strength or wear strength. If in any case during program execution design fails, then user can change 3.4 Development of Design Matrix and Conduction module and manufacturing grade number to design Experiments. foolproof gear pair. 3.5 Development of Mathematical Models. Range of input design parameters is follows: 3.6 Checking Adequacy of Models (ANOVA). power variation from 5 to 9 kW, shaft center distance 3.7 Development of Final Model. varies between 200 to 300 mm, rpm of pinion varies 3.1 Identification of Process Parameter between 1000 to 1400 rpm, and for gear it varies within 250 to 350 rpm. Other constant design parameters Center distances, rpm of pinion and gear and are starting torque multiplication factor = 1.2, bearing power source are independent factors. While module, stress = 200 N/mm2. Factor of safety = 1.5, Grade of number of teeth are depending upon above four factors gears = 6, Tooth form = 200 full depth, Material for and depends on each other. both pinion and gear is steel, Elastic limit 206000 According to M.F. Spotts [10] beam strength is N/mm2, tooth width = 10 *m. given by S = m*b*ó*y, where m = module, b = tooth b 3.3 Finding the Limits of the Process width, ó = bearing stress (= ultimate tensile strength/ Parameters 3). Our objective is to maximized module to increase beam strength because of direct proportionality and Trial runs were carried out by varying one of the to minimize number of teeth which is inversely process parameters while keeping the rest of them at proportional to module. constant values. The upper limit of a factor was coded as +2 and the lower limit as -2. The coded values for While other geometrical factors like addendum, intermediate values were calculated from the dedendum, tooth thickness are according to following relationship given below: Xi = 2 [ 2X - (Xmax empirical standard relationship. But they are not + Xmin) ] / [ Xmax – Xmin ] Where Xi is the required considered as deciding factors for gear design. So A.G. Kamble et al. : Prediction and Optimization 25 of Spur Gear Pair by Response Surface Method coded value of the variable X; X is any value of the Input factors Output factors variable from Xmin to Xmax; Xmin is the lower level of SD RP RG PO MO NP NG the variable & Xmax is the upper level of the variable. 225 1,100 275 6 3 31 21 The process parameters levels with their units and 225 1,100 275 8 4 23 91 notations are given in table1. 225 1,100 325 6 3 35 116 225 1,100 325 8 4 26 87 225 1,300 275 6 3 27 24 225 1,300 275 8 4 20 93 225 1,300 325 6 3 31 21 275 1,300 325 8 4 23 91 275 1,100 275 6 3 37 47 275 1,100 275 8 4 28 111 275 1,100 325 6 3 42 42 275 1,100 325 8 4 32 07 275 1,300 275 6 3 33 52 275 1,300 275 8 4 25 114 TABLE 1 275 1,300 325 6 3 37 47 3.4 Developing Design Matrix and Conducting 275 1,300 325 8 4 28 111 Experiments 200 1,200 300 7 4 21 81 The selected design matrix, shown in table 2, is 300 1,200 300 7 3 41 61 four factors, five levels composite rotatable factorial 250 1,000 300 7 4 29 97 design was selected for conducting the experiment. 250 1,400 250 7 4 23 03 The design matrix comprises a full replication of 250 1,200 350 7 4 22 04 4 2 = 16 factorial design, plus seven centre points and 250 1,200 300 7 3 38 30 eight star points. Therefore, experimental design 250 1,200 300 5 3 34 34 consist of 31 (16+7+8=31) experimental runs. 250 1,200 300 9 4 26 01 Observations in table 2 are modified to nearest higher 250 1,200 300 7 4 26 01 values; viz. for module if it is 3.56 then we are choosing 4, similarly for number of teeth also means 250 1,200 300 7 4 26 01 for 36.67 we are taking as 37. After changing such 250 1,200 300 7 4 26 01 values of teeth and module we are getting 2 to 4 mm 250 1,200 300 7 4 26 01 more center distance than we are giving as input. 250 1,200 300 7 4 26 01 Moreover above observations can be taken readily if 250 1,200 300 7 4 26 01 our gear material is changed; viz. elasticity, bearing 250 1,200 300 7 4 26 01 strength, starting torque multiplication factor, factor of TABLE 2 safety etc. The second order response surface model for the 3.5 Development of Mathematical Models three selected parameters is given by the equation Mathematical model developed is given by Y = f (SD, RP, RG, PO) where, Y is the measured response (MO = Module, NP = No. of teeth on pinion, NG = No. of teeth on gear) and The second order response surface model [12] SD is shaft distance, mm. could be expressed as: RP is rpm of pinion, rev/min Y = b0 + b1 SD+ b2 RP + b3 RG + b4 PO + b12 SD.RP + b13 RG is rpm of gear, rev/min SD.RG + b14 SD.PO + b23 RP.RG + b24 RP.PO + b34 RG.PO PO is power output, kW + b11SD.SD + b22RP.RP + b33RG.RG + b44PO.PO 26 Journal of Advances in Engineering Science Section B (1), July - December 2010
Where, b0 – Free term coefficient, b1, b 2, b3 and b4 1) Interaction effect on module due to center
– Linear coefficients, b11, b22, b33 and b44 – Quadratic distance and rpm of pinion: coefficients and b , b , b , b , b and b – 12 13 14 23 24 34 As shown in Fig. 1, as shaft distance increased, Interaction coefficients. module increases. While if we keep our center 3.6 Checking Adequacy of Models distance constant and increase rpm, module line is Adequacy of the models [14] was then tested by almost vertical. It means that rather rpm of pinion, analysis of variance (ANOVA) given in Table 3. The shaft distance is a major deciding factor for module. value of R2 (adjusted), squared multiple R and F-ratio is high enough to be confident for better fit. Half (linear) model Factors %R2 %R2adj. SS SEE F-ratio MO 61.4 55.5 4.36 0.324 10.36 NP 66.7 61.5 680.04 3.617 2.99 NG 73.7 69.6 9330.6 11.32 18.19 Full model MO 74.1 51.5 5.26 0.339 3.28 Figure1 Figure 2 NP 76 55.1 775.82 3.91 3.63 NG 83.9 69.8 10625.2 11.29 5.95 2) Interaction effect on module due to power TABLE 3 and shaft distance The estimate of error is also seemed to be less. Fig 2 shows, module is highly sensitive to power Linear regression models are fitting best than any input. Increase in power lead to increase in module other model and also reduce cumbersome (near to 5). If we are keeping power input constant and mathematical calculation for prediction. try to increase our center distance, a little reduction 3.7 Development of Final Model in module is seen. Maximising module needs to increase in power and at the same time decrease After certain experiments, it has been found that shaft distance. Thus both have opposite effect on linear regression model suits best than full model so module. we have chosen model given below: 3) Interaction effect on module due to rpm of Y= b + b SD+ b RP + b RG + b PO 0 1 2 3 4 pinion and gear Final mathematical models are as below. These models are developed with the help of SYSTAT12 We can see from the Fig. 3, that variation in rpm software package [13]. The input design parameters of pinion and rpm of gear is not affecting so much to are in their uncoded form. the module. Module varies within 3.6 to 3.7 for the entire range of rpm of pinion and gear. MO = 1.37-0.00365 SD+0.000091 RP+0.00036 RG + 0.424 PO NP = 36.6 + 0.142 SD - 0.0195 RP + 0.0218RG - 3.80PO 4) Interaction effect on number of teeth due NG = 86 + 0.566 SD + 0.0022 RP - 0.0940 RG - 15.0 PO to center distance and power 4. Results and Discussion Fig.4 shows that, if shaft distance is decreased Standard error is an inherent error in the model number of teeth decreases. Power is having negative developed which is less in this given model than any and shaft distance is having positive effect on number other model previously developed. It is necessary to of teeth. This means, increase in power will definitely minimize standard error. Higher F-ratio and almost lead to reduce number of teeth while there is a zero P-value in ANOVA (Table 3) describe better reverse phenomenon when center distance is to be conformity of results. increased. A.G. Kamble et al. : Prediction and Optimization 27 of Spur Gear Pair by Response Surface Method
Global or Multi Objective Optimization Optimum value of maximum module to transmit maximum power and with minimum number of teeth of pinion & gear are given in table 5.
Input Output SD RP RG PO MO NP NG 200 1400 300 9 4.71 -- -- 249.85 1400 350 9 -- 18.17 -- Figure 3 Figure 4 200 1000 350 9 -- -- 8.37 5) Interaction effect on no. of teeth of pinion TABLE 4 due to rpm of pinion and power Input Output From Fig. 5, we can understand that rpm of pinion SD RP RG PO MO NP NG and power has negative effect on number of teeth of 200 1000 350 9 4.5 18 77.56 pinion. Increasing one or both will gradually reduce number of teeth on pinion. It is similar for gear also. TABLE 5 The range for predicted response shows 95% confident results. If we calculate for shaft distance according to predicted no. of teeth of pinion and gear; [(4.5*20) + (4.5*90)] /2 = 247.5 mm, while given shaft distance is 250 mm. So there is little variation between predicted and actual values. By above results we can conclude that our shaft distance in not be minimum (200 in experiment) every time to optimize gear pair design but at the same time it could transmit maximum power with enough beam strength. Moreover observations are also in modified form means rounding off to nearest higher value as Figure 5 explained previously.
