Design and evaluation of a model-driven decision support system for repurposing electric vehicle batteries

Abstract

The diffusion of electric vehicles suffers from immature and expensive battery technologies. Repurposing electric vehicle batteries for second-life application scenarios may lower the vehicles' total costs of ownership and increases their ecologic sustainability. However, identifying the best-or even a feasible-scenario for which to repurpose a battery is a complex and unresolved decision problem. In this exaptation research, we set out to design, implement, and evaluate the first decision support system that aids decision-makers in the automobile industry with repurposing electric vehicle batteries. The exaptation is done by classifying decisions on repurposing products as bipartite matching problems and designing two binary integer linear programs that identify (a) all technical feasible assignments and (b) optimal assignments of products and scenarios. Based on an empirical study and expert interviews, we parameterize both binary integer linear programs for repurposing electric vehicle batteries. In a field experiment, we show that our decision support system considerably increases the decision quality in terms of hit rate, miss rate, precision, fallout, and accuracy. While practitioners can use the implemented decision support system when repurposing electric vehicle batteries, other researchers can build on our results to design decision support systems for repurposing further products.

Transactions for trading used electric vehicle batteries: theoretical underpinning and information systems design principles

“…Performing an efficient matching using an information system lowers the efforts required for selecting appropriate batteries and, thus, reduces a second life manufacturer's transaction costs. Associating EVBs and scenarios can be described as a bipartite matching problem that can be solved using linear programming decision models to increase the quality and speed of the decision-making process (Klör et al 2018).…”

Untitled

“…Obtaining a high response rate from participants of Delphi studies seems crucial since it significantly enhance research rigor and validity . The studies (Ågerfalk et al, 2009;Chiravuri et al, 2011;Chou et al, 2014;Daniel and White, 2005;Klör et al, 2017) in our sample show a high response rate of experts reflecting a successful procedure implemented to engage the experts in the Delphi process. In (Daniel and White, 2005), the authors link this success to the involvement of experts that were directly recommended to them.…”

“…The Delphi method was often used for exploratory purposes in the analysed IS studies. However, some authors apply the Delphi method to test hypotheses ( (Aloini, Dulmin and Mininno, 2012;Chen et al, 2014;Chou, Zahedi and Zhao, 2014;Görg and Bergmann, 2015; as well as experimental (Chiravuri, Nazareth and Ramamurthy, 2011;Klör et al, 2017;Nah and Benbasat, 2004) and descriptive studies Saunders and Benbasat, 2007). Some studies used the Delphi method to elicit the as the basis for the development of a theoretical model or an assessment instrument that was later tested using the survey method Chou, Zahedi and Zhao, 2014;Görg and Bergmann, 2015).…”

“…In our sample, descriptive and explanatory research tend to recruit large panel sizes including 73 experts in single panel , 54 and 76 participants in two panels (Saunders and Benbasat, 2007), and 64 participants in two panels . Exploratory studies, however, have implemented small panel sizes, which range between 10 to 23 experts as founded in the (Ågerfalk, Fitzgerald and Slaughter, 2009;Aloini, Dulmin and Mininno, 2012;Görg and Bergmann, 2015;Kasi et al, 2008;Kasiri, Sharda and Hardgrave, 2012;Klör et al, 2017) studies. The only exception is (Daniel and White, 2005), that while being an exploratory study it involved a panel of 35 experts.…”

“…For most of the studies analised this was achieved in 2 or 3 Delphi rounds (Ågerfalk, Fitzgerald and Slaughter, 2009;Chou, Zahedi and Zhao, 2014;Daniel and White, 2005;Görg and Bergmann, 2015;Kasiri, Sharda and Hardgrave, 2012;Liu et al, 2010;Nah and Benbasat, 2004;Saunders and Benbasat, 2007;Singh, Keil and Kasi, 2009). Four examples deserve to be noticed; while in the studies of Klör et al, 2017) researchers stopped their Delphi rounds due to the high stable responses they were receiving from all experts. (Singh, Keil and Kasi, 2009) stopped their study after the second round due to some degree of panel fatigue and to avoid potential panel drop-off from the study.…”