Ontology-Based Requirements Management

Intervenant⋅e⋅s

Résumé

Requirements management is a primary challenge for the industrial enterprise. Development and operation of a capital asset involves tens of thousands of requirements, spanning multiple engineering domains. Requirements are issued by a variety of sources, including public authorities, standards bodies, operators, contractors and suppliers. With manual work processes and unsatisfactory coordination across stake-holders, requirements are a major source of cost. In this talk, we will describe how an ontology-based approach to digitalisation of requirements has been developed for the Norwegian Oil & Gas industry. We outline the methodology, and demonstrate a working prototype for developing, testing, and implementing digital standards. We also make a bold claim: that the standards of the future will be provided as ontologies.

In response to a well documented need for more cost effective requirements management in Oil & Gas, a joint industry project (JIP) has been established for the Norwegian Continental Shelf, to revise, restructure and digitalise the NORSOK Z-standards for Technical Information. Initiated by Standards Norway and DNV GL, and in close collaboration with the academic community and the Sirius research centre in Oslo, the JIP has found broad industry support. O&G authorities, the main operators, and the main engineering, procurement, and construction (EPC) contractors have signed up for a two-year common effort.

An efficient representation of requirements must be at the same time generic and sensitive to the characteristics of special subjects, whether in technical domains or in the management of asset information. This means that we need a common, multidisplinary vocabulary, uniform representation of requirements as rules, and a precise way to validate consistency and correctness of a set of requirements.

Ontologies in OWL DL fit the description for going from documents to databases. The distinguished feature is automated reasoning with description logic semantics. A modular ontology provides a multi-disciplinary vocabulary of class definitions, suitable for automated consistency checking against an asset described in that vocabulary. In the Norwegian O&G space we already have large-scale systems in production, applying reasoning over ontologies to support EPC projects.

Extending this basis to requirements, we need to manage normative modalities

  • requirements, as norms, are not always satisfied. We need algorithms to mangage defeasible constraints and agreed deviations from specifications. For this, a methodology has been developed, drawing on ISO/IEC 81346 for lifecycle perspectives and on the INCOSE Guide for Writing Requirements. A generic ontology for representing requirements has been created. OWL DL reasoning lets the computer derive which requirements apply to a type of equipment, or which requirements apply to a selected component; and in cases of deviation, the reasoner mechanism can identify where requirements are violated, and why.

A prototype implementation to support this methodology has been built, and will be demonstrated during the presentation at SemWeb.Pro. This includes a modular ontology, an expert friendly Excel format for building ontolgies from straightforward templates, and end-user services for stating, verifying, and testing a body of requirements. We obtain a tool for creating digital standards that are suitable for for use with existing software application and databases, as well as for information exchange.

Références

Auteurs/Autrices

Johan W. Klüwer, Dr. philos., is a principal specialist with DNV GL Digital Solutions. Since 2006, he has been engaged in making ontologies work for industrial applications. This includes implementation for Oil & Gas capital projects, adopting the ISO 15926 standard for OWL DL, and methodological work, most recently in contributions to Reasonable Ontology Templates (OTTR).