Using data effectively for the sound management of concealed assets

The goal of the Midas (Multiple Data Sources) project was to improve data for use in pipe condition models. The research focus was on the interplay between individual data components and their influence on the margin of uncertainty in predictions. Three technology suppliers were closely involved in the project. HDM Pipelines – an operator and asset manager of underground infrastructure – played an important role in connecting research, practice and services. Acquaint – a company that inspects underground pipes – supplied and prepared non-public data and services. And Spatial Insight was responsible for integrating data about pipe quality, and quality control for those data. KWR managed the research component of the project.

“The main benefit is that this study shows how important it is to collect data properly and reliably.”

Data science

For Spatial Insight – a company that helps drinking water companies to manage underground assets optimally – the goal of the project provided an interesting way to approach this area, explains Den Besten. “Data science allows us to use knowledge about malfunctions in the past to anticipate malfunctions in the future. Midas had a slightly different focus. The idea wasn’t to predict malfunctions but to determine the quality of pipes without having to dig them up or inspect them. As well as drinking water companies, water authorities are also increasingly interested in data science. If we could predict pipe quality for the whole of the Netherlands, and then compare that information with actual malfunctions, we could superimpose these two categories of data to identify overlaps. That would deliver enormous added value for asset management.”

Complex puzzle

Den Besten emphasises how important it is to organise data integration – tying different data sources together – in the right way. “You can try to conduct analyses, but if the quality of the underlying data isn’t good enough, that won’t work.” But even if data organisation is good, predicting what happens underground is a complex puzzle, Den Besten explains. He blames this on the inadequate documentation of the biography of the mains. The construction work in the 1960s – with intense time pressures because of the acute housing shortage – was an excellent achievement. But there are no data about changes in the condition of the pipes after construction was completed. So there are extensive gaps in our knowledge about the factors that affect the life cycle of pipes. The Midas project looked at whether the data that are available can be used to predict the condition of pipes.

Limited correlation

Den Besten explains. “If you know how parameters are related to measurements of the pipe condition at one location, you can use those parameters at another location to predict the condition of pipes without making actual measurements. For example, if a pipe is very thin somewhere, and you also find acidic groundwater at that location, you can surmise that there is a connection. So you will expect wall thickness to be affected in other places with acidic groundwater. But things aren’t that simple, unfortunately. This study has shown that the correlation between known parameters such as groundwater acidity, and target variables such as wall thickness, is actually limited. This may be due to differences in pipe quality, for example. However, as we have already seen, the data aren’t available. We simply don’t know what we don’t know. That’s quite often a problem.’

“The dashboard is a support tool that helps us to identify the right focus for the time ahead.”

Dashboard

Nevertheless, enough data were collected to allow Midas to make progress on the desired integration. A concrete result emerged: a dashboard developed by KWR for inputting variables in order to determine their effect on the condition and performance of the pipe. End user George Galama of the Zuiderzeeland water authority is enthusiastic about this practical tool. “Since 2016, we have been taking an active approach to designing the management of our pipelines. We got on board with this project with the aim of being involved with new developments in this field. The dashboard is a support tool that helps us to identify the right focus for the time ahead. We have acquired a picture of which data we collect are accurate enough, and where we still have to do our homework. Moreover, we can use both the dashboard and the report – with a specific chapter for end users – to initiate discussions with management. It is often difficult to get people thinking about subsurface issues so that funds are earmarked for this purpose. Obviously, the dashboard isn’t yet an off-the-shelf end product that tells us which investments are needed, or where. But it’s certainly a useful addition to our toolbox.”

Balancing act

Den Besten says that the dashboard is primarily a useful instrument that shows the extent to which the variation in source data affects the final outcome. He describes it as an ‘awareness instrument’ that teaches end users about how data uncertainty in successive stages can have a knock-on effect. “It makes you realise how important it is to collect high-quality data. Even though it remains to be seen whether we will achieve reliable predictions of pipe condition with data integration. As I said, we have found that it is very difficult to make a good estimate based on data. Here at Spatial Insight, because using data about pipe condition to predict leaks is so difficult, we primarily base our calculations on historical leaks. These data are recorded very well by the Dutch drinking water companies. Sometimes, they opt for malfunction-dependent maintenance but that is a very delicate issue in the public arena, and rightly so. If, on the other hand, you opt for preventive replacement, you have to accept the need to invest money and resources to ensure that all the pipes are safe. That trade-off between economic value and public perception can be quite a balancing act. In general, it comes down to the question of the best way to tackle the challenge of ageing assets.”

“ … we were given enough opportunities to describe where we had questions, and then steps were taken accordingly.”

Theory to practice

Galama of the Zuiderzeeland water authority is very happy with the alliance with Spatial Insight, Acquaint, HDM Pipelines and KWR. Alongside the drinking water companies, he was involved as an end user in the work that has been done in the meantime. “The study was very theoretical,” he says. “But the switch to practice was done well. We were given enough opportunities to describe where we had questions, and then steps were taken accordingly.” When asked how to implement the next step in practice, Den Besten’s advice is clear. “Establish a sound and structured approach to asset data. And during that process, don’t forget about the basic layer of information, with data about the diameter, material and year of construction of pipes, because these data are precisely the data that are extremely valuable.”

TKI as a valuable environment

From the technology suppliers’ point of view, Den Besten agrees that Midas has provided enough ammunition for them to continue working on deploying pipe condition models to interpret locality data and the inspections made by their clients in the years ahead. He does, however, have a reservation to make. “The project is a good place to start in terms of being aware that we don’t know everything. And I have my doubts about whether it is possible to acquire the information we don’t have at the moment. It is precisely for this type of research that TKI is a valuable environment. It gives you the room to identify gaps in knowledge. These studies can sometime be seen as failures. But this kind of research is actually useful in terms of deciding where to go next. That’s something we can be proud of.”

The MIDAS project involved the following partners: Acquaint, Brabant Water, HDM Pipelines, KWR, Spatial Insight, and the Limburg and Zuiderzeeland water authorities.