Interconnection Innovation e-Xchange (i2x) Open Test Systems

This repository contains open grid network models and modeling scripts to test and compare different factors/measures (e.g., technoeconomic, social, equity, and engineering) that examine practices and policies for interconnection queue management and cost allocation. This repository is used for engineering training bootcamps, interconnection study guides, and an interconnection roadmap for the i2x project.

Users

The i2x DER package has been tested on Windows only, with Python 3.10. It does net yet support Mac OS X or Linux. During the installation process, a version of OpenDSS will be installed to work with the Python interface, i.e., you do not have to install OpenDSS separately. The steps are:

1. Install Python 3 if necessary. This is available from Python Site, Anaconda/Miniconda, or the Microsoft Store.
2. On the second panel of Python 3's installer, select the option that adds Python variables to your system environment, which includes the path.
3. From a command prompt¹, pip install i2x --upgrade

Once installed, invoke the GUI from a command prompt¹: i2x-der

Sources of background information include:

1. Slides from the first DER interconnection study boot camp, PNNL, NRECA and GridUnity.
2. CIGRE Canada Paper on hosting capacity analysis methods, focused on North America.
3. Multi-country Survey of Hosting Capacity

User Interface

When you start the program, six tabbed pages appear in a notebook format:

1. Network shows an overview map of the circuit selected from a library. Buses are labeled when they have either a substation source, or DER at least 100 kW in size. X and Y coordinates are arbitrary. You can pan and zoom this map using the toolbar, but can not modify it or obtain more data from it. A key to the legend is:
   • LN = line segment
   • XFM = transformer segment
   • REG = voltage regulator segment
   • SWT = switch segment
   • NWP = network protector segment
   • RCT = series reactor segment
   • SUB = substation
   • GEN = (conventional) generator
   • PV = photovoltaic generator
   • CAP = capacitor bank
   • BAT = battery, i.e., storage
2. DER summarizes DER and load on the circuit. kVA refers to the size, while kW refers to the requested output. The Available Residential Rooftops include single-phase load points without existing DER, with 120/240-volt service. You can choose a percentage of these to populate with new PV each time a case is run. The size at each location depends on local load:
   • The PV size matches the local load size, rounded to the nearest 5 kW.
   • When the rounded PV size is 0 kW, the minimum size is 3 kW instead. The rooftop locations change each time due to randomization, which usually produces a slight variation when you repeat a case, unless you set the useage rate to 0%. You can change the kW, kvar, and even the type of existing large DER, but you can't add new large DER at a different location. You can effectively remove DER by setting kW to 0.
3. Solar offers a choice of time-dependent solar output profiles, which apply to each PV in the circuit.
4. Loads offers a choice of time-dependent load output profiles, which apply to each load in the circuit. You can also apply a global scaling factor to the loads, which acts along with the load profile.
5. Inverters offers a choice of 7 inverter functions that respond to voltage, which apply uniformly to every PV and storage inverter in the circuit. Please see IEEE 1547 and its application guides for more details. The Power Factor applies to both CONSTANT_PF and VOLT_WATT modes. Enter a negative number to absorb reactive power, positive to inject reactive power. In other modes, the red curve shows how reactive power varies in response to voltage.
6. Output allows you to run a simulation and view the results. Click the blue Run button to run a new case. The other widgets on this page are:
   • Solution Mode should be DAILY for a 24-hour simulation using the pcloud or pclear solar profile. Use DUTY with the pvduty solar profile to focus on rapid voltage regulator and capacitor switching response. Less often, use SNAPSHOT to run a single power flow, with limited results shown.
   • Control Mode should be STATIC. Change this only if you are familiar with how it works in OpenDSS.
   • Stop Time specifies how long a period will be simulated. 1440 minutes covers one day, while 48 minutes covers the pvduty solar profile.
   • Time Step specifies the period between each power flow solution. There is a tradeoff between precision of the voltage fluctuations, and the simulation time. The software requires a value from 1 to 300, inclusive.
