In recent years, almost every manufacturing site has been supported by a lot of part-time, temporary, or mid-career personnel, given the poor state of the economies all over the world. However, expert managers of front-line workers have to design more complex methods of production planning, when urgent orders due to uncertain elements like a disaster are taken into consideration. Therefore, this paper proposes a sustainable production planning model using an inference methodology. The method was suggested by Mahadevan et al. in the field of architecture, and some cases of its effectiveness have been shown. In this paper, we try to incorporate that method into our sustainable production planning model. First, a work element that overflows into another process is assumed to be a "Dummy Moving Element (DME)". Next, DME estimation analysis and improvement using Bayesian estimation are discussed. Finally, the effectiveness of our model is verified by a numerical experiment.