What Does Sensor Data Really Mean and How Do You Analyse It?

Why all these sensors?

A wave of sensors has hit the market. FM operators and building owners have become avid users. Multi-sensors measuring 3-12 values are commonly delivered, but are we using all the data we collect correctly? How does it give you insight?

To analyse data, understanding the values is important, as is getting to grips with how to analyse collected data.

In this article, I intend to look at 3 different levels of analysis, while also addressing the most important sensors, what they actually measure and how we can make use of the measurements.

Once we have started a sensor project, data over time is of course important, and we need an interface that makes it easy to view history with a good time axis — as it is the history that is most interesting, not current values.

The first part of the analysis is to look at the various values individually. How high is the CO2 level in the meeting room, how high is the temperature in the work zone, and is there a lot of noise in the common areas?

The second part of the analysis is to present and compare data. Data alone may look normal, but it is only when you present it together that it can take on significant meaning. There was an abnormally high decibel level in the meeting room at the same time as CO2 and VOC were sky-high — could it have been an "unusual" meeting with a high level of activity?

The third part of the analysis is that this should be done automatically, based on rules and self-learning — which is machine learning. Data must find you as a user; you should not have to seek out a system. As I wrote in an earlier article, you can integrate sensor data into systems you already use, such as an FM system.

Sensor data

We have received many different sensors in recent years, but what does this actually mean for our analysis?

The Norwegian Labour Inspection Authority and Directive 444 have some guidelines on what constitutes a good indoor climate, but they are fairly broad and there is also room for interpretation.

According to Directive 444, for light activity we should have a temperature between 19 and 26 degrees. 19 degrees will probably feel cold for many, and 26 degrees will often feel like heavy air, even if the air quality is good — it is simply warm.

Is indoor climate important? With the focus on the WELL standard (to be explained in a later article) coming in full force to Norway, it is about "wellness" and ensuring that building users feel well. At the same time, it is well known that productive users are good for the bottom line — which is perhaps one of the reasons sensors have become so popular.

Director of BI in Stavanger, Dr. scient Ragnhild Wiik, conducted an extensive research project in which she found that a good indoor environment can increase productivity by a minimum of 3 percent. This has a significant impact on the bottom line.

Temperature can be a subjective feeling — some thrive at 18 degrees, others need 26 degrees. The Labour Inspection Authority Directive 444 recommends 22 degrees, which can be a reasonable starting point. Warm air often feels heavy and stale. This is likely due to both a direct effect on the mucous membranes and increasing evaporation and microbial activity at elevated temperatures.

CO2 stands for carbon dioxide and is probably the most commonly used sensor for air quality. CO2 is produced by the body's metabolism and is found in the air we exhale. Several studies show links between high concentrations of CO₂ in indoor air and health problems such as headaches and mucous membrane irritation, reduced work capacity and dissatisfaction. The threshold for schools and office buildings is generally 800-1000 ppm, and the Norwegian Institute of Public Health's standard is 1000 ppm. CO2 responds quickly to the presence of people, and in some cases sensor companies manage to use CO2 sensors for presence detection and usage pattern analysis.

Humidity should ideally be around 20-40%. We humans have no sense of whether humidity is at 20 or 40%, which is why sensors are important. Each adult emits approximately 40 grams of water per hour through exhaled air and is effectively a "humidifier" themselves. Too high indoor humidity very often leads to condensation and moisture damage and/or an increased tendency for mould growth, better conditions for house dust mites, and more off-gassing from various materials. Low humidity, on the other hand, is a major problem in winter. When we heat the cold and dry outdoor air, the result can be that one becomes more susceptible to illness and at higher risk of infection. You may experience dry mucous membranes, dry eyes and dry skin. Many also experience problems with static electricity.

Radon is, according to the WHO (World Health Organisation), the greatest risk factor for lung cancer after smoking, and it is estimated that 3-14% of all lung cancer cases in a country are caused by radon. The percentage depends on the country and how widespread it is. The action level for radon is 100 Bq/m³, and the risk of lung cancer increases by 16% with each increase of 100 Bq/m³. In buildings where radon is present, simple measures can be taken, and where there is a ventilation system, controlling it based on radon levels and occupancy is a good solution. Radon can also be used as one of the factors ensuring the ventilation system does not stop, as a marked increase is observed when the ventilation system stops or when air exchange is insufficient.

VOC stands for volatile organic compounds, or total volatile organic compounds (TVOC). These are generally gases we can smell. This sensor has not fully caught on previously, partly because we do not get to know which gas is triggering the reading — only that we have a high value. But recently, as history and data have come into focus, the VOC sensor has become an important component. As mentioned in an earlier article on data-driven facility management, large correlations have been observed between ventilation stopping and an increase in VOC, which can indicate a fault. Furthermore, you can use history before and after refurbishment — was the correct paint used? Are the furnishings as ordered?

PM sensors are coming in full force, and it will be exciting to see what they can do for our commercial buildings. Particle measurement is considered an important indicator for measuring dust and particles in commercial buildings. I believe this sensor will be a key piece for replacing filters based on condition rather than calendar, where differential pressure across the filter can be used while monitoring indoor climate and verifying particle ingress if you choose to postpone filter replacement.

Decibel sensors have been introduced in schools to ensure the correct learning environment. Noise is a common cause of reduced hearing, tinnitus, stress, muscle tension, poor well-being, and can contribute to nervousness, increased blood pressure, headaches and sleep difficulties. Noise in work environments such as schools and nurseries can reduce productivity and learning. It can also lead to hoarseness, coughing and vocal strain among staff. Surveys in some schools show that over 70 percent (on average approximately 50%) of pupils are troubled by noise, with the most troublesome noise coming from fellow pupils. Decibel sensors can also be used to verify that technical installations do not produce more noise than designed, while allowing us to detect changes in noise that may relate to problems with technical equipment. A lot of research is ongoing in this area.

Lux sensors measure how much light we have at the sensor location. Research shows that melatonin production increases at low light levels and in darkness, while cortisol production increases at high light levels. Too little light can make us more prone to depression, particularly in the winter months (SAD – Seasonal Affective Disorder). In healthcare, light therapy has long been used to reduce the effects of seasonally related mild depression. The most positive effects in terms of alertness, well-being and productivity are achieved at levels above 300 lux on walls with a horizontal illuminance of 500 lux on the work surface. Where there is no automatic light control, a light sensor can also indicate when rooms are active, whether lights have been left on in larger zones, and can contribute to simple energy management measures.

Push buttons are perhaps what Disruptive Technologies is best known for. They require some creativity to use correctly. At Proptech Bergen, we have used these for checking in/out of guests, accessing the premises during opening hours, and reporting faults to catering, cleaning, facility management and the site manager. As mentioned, this needs to be integrated into the system you use to get the full effect. We connected it further to Google Home, which also verbally announced that a fault had been reported from building users.

An important factor in the popularity of sensors is the creativity that makes data mean so much more. Use a temperature sensor at a desk and you suddenly get an indication of whether someone is sitting there. Use the same sensor on pumps and you can create a simple algorithm that says something about whether the pump is running or not.

A sensor can be much more than just the value — we must use them smartly and in combination with other data.

This article is intended as a basic introduction; I hope it proves useful to some. I will try to update it with more sensors — please provide feedback and requests for additional sensor types, and I will update it.

‍