The famous artificial intelligence that has been served to us in every possible way since the release of ChatGPT in November 2022. I had to bring up this topic sooner or later, applied to the energy management of tertiary & industrial buildings.

When even your banker starts saying, "Ah! You do this thanks to AI?" it means that the marketing hype is at its peak right now.

Few people really know what AI is, and even fewer are able to explain its role in Building Management Systems (BMS). Nevertheless, we are seeing more and more AI innovations dedicated to energy management. It’s not just a bunch of carpet salesmen, far from it! The purpose of this article is to explain AI's contribution to building management and present some of the solutions currently available on the market.

How Does Energy Regulation Work Today?

To start, let’s look at how the current automation systems, sensors/actuators in energy management solutions (BMS), work.

Many Building Management Systems (BMS) primarily use PID regulation to control various equipment such as boilers or air handling units. PID regulation is chosen for its ability to provide precise and stable control, which is crucial for managing heating, ventilation, and air conditioning (HVAC) systems.

A bit of history: Elmer Ambrose Sperry (1860 – 1930), an American inventor and industrialist, started working in 1911 on a system to make ships navigate more autonomously. To design his gyrocompasses (gyroscopic stabilizers), he observed the work that a helmsman does intuitively to compensate and anticipate deviations and errors to keep the course steady.

Later, around 1922, mathematician and engineer Nicolas Minorsky (1885-1970) proposed a mathematical formula that would form the basis for PID controllers. The PID logic would be used in automatic pilot systems for the US Navy’s ships as early as the 1930s.

The term "regulation" is used when controlling unwanted variations and maintaining a stable value, such as temperature, pressure, flow, or humidity. This process involves continuously measuring the system via probes or sensors. The collected data is then sent to a controller that compares these measurements to a target value or setpoint. According to its program, the controller sends commands to different actuator devices such as valves, dampers, or motors to adjust and stabilize the system against the detected disturbances.

Let’s continue by explaining in detail how a PID regulation works ⤵️

PID Regulation Example: Boiler Control with an Automaton

Context: Imagine a boiler used to heat a building. The boiler must maintain the water temperature at a defined level to ensure effective and comfortable heating of the indoor spaces.

PID Regulation Functioning:

• Proportional (P): If the water temperature is lower than the setpoint (the desired temperature), the regulation increases the heat produced by the boiler. If the gap between the current temperature and the setpoint is large, the increase in heat is more significant. This ensures rapid heating of the water.
• Integral (I): This part of the regulation deals with small persistent deviations between the current temperature and the setpoint that haven’t been corrected by the proportional part. It finely adjusts the heat to ensure these small deviations are eliminated over time, preventing the boiler from "stagnating" below the desired temperature.
• Derivative (D): It responds to rapid changes in temperature. For example, if the water temperature rises quickly and is at risk of exceeding the setpoint, the derivative control can reduce the heat production to avoid surpassing the target temperature.

Application with an Automaton: Let’s take an automaton like the PXC4 from Siemens, which can be configured to perform PID regulation on a boiler. The automaton continuously measures the water temperature and adjusts the combustion in the boiler to reach and maintain the desired temperature setpoint.

• Water Law (Heating Curve): The automaton adjusts the water temperature based on the outdoor temperature. For example, the colder it is outside, the more the boiler needs to heat the water to achieve the desired indoor temperature. This relationship is often represented by a curve, where the vertical axis represents the water temperature and the horizontal axis represents the outdoor temperature. The automaton uses this curve to determine the ideal water temperature based on outdoor conditions.

Think of PID regulation like a car’s cruise control, which constantly adjusts the speed to stay within the set limit while adapting to the inclines or declines of the road to maintain that speed. Similarly, the automaton adjusts the boiler’s heat output so that the water temperature stays constant and suited to the building's needs, no matter the outside conditions.

This regulation allows for significant energy savings and works perfectly today. However, it acts "in response" to changes in climatic parameters; it doesn’t anticipate or learn. This is precisely where AI brings a significant innovation.

The Role of AI in Energy Management (BMS)

First, let’s distinguish artificial intelligence (AI) from simple algorithms. Many companies misuse the term AI to sell solutions that do not actually work with real AI.

The difference between artificial intelligence (AI) and sophisticated algorithms (like those used in PID regulation) but not qualified as AI might seem blurry, but there are clear distinctions in terms of design, features, and applications. Here are a few points to clarify these differences:

1) Artificial Intelligence (AI)

• Adaptability: AI can learn from building behaviors, occupant habits, and external environmental conditions. This allows it to predict future needs for heating, cooling, or ventilation and make adjustments even before the changes are needed. For example, if AI predicts that a meeting room will be used, it can start adjusting the temperature in advance to reach the desired comfort level as soon as the meeting starts, rather than reacting once temperature changes are already perceived.
• Complexity: AI can effectively handle the complexity and variability inherent in large buildings or campuses with multiple buildings serving diverse purposes. It can simultaneously optimize many interconnected systems (heating, cooling, lighting, ventilation) more effectively than PID controls, which may struggle to balance conflicting objectives without human intervention.

Example: An AI system could predict a building's energy demand by analyzing weather data, occupant usage patterns, and other variables to optimize energy use.

2) PID Regulation Algorithms

• Stability: PID algorithms are designed to maintain a specific variable (like temperature or pressure) at a constant target value. They do not "learn" new information but follow predefined rules to minimize the gap between the current measurement and the setpoint.
• Simplicity: PID uses a simple mathematical method based on three terms (proportional, integral, and derivative) to adjust control. They are less flexible than AI and are primarily used where a stable and predictable response is needed.

Example: A PID system might control a boiler to maintain a constant temperature by simply adjusting the response based on the gap between the measured temperature and the setpoint.

In summary, AI is adaptive and capable of processing complex scenarios by learning from the environment, while PID algorithms offer a stable and predictable response based on fixed rules to maintain a variable at its target value. Let’s now discover the different AI solutions available on the market and how they function.