3D Animation of Power System Data

3D Animation of Power System Data

Sean Indelicato, Blair Johnson, Shutang You

Abstract: The goal of this project was to create a 3D model of the electric power grid using data from FNET. This form of visualization allows for better analysis of disturbances. The program we developed in MATLAB can read this FNET data and interpolate it on a map using the latitude and longitude of the frequency disturbance recorders. We were able to gain better insight into the dynamics and trends of the power grid through observation of the visualized data. Location of disturbances and propagation speed of instability were both easy to estimate using this 3D method of visualization.

Key words: Frequency, Angle, FNET, FDR, MATLAB, visualization

1. Introduction

FNET is a frequency monitoring system. It uses the Frequency Disturbance Recorder (FDR), as shown in Fig.1, as its sensor. FDRs spread across the United States as well as other parts of the world. The FDRs measure the change in frequencies to the nearest ten-thousandths place. The frequency should be very close to 60.00 Hertz, as that is the standard in the United States. Each FDR can measure the frequency and voltage angle at household outlets while timestamping the measurements by GPS information. The accurate GPS timestamp realized wide-area synchronized measuring of the power grid.

The US power grids consist of three synchronized interconnections: the Eastern Interconnection, WECC, and ERCOT. Fig.3 shows the FDR deployment map in the three interconnections in the U.S. The data collected by FDRs are transmitted to data centers running at the University of Tennessee and Oak Ridge National Laboratory.

Figure 1. Frequency Disturbance Recorder [2] Figure 2. FNET system architecture [1]

Figure 3. FNET FDR Deployment Map [1]

The data allows us to triangulate the location of any disturbance in the grid, and generate a map to visualize how a disturbance affects the grid. A disturbance is caused by either a fault of the network, a sharp increase of load, or a malfunction in a power plant that causes less electricity to be available to the consumer. The frequency is an indicator of the balance between the energy producers and consumers. A frequency higher than 60Hz usually means a generation surplus while a lower than 60 Hz frequency means generation insufficiency. Our objectives included visualizing the frequency and voltage angle data from FNET as an animated 3D contour map, creating an automated program, and understanding how this new imaging can provide better insight into the electrical grid.

2. Visualization of power system measurements

Frequency disturbance recorders provide four important types of data: Location of the FDR, frequency of the grid at the FDR’s location, voltage angle at the FDR’s location, and the exact time of frequency and voltage angle readings. This data is critical to the visualization process. The information is recorded in the format of a text file, as shown in Fig.2.

Figure 2. Format of the recorded FDR data

2.1. Methods

We wanted our program to be a versatile, automated, and efficient as possible. The program needed to be capable of automatically importing and sorting the FDR data, visualizing frequency and voltage angle data, and storing and recalling previously run animations. The workflow of the project followed the following outline.

Figure 4. The structure of the MATLAB animation program

1) With help from our mentor, we added code to the program that imports data from the FDRs into MATLAB and sorts it into variables. The complete automation of this process was not achieved, as errors such as missing data, and FDR location data needed to be corrected.

2)A script was written to associate the FDR data with the corresponding FDR locations within a unified matrix.

3)A grid was then constructed using the latitude and longitude values from the imported FDR location data. This grid was created using a “meshgrid” function.

4)The FDR data was then interpolated over the previously created meshgrid using the “griddata” function.

5)The “vec2mtx” function and the built in “coast.mat” file were used to create a logical matrix in the shape of the United States’ coastline. The matrix was then saved as a .mat file in the project folder where it could be read from as needed. This greatly improved the efficiency of the program.

6)The coastile matrix was compared to the interpolated FDR data, and the data points that fell outside of the coastline were removed.

7)The “surf” function was then used to create a 3D contour map of our data. This “surf” function was placed inside a FOR loop to animate it, with each tenth of a second of data corresponding to a single frame, and each frame being stored in cell matrix. This matrix was then saved as a .mat file in the project folder until manually reset.

8)When run, the program retrieves and plays each frame from memory. This enables the animation to play much more smoothly than if it were being calculated in real-time.

9)Each individual case was then manually debugged. This was done by running the program until it encountered an error and then deleting the FDR with missing data, or manually adding location data for FDRs with missing GPS locations.

10) A “line” function was used to draw a 2D map of the U.S. state borders on the ideal z axis value. For the frequency graphs this map was at 60hz. On the phase angle graph, this reference point was placed at 0 radians.

Visualization of phase angle measurements required more processing, as unlike the raw frequency data, the raw angle data is unusable in its original state. In addition, further processing was required to isolate small oscillations in the phase angle data.

1)Phase angle data was unwrapped using the “unwrap” function.

2)The mean of the unwrapped angle data was then calculated and stored.

3)The difference between the unwrapped data and the mean was found and visualized using the same method as frequency data.

4)To isolate oscillations in the phase angle data and remove noise, the “memd” function was used to decompose the difference from the mean phase angle data into component channels. Only the channels containing the smaller oscillations were stored and subsequently visualized using the previous method.

2.2 Results

The completed program displayed a 3D animated contour map of the United States. The x and y axes represented latitude and longitude, and the z axis represented either frequency in hertz or angle deviation from the mean in radians. This 3D method of visualizing the data made locating disturbances very easy, and in every case tested, we could clearly see the beginning of the disturbance and the propagation of instability across the grid. Using a 2D plot of our data, it was easy to find the time of the location as each frame corresponds to .1 seconds. Using the location and time of the fault it was possible to estimate the propagation speed in miles per second using the formula:

Ps = (La - Le) / ( (Ta - Te) * .1)

where La is the location of a given point of arrival, Le is the location of the event, Ta is the time of arrival, and Te is the time of the event. Analyzing the animated contour map allowed us to identify which regions of the power grid were more susceptible to instability than others.

Figure 5. 2D plot of frequency data (left), and 3D plot of frequency data (right)

Figure 6. 2D plot of phase angle data (left), and 3D plot of phase angle data (right)

Figures 5 and 6 show data from the same case. On April 27, 2011, several large hail storms hit East Tennessee. These storms took out eleven 500kv transmission lines. The storm resulted in numerous extended outages and damage to roofs and cars. Examination of the 3D map in Figure 6 reveals large phase angle oscillations beginning in East Tennessee and spreading outward. It is also apparent that aside from the location of the initial disturbance, the midwest and Florida are particularly susceptible to large oscillations.

4. Conclusions

The use of 3D animation provides a superior method of viewing and analyzing power system data. This kind of visualization allows the user to easily see the time and location that an event occurred, as well as estimate the propagation speed of a disturbance. The use of 3D visualization also makes it easy to observe the dynamics of a power system, as well as trends such as major oscillation regions.

Acknowledgements:

Special thanks to Dr. Chen and Erin Wills.

References

[1] FNET Web Display. University of Tennessee.

[2] Frequency Disturbance Recorder. CURENT.