Extention Request Report
Approved duration of the project
: From November 26, 1998 to November 26, 2001
Requested extension
: From November 26, 2001 to November 26, 2003 ( 2 years )
Original Objectives ( quoted
from the project proposal)
-
To provide viable mathematical framework
to analyse and comprehend efficient novel methodologies in technological
processeses.
-
To design appropriate mathematical
software for industrial problems supplemented with {\it Courseware} and
{\it Teachware.}
-
To establish computer aided instructional
environment equipped with a software library developed in-house.
-
To train scientists and engineers
in the realm of industrial mathematics for technical competence and human
resource development.
-
To develop a Network of Industrial
Mathematicians for collaborating with local industrial units on a continuing
basis.
-
To create an FTP site for public
domain numerical software packages to function as a mirror site of similar
centres in the world.
Results achieved so far (in relation
to attainment of objectives)
Summary of Progress
For the result accomplished
during November, 1998 to May, 2000, please refer our progress report mailed
to you earlier. We reproduce here the summary.
Research Accomplishments. Under
this project the following five classes of problems have been integrated
with the sole aim of linking them to industrial applications. The summary
of these classes of problems is given as under.
-
Genetic Algorithms (GA's) and Economic
Dispatch Problem
A survey on Genetic Algorithms
has been completed. This includes some hybrid algorithms alongwith Simulated
Annealing. The CAI Lab of the Industrial Mathematics Group now has a good
glossary of algorithms related to GAs.
These algorithms have been subsequently
used for economic dispatch problems posed as Travelling Salesman Problem
and Bus Scheduling Problem. A comparative study (on Bus Scheduling) between
GAs and Classical Techniques has also been undertaken.
Recently, the IMG group is working
at the use of neural network algorithms for solving Stiff Differential
Equations and Differential Algebratic Equations.
-
Dynamic Simulation by Differential
Algebraic Equations (DAE's)
The following developmental
works have been undertaken :
-
Role of index in numerical solution
to the system of DAE's:
When a class of numerical methods
say-- backward difference formula -- is applied to the differential algebraic
equations, it has been noted that the index of the problem is instrumental
in amplifying the truncation error (the amount by which the exact solution
does not satisfy the difference scheme) and the algebraic error due to
early termination of the iterative methods for solving the resulting system
of nonlinear algebraic equations. Further, it is observed that for higher
index problems, the algebraic variables can be categorized as index one,
index two, index three and higher index. Depending on the types of algebraic
variables, special scaling is needed in order to preserve the overall order
of convergence.
-
Automatic index determination:
Most of the existing software
packages like DASSAL, RADAU etc. for solving DAEs numerically, requires
the supply of index by the user. By adopting a Graph theoretic algorithm
of Pantelides, which has been developed for consistent initialization of
DAE's, the index has been detected for a wide range of problems.
Research Outcome
The mathematical model developed
in the project for an industrial slurry high density polyethylene reactor
gives rise to a class of discontinuous dynamical systems consisting of
DAE's. This problem which is worked out in collaboration with NOCIL results
in
-
Simulation supporting the prediction
-
Sensitivity analysis showing that
the manipulation of pressure yields a faster change in the reactor conditions
as compared to manipulating the catalyst feed rate
-
Tuning of the model parameters
Recently, following problems have
been investigated by the members of the IMG:
-
A literature survey on the tracking
across discontinuties in dynamical simulation problems.
-
Enhancement of general purpose public
domain DAE package called RADAUS by supplementing with inhouse developed
software autometic index detection has also been carried out.
-
Implementation of existing software
on tracking across the discontinuties in ODEs.
-
On-line Optimization: Algorithms
and Implementations
On-Line Optimization involves
networking of various modules. Each module requires a thorough study of
techniques used therein - Data Reconciliation, Parameter Estimation and
Error Detection.
A theoretical study of these
modules has been completed. Subsequently, two robust algorithms - Generalized
Reduced Gradient and Sequential Quadratic Programming - have been tested
on standard examples. They now form part of Resources generated under the
Project.
