About > As we enter 21st century
 
MJ

As we enter the Twenty-First Century

Mathai Joseph

The TRDDC of the mid-1990s, preparing for its role in the twenty-first century, went through a number of experiences that led to internal changes and to a major redefinition of its face to the world.

MATURITY AND CHANGE GO HAND IN HAND, WITH PEOPLE AND WITH institutions.  Change may happen organically, as with a natural coming of age, but the more interesting changes are inorganic, such as those produced in response to external stimuli.  The TRDDC of the mid-1990s, preparing for its role in the twenty-first century, went through a number of experiences that led to internal changes and to a major redefinition of its face to the world.

There were three familiar and persistent problems in large-scale software development: the difference between what the software does and what is actually required, the long development time resulting from the relatively slow pace of manual software development, and the unnecessarily close dependence of the software on a particular computer system. In 1994-95, work started in the Software Engineering group in TRDDC on finding new ways to overcome all three problems, using model-based software development.  The first version of this new software development environment was used for the TCS Quartz banking product, then under development by a TCS group in New Delhi.  Various changes were made in the environment and, by 1997, its effectiveness was established and Quartz became its first major user. In 1999, the development environment was branded as MasterCraft and launched simultaneously in the US and in India.  Measurements made showed that systems developed using MasterCraft were matched well to the actual requirements, programmer productivity went up five to ten times, and systems could be easily re-implemented for different execution platforms, thus meeting all three objectives set at the start of R&D project.

Around the same period, work in Process Engineering group in TRDDC on developing mathematical models for industrial plant operations had progressed to the stage where it was ready for industrial deployment. The first large-scale test of this capability came in 1996 when work started on developing models for optimization of some critical operations in a major cement plant.  The early results showed enough promise for full-scale industrial automation system to be build using the models.  The performance of this system was closely monitored and showed that there were  significant improvements in productivity (up six to seven percent) and product quality (variation in quality reduced by forty to fifty percent) and accompanied by a reduction in energy consumption (down five to six percent).  It was evident that similar solutions could be used for other cement companies, in India and elsewhere.  This led to the solution being branded as CemPac; the development team in TRDD moved into TCS as the CemPac business group and later grew into the broad-based Manufacturing Systems Group (and more recently into the large TCS Engineering and Industrial Service Group).

The creation of MasterCraft and CemPac were decisive steps in answering the question; ‘How does R&D in TRDDC affect the business activities of TCS?;  If these were the only steps, however they would have remained as isolated events in the twenty-five year history of TRDDC.  Instead, they became part of a process by which, over the years, in small steps and large, the results of R&D were applied to live TCS projects.  And they went beyond that: there were, and are, many TCS projects that for reasons of cost, feasibility or time-scale, could be undertaken only because of the high degree of automation made possible because of the software tools supplied by TRDDC.

Building good software tools requires a good understanding of the nature of the problems to be solved and application of the right results from computer and engineering science.  The challenge is not just to put together a prototype tools, but to do it in a way that can be scaled up tot meet industrial demands for performance and reliability.  For example, advanced software engineering tools must be build using techniques that will allow the tools to be used by hundreds of programmers and for developing programs of millions of lines of code, by teams that may be distributed over different geographic locations  Each of these brings major challenges and devising a solution that meets all of them requires both skill and experience.

At one level, the design and development of such tools provide a good test of the use of computer science in a practical context, at another level, the experience of the development can help to identify new problems that need to be addressed by computer science.  For example, elegant techniques that work well in isolation may often not be suitable to solve problems that appear in complex situations where a variety of different factors come into effect.  Solutions to one problems may in principle be applicable to a class of problems, but each solution may need so much reengineering that devising a single more general solution becomes infeasible.  This raises the question: ‘Is it possible to automatically generate specific solutions from a higher-level description or model?’

The TRDDC Software Tools Foundry was created for building a variety of customized software engineering tools, ranging from relatively simple tools for automatically checking programme conformance to standard to complex programme analysis tools for software transformation and re-engineering.  The first workbench for the foundry was Darpan.  In technical terms, Darpan was a language-independent programme analysis tool generator.  More simply, Darpan allowed customized software tools to be generated quickly to meet the specific needs of particular TCS projects, quickly, in this context, meant that in one week a team could product a tool that would otherwise have taken them one or two years.

