GridDuck’s Current and Potential Use Cases

For all of us here at GridDuck, one of the most exciting aspects of our work is seeing the variety of use cases in which our system can benefit clients. 

Amongst our current clients, GridDuck’s use cases range from submetering for compliance and accounting purposes to sustainability credentials or financial savings. For example, dark kitchens can use GridDuck to identify how much each tenant owes for utilities, or large corporations/government bodies can use GridDuck to gain data for their sustainability/energy efficiency reporting. National Highways are a key example of this use case and benefited from GridDuck monitoring for this reason. 

Other businesses benefit from GridDuck as a means of identifying consumption habits and shifting their operations in order to make savings. Priors Grove farm did just this and saved 45% through GridDuck’s system this way. 

There are also different use cases for GridDuck based on our automation capabilities. For example, London Cocktail Club was able to save 35% on energy through automating drinks fridges to switch off out of hours. 

As we develop our technology further, we are only seeing more potential use cases, some of which we will be sharing in this article. So how can we see GridDuck being used in the future, and who can it help?

Balancing Renewables and Carbon Reporting

We have already been working on balancing renewables, for example, in agricultural sites which use GridDuck to automate their cold store usage in accordance with available solar energy onsite. Similarly, as National Highways demonstrate, GridDuck’s system can be useful for carbon reporting if businesses have certain sustainability compliance requirements and legislation they must meet. 

However, our technology also has the ability to monitor solar battery usage, HVAC systems and EV chargers on a larger scale than we have done so far. This can also be combined with carbon reporting requirements to give a complete overall picture of how much carbon a business is consuming compared to how much it is saving through renewable generation at any given time. There is no reason that carbon emissions cannot be a highly considered currency when analysing data, just as much as energy in kWh or money in £.

Anomaly Detection and Predictive Maintenance

Another potential use case for GridDuck moving forward is anomaly detection and predictive maintenance, which involves detecting breakages and/or operational errors. For example, if a fridge is left open after closing time in a restaurant, our data would identify that the fridge’s compressor cycle is outside of the normal data pattern, and flag this for the business virtually. This way, accidental wastage can be reduced. 

The graphs below show normal compressor data for a fridge and one anomalous data set. Each peak and trough in the two graphs represents the compressor cycle of the fridge. As you can see, it is around the same size each time, which we would expect to see. 

In the second graph, however, the last peak is almost twice as long as we would expect along the X axis. This means that the fridge has been left open or the compressor is malfunctioning in some way. 

Anomalies are also represented by the graphs below, in a more complex representation. 

In these graphs, there are two groups of dots, representing good (or non-anomalous) and anomalous data respectively. Good data (data that meets the expected consumption level) is data we have read historically and bad, or anomalous data, has in this instance been randomly generated/modified to make it anomalous.

These points on the graph represent complex vectors taken for data points within a given window of time which are then simplified into singular points. The window of time is calculated using an algorithm which identifies cycles and repetitions within historic data, in this case an appliance’s consumption. This is known as a fourier transform, which is used commonly in music to identify the dominant frequencies in sound and identify notes/keys. 

For our purposes, a fridge’s compressor cycle serves as the most common data set. However, with each appliance our system analyses, the time window will be slightly different.

Our energy saving system can then place these points on a graph to proportionally mirror the difference between vectors for anomalous and ‘good’ data. 

The blue and red graph represents the data in separate coloured zones. This is a mathematical estimation of the threshold for ‘good’ and anomalous data, making it easier to identify issues worth our clients investigating. 

Predictive maintenance is possible simply through using all of this information to notify our clients about potential maintenance issues. 

For example, if a fridge’s compressor has risen in consumption dramatically over several days, this may signal that the machine is breaking and will soon need to be replaced. Identifying such issues early saves businesses more in wasted energy, as well as helping to give greater warning for large maintenance costs. Furthermore, predictive maintenance can allow issues to be fixed before machines stop working completely, potentially holding up our client’s operations and thus costing them more money.

Overall, we look forward to integrating these processes into our platform and finding even more ways of helping our clients to save energy, money, carbon emissions and time. 

If you would like to learn more about our system and how we may be able to help your business, please book a 15-minute meeting using Calendly.