KleeneTDDs are useful for evaluating logic functions in the presence of unknown inputs, 0 or 1. Although KleeneTDD-based logic simulation is promising, the size of KleeneTDD for an n-variable function is O(3^n/n). Thus, when n is large, the KleeneTDDs are often too large to build. In this paper, we propose several methods to decompose KleeneTDDs. By using this method, we can generate smaller KleeneTDDs for sub-functions independently to reduce the necessary memory. Preliminary experimental results show that the effectiveness of the presented approach.