The Sino–US trade war has prompted China to re-examine the development of manufacturing, while the energy crisis restricts such development. Scientifically planning industrial energy allocation is important for supporting industrial transformation and the upgrading of manufacturing. The embodied energy flow in China’s manufacturing was investigated by reconstructing the energy flow network; taking a systems perspective, a fine-grained analysis of the emerging patterns and evolution of these flows in the internal and external manufacturing industries was performed, thus providing useful insights for energy planning. The results show that in the internal and external networks of Chinese manufacturing, most of the embodied energy convergence and transmission is concentrated in a few industries Moreover, it is clear that industries with stronger embodied energy convergence and conductivity are generally more likely to be associated with industries with weak convergence and conductivity. Preferential selection is an important mechanism for the generation of embodied energy flow paths. The choices of the embodied energy flow paths of various industries exhibit the preference that ‘the rich get richer,’ and newly generated flow paths are more likely to be chosen for connectivity to a path of strong convergence or conductivity. The embodied energy flow patterns of the internal network of manufacturing mainly include two-focus and multi-focus convergence patterns, while that of the external network of manufacturing is mainly a two-focus transmission pattern. Within in-edge networks, communities of high-end manufacturing have gathered most of the embodied energy, while in out-edge networks, communities of traditional manufacturing have been key in the transmission of embodied energy. The impacts of the internal and external network types, and of the in-edge and out-edge types on the stability of the embodied energy flow pattern are separate, and the embodied energy flow pattern is stable. Based on these findings, an ‘energy-related industrial cluster’ model is proposed here to aid in energy convergence and transmission, as well as to realize network cluster synergy.

University-industry collaboration networks consist of heterogeneous subjects; in fact, the interactions among these subjects lead to the emergence of the complex characteristics of university-industry collaboration networks and even to the emergence of the networks themselves. Thus, understanding the behavior that generates university-industry collaboration networks is essential to their construction and optimization. To more fully comprehend this generating behavior, we considered a university-industry collaboration network in the ocean energy industry in China as a case study. Employing the motif analysis method, we analyze the network’s generating behavior from four perspectives: motif characteristics, motif classification, motif combination and motif clustering. Furthermore, we explore agents’ modes of adaptive collaboration — and the causes of such collaborations — and summarize the characteristics and modes of networks’ generating behavior. The results indicate that universities and industries tend to form highly stable and close cooperative relationships. The constraints of the network’s generating behavior — which influences the subjects’ interactions through relational and structural embeddedness — include asymmetric information, resource and cost constraints, risk aversion and geographical constraints. Our research presents the benefits of using the motif analysis method to study the network’s generating behavior and its implications in constructing and optimizing a university-industry collaboration network.