Developing Modeling Languages - Two Models

Published Apr 24, 2013

A major problem facing many engineers today is the sheer scale of information exchange and communication required in product and process development projects. This particularly affects Mechatronics engineers, who are not only reliant on their multidisciplinary knowledge and skills to tackle the task in front of them, but are also an increasingly in need of a broader awareness – alertness, even - when interacting across disciplines with other design teams and stakeholders.

The many variables involved in efficient engineering project management make life hard enough, but as the level of complexity rises and the number of people involved increases, how do you ensure that the human element is reliably supported by automated, accurate data - to avoid design errors, inconsistencies and product delays - without the brain being overloaded by the immense volume of information?  That’s without going into the issue of optimizing knowledge transfer and  preparing in advance for product life-cycle issues.

Implicit dependencies between documents of multi-domain engineers (courtesy of Chris Paredis, 2010)

A model-based design process can promise better complexity management, improved design quality, better knowledge transfer and re-use, and improved communication (Friedenthal et al, 2008).  However, challenges such as tool integration, model management, data exchange, model interoperability and consistency management do not address themselves (or so we have found here at the KTH Mechatronics Division!).  Two of our graduate students have recently successfully defended their research into improving the current situation – which will make life easier for future generations of engineers.

has devised a modeling language to describe and develop tool chains for embedded systems. The Tool Integration Language (TIL) aims to increase the level of automated support available to those working in complex product development.   His thesis also outlined the need for a more systematic description of the tool chains themselves, since these are often highly individual, evolving from general purpose modeling languages and then tailored to suit a specific problem in a specific domain – i.e. usually not easily transferable or transparent.  In November last year, Matthias and Martin Törngren won the International Academy, Research, and Industry Association 2012 Best Paper Award for their contribution "Constructing Tool Chains based on SPEM Process Models".

has approached the problem from a different angle, focusing on how to reliably manage consistency in the decision making process when designing complex engineering systems.  He proposes a Dependency Modeling Language (DML) that - among other things - identifies and maps patterns that can aid in filtering vast amounts of data into useful and usable human-sized chunks.  For example, DMLs can generate the different views and levels of detail needed to perform different tasks, implement efficient consistency checking, cross-domain model transformation sand more – essentially offering a less labour intensive (and therefore less error prone) management of the details, freeing up the human engineer to focus on the more subtle aspects of complex product and process development.

Drs Biehl and Qamar are currently working together on ways to co-ordinate the development of these two languages.   Ahsan has also written a short popular science article , using examples from the Aerospace industry to illustrate the scale of the problem and the ultimate human cost when things go wrong.