-
Market Assessment and Forecasting
Models
The developmental work undertaken
by the group for three different industries Pfizer Ltd., Associated Cement
Company Pvt. Ltd. (Thane), A.F. Fergusson & Co., Mumbai resulted in
-
Studying market conditions, evaluating
existing brands and comparing with new brands
-
Monitoring the relative positioning
of all brands and change in the market mix
-
Estimating and forecasting market
performance of the existing brands
-
Literature survey on market segmentation
-
Computer Aided Geometric Design
(CAGD) (Convexity Issues in Surface Designing).
Under this category, the main objective is to provide user friendly software
package on convex surfaces to the designer. The software module developed
by our group provides the Graphical User Interface(GUI) to determine whether
surface is convex. Graphical User Interface displays the surface along
with its Gaussian Curvature plot. If the Gaussian Curvature plot goes below
the X-Y plane along Z-axis, then the designer can safely conclude that
the surface is non-convex. The GUI uses freely available public domain
software package called `Geomview' to display the surfaces. Then if the
user wants, the module will search for the convex surface around the given
surface.The search algorithm used in the module is heuristic in nature.
The group has also developed a software which can compute the surface interpolating
a given set of points, which makes designers job to provide the input for
the prospective surface extremely easy.
The Software has been written in JAVA programming language. It provides
applets to the designer to give the control points of the surface and has
control over Geomview camera.The designer is also provided with the facility
to indicate the control points by just clicking on the Geomview camera.
Recently, the group has carried
out following :
-
Designed and implemented an algorithm
which can work as a Geometric modeller for the designers in the industries
to generate good convex surfaces interpolating the given points.
-
Designed algorithm to find an approximate
convexification of the given points, which is then interpolated by a patchwork
of surface patches. Patchwork satisfies the compatibility conditions and
convexity conditions so that the final surface is convex and smooth. The
algorithm developed in the first stage has been used as one of the modules
in this modeller.
IIT(IMG) - INDUSTRY MEET
-
IIT-Industry Meet between NRB and
IMG at IIT Bombay on November 20, 1999.
After a few rounds of discussion
with NRB ( NRB Bearing ), the IMG organized a meeting on 20th November
1999 at IIT Bombay to discuss the mathematical aspects of the industrial
problems pertaining to NRB. A total number of 31 participants out of which
seven senior executives from NRB attended the meeting.
The participants from NRB presented
some typical problems faced by their industry. After a due deliberation,
eight possible projects (see the report for details) were identified. However,
on priority basis the following three projects are taken up by IMG.
-
Statistical analysis of distortion
of shells due to heat treatment process.
-
Production planning and scheduling
of heat treatment operation.
-
Bend test for needle rollers.
-
IIT - Central Railway Meet
Mr. S. C. Sengupta, Chief Mechanical
Engineer, Central Railway briefly described the specific problems with
wheel breakage (fracture, bending etc) faced by Central Railway after adopting
air breakage system. he concluded his talke with broad outline of areas
such as
-
Optimization of the breakage system
-
Stress due to the break
-
Characteristic break loss
-
Composites for composition break
blocks
in which Central Railway is looking
for some co-operation.
After detail discussion, the
following possible broad based projects are identified with group members.
-
Mechanism and Optimization
-
Excess stress including thermal
stress
-
New material characteristic of composite
breaks
-
Statistical Data Analysis
The group alongwith other faculty
members from Metallurgical Engineering Department are working on some time
bound viable project with Central Railway.
Study Group Meetings
Study Group Meeting held on 19th
February, 2000, in the MS University Baroda, Vadodara.
The main theme of the meeting
was Eco-Modeling. A total number of 30 (out of which there were 9 senior
scientists and executives from industry) participants drawn from IIT Bombay,
MS University, Industry and R & D Organizations in and around Baroda
attended the meeting.
After a day long presentation
and deliberation, the following two projects were proposed:
-
Dynamic model for salinity distribution
in estuaries.
-
Statistical study of salinity changes
on populations density in estuaries.
Our resource persons in MS Univ.
