As an engineering company with a 100-year history, we understand the importance of ageing infrastructures, such as water distribution networks, sewer systems and drinking water wells. These infrastructures are valuable public assets with high monetary worth, ranging from millions to billions of euros, depending on the size of the cities they serve. However, as these infrastructures lay buried underground over decades, they are usually beyond public perception. Nonetheless, we need to recognise that making wrong decisions in the management and rehabilitation of these infrastructures can pose significant risks to public services and the environment.

Failures of water distribution pipes, for example, can lead to interruptions in water supply, which become particularly critical when hospitals or large industries are affected. In addition, cracks in wastewater pipes can lead to contamination of soil and groundwater and, in the worst-case scenario, drinking water resources. In addition, ageing infrastructures poses financial risks, as operating costs for maintenance and repairs increase over time. This also holds true for drinking water wells, where ageing processes lead to higher energy demand and pumping costs. Finally, we must acknowledge that investments in infrastructure that are suspended today will fall on the next generation and may jeopardise future public welfare.

As we transition into digital engineers, we have learned to embrace modelling tools to support operational and strategic asset management and minimise risks. These tools have become increasingly sophisticated over the past two decades and are now state of the art. Historically, we have relied on statistical approaches to simulate the probability of asset failure or performance indicators for different groups of assets over time. Recently, machine learning algorithms have appeared on the scene. These algorithms aim to predict the condition of specific assets but require comparably extensive datasets for accurate model setup.

We have witnessed significant improvements in data availability over the years. Most water utilities now use geographical information systems which can provide asset information in terms of the year of construction and the material used, among others. Failure statistics for water distribution networks are typically available with recorded times of ten years or more. For wastewater infrastructures, inspection data is usually available for large portions of the network, providing information on the number and severity of defects. Environmental data, such as soil types or groundwater levels, which play a crucial role in ageing processes, is also openly accessible in many states and countries.

Nonetheless, there are still some major challenges ahead, especially with regards to data quality. For example, recorded failures are often not properly linked to the associated asset, making the data unusable for model calibration. In the case of wastewater infrastructures, certain asset groups or pipe ages are frequently underrepresented in inspection data. Consequently, we need to rely on assumptions when building models, which can lead to incorrect decisions by asset owners. Lastly, it has been common practice to overwrite data of defective pipes that previously required replacement in asset databases. As a result, valuable information about assets that have reached the end of their lifespan is often missing or incomplete.

These examples highlight the long-standing underestimation of the value of data. However, we are now witnessing significant progress in recognising and prioritising data acquisition and processing efforts. Many utilities have already taken proactive steps to address the challenges mentioned above. Furthermore, confidence in data-driven modelling techniques, as a supplement to engineering experience, has largely increased. In a nutshell, the foundations for making informed decisions and ensuring a reliable public infrastructure – a priority action of the United Nations – have been established and are ready to drive positive change.