Assessing the effect of PV generation on distribution system voltages

October 13th, 2014, Published in Articles: Energize

 

Utility distribution systems have seen a huge increase in distributed energy resources (DER) in recent years not only due to renewable portfolio standard requirements but also government financial support. Furthermore, it is becoming much more common for DER generation levels to significantly exceed the local load on the feeder to which it will be connected, which creates many more situations requiring investigation due to significant reverse power flow. For solar photovoltaics (PVs), different seasons and hours of the day need to be researched to gather and develop best fit insolation curves to assist in the power flows with the distribution system local loads.

Solar PV distributed energy resources (DER) integration assessment has become increasingly challenging as larger PV plants with capacities exceeding local load levels are becoming quite common. The required areas addressed during the impact study are much more vast and complicated than is the case with conventional induction or synchronous generators, resulting from the different operational mode capabilities. Each mode may cause different power flows to occur and may impact the system with tradeoffs between voltage improvements and loss increases. Developing reasonable power generation profiles for intermittent DER is another complication as it has a direct effect on voltage profiles and how line voltage regulators or any other type of active voltage regulating devices will operate to maintain sufficient voltages. Investigating a variety of settings for this equipment in tandem with the PV inverter control modes, such as voltage following versus voltage controlling, is critical to system reliability.

Case studies with 1 MW PV plant

The possible impact a PV solar generation plant may have on the distribution system voltages is now presented in an example case study, currently in the design stage, by addressing various voltage regulator control and PV inverter operating modes. To fully evaluate local circuit load and coincident generation levels, customer load has been allocated throughout the circuit for the minimum and maximum summer and winter demands. Commercially available software is used for the power flows and voltage drop profiles for the study results. Fig. 1 represents the case study circuit.

Fig. 1: Case study distribution circuit diagram.

Fig. 1: Case study distribution circuit diagram.

Inverter operating modes

The PV inverter is another important modeling factor for addressing distribution system voltage impacts. The various operating modes function very differently from an active power and reactive power standpoint. Traditionally for smaller PV installations, active voltage regulation on the distribution system has not been accomplished, and thus inverters have been operated in a very static unity power factor mode. The amount of active power output is directly related to the insolation level at that particular time. Fortunately this is changing as DER sizes are increasing and thoroughly evaluating the system to determine set points for the inverters reactive power modes prove to be very beneficial and may reduce line losses and help hold voltages within tolerances [1].

Step voltage regulators

This circuit traditionally contained step voltage regulators (SVR) at two locations, one at the substation feeder bus and the other approximately half way out on the circuit. Reverse power flow due to DER, along with multiple SVRs in series, as in this case, may have adverse voltage impacts on the system and therefore their settings must be known to complete a thorough impact study to ensure safe and reliable operation. SVRs may contain seven or more operating modes such as forward and reverse bi-directional, reactive bi-directional, and co-generation modes. Line drop compensation (LDC) also behaves according to the level of real and reactive current through the regulator control. The forward or reverse operation depends upon the measured real and reactive power flow through the regulator control circuit. Fig. 2 shows which side of the regulator will have the voltage held based upon these real and reactive flows.

Fig. 2: Voltage regulator control modes.

Fig. 2: Voltage regulator control modes.

Solar collector tracking system

Research and analysis of the kind of tracking system to be installed for tilt and angle adjustments of the PV panels is extremely important to most accurately develop solar insolation curves for all daylight hours throughout the entire year. There are dozens of publicly available databases containing this data. These references generally contain the actual insolation levels experienced throughout the day, and may also contain other valuable data such as the insolation levels for indirect sunlight. Fig. 3 illustrates the three most common types of tracking systems [2].

Fig. 3: Collector tracking systems.

Fig. 3: Collector tracking systems.

Feeder load and PV generation coincidence

Energy, demand, and load class shapes for the various customer classes were used to allocate load throughout the circuit for the minimum and maximum summer and winter demands from the most recent year. Presented in Fig. 4 are the historical demands for the summer minimum and maximum. Also shown are the peak insolation and therefore a peak generation profile to reflect a clear sky day according to data gathered at this site. The net minimum and net maximum load profile curves, which are the circuit load minus the PV generation, are also reflected in the figures. These net load profiles help to determine which periods to analyse.

Fig. 4: Feeder load and PV profiles.

Fig. 4: Feeder load and PV profiles.

Note the possible load on this circuit in the peak summer month may lie anywhere between the summer maximum and the net minimum curves. For the case study, the National Renewable Energy Laboratory (NREL) and typical meteorological year (TMY) databases were accessed to develop the insolation curve profiles for the case study. These databases contained data for both a fixed horizontal collector system as well as a two axis tracking system. Since this case study installation was to use a fixed tilt with one axis tracking system, interpolation between the two data sets allowed for creation of the insolation profile for this case study system. These profiles were used for investigation for worst case voltage issues since maximum reverse power export will occur during these periods.

Fig. 5: Net load (min. load – max PV).

Fig. 5: Net load (min. load – max PV).

