||Recent disasters have revealed growing numbers of citizens who participate in responses to disasters. These so-called spontaneous unaffiliated on-site volunteers (SUVs) have become valuable resources for mitigating disaster scales. However, their self-coordination has also led to harm or putting themselves in danger. The necessity to coordinate SUVs has encouraged researchers to develop coordination approaches, yet testing, evaluating, and validating these approaches has been challenging, as doing so requires either real disasters or field tests. In practice, this is usually expensive, elaborate, and/or impossible, in part, to conduct. Simulating SUVs' behaviors using agent-based simulations seems promising to address this challenge. Therefore, this contribution presents a conceptual model that provides the basis for implementing SUV agents in simulation software to perform suitable simulations and to forecast citizens' behaviors under a given set of circumstances. To achieve adequate simulations, the conceptual model is based on the identification of 25 behavior-affecting attributes.