NP actual By Model. NG actual By Model. Thus we can say that above results are to be modified to suit standard values of design data books. 28 31.5 113 110.5 Table 6 shows comparison of actual values calculated 24 27.89 107 101.2 by design and values obtained by model, they have 26 29 90 86.5 little variation but still under acceptable range. TABLE 6
5. OPTIMIZATION 6. CONCLUSION Single and multi objective optimization were done 1. Mathematical models for module, number of with the help of MINITAB software package. teeth of pinion and gear can give direct values without going for long design process, thus it Single Objective Optimization saves designer’s time and efforts. Maximum value of module, minimum number of 2. Predicted results are verified by testing, they gear teeth and minimum number of pinion teeth is are reasonably accurate for new observation obtained when single objective optimization problem within range. Optimized parameters for is considered. Optimum values are given in table 4. 28 Journal of Advances in Engineering Science Section B (1), July - December 2010
Mechanism and Machine Theory, Volume 29, maximum power transmission for given Issue 7, October 1994, PP 1071-1080 conditions are SD, RP, RG, PO simultaneously 250 mm, 1400, 350 and 9 kW 5 “Practical approach to optimum gear train design”. By S. Prayoonrat and D. Walton 3. Designer can easily predict dimensions of Computer-Aided Design, Volume 20, Issue 2, spur gear pair. Little modification can finalize March 1988, PP 83-92 design keeping other conditions same. 6 “A solution method for optimal weight design 4. This study and relations can be very useful problem of the gear using genetic algorithms.” to designer to have optimum spur gear pair By Takao Yokota, Takeaki Taguchi and Mitsuo with maximum power output. Gen. Science direct online journals. 5. There are future prospects for optimization 7 “Genetic approach to automate preliminary of other automotives, machine tools and design of gear drives” By Cevdet Gologlua and industrial transmission systems. Metin Zeyvelib. Science Direct online journals. 8 “Relation between wear and tooth width REFRENCES modification in spur gears” By Hüseyin Ýmreka, 1 “Computer-aided design of spur or helical gear and Hayrettin Düzcükoðlub. Science direct online train.” G. Cockerham, D. Waite. Computer-Aided journals. Design, Volume 8, Issue 2, April 1976, 9 Design data handbook for mechanical engineers. PP 84-88 Coimbatore institute of technology, Coimbatore 2 “The optimisation of internal gears” By B.S. 10 Machine element design. Prof. M.F. Spotts2003. Tong and D. Walton. International Journal of Machine Tools and Manufacture, Volume 27, 11 Fisher R.A. 1952 statistical methods for research Issue 4, 1987, PP 491-504 Workers, 12th edition. Edinburgh, Oliver and Boyd. 3 “A computer design aid for internal spur and helical gears” By B.S. Tong and D. Walton. 12 J. Arora. Introduction to optimum design. International Journal of Machine Tools and McGraw hill, 1989 Manufacture, Volume 27, Issue 4, 1987, 13 SYSTAT version 12, Systat, Inc. PP 479-489 14 Davis. O.L. 1978. The design and analysis of 4 “Optimal engineering design of spur gear sets” industrial Experiments. Longman. By Hunglin Wanga and Hsu-Pin Wanga, Jaya T. et al. : The Buried Gate MESFET Journal of Advances in Engineering Science29 with Turning on Characteristic Section C (4), July - December 2010, PP 29-32
THE BURIED GATE MESFET WITH TURNING ON CHARACTERISTIC
Jaya T.1* and Kannan V.2
1 - Research Scholar, ECE, Sathyabama University, Chennai, INDIA 2 - Prof & Head of VLSI Dept, Sathyabama University, Chennai, INDIA
ABSTRACT
The buried-gate GaAs MESFET with front illumination using turning ON channel current been analyzed by solving continuity equation. This analysis includes the ion implanted buried-gate process. At time ‘t’ is equal to zero, the light through the optical fiber is turning ‘ON’ has been considered. The channel current has been evaluated and discussed. Buried-gate optical field effect transistor (OPFET) will be highly suitable for microwave communication.
1. INTRODUCTION 2. OVERVIEW OF BURIED GATE MESFET The Metal-Semiconductor-Field-Effect-Transistor Basically, the characteristic of he buried-gate (MESFET) consists of a conducting channel GaAs MESFET had been analyzed in [2],[6].In that, positioned between a source and drain contact they are mentioned only the characteristic of region. The control of the channel current is obtained MESFET with and without illumination. The sensitivity by varying the turning the light through buried -gate of the device depends on the absorption coefficient of GaAs MESFET with front illumination. light. There are different ways by which light may be absorbed within the material. The use of GaAs rather than silicon MESFETs provides two more advantages that are significant: The conventional way of illuminating the MESFET first, the electron mobility at room temperature is more is the front illumination with transparent/ semitransparent gate or opaque gate [5]. However, for than 5 times larger, while the peak electron velocity enhanced absorption in MESFET, de Salles [3] has is about twice that of silicon. Second, it is possible to suggested two alternatives: 1) the device may be fabricate semi-insulating (SI) GaAs substrates, which illuminated from the back where the fiber may be eliminates the problem of absorbing microwave power inserted partially or fully into the substrate of the in the substrate due to free carrier absorption. device and 2) the buried gate MESFET with front In this paper, we have analyzed theoretically the illumination. effect of the time dependent characteristics of ion- 3. RELATED WORK implanted buried gate GaAs MESFET with front illumination. We consider the ion-implanted profile in The work has been developed in the field of studying the active region. and analyzed the field of optically controlled GaAs Metal Field effect Transistor (OPFET) [1] and [2]. In [1] DC Present calculation shows that the buried gate characteristics of buried-gate GaAs MESFETs are GaAs MESFET with back illumination represents still studied by solving dc continuity equation. In this work, better performance compared to the results of the the optical fiber has been inserted up to active layer of conventional front illumination. buried-gate GaAs MESFET.
*[email protected] 30 Journal of Advances in Engineering Science Section C (4), July - December 2010
In this work, the drain –source voltage controlled The generated rate and recombination rates is, the current –voltage characteristics of buried-gate GaAs MESFET. Shuba et al.[2]developed the first dc (1) model of an ion-implanted MESFET. The effects of varying the light intensity on Si OPFET parameters and J and J are the electron and hole current were studied by Singh et al. DC models was further n P densities, respectively defined by, developed introducing the effect of surface recombination on the device parameters of an ion- implanted OPFET S. H. Lo and C. P. Lee. (2) 4. PERFORMANCE METRICS And The optical effect on the buried-gate MESFET is shown in Fig. 1. The optical fiber is inserted in to the buried-gate MESFET. The light is incident on the (3) buried-gate MESFET without any deviation through optical fiber. In the above equation include drift and diffusion terms. 4.2 Channel Charge current due to Light is Turned ON In the depletion region, the continuity equation is applied in the first order form
(4)
Let us take (4), we consider two cases: When light is turned on at t = 0, the boundary condition becomes at t = 0, (5) Using this, (4) gives the solution
Figure 1 : Schematic Structure of MESFET (6)
Where The fundamental physical mechanism arising in optical illumination of the MESFET is the production and may be defined as the optical of free carriers within the semiconductor material when light of photon energy equal to or greater than relaxation time for the OPFET. the semiconductor material band gap energy is The corresponding channel charge due to ion- absorbed. implantation is obtained at t = 0, light is turned on is 4.1 Continuity Equation (7) Under illumination the photo generated electrons and a hole in the neutral and depletion regions has In the above (7), N(y) is obtained from (6) and is obtained by solving the dc continuity equations. obtained from (4) is obtained as Jaya T. et al. : The Buried Gate MESFET 31 with Turning on Characteristic
4.3 Total Channel Current with Ion-Implantation (8) The semi-insulating substrate is p-type doped and has uniform doping profile, which is represented by
Where Y11 is the distance from the surface to the modified edge of the gate depletion region, due to (10) photo voltage developed across the Scotty junction of the buried gate. Where Q, Rp and σ are the implanted dose per unit area, projected range and straggle parameter in (9) length respectively. The channel current due to ion-implantation is obtained using the relation at t = 0, light is turned on Table I : Parameter Values
Device Parameter Symbols Values (11)
Photon absorption á 1.0x106 m-1 Substitutes (7) in (11), we get coefficient
5 (12) Carrier velocity in vy 1.2x10 m/s y-direction Where
23 -3 Equivalent constant N Dr 0.658x10 m VD is the channel voltage doping profile 5. RESULTS AND DISCUSSIONS The Fig. 2 gives the plot of channel current versus Life time of electrons ô 1.0x10-6 s n time when the light is turned ON at flux density 16 2 -8 (Ô = 10 / m s) and at a fixed absorption coefficient Life time of holes ôp 1.0x10 s (á= 1.0x106 m-1)). The time for reaching steady state Straggle parameter ó 0.383x10-6 m value varies from 10-20 ps. For this case of “light ON,” the channel current required to reach the steady state Permittivity å 1.04x10-10 f/m corresponds to time above 30ps.
Total effective thickness ts 0.15 ìm of active layer
Projected range Rp 0.861x10-7 m
Schottky barrier height ÔB 0.9 eV
15 -2 Trap density Nt 1.0x10 m
Channel width Z 100x10-6 m
Channel length L 3x10-6 m
2 Electron mobility ì n 0.85 m /v.s
2 Hole mobility ì p 0.04 m /v.s
Active layer thickness a 0.25 ìm Figure : 2 Channel current versus time when the light is turned” “ON” 32 Journal of Advances in Engineering Science Section C (4), July - December 2010
2. Nandita Saha Roy, B.B. Pal and R.U.Khan, 6. CONCLUSIONS & FUTURE WORK “Analysis of GaAs OPFET with Improved Optical The time – dependent characteristics of Absorption under Back Illumination,” IEEE buried-gate GaAs MESFET with ion-implanted Transactions Election Devices,Vol.46, profile under front illumination has carried out. PP. 2350-2353,Dec. 1999. The light is incident on the buried-gate MESFET 3. A. A. A. de Salles, “Optical control of GaAs without any deviation through optical fiber. At MESFET’s,” IEEE Trans. Microw. Theory Tech., time‘t’ is equal to zero, we have been studied: Vol. MTT-31, PP. 812–820, Oct. 1983. light is turned “ON,” .When light is turned “ON,” the parameters reach the steady state value at 4. S. H. Lo and C. P. Lee, “Numerical analysis of a lesser time than that when light is turned off. the photo effects in GaAs MESFETs,” IEEE This results in 1) additional generation of carriers Transactions Electron Devices, Vol.39, PP. due to photogenation, and 2) widening of channel 1564-1570, July 1992. width. Buried-gate optical field effect transistor 5. Shubha, B. B. Pal, and R. U. Khan, “Optically (OPFET) will be highly suitable for power device controlled ion-implanted GaAs MESFET application and microwave communication. characteristics with opaque gate,” IEEE Trans. In future, we will planed the buried-gate GaAs Electron Devices, Vol. 45, PP. 78–84, Jan. 1998. MESFET with front illumination to analysis with 6. V. K. Singh, S. N. Chattopadhyay, and B. B. light is turned “OFF,” Pal, “Optically-controlled characteristics of an ion-implanted Si MESFET,” Solid-State 7. REFERENCES Electronics, Vol.29, PP. 707-711, 1986. 1. Jaya.T, Kannan.V,” Analytical Model for I-V 7. Walter R.Curtice, “A MESFET model for use in Characteristic of Buried Gate MESFET,” IJAST, the design of GaAs integrated circuits” IEEE Vol.17, PP. 2350-2353, JULY 2010. Transactions Microwave Theory Tech., Vol. MTT- 28, PP. 448-456, may 1980. Sachin R. Jadhav et al. : Discriminative Common Journal of Advances in Engineering Science33 Vectors for Face Recognition Using Iterative Approach Section C (2), July - December 2010, PP 33-40
DISCRIMINATIVE COMMON VECTORS FOR FACE RECOGNITION USING ITERATIVE APPROACH
Mr. Sachin R. Jadhav1 *, Prof. D.R.Ingle2, Prof. Naresh Kumar Harale3 and Prof. Vijay Bhosale4
1 - Sinhgad Institute of Technology, Lonavala, Pune, Maharashtra State, India. 2 - Bharati Vidyapeeth’s College of Engineering, Navi Mumbai, Maharashtra State, India 3 - M.G.M. College of Engineering & Technology, Kamothe, Navi Mumbai, Maharashtra State, India 4 - M.G.M. College of Engineering & Technology, Kamothe, Navi Mumbai, Maharashtra State, India
ABSTRACT
Face recognition is the process of identifying individuals from images of their faces by using a stored database of faces labeled with people’s identities. Since face images are similar, they are correlated; therefore can be represented in a lower dimensional subspace called feature space without loosing a significant amount of information. LDA method cannot be directly applied because of “small sample size problem”. To overcome this problem the discriminative common vectors (DCV) approach is proposed. DCV approach is based on a variation of Fisher’s Linear Discriminant Analysis. In this the common vectors are extracted by eliminating the differences of the image samples in each class of images. Then the DCV which will be used for classification are obtained from the common vectors. In this paper, Iterative hierarchical classification by using Discriminative Common Vectors approach is suggested. In this method the Common Vectors obtained from the earlier iteration are grouped together according to their classes and then used as input for the next iteration to obtain the DCV of these classes. The number of iterations required is equal to the number of hierarchy levels in the training set. Key words : Discriminative Common Vectors, face recognition, common vectors, Iteration Matrix.