Case study 1: PV effects on regulator LDC

As seen in the circuit diagram, this circuit traditionally deployed line drop compensation (LDC) on the line SVR. Prior to the PV integration, the worst case voltages occurred during the 18h00 at a circuit demand of 3600 kW. Therefore, this load level is used to calculate the base output voltage and LDC settings. The nominal voltage of 120 V is desired at the voltage regulating point load centre near the end of the circuit. The X/R ratio from the SVR to the load centre regulation point is approximately 1,2. This results in a
120 V setting with LDC of 12 V for X and 10 V for R. These established settings are then evaluated at circuit minimum load levels as well to ensure desired voltage levels are met. Introducing the DER generation contribution creates a desensitising effect on the LDC since the active power export from the PV plant creates a situation which reduces the R component of current, in turn, causing the regulator to operate on a lower tap potentially decreasing circuit voltage outside standards. This has a chance of occurring during peak insolation at 12h00 as the generation reaches 1 MW. Fig. 6 represents the voltage profiles to node 10 for minimum and maximum circuit load without PV and then again with the PV contribution. Violations occur since voltages fall outside the 118 V minimum operating constraints.

Fig. 6: Regulator LDC desensitising.

Fig. 6: Regulator LDC desensitising.

Case study 2: PV generation exceeding circuit load

As seen in Fig. 5, the potential for net negative demand on this circuit may occur anytime from 07h00 until 15h00 depending on the PV insolation levels and local load demands. Nonetheless, whenever reverse power flow occurs through the line regulator, other control modes need to be deployed to limit the potential of the substation and line regulators fighting to hold voltage between them. The line regulator has traditionally been set with a bi-directional reverse power setting for contingencies. As can be seen in the bi-directional mode in Fig. 2, during reverse power, the incoming substation side of the line regulator is held as the voltage control point. Since the substation SVR is also trying to maintain a set voltage at its downline side, each regulator is controlling the same “zone” thus excessive tapping will occur such that one or both will tap until its maximum or minimum tap is reached. Fig. 7 is an illustration of what will occur with the substation regulator set with 122 V, and the line regulator set with a reverse power setting of 122 V with no LDC. There is a slight voltage rise from the substation to the line regulator during a reverse power state, therefore the line SVR substation side controlled voltage point will then always be higher than the 122 V substation regulator setting, and will attempt to tap down or buck the voltage to its 122 V level. Since the substation SVR is holding the “rigid” voltage, 122 V cannot be maintained at both regulators in this intermediary zone. As a result, the line SVR will then tap up the required number of steps trying to achieve 122 V. Initially, if the line SVR is sensing more than 122 V, it will step to obtain its desired voltage, but will actually be raising the voltage on the downline PV side of the line SVR, just the opposite of what the control algorithm expects. This will continue until the regulator reaches its minimum step. In this case, the line regulator has a 10% buck/boost capability. Fig. 7 represents the profile for this case assuming enough time constants have occurred for both regulators to step through the necessary tapping sequences.

Fig. 7: Reverse power case with line regulator in reverse power mode.

Fig. 7: Reverse power case with line regulator in reverse power mode.

Case study 3: Regulator cogeneration mode

The cogeneration mode can be set at the line SVR to help eliminate operational issue between both voltage regulators as in case 2. Refer to the cogeneration portion of Fig. 2 and notice a cogeneration mode regulates the PV side, or downline side, of the regulator. With this set, the PV side of the regulator will be held as the control point during not only forward, but also reverse power flow. For this situation, a cogeneration voltage setting between 118 and 124 V will be applicable to hold the voltage at node 10 to within 118 to 126 V since the voltage swing from the line regulator to node 10 is 2 V. Fig. 8 illustrates the profile with a cogeneration mode of 122 V.

Fig. 8: Line regulator with cogeneration mode.

Fig. 8: Line regulator with cogeneration mode.

Case study 4: Intermittency, ramping and PV disconnect on faults

Also consider the ramping which occurs due to cloud shading. This is not an easy task to determine, since many factors such as cloud patterns, wind speeds, PV cell angle, azimuth tracking, plant land area, total surface area of PV arrays and shading between arrays convolute the task. These ramp rates ultimately need to be predicted or calculated to improve the accuracy in generation profile determination. Traditionally step voltage regulators or load tap changers have done the job of reacting fast enough to maintain satisfactory voltage on the system, but with the large generation amounts and quick ramp rates, these active regulation devices may not be able to adequately operate or catch up with voltage changes on the system. To calculate ramp rates, the cloud travel direction needs to be known in addition to the PV array length in parallel with the cloud direction.

Dividing the speed of cloud travel into the PV array linear distance is one simple way to develop a “back of the envelope” estimated rate of change. A reasonable ramp rate for this installation is 30 s and was developed through discussions with the PV plant developer, and also was considered a reasonable worst case according to industry research efforts when taking the effects of the physical site size for the array coverage area into consideration. The SVRs on this circuit traditionally deployed longer time delays in the 30 to 60 s range. Comparing that to a 30 s ramping possibility from full PV generation output to no generation, or vice-versa, indicates that this is comparable to a situation where the PV plant completely disconnects from the circuit due to a temporary line fault causing the substation recloser to trip followed by a disconnect of the PV plant.

