Abstract Summary
The infrastructure manager (IM) DB Netz AG (Germany) is responsible for a safe infrastructure under the traffic operated. The infrastructure is on average 80 years old and designed for the traffic and the respective load models known at the time of construction. Traffic changes, not in the maximum axle loads 22.5 t in passenger traffic), but in the utilisation of the permissible minimum load bearing capacity expressed by the EN line categories [EN 15528]. Whereas in the past, in passenger traffic, heavy locomotives or power cars were used with light wagons (e.g. 16 t), wheelset loads of up to 22 t are achieved over the entire train length. In consequence, the known dynamic coefficients lose their validity. In addition, train sets are becoming more efficient, so that regional multiple units are now expected to operate at speeds of up to 200 km/h and locomotive hauled passenger vehicles at speeds of up to 230 km/h. Due to cost pressure for railway manufacturers, different customised vehicle concepts with different excitation patterns (locomotive + wagon pushed, multiple units with/without power car, different wheelset load patterns AB, CB, SA, excitation of different wavelengths) are developed. An adaptation of infrastructure to the new rolling stock is possible for new bridges (built for other reasons than dynamics), however, the renewal of existing bridge structures is very costly (disruption of traffic, planning and construction costs). The IM has, therefore, always strived to maintain the infrastructure as long as possible. When introducing new trains such as ICE4 or ECx, verifications are necessary at a large number of bridges of the overall network or at least the operational network. As simple estimates are becoming increasingly difficult in this context, the IM calculates individual existing structures dynamically (step by step concept DB Netz in [TNB, Grunert 2022]). It becomes apparent that the hidden reserves of the infrastructure (difference between model and reality) must be proven through measurements in order to update the models and prolong the lifetime of certain structures. There are also bridges in the inventory which, for historical reasons or due to ageing, have less load bearing capacity than required for LM71. The heterogeneous DB bridge stock with its characteristics is challenging, especially if verifications are only scarce. For this purpose, databases are used to hold the structure’s data, to supplement it with load data (e.g. passing vehicles’ timetable), to complement the data with calculation results and to assign it to an assessment. In addition, a post processing presentation method of the results was also developed in order to derive simplified verifications of existing or new structures. The contribution will consist of the following parts: 1) short overview of existing bridge stock 2) introduction of the used data landscape 3) presentation of the static and dynamic effects in v/n diagrams, derivation of monitoring installations 4) application of Machine Learning to the 1st bending natural frequency 5) recent developments of digital twins in the DB environment (overview of research projects) 6) development of model and result interfaces
