Abstract for: Agile SD - fast, reliable, effective

This workshop explained and offered practice in using an "agile" method for developing system dynamics models - a means of escaping the common worry that building SD models takes too long, costs too much, and is unreliable. Participants brought along real issues they wished to model, or models they had already built, so they could see how the agile method compared. The agile approach starts by creating a very simple but working numerical model, developed quickly, hand-in-hand with users. This may take as little as 1-2 hours, but already provides insights. The process applies the basic principles of SD theory - performance depends on stocks, which are changed by their flow-rates, which are driven by decisions, external factors and existing stocks, giving rise to feedback. Being data-driven, the model is easily validated. This simple model identifies important unknowns, so the principles are repeated until users are satisfied that the model captures everything significant. During this development, important data-items are identified, and extensions may also be added, such as segmentation of important sections. The workshop ended by considering how established, standard models can add further to fast, effective SD modelling.