SEMA-Berlin enters the second phase: Two selected model approaches, which can describe the state of the sewage system with particularly high accuracy, are further developed and put into practice
Recent studies about wastewater infrastructure in Germany show that current investments for sewer rehabilitation are not sufficient to tackle the aging and deterioration of the networks. In order to be able to forecast the future development of the structural condition and to support sewer rehabilitation planning, various deterioration models were tested on Berlin data considering their prediction quality. These models can simulate the condition of uninspected pipes and predict the future development of the network condition. They consider the results of more than a hundred thousand CCTV inspections as well as data on individual sewer properties and the city’s surrounding conditions.
During the project’s second phase two selected model approaches, which can describe the state of the sewage system with particularly high accuracy, will be further developed and put into practice.
The models focus on two application areas:
Supporting the planning of sewer inspection in line with the actual demands and
Supporting the long-term sewer rehabilitation and investment strategies.
As to item 1, priority pipes and areas are identified and visualised by means of machine learning models. The results will be integrated into the current inspection planning routines. As to item 2, long-term predictions of the sewer condition development are prepared by means of a statistical deterioration model taking into account various investment scenarios. In this context, possible synergies with other infrastructures, e.g. the drinking water or district heating network, are also examined. In a next step, uncertainties in data and models are quantified. The further development of the models and the consideration of additional inspection data will lead to increased prediction accuracy.
Selected publications produced in the SEMA-Berlin-2 project are available here.