Anirban Guha

 

Office

Department of Mechanical Engineering

IIT Bombay

Powai, Mumbai - 400076

India

 

Permanent Residence

39B, Dilkhusa Street

Kolkata - 700017

India

Personal Information

Nationality: Indian

Marital Status: Single

Date of Birth: 22nd February, 1972

Contact

E.Mail: anirbanguha1@gmail.com

Phone (Off.): +91-22-25767590

Phone (Res.): +91-22-25768590

Phone (Mob.): +91-9619057590


Current Occupation

Assistant Professor in Department of Mechanical Engineering, IIT Bombay since July 2005.

 

Education

Ph.D. - IIT Delhi

Completed in 2002

Topic: Application of Artificial Neural Networks for Prediction of Yarn Properties and Process Parameters

M.Tech. - IIT Delhi

Requirements completed in December 1995

Topic: Investigation of properties of jute and coir fibres

Application: Biodegradable geotextiles

B.Sc.(Tech.) - Calcutta University

Requirements completed in July 1994

 

Prior Work Experience

Worked in Marketing and Technical Services division of Futura Fibres Division of Indian Organic Chemicals Ltd. from February 1996 to July 1997.

 

Patents

Design of a robotic arm (in the process of being patented)

 

Loom for 3D fabrics (in the process of being patented)

 

Selected Publications

Predicting Yarn Tenacity: A Comparison of Mechanistic, Statistical and Neural Network Models, Anirban Guha, R. Chattopadhyay and Jayadeva, Journal of The Textile Institute, Vol. 91, Part 1, No. 2, 2001, pp. 139-145.

Investigations into Homogeneity of Coir Fibres, P. K. Banerjee, R. Chattopadhyay and A. Guha, Indian Journal of Fibre & Textile Research, Vol. 27, June 2002, pp. 111-116.

Artificial Neural Networks: Application to Textiles, R. Chattopadhyay and A. Guha, Textile Progress, v35, n 1, 2003.

A Study on Neural Network’s Capability of Ranking Fibre Parameters Having Influence on Yarn Properties, Jayadeva, Anirban Guha and R. Chattopadhyay, Journal of The Textile Institute v 94, part 1, no. 3/4  2003, pp. 186-193.

Performance of Neural Networks for Predicting Yarn Properties Using Principal Component Analysis, R. Chattopadhyay, Anirban Guha and Jayadeva, Journal of Applied Polymer Science, v 91, n 3, Feb 5, 2004, pp. 1746-1751.

A Novel Mechanism for Driving the Sley in a Shuttle Loom, Anirban Guha, C. Amarnath, M. K. Talukdar and Manoj A.G, Journal of The Textile Institute , 97(6), 2006, pp 513-518.

Cloth Roll Doffing Machine for Indian Powerloom Industry, Anirban Guha, N. Ramakrishnan, M. K. Talukdar and Prateek Sharma, Indian Journal of Fibre and Textile Research, Vol. 33, March 2008, pp 97-100.

A Method for Validation of a Model of Balloon in Ring Frame, Milan Sharma and Anirban Guha, Indian Journal of Fibre and Textile Research, Vol. 33, December 2008, pp. 383-387.

Measurement of Yarn Hairiness by Digital Image Processing, Anirban Guha, C. Amarnath, S. Pateria and R. Mittal, Journal of The Textile Institute, 25 February 2009, pp 1-9.

Linkage Driven Dobby With Dwell, Anirban Guha, C. Amarnath, K. Kurien Issac, M. K. Talukdar and Shrinivas Joshi, Textile Research Journal, Vol. 79, Issue 9, 2009, pp. 804-809.

 

Papers Presented in Conference (Selected)

Prediction of Properties of Friction Spun Yarns by Using Artificial Neural Networks, Anirban Guha, Jayadeva and R. Chattopadhyay, 13th National Convention of Textile Engineers, Solapur, Maharashtra, India organised by Textile Engineering Division of The Institutuion of Engineers (India), October 3 - 4, 1998.

 

Simulation of Yarn Manufacturing Process Using Neural Network, R. Chattopadhyay and Anirban Guha, 31st Textile Research Symposium at Mt. Fuji, Susono City, Japan, August 2 - 4, 2002.

 

Simulation of Spinning Process: A Comparative Assessment, R. Chattopadhyay, Anirban Guha and U. Mahapatra, Proceedings of Asian Textile Conference, Delhi, India, December 2003.

 

Double Dwell Linkage Mechanisms for Prolonged Dwells, C. Amarnath and Anirban Guha, 12th IFToMM World Congress on Mechanism and Machine Sciences, Besancon, France, 17th to 21st June, 2007.