   • Output PV Details will show the production and voltage results for each PV in the circuit. There may be a lot of these, especially on residential roof tops.
   • Clear Old Output will erase prior results before displaying new ones. This is the default, so new results appear right at the top whenever you run a new case. If you unselect this option, please remember to scroll down to the bottom of the results each time you run a new case. The advantage is that you will now have a log of all cases run. Use copy-and-paste to another program to save any of these results.
   • Summary Results is a row of labels that will show in red when important limits are violated in the simulation results. See below for more details.
   • Detailed Results appear in the large white area below the other widgets, categorized as follows:
     • Number of Capacitor Switchings: the number of times a capacitor bank switched on or off. Expect no more than 2 per capacitor bank per day. If higher, PV fluctuations may be the cause.
     • Number of Tap Changes: the total number of voltage regulator tap movements. Expect one or two dozen per day per regulator. If higher, PV fluctuations may be the cause.
     • Number of Relay Trips: if not zero, PV reverse power flow may be the cause. Any time a relay trips, some load has likely lost service. Furthermore, the following voltage and energy results may be unreliable. When this is not zero, the DER hosting capacity limit has been exceeded.
     • Nodes with Low Voltage: how many node voltages fell below the ANSI C84.1 A Range, i.e., 0.95 perunit
     • Nodes with High Voltage: how many node voltages fell above the ANSI C84.1 A Range, i.e., 1.05 perunit
     • Load Served: total energy delivered to loads in the circuit
     • Substation Energy: total energy from the substation, i.e., the bulk electric system
     • Losses: total losses in lines and transformers
     • Generation: total energy from conventional generation
     • Solar Output: total real power production from PV
     • Solar Reactive Energy: total reactive power production from PV, in response to local voltage
     • Energy Exceeding Normal: EEN is an estimate of the load energy delivered under conditions of voltage outside normal limits, and/or conditions of line or transformer current above normal limits. This may indicate the need for grid infrastructure upgrades. It indicates that a DER hosting capacity limit has been exceeded. See the OpenDSS documentation for more details.
     • Unserved Energy: UE is defined like EEN, but with emergency limits rather than normal limits. Load would not be disconnected, but non-zero UE is a stronger indication that grid upgrades are needed, that DER hosting capacity has been exceeded, and that operational problems are more likely.
     • Minimum PV Voltage: among all PV, in per-unit.
     • Maximum PV Voltage: among all PV, in per-unit.
     • Maximum PV Voltage Change: the voltage change, in percent, is measured as the largest difference in PV voltage magnitude between consecutive time points. There is some sensitivity to the choice of Time Step. In more detailed OpenDSS modeling, signal processing techniques are applied to mitigate the sensitivity, but for illustrative purposes in the i2x-der software, that's not necessary. The voltage change, Vdiff, should be limited to 2% or 3%, depending on the local electric utility guidelines. Otherwise, nearby customers may complain. The Vdiff results consider only the PV locations, as the load Vdiff values should all be equal to or less than the worst PV value. The use of inverter control modes could mitigate Vdiff without having to reduce the amount of DER.
     • PV Details: if requested, shows the real and reactive energies, and the voltage results, for each PV in the model.
   • Check for Updates will compare your installed software version to the latest on PyPi. Requires an Internet connection.

Some other important notes about the program:

• The main window is resizable. The graphs and the output results display may increase in size.
• Close the program by clicking the X in the top right corner.
• As you run simulations, some logging messages appear in the Command Prompt. You don't need to pay attention to these, unless an error occurs. If there is an error message, please copy-and-paste the message into your issue report.
• Please report any comments, suggestions, or errors on the issues page. Before submitting a new issue, check the others listed to see if the problem or suggestion has already been reported. If it has, you might still add new information to the existing issue as a comment. The issues page is better than emailing for this purpose, as it helps the team organize these reports and updates. It also creates a public record that may help other users.

