Meeting customer demands for order-based production and make‐to‐stock production policies against holding and non-holding costs are fundamental functions for businesses to ensure. For these policies, finite capacity buffers between machines is of great importance. WIP, production rate and profit values, the key performance indicators of the transfer line, affect the sustainable economics of companies. It is important to investigate how the production rate, one of the most important performance indicators, and its CPU time are affected by the reliability parameters of the machines, the convergence rate and the analytical methods applied. In this study, the theoretical computational convergence analysis of the Dallery-David-Xie (DDX) algorithm is conducted on balanced transfer lines consisting 20, 30 and 50-machines with four different reliability parameters, each having finite buffers. The results show that the performance of the DDX algorithm is very sensitive to the convergence rate. The CPU times spent based on the different convergence rates used in the applied DDX algorithm significantly differ from each other at a 95% confidence interval. Additionally, the study investigates uniformly, ascending order and descending order buffer distributions to maximize the profit value and minimize WIP in the transfer line. The initial buffer configuration affects the key performance indicators on balanced transfer lines with different reliability parameters.