Meeting Air-Quality Legislation without Increased HVAC Use Demands More Efficient IoT CO2 Sensors

­Legislation on air quality has recently been added to the statute books of multiple states including California and Massachusetts, and is being considered in several others, as well as at the federal level. The legislation sets a maximum in-building CO₂ concentration and requires air to be replaced regularly, but this adds to energy requirements for heating.

IoT sensors have the power to better monitor air quality on a room-by-room basis and reduce the number of replacement cycles as well as the volume of air in each refresh, but traditional sensors have been too power intensive to enable these to be rolled out in battery-powered systems. Here GSS’s Neill Ricketts examines a new sensor design that will reduce the legislation’s unintended effect on net-zero commitments.

While not directly toxic, elevated levels of CO₂ can be a strong correlative indicator of airborne pollutants such as volatile organic compounds (VOCs) and other bio-effluents, which are emitted or brought into a space by its occupants. And even though they are completely natural, CO₂concentrations also correlate with the risk of COVID-19 infections, which is leading / has led to legislation being enacted in multiple US states (and countries across the globe) to stipulate maximum CO₂ limits in buildings.

In the US, California is among the most-proactive states, with several acts of legislation introduced to limit indoor pollutants and mandating the use of CO₂ monitoring in schools. Similar laws exist in Massachusetts and New York as well as multiple other states, with federal laws being debated. Indeed, the Department for Labor’s Occupational Safety and Health Administration (OSHA)has already established a permissible exposure limit (PEL) for CO₂ of 5,000 parts per million (ppm) averaged over an 8-hour workday.

Elsewhere, laws dictating indoor air quality and limiting the concentration of in-building CO₂ have been enacted by legislatures from the EU, Japan, Canada, Australia, and China, with proposals underway in multiple others including the UK and India.

This legislation, though vital, will significantly affect the environmental impact of these buildings through increased energy usage, and (as a result) significantly increase the cost of running schools, hospitals, housing facilities…

The reason being, all fresh air will need to be heated / cooled by HVAC systems (heating, ventilation, and air conditioning) and these are already responsible for approximately 40% of the total energy used by a building (Source: Australian department of Energy). It is somewhat ironic that legislation intended to significantly improve air quality will at the same increase CO₂ emissions elsewhere in the process.

IoT sensors

This unintended consequence doesn’t have to exist and IoT sensors are an obvious answer to the problem.

Not all the air in a building needs to be refreshed each time, and IoT monitoring would enable the identification of localised peaks in CO₂ concentrations, with only these areas being refreshed and heated or cooled. Such an approach would significantly reduce the required heating/cooling. But… sadly not in this case.

It is true that IoT sensors are already used to monitor gas levels. And CO₂ sensors exist. These measure concentrations of CO₂, which absorbs the mid-IR 4.25 µm wavelength. Traditionally, these sensors worked by illuminating gas via filtered incandescent sources of IR light and measuring absorption via a light detector. The power required for these sensors was significant and prevented their use in anything other than mains powered IoT equipment.

More recently, solid-state NDIR (non-dispersive infrared-based) sensors have been developed, but while significantly more power efficient than their incandescent equivalents, the power is still too high for use in battery-operated systems that last for long enough to be worth it.

And while modern buildings may have the power-supply networks to roll out such systems in every room, in older buildings such as hospitals and schools this may require battery-powered IoT systems, transmitting data wirelessly via LoRaWAN, Bluetooth Low Energy (BLE) or a similar protocol.

This means even using solid-state sensors would require very regular battery changes, with the time/cost of regular checking and replacing batteries making this prohibitively expensive.

Solid state NDIR sensors

Sensors are moving away from incandescent IR emitters, with solid-state LEDs and emitters now being used.

LEDs are the most used light source and, like elsewhere, offer the lowest power consumption and longest mean time to failure. But even using LEDs, the power consumption is typically in the region of 50 to 150 mW average, with a peak of 300-400 mW.

While this doesn’t technically put it beyond being used as part of an IoT system running from AA batteries – or similar, it would require very regular checks and replacements, which adds to the workload (and cost) of the building’s maintenance staff.

A more power efficient design

Significant energy savings can, however, be made.

Click image to enlarge

Figure 2: The GSS CoziR-Blink uses an alternative topology to significantly reduce power consumption. When the sensor is switched on, a measurement is automatically initiated. Data can be read out once the READY pin is pulsed high, approximately 3.3s after application of power. The sensor can then be subsequently switched off, saving power

If we look at the biggest power drain, the light source, we can see that reading frequency matters. For this application, the CO₂ sensors do not need to run several times per second. Indeed, as we said at the start, carbon dioxide is not toxic so, unlike a carbon monoxide or smoke detector, sensors can be set run at far less frequent intervals.

If we instead adjust the duty cycle and have it run at a much lower frequency, one that still comfortably meets the needs of the application, to ensure that it only runs when required it is possible to bring this down significantly. For an application like in-building CO₂ level monitoring, every 10-15 minutes (or less) should be fine.

There are also additional efficiency gains to be made. Notably these can come through optimisation of the sensor’s signal processing algorithms and hardware in order to reduce the computational load.

Up next, there is the introduction of sleep modes as well as the incorporation of more energy-efficient components elsewhere on the board. Combined, these elements can make it is possible to create a more efficient IoT sensor. Indeed, by applying these techniques, we’ve been able to show it’s possible for an NDIR CO₂ sensor to run at sub 1.5 mW levels and with a peak of between c.33 mW.

For example, the GSS CozIR-Blink requires just 60 µJ per reading and delivers an accuracy of 30ppm.This places IoT sensors based on it not just in the range of AA batteries operation, but also in the range of energy harvesting supplies too.

IoT system development

Click image to enlarge

Figure 3: A schematic for an IoT CO₂ sensor for use in air quality monitoring. Switching to the new generation of low-power CO₂ sensors, with an appropriately set duty cycle, would enable this to run for years before batteries are replaced

A wireless CO₂ sensor suitable for use in IAQ systems consists of four major building blocks, a CO₂ sensor, microcontroller, low power radio module and a power source. As we’ve highlighted, advanced sensors can run at 60 µJ per cycle.

Next in the system is the wireless data transmission protocol, the choice of which can vary according to both power budgets and system requirements. Given these will often be implemented in buildings with solid walls and across wide areas, LoRaWAN’s increased range makes it a more natural choice. It is also more energy efficient than BLE.

Likewise, there is a seemingly limitless choice when it comes to the microcontroller, but ST Micro’s STM32L0, with its low power design and exceptionally-low-power sleep mode might make a good starting point.

This combination could be powered for many years via either AA batteries, or (depending on conditions in the room being monitored) by piezo / or solar energy harvesting.

Conclusion

Without localised testing across a building, health legislation intended to improve in-building air quality and reduce our exposure to CO₂ will directly lead to an increase in energy consumption from HVAC systems and therefore greater levels of pollutants emitted into the atmosphere.

If we are to minimise these effect, upcoming legislation needs to consider the use of IoT sensors throughout buildings to minimise the volume of air that requires refreshing and reheating/cooling.

Such sensors exist, however their high-power consumption levels make them badly suited to many facilities, especially schools / hospitals and other older buildings where power networks are not accessible in every room.

However, our work has shown that by adapting solid-state LED sensors to better meet the demands of the application and the needs of IoT systems – via changes to the duty cycle, through the implementation of more efficient algorithms, and by considering the power consumption of every on-board component – it is possible to reach the power levels required for such systems to operate either from AA batteries or even from energy harvesting.

GSS