Abstract for: Elasticity formulations: theory and practice in System Dynamics

The adequate representation of non-linear relationships (NLRs) in SD models is often challenging. The implications of using alternative formulations for formalizing NLRs can affect in a significant way both the efficiency of the modeling and calibration process and the accuracy of the results produced. The use of elasticity formulations to represent such relationships is not new to SD as they have been used from the inception of the field, including by Jay Forrester (1956, published in 2003) and are covered briefly in Sterman’s popular SD textbook (2000). Elasticity formulations can be very effective and efficient in some circumstances, and their low occurrence in the system dynamics literature may indicate a lack of knowledge of their characteristics and usability that favor the more popular alternative: table functions. Our review indicates that sufficient guidance about how to use elasticity formulations in SD models and about the conditions under which this appears justified/advantageous has thus far been lacking. This paper attempts to address this gap by a theoretical discussion about the properties of elasticity formulations, a brief literature review including criticism of their use, prerequisites for use (including testing), and a comparison of the approach to table functions in terms of advantages and limitations and in terms of practical modeling examples.