Examples: 9500-Node Network

When you first start i2x-der, the IEEE 9500 node circuit is displayed. We can use this to examine the effect of inverter controls on solar-induced voltage fluctuations:

• Go to the DER tab, and reduce the usage of residential rooftops to 0%. This makes the results repeatable.
• Go the the Output tab and run a case. You should find the maximum PV voltage fluctuation to be at or near 0.8656%. This is less than 2%, and should be acceptable, but that's on a clear day.
• Go to the Solar tab and select the pcloud profile. The graph shows much more variation in output. Use this profile for the rest of the example. If you run the case again, the voltage fluctuation should exceed 3%, which is not acceptable.
• Go to the Inverters tab and try non-unity power factors, e.g., 0.9 and -0.9. One of these improves the voltage fluctuation, while one makes it worse. Both choices result in significant levels of PV reactive energy.
• On the Inverters tab, try the VOLT_WATT function, which is designed to mitigate steady-state voltage rise. It doesn't affect the voltage fluctuations in this case, i.e., you should get approximately the same result as you did with the same power factor in CONSTANT_PF mode. The IEEE 9500-node circuit doesn't have significant voltage rise problems, even if you were to add much more PV.
• On the Inverters tab, try the other functions. Results are tabulated below.
  • VOLT_VAR_CATA has a small beneficial effect, but it's not very aggressive in using reactive power.
  • VOLT_VAR_CATB is more aggressive, but only outside a "deadband" of zero response (see its graph). In this case, the voltage fluctuations occur mostly within the deadband, which spans 4%.
  • VOLT_VAR_AVR uses the most aggressive response allowed in IEEE 1547-2018, along with "autonomously adjusting reference voltage" as described on page 39 of IEEE 1547-2018. There is no deadband, but the setpoint is not fixed at 1 perunit reference voltage. Instead, the se_VOLT_VAR_CATB_ point follows the grid voltage with a response time of several minutes. The effect is to resist sudden voltage changes, while not resisting longer term changes in grid voltage. In this case, it reduces the voltage fluctuation below 2%, and the PV reactive energy is only 0.80% of the PV real energy. There are higher short-term transients in PV reactive power, but over the day these net to nearly zero. On the other hand, the CONSTANT_PF result with -0.9 power factor also reduced the voltage fluctuation below 2%, but the PV reactive energy was 48.4% of the PV real energy, i.e., the PV absorbed reactive power all the time.
  • VOLT_VAR_VOLT_WATT uses both VOLT_VAR_CATB and VOLT_WATT, at unity power factor. Because of the deadband, it doesn't help with voltage fluctuations in this case.
  • VOLT_VAR_14H uses both volt-var and volt-watt characteristics according to Hawaii Rule 14H, which was developed for an area that has high steady-state voltage rise on some long secondary circuits. The volt-watt characteristic is more aggressive, but the volt-var characteristic has a wider deadband, of 6%. As a result, it helps even less with voltage fluctuations in this case.
  • Although not illustrated here, VOLT_VAR_AVR may be combined with VOLT_WATT to address steady-state voltage rise along with voltage fluctuations. This is the same combination in IEEE 1547-2018 that allows the VOLT_VAR_VOLT_WATT and VOLT_VAR_14H modes.

Profile	Inverters	Max Vdiff [%]	Notes
pclear	CONSTANT_PF=1.0	0.8656	No problem on a clear day.
pcloud	CONSTANT_PF=1.0	3.1382	With clouds, too much voltage fluctuation.
pcloud	CONSTANT_PF=0.9	4.5609	Injecting reactive power makes it worse.
pcloud	CONSTANT_PF=-0.9	1.6858	Absorbing reactive power all the time.
pcloud	VOLT_WATT, PF=-0.9	1.6999	Close to CONSTANT_PF result at same power factor.
pcloud	VOLT_VAR_CATA	2.8752	Helps a little.
pcloud	VOLT_VAR_CATB	3.0721	No help in the deadband.
pcloud	VOLT_VAR_AVR	1.5747	Setpoint adjusts to grid voltage over several minutes.
pcloud	VOLT_VAR_VOLT_WATT	3.0721	Still no help in the deadband.
pcloud	VOLT_VAR_14H	3.1209	Still no help in the deadband.

