Industrial symbiosis (IS) is one of the alternative ways of using natural resources in industrial processes. Eco-industrial parks (EIPs), as commonly known areas of IS practices, increase resource efficacy and reduce environmental effects by implementing waste/by product exchanges among tenant plants. Although there is an increasing but limited number of EIPs around the world, their circularity is not ensured due to high dynamic market and business conditions. This paper aims at offering an innovative design approach for EIPs taking into account the potential waste exchanges between the plants potentially to be co-located within EIPs with the goal of eliminating adverse impacts of market and business dynamicity. To this end, first an analysis of an existing IS database is conducted and the sectors potentially to be co-located are identified. Second, inspired by natural ecosystems, the food web (FW) metrics are defined to measure the potential EIPs' circularity. Third, a non-linear optimization method, namely branch and bound algorithm, is adopted to decide which plants should be included in the EIP designs to maximize the cyclicity of the networks. Lastly, a location analysis is conducted in order to co-locate the plants and to minimize the operational costs of implementing and running the EIPs. The use of this integrated approach is illustrated in a scenario analysis for four theoretical EIPs, two taking the construction industry as an anchor industry and two considering the random inclusion of various industries that can exchange wastes with the constructions industry. These EIPs' FW metric values are compared with the biological FW averages of natural ecosystems. The results support the method's ability to design sustainable and circular EIPs and point out practical implications for practitioners and policy-makers. The study is a seminal one integrating three methodologies for the first time to design IS networks in the form of EIPs.