I. INTRODUCTION recognition problem, like recognizing only frontal static images taken in controlled conditions etc. Machine recognition of faces from still and video images is emerging as an active research area There are two major types of feature extraction spanning several disciplines such as image algorithms. One type of these algorithms, extract processing, pattern recognition, computer vision and statistical features such as Karhunen- Loeve neural networks. Although humans seem to recognize transform and singular value decomposition faces in cluttered scenes with relative ease, machine coefficients. They are also known as appearance recognition is a much more daunting task. The general based methods. The other type of these algorithms, statement of the problem is as follows: Given still or extract structural features like eyes, nose, lips, and video images of a scene, identify one or more persons points of high curvature. In this project we are in the scene using a stored database of faces. interested in appearance based feature extraction Available collateral information such as race, age and methods. Two major appearance based feature gender may be used in narrowing the search. extraction methods are Principal Component Analysis (PCA) and Linear Discriminant Analysis Face recognition is a complex task due to various (LDA). Both these methods have their own factors. Extensive research is being conducted to disadvantages and are overcome in the Discriminative overcome each of these factors. To simplify the design Common Vectors approach PCA also known as of a face recognition system most current face Eigen face method is the conventional method used recognition systems handle only a subset of the face
*Mr. Sachin R.Jadhav 34 Journal of Advances in Engineering Science Section C (2), July - December 2010 for dimensionality reduction. This yields projection projection vectors. Therefore, this method can only be directions that maximize the total scatter across all used when the dimension of the sample space is classes i.e. across all images of all faces. This larger than the rank of SW. method also maximizes within class scatter which is not suitable for classification. The projected feature Let, space retains variation due to lighting and the classes may be smeared together. LDA attempts to find projection directions that minimize within class scatter and maximize between class scatter. But since the dimension of the sample space is typically larger than the number of samples Now, we define S , S and S as follows : in the training set the within class scatter matrix is W B T singular. Therefore LDA method cannot be directly applied. This is known as the “small sample size problem”. To overcome this problem the discriminative common vectors approach is proposed. The discriminative common vectors approach is based on a variation of Fisher’s Linear Discriminant Analysis. In this the common vectors are extracted by eliminating the differences of the image samples in each class of images. Then the discriminative common vectors which will be used for classification are obtained from the common vectors. The criterion used by the Discriminative Common Finally a method for performing hierarchical Vectors approach is classification by using Discriminative Common Vectors approach iteratively is suggested. In this method the Common Vectors obtained from the earlier iteration are grouped together according to their classes and then used as input for the next iteration To find the optimal projection vectors W in the null to obtain the Discriminative Common Vectors of these space of S , we project the face samples onto the null classes. The number of iterations required is equal to W space of S and then obtain the projection vectors by the number of hierarchy levels in the training set. W performing PCA. 2. PROPOSED DCV APPROACH 2.1.DCV BY USING THE RANGE SPACE OF S The common vectors approach was originally W used for isolated word recognition problems where the For projecting the face samples onto the null number of samples in each class was less than or space of SW, vectors that span the null space of SW equal to the dimensionality of the sample space. In must first be computed. But, the dimension of the null these methods common properties are obtained by space can be very large (as all human faces have eliminating the differences in the samples of the same similar appearance in general); therefore it is very class. After finding the common vectors, a new set of difficult to find vectors that span the null space of SW. vectors called ‘the discriminative common vectors’ are A more efficient way to achieve the projections of face obtained from the common vectors. These are used samples onto the null space of S is to use the for classification. This method addresses the W orthogonal complement of the null space which is limitations of other methods that use the Null space generally a significantly lower dimensional space. of SW (like the Null Space method) to find the optimal Sachin R. Jadhav et al. : Discriminative Common 35 Vectors for Face Recognition Using Iterative Approach
2.1.1 Finding Common Vectors 2.1.2 Finding Discriminative Common Vectors
Let, After obtaining the common vectors , optimal projection vectors will be those that maximize the total scatter of the common vectors,
Where W is a matrix whose columns are the orthonormal optimal projection vectors w , S and is an orthonormal set, and k com is the set of eigenvectors corresponding is the scatter matrix of the common vectors, to the nonzero Eigen values of .
Consider the matrices and Where µ is the mean of all common vectors, . Since , every com face image has a unique decomposition of the form
Now the optimal projection vectors can be
found by an Eigen analysis of Scom. In particular, all eigenvectors corresponding to the nonzero Eigen where values of Scom will be the optimal projection vectors. There will be C-1 optimal projection vectors since the and P and are rank of Scom is C-1 if all common vectors are linearly the orthogonal projection operators onto V and , independent. If two common vectors are identical, respectively. Our goal is to compute then the two classes, which are represented by this vector, cannot be distinguished. Since the optimal projection vectors belong to the null space of , it follows that when image samples of the ith class are To do this, we need to find a basis for V, which can projected onto the linear span of the projection vectors be accomplished by an Eigen analysis of SW. In particular, the normalized eigenvectors αk , the feature vector corresponding to the nonzero Eigen values of S will W of the projection coefficients will also be be an orthonormal basis for V. There will be at most M - C eigenvectors corresponding to the nonzero Eigen independent of the sample index m. Thus we have values of SW (Since rank of SW is at most M - C).
We call the feature vectors discriminative common vectors, and they will be used for classification of face images. The fact that does In this way we obtain the same unique vector for not depend on the index m guarantees 100% all samples of the same class, and we refer to the accuracy in the recognition of the samples in the vectors as the common vectors. training set. 36 Journal of Advances in Engineering Science Section C (2), July - December 2010
2.2.DCV by Using Difference Subspaces and normalized by using the Gram-Schmidt Gram-Schmidt Orthogonalization procedure. orthogonalization procedure to obtain orthonormal basis vectors . The complement of Discriminative Common Vectors approach using is the indifference subspace such that the range space of SW may become computationally expensive and numerically unstable for large values of M. Since we do not need to find the Eigen values of SW, the difference subspaces and Gram-Schmidt orthogonalization procedure can be used for finding the common vectors efficiently. Where P and are orthogonal projection operators onto B and , respectively. Thus 2.2.1 Finding Common Vectors matrices P and are symmetric and idempotent, In this method we choose any one of the image and satisfy . Any sample from each class vectors from the ith class as the subtrahend vector and can now be projected onto the indifference subspace then obtain the difference vectors of the so called to obtain the corresponding common vectors of difference subspace of the ith class. Thus assuming the classes. that the first sample of each class is taken as the subtrahend vector, we have
The difference subspace of the ith class is The common vectors do not depend on the choice defined as of the subtrahend vectors and they are identical to the common vectors obtained by using the null space of . These subspaces can be summed up to form the 2.2.2 Finding Discriminative Common Vectors complete difference subspace as defined below : After calculating the common vectors, the optimal projection vectors can be found by performing PCA. The eigenvectors corresponding to the nonzero The number of independent difference vectors eigenvectors of Scom will be the optimal projection will be equal to the rank of . For simplicity suppose vectors. However, optimal projection vectors can be there are independent difference vectors. obtained more efficiently by computing the basis of the Since it can be proved that B and the range space V difference subspace Bcom of the common vectors, of are the same spaces, the projection matrix onto since we are only interested in finding an orthonormal B is the same as the matrix (projection basis for the range of Scom. matrix onto the range space of ). This projection matrix can be calculated as shown below: 3. HIERARCHICAL CLASSIFICATION USING DCV When presented with a picture of an individual we where is a can identify that individual if he is already known to us. matrix. This calculation involves finding the But when the picture belongs to an unknown person inverse of a non singular, we at least are able to obtain other information like that individual's apparent age group, race, etc. This is positive definite symmetric matrix A possible because we have a prior knowledge computationally efficient method of applying the (acquired from earlier experiences) of features that projection uses an orthonormal basis for B. In belong to an individual of a particular age group, or an practice, the difference vectors can be Ortho individual of a particular race, etc. Sachin R. Jadhav et al. : Discriminative Common 37 Vectors for Face Recognition Using Iterative Approach
In hierarchical classification using DCV, we first common vectors of individuals of the same class as attempt to determine whether the given test image different images belonging to the same individual and belongs to any of the individuals within the training set. each class as one individual and then obtain the If it is found that the given test image does not belong common vector for each class. This is graphically to any of the individuals within the training set, an shown in figure 2. attempt is made to obtain collateral information about the individual within the test image. For performing hierarchical classification using DCV, we arrange the training set as a hierarchical collection of sets if image vectors. One example for such an arrangement with two levels of hierarchy is shown in figure 1.
