An important task in railway planning is to construct the annual timetable. In Sweden, this is coordinated by the Swedish Transport Administration (Trafikverket) through a process in which operators apply for capacity for their services. The Swedish railway system is characterised by a heterogeneous traffic mix. This leads to planning challenges related to mixed speeds and varying requirements for regularity and flexibility, as well as a need to prioritise between different traffic patterns. One practical question that may arise in this process is how the cyclic requirements of periodic services should be valued relative to other traffic and other aspects that are also important for railway operations.

This study therefore examines how regularity requirements affect timetable robustness and how these properties are related. A key question is whether timetable robustness can be improved by allowing controlled deviations from strict regularity requirements, and how this in turn affects traffic punctuality and the social welfare of the railway system. To investigate this, the study focuses on double-track lines where traffic is separated by direction. As the conditions differ between passenger and freight traffic, the study is primarily limited to passenger traffic.

For the above purpose, the study proposes an extension of a non-periodic optimisation model so that chosen trains can also be planned in a semi-periodic manner. The model is used to generate Pareto-optimal timetables with respect to regularity and robustness, and their punctuality is analysed by simulation. To evaluate how social welfare varies across the generated timetable alternatives, an agent-based socio-economic model (AGSIM) is also proposed.

The main results of the study show that semi-rigidity may be important for scheduling heterogeneous traffic with both periodic and non-periodic services. If no deviations from strict regularity are allowed, feasible solutions may be unavailable, whereas the intended regularity of services is lost if all traffic is treated as non-periodic. The case study also indicates that relatively small deviations may be sufficient to achieve better robustness, and that the marginal benefit of larger deviations declines rapidly. This implies that there are attractive compromises that naturally balance regularity and robustness. The simulations also showed that punctuality could be improved, while the social welfare of the optimised timetables was marginally lower than that of the given timetable. However, the differences in social welfare were relatively small.

The case study demonstrates that the proposed method, that combines timetable optimisation, simulation and socio-economic analysis, can be used as a decision- support tool for evaluating timetable alternatives. The results also indicate that increased punctuality should not automatically be equated with increased social welfare, as this may depend on the measures required to achieve it. It may therefore be justified to combine traffic simulation with socio-economic evaluation in order to investigate the effects of different decisions more broadly. Further research should  focus on further integrating the models, extending them so that they can be applied to networks, and including freight traffic.