STRABAG Data Science Hub
What is a Data Science Hub?
Our approach is to use data differently and better. We still make far too little use of this resource, even though huge amounts of data are created every day throughout the Group. In the Data Science Hub we collect analysis data in a standardised way and prepare it for every conceivable utilisation.
Can you be more specific?
I’m talking about analysis data, for example from our machines, from the scheduling of our construction projects or the project calculations.The cranes, the construction machinery, but also the commercial data on the administrative side.
And the goal is to bring this data together and evaluate it?
This is the only way to give data value. We need to put relevant data together. Only when we succeed, can we think about everything else. Data-driven risk analyses based on artificial intelligence, optimisation of work processes – whatever we have in mind, we need data from the construction sites, our machines, the customers, from finance.
In the group, the term ‘data bathtub’ was created some time ago; I prefer to use ‘data hub’ myself, because not everything necessarily has to be stored there.
Often, an interface is enough to connect systems and create added value.
What is meant is an imaginary container into which relevant analysis data and also external data such as weather data flow into – and then everyone can use it for their application …
… something like this, it’s a place in the cloud, whoever brings the data here doesn’t necessarily know what exciting applications others will come up with for it. That can be a very creative process.
That sounds quite abstract. What data-driven applications is the Group working on?
Obvious applications, which are already very well-advanced, are, for example, the utilisation of the cranes on our building construction sites. Automated crane data collection provides parameters that are an important control element for construction sites. Cranes are utilised to varying degrees depending on the construction process. There are always workload peaks that cannot be predicted due to the complexity of construction sites. An automated evaluation of the utilisation can help to recognise these peaks and counteract them. This includes using the resources more economically than before, so that the crane is utilised in all areas but not overloaded.
And if you combine all this with other data, for example, the weather report ….
… then you could optimise construction processes in the face of an approaching low. This is the aim of the digital tact control project, for instance. The weather data is already integrated in order to get an even better view. Through consistent construction data collection in real time, use cases such as digital action management in the event of disruptions as well as cross-project comparability can be implemented in the future.
And the systems also learn?
Right, then an artificial intelligence may suggest to us that we don’t need so many devices or resources on construction site X, because that was already the case in hundreds or thousands of cases in the past.
Why don’t we just do it then?
There are still many hurdles, some barriers are simply based on cultural issues, such as thinking that my data is my business. Other barriers are numerous data silos that still exist. Technologically, many things could be implemented, but we can’t access them yet because access is often not possible or data maintenance is not so advanced that one could easily carry out an analysis. With SID, however, we now have a major driver who has put the topic of data at the top of the Group’s agenda.
Nevertheless, the SID is only the enabler of these topics; the identification of suitable use cases and the willingness to tread these new paths can only work with the help of our colleagues from the individual corporate and central divisions.
Do you have an example?
Currently, for example, we collect and analyse very little data in our own properties. But if we want to have – or should I say have to have – a green, climate-neutral administration in the near future, then we should at least be able to call up such things as electricity consumption precisely and in real time and – in step two – also control them. What is often missing, is the view of the big picture, our view is often too small. It usually remains with the narrow horizon of those involved in their area, their tasks. Collecting and consolidating the data, i.e. the current status, – in this example using IoT sensors – is the necessary first step. Once this step has been taken, it is possible to think about intelligent control of the heating, lighting or other things such as needs-based cleaning. We can then also intelligently evaluate our project calculations, including pattern recognition and anomaly identification, to predict project risks using trained models. This then pays directly towards the two central goals of increased efficiency and sustainability. Only step two brings the actual benefit.
A data-driven risk analysis using artificial intelligence, for example, is only useful if the data basis at least covers the entire group.
Of course, because in order to analyse data using state-of-the-art methods, the amount of training data is crucial. Anyone who believes that the digital tact control only serves to control the tact does not understand the world of data. In the future, it will help us to become better, because we can also evaluate the process data generated there. One question we should always ask ourselves: What could others do with our data?
Or vice versa: How could we use data from other Divisions?
Not only other Divisions, our many subcontractors are also not yet sufficiently considered from a data perspective.
However, it has worked without them so far ...
But we mustn’t rely on that. In the construction industry, we achieve margins of 3–4% when things go well. With only a few flops, we would quickly strain our results. If you use data smartly, you can avoid the kinds of mistakes that generate serious consequences. That’s why data is the basis and central element of our digital strategy.
STRABAG is not the only one that wants to do new business with data. Is a new competition for data now emerging on construction sites?
Of course, others also want to provide and sell data. In future, device manufacturers would prefer to sell us their devices ‘as-a-service.’ This means that we get ready-made analyses, while the raw data remains with the manufacturer. We, on the other hand, naturally want to collect and analyse data ourselves with the devices and are often interested in the raw data. Service providers need our data to train their algorithms. In the long run, we will have to find a solution. The decisive question is: Who will own the data in the future?