We present a modeling framework for a production line consisting of multiple robotic assembly cells by integrating probabilistic modeling and temporal logic verification techniques. A production line is a complex system comprising of multiple assembly cells which in turn comprises of multiple assembly stations. We evaluate the performance of the proposed operations plan by decomposing the whole system into smaller subsystems which are analyzed separately. This allows us to design optimized systems. We first model an assembly cell in a probabilistic model checking tool. Contingencies may arise in each cell during the production operations and are resolved by humans or the assembly robots themselves. We simulate multiple cells operating cooperatively to finish production of parts. We perform formal analysis for each cell, which is a sub-system of the production line and then study the aggregate behavior of the resulting production line. Based on our analysis of individual cell behavior, we can determine how to change cell settings to achieve the best production performance.