Modern building automation increasingly relies on artificial intelligence, user-behaviour analysis and cloud services. These solutions work very well where the goal is convenience, optimising utility consumption or providing information to the building owner.
But this raises a question: is the same operating philosophy right for systems whose primary task is to protect property from damage?
This is not a dispute between modernity and tradition. It is a clash of two entirely different philosophies and ways of understanding safety.
01Scenario: a weekend, an empty house
The homeowner goes away for the weekend. The building is empty. At some point a seemingly insignificant component of the water installation fails — a flexible hose supplying a tap or a household appliance bursts.
At a mains pressure of around 4 bar, such a leak can reach a flow rate of several dozen litres per minute. From the point of view of physics, the event is unambiguous:
- water is leaving the installation,
- the flow continues without interruption,
- every minute increases the extent of the damage.
The problem is that, to some monitoring systems, such a flow may look like normal water use by the occupants. After all, similar consumption occurs during:
- bathing,
- running the washing machine,
- washing dishes,
- watering the garden.
The system therefore has to answer a fundamental question: is what it is observing normal use, or the start of a disaster?
02The dilemma of behaviour-analysis systems
Modern platforms that use behavioural analysis try to interpret events. Their job is not just to measure flow — they try to understand the context. They compare current data with usage history, look for anomalies, reduce the number of false alarms and try not to disrupt the occupants’ comfort.
From a convenience point of view, this makes sense. Nobody wants the system to shut off the water during a normal bath just because consumption was higher.
However, a problem arises that is typical of all solutions based on data interpretation. Every algorithm must first weigh up different scenarios and ask itself: is this still normal consumption, or already a failure? Meanwhile, the damage keeps growing.
For the installation — more litres of water on the floor.
03Comfort and safety are not always the same thing
In modern systems, the priority is very often to reduce the number of false alarms. That is a logical approach — false alarms annoy users. In safety engineering, however, a different principle applies:
An undetected failure is a loss.
These are two different categories of risk. The more a system tries to interpret a situation, the greater the chance of misclassifying an event. This applies to every statistical model — regardless of the manufacturer, brand or technology used.
04The philosophy of classic automation
For decades, industrial automation has developed according to a different logic. There, the question is not “what is the user probably doing?”, but “what is physically happening in the installation?”
If a parameter exceeds a set threshold, a response follows:
- without interpretation,
- without waiting,
- without consulting a server,
- without analysing user behaviour.
This approach is called deterministic: if condition X is met, action Y is executed automatically. Every time. In exactly the same way.
05A hybrid system: safety based on physical facts
The philosophy behind H-corePLC® class controllers is based on local decision-making. There may be internet. There may be an app. There may be a cloud. But none of these is needed to make the decision that protects the building.
All the critical logic sits locally — where the problem arises. Exactly where you find:
Records every litre flowing through the main water connection.
Physically shuts off the water supply to the building.
Makes decisions locally, in a fixed program cycle.
React to the physical presence of water.
06Four lines of defence
The most effective failure is the one that never happens. If the system is integrated with the building’s alarm, locking up the property can automatically close the main water valve — while the owner is away, the building is physically isolated from the water mains.
It does not matter where the user is, whether the internet is working or whether the phone has signal. The installation remains protected regardless of these factors.
If the valve remains open, protection moves to the next level. The controller monitors the flow in real time. It does not analyse the occupants’ behaviour — it analyses the water.
If the flow characteristics match failure parameters, the valve closes immediately once the set threshold is reached. Not after sending a notification. Not after getting the user’s consent. Not after consulting a server. Only once the logical condition is met.
Not every leak is sudden — some failures develop slowly. That is why the system also monitors the total volume of water flowing through and automatically shuts off the supply once a defined value is exceeded.
This mechanism does not replace flow monitoring. It is an independent second-level safeguard.
The last layer of protection is water-presence sensors. Their advantage is obvious: they do not analyse trends, forecast or predict. They react to a fact — if water physically appears at the measuring point, the emergency procedure is triggered.
07The real difference is not the technology
It would be wrong to claim that cloud solutions are bad. They are not. It would be equally wrong to conclude that artificial intelligence has no place in buildings — it does, and its role will keep growing. What matters, however, is distinguishing between two areas:
- optimising utility consumption,
- data analysis,
- reporting,
- forecasting,
- automating everyday tasks.
- deterministic logic,
- local decision-making,
- minimal response time,
- independence from external infrastructure.
08Conclusion
In systems responsible for protecting property, the most important parameter is not intelligence in the sense of the ability to interpret events. What matters most is the ability to respond immediately to a physical phenomenon.
In industrial automation, it has long been a rule that safety functions must work even when the internet, the app, the server, the user’s phone or any other element of the supporting infrastructure fails.
It is the system that makes the right decision based on the installation’s physical parameters — and carries it out immediately.
