In the evolving landscape of technology, Artificial Intelligence (AI) has emerged as an important feature that will affect all industries. Encompassing a vast array of cutting-edge systems, AI aims to replicate and even surpass human intelligence and behaviour.

Among the many applications within the realm of AI, Machine Learning stands out as a key, developing capability. By enabling machines to learn from data without explicit programming, machine learning has revolutionised the way systems operate. It is a technique that empowers systems to automatically improve their performance and make accurate predictions or decisions based on patterns and examples found in the data they are exposed to.

Unlike traditional software development, where explicit instructions are hand-coded for specific tasks, Machine Learning trains machines by exposing them to vast amounts of data and utilising algorithms that equip them with the capability to learn and execute the task effectively. Through this approach, machines can autonomously adjust their performance and adapt to various scenarii without the need for manual programming.

Machine Learning can play a crucial role in HVAC applications, exerting a profound impact on the performance of systems while offering significant opportunities for cost savings. Leveraging Machine Learning capabilities can reduce building construction costs by up to USD 3 per square foot, and operating costs by approximately 40 US cents per square foot.

Kevin Laidler

Another significant aspect of Machine Learning in HVAC applications is its ability to enhance occupant comfort. While various solutions exist to maximise occupant comfort in buildings, these solutions typically incur additional costs. The ultimate goal is to achieve 100% occupant comfort without incurring expenses for idle assets. This means that building developers and owners seek to avoid the financial burden of additional equipment that remains underutilised for extended periods. Machine Learning offers a highly effective and efficient approach to address this challenge. It enables alignment of the HVAC system with occupants’ comfort requirements while minimising unnecessary costs associated with underutilised assets, at a level of precision that a human operator would not have the patience for. Essentially, Machine Learning is a highly effective way to improve occupant comfort while reducing costs. However, the question remains: How does it do this?

As mechanical systems age, their operating performance tends to diminish compared to when they were initially commissioned. This gradual decline, known as performance drift, occurs as a result of minor changes in equipment behaviour. These changes can be attributed to issues like fouling or insufficient maintenance of specific components, such as clogged strainers or dirty filters. Consequently, the overall performance of the system gradually declines. On average, one can anticipate a 25% increase in power consumption for the same level of output.

However, the integration of Machine Learning goes beyond simply maintaining HVAC performance. It has, in fact, been proven to enhance system operation and efficiency over time, achieving an impressive increment of eight per cent. This is more than any previous approaches have been able to unlock before. This is achieved through learning and adaptation to various combinations and configurations to ultimately achieve performance and efficiency levels that surpass those during the initial commissioning phase.

For this process of enhancing HVAC system efficiency, Machine Learning capabilities manifest in the form of digital twin technology. This is where data collected from Internet of Things (IoT) devices is processed and utilised to create an exact digital replica of the live model — a digital twin. The sophisticated technology allows for the generation of insights that enhance operations, improve efficiency and identify potential issues in advance. It even enables testing and comparison of optimisation strategies before applying any changes to the physical system, thereby minimising the risks.

A digital twin continuously captures new data and updates all relevant variables over time, allowing for real-time monitoring of changes in the digital replica. Leveraging this dynamic information, a predictive model can be developed using digital twins to gain insights into the potential outcomes of operating the system differently. By simulating alternative scenarii, organisations can proactively assess the impact of different strategies and make informed decisions based on the predictive capabilities offered by digital twin technology. To put it simply, how the HVAC system automatically adjusts performance is similar to the way a Sat NAV system recalculates a new route when a change of direction is observed. This ongoing, persistent optimisation strategy retains an impressive 25% of HVAC efficiency that is normally lost due to performance drift.

In addition, Machine Learning brings new opportunities for condition-based maintenance. Using automatic notifications, any equipment failures, service requirements and performance deviations can be detected well before performance drift occurs. This advancement opens up opportunities to improve service protocols, transitioning from a scheduled maintenance approach to condition-based maintenance. By proactively addressing maintenance needs, based on the actual condition of the equipment, operators can optimise efficiency and minimise downtime. In fact, it is possible to generate life-cycle cost savings of between 30% and 40% through this method of condition-based maintenance.

The integration of Machine Learning technology into HVAC systems offers great potential for building owners, operators and occupants alike. AI-optimised pump systems play a crucial role in lowering energy consumption, which leads to substantial reductions in operating costs and, ultimately, results in significant long-term savings for building owners and operators. These systems also necessitate less maintenance, effectively minimising the time and resources required for upkeep. And most importantly, the lower risk of failure associated with optimised HVAC systems ensures a more reliable and uninterrupted operation, providing enhanced comfort for building occupants.

The writer is Sales Director – Middle East & Africa, Armstrong Fluid Technology. He may be contacted at klaidler@armstrongfluidtechnology.com