Baroda are working on these two issues and we hope to provide some preliminary
results in future. Study Group Meeting held on 11th March, 2000 at IIT
Bombay.
The general theme was to understand
the types of problems faced by TRDDC, ABB Ltd., Crompton and Greaves Ltd.,
ADE, TIL etc. A total number of 15 participants from IIT Bombay, MS University
Baroda and Industries attended the meeting. Out of 15, there were 7 senior
scientists from industries represented the meeting.
The problems presented by the
participating industry and R & D organizations were technical in nature.
We expect that some of these problems will lead to research and developmental
projects at least of the level of M.Sc as well M. Tech.
Industrial Interactions through
Individual Contacts.
Apart from these group meetings
mentioned above, from time to time, the members of the IMG are interacting
with senior executives of various industries, namely; Hindustan Lever Ltd.,
Cytel Software (India) pvt. Ltd. (Pune), Associated Cement Companies Ltd.
(Thane), Pfizer Ltd., A. F. Fergusson & Co. (Mumbai), Hindustan Motors,
Godrej Soap.
Some of the major developmental
works already completed or in progress are as follows:
-
Development of Algorithm for Generalized
Estimating Equations.
The main objective is to design
statistical software which can be used for fitting the generalized linear
model to both longitudinal as well as non-longitudinal data. ( (with Cytel
(India ) Pvt. Ltd., Pune, Contact Person : Ajay Sathe, Vice-President,
Cytel, Project status: Completed)
-
Development of Algorithm for Log-Linear
Models.
The major theme is to develop
a module that does all the analysis that can be performed by LOGLIN statement
of PROC CATMOD (SAS procedure) using MS-Excel. (with Cytel Software (India)
Pvt. Ltd., Pune, Contact Person : Ajay Sathe, Vice-President, Cytel, Project
status: Completed).
-
Development of an Algorithm for
Logistic Model.
The main focus is to develop
a robust and efficient Java code that can take into account ``stepwise
selection'' and ``best subset selection'' in logistic regression. (Cytel
Software (India) Pvt. Ltd., Pune, Contact Person : Ajay Sathe, Vice-President,
Cytel, Project status: Completed).
-
Application of Statistics to Mineral
Processing Industry.
The objective here is to predict
the strength of cement from its micro structural characteristics. (with
Associated Cement Company Pvt. Ltd, Thane, Contact Person: Dr. D. Venkateswaran,
Managar (Research), Status: Project Completed)
-
Forecasting Sales and Product Positions
for Pharmaceutical Products.
The primary aim is to compute
brand index for various pharmaceutical products in order to judge the position
of the product in the market. (with Pfizer Limited, Contact Person: Mr.
Sanjeev Walawalkar, Status: Ongoing)
-
Identification of successful business
practices of Indian exporting firms.
The objective is to identify
business practices that explain export performances in apparel, fisheries
and spices industries. (with A. F. Fergusson \& Co., Mumbai, Contact
Person: Mr. Sameer Wadhwa, Status: Project Completed).
-
Issues in Photomask Generation-I,
II & III.
Developmentof algorithms. (with
Silicon Automation Systems, Bangalore : completed)
After the submission of the report,
the group is involved in following two projects :
-
Organization: Mahindra and Mahindra
Ltd., Tractor Division, Mumbai.
Basic Theme: Failure Data Analysis
Present Status: Ongoing
-
Organization: Indira Gandhi National
Centre for the Arts, New Delhi.
Basic Theme: Identification
and Enhancement of Cultural Heritage in Indian Villages- A Statistical
Analysis
Present Status: Ongoing
The group had some discussions with
Hindustan Lever Ltd. to explore the possibility of doing a project on crystallization
for them. Hindustan Lever is using crystallization as a way of purifying
speciality chemicals. The main focus of this proposed project is to obtain
good purity in shorter time. Substantial ground work has been done in this
direction.
The IMG had Some preliminary
discussions with Foxboro in Chennai to explore the possibilities of working
on problems of quality control in sugar industry. The objective in this
case is to produce the desired crystal size distribution without material
loss and in a short time.
