How modelling and AI can help manage ageing urban water infrastructure

Managing ageing infrastructure and large networks of water systems are big challenges facing water companies all over the world. In 2021, a study led by a team at United Nations University identified ageing water infrastructure as an emerging global risk. Most of the engineering solutions for water management still in use today, such as dams and sewer networks, were built between the 1930s and the 1970s. These are now posing risks and challenges to infrastructure managers striving to expand services and embrace the digital revolution.

Berlin’s sewer is 9,700 km long, enough to cover a journey from Berlin to Bogotá! Managing such a vast network of ageing infrastructure is becoming increasingly difficult due to an insufficient rehabilitation budget. The Berlin Centre of Competence for Water (KWB) is rising to the challenge and researching a promising smart solution to improve asset management: deterioration models. Key research activities on modelling are being developed to address both the short and long-term operation of sewer networks and offer a better, cost effective solution for asset management.

Deterioration models can be applied to forecast the evolution of the condition of the entire sewer network or of specific groups of sewers with similar characteristics. The models need to be as accurate as possible so that utilities and municipalities can trust their predictions and use them to plan efficient inspection, rehabilitation and investment strategies.

In the past decade, KWB has analysed meticulously the prediction accuracy, the operational benefits and the limitations of deterioration models in different countries of the world, such as Germany, France, Colombia and the United States. Working closely with utilities, this work also addressed key issues such as the improvement of condition assessment from CCTV inspections and the consideration of uncertainties in the decision-making process.

How do deterioration models work?

Inspection data from the sewers is the basis for modelling. In a KWB case study from Berlin, all kinds of data  about the condition of sewers were used: for example, the material of a pipe, its diameter, the age, its conditions etc. In addition, we learned that open data is particularly important in this context: by using publicly available urban data such as soil type, groundwater level or the presence of trees, the accuracy of the predictions can be significantly improved. The results from KWB are very encouraging for the industry. The models can simulate the condition of the entire network with excellent accuracy. It is also interesting to note that machine learning algorithms perform better than statistical models in predicting the condition of individual sewers.

Based on these results, researchers at KWB and Berliner Wasserbetriebe have developed a new prediction tool for the management of Berlin´s infrastructure called SEMAplus. The aim of this project is to investigate the suitability of sewer deterioration models in predicting sewer conditions, and to identify the relevant specifications of sewer deterioration models and input data needed for successful utilisation.

In the first project phase, data from more than 100,000 sewer pipes in Berlin was used to test various statistical and AI-based approaches to modelling for the prediction of sewer deterioration. Using this new tool, the current condition of the network can be determined more precisely, giving an accurate picture of what to expect in the future. Thanks to this, the sewer utility can now carry out predictive maintenance and efficiently renew the infrastructure in a timely manner – keeping investments and expenses as low as possible while maintaining the condition of the infrastructure.

Tools using AI, modelling and statistics like SEMAplus can be applied worldwide, even in cities where there is less available data from sewer utilities. Recent applications of these tools in Sofia, Bulgaria and Bogotá, Colombia, have shown that models can be beneficial even when only a small portion of the network has been inspected. Due to its efficiency, wide scope and potential, SEMAplus was awarded the 2019 Innovation Prize by the Association of Municipal Utilities (VKU), in the category of outstanding innovations by municipal utilities.

The project still has a long way to go – currently the KWB is building a SEMAplus Community with operators and municipalities which further develops and optimises the tools together. A key priority on the horizon to improve asset management across the whole water cycle is the development of predictive maintenance solutions to optimize the rehabilitation of water wells, among other new digital solutions for water management. To stay up to date about the latest innovations in urban water infrastructure and asset management, visit IWA’s Digital Transformation Hub and, a project involving more than 24 partners in Europe, including KWB, who are investigating new digital solutions in the field.