Actus Integrated Management (AIM) and Partners in Development (PID), two small civil engineering companies, were successful in a bid to investigate the sanitary sewers (wastewater network) of the Msunduzi municipality. The project involved mapping the pipe network comprising 1450 km of pipe and making an assessment of the opportunities for recycling wastewater. A GIS solution was implemented for the mapping task and use was made of two FOSS GIS systems. In this paper the project team's selection of a FOSS GIS solution over a proprietary solution is explained. The team's experience with the two software suites is described and some recommendations are made to assist similar organisations considering a FOSS GIS solution.

A limited review of FOSS GIS systems was conducted and this led to the short-listing of two software suites. The products selected for the project were the Geographic Resources Analysis Support System (GRASS) and Quantum GIS (QGIS). The two software suites were deployed on a personal computer running the Ubuntu Gutsy operating system and concurrent graphical user cessions were provided by means of the proprietary NX system by NoMachine. This solution was chosen over a proprietary GIS solution involving multiple copies of ESRI's ArcView 9.2 software. The choice of the GIS solution was influenced by cost, deployment time, capability of the software, the software's ease of use and the perceptions of the team leaders based on limited exposure to FOSS GIS.

At the outset the project team had fairly limited GIS experience, but fairly well developed CAD skills. QGIS was found to be particularly easy to use and GRASS provided all the analytical tools required by the project. These two products are complementary, which enabled the team to leverage the best of both products. The scope of the GIS task was to produce a single plan of the city's sanitary sewer network. The pipes were to be digitized from 1 : 2500 scale paper plans and the pipe sizes were recorded in attribute tables. Some work had already been done and this was provided as ESRI shape files. Substantial cleaning of the supplied data was needed to eliminate dangles and overlapping pipes. The paper plans were several years old and only covered a portion of the city. Various engineers' drawings in CAD formats were used to patch in missing or updated information. Particular challenges included working with a large set of high resolution aerial photography, working with data in different co-ordinate systems and working with data in various CAD formats. Neither of the software suites were able to satisfy the team's requirements for large format, high resolution paper plots. Thus a solution involving a proprietary CAD product was implemented. The ready availability of user manuals, tutorials and support via the project mailing lists is considered to be one of the primary reasons for the success of the project.

The team successfully completed the GIS task of their project within the imposed budget and time constraints. The FOSS GIS software suites performed as expected and no unexpected limitations were encountered with the products. A review of the project has highlighted aspects of the implementation that could have been improved. Drawing on personal experience from this first FOSS GIS project the authors offer some recommendations to other organisations, which may be evaluating the FOSS GIS alternative.