In search of a perfect world (model)
Over the last year, we’ve been very busy at Demand Logic rethinking our model of the world. We’ve known for a long time that building services are incredibly complex -- indeed, that is why we exist, because out of this complexity comes gentle chaos, and our service helps make sense of it, and fix it. Because they are so complex, our first step on the road to improving them was to make a relatively simple (but quite powerful) model of them.
This model largely depended upon the fact that both terminal units and major plant have a lot of data available from the Building Management System, such that they can be analysed in quite high detail independently of each other. While we know that they are physically connected (e.g. ceiling-mounted air-conditioning units or "terminal units" are connected via water pipes to boilers and chillers), a lot of analysis and fault finding can be done without having to model this fact. We can see if a boiler is running over night, or on a hot day, if a chiller’s strategy is demand-led and there are several terminal units always demanding of it, and we can see if a terminal unit is fighting itself. This has served us very well in reducing customers’ bills and increasing comfort levels.
However, we’ve also always known that it’s not as good as it could be, that actually even though this hot coil is reporting itself to be on, the boilers are all off, and that analysing the data independently misses these subtle but crucial insights. Our previous world-model has focussed on the most common properties of plant, those that give us very clear insights very quickly - things like output power, on-status and on-command. Again, this focus has served us very well in getting to the crux of major inefficiencies very quickly. But it’s also bit like paddling in Hampstead Ponds, when the Thames is there, rushing by… that is to say that it’s a safe bet, but we’re actually able to get our hands much dirtier, and feel the power of it too…!
So what does this mean for our new world view?
Over the last year, we’ve created a brand new internal architecture that not only closely maps onto what we know about in the torrents of data we already analyse, but is also powerful and flexible enough to be expanded and built upon as we discover new data points, encounter new weird control strategies, as vendors upgrade and introduce new sensors. In our new model, we have full end-to-end systems mapping, meaning that if your BMS thinks a terminal unit is heating (because its hot water valve is open), but actually the pumps are off at the moment, or the boilers are all off, then that unit will no longer tell you that its output power is anything more than what you’d expect ambiently.
We’re in the roll out phase of this new model, and will be pushing it to all our customers over the next months. As we go, we’ll be enhancing views and giving deeper insights into what your building is actually doing. You’ll see graphs becoming more honest about what they know and why.
Stay posted for some upcoming examples of what you can expect when you’re upgraded.