WORKSHOPS AND CONTINUING EDUCATION
PROGRAMME
Workshop on Mathematical / Statistical
Software Packages for Data Analysis, Modeling And Simulation, April 22-24,1999
in the IMG Lab at IIT Bombay.
A total number of 25 from various
industries and R \& D Organizations - (NRB, TRDDC, ABB Ltd, Crompton
Greaves, ADE, CDAC, TIL, CICFRI, GES, Kirloskar Consultancy, GUIDE, WWFI,
ORG-MARG, MSU Baroda) registered for the workshop. The registration was
on payment basis.
Objectives : To train the participants
in the following aspects:
-
Demonstration of the power of some
of the modern Mathematical/Statistical Software Packages like MATHEMATICA,
MATLAB, OPTIMIZE, SAS etc.
-
Exposition to mathematical and statistical
techniques used in industrial problems.
-
Online computing experience through
tutorial sessions.
The workshop was conducted under Computer Aided Instructional (CAI ) environment
using Industrial Case Studies and Algorithms. Professional Development
Workshop in Engineering Optimization, February 29th - March 3rd, 2000
This workshop has been exclusively designed to cater the need of Optimization
Techniques in Industry. A novel feature of this workshop was real life
problem solving session. The main features were: Constrained and Unconstrained
optimization, Linear and Nonlinear Programming, Evolutionary Algorithms
(Genetic Algorithms, Simulated Annealing). Most of the techniques were
motivated through case studies. The tutorial sessions were supplemented
with hands-on experience on the in-house developed educational software
``OPTIMIZE''.
A total number of 13 participants from Industry were trained and this was
done on payment basis.
In-House Workshop on Industrial
Statistics held in the premises of Sterlite Copper company Ltd, July 22--23,
2000
A two days inhouse workshop
was organized in the premises of Sterlite Copper company Ltd. during July
22--23, 2000. About 23 employees participated in the programme. The main
focus was on the Sampling and Statistical Quality Control. Starting with
basics, a large number of case studies were presented in the workshop.
Proposal for In-House Course
on ``Engineering Optimization'' to be conducted in Hindustan Motors Ltd.
On a specific request from the
General Manager-QA & TQM, Hindustan Motors Ltd. a proposal on the above
In-House Course has been submitted.
The prime objective of the proposal
is to share with participants the power of state-of-art Optimization Techniques
that are available in the form of Software packages. It is also intended
to describe to the participants the systematic approach of optimization
which includes the problem definition, its mathematical formulation, selection
of a suitable method (algorithm) and a detail analysis of the solution
obtained.
CONFERENCE / WORKSHOP PARTICIPATION
PERTAINING TO THE PROJECT
Sanjeev Sabnis:
Participated the Workshop on
"Modeling Real Systems : a Hands-On First Encounter with Industrial Mathematics",
during 27th September-22nd October, 1999 held in ICTP, Trieste, Italy.
During his stay there, Dr. Sabnis also participated in the following two
projects:
-
Forecasting ozone level.
The objective was to use time-series
modeling techniques in order to
(i) study the main characteristics
of the variables and their linear relationships
(ii) make a one step ahead prediction
for ozone
(iii) make a prediction of daily
maxima of ozone.
-
Application of Inverse Problem Methodology
to Steel Industry.
The main theme was to use inverse
problem methodology to determine the unknown boundary of the furnace when
the inside temperature at a few locations is known.
A group of three faculty members
( Amiya Kumar Pani, Sanjiv Sabnis, Alladi Subramanyam and one JRF (Mr.
Vishwas Jadav ) from Industrial Mathematics Group attended a series of
Lectures on `Stochastic Process and Option Pricing', given by Professor
Rajiv Karandikar (ISI, Delhi) in the IGIDR, Goregaon during Nov. 23- Nov.
26,1999.
Ashish Misra (SRA, DST Project)}:
Attended three days short term
course on "Optimizing Engineering Design Using Evolutionary Algorithms",
held in I.I.T. Kanpur ( Organized by Prof. K. Deb) during 24th Nov. to
26th Nov. 1999.
