CSIRO has a long history of modelling arc plasma processes, including arc welding, plasma torches and plasma waste conversion. The models use the methods of computational fluid dynamics applied to viscous incompressible flows, extended to include an energy conservation equation, Maxwell's equations, and additional source terms to take into account plasma effects. The models have been used to gain a scientific understanding of arc plasma physics and chemistry and, as they have gained sophistication and accuracy, to improve and optimise industrial processes. The computer codes that have been developed have, until recently, relied on expert users with familiarity with the detailed operation of the code, and access to a FORTRAN compiler. In a typical implementation, input parameters are provided using user-editable text files, progress towards convergence is monitored by viewing a continuously-updated text file, and the results of the calculations are written as text files and as data files readable by graphics programs. Input and output file handling is not automated; the user is required to rename files to avoid overwriting, and to ensure that all relevant files are stored together. The provision of an appropriate start-up file (for example, a solution obtained for a similar set of input parameters) has required the user to select the file; this relies on the user carefully documenting the input parameters for all solution files. One of CSIRO's arc welding codes has now been packaged into a software product, ArcWeld, using CSIRO's Workspace workflow framework. This advance was motivated by a customer, General Motors, requesting that the code be easily usable by its welding engineers and technicians. The model is three-dimensional and treats the full arc welding process, with solid, liquid and plasma regions included in the computational domain. Despite these factors, physically-based simplifications are used to ensure that ArcWeld can be run under 64bit Windows on standard desktop computers. A simple GUI is used for entry of input parameters, starting the computer code, displaying progress towards convergence, and access to graphical output. The most appropriate start-up file is automatically selected based on the input parameters. All input and output data are written to a user-selected directory. This work has had important benefits. For the scientist, using a workflow platform means that common components and functionality are available as prewritten and pretested operations, so scientists can focus more on their core science and not as much on software development. For end users, the code can now be run by non-experts, who require only very basic training and access to a desktop computer. The input parameters are automatically stored with the output data, reducing the reliance on user documentation, and the user can easily visualise the results. These changes mean that the science can be advanced more quickly and the software can be used 'on the factory floor' to help optimise welding processes. More broadly, t...