Figure 1: Hierarchical arrangement of image vectors with two levels of hierarchy. Here (that is in figure 1), at the lowest level of the hierarchy of image vectors, we have all image vectors Figure 2 : Calculating common vectors assigned to various subjects. At the next higher level, hierarchically by eliminating we have all subjects assigned to various classes. A training set can have any number of such hierarchical distinct features. levels. 3.1. ALGORITHM Training is performed by using the DCV algorithm once for each level starting with the lowest level. The Based on this an iterative training algorithm to common vectors obtained from the earlier iteration are perform hierarchical classification is as stated below: used as input for the next iteration except for the first 1. Store the input image vectors in a matrix iteration where input image vectors are used as input. called X(1). Figure 3.2 shows how common vectors of an earlier iteration are used as inputs for the next iteration. The 2. Set a variable level_count to 1. common vectors are also used to find discriminative 3. Set a variable hierarchy_levels to number of common vectors at each level. levels in the training set. Hierarchical classification using DCV is based on 4. While level_count <= hierarchy_levels + 1 the idea that common vectors of subjects belonging perform the following to the same class will include features that are present in all individuals of that class and some 4.1.Find orthonormal basis for range space additional features that are unique to that individual. constructed using X(level_count) and If we eliminate the unique features of an individual from save them in a matrix U(level_count)(This his common vector we obtain the common vector of is sam as the matrix W of level_ his class. This can be done by considering the count -1). 38 Journal of Advances in Engineering Science Section C (2), July - December 2010
4.2.Find the common vectors for each 1.3.4. else subject or group at the level and save 1.3.4.1. Assign the test image to the them in the matrix X(level_count + 1). nearest subject or group. 4.3.If level_count > 1 then 1.3.4.2. Set identified to true. 4.3.1. Find feature vectors of previous level using U(level_count) and one image 1.3.5. Increment current_level by 1 vectors per each subject. 1.3.6. Set image vector to image vector 4.4. Increment level_count by 1. U(current_level)U(current_level)T Now the iterative testing algorithm to perform (image vector). hierarchical classification is as stated below: Threshold of highest level is set to maximum 1. For each image vector in the test set perform: possible value to ensure that all test images are at 1.1. Set identified to false. least assigned to any one of the groups at this level. 1.2. Set current_ level=1 1.3. While identified is false do 4. RESULTS AND CONCLUSION 1.3.1. Find the feature vector of the test In this research we have studied the image using the following equation: Discriminative Common Vectors approach for U(current_level+1)T* image vector. face recognition applications and its use to over 1.3.2. Find Euclidean distance between the come the small sample size problem feature vector of the test image and encountered in Linear Discriminant Analysis for feature vectors of subjects or groups face recognition. We have also proposed the used at that level. of Discriminative Common Vectors approach for hierarchical classification. Hierarchical 1.3.3. If the minimum of these distances is Classification allows us make best use of grater than a threshold(current_level) information available in a database allowing us to then even classify individuals not present in the 1.3.3.1. Set the identified group at database to obtain collateral information like age this level as stranger or group, race, etc. unknown.
Figure 3 : Operations performed while training on a test set with 2 levels of hierarchy. Sachin R. Jadhav et al. : Discriminative Common 39 Vectors for Face Recognition Using Iterative Approach
classification is attempted at a higher level. It is assumed that all input images belong to any one of the groups present in the top most class. The face detection and segmentation algorithms can be used to ensure this requirement in a face recognition system. An implementation of Iterative hierarchical classification using Discriminative Common Vectors was constructed using Matlab. This implementation can only handle two levels of hierarchy but implementations that can handle any number of hierarchy levels can be constructed. The constructed implementation was tested using on Yale University Databases to check its performance in test sets constructed with variation in illumination direction and expression. It was found that the algorithm worked as expected and the recognition efficiency is good when the illumination and expression variation are mild.
5. REFERENCES
1. Alex M. Martinez, (2001), "PCA versus LDA", IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 23, No. 2, pp. 228-233. 2. Bernard Kolman, (2001), "Introductory Linear Algebra with Applications", 7th Edition, Prentice Hall International, Inc. 3. Daniel L. Swets, (1996), "Using Discriminant Eigen features for Image Retrieval", IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 18, No. 8, pp. 831-836. 4. Hakan Cevikalp, (2005), "Discriminative Common Vectors for Face Recognition", IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 27. No. 1, pp. 4-13. 5. Jonathon Shlens, (2005), "A tutorial on Principal Figure 4 : Operations performed during testing with Component Analysis", http://www.snl.salk.edu/ 2 levels of hierarchy. ~shlens/pub/notes/pca.pdf
This is made possible by obtaining the features 6. M. Bilginer Gulmezoglu, (1999), "A Novel of individuals in one level (i.e. the lowest level), then Approach to Isolated Word Recognition", IEEE obtaining the features of groups in the higher levels. Trans. Speech and Audio Processing, Vol. 7, When classification fails at a lower level then No. 6, pp. 620-628. 40 Journal of Advances in Engineering Science Section C (2), July - December 2010
7. M. B. Gulmezoglu, (1999), "A Novel Approach to Isolated Word Recognition", IEEE Trans. Speech and Audio Processing, vol. 7, No. 6, pp. 620-628. 8. P. N. Belhumeur, (1997), "Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection", IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 19, No. 7, pp. 711-720. 9. Rama Chellappa, (1995), "Human and Machine Recognition of Faces: A Survey", Proc. IEEE, Vol. 83, pp. 705-740. 10. W. Zhao, (1998), "Discriminant Analysis of Principle Components for Face Recognition", Proc. Third IEEE Int'l Conf. Automatic Face and Gesture Recognition, pp. 336-341. S. B. Nimbekar et al. : Recognize Human Journal of Advances in Engineering Science41 Emotions: Recognition from DCT Section C (5), July - December 2010, PP 41-44
RECOGNIZE HUMAN EMOTIONS: RECOGNITION FROM DCT
Prof S. B Nimbekar1* and Prof D. D. Badgujar2
1 - Computer Engineering Department, Sinhgad Institute of Technology, Lonavala, Affiliation with Pune University, Maharashtra, India 2 - Information Technology Department, Sinhgad Institute of Technology, Lonavala, Affiliation with Pune University, Maharashtra, India
ABSTRACT
The ability to recognize emotion is one of the hallmarks of emotional intelligence, an aspect of human intelligence that has been argued to be even more important than mathematical and verbal intelligence. This paper proposes that machine intelligence needs to include emotional intelligence and demonstrate results toward this goal: developing a machines ability to recognize human emotion from given facial parameter. We describe difficult issues unique to obtain reliable affective data and collect a large set of data from a subject and experience each of two emotional states. This paper presents techniniqus for feature extraction and use algorithm for classification, which is on kernel based. We got 100% recognition accuracy on three classes of emotion, including neutral. Keywords : DCT, Mean, Entropy, Energy, SVM.
1. INTRODUCTION recognition in section3 and in section 4, simulation, result by using Neurosolution[11]. Human- Computer interaction will be more affective if a computer knows the emotional state of 1.1 Related work human. Facial emotion contains much information Pantic Rothkrantz identify three basic problems about emotion so if we can recognize facial emotion. in facial Expression analysis approach needs to deal However, it is difficult to categorize facial expression with: face detection in facial image, facial expression from static images. Neural Network may be suitable data extraction and facial emotion classification. Most in this problem because it can improve its previous system assume presence of a full frontal face performance given more examples. More over, we do view in the image being analyzed to give the a boost not need to know much about the features of the facial algorithm to exhausticaly pass a search sub-window expression to build the system. The system will over the image at multiple scales for rapid face generalize the features itself, given enough examples. detection. Scales for rapid face detection. Essa & In this paper, we investigate the performance of Pentland uses the eigenfaces method via principle neural network on this problem; at the same time component analysis (PCA)[2] To perform data compare different ways of training the network. extraction Littlewort et al use a bank of 40 Gabor wavelet filters at different scales. In the final step of I found that system gives 100% recognition result expression analysis expression are classified for known & unknown examples. We have tried for according to some scheme. The most prevalent different network & found that svm is most suitable for approaches are based on the existence of six basic this pattern recognition problem with learning rate emotions as argued by Ekman and facial action 0.01. coding system (FACS)[5] developed by Paul Ekman In section 2, I will talk about development of facial & Frisen, which codes expression as combination of parameter set, Kernel based classifier use for 44 facial movements called Action unit.
*[email protected] 42 Journal of Advances in Engineering Science Section C (5), July - December 2010
1.2 Development o Facial Parameter The Japanese female facial expression Face parameter can be defined in many levels of database[7] has been selected as basis for doing the details. They range from detailed description of the training and recognition of the expression. The materials property of the face, to the underlying cropping techniques have been carried out to structure and material property of the facial determine the best parameter to recognize physiology. In one extreme. The parameter can expression. describe the appearance of the face such as how 2. KERNEL BASED CLASSIFIER[8] much the brow is raised, how many wrinkles are on the forehead and whether there are side burns. Since This is more sophisticated learning m/c is obtain the space of the facial appearance is very large the by implanting a nonlinear mapping from the input to number of parameter to describe all the variation in another space followed by a, linear discriminate details often becomes in tractable. Various statical function. This is based on ‘Cover Therom’ Any pattern techniques have been developed for compressing the recognition problem is linearly separate in a large amount of information to some manageable sufficiently high dimensionally space called feature size. For instance cropping of image and calculation space, using nonlinear transformation. of mean, dct entropy & variance parameter of that image. These parameters are very useful for application like compression, filtering and facial recognition. If the physical property of this parameter can be simulated accurately, we can produce any kind of facial appearance using physical lows. Physical simulation of biological material is skill an active research areas.
Figure 2 : Optimal hyper plane
2.1 Support vector machine[8] Goal is to find the optimal hyper plane for non separable Pattern. According to covers Therom Multidimensional space may be transformed into a new feature space where the patterns into linearly separable with high probability by satisfying two condition 1) Transformation is nonlinear 2) Dimensionality of the feature space is high. Hyper plane acting as the decision surface as follows.
