Open jinningwang opened 1 year ago
Jinning, For the nodal variables, such as LMP and nodal voltage magnitude, even nodal generation injection, the contour map can be a good way to visualize it. For the line power flow, the branch arrow is good.
@zmalkmus Zack, would you mind reminding me which file or where to start trying to build a HTML file for static layer?
@zmalkmus Zack, would you mind reminding me which file or where to start trying to build a HTML file for static layer?
Hi Jinning,
Sorry for the late reply. I wanted to think about this problem before answering.
The approach you want to make can work, but it would require solving a couple things. To create an html snapshot, you would probably have to find an image of where the simulation is supposed to take place and figure out how to map the static results to it (which would probably involve making your own coordinate system).
I believe it would be better to try and export the data from AMS in snapshots (like a multipage excel file or multiple csv files) rather than create html files for each step in the simulation. This way it we can build a seperate view on AGVis to model the results within leaflet and refrain from messing with the coordinates.
I have not looked at the AMS program yet so I could be missing some things. Would you be available to zoom sometime this week to discuss this further?
AMS has been developed to v0.8.1, where basic dispatch models has been finished but the output functionality remains to be done in the near future.
There is one trouble that, AMS has two major types thing to display, i.e., bus-type, and line-type.
Bus-type refers to the elements defined by Bus, such as bus voltage, bus angle, and bus omega/frequency (this is actually calculated from bus angle). This is also what we show for in the current version AGVis, that it animates bus-type variables in the map. In contrast, line-type refers to the elements defined by Line, such as line flow.
In the development of ANDES and AGVis, there is a dynamic model BusFreq
served as the PMU. The bus-type data is actually collected by PMU, and then broadcasting to AGVis through DiME. The ANDES output CSV has columns to be different variables and rows to be timestamps.
Back to the AMS output design, I guess the most intuitive design would be follow the same output style of ANDES, which means that, columns to be variables and rows to be timestamps. This design is okay for multi-time period dispatch, such as Economic Dispatch (ED) and Unit Commitment (UC). A potential trouble might be that, for the single-time period dispatch, such as DC optimal power flow (DCOPF) and real-time economic dispatch (RTED), the output file would be in single row.
Not sure if this will be a trouble, but I feel it is definitely doable. I'll try to work on the AMS output functionality and let you know if there is an example output file.
@zmalkmus Zack, would you mind reminding me which file or where to start trying to build a HTML file for static layer?
Hi Jinning,
Sorry for the late reply. I wanted to think about this problem before answering.
The approach you want to make can work, but it would require solving a couple things. To create an html snapshot, you would probably have to find an image of where the simulation is supposed to take place and figure out how to map the static results to it (which would probably involve making your own coordinate system).
I believe it would be better to try and export the data from AMS in snapshots (like a multipage excel file or multiple csv files) rather than create html files for each step in the simulation. This way it we can build a seperate view on AGVis to model the results within leaflet and refrain from messing with the coordinates.
I have not looked at the AMS program yet so I could be missing some things. Would you be available to zoom sometime this week to discuss this further?
Having a single row should work perfectly fine. Would multi-time dispatch need to be viewed differently than single-time? If this is the case we can add a value in the csv to distinguish which type it is and handle each type based on this value.
There are some differences between single-period (SP) dispatch and multi-period (MP) one.
For the SP dispatch, basically it is a static visualization, as shown in the figure "Power Flow" at the very beginning of this discussion. Regarding the MP dispatch, the user might need to switch between several timestamps. Let say, hour1 - hour24, or day1 - day7. Each of them is a static visualization, just the same as SP one.
However, there is another scenario that we might need to develop. In energy market cyber attack scenario, we might need to visualize the results in the dynamic manner, which means the ANDES animation manner.
Excellent. Visualizing both MP and SP should be almost identical then.
I've included the output function in the AMS latest development branch, https://github.com/CURENT/ams/tree/develop An Jupyter notebook example is here for more details, https://ltb.readthedocs.io/projects/ams/en/latest/_examples/ex7.html
Here PJM 5-bus system is used, and the example files are attached: Case file: pjm5bus_demo.xlsx SP Output: pjm5bus_demo_DCOPF.csv MP Output: pjm5bus_demo_ED.csv
Not sure if the PJM system coordinates are available in AGVis right now, maybe you can make up some pseudo coordinates given the bus name if necessary. Sorry for the inconvenience...
Visualization of static results
For now, AGVis is designed to show dynamic results from ANDES. In the future, we may also integrate dispatch results, or named static results, from AMS in this visualization. The dispatch results typically include power flow results and nodal price.
In the future, we may also integrate with AMS to visualize the dispatch results. The button to visualize power flow and nodal price may need to be added.
Power flow
We can use arrows on the transmission lines to show the power flow between buses. We can use the boldness or color of arrows to represent the numerical value of the power flow. An example is shown in the picture:
Picture source: Schäfer, Mirko, et al. "Principal Cross-Border Flow Patterns in the European Electricity Markets." 2019 16th International Conference on the European Energy Market (EEM). IEEE, 2019.
Another practice in the industry can be seen from the following examples:
However, this might not be compatible with AGVis implementation.
Price map
Another practice in the industry to visualize prices is to fill the price signal areas on the map in a pattern, as shown in the two following examples
However, similarly, this might not be compatible with AGVis implementation. We can stick with the frequency contour map as shown below, but discussions are very welcomed.
Regards, Jinning