Kannan M. Moudgalya:
-
One hour Invited talk on `` Dynamical
model of an industrial slurry high density polyethylene reactor'' in the
Workshop on Simulation and Computing for Industrial Applications under
Technology Appreciation Programme in IIT Madras in February'2000.
-
Attended, as a representative from
IMG, an Industrial Mathematics Workshop organized by Institut Teknologi
Bandung in collaboration with University of Twente, Netherlands in Bandung,
Indonesia, July 10-14, 2000.} Dr. Kannan worked on the problem of "The
effects of increasing the frequency and velocity on the operation and structure
of trains" for the company PT Kereta Api. The objective of this work was
to increase the throughput of trains between Jakarta and Bandung. A smaller
version of this case study was posed in the form of a linear programming
problem.
Amiya K Pani:
-
One hour Invited Bama Charan Das
Memorial Lecture on ``Industrial Mathematics: Key to Key Technologies and
Its Impact on Mathematical Research and Education in Mathematics'', National
Conference on ``Recent Trends in Mathematics \& Computing'' at ``27th
Annual Conference of Orissa Mathematical Society'', held in Bhubaneshwar,
during 11th-12th March 2000.
-
One hour Invited talk on `` Particle
Size Distribution in Emulsion Polymerization: Some Computational Issues'',
in the session `Applied Analysis and Industrial Problems' of the third
World Congress of Nonlinear Analysts held in Catania (Italy) during 19th
- 26th July, 2000.
Mohan C Joshi:
One hour invited talk on ``
Doing Industrial Mathematics via Optimization, DAE, CAGD and Market Forecasting'',
in the Third World Congress of Nonlinear Analysts held in Catania (Italy)
during 19th - 26th July, 2000.
Sachin B. Patkar:
Presented a talk on ``Applications
of Submodular Functions to Algorithms in VLSI CAD'', in the 17th International
Symposium on Mathematical Programming-ISMP2000, held in Georgia Institute
of Technology, Atlanta, GA (USA) during August 7-11, 2000.
Kirankumar R. Hiremath (SRF),
A. K. Pani, \& Kannan Moudgalya :
Presented a talk on ``Semi Implicit
Finite Difference Method for Particle Size Distribution in Emulsion Polymerization''
at International Conference on Nonlinaer Systems : Modelling, Simulation
and Applications held at Nanded, Maharashtra during December 26-29, 2000.
M. K. Ramaprasad (SRF) \&
M. C. Joshi :
Presented a talk on ``Online
Optimizaion Algorithums and Implimentations '' at International Conference
on Nonlinaer Systems : Modelling, Simulation and Applications held at Nanded,
Maharashtra during December 26-29, 2000.}
Ravi Shankar Gautam (SRF) \&
S. B. Patkar :
Presented a talk on ``Geometric
Modeller for Convex Surfaces'' at International Conference on Nonlinaer
Systems : Modelling, Simulation and Applications held at Nanded, Maharashtra
during December 26-29, 2000.}
Techniques / Software Developed
In-House.
-
Genetic Algorithms: Single Depot
Vehicle Scheduling Problems in Mumbai
-
Online Optimizations Techniques
in Chemical Processes (viz, Data Rectification, Parameter Estimation and
Economic Optimization).
-
Market Forecasting: Band Switching
Analysis, Algorithm developed for Generalized Estimating Equations; Algorithms
to predict cement strength from its micro-structural characteristics; Forecasting
Sales and Products for Pharmaceutical company; Algorithm for Logistic Model;
Algorithms for Log-linear model etc.
-
Generation of Convex Surfaces. Software
with Graphical User Interface to find out the convexity of a surface.
-
Graph Theoretic Algorithm for Automatic
Index Determination of DAEs.
-
Tracking Across Discontinuties in
ODE using Gears Algorithm.
-
Algorithm for Adaptive refiment
in the multiscale dynamical optimization using wavelets.
