Skip to main content

AGENT-based arChitecture fOr numerical Simulation

Objective

We propose to build a new approach to numerical simulation using agent technology, software components and web repositories, with an example in process modelling, thanks to the existing CAPE-OPEN standard. We will demonstrate the use of cognitive agents to support dynamic and opportunistic interoperability of CAPE-OPEN compliant process modelling components over the Internet, facilitating the provision of application services in process modelling, in a generic framework that can be applied to other numerical simulation domains. One industrial user, IFP (France), is the co-ordinator of the project; one major vendor of modelling software, Hyprotech (Spain), has a commercial offer based on this technology and is interested in becoming an ASP and components provider; three university departments: RWTH.LPT (Germany), UCL (UK) and LIP6 (France) are among the world leaders in modern computer-aided process engineering and in agent technology.

Objectives:
New distribution mechanisms for CAPE software : define a framework allowing simulation components to be distributed and referenced on the internet and intranets, facilitating provision of application services in process modelling; demonstrate it with Process Modelling Environments and Process Modelling Components. Knowledge-based dynamic configuration of simulations: define representations of requirements and services in form of an ontology of process modelling, that supports opportunistic configuration of simulators from internet components. Improvement of ASP business with knowledge-intensive middleware agents: design new facilities for supporting the dynamic matchmaking of modelling components. Further development of CAPE-OPEN: develop the CAPE-OPEN standards in the semantic and knowledge-based dimension. Example for other domains: a concrete example of supporting ASPs with middleware agents will be relevant for application service providers in numerical simulation.

Work description:
We will demonstrate the use of cognitive agents to support the dynamic and opportunistic interoperability of CAPE-OPEN compliant process modelling components over the internet, facilitating the provision of application services in process modelling, in a generic framework that can be applied to other numerical simulation domains. The work is organised in four work packages. Framework design and case studies: We will define the application framework for and propose relevant case studies. We will choose the software platforms needed for developing the cogent, we will define how these have to be modified in order to cope with our application needs, we will select among a number of architectural alternatives. In parallel we will identify a small set of case studies that will guide the selection of application-oriented agents, and found the basis for the prototype assessment. CAPE-ML development: We will design and develop of a formal description of process models in order to support the reasoning about model configurations, services and requirements. This will be the link between users request and services provided by the available cogents. A first version of the object model will be drafted in the first part of the project; it will be finalized after subsequent testing in the assessment phase. Application development: We will develop the prototypes along the proposed framework, including all tools and all agents. We will adapt the software platforms, develop three categories of cogents (PMEs, PMCs and utilities), and test them individually and collectively, independently of the full case studies. We will validate the architectural choices, debug the various components, and prepare the full scale testing. Assessment on case studies: We will extensively test the prototype infrastructure and agents on real-world application cases, and report on these testing in order to finalise CAPE-ML and to justify the exploitation perspectives.

Milestones:
COGENTS is building technologies for opportunistic combinations of software components on networked computers. This is a facilitating infrastructure for new business models in process modelling, and an opportunity for SMEs and niche providers to offer components in conjunction with environments of major suppliers. Milestones (months): 6: framework defined; 9: draft CAPE-ML exchange language; 18: application infrastructure and cogents prototypes; 24 (end); tests done on case studies.

Funding Scheme

CSC - Cost-sharing contracts

Coordinator

GROUPEMENT EUROPEEN DE RECHERCHES TECHNOLOGIQUES SUR LES HYDROCARBURES GERTH
Address
4 Avenue De Bois Preau
92500 Rueil-malmaison
France

Participants (5)

HYPROTECH EUROPE, S.L.
Spain
Address
Paseo De Gracia 56/4
08007 Barcelona
INSTITUT FRANCAIS DU PETROLE
France
Address
1 Avenue De Bois Preau
92500 Rueil Malmaison
RHEINISCH-WESTFAELISCHE TECHNISCHE HOCHSCHULE AACHEN
Germany
Address
Templergraben 55
52056 Aachen
UNIVERSITE PARIS VI PIERRE ET MARIE CURIE
France
Address
Tour Centrale 21Eme Etage - 4 Pl Jussieu
75252 Paris Cedex 05
UNIVERSITY COLLEGE LONDON
United Kingdom
Address
Gower Street
WC1E 6BT London