||In recent decades, numerous univariate/multivariate techniques have been developed and used for process monitoring and diagnosis. However, these techniques fundamentally focused on discriminating whether a process is in a normal or abnormal states. Therefore, it is difficult to use them for a diagnosis purpose indeed—detecting the root cause and identifying the disturbance propagation pathway. In this paper, we conducted causality analysis based on convergent cross mapping (CCM) to find correlation among variables in various relationships such as unidirectional, bidirectional, pan-in, pan-out, and combined system. We also investigated its applicability to the actual process through the simulation of the process model with noises.