 

On Evolution of Textile Machinery, C. Amarnath and Anirban Guha, IFToMM Workshop on History of Machine and Mechanisms Science, Bangalore, India, December 14, 2007.

 

Current Research Interest

Design of machines for the textile industry with focus on kinematics.

Application of artificial neural networks in industrial problems.

 

Interaction with Industry

Worked with M/s Aditya Birla Nuvo for development of filament yarn twisting device.

 

Worked with M/s Bond Safety Belts for design of emergency locking retractor for restraining vehicle passengers during accidents.

 

Ongoing work with Ph.D. students

Design of an economical mechanical face seal. (Thesis submitted in January 2010)

 

Studies on cotton ginning.

 

Studies on impact simulation.

 

Ongoing work with other students

Development of an experimental setup to simulate stresses generated by high speed trains

 

Development of an experimental setup to simulate rapid construction of embankments

 

Determination of aerostat position and orientation

 

Determination of aerostat shape

 

Multi-container loading problem for stenter packaging

 

Studies on image analysis for yarns

 

Improvements in loom for 3D fabrics

 

Design and analysis of rod cutter for biomedical applications

 

Modelling knee mechanism using four bar linkage

 

Optimizing aids for paraplegics

 

Earlier projects with students

 

Core yarn attachment for ring frames

An attachment for spinning core yarn was retrofitted on an LMW G5/1 ring frame. A yarn tension measuring device was designed to continuously monitor the tension in the core yarn. Effect of different factors including tension and manner of feeding on the quality of core yarn was studied.

 

Automation in Garment Cutting

This project, apart from automation of the cutting process, investigated an alternative cutting tool for fabrics namely laser. Software has been developed which takes the essential dimensions to generate a shirt as input and gives the CNC code to cut the pieces in a particular fashion, to minimize waste under the given constraints, as the output. A design has also been proposed for an automatic laser cloth cutter which read the CNC code generated by the software and cut the fabric in that fashion using laser as the cutting tool.

 

Study of Existing Sley Drive and Development of Novel Devices

A linkage based mechanism which will give a dwell to the sley near its backmost position was designed. This is expected to be cheaper to incorporate into a loom compared to cam based sley drives. It will give lower abrasion between warp threads and shuttle during weaving.

 

Automatic Doffing of a Cloth Roller on a Loom

A hand operated wheel mounted device was designed for loom auto-doffing. All the operations, disengaging the cloth roller from the loom, removing it from the loom, loading an empty cloth roller and carrying away the full cloth roller can be done by a single operator by operating the mechanism manually. Three different designs were studied. The final design was can pass through narrow aisles between rows of looms.

 

Design of New Yarn Hairiness Tester

The aim of this work was to explore advance techniques of Digital Image Processing (DIP) for hairiness measurement. Evolution of high speed processors and optical instruments over last two decades has brought revolutionary changes in scientific measurement techniques. Various issues related to hairiness measurement with reference to DIP such as selection of appropriate optics, mathematical formulation, and use of appropriate DIP tools were analysed. Existing hairiness measurement techniques and their drawbacks were also studied.

 

Design and Fabrication of a Machine for Making Prefabricated Vertical Drains

The most widely used prefabricated vertical drains (PVDs) use a corrugated (or suitably shaped) polymer sheet as the core and nonwoven material as sheath. Both of these are manufactured in India. But there is no company in India which combines these to make the PVD. As a result, the PVDs, which are being used in increasing numbers, have to be imported at a very high cost. Such a machine was designed and fabricated under this project.

 

Programmable Ring Traversing Device for Ring Spinning

Raising and lowering of the ring rail of a ring frame is usually done with a cam. This precludes the possibility of changing the building mechanism in a ring frame. A microprocessor based programmable traversing system was designed and fabricated. The capability of the system to perform for an extended period was demonstrated by building full cops. This will give the spinner the option of changing the build from cop build to roving or combination build.

 

Development of Novel Mechanisms for Dobby

A mechanism has been designed to provide dwells to the knives of a dobby at their extreme positions without using cams. This will allow the maximum shed depth to be reduced, leading to less strain on warp sheets compared to current linkage driven dobbies.