Currently, this is a requirement of the IEEE 1547 standard [1] which is currently being scrutinised since the industry is finding low voltage ride-through and voltage regulation a benefit rather than a hindrance. Both summer and winter load and generation profiles were referenced to determine the worst case periods for both months in which the PV could be either disconnected or fully ramped producing excessive voltage flickers on the system.

Fig. 9: Feeder voltage change before and after system fault or full ramping event.

Fig. 9: Feeder voltage change before and after system fault or full ramping event.

Load flows for the following scenarios were investigated:

  • Peak load with maximum coincidental PV kW
  • Minimum load with maximum coincidental PV kW
  • Peak load with no PV
  • Minimum load with no PV
  • Peak PV at 12h00 with maximum coincidental circuit load

The process for each of the above was to run a load flow in order to find the tap position of the regulators for the steady-state condition. Next, the regulators tap positions were locked, the PV generation dropped, followed by a new power flow to find the voltage change. During reverse net power flow there will be a voltage rise along the circuit from the substation out to the PV plant location. During these periods, the line SVR is operating in a 122 V cogeneration mode with indicating that the voltage at the downline side of the regulator is being controlled to hold the voltages along the circuit to within standards. A temporary fault occurs, the substation recloser trips to clear the circuit fault, which is then followed by a loss of PV generation. After the substation recloser closes to restore power, the total circuit demand from the power supplier has now been increased by 1 MW.

During this new steady-state condition both voltage regulators on the circuit have not yet had a chance to respond to the change in load, and thus will remain on the same tap prior to the feeder disturbance. This can have a very dramatic effect on system voltage.

Fig. 9 shows the worst case voltage change according to the five scenarios listed above occurred during minimum feeder load with maximum coincidental PV generation. As expected, the voltage rise prior to the disturbance has now transformed into a voltage drop throughout the circuit thus producing a worst case voltage change of approximately 5 V at node 10. Furthermore, this voltage deviation which would be sustained until the regulator’s time delay has been met, is below the minimum voltage limit of 118 V.

Fig. 10: Inverters controlling voltage at POI.

Fig. 10: Inverters controlling voltage at POI.

Case study 5: Inverter voltage control mode

The previous cases were run with the PV inverters operating with a fixed unity power factor, in other words generating real power based on the amount of insolation only. Many inverters used in large PV installations have the capability to export and import reactive power almost instantaneously. The levels of export or import are based upon the control mode and settings of the inverters. For the case study, allowing the inverters to operate in a voltage control mode at the point of interconnection (POI) creates much more stable voltage profiles along the circuit. The worst case loading and thus voltage drop, neglecting any regulation, is about 9 V and occurs with the 3600 kW load and no PV generation.

Conducting power flows for this state and the other varieties of circuit demand levels coincidence with the PV generation levels concludes that the line regulators could be removed if the PV plant is allowed to control the voltage. Allowing the inverters to export 1 Mvar reduces the volt drop by about 2 V, therefore if the substation regulators are set to 125 V, the voltage at node 10 will be above the 118 V minimum. For this case, an inverter voltage control setting of 120 V was established at the POI on the distribution circuit. Fig. 10 represents the profile with inverters allowed to regulate voltage at 120 V.

Conclusion
Determining the potential power generation curves from PV at various times through the daylight hours and seasons is not a trivial task due to many dynamics. The inverter ratings, design and operating modes also need to be considered since they can vary greatly. Coupling this with the local load profile to develop a wide variety of net circuit load levels with and without generation is needed with follow-up to determine any issues with negative interactions between voltage regulation devices, which in turn will assist with the development of new control settings to accommodate the DER enhanced system. Examples were presented with traditional voltage regulator settings normally used for one-way steady-state load flows with only one source of normal utility supply, resulting in voltage problems when introducing intermittent generation, especially with reverse power flow during periods of peak insolation. Also demonstrated were the benefits of operating PV inverters in voltage regulating mode rather than the more commonly used unity power factor mode.

References

[1]    T Ortmyer, R Dugan, D Crudele, T Key and P Barker: “Utility Models, Analysis, and Simulation Tools”, US Department of Energy Renewable Systems Interconnection Series, 2008.
[2]    W Jewell and R Ramakumar: “The Effects of Moving Clouds on Electric Utilities with Dispersed Photovoltaic Generation”, IEEE Transactions on Energy Conversion, Vol. EC-2, No. 4, 1987.
[3]    IEEE Standard 1547TM-2008: “IEEE Standard for Interconnecting Distributed Resources with Electric Power Systems”.

Acknowledgement

This paper was presented at Cigré’s Canada conference in 2012 and is republished here with permission.

Contact Greg Shirek, Milsoft Utility Solutions, greg.shirek@milsoft.com

Related Articles

  • South African Government COVID-19 Corona Virus Resource Portal
  • Ministerial determinations propose 13813 MW of new-build by IPPs, none by Eskom
  • Crunch time for South Africa’s national nuclear company, Necsa
  • Dealing with the elephant in the room that is Eskom…
  • Interview with Minerals & Energy Minister Gwede Mantashe