Figure 1 : Japanese Female S. B. Nimbekar et al. : Recognize Human 43 Emotions: Recognition from DCT
- Linear weights connecting to feature Input PEs = 31 —— no of parameter space to the output space Output PEs = 2 —— neutral / happy, fear / neutral , sad / neutral b = bias Exemplar = 12 —— Training input mi dimension of feature space Step size / learning rate 0.01 By using Langrangian function w is define by Epoch = 1000 Stopping Criteria – minimum – for training set MSE = feature space Xi = input pattern CONCLUSION Bi taking inner product of two vector the optimal In this paper, we discussed recognition of hyper plane is given by emotional expression on the face. By observing the control parameter over a wide range of facial motion. We can then extract a minimal parameter representation of facial control. We also developed Intelligence machine, which recognize the mood of operator
3.1 Result Table 1 : Neutral & Happy emotion classification
Output / Desired Happy Neutral Happy 60 Neutral 06 Performance Happy Neutral MSE 0.044407372 0.048359549 Max Abs Error 0.291147709 0.291432451 Percent Correct 100 100 Training Percent Correct 50 100 Testing
Figure 3 : SVM Table 2 : Neutral & fear emotion classification Output / Desired Fear Neutral 3. SIMULATION Fear 60 Import dataset – Input vector x Neutral 06 Tag data for input as well desired Performance Fear Neutral Save Tag rows by % MSE 0.020783105 0.022577808 Training 60% Max Abs Error 0.197532131 0.198530696 CV 20% Percent Correct 100 100 Test 20% Training Create Open Network Percent Correct 100 100 SVM Testing 44 Journal of Advances in Engineering Science Section C (5), July - December 2010
Output / Desired Sad Neutral 3. Neuro Net: Roadmap: Selected application . Neural Network for Emotion Recognition. Sad 60 Neutral 06 4. Ben Wong, Facial expression Recognition using Neural network, 2000. Performance Sad Neutral 5. Paul Ekman & Frisen, FACS (Facial Action MSE 0.215462971 0.233563152 coding system) 2002. Max Abs Error 0.524326852 0.527594636 6. Montse Pardas, Antonio Boafonte, Facial Percent Correct 100 100 animation parameter, Extraction & Expression Training Recognition using HMM. Percent Correct 100 100 Testing 7. http://www.mis.atr.w.jp/~mlyons/jaffe.html. 8. Simon Haykin, Neural network IInd edition 4. REFERENCES 9. Milan Sonka, Vaclav Hlavac, Image Processing 1. Philip Michel, Rana El Kaliouby, Real time facial Analysis & Machine Vision IInd edition. expression recognition in video using Support 10. R.C Gonzalez, R.E woods, Digital image vector Machine, ICMP03 Nov 03 Vancouver processing using Matlab. British Columbia Canada. 11. NeuroSolution 5.04, MATLAb 7.0.4. 2. Irfan Aziz Essa, Analysis, Interpretation & synthesis of Facial expression, Feb 95. D.K.Shende et al. : Embedded Linux Based Journal of Advances in Engineering Science45 Graphics LCD Application Development Section C (1), July - December 2010, PP 45-48
EMBEDDED LINUX BASED GRAPHICS LCD APPLICATION DEVELOPMENT
Mrs. D.K.Shende1*, Mrs.S.S.Barve2 and Mr.S.B.Mule3
1 - Assistant Professor, SIT, Lonavala 2 - Assistant Professor, MAE, Aalandi 3 - Assistant Professor, SCOE, Pune
ABSTRACT
Linux is a popular operating system for embedded systems. Linux has come a long way since its humble beginnings in 1991. Today Linux supports a very wide range of platforms, from Embedded Systems based on ARM, PowerPC, Intel, and Hitachi microprocessors. It also served as a launch pad for the open source movement, and consequently leads to great interest. This paper describes one such application of Linux of generating software for interfacing Graphics LCD using hardware ARM9 Single Board Computer. Uses sensor card to get data. This data will be from various sensors. This data is to be shown on the LCD. For some data will have to do perform some calculations and then display on the LCD. Data will be updated every second. This will involve interface of Color LCD, keypad, encoder. This will also require interface with PC for future expansion to create interface with Webpage.
1.1 INTRODUCTION Embedded Linux Based GLCD application pulses which are directly dependant on the knob development project is used for interfacing the sensor position. GLCD also shows indication of over and data card with ARM9 processor based Single Board under measurement status after doing respective Computer and displays the records on Graphics LCD calculation. (GLCD) with read-through of upper and lower limit of We can develop all of our code on your PC and the incoming data through two encoders and indicating the embedded system has full access to all of the files on same on GLCD. your PC. This is not helpful in all systems but under Sensor data card consist of all together four certain circumstances it is very helpful. Linux sensors-Temperature, Pressure, Humidity, Distance provides such access. Due to its open source nature, respectively. Sensed parameters coming from Linux has a highly qualified code base. The Kernel can Temperature, Pressure and Humidity sensors are be very small; it could fit onto a single 1.4MB floppy analog data which is given to SBC9302 processor disk drive, while including all the fundamental through ADC and Distance sensor data is serial data operating system tasks. It is highly portable; it is directly in centimeter which is given to processor available for almost every microprocessor system in through serial port. The precise data coming from all existence today. It is highly supported; it draws on the four sensors are displayed on 320 X 240 Graphics open source community across the globe for both LCD in different graphical format. Before displaying development and support. It supports a multi-user data on GLCD all sensed parameters are first set to environment with a built in Capability to concurrently Upper and Lower Limit by two encoders. These two execute applications belonging to 2 or more users. It encoders set the upper and lower limit by two Knobs also supports multiprocessor systems, is well respectively. Output parameter fluctuate number of documented.
*[email protected] 46 Journal of Advances in Engineering Science Section C (1), July - December 2010
1.2 GENERAL SYSTEM REPRESENTATIONS Feature of the selected single-board The basic block diagram is as mentioned below. computers (SBC):- The unit consist of the ARM9 processor based Single Hardware: EP9302 Board Computer. This single board computer is having embedded Linux preloaded. SBC9302 is SBC9302 is based on Cirrus Logic EP9302 based on Cirrus Logic EP9302 processor, which has processor, which has ARM920T core. ARM920T core. Although stand-alone (no OS) 1. 200 MHz (184 MHz for industrial) ARM9 applications can be run on SBC9302, it is primarily CPU (EP9302) intended to run Linux based applications. 2. On-board 8 MB Flash, 32 MB SDRAM 3. Optionally, upto 16 MB Flash and upto 64 MB SDRAM 4. 2 UARTs with option for RS232 / RS422 / RS485 / TTL (3.3V level) 5. RJ45 Ethernet LAN interface 6. 2 USB Host interface ports 7. 1 USB device interface port 8. PC/104 expansion bus 9. Channel 12 bit ADC 10. Upto 23 GPIO (3.3 Volts TTL) 11. RTC with battery-backup 1.3 SINGLE BOARD COMPUTER (SBC) 12. SD-Card interface Single board computers are most commonly 13. Standard 20 pin JTAG interface used in industrial situations where they are used in rack mount format for process control or embedded 14. Wall type power supply included within other devices to provide control and interfacing. 15. One serial cable and one USB cable Because of the very high levels of integration, included. reduced component counts and reduced connector 16. Small size - (90 x 95 mm) counts, SBCs are often smaller, lighter, more power efficient and more reliable than comparable multi- 1.4 BLOCK DIAGRAM board computers. The basic block diagram is as mentioned below. The unit consist of the ARM9 processor based Single Board Computer. This single board computer is having embedded Linux preloaded. Sensor data card consist of all together four sensors-Temperature, Pressure, Humidity, Distance respectively. Sensed parameters coming from Temperature, Pressure and Humidity sensors are analog data which is given to SBC9302 processor through ADC and Distance sensor data is serial data directly in centimeter which is given to processor through serial port. D.K.Shende et al. : Embedded Linux Based 47 Graphics LCD Application Development
The precise data coming from all four sensors are Serial connection is used to bring up shell in host displayed on 320 X 240 Graphics LCD in different pc. Ethernet connection is used for downloading graphical format. Before displaying data on GLCD all kernel and debugging. Host terminal must have sensed parameters are first set to Upper and Lower following tools: Limit by two encoders. 1. IDE These two encoders set the upper and lower limit 2. GCC tool chain for Embedded Linux. by two Knobs respectively. Output parameter 3. TFTP (Trivial File Transport Protocol) server, fluctuate number of pulses which are directly This is for downloading of modules. Linux is free dependant on the knob position. GLCD also shows to download, user is free to modify source code, user indication of over and under measurement status after is free to distribute its modified code to everyone. doing respective calculation. But, as in order to use it professionally, we still need Following are the essential for embedded Linux following, setup: · Integration 1. Embedded system development board (like · Development of Board ARM9 board) · Support Package (BSP) · Maintenance 2. Host PC · Support 3. Serial cable All above things are not free of cost but user has 4. Ethernet cross cable to pay for the same. 5. Embedded Linux kernel running in board. Linux is one of the favorites Operating System for Embedded development. A] Benefit of using Embedded Linux 1. Vendor independent Using Linux means you are no longer depend on particular vendor for supply of tools. In Linux everything is available from open source community. Even service model of all Linux vendors is almost same they used to provide Linux kernel, libraries etc. So, user can easily switch from one vendor to another. And even if user wants to go without vendor, everything is freely available. But in that case of the work of integration, BSP development has to be done by use itself. 2. Easy availability of used tools In embedded Linux so many development tools and utilities are easily available. User can download them and use them freely. So this result in fast development time for embedded system products. 3. Various hardware supports Linux community is very active. They regularly add support of new hardware. Linux is used in various research laboratories and universities worldwide, so 48 Journal of Advances in Engineering Science Section C (1), July - December 2010
Linux is always up to date with latest hardware 4. make: This is Linux standard make utility, support. useful for building complex projects. This is 4. Low cost development generally part of the Linux installation on the computer. By using Linux in embedded system product, we can development low cost products. Linux 5. Minicom: This is a serial terminal utility – development tools are free and easily available. Linux similar to HyperTerminal of Windows. This can be part of Linux OS installed on the is royalty free. There is no need to pay royalty for computer. making any number of products. B] Development Tools 6. TFTP: Stands for Trivial File Transfer Protocol – it is a useful tool for transferring files over Development tools are important. They save Ethernet LAN. This can be also part of Linux development and debug time. But most importantly, OS installed on the computer. they make developers more happy and productive by 7. Many useful make files, configuration files automating many routine, boring, and time- and some shell scripts can be also consuming tasks. It’s painful to see programmers considered part of the development tool. spend a significant percentage of their valuable time on such routine tasks as downloading their code to 1.6 FUTURE EXPANSION the embedded target. This situation is not uncommon · The GUI on GLCD will transmitted through even with traditional embedded systems, but it’s far network and displayed on demand through worse with embedded Linux, where the lack of good webpage. development tools is evident. · Includes serial interface and remote access Commercial tools with unit. Embedded Linux integrated development environment (IDE) software suites are usually available 1.7 References from the same companies that sell embedded Linux. 1. IEEE paper on “Embedded Port Scanner (EPSS) Wind River, MontaVista, TimeSys, LynuxWorks, and System Using Linux and Single Board a dozen other vendors come to mind. Computer” published in year 2008. However, these are open source tools and can be 2. IEEE paper on “Code generation for Linux device downloaded from Internet. Important tools required driver” published in Advanced Communication are: Technology, ICACT 2006. The 8th International Conference in year 2006. 1. gcc compiler: This is native compiler - i.e. runs on Linux computer and generates 3. IEEE paper on “A Device Driver Engine Based executables for Linux computer. This is on Components for the Embedded Prototyping System” published in Ubiquitous Information generally part of the Linux installation on the Technologies & Applications, 2009. ICUT ’09. computer. Proceedings of the 4th International Conference 2. arm-elf-gcc: This is cross compiler - i.e. in year Dec 2009. runs in Linux computer and generates 4. IEEE paper on “Embedded Linux implementation executables for standalone ARM targets. This on a commercial digital TV system” published is is useful for generating executable file of in Consumer Electronics, IEEE Transactions in boot-loader. Boot-loader is a standalone year NOV 2003. application, since it starts before Linux. This 5. IEEE paper on “An embedded Linux based can be downloaded from Internet. navigation system for an autonomous 3. arm-linux-gcc: This is cross compiler – i.e. underwater vehicle” published in Southeast on, runs in Linux computer and generates 2007. Proceedings. IEEE executables for ARM Linux targets. This is 6. IEEE paper on “Building distributed embedded useful for generating Linux applications to run systems with RTLinux-GPL” published in on SBC9302 board. This can be downloaded Emerging Technologies and Factory Automation, from Internet. 2003. Proceedings. ETFA ’03. IEEE Conference. S. V. Pingale et al. : Cost Estimation for Journal of Advances in Engineering Science49 Distributed Systems using Use Case Diagram Section C (3), July - December 2010, PP 49-56