Suggested exercises for this circuit:

• Add more residential rooftop PV without exceeding the hosting capacity.
• Use the pvduty solar profile to explore the effects of inverter control mode on regulator tap changes.
• Use the DER tab to replace as much of the conventional generation as possible with PV. How could you quantify the effect on local air quality?
• Use the DER tab to increase the existing pvfarm1 size as much as possible.

Examples: Low-Voltage Secondary Network

The second available circuit is an IEEE Low-Voltage Network Test System. It comprises 8 radial primary feeders that supply a grid of 480-V and 208-V secondary cables in an urban, downtown area. This design provides economic, high-reliability service to dense load areas, but it does not support very much DER. The network protectors (NWP) trip on reverse power flow, as intended for faults on a primary feeder. DER can also cause NWP trips under normal conditions, which is not intended. To explore this effect:

• On the Network tab, select ieee_lvn and review the locations of NWP with respect to the primary feeders and the secondary grid. There are 8 fixed PV locations indicated in yellow. In a dense urban area like this, there are no available residential rooftops for single-phase PV.
• On the DER tab, the PV are dispatched to a total of 800 kW, which is only 1.9% of the peak load. IEEE 1547.6-2011, which is an application guide for DER on secondary network systems, refers to such a limit as "de minimus".
• On the Output tab, run the case. There should be no voltage problems because of the strong grid, and no relay trips because of the "de minimus" quantity of DER.
• On the DER tab, change each DER to dispatch at 1000 kW, as might have been intended for 1095 kva ratings. This is still only 19% of the peak load.
• On the Output tab, run the case again. Now, you should see 4 relay trips, and some of the loads are unserved. Two of the eight PV were also disconnected. This result is not acceptable.
• On the DER tab, adjust the individual DER kW and kva parameters to achieve as high a hosting capacity as possible.
• Some changes to the traditional NWP scheme have been investigated to increase the DER hosting capacity, but these are advanced topics and not considered in the i2x-der software.

Developers

Familiarity with git and Python is expected. Experience with OpenDSS is also helpful. The steps for working on the i2x Python code are:

1. From your local directory of software projects: git clone https://github.com/pnnl/i2x.git
2. cd i2x
3. pip install -e . to install i2x from your local copy of the code.

The steps for deployment to PyPi are:

1. rm -rf dist
2. python -m build
3. twine check dist/* should not show any errors
4. twine upload -r testpypi dist/* requires project credentials for i2x on test.pypi.org
5. pip install -i https://test.pypi.org/simple/ i2x==0.0.8 for local testing of the deployable package, example version 0.0.8
6. twine upload dist/* final deployment; requires project credentials for i2x on pypi.org

Bulk Electric System (BES) Test Cases

Two BES test systems are under development at CIMHub/BES. These will be used in BES boot camps and i2x sprint studies.

License

See License

Notice

This material was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor the United States Department of Energy, nor Battelle, nor any of their employees, nor any jurisdiction or organization that has cooperated in the development of these materials, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness or any information, apparatus, product, software, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof, or Battelle Memorial Institute. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

PACIFIC NORTHWEST NATIONAL LABORATORY
            operated by
             BATTELLE
             for the
 UNITED STATES DEPARTMENT OF ENERGY
  under Contract DE-AC05-76RL01830

Copyright 2022-2023, Battelle Memorial Institute

Footnotes

1. On Windows 10, this may be found from the Start Menu under Windows System / Command Prompt. On Windows 11, one method is to search for Command Prompt from the Start Button. Another method is to find Terminal under All apps from the Start Button. ↩ ↩²