Skip to main content

Demonstration of Essential Reliability Services by Utility-Scale Solar Photovoltaic Power Plant: Q&A

Webinar Questions & Answers

April 27, 2017

Is photovoltaic (PV) generation required to provide grid supportive services anywhere in the U.S. or worldwide? 

  • There are a few things that the federal government has established in terms of requirements: the plant needs to be able to provide voltage and frequency ride through capabilities that’s appropriate for the plant, and it needs to be able to provide reactive support.
  • If you look at the features that are required by NERC (the voltage regulation, real power control, curtailment capabilities, frequency regulation, droop response, short-circuit duty, fault ride through), these are features that we see many of the utility-scale PV plants, especially in the U.S., but also more recently in other places, are increasingly either encouraged or already incorporated in the LGIA agreements. In particular, we should be able to support up to 0.95 power factor, leading or lagging at the high side of the collector step-up transformer. 
  • A broader question might be, “are the ancillary services required?” I don't believe that these capabilities are contemplated as being required, however the effort in this particular webinar was to illustrate that these kind of capabilities, now that they are available, can be leveraged. This is especially true in cases where PV plants are being curtailed for other reasons to take advantage of them and in some ways reduce the constraint that only conventional generation plants can provide the services. 

Are the specifications of the inverter available publically?

  • In general, the utility-scale inverter manufactures that we work with has matured to a level where they are providing many of the capabilities that we leverage in utility-scale plants, in particular, the ability to respond to signals to control P&Q (active power and reactive power) in a fast manner. Increasingly other inverter manufacturers of utility-scale plants are certainly able to do so. Those specs are very generic in that sense.
  •  When we started this development about 6-7 years back, we had to work with the inverter manufacturers to establish a faster communication interface to respond much faster than they were required to in the past. The fact that we have control loops of 100 milliseconds means that the inverter has to be able to receive that signal and respond to that signal very rapidly. The specifications were being met by a small sub-set of the inverter manufacturers at that time. 

What is the power ramp rate limit of the plant? And what is the power ramp rate requirement for CAISO?

  • Utility scale plants can actually respond much faster than conventional generation. The challenge is to have the control system so that the plant responds at the rate that the grid operators desire. So under steady state conditions if the system operator requests a 10% (of maximum capability of the plant) per minute ramp rate, then