COST ESTIMATION FOR DISTRIBUTED SYSTEMS USING USE CASE DIAGRAM
Prof. S. V. Pingale1*, Prof. R. S. Badodekar2 and Prof. H. S. Badodekar3
1 - Department of Computer Engg, Sinhgad Institute of Technology, Lonavala, Dist-Pune, INDIA 2 - Department of Info-Technology, Sinhgad Institute of Technology, Lonavala, Dist-Pune, INDIA 3 - Department of E&TC, A. C. Patil College of Engg, Kharghar, New Mumbai, INDIA
ABSTRACT
Estimating the cost of development is based on a prediction of the size for future systems. A lot of cost estimation models were reported in the literature but many of these models became obsolete because of the rapid changes in technology. Reliable estimations are difficult to obtain because of the lack of detailed information about the future system at an early stage. Cost models like COCOMO (COnstructive COst MOdel) [5,6] and sizing methods like Function Point analysis are well known and in widespread use in Software Engineering. These models were applicable only to procedural paradigm, and are not directly applicable to software products developed using the object oriented methodology or real time systems. It is this idea that gave birth to the creation of Use Case Point (UCP) metrics, originally developed by Gustav Karner [3]. UCP uses use cases as the primary factor, use case model is the first model developed in an object-oriented design process using UML. In this paper we propose a novel approach to map the object oriented systems from their function points and converting use case point counts on the basis of actor (Customer’s) interaction to the software and to estimate cost of development by using Synthesized UCP (s- UCP) model with additional information obtained from synthesized use case attributes. Key words : Use Case Point, COCOMO, Cost Estimation
1. INTRODUCTION used LOC metrics. The major problem with these metrics was two-fold: (i) the lack of precise definition This Software is an intangible product and hence of LOC or DSI, and (ii) there is no reasonable probably the most crucial difference between the methodology by which one can estimate the number manufacturing industry and the software industry is of source code lines or instructions until the coding is that the former is able to stick to schedules and cost over. Further, these metrics were defined for most of the time. Even when using a well-defined procedural, possibly line oriented, languages such as methodology, the development cost of a well-defined FORTRAN and COBOL. With the development of application is not easy to predict. Some key factors block structured languages such as Pascal, Algol, C that contribute to this difficulty include the precise set etc., the notion of LOC or DSI was difficult to define of functionalities to be implemented, the various risks precisely. associated with the development process, the knowledge and experience of the development team Albrecht came up with the Function Point (FP) and lack of detailed information about the system at metrics in 1979[1]. FP uses five parameters: number an early stage. Among these, the set of functionalities of inputs, number of outputs, number of inquiries, to be implemented is the most crucial factor. Software number of internal logical files and number of external cost estimation models developed in the 60’s and 70’s logical files. Consequently, FP is based on the used the Line of Code (LOC) or Delivered Source number of interactions and the size of data to be used Instruction (DSI) metrics. For example, COCOMO in the end product. While FP eliminated the need for model [5, 6] used the DSI metrics while many others lines of code or delivered source instructions, and was
*[email protected] 50 Journal of Advances in Engineering Science Section C (3), July - December 2010 widely used because of its independence on Section 4 with the discussion on continuing work in development platform and environment, it does not this direction in Section 5. Due to space constraints, seem to be applicable to software products developed this paper does not include descriptions of a use case using the object-oriented methodology [8] for model such as notations and their semantics. languages like C++ and JAVA. In particular, the notion Interested readers are referred to any UML book such of internal and external logical files is somewhat as [10] or UML manual published by OMG [9]. harder to identify in the object-oriented paradigm as 2. PROPOSED METHOD : SYNTHESIZED UCP well as real time systems. (S-UCP) Gustav Karner [3] came up with the notion of Use Case Point (UCP) which is somewhat similar to the The s-UCP methodology uses every aspect of a notion of Function Point but based on use cases. use case model such as actors, use cases, Model based estimation will help in a greater extend interaction between actors and use cases, to reverse engineering and maintenance of legacy relationships between actors, relationships between software. The use case model is the front end model use cases and finally the synthesized attributes of of the Unified Modeling Language (UML) [9]. With the each use case. The last one is important because it emergence of the UML as the most commonly used describes the missing details of a use case diagram, notation to model and design object-oriented software while others can be directly extracted from a use case products, the application of use cases for size and diagram itself. This makes the significant difference hence cost estimation seems to be a perfect fit. between UCP method given by Karner [3] and s-UCP. However, Karner’s [3] method does not take into The attributed list also contains few use case account some of the application domain details such narratives suggest by Periyasamy [4]. In order to as the number of interaction between actors, and concretize the methodology, the authors used the relationships between use cases. template for use case synthesized attributes in Table I. The objective of this research is to elaborate the UCP model by synthesizing the use cases with a Table 1 :Sample use case synthesized attributes focus on internal details of each use case as well as Use Case A Descriptive Name of the Use interaction between the use cases. It is common that Name Case every use case is supported by a use case synthesized attributes that explains the internal Purpose A brief description of the tasks to details of the use case. This paper describes a be implemented by this use case method for cost estimation based on the use case Input List of Input Parameters to the diagram at an early stage of software development. It parameters use case focuses on the use case attributes, uses the Output List of Output parameters returned interaction between the entities in a use case parameter from the use case. diagram, and hence closely estimates the size of the software product to be developed. The methodology Primary The list of actors who invoke this was tested with the use cases developed for an actor use case automated indexing and searching system for a book Secondary The list of actors that are used publishing company. The estimated value for Actor by this use case development efforts seems to match with the actual Precondition Conditions that must test true to development efforts spent by the company. use the use case. Unlike The rest of the paper is organized as follows: assumptions, these conditions Section 2 describes the step by step application of the are tested by this use case before doing anything else. If the synthesized UCP methodology. A case study with a conditions are not true, the actor step wise evaluation of UCP using s-UCP model is or other use case is refused entry. described in section 3. The paper concludes in S. V. Pingale et al. : Cost Estimation for 51 Distributed Systems using Use Case Diagram
TABLE II : Actor Weight Classification Post Conditions that must test true condition when the use case ends. You Actor type Classification of Actors Weight may never know what comes after Weight the use case ends, so you must guarantee that the system is in a Very simple Specialized Primary/ 1.0 stable state when it does end. Secondary actor Process A step-by-step description of the Simple Simple Primary actor 1.5 dialog between the use case (the with 1 < number of system) and the user (actor or other use case). Very often it is transactions <= 3 helpful to model this sequence of Less Primary actor with 3 < 2.0 events using a flowchart or average number of transactions activity diagram just as you might model a protocol for communication <= 5 between two business units. Average Primary actor with 2.5 Successful A sequence of instruction that number of transactions>5 Scenario explain the successful scenario Secondary actor with 1 2.5 of invoking this use case. transaction Exceptions A set of conditions that may make Complex Complex Secondary actor 3.0 the use case fail when invoked. with 1 < number of Includes List Use cases that can be included transactions < = 3 in this use case. Extends List Use cases that can be extended This, in turn, is based on the number of from this use case. transactions the actor has with the use cases. As the number of transactions increases, more effort is 2.1 Use Case Estimation involved in coding and hence the complexity of the The s-UCP method first calculates unadjusted corresponding actor increases. The complexity of a UCP values, referred to as UUCP in this paper. This secondary actor (such as a database) with ‘n’ includes unadjusted actor weights, unadjusted use transactions is higher than that of a primary actor case weights. The unadjusted UCP values are then (such as a user) with the same number of adjusted using three additional factors, namely transactions because generally more effort is involved technical complexity factor (TCF), environment factor in coding transactions with the secondary actor (EF), and synthesized factor (SF). The final result is compared to those with a primary actor. For example, the adjusted UCP values, referred to as AUCP. The s- if the secondary actor is a database, then coding UCP method uses the same TCF and EF weights as efforts are required for checking the connection to the given in the original UCP method [3], but it uses a database, writing SQL statements for transactions, different set of calculations for UCP values. The checking for database commit actions and so on. authors of this paper believe that the final UCP value 2.1.2 Use Case Weight Classification (s-UCP) returned by AUCP is more precise compared to those returned by UCP. Similar to actors, each use case is assigned a different weight. The use case type is defined based 2.1.1 Actor Weight Classification on the number of transactions (the number of direct Table II shows the assignment of weights to connections between actors and the use case). Table various actors based on the actor type and the III lists the classification of use cases and their number of transactions performed by an actor. associated weights [7]. 52 Journal of Advances in Engineering Science Section C (3), July - December 2010
TABLE III : Use case weight classification Thus, a complex precondition such as Bank account number must be valid ^ PIN code must be Use Case Classification of Use Weight type Cases valid will include two simple predicates. The justification of assigning separate weight for each Simple Number of transactions <= 3 5 individual predicate comes from the fact that each Average 4 < number of transactions <= 7 10 simple predicate is required to be implemented to Complex 7 < number of transactions <= 15 validate the precondition.