Darpan served as the workhorse of the TRDDC Software Tools Foundry for about seven years. Numerous programme analysis tools were developed using Darpan and shipped out to TCS projects.  This experience provided many opportunities to study the strengths and weaknesses of Darpan and to think of ways in which it could be improved.  In 2004, work started on developing a more sophisticated framework called Prism. By 2006, all software tool production moved to Prism to allow a new set of software Tools to be generated.

For customizing process engineering models and tolls, a different kind of generalized solution was required.  Each industrial plant is unique, both in its performance characteristics and in its selection of equipment: there a probably no two plants anywhere in the world that are identical.  However, different plant that use the same raw materials and the same processes to product similar products obviously do have much in common.  How can the common features be modeled in a uniform way even if the plants differ in equipment?

In many cases, the differences in plant equipment just meant that input data came from devices with different characteristics and outputs hat to be sent to different devices to control the plant processes. Once the individual  characteristics of these devices are abstracted away, the devices can be treated uniformly.  The TCS plant automation systems therefore use a common set of models to characterize and predict plant behaviour and a layer of software to isolate these models from the characteristics of the input and output devices. This allows the models to be made sufficiently general to be used in many different plants.  On the other hand, the layer of software used to connect models to the plant equipment can take into account all the different kinds of devices that may be used.  This layer later become the APC Toolkit, now  a critical part of the TCS plant automation technology.

Abstraction, automation, models and generation are critical facets of work at TRDDC. They are used to devise solutions to a variety of problems, from software development and maintenance to industrial plant operation.  Some applications of this technology have been immensely successful, other have anticipated needs that are still in the future.  For example, work in the TRDDC Process Engineering group on renewal of different kinds of industrial waste, from tailings and slag generated during mineral processing to linings used for smelting, has in many cases progressed up to the feasibility analysis stage only to find that there is not yet sufficient interest in industrial take-up.  On the other hand, work in the TRDDC Software Engineering group on tools for data privacy have found immediate interest and solutions have been demanded almost before the research process has been completed.

Research work at TRDDC is one part of a large spectrum of scientific research, ranging from the theoretical to the purely practical. And there are strong scientific connections between different parts of the spectrum: what is not even solvable in today’s theory may well find a solution and applications in the near future.  So it is very important that scientists at TRDDC keep active links with their peers in the academic world and in the industry.  Over the years , TRDDC has established strong academic research collaborations in a number of institutions in different countries. There have been regular exchanges of people in both directions and the results of the joint research have been published widely.

The division of interests between the academic and industrial worlds has narrowed considerably.  The time needed to bring academic work into practical use is now far smaller than it has ever been , and is likely to reduce even further.  While solving major scientific problems continues to be the focus of academic research, many of these problems are now seen emerging from practical and industrial contexts.  New technologies provide new ways to meet the needs of existing industrial applications and to make possible new areas of application, but equally their use may reveal new problems.  For example, advances in networking have made enormously fast communication systems possible but have also demanded new high-performance algorithms that can work in real time to solve routing problems.

With many new problems coming into focus, some of the older research areas have become less important.  The emphasis in software development has moved away from programmes per se to systems.  Good code generation has meant that attention can now be focused on modelling requirements and with integrating systems composed  of difference software components. Underlying all this are the emerging standards of Model-Driven Architectures, which will certainly dominate software development in the next decade. The growth and proliferation of large data repositories will renew demands for more control over privacy and security, and these are areas in which major changes can be expected.  And the ability of systems to use past knowledge and to infer new rules will lead to a generation of far more flexible and agile ‘intelligent systems’ than anything we have seen till today.  Human interaction with systems will move away from keyboards to writing devices and to speech and visual input mechanisms that work with mobile devices.

Work in process engineering is on the brink of being transformed through nano-technology.  New materials with highly efficient properties are certain to make profound change in the way many common products are made, ranging from automotive tyres to new kinds of fabrics and the coatings.  Nano-technology  combined with microscopic Mems devices will be able to manage patient conditions very precisely and to provide measured levels of treatment.

These are all areas in which work has started in TRDDC.  The coming years will bring in more focus on specialized applications that make use of these new technologies to create new materials and devices that can see wide-scale proliferation.  With its wide-ranging achievements over the past twenty-five years and a new definition of its place in the scientific and industrial world, TRDDC is now poised to play its role in the twenty first century.