Release notes part 2 - June 2022

Signatures are now available in Data Library!

This week we’re very excited to be releasing a feature that has been long-awaited and requested - being able to generate, look at and download energy signatures for each substation in a network! We even got a little bit excited and made flow, return and supply temperature signatures available while we were at it. This will come in handy for all of you who’d like to get a quick glance at what the load, flow or temperature will be for a substation at a given outdoor temperature. It is also a great opportunity for you to compare the design load calculated using machine learning (which you’ll find in the “Design Load”-section) to the design load calculated using energy signatures!

How do I get it?

Energy signatures are available to all our platform users. To get them, navigate to “Data Library”, select a year (signatures are only available for calendar years) and select the signatures you want to see from the list in “Manage Columns”.

Come and get your signatures

You’ll find energy signatures in the “Energy” section, flow signatures under “Flow”, while return and supply temperature signatures are found in the “Temperature” section.

When you’ve selected the signature you would like to see, hit confirm and you’ll have three columns to look at in the data library; a, b and t0

There you go, energy signatures for every substation in your network

The signatures are given as the parameters that make up a linear regression based on the measured hourly data from the selected calendar year. The signatures themselves look like this:

where Energy can be replaced by flow or temperatures. a is the slope of the signature, t is outdoor temperature, b is the energy when t=0, and t0 is the highest outdoor temperature for which the signature of each substation is valid.

The math behind it all

How do Utilifeed generate signatures?

For each substation in the network, we take each measured value and pair it with the corresponding outdoor temperature that was observed when the value was measured. So for a year, we will have a list of 8760 measured values and 8760 outdoor temperatures. The signatures are thereafter generated by performing a linear regression on that data, which outputs the parameters described earlier. 

Here’s the twist though; linear regression only works well if there is a quite obvious correlation between the outdoor temperature and the measured parameter, for example, energy. However, in district heating, it is common that you can’t create one linear regression that fits all data well during the entire year. This means that one can typically generate (at least) two different energy signatures for a single substation depending on the time of year. This is due to the fact that at higher outdoor temperatures above a certain threshold temperature, the energy use is close to zero.

Since the signature would be skewed by all the zeros at high outdoor temperature, we exclude data from above a certain threshold temperature. This threshold is found by re-calculating the energy signature several times, each time excluding data above different temperatures, until the error between the energy signature and the measured values are at its minimum. Because of this calculation method, each substation will have a different threshold temperature, since each substation's data is different.

Load Forecasts just got better!

One of the coolest things about the Utilifeed platform is that it can create load forecasts for each individual substation in a network, or for the network as a whole. If you’ve been hanging out in the Production section, you already know this. In our latest release, we’ve made the forecasts quicker to get and updated the interface so that it is a bit easier to understand what’s going on when you create a forecast. You are still able to forecast or simulate energy, volume flow and return temperature for a cluster or single substation!

The brand new forecast view

The new forecast defaults to use the latest stored forecast that has been generated. If you want to dig deeper, you can still create a custom forecast where you set the training data set used in the forecast. You can also create a custom simulation, if you’re interested in comparing old measured data to simulated data

Select which type of forecast you want

Some new symbols explaining which cluster or substation the forecast is for, which training data was used and which time period is simulated.

Bug fixes and minor updates

Delta temperatures

We found out a couple of weeks ago that delta temperatures were not visible in some views in Return Temperature Analysis. This has now been fixed and weekly delta temperatures are available to download from the return temperature table view again.

Design load columns in Data Library

While we were busy adding signatures to the data library, we snuck in a few more columns that are selectable there. They are:

  • Return Temperature at Design Load 1h
  • Return Temperature at Design Load 24h
  • Return Temperature at System Design Load 1h
  • Return Temperature at System Design Load 24h
  • Flow at Design Load 1h
  • Flow at Design Load 24h
  • Flow at System Design Load 1h
  • Flow at System Design Load 24h.
  • System Design Load 1h
  • System Design Load 24h

As you can see, they are all related to a substation's values at the system design load, or the design load of the single substation. The 1h denotes hourly values and the 24h denotes daily values. Go ahead and compare these values to the values you’d get from a signature at a certain outdoor temperature!

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Thank you! Check your email for a link to the report.
Oops! Something went wrong while submitting the form.