2.1.3 Weights for Synthesized use Case 3. EVALUATION OF s-UCP: A CASE STUDY Attributes In the original method the total unadjusted actor A use case diagram must be supported by weight (UAW) is calculated by counting how many Synthesized Use case Attributes. Table I shows the actors there are of each kind (by degree of structure of a synthesized use case attributes used complexity), multiplying each total by its weighting in this methodology. Though there is no standard for factor & adding up the products. Each use case is the structure of a use case attributes, the authors then defined as simple, average complex depending found that the use case structure illustrated in Table on number of transactions in the use case I contains all information that many practitioners use. descriptions, including secondary scenarios. With this assumption, Table IV describes the weights 3.1 Proposed Estimation Method associated with the different parameters of a use case The primary goal of proposed technique is to attributes. estimate the cost of software from number of different Notice that all the use case attributes shown in actors and its use case interactions between both of Table I are used in Table IV because all of them them during the time of implementing the application. contribute to the coding efforts, and others such as In this proposed technique, all the actors are actors are already taken into consideration. Moreover, classified into very simple, simple, less average, it should also be noted that the weight associated average, complex and then the use case are classified with ‘Precondition’ in Table IV must be used for only into simple, average and complex based on the one predicate in the precondition; the same applies number of actor interacting with that use case along to post condition as well. with use case interacting with another use case. Hence use case complexity can be defined easily. Table IV : Weight for synthesized use case The s-UCP method also uses synthesized weight attributes factor (SF) to improve the effectiveness of this Use Case Classification of Use Weight technique. Further classification is made for all the use type Cases cases that interact with another use case. If the S1 Input parameter 0.1 current use case is interacting with a simple use case then we will classify it as average and if the current S2 Output parameter 0.1 use case is interacting with an average use case then S3 A condition to execute each 0.1 we will classify it as complex. Then we will assign the process weighting factor in the above said manner. S4 A predicate in Precondition 0.1 3.1.1 Estimating s-UCP S5 A predicate in Post-condition 0.1 Finally, the synthesized use case point (s-UCP) S6 An action in Successful 0.2 is calculated by multiplying all the three values UUCP, scenario TCF, EF & SF. That is, S7 An exception 0.1 S8 An Includes conditions 0.1 s-UCP = UUCP * TCF * EF*SF (1) S9 An Extends conditions 0.1 CASE STUDY: - A Book Indexing Project. S. V. Pingale et al. : Cost Estimation for 53 Distributed Systems using Use Case Diagram
Given bellow is the use case diagram of indexing project for which we will calculate the use case points.
Figure 1 : Use case diagram for book indexing project
3.2 Weight Based Estimation TCF = 0.6 + (0.01 * T Factor) (2) The method employs a technical factors multiplier The Environmental Factor (EF) is calculated corresponding to the technical complexity adjustment accordingly by multiplying the value of each factor (F1 factor of the FPA method, environmental factors – F8) by its weight and adding all the products to get multiplier in order to quantify non-functional the sum called the E Factor. The formula below is requirement & synthesized factor in order to quantify applied: the use cases. Various factors influencing productivity are associated with weight and values are EF = 1.4+ (-0.03 * E Factor) (3) assigned to each factor, depending on the degree of The Synthesized Factor (SF) is calculated influence. accordingly by multiplying the value of each factor (S1 The Technical Factor (TCF) is calculated – S9) by its weight and adding all the products to get multiplying the value of each factor (T1 – T13) by its the sum called the S Factor. The formula below is weight and then adding all these numbers to get the applied: sum called the T Factor. Finally, the following formula is applied: SF = 1.1+ (0.01 * S Factor) (4) 54 Journal of Advances in Engineering Science Section C (3), July - December 2010
3.2.1 Weight based on Technical Factor Table II : Environmental Factors in Project and The technical factors and their weighted values Their Weights are shown in the following table:- Factor Description Weight Assessment Impact Table I : Technical Factors in Project and their Number Weights F1 Familiar with 1.5 2 3 RUP Factor Description Weight Assessment Impact Number F2 Application 0.5 1 0.5 experience T1 Distributed 2 0 0 System F3 Object-Oriented 1 2 2 experience T2 Response 1 4 4 adjectives F4 Lead analyst 0.5 1 0.5 capability T3 Response 1 5 5 adjectives F5 Motivation 1 2 2
T4 Complex 1 2 2 F6 Stable 2 3 6 processing requirements
T5 Reusable code 1 3 3 F7 Part-time -1 0 0 workers T6 Easy to install 0.5 4 2 F8 Part-time -1 4 -4 T7 Easy to use 0.5 4 2 workers T8 Portable 2 4 8 E-FACTOR TOTAL WEIGHT 10 T9 Easy to change 1 3 3 The EF value is calculated using equation (3): T10 Concurrent 1 3 3
T11 Security 1 3 3 EF = 1.4 + (-0.03 * 10) = 1.1 features 3.2.3 Weight based on Synthesized Factor T12 Access for 1 1 1 TABLE III : Weight for use case on Synthesized third parties Attribute T13 Special training 1 4 4 required Attribute Use case Weight Assessment Impact Number Synthesized T-FACTOR TOTAL WEIGHT 40 Attributes
S1 Input 0.1 2 0.2 The TCF valued is calculated using equation (2): parameter
TCF = 0.6 + (0.01 * 40) = 1 S2 Output 0.1 2 0.2 parameter 3.2.2 Weight based on Environmental Factor S3 A condition to 0.1 1 0.1 The environmental factors and their weighted execute each process values are shown in the following table:- S. V. Pingale et al. : Cost Estimation for 55 Distributed Systems using Use Case Diagram
S4 A predicate in 0.1 1 0.1 4. CONCLUSION Precondition This paper evaluates a use case diagram for S5 A predicate in 0.1 1 0.1 an indexed project and calculated the Post-condition synthesized use case points (s-UPC) according S6 An action in 0.2 1 0.2 to proposed method which help us to calculate Successful the cost of the project. Here the proposed method scenario constructed a simple way to use minimal set of formula to calculate the cost of objects oriented S7 An exception 0.1 1 0.1 software development. This motivates a S8 An Includes 0.1 1 0.1 relationship between some of the existing use conditions case methods. With classifying the actors here, the approach was able to classify the software S9 An Extends 0.1 1 0.1 system with respect to the level of code quality conditions and the better way to calculate the use case s-FACTOR TOTAL WEIGHT 1.2 points. This approach is very promising and more The SF value is calculated using equation (4): applicable to identifying low quality code than high and use case points due to specific SF = 1.1+ (0.01 * 1.2) = 1.112 theoretical concept employed to develop the The UUCP value is calculated using the following approach. However, this is a mammoth step in the equation: right direction in reducing the turnaround time. It takes to perform a code analysis on industrial UUCP = UAW + UUCW (5) software cost estimation. UUCP = 1*2 + 15*5 + 3*10 + 0*15 = 107 5. REFERENCES Using the above values the use case points of this 1. Albrecht, A.J., 1979. Measuring application project is calculated using equation (1): development productivity. IBM Applications s-UCP = 107 * 1 * 1.1 * 1.112 = 130.882 Development Symp, GUIDE Int. and SHARE Inc., IBM Corp., Monterey, CA, Oct. 14-17, 1979. 3.3. Result Analysis 2. Anda B. et al. Estimating software development The UCP count using s-UCP method will give efforts based on use cases - Experience from more accurate and effective estimation as compared industry. In M. Gogolla, C. Kobryn (Eds.): UML to the UCP counts done by Bente Anda [2] and 2001 - The Unified Modeling Language. Springer- Periyasamy [4]. The comparison made is listed in Verlag. 4th International Conference, Toronto, Table VI. Canada, October 1-5, 2001, LNCS 218, 2001.
Table IV : Comparative Study 3. Karner, G. 1993. Metrics for Objectory. Diploma thesis,University of Linköping, Sweden. No. Sr. No. METHOD UCP Count LiTHIDA- Ex- 9344:21. December 1993. 4. Periyaswamy,K., Ghode, A. “Effort Cost 1 Bente Anda [2] 107.23 Estimation using extended Use Case Point 2 Periyasamy [4] (e-UCP) 118.91 (e-UCP) Model”, IEEE 2009
3 s-UCP (Proposed method) 130.882 5. Boehm, B., 1981. Software Engineering Economics, Prentice Hall, 1981. 56 Journal of Advances in Engineering Science Section C (3), July - December 2010
6. Boehm, B. et al., Software Cost Estimation with COCOMO II, Prentice Hall Englewood Cliffs, NJ, 2000. 7. Edward, C. R. “Estimating Software Based on Use Case Points.” Proceedings of the Object- Oriented, Programming, Systems, Languages, and Applications (OOPSLA) Conference, San Diego, CA, 2005. 8. Fetke, T., Abran, A., and Nguyen, T. 1997. Mapping the OO-Jacobsen approach into function point analysis. In proceedings of Technology of Object-Oriented Languages and Systems (TOOLS 97), 1997. 9. UML 2.0 Reference Manual, Object Management Group, www.omg.org, 2003. 10. Schneider, G., and winters, J.P., Applying Use Cases, Second edition, Addison Wesley, 2001. M.S. Jadhav et al. : Limnological Study of Journal of Advances in Engineering Science57 Pashan Lake, Pune (Maharashtra, India) Section D (1), July - December 2010, PP 57-62
LIMNOLOGICAL STUDY OF PASHAN LAKE, PUNE (MAHARASHTRA, INDIA)
Mrs. M.S. Jadhav1 * and Dr. Mrs. K.C. Khare2
1 - HOD, Department of Civil Engineering, Sou. Venutai Chavan Polytechnic, Pune 411041, Maharashtra 2 - Professor of Civil Engineering, Sinhgad College of Engineering, Pune 411041, Maharashtra
ABSTRACT
Pashan Lake is a important Lake in Pune city which attracts migratory birds. Deforestation on nearby hills has caused heavy siltation resulting in decrease in the depth of the lake. This has reflected in reduction in the number of deep diving ducks which prefer to occupy the central deep portion of the lake. They are now outcompeted by dabbling ducks which prefer shallow water. The increasing number of these ducks in the central position is a clear indicator of decrease in the depth of the lake. It was found that depth of the lake in 1980 was 30-40 ft. which was reduced to 15-20ft in year 2001-02. The surface water quality of Pashan Lake is severely degraded due to the pollution from surrounding areas directly entering the water. Three surface sampling points were selected to evaluate the water quality. The study presents the. Phyisico_chemical characteristics of the lake water and suggests the means to improve the water quality through eco remediation measures for restoration.
