by Ivan Kourza, Head of R&D at Perfectial
IoT might still be in its infancy but it’s already transformed various industries in a meaningful way. For example, Heating, Ventilation, and Air Conditioning manufacturers increasingly embed intelligence into their machines and connect HVAC devices to the cloud. This allows to capture valuable operational data, unlock insights from it, optimize the performance of equipment and reduce, substantially, HVAC energy consumption.
IoT’s total economic impact is expected to reach $11.1 trillion by 2025 (McKinsey Institute)
Perfectial hires hundreds of people; we care deeply about their well-being as the health and comfort of each employee is linked directly to their productivity. We’re also a firm that builds its entire business strategy around increasing efficiency, so updating the components of the systems we rely upon (this includes both software and hardware tools) is something we do regularly.
The hype around IoT in the energy sector has become too loud to ignore in recent years so we’ve decided to conduct our own research on the matter; our R&D department has been tasked with researching the feasibility of Perfectial’s legacy HVAC equipment modernization and developing a prototype of a custom, IoT-enabled HVAC infrastructure.
How IoT Reshapes HVAC
The Internet of Things is about getting physical objects to use the Internet backbone to transmit data about their condition, location, and various other attributes. HVAC providers seeking to leverage the power of IoT are integrating equipment (air-conditioning units, heating radiators) with sensors that monitor air pressure, volatile compounds, CO2 emissions, etc., and then sending that feedback along with machines’ operational data to the cloud (through intelligent gateways) where experts can draw insights from it.
Having complete operational visibility into devices, building managers can spot anomalous activities and detect faults early; seeing a part’s decreasing performance, for instance, they can predict its maintenance needs and fix the equipment before it shuts down, thus preventing costly repairs.
This switch from reactive to a preventive mode of maintenance opens up an enormous savings potential for the manufacturers; it allows to increase the lifespan of HVAC appliances, drive down maintenance costs, and deliver new value to clients.
This is a huge issue in our case; Perfectial has been using ventilation and heating units made by various vendors for years and rather chucking all the equipment away and replacing it with readily connected devices, we’d much rather see if anything could be done to improve the existing hardware.
Hence, our R&D team has been assigned to build a custom, intelligent, connected solution from scratch.
Perfectial’s Custom HVAC Infrastructure
First, we conducted thorough research of the infrastructure components. We prepared and validated the hardware building blocks (sensors, actuators, etc.) and engineered a complete software infrastructure that includes:
- software that goes into HVAC devices
- gateway software
- software at the cloud level that allows managing the equipment
- software that enables remote control and monitoring.
We built custom controllers based on ESP32 development boards, which we picked due to their impressive performance, relatively low cost, and a wide variety of built-in interfaces such as:
Each board has a limited number of I/O interfaces though, so we had to incorporate a few of them per each control unit to accommodate all our sensors and actuators and enhance the solution’s performance.
To truly harness the IoT, it’s crucial to ensure seamless data accumulation, filtering, and sharing. We’ve decided to use the lightweight MQTT protocol, which sits on top of TCP/IP protocol, to establish the two-way data flow.
As for the MCU software, we wrote the code in the Arduino IDE and C programming languages and uploaded the programs to the boards via a USB interface.
The cloud provider we ended up using is Amazon. We picked AWS due to it having a secure gateway, a dedicated AWS IoT communication library for ESP32, and an exhaustive set of databases, storage and computation resources for analytics.
At first, our efforts were focused on maintaining a high quality of air in the office; we’ve integrated our old HVAC equipment with sensors that keep track of:
- PM 2.5
- Air pressure
However, later we started experimenting with various machine learning architectures that could be applied to HVAC controls to teach them how to minimize energy consumption automatically. At the point being, we’re actively building functionality for real-time power usage monitoring (and logging the stats in different granularities – offices, floors, the entire building) as well as some advanced modeling and evaluation tools.
The evaluation and modeling software will help us see clearly the current patterns of power consumption, the factors that affect it the most and the extent of their impact. Having this data, we’ll be able to quickly identify key energy components of our building, apply necessary adjustments and devise an overall strategy to lower HVAC expenses.
Afterward, our plan is to incorporate third-party data sources, such as weather feeds, into the infrastructure to push the connected machines into a proactive mode and have them make decisions on their own, such as raising or lowering air intake, to enhance the facility’s energy efficiency and bring greater degrees of comfort to the tenants.
IoT enables us to move to predictive maintenance which can increase the longevity of our equipment and is substantially less labor and cost-intensive than traditional maintenance. As opposed to conventional BMS systems, IoT technologies can be installed in any type of building; they are easily scalable to other building assets (security, lighting, etc.)
Legacy HVAC equipment tends to have a long life-span and, for some companies, it might not be economically feasible to upgrade every embedded HVAC device. The key to solving this, we’ve found, is establishing a flawless integration of the equipment with sensors and using intelligent gateways (that must have enough processing power to enable end-to-end- analytics) to connect the machines to the cloud directly, circumvent a building’s IT networks.
Generally, we recommend taking a phased approach to IoT implementation; this allows addressing efficiently any unforeseen requirements that might pop up and ironing out any process gaps.