The Internet of things (IoT) is set to generate vast amounts of new data, with some predictions at over 100Billion ‘things’ sending out data on the Internet by 2020. Without new analysis methods, the sheer volume, at best risks the value of data being lost, and at worst, may overwhelm a business completely. So to make sure we create value from all this data, we need to concentrate on generating insights that businesses can use.

Internet of Things

So what is an insight in IoT? Insight is defined as ‘an accurate and deep understanding’ of something. In IoT, insights use data to generate an accurate and deep understanding of the processes that drive value across a business. These insights can influence key business decisions and enable feedback in a way that was previously impossible.

In looking at what insights are needed, a business needs to first understand where it has a gap in its knowledge that is costing money, reducing efficiencies or preventing new markets from being explored. This could be anything from being able to calculate when it is cheaper to replace a piece of equipment rather than continue to maintain it, through to understanding why some regions perform better than others.

So how does IoT insight apply in the real world? To illustrate this, let’s look at building HVAC (heating ventilation and air conditioning). Heating or air conditioning always seems to break down just when you need it most. Air conditioning systems always seem to break down in late spring and heating always seems to break in the autumn. If you look at failure statistics this is actually the case, as problems become obvious when the stress on a system increases. Keeping the HVAC systems in buildings and offices running is the job of the service companies, who are contracted to service and repair the systems. Imagine being the CEO of such a company. You would want to minimise the seasonal breakdowns as that can reduce the number of service teams required. From a business perspective it would be better to perform servicing throughout the year and schedule visits based on when a service is required. This requires insights. Let's walk through the Insight development that this company might need.

Before working out how to generate the insight, first you need to understand the goals. The first goal we know, which is to keep the number of service teams at the optimal level for the number of buildings. The second goal is to drive additional revenue through enhanced services. These goals then define the required insights.

The first insight is from analysing existing data to understand where to start. Using advanced analytics from reported faults, it was calculated that 40% were mechanical and 60% electrical. The top mechanical faults were:

  1. loss of refrigerant
  2. failure of refrigerant pump
  3. blockage of condenser


Analysing the data to deliver meaningful insights. Then analysis of the causes of these faults was performed to understand the key reasons for these in terms of serviceable items, age of items before failure, location and time of year. For these there are simple parameters that can be measured (refrigerant pressure, pump vibration, oil condition, power consumption and airflow) that can be used to feed healthcheck data into the system, with a risk/cost weighting, to show which of these has greatest cost to the service provider. This highlights the fact that, for certain types of unit, the refrigerant pump is NOT field replaceable, so the fault requires an entire unit replacement. In many cases this needs to be lifted in by crane, so is the highest risk item.

The electrical faults turn out to be related to lightning strikes and power surges. For these the analytics were more complex and required the use of machine learning algorithms that were given information on power quality in an area and incidence of lightning strikes as well as type of equipment , electrical test results and any protection equipment installed. From that it was possible to generate a risk profile for any site that would tell how susceptible to power related faults it would be and the risk of such an event in a 12 month period.

Get the basics right, and business benefits will follow. Without adding any ‘things’ there are already rich insights into costs and risks by combining internal data with external sources. Most importantly there is clear indication of the critical parameters to monitor to be able to accurately calculate when a service is due and the potential failures if this does not happen.

Monitoring systems were then fitted to sites where there were significant risks. Thresholds were set to visit site for a service, and new data started to be gathered to create new insights on how effective the service offerings are. Once sufficient field data was gathered, new insights were possible to accurately calculate the projected cost of service for units and calculate on the spot quotations for new service contracts. The insights on projected service requirements also allowed discussions on projected lifetime for a unit, and the opportunity to have robust data to show a client where buying a new unit is cheaper than continuing to service an old one.

It’s like new for old. From a simple beginning the development of insights means that air conditioning servicing is now like buying a second hand car, but getting the fixed cost servicing and variable service intervals that a new car would give you. The service company can offer new products and the building owners can get accurate data on service/fix/replace questions. Insight creates IoT value.

Add new comment

Comment editor

  • No HTML tags allowed.
  • Lines and paragraphs break automatically.
Blog moderation guidelines and term of use