Summary Master Thesis Simone Köhler PostNL, the largest Dutch postal company, is facing lowering mail numbers and strong competition. To deal with these facts, a strategy is formed to pursue cost reductions in the postal supply chain to help mitigate the volume decline. One of the bottlenecks in the supply chain is the facing process of mailbox mail. The machine used in this process, the Culler-Facer-Canceller (CFC), is the reason of the bottleneck. By the use of a network agreement the collection process is adjusted to fit to the needs of the facing process (and therefore to the needs of the CFC's). Currently, there is a discrepancy between supply pattern imposed by the network agreement and actual supply pattern. This should lead to problems in the facing process, but this is usually not the case. Therefore it is expected that the coupling of both processes can be improved, leading to the main research question of this thesis: How can the coupling of the collection and facing process of mailbox mail be improved to increase the efficiency of the postal supply chain at PostNL by analyzing and modeling the processes? Before looking into the case of PostNL a literature study is performed. It revealed the postal supply chain satisfies the four characteristics of a service supply chain: intangibility, heterogeneity, inseparability and perishability. A service supply chains can be classified by many frameworks. Most of them classify the postal supply chain as a \textit{"service factory"}, implying that the operational challenges lie in capital decisions, technological advantages, managing the demand, scheduling the service delivery, fighting cost increases and maintaining quality. Methods to tackle the challenges are, for instance, Swift, even flow, Lean Manufacturing, Six Sigma and the Theory Of Constraints. What is taken away from these methods is that a bottleneck is a good starting point for supply chain improvement, and that even though the primary focus point is different in each method, the effects are similar. With this knowledge the PostNL case is investigated thoroughly. This case focuses on the sorting center of Amsterdam, and uses data from January to May. After a systematic process analysis and simulation with the help of a SIMIOmodel is concluded that the facing process showed three different states: a filling-, processing- and emptying-state. These are visible when assessing the input buffer of the system. However, models currently used at PostNL take the CFC's as a starting point to couple the collection process to the facing process, neglecting the presence of different states. Additionally, in current practice the belt conveyor, which transports mail to the CFC, delivers more mail to the CFC's than necessary. This results in a belt conveyor that often stops, preventing a swift even flow of mail through the system. With this knowledge, the current coupling model is transformed to a model that implies the different states of the process. Using the input buffer as the starting point is not the only improvement: the actual collection rides are used in the model to see tacit possibilities in the actual supply pattern. The model shows that, with the predicted decrease in volume, 8 rides can be bundled without creating idle time at the CFC's, resulting a decrease in costs of 120.000 euro/year in Amsterdam. There are also possibilities to shift rides without creating idle time at the CFC's, but the impact and feasibility is determined by factors out of the scope of this study. Next to changing the collection process, the process time of the facing process can be decreased caused by efficient coupling and decreasing mail volumes. This result is promising for the other five areas and sorting centers of PostNL. In total four main points for improvement can be derived from this study, which have a positive impact on the daily operations. The essence of the recommendations can be applied in all sorting centers of PostNL. Still, the exact degree of improvement might vary throughout the regions as each sorting center has a different hinterland and a (slightly) different setup for the facing process. The four points for improvement are listed below: By adapting the speed or fill rate of the belt conveyor to the CFC's, a more swift and even flow is created. This gives room to improve the manual culling process, causing the number of CFC failures to decrease. It is best to implement a new setup for the belt conveyor which uses a low belt speed in the first part (where the manual culling occurs) and a high speed in the other belt parts (as they only have a transportation purpose). This way the filling state remains short, even though the overall process performance increases due to the improved manual culling process. With the decreasing mail volumes it it possible to bundle rides. In the region of the sorting center in Amsterdam 8 rides can be bundled, saving 120.000 euro/year. This gives great promises for the other PostNL sorting centers. There are also several possibilities to shift rides when only mailbox mail is taken into account. To identify the actual opportunity for ride shifting also other mail flows (mail from retailers and mail from companies) and processes (for instance the delivery of parcels) have to be taken into account. Due to an improvement in the coupling of the collection and facing process, the process time of the facing process can be decreased. Unfortunately the working hours scheme, determining the working hours of the employees, can only be changed once a year. As the working hours have a direct link to the process time, the process time can as well be changed only once a year. For the new working hours schedule it is therefore advised to create a tight schedule, as the decreasing volumes over the year will make it fit soon. For sorting center Amsterdam is shown by this research that the process time can be decreased up to 25 minutes on a Monday. The developed model improves calculations concerning the coupling of the collection and facing process. It can be used for various purposes, from creating a network agreement, to giving insight when making decisions on assets, to giving support to a team leader during the facing process. Due to the wide application it is advised to use this model nationwide throughout the organization to obtain more accurate insights and results. The model has to be turned into an online tool, which can be maintained by a single administrator while the latest version can always be used throughout the organization. Based on this thesis there are several additional recommendations for PostNL. In the coming years there are plans within the organization of PostNL that might influence the volume of mailbox mail (for instance the plan to downsize the number of mailboxes). As a change in this volume directly influences the processes in the supply chain, it is important to keep track of the changes in volume. The same goes for the trend of the fluctuating volume throughout the year. For this purpose it might be necessary to improve the collection of data and the systems containing the data. The improved coupling model could be further improved by implementing stochastic behavior. Also, an automatic solver could be build to improve the ease of use of the model. Currently the data of the collection process has to be entered in the model manually, therefore a big improvement would be to automate this process. Finally, more insight in the process could be given to the Car Company by not only present the network agreement (based on percentage of a volume) to them, but also the necessary absolute volume at each point in time. Even though this research focuses on the service supply chain of PostNL, the drawn conclusions can be applied to more organizations. All systems with a high dependency within links can learn from the used coupling method, by acknowledging the presence of different states. When making agreements based on behavior of a single element, it must be an element close to the coupling point. This way the needs of one link in the system can be closely approached and translated to demands for the other link in the system. Even when only one of the links can be adapted, the improved coupling model can give insight how to adapt this link optimally to the behavior of the other link. When it comes to supply chain research in general it is recommended to not study the reasoning behind improvement points for to long, but focus on creating a wide base of possible improvement points and methods for supply chains with certain characteristics. The main goal of this, is to create a good link between the cluttered literature in this feel of study and the actual practice.