BACK TO ALL BLOGS
24 Oct

Working out the numbers – how many sensors do you need?

Working out the numbers – how many sensors do you need?

Often, when we’re talking to customers they will ask us how many sensors they might need. This is a great question because it lets us talk a bit about the applications customers are using the sensors for.

Honestly, we don’t have all the answers on this because we’re still breaking new ground here. There are a lot of potential applications for Purrmetrix that haven’t been tested thoroughly. That said I thought it might be helpful to explain a few rules of thumb we tend to use in answering this question.

TL:DR – it depends upon what you want to achieve with your project. Contact us if you want to talk through the specifics of what you are measuring

How specific is your HVAC analytics project?

Many customers start working with us in exploration mode. They want to identify and pick apart all the problems in their estate. In that case we suggest a fairly high density to start with – a sensor every 10 sq m or one for every cluster of desks. So in a fairly average 60 person office we’d be thinking about 15 – 20 sensors.

In a case where you want to collect data around a known problem, it’s generally possible to be a bit more precise about the numbers, depending on the type of problem you’re looking at.

What sort of problem are you hoping to test?

If you are looking at problems with specific parts of your building services – for example in each fan coil unit – then you have a fairly obvious guide of one or two per FCU. Although kittens can be redeployed its always better in our experience to test all parts of the system simultaneously so allow enough numbers to do that.

On the other hand, you may be interested in how the building’s fabric is performing – how quickly certain parts of the building heat and cool compared with outside temperature. If you think you have generalised insulation problems then 3 or 4 sensors along each aspect can generate quite a lot of information on the rate of heat loss, although you should allow for more if the materials change significantly along each aspect.

Are you analysing or influencing?

If you are hoping to influence behaviour (whether to save energy or helpdesk time) then you need to be presenting data at the hyperlocal level for each person. The ideal extreme would be one for every desk or working area, but in practise we find that a sensor within the same 10 sq m is generally adequate. Remember they can always be moved to accommodate sceptics!

How are you displaying it?

Because monitors are limited in how many pixels they can display the webservice has limitations in the way it displays the image of your building/project, which will be sized to 600 px wide. This generally means it is tricky to display very large areas with a lot of kittens, or make very precise placement of kittens on a low scale (zoomed out) plan.

If you do have a project requiring a high density of sensors then make sure you zoom in and use the largest scale plan you have.

How the temperature gradients work

The colour gradient between kittens is not a reflection of the actual temperature in the gradient but of the confidence that it reflects the correct temperature. We don’t vary the size of kitten icon or the spread of the colour gradient so its spread is determined by the scale of the plan or image you upload the kittens above appear to be covering a floor area of around 6 sq m.

You can make the kittens disappear from the plan to better understand what is going on. Do this by clicking the arrow to the left of the view titla:

In general we don’t recommend uploading plans of any building of more than 60 meters.

In summary

  1. You don’t have to get it right first time, because you can redeploy your sensors
  2. Size your order to your use – general exploration and influencing behaviour will take more kittens
  3. Don’t try and display too many kittens in one heatmap view. Use other views (graphing) for that.
  4. Talk to us if you have a problem building and you’re not sure what the right level of diagnostics might be.