The overall research objective has been to develop a high reliability safeguards model for a full, commercial-scale pyroprocessing facility based on the fundamental principle of safeguards-by-design. To this end, a quantitative safeguards model with the metric of false alarm probability has successfully been developed for the fuel fabrication process to the point of being useful and flexible in practical nuclear material accounting (NMA), particularly with weight measurements. Fuel fabrication was selected for study because a large amount of special nuclear material is included in this process. The current model allows for adding key measurement points (KMPs) or sub-processes, and it updates an inventory difference (ID) calculation at each KMP. A facility design (the baseline design), which contains a KMP between the melting and trimming processes, reduced the product throughput compared to the equipment design without the KMP, given the same false alarm probability of 0.05 (It is based on the IAEA recommended target value). Therefore, the operator will prefer the equipment design to process more material while the baseline design can tell the exact location at which a diversion was attempted. In this paper, a sensitivity analysis has been developed for various facility parameters, such as false alarm probability, melter failure rate, alarm threshold at each KMP, etc., to observe the results in the productivity of operation. With a false alarm probability of 0.1, about ∼8% less material had been processed compared to that of 0.05. In addition, if the melter failed more often than once per month, less material was processed, and vice versa. Changes related to the heel amount accumulated in the melter did not affect the number of campaigns because the model correspondingly calculates the alarm thresholds to achieve the same desired false alarm probabilities. If a different false alarm probability was given for each sub-process unit; i.e., an alarm threshold was set differently at each KMP, safeguards on sub-processes could effectively be applied. For instance, a sub-process, which includes more amount of special nuclear material than other processes, can use a high false alarm probability to detect the diversion attempts following the higher detection probability. In addition, the functionality of random sampling for physical or interim inventory verification was tested. This concluded that a smaller container of Pu compared to 20 kg would be necessary for reducing the time for physical or interim inventory verification, given the relatively high detection probability.