Research, research, research… it is whats paying my bills this summer and it’s also something that we don’t do enough of in the USA. Karen and I have roughly six weeks left in Norway so I thought I would give everyone a little update on the things I have been working on since I started.
When I first started at work this summer I was focusing on continuing a project I started during the spring semester at MTU. This project was work on the single pole tripping and reclosing that I had described in an earlier post (click here for the old post). In 2005 a previous master’s student at MTU had written code for a program called Alternative Transients Program (ATP) to more topologically correctly model a transformer and transformer core. The project I had been working on during the spring semester was to create/update models of the same power system using a program called ATPDraw. ATPDraw is basically a higher level graphical programming language for ATP. ATPDraw provides a user interface that allows easier programing and implementation of models based on something that would be similar to a cad package. By this I mean you can visually see symbols and graphics to do the programming instead of just writing text lines of computer code. The newest versions of ATPDraw contain many new models, features, and updates over the programs used in 2005 by the other masters student. So, my project (working with 2 other students) was to create, update, and correct the models built by the student in 2005 using the latest ATPDraw version. We had those models built, partially benchmarked, and ready to roll by the time the spring semester was over.
This leads into my first task here at NTNU this summer. I wanted to see how the results compared from the original hand-coded models to the new models we had built. The purpose was to show that although we made changes, corrections, and updates the older models still provided valid results and that the conclusions and suggestions made based on those results were still OK and valid. The main problem in doing this was to learn how to run the old files in the new software. This took me a while to figure out but luckily once I realized how to do it things went rather smoothly. These old files are also not as fast to run and require a larger amount of tedious work to run the number of simulations and then tabulate the results. I’m not sure on the exact number of days or hours that I spent working on this part but it did fill a good chunk of my time for the first few weeks.
After I got all of the files ran it was time to tabulate some results and see how everything compared. In short, everything matched up really well. Across all of the cases the largest difference I saw was about 6.3% and that was just for one measurement in one case. All of the other measurements in the other cases were typically around 2% or much less. This is a very good match considering we made several updates, corrections, and small improvements. This shows that the original work provided valid results that are still usable and models that can be used in future projects if needed. Although, in all likely hood any future work would be done with our newer models and the latest version of the program.
After this project was done and I had a chance to present my results to my adviser it was time to move onto some new work. During the weeks leading up to this point my adviser, Nils (the other MTU student), Nicola (PHD student at NTNU), Hans (the professor we are working with here), and myself had been meeting to discuss the work that we needed to perform in the lab to get things rolling with the papers we want to write. Since I was done with my first project and Nils had also completed work on a separate project related to ATPDraw modeling of Static Var Compensators (SVCs) it was time to move forward with the lab work. We took a couple of days just to digest all of the information from the previous meetings, locate all needed testing equipment, and setup everything up for the tests.
For the first cases of study we choose to use a 22kVA single phase isolation transformer. This allows us to have a simple “base case” and then launch into more complex 3 phase transformers. The tests we were to perform are known as no load tests. This means that the low voltage side of the transformer is connected to the voltage source and nothing is connected to the high voltage side – hence the “no load”. In the case of this transformer both sides of the transformer are at the same voltage (its an isolation transformer) so we connected the voltage source to the terminals with the winding closest to the core material.
A total of 8 different tests were ran on the transformer each with varying levels of input impedance. This input impedance ranged from 9-24 Ohms and was provided by two different methods. First we used a resistive load cart and then we used a reactor (inductor). We did this because we wanted to look at data from each way of doing it. For each level of added impedance we also energized the transformer at different voltage levels ranging from 50% to about 130% excitation.
It took about a week or a little more to setup and run all of the tests. After the data was collected it was then time to do some post processing to see what it all meant. Nicola, the PHD student here, already had some initial programs written in MATLAB to do some of the first processing. He was able to quickly process the data so we could take a look at what was happening. His results showed how the magnetization curves were differing based on the distortion level of the voltage which in turn is tied to amount of added input impedance. The higher the input impedance the more the voltage is distorted. This is what we were expecting to see. He also showed that some routines written previously by my adviser worked for this application as well. After his results it was time for us go a step further and take a look at the exact harmonic content of the data we gathered.
Nils and I partnered up and wrote a routine to do this in MATLAB using Fast Fourier Analysis (FFT). MATLAB stands for Matrix Laboratory and is a very useful engineering program. Between the 2 of us this took a few days to get to work properly but we are now able to see the exact frequency content of the voltage waveform used to energize the transformer. This was not exactly an easy task due to the way the original data was required. Without getting into the exact details I will just say that there was a lot of manipulation of the data that needed to be done first before the final results could be calculated. This routine returns the exact frequency, magnitude, and angle of each voltage harmonic present in the applied waveform. In the last few days I worked on adding some additional functionality which allows us to calculate and see some other properties that both the IEEE and IEC standards for transformer no load testings are concerned with. All of the results will be used to write our journal papers and make our case for the needed changes.
So, those are the basics and highlights of what I have been doing so far this summer. It will be interesting to see what I end up working on during my remaining time here but I suspect it will more geared towards the completion of the first journal paper. Comprehensively compiling test results, creating circuit diagrams for our test setup, and writting/reviewing written material for the paper. Stay tuned for more updates and if you have any questions or want clarification on anything written here please don’t hesitate to ask. Leave a comment or shoot me an email.