(590b) A Problem Solving Environment for Integrated Solvent and Process Design on the Grid

Grid technologies have received a lot of attention by the scientific community recently; however, developments are at their infancy and very few engineering applications have been reported to date. A typical application in the domain of process engineering that would benefit from grid computing is the integrated computer-aided solvent and process design. The design problem makes an ideal candidate for grid computing as it integrates a number of software resources such as computer-aided molecular design (CAMD) tools, data mining tools for the analysis of molecular design information, process synthesis tools and web mining tools with data repositories such as in-house and web databases required in the different design stages. At the same time all the above tools may be parts of different design stages spanning from conceptual design to detailed simulation models and may belong to various research groups or companies spread around the world. Furthermore, the optimization problems involved require vast screening of design options, which in its turn requires high computational effort. Grid technologies enable the development of an integrated solution environment for such a problem, where software and hardware heterogeneity is addressed through the use of appropriate tools and vast computing and storage resources become available.

In this context we develop a problem solving environment that interfaces and manages a number of available heterogeneous resources to be used in the integrated solvent and process design. The proposed architecture of the developed problem solving environment involves the following; The available in-house design and optimization tools exist in the form of modules and we develop a web-portal to render them available to the end user. The developed web portal further serves as a connecting node between remote, geographically distributed computing resources and external databases available through web channels. Such resources involve exploitation of computer networks for distribution of the design calculations and available data repositories for storage of design related information. The utilized computing networks include over 3000 processors while the data repositories have a storage capacity of over 100TB. The geographically distributed computing resources are managed by middleware available through a third party organization, which provides generic functionalities, is hard to use and lacks the automation levels matching particular requirements of a problem such as the integrated solvent and process design. In this respect, we develop a set of services that allow user friendly access of the service functions available at the third party middleware through the web portal, focus specifically to the requirements of the integrated solvent and process design problem, while also maintaining functionalities that facilitate the use of the existing services. One major service developed is the workflow generator, which allows the user to select from a number of developed solvent and process design workflows that render the design and optimization tools available to the user, depending on the design problem at hand. Following preparation of the workflow, the job scheduling service exploits a collection of services available at the third party middleware to launch the workflow in the distributed resources for execution, while it dynamically manages the selection of appropriate execution resources, the storage of generated data, and the termination of the workflow execution. Data management and resource allocation services are also available independently of the job scheduling services that allow treatment of the generated data (i.e. storage, retrieval etc.) in the available storage resources as well as querying of the status of the available storage and computing resources. Furthermore, we develop appropriate service functions that allow querying and retrieval of results in third party web-based databases through the web portal. The functionalities of the developed problem solving environment are illustrated through the solution of an integrated solvent and process design problem, while usability issues of the developed portal, computational characteristics and time performance of the workflows execution in the geographically distributed resources are also discussed.