What can we help you with?

Why does energy data contain so many errors?

 

You don’t need a lot of errors before everything you see is garbage. Without you knowing what you can trust or where you should start tracking down the fault. Why is that so?

For many people, getting correct energy data is a nightmare. Often, even tiny mistakes can turn everything you get into meaningless gobbledygook. Typical challenges are phase displacement, incorrectly configured data loggers, swapped data logger IDs, time variances, plain old installation errors, loose cables, etc.

There can be many causes but, in our experience, most of them stem from the enormous complexity of the task. Without being able to verify data in arrears, how are you supposed to know whether the figures you receive make sense or not?

In this context, hierarchy is key. If you know what belongs to what, you will have a better chance of seeing if the measurements have any basis in reality. Grouping things together, by means of tagging for example, is important to enable you to perform large aggregations and “sanity check” the data after it has been generated.

We like to sort energy data errors into three different categories:

  • Structural errors

  • Electro-specific errors

  • Thermo-specific errors

1. Structural errors

These relate to how energy data logging works in practice. Typical errors here include a mismatch in time between data loggers – so-called time variances, or that the IDs of a whole set of data loggers have changed.


Time stamp errors

Here, you will be able to see a mismatch in time. Everyone involved may have done everything right but we still see complications that can result in different data loggers having different time zones. This is one of the risks of going into an existing installation.

 

Data logger IDs change

Imagine that you have one set of data loggers but two sets of ID numbers for them. If the ID number of one or more data loggers changes, you no longer know what is being measured. The result is chaos and you will have quite a jigsaw puzzle to piece together so that you can measure things correctly again.

In this kind of situation, it will look as though you have one set of new data loggers measuring something, but you do not know what. It can feel like this: “Before, I had one set of time series and I knew where they belonged. Suddenly, they all changed their names, without indicating to what, so now I don’t know which measurement the time series represent.”

How to solve the problem: We check that the data loggers and what they are measuring make sense when we start up.

Here, it is important to perform a thorough check. The idea is that if we have continuous data, we can make a rough estimate by matching the last valid measurements before any changes took place. We can also analyse patterns before and after a change, to see what has changed. We are working continuously to automate this task and find a simpler solution that solves the problems caused by changes to data logger IDs.



2. Electro-specific errors

Some errors are specific for electrical energy. One example is a phase error from installation, a so-called phase displacement. Phase errors are physical errors that arise when the clip tightens around the phases (typically 1,2,3) in the wrong order. You will then get a curve that looks correct but is at least 30 per cent wrong.

 

Example of a phase error

A power cable has three phases: 1-2-3. A typical error which occurs at installation is that the phases become 2-3-1 instead. In that case, the data logger generally measures incorrectly and the data you receive makes little sense.

Our algorithm identifies and corrects this type of error so that you obtain the correct data. We can, for example, detect incorrect connections, given certain conditions. If we detect that it has been connected incorrectly, we can also find the most probable wiring diagram and reconfigure it so that the data output is correct.

 

 

The pictures above are a before/after example, where the “before” is how it was set up and the “after” is how we believe it should be. We compare all the currents on each of the two phases (only two because it is a three- phase IT network) and examine how the sum of the children compares to the parent (in the hierarchy of data logging points). The sum of the children is not exactly the same as the parent because not all the children are logged. In examples where absolutely everything is logged, the sum of the children must be identical to the parent (apart from deviations caused by metering noise).

 

Two different “networks” – and faulty connections on them

In our work, we generally encounter two different networks: Terra Neutral (called TN) and Isolée Terre (called IT). We think TN is the nicest to work with. TN has three curves for voltage and three for current. With IT, you are only supposed to have two data loggers on a three-phase, and this creates plenty of room for errors when you have three phases and two directions. If you connect three, you will get an error. In that case, you are measuring different values to those previously expected. There are quite a lot of opportunities for incorrect connections even with TN (which is the easy one).

How to avoid this kind of error? Our developers are working continuously to avoid such connection faults because you can end up with substantial errors if there are displacements. In these cases, the measurements can look correct during the day, with errors often not revealing themselves until things are turned off or turned down. At night, the data is obviously incorrect. For example, it can look as though we are producing electricity at night – which is not possible when things are turned off/down. Errors like this generally happen because the person who installed the system has put the clip on the wrong phase. In that case, all the data will be corrupt.

We create systems that enable the installer to troubleshoot on site. As far as possible, we must ensure that the installer is provided with all necessary information . It is the installer who has the very best foundation for doing things correctly. They must be able to check that the phases are where they

are supposed to be. Providing an installation guide is of great benefit for the installer, since they can verify or make changes during the installation if an error is detected.

 

3. Thermo-specific errors

Some errors are specific to thermal energy. For example, an accumulator may stop increasing. Kilowatt hour (kWh) is a value used by such a data logger. Sometimes, it simply stops – without explanation and without increasing, even though the energy is increasing rapidly. In that case, you get zero in consumption. We don’t know why this happens, and this error also sometimes occurs in electrical installations. The advantage is that in thermal systems we can still work out how much energy has been used.

How do we solve this? To measure thermal energy, we must calculate the amount of heat exchanged from district heating water to the recipient. So we need a temperature measurement in/out and a measurement of how much water has passed through during the same period.

Many things may be logged to find out how much energy has been used. In other words, we reconstruct the missing data by using other available data about energy, load, temperature, volume, flow rate, etc. As long as we have specific combinations of this data, we can reconstruct what the data logger should have recorded if it had been working properly.

 

What is high-quality energy data? Read our article.

 

Layer

Contact Glitre Energy Solutions to learn more about how we can help

Recent posts

What do we mean by worry-free energy data?

What do we mean by worry-free energy data?

Energy data has been a challenge that has slowed down the development of EMSs and energy ... Les mer »

What is high-quality energy data?

What is high-quality energy data?

Smart EO gives you energy data that is ready-processed to a high level of accuracy. We call it ... Les mer »

Why does energy data contain so many errors?

Why does energy data contain so many errors?

You don’t need a lot of errors before everything you see is garbage. Without you knowing what you ... Les mer »