 

Design of a Yarn Hairiness Tester Working on the Principle of Image Analysis

The current generation of yarn hairiness testers aim to conduct the test at higher and higher speeds. As a result, the hairs have to be detected by indirect methods and some information about hairiness is lost. An earlier project has demonstrated that it is possible to use image analysis techniques to extract all possible information about hairiness from static yarn images. In this work, the possibility of capturing images of moving yarn and extracting complete information about hairiness was explored. Improvements in the image analysis algorithm used earlier allowed the images taken from moving yarn to be analysed. Improvement in hardware necessary for image capturing is likely to make this technique viable in future.

 

Modelling Yarn Balloon During Ring Spinning

In this project, the image of a yarn as it forms a balloon during ring spinning was captured. This was used to validate the currently accepted models of yarn balloon using a new procedure. The effect of changing the spinning parameters on yarn balloon was also observed. This may lead to a better way of positioning the balloon control ring.

 

Design of a Novel Drive for Rapiers

Most of the current mechanisms used to drive rapiers use prismatic pairs or a rack and pinion mechanism, none of which are suitable for high speed operation. The current project explored different design solutions which can eliminate these problems. The best possible straight line approximation of the coupler curve of a crank rocker mechanism was obtained and was suggested as a rapier drive.

 

Reducing Energy Required for Running a Loom

As the speed of a loom increases, the power requirement goes up due to losses in deceleration of oscillating links. This power loss is directly proportional to the cube of the operating speed. The current project established the theoretical basis of saving this power in rapier looms with use of springs and inertial elements. The method was demonstrated with simulations.

 

Warp Breakage Detection Mechanism

A wide range of options were explored for designing a warp breakage detection mechanism which would have some advantage over the drop-wire based system. An image analysis based system was demonstrated on a set of parallel threads which was assumed to simulate a warp sheet. The accompanying algorithm was also developed.

 

Novel Design of a Positive Rapier Head

Positive rapier heads can give higher weft transfer efficiency compared to negative rapier heads. However, they may cause damage to warp yarns and have a lower working life because of repeated impact between cam and follower. This work suggests an alternative design in which these deficiencies are removed. A model of the mechanism was made and run at speeds close to loom speeds. Analysis of clamp openings with high speed photography seems to suggest that the proposed design will be easier to work with at high speeds than at low speeds.

 

A Device for Saving Energy of Shuttle Picking

A magnet based device for saving energy of picking the shuttle of a loom has been investigated. Experiments on a shuttle like device indicated the possibility of 50% energy saving. The use of solenoids for energy regeneration has also been investigated.

 

Optimum Packaging of Stenter

The large size of the stenter requires individual components to be transported from the manufacturer to the user in multiple containers. A study of this topic revealed that the more general “container loading problem” needs to be solved for improving stenter packaging. This has also been described in literature as “Packing of small boxes in large boxes.” A search of container loading problems reported in literature led to the identification of two promising approaches. These two approaches were evaluated on a number of simulated packaging problems. The better amongst the two was identified. This was further modified by simplification of the solution procedure and addition of other constraints. Evaluation of all the methods revealed the superiority of the method developed in this work particularly in terms of time taken to solve problems involving large number of constraints. This is expected to lead to better algorithms for solving the “multi-container loading problem” which will allow the stenter packaging problem to be addressed.

 

Modification of Cotton Flyer

This study was undertaken to explore the possibility of reduction of material used for manufacture of the cotton flyer while keeping the flyer leg deflection due to centrifugal forces within acceptable levels. Simulation of different designs with the aid of finite element analysis indicated the possibility of 20% reduction without significant ill effects. Increase in air resistance in some of the proposed designs was estimated with the help of computational fluid dynamics. The corresponding increase in total power requirement of the speedframe was estimated to be well within acceptable limits.

 

Fault Detection in Yarns by Image Analysis

The capacitance based yarn fault detection used by the industry allows fault detection at very high speeds but does not allow any feature of the fault except length and diameter to be measured. Image analysis based yarn fault detection offers the possibility of feature/shape based classification of faults and may also lead to machine defect based classification of faults. Deliberately generated faults in a ring frame were analysed with image analysis techniques and classified with artificial neural networks. An accuracy of 70% was achieved and will be improved upon with ongoing research.

 

Energy Optimization of Stenter

The mass and energy flows through a stenter were modeled. A scheme for operating the primary change points of a stenter which minimizes energy consumption was suggested.

 

Fast Non-Invasive Methods for Yarn Cross Section Imaging

Visualisation of the cross section of a yarn by traditional methods requires 24 to 48 hours of sample preparation. Absence of a faster method has precluded its use as a quality control measure. In this work, two attempts to obtain the cross sectional view of a yarn has been reported – one based on optical coherence tomography and the other based on confocal microscopy. Limited success has been reported for both attempts.