-
Algorithm for option Pricing
-
Algorithm for computing Volatility
-
Algorithm for Designing of controllers
for DAE systems
Educational Technology.
-
Teachware.
Enhancement of MATHEMATICA for
the purpose of teaching Optimization Techniques and Ordinary Differential
Equations. Using MATHEMATICA support system, the IMG has developed subroutins
of various optimization algorithms pertaining to one dimensional and multidimensional
unconstrained as well as constrained optimizations with graphical interface.
This module has been used for the development of an Institute Course--Engineering
Optimization. In view of very effective computational platform, MATHEMATICA
is ideally suited for illustrating various facets of techniques, theorems
and phase portraits of a system described by ODE. Efforts are on to develop
MATHEMATICA based subroutines for this purpose.
-
Course Materials.
The novel feature of the Lecture
Notes on Engineering Optimization developed by some members of IMG are
* Motivation through Case Studies.
* Explanation of Mathematical Techniques.
* Algorithmic Approach using MATHEMATICA.
*Graphical Interface.
Resource Persons
We have a network of resource persons for popularizing industrial mathematics
and arranging locally Study Group Meetings at MS University, Baroda, IISc,
Bangalore, IIT Madras and IIT Kharagpur etc. One of our resource persons
Professor P. R. Parthsarathy from IIT Madras conducted a workshop on Simulation
and Computing for Industrial Applications under Technology Appreciation
Programme ,held in IIT Chennai Feb'2000. Our representative Professor Kannan
Moudgalya attended the meeting and presented a talk.
New Initiatives
Initiatives with Industry
As a byproduct of this project,
the group is also actively engaged in the following Research and Development
programmes with Indian Industries.
-
High Density Polyethylene Reactor
(Source of Problem; NOCIL)
Apart from modeling and simulation,
the group is presently working at a new research and development problem
on tracking discontinuities efficiently in DAE modeled by the HDPE.
-
Emulsion Polymerization (with Hindustan
Lever Limited)
Emulsion polymerization is a
process of great industrial importance which is used for manufacture of
latex paints, adhesives, coating, binders in papers, textile products and
synthetic rubber. The mathematical model gives rise to a system of nonlinear
ODE, integro-differential equations and single first order nonlinear hyperbolic
integro-differential equation. The group has been working on a numerical
simulator for the last one and half years. Apart from computational issues,
this class of problems gives rise to some interesting theoretical questions
related to convergence of numerical schemes, existence and uniqueness results
for large time. More importantly, the question of asymptotic behaviour
remains a challenge in the field of analysis.
-
Modeling and Control of a Crystal
Size Distribution in an Industrial Crystallizer. (interested Industry:
HCL, Foxboro in Chennai)
The first part of the project
involves modeling the system. This yields a mathematical model consisting
of a first order hyperbolic PDE (population balance of the crystal size
distribution) coupled through its boundary condition with an ODE (mass
balance of concentration) to which a control can be applied. Solving the
PDE yields a nonlinear hereditary system for which standard control theory
can be applied. The output quality (shape of the crystal size distribution)
and yield (volume of crystals produced) depends strongly on the concentration
evolution. The general case (which is nonlinear and where the control is
to yield a concentration which is optimal over a range of time, instead
of at some fixed time which is the usual case) is not yet documented in
literature from a control point of view.
-
Online Optimization : Algorithms
and Implementation
It has been observed that in
Error Detection Module, one can use Neural Network Algorithms in conjunction
with Genetic Algorithm to obtain faster and more accurate convergence.
It is perceived that this approach will lead to some interesting problems
involving differential equations, optimization and numerical analysis--culminating
to new modular development techniques pertaining to On-Line Optimization.
-
Financial Engineering
The main focus is on the Numerical
Simulation of Option Pricing Models and validating it using data from Indian
Financial Markets. By adopting the existing models to compute the volatility
(roughly the market trend), some experiments were conducted on the data
obtained from some Indian companies and the option pricing using Black
-Scholes Model has been computed with the help of finite difference schemes.
