How data models, algorithms and sensor systems are vital assets

It starts with master data

“It starts with master data, which you could say is the most important kind of data. Master data is the static data about the property and the customer: what sort of building it is, what assets are in the building, its maintenance history and maintenance plan, and other factual information like that,” says Maikel Nabuurs. “It can be a challenge to obtain all of this data in full and make sure it is correct,” adds Timo van Drenth, contract manager at SPIE Nederland. He knows from experience that this information is often highly segmented and sometimes difficult to find. “It is not unusual for us to have to go to quite a lot of effort to get this information, especially for older buildings.”

Maikel Nabuurs: “In general, at the start of a contract, we carry out a baseline measurement where we collect and check the master data. We need to know whether the assumptions we intend to use in our work are correct. That is why that initial data collection phase is so important. It is also a very time-consuming and expensive process. But if the data isn't right, that can lead to costly failures. Increasingly, the value of a real estate property is determined by the demonstrability and accuracy of master data.”

New building as a baseline measurement

With a new building, it seems like it would be a lot easier. After all, the design and implementation of building processes are increasingly—and sometimes fully—digitalised. And yet this still poses challenges. Timo van Drenth: “With a new building, you immediately have a great baseline measurement, or so you would think. In our experience, however, there is still too little attention being paid to the management and maintenance phase when designing and constructing a building and all of its installations. This is mainly because designers and builders generally do not have a long-term interest in the building once it has been delivered. Also, when clients take ownership of the property, they are not sufficiently aware of how to organise the data properly so that it can be used as it is in the post-delivery phase.”

“This is sometimes possible,” says Maikel Nabuurs, “but mainly in Design, Build, Maintain (Operate) – DBM(O) –  contracts, as we found with the Knoopkazerne. During a DBM contract, there is a lot more integrated planning involved at the start of the construction process and, ideally, the maintenance and management people will also be sat around the table with the designers. In this way, we can ensure that the correct data, which we need after the delivery, is recorded properly right from the design phase and during implementation.”

Although building information modelling (BIM) often comes with high expectations, it is not necessarily useful for those carrying out the maintenance and management, according to Maikel Nabuurs and Timo van Drenth. “In our experience, BIM is still too often mainly for the designers and builders. Although it is starting to grow, it is often the case that only part of it can be used after delivery.”

Working with data scientists

Once the master data is available, it is simply a matter of adding real-time data to your systems, according to the specialists from SPIE. This data originates from many different systems. Often, these are familiar systems such as building management systems, access systems and other building-related installations, all of which generate data to an increasing extent. “It is one thing to collect data, but an entirely different thing to actually use that data in a meaningful way,” says Maikel Nabuurs. “That is why we are increasingly working with data analysts and data scientists. Ultimately, they will help us use all this data to take our business models to the next level. At the moment, they are easily spending sixty percent of their time cleaning up this data and collating it in a single model. We need to bring that percentage down as soon as possible. One important way we can do this is by making industry-wide agreements on standards—we and a number of other major players have opted for Project Haystack—and on how to make data available. At the end of the day, this is important for everyone.”

Needs and priorities

For Maikel Nabuurs, “it is certainly not the case that we immediately collect as much data as possible for every building or for every contract. First, we look at the client's needs and priorities. That might be sustainability, which means that we will prioritise energy monitoring and metering. Currently, however, comfort and indoor climate are increasingly becoming priorities, especially since the coronavirus pandemic has taught us the importance of sufficient ventilation and fresh air. We then start to collect data from existing systems, the low-hanging fruit. But eventually we may decide to install extra sensors, for example so that we can monitor CO2 everywhere. Sensor networks of this kind are also becoming better value.”

“At TU Delft, the Technical University, we conducted a pilot for monitoring Air Handling Units (AHUs) at the request of the university. We tested new filters in four more or less identical AHUs of the same capacity. These filters should improve the quality of indoor air and lead to lower energy consumption. The sensors that we installed inside the filters provide us with a huge amount of information, which allows us to further improve the way we carry out predictive maintenance', concludes Timo van Drenth.

Essential assets

“Predictive maintenance is one of the main reasons why it is important to focus on building the most complete data model possible. As mentioned, this requires specialists, such as data scientists, but also reliable hardware and secure networks. Just like a boiler or an air handling unit, we will have to consider sensor networks and algorithms as essential assets. Hardware and software need maintenance and calibration. A few years ago, the quality of sensors was still sometimes questionable, but that is rapidly improving. In addition to the quality of the sensors, their placement requires attention. Also make sure that the sensors are not moved without being noticing. The raw data is always owned by the property owner”, say Nabuurs and Van Drenth. The larger property owners are aware of that. And we respect that, too, because those data have to be available to another maintenance party if they take over a contract. We also want to have that data available when we start working on a property. However, as soon as we add intelligence to that data, by using it in our systems and models via algorithms, then those are our assets. But the raw data is always kept, so that it can be transferred when the owner wishes.

Work with universal standards

We almost always ensure that the data we process, via our PULSE Core platform, is accessible to the owner. Also for other parties who are involved in the object in a maintenance or management role. In any case, the client can see what we are doing and what we base our conclusions on; it is good to be as transparent as possible in this regard. What does make our work more difficult sometimes, Van Drenth agrees with his colleague, is the installation of closed systems. Unfortunately, in our practice we still sometimes come across suppliers who install smart systems, but the data is not available or only in such a format that we can't use it. This is something that clients, in particular, should be aware of. In any case, they could set conditions on how and in which format a supplier should offer the data from its system. We will also play an advisory role in this, continues Nabuurs. We obviously have nothing against someone who is a specialist in a particular technical field supplying and installing a smart system. But you have to make sure that the data can be integrated into models which work with universal standards.

On the road to digital twins

Ultimately, Nabuurs and Van Drenth foresee that the models they are already building will lead to so-called digital twins. At SPIE, certainly since the integration of Worksphere within the company, they have already made great strides towards becoming a digital twin specialist with data, machine learning, and artificial intelligence. In cooperation with TU Delft and others parterners, we are already using machine learning and AI to a certain extent in the B4B (brains for buildings) research project to make our maintenance and management more efficient. We can also predict energy consumption. Furthermore, we use these techniques for the soft side of building management, i.e. what does the user think about his surroundings and indoor climate. We can respond to this better with these techniques, says Van Drenth. The most important thing is that, in addition to the hardware and the smart software systems, we attract the right people. In addition to data scientists, these include maintenance consultants, building physics specialists, facility and logistics specialists. These people are hard to find. At job fairs, we stand among Philips, ASML and Microsoft, who are looking for more or less the same specialists. Our advantage is that these specialists are not one in a hundred or one in a thousand in the same job. Those specialists can really make a difference in our profession. They are working on real issues; improving people's health, tending to reduce fossil energy consumption to zero. Fortunately, such challenges appear to attract many people, concludes Nabuurs.