1. INTRODUCTION from surrounding areas directly entering the water. Three sampling points were selected to evaluate the Pashan Lake is situated between 18°32’7"N and water quality. Surface water samples are collected 73°46’58"E near Mumbai - Pune by pass highway in from established sites .Water analysis was done for Western India. Pashan lake is manmade lake built by the parameters like PH, Dissolved oxygen (DO), bunding Ram river. The catchment area is 40 sq.km. Biochemical oxygen Demand (BOD), Chemical Silting of lake occurs due to deforestation, road Oxygen Demand (COD), Total dissolved solids (TDS), construction, and other land disturbances which Chloride, Calcium and Magnesium and Hardness for results erosion. The surface water quality of the testing the suitability for drinking, agricultural Pashan Lake is severely degraded due to the pollution purposes.
Figure 1 : Location of Pashan Lake
* [email protected] 58 Journal of Advances in Engineering Science Section D (1), July - December 2010
The lake does not seem to have received care and · Three depths (surface, thermocline and attention during last few years as a result, the lake as bottom) for lakes not deeper than 30 m; well as its catchment is facing serious threats from · Four depths (surface, thermocline upper encroachment and pollution, in this connection Pune hypolimnion, bottom) for lakes of at least 30m Municipal corporation is going to undertake the lake depth; in lakes deeper than 100 m additional improvement project which includes desiltation, depths may be considered.[3] beautification and removing of aquatic weeds and its disposal with the help of government and NGOs. 4. MATERIALS AND METHODS 2. CHARACTERISTICS OF LAKE For the present study, the water samples were collected in sterilized bottles using the standard When selecting a lake/reservoir station, there procedure in accordance with the standard method of should be a comprehensive collection of information American Public Health Association (1995). The and an appraisal of the information requirements. There samples were collected for a period of 6 months from is a need for data on the lake characteristics such as December 2006 to June 2007, at three sampling volume, surface area, mean depth, water renewal time stations. The samples were brought to the laboratory together with such information as is available on the with due care and were stored at 20oC for further thermal, bathymetric, hydraulic and ecological analysis. The physico-chemical parameters such as characteristics. Pashan lake area is around 144 Acers. pH, Biochemical Oxygen Demand, (BOD), Dissolved Depth is 25-30 ft. after desiltation. The source for the 340 to 960 m/m) was used for analysis and chemical lake is water from Ram river and under ground water used were of analytical grade.[2] from wells which are situated in lake. 3. LAKE WATER SAMPLING Table 1: [1] Parameters and methods employed in the chemical examination of water The choice of sampling stations is influenced by samples: the various uses of the water and their location, relative magnitude and importance. It is important to note that Sr.No. Parameter of Methods any information obtained from the survey of a water water analysis intake from a lake for drinking water, industry or agriculture cannot reflect the overall quality of this 1 pH Potentiometric water body which should be determined from vertical profiles. 2. DO Azide modification Lake sampling is normally carried out from a boat. 3. BOD Azide modification The station is usually identified from a combination of landmarks on the shore and depth profiles with echo 4. COD Dichromate reflux sounding. Precise identification of the station each time is not easy but this is usually immaterial 5. Chloride Gravimetric because of the good lateral mixing. A number of samples will need to be taken at 6. Calcium Titrimetric vertical intervals. The following minimum programme is recommended. 7. Magnesium Titrimetric
· Two depths (surface and bottom) if lake depth 8. Hardness Titrimetric does not exceed 10 m;· M.S. Jadhav et al. : Limnological Study of 59 Pashan Lake, Pune (Maharashtra, India)
Table 2: Physico – chemical analysis of Pashan Lake During December 2006.
Station I Station II Station III
pH 7.9 8.07 8.2
DO mg/L 6.8 6.9 6.6
BOD mg/L 128 130 133
COD mg/L 134 136 137
Chloride mg/L 51.2 47.50 52.60
Calcium mg/L 80 78 83 Figure 2.1 : BOD During Different Seasons
Magnesium mg/L 16 18 19
Alkalinity mg/L 172 175 186
Hardness mg/L 203 209 228
Total 295 300 310 dissolved solids mg/L
Table 3 : Physico – chemical analysis of Pashan Lake During June 2007.
Station I Station II Station III
pH 8.3 8.8 8.5 Figure 2.2 : DO During Different Seasons DO mg/L 6.2 5.8 5.9
BOD mg/L 119 120 124
COD mg/L 124 128 126
Chloride mg/L 50.8 56.2 54.5
Calcium mg/L 84 88 86
Magnesium mg/L 14 21 18
Alkalinity mg/L 186 189 184
Hardness mg/L 198 189 194
Total dissolved 311 312 308 solids mg/L Figure 2.3 : COD During Different Seasons 60 Journal of Advances in Engineering Science Section D (1), July - December 2010
Figure 2.4 : Chloride Concentration During Different Seasons Figure 2.7 : Magnesium Concentration During Different Seasons
Figure 2.5 : Hardness During Different Seasons
Fig 2.8 PH During Different Seasons
5. RESULTS AND DISCUSSIONS The seasonal variation in Physico- chemical parameters are given in table No. 2 and 3 respectively. In the present study PH range was recorded 7.9 to 8.2 in Dec. 2006 and 8.3 to 8.8 in June 2007. The high PH range was recorded in summer and low range in winter. PH of water is important for the biotic compound because most of the plant and animal species can survive a narrow range of PH. According H Figure 2.6 : Calcium Concentration During to literature P is considered to be the most important Different Seasons factor particularly in the case of the green algae. The M.S. Jadhav et al. : Limnological Study of 61 Pashan Lake, Pune (Maharashtra, India) lower value of PH during rainy season may be due to 6. RESTORATION OF LAKE the dilution of alkaline substances. The lake restoration actions can be categorized The dissolved oxygen varied from 6.6 to 6.9 mg/l under five broad heads, i) in- lake treatment ii) shore (Dec. 2006) and 5.8 to 6.2 mg/l in June 2007. line treatment iii) source control iv) people’s Dissolved oxygen in water at a given temperature participation v) environmental awareness. depends on factors like temp. of water. Almost all For Pashan Lakes following restoration plants and animals need Dissolved oxygen for techniques can be used. respiration . The biochemical oxygen demand was recorded in the range 128 to 133 mg/l (Dec.2006) and 1) Diversion of sewage line from the lake. 119 to 124 mg/l (June 2007). The values of Dec.2006 2) Dredging of silt and removal of weeds. are somewhat higher as compared to June 2007. The 3) Bio- remediation Chemical oxygen demand was recorded in the range 134 to 137 mg/l (Dec. 2006) and 124 to 128 mg/l (June 4) Planting native trees around the lake to 2007). COD is a measure of any kind of oxidisable control the erosion of soil impurities present in the water. 5) Introduction of composite fish culture to COD is a measure of both the biologically control mosquitoes oxidisable and biologically inert organic matter 6) Creation of landscape or artificial land at centre present in the water sample. The chloride was of the lake will provide many benefits, which recorded in the range 47.5 to 52.6 mg/l (Dec. 2006) include increased land value, recreation and 50.8 to 56.2 mg/l (June 2007). The total hardness facilities, water for gardening, jogging track [4]. ranged from 203 to 228 mg/l (Dec. 2006) and 189 to 198 mg/l (June 2007). The calcium levels varied from 7. REFERENCES 78 to 83 mg/l (Dec. 2006) and 84 to 88 mg/l (June 1. R., Azhagesan, “NWA Water Quality parameters 2007). The magnesium levels varied from 16 to 19 and water quality standards for different uses” mg/l (Dec. 2006) and 14 to 21 mg/l (June 2007). The maximum values were during monsoon while 2. K.K.S., Bhatia and O., Singh, “Water quality minimum values were during winter. The calcium and assessment for management of a typical lake magnesium along with other salts are responsible for in South India” NCEC, 2006. the hardness of the water. The calcium is not known 3. K.K.S., Bhatia, “Lake water quality monitoring to indicate or produce any hazardous effect on and Modelling Chaturvedi Samiksha; human health. The magnesium has higher solubility Dineshkumar and R.V.Sing” Res. j.chem. than calcium. The calcium and Magnesium Environ, Vol.7 (3), 2003 hardness are the two elements which form the most 4. S., Joshi, “ Monitoring of Environmental factors abundant ions in fresh water. The average values of and lake water quality Pune Municipal calcium and magnesium hardness never exceed the Corporation Report”, Renewal and management standard limits of [7]. i.e. 200mg/l and 100mg/l. The of sewage and drainage disposal in city of Pune total dissolved solids ranged from 295 to 310 mg/l under JNNURM . (Dec. 2006) and 308 to 312 mg/l (June 2007). The excessive total dissolved solids generally affects 5. N., Shaikh, and S.G. Yeragi, “ Seasonal potability of water. Temperature changes and their influence on free 62 Journal of Advances in Engineering Science Section D (1), July - December 2010
carbondioxide, Dissolved (DO) and pH in Tansa 7. “Standard Methods For Examination Of Water river of Thane District, Maharashtra”, J.Aqua. and Waste”, 20th ed; American Public Health Biol., Vol.18(1), PP. 73-75, 2003. Association,Washingaton;D.C.1995.(APHA)
6. K., Sahastrabuddhe, and A., Patwardhan, 8. R. K., Trivedi, “Assessment of water quality of “Changing status of urban water bodies and some Indian Rivers based on factor analysis associated health concern in Pune India”, WHO Guidelines for Drinking Water Quality”, International Conference on Environment and Second Edition Vol.1 PP. 52-82, Geneva, 1993. Health, Chennai, 2003