Rijkswaterstaat, the Department of Public Works in the Netherlands, produces slightly more than one lock a year but wants to have renovated 52 of its 137 locks in 20 years. A search with Eindhoven University of Technology for greater efficiency yielded a standardized language and the new startup Ratio.
The vast majority of locks in the Dutch waterway system date back to the last century. Many of them are at the end of their lifespan. The increasing intensity of waterway traffic also calls for expansion. That’s why Rijkswaterstaat launched the Multiwaterwerk initiative in 2014. MWW aims to renovate or replace 52 of the 137 locks by 2040. This will require a substantial efficiency drive, as the the Dutch Department of Public Works produces just over one lock per year.
“We were faced with the problem of renovating a large number of our locks,” says Han Vogel of Rijkswaterstaat. “Doing that one by one is very inefficient.” Vogel was a member of the team that had to address the problem. The solution had to come from standardization and innovation. Vogel stumbled upon an article on standardizing production processes by Eindhoven University of Technology professor Koos Rooda.
That eventually led to a doctoral study by Tim Wilschut. Vogel: “In the exploration phase, it quickly became clear that we needed to address two things at once: first, the standardization of the lock components and associated control technology and second, the standardization of the components’ production processes.”
“We started by looking at the extent of the variations in functionality, operating principles and implementation forms of lock systems,” recalls Tim Wilschut. “By comparing all the different lock types, we were able to arrive at a matrix and by clustering that matrix, we found seven groups of lock types that are very similar. Those groups form the basis for a lock family platform that you can use to efficiently organize maintenance, renovation or replacement in the future. We got a better overview of the common core components that are primary candidates for a standardization effort to save cost, time and effort.”
Wilschut also developed a common specification language, essential for engaging parties such as contractors. Vogel: “We want to make sure that contractors follow a standardized process. For that, you need a standardized language.”
In October 2018, Wilschut founded the company Ratio CASE (Computer Aided Systems Engineering) together with Tiemen Schuijbroek to deploy the developed tooling with the underlying Elephant Specification Language (ESL) at other customers. Wilschut: “ESL is a language for writing highly structured multi-layer system specifications from which you can derive automatic system architecture models.”
Wilschut says that in doing so, he’s taking a fundamentally different approach from existing methods for requirements management and modeling system architectures. “We’ve formalized the grammar of function, behavior and design requirements,” he explains. “This has a number of advantages. With ESL, it’s simply not possible to write very woolly requirements. In addition, the semantics of each sentence ensures that the specifications are readable by the computer.” This, according to Wilschut, allows the ESL compiler to automatically perform all kinds of quality checks.
Moreover, ESL captures dependencies between function requirements, behavior requirements and design requirements, allowing components and design variables to be automatically derived throughout the system decomposition. “This network of dependencies forms the basis of the system architecture model,” says Wilschut. “The architecture model is thus one-to-one equivalent to the written system specification.”
According to Wilschut, this means his specification language brings model-based systems engineering (MBSE) back to the early stages of the design process. “As soon as you start writing a specification, you’re already creating an architectural model,” he states. “In many existing SysML-like methods, requirement blocks generally contain purely natural language and must be manually linked to other elements. This linking is labor intensive and error prone, so the quality of the resulting model is often low and its usability limited.”
Wilschut finds that his approach often generates discussion, as it causes Ratio to kick against “a number of sacred cows” within classical systems engineering such as the use of separate function requirements and system decompositions.
At the Bits&Chips System Architecting Conference, 22 June in ’s-Hertogenbosch, Tiemen Schuijbroek of Ratio CASE and TUE’s Sjoerd Knippenberg give a talk on the MWW research.