-
Bus Scheduling
It concerns with the problem
of scheduling a fleet of buses to a given set of trips encountered by BEST
in Mumbai. As a preliminary step, a single depot vehicle scheduling problem
has been formulated as Zero One integer linear programming problem and
is solved using primal-dual algorithm. The model is tested on real world
data obtained from BEST.
Collaborative Research with Academic
Institutes.
-
Interaction with Foreign Universities
Professor Mohan C Joshi visited University of Kaiserslautern, Germany and
University of Twente, Netherlands during May-June, 1999 to explore the
possibility of collaborative research programmes between IMG, Bombay and
these universities. Both institutions are receptive to the idea of collaborative
programmes and efforts are on progress to come up with a joint programme.
It may be mentioned that University of Kaiserslautern has a very strong
group on Industrial Mathematics and is one of the core universities among
European Consortium on Mathematics for Industry ( ECMI). A linkage with
ECMI will not only help IMG in generating solutions for industry based
problems but might also lead to new academic programmes involving both
institutions-IIT Bombay and ECMI Universities. As a result of the IMG effort
towards cooperation, the following students worked with members of IMG
for their dissertation.
-
Jitendra V. Kadam, a M. Tech. student
in Chemical Engineering Department of IIT Bombay spent a year in the Process
Technology, RWTH, Univ. Aachen (Germany) for his dissertation on `Index
Analysis and Adaptive Refinement in the Multiscale Dynamic Optimization',
see for details, M. Tech. Project Report 2000-01.
-
Frederik H.W. Litjens, a five year
integrated Master of Science student in Applied Mathematics, Department
of Systems, Signals and Control from the University of Twente, Netherlands,
worked with the members of IMG for his final year project work on
Modeling and control of a crystallization system.
-
Generation of Possible Research
Projects
EU-Asia Research Projects. A
joint project entitled `Strengthening the Application of Mathematics
in Industry and Society' was generated in collaboration with
University Twenty (Netherlands), University of Kaiserslautern (Germany),
and Institut Teknologi Bandung (Indonesia). The main objectives of this
proposal are:
-
To start cooperation between the
leading research institutes in Europe and Asia in contact with industry
and society.
-
To work in the frontier areas of
science and technological developments through management of mathematical
knowledge.
-
To provide expert services in the
areas of Modeling & Simulation, optimization & Control, Scientific
Computing, and Industrial Statistics for specific industrial and technological
questions.
-
To train scientists and engineers
in the realm of industrial mathematics for technical competence and human
resource development.
APPLICATION POTENTIAL
LONG TERM
-
Optimal Scheduling of Airlines.
-
Modeling and Simulation of Particle
Size Distribution in Emulsion Polymerization: ( potential user industry
: HCL)
-
Crystallization Model and Simulation.
-
Computational Statistical Techniques.
-
Simulation of Option Derivatives
for Financial Industries.
IMMEDIATE
-
NRB work in progress related to
the three projects identified by the IMG.
-
CYTEL, Statistical Software.
-
BEST. Bus Scheduling in Mumbai.
-
Market Forecasting, Pfizer.
-
Computer Aided Geometric Designs
(Convex Surface Generation). (Godrej Soap)
Project Progress Indictors:
-
Ph.D Students:
Mr. Gautam, one of the RA's has joined our
Ph.D and has been working on the project related problems.
Ms. Jyoti Agrawal, a Ph. D. student from Chemical Engineering is presently
working on Discontinuity Detection for DAE System under regular Institute
Fellowship.
-
M.Tech:
Two completed. One is on Multiscale Dynamic Optimization & the
other is on Control of DAE jointly with Thomas Binder & Wolfgang
Marquardt from Aachen (Germany).
-
M.Sc Projects:
Seventeen Completed.
-
B.Tech Projects:
One on Emulsion polymerization (completed)
-
Technical
Persons Trained. (A total of 120 persons trained
from Industry, M.Sc. + B. Tech + M. Tech. etc.)
-
Research Publications Arising out
of the Present Project.
-
Patkar, S. and Narayanan, H.: Fast
On-line/Off-line Algorithms for Optimal Reinforcement of a Network and
its Connections with Principal Partition, to appear in Proc. of ann.
Conf. on Foundations of Software Technology and Theoretical Computer Science,
(FST TCS-20), Lecture Notes in Computer Science, Springer, India, 2000.
-
Venugopal, C.R., Rao, S.S.S.P. and
Patkar, S.: Priority Scheduling in Parallel I/O Systems, in Proceedings
of 1999 Intl. Conf. on Parallel and Distributed Processing Techniques and
Applications (PDPTA'99), June 28 - July 1, 1999, Las Vegas, Nevada, USA,
1999.
-
Venugopal, C.R., Rao, S.S.S.P. and
Patkar, S.: Parallel I/O : Modelling and Scheduling Policies, in
Proceedings of TENCON'99, Korea, 1999.
-
Patkar, S. and Narayanan, H.: Applications
of Submodular Functions to VLSI CAD, presented at ISMP2000: International
Symposium on Mathematical Programming, Atlanta, USA, August 7-11, 2000.
-
Moudgalya K. M., Ryali Venkatarao,
Modelling a discontinuous dynamical system I. An ideal gas - liquid
system. (in print) Chemical Engineering Science}, (IMG-RR-1999-1)
-
Moudgalya K. M., Jaguste Shubhangi,
Modelling a discontinuous dynamical system I. An Industrial Slurry High
Density Polyethylene Reactor,
(in print) Chemical Engineering Science, (IMG-RR-1999-3)
Clear statement of the objectives
that have not been achieved so far but will be achieved during the extension
period :
-
Parallelization of GA's for Economic
Despatch Problems pertaining to scheduling of buses and aircrafts.
IMG has investigated the solution
to bus scheduling problem from the point of view of application of Genetic
Algorithms and Linear Programming Problems. In view of enormity of bus
routes , it is very pertinent to investigate the efficiency of the algorithm.
This demands that we examine the possibility of parallelization of these
algorithms. This remains to be done and will form a major part of further
investigation.
-
GA's and ANN algorithms in evolving
new algorithms for Error Detection Module.
On-line optimization has three
key modules - Data Reconciliation, Parameter Estimation and Errror Detection.
Theoretical studies pertaining to these modules form a major part of the
objective. But , equally important is the use of GA's and NNA's in evolving
new algorithms as they do not require the knowledge of the gradient. In
On-line optimization functional representation is not known and hence the
absence of the knowledge of its gradient. We intend to surmount this difficulty
by resorting to GA's and NNA's in the remaining part of the project.
-
Tracking across discontinuities
in differential algebraic equations and development of efficient algorithms
based on local error estimates.
-
Hazardous & operability analysis
in a batch or semi-batch processing plant and its quantification with the
help numerical simulation of DAEs.
-
To design and implement algorithms
for interpolating given points in space by convex surface.
Since large surfaces are computationally
difficult to design, the group could develop it as a patchwork of surface
patches. We intend to investigate better convexity conditions for finding
the control points for the surface. It is most probably possible to get
a linearized convexity condition for each local patch which can set well
with the compatibility conditions for the patchwork leading to facilitate
the design of fast and efficient algorithm for the modeller. It is also
available to use techniques like subdivision algorithms and its analysis
via Fourier analysis and convex splines.
-
To complete the developmental work
of a Graphic modeller with features for surface fast volume computation
and interactive curves \& surface editing.
-
Construction of Brand Index ( it
is being pursued with Pfizer Ltd.).
The data collected using pilot
study has been analysed to identify important factors w.r.t. pharmacutical
products. Further, this analysis based on brands has been used to rank
the brands. The similar study will have to be carried out to on main data
sets to arrive at the brand index which will be used to track the pharmacutical
products.
-
Use of existing models and possible
development of new model for market segmentation \& their applications
to real data.
Various aspects such as modelling,
estimation related to diffusion and substitution phenomena have been studied.
In future, attempts will be made to apply the same to real life data and
to propose new models and better estimation procedures.
Back
to top