top of page

Eddie's first month at Full Matrix

Eddie

Updated: Dec 21, 2023

My name is Eddie Wilkinson and I have just finished my first month working as a Research Scientist at Full Matrix Ltd, Cambridge (UK). I studied Physical Natural Sciences at university, specialising in condensed matter physics and materials science. I felt a moral imperative to work on technology that would, in some way, contribute towards climate change mitigation. Fusion power will be a big part of either reaching or sustaining net-zero emissions in the next 100 years and the work being done by Full Matrix is necessary to the future commercialization of Fusion power.


This article will give an insight into what life and work are like as a graduate at Full Matrix.


I have been working, predominantly, on the repeatability of signal baselines in my first full month at Full Matrix. As explained in my recent blog article “A beginners guide to Full Matrix's tech”, the resolution achievable in crack/fault identification is completely dependent on how repeatable the baseline signals are for the experimental set-up.


With this in mind, I have been conducting a number of experiments to investigate how different factors affect this “repeatability”. This involved a number of studies, including:

  • The effect of temperature (on the properties of the metal pipe, of the electronics, on all the component impedances)

  • The effect of rogue electromagnetic and acoustic signals

  • How the measurement of the received signal is triggered

  • How the position of the EMATs on the pipe affects the form of the received signal

 

Temperature is a factor that affects almost every single physical process - on an atomic level it is an average amount of energy that an atom or molecule possesses. I set about trying to set up experiments to control for temperature. This was to see to what extent the difficulties with repeatability were due to temperature. I ran an experiment to investigate how the repeatability of tests at set temperatures varies as a function of temperature (Figure 1).


Fig.1: This scatter plot shows the standard deviation of received signals at integer temperatures. The value of standard deviation is normalised by the signal amplitude. Higher standard deviation means worse repeatability. Data were received from two separate EMATs, shown as EMAT A (red) and EMAT B (green). For each temperature there were either 4, 5 or 6 measurements in the dataset.

 

An example of what the raw data for each given integer of temperature look like is shown in Figure 2.


Fig.2: This plot shows the 6 received signals, using channel B, at 17 degrees across three days. Slight differences can be seen in the received traces, they are especially visible at lower amplitudes.

 

This shows an interesting increase of the average normalised standard deviation of the runs at higher temperatures. A high value of the average normalised standard deviation means worse repeatability. There were a number of hypotheses to explain this, the most interesting of which being that this change in repeatability was due to another factor that was correlated with temperature but not caused by it. I set out to test this hypothesis.

 

The times for which the repeatability was worst was the time that humans were in the lab. It was therefore a reasonable assumption this the poor repeatability could be due to something that we were doing. I then investigated what rogue electromagnetic and acoustic signals might also be present in the lab and being picked up by the equipment. A spectrogram was decided as the best way of presenting these data, so I went about automating the production of one. The results of the first spectrogram of the lab are shown below:


Fig.3: A spectrogram of the lab. Visible signals include: generated signal in the pipe (at and around 10kHz), CFL lights in the lab (40kHz-60kHz), and the heaters (25kHz and 26kHz). The colour shows the spectral amplitudes of the Fourier transforms of the received signal every 10 minutes for two hours.


During the data collection process I went around the lab switching various things on and off. I was able to show that the incredibly broadband signal from about 40-60kHz was due to the CFL lights in the lab, the very specific signals at 25kHz and 26kHz were due to two heaters in the lab, and the low frequency signals up to about 2kHz were due to audible acoustic signals and vibrations due to walking in the lab. These signals were only being created while people were in the lab with the lights and heaters turned on. These rogue signals are therefore correlated with temperature but not caused by it.

 

As mentioned before, for this technology the repeatability of measurements will likely be the limiting factor for the resolution achievable for fault identification. It is essential, by optimising hardware and software, to get as close to perfect repeatability (all received signals in the same environmental conditions being identical) as possible.

 

Accordingly in the next few weeks I will be investigating the use of hardware (resonant capacitors) and software (rogue signal removal by refining the Fourier transform of the raw received signal) in improving the repeatability of measurements.

 

My first month at Full Matrix has been incredibly enjoyable. Every day I find myself doing different and exciting science, with the opportunity to apply the knowledge and skills that I have to novel problems. Not just problems that are novel to the Full Matrix team but to the world of science. It is exciting and a privilege.


Fig.4: Me in the lab, early December 2023.

3 views0 comments

Recent Posts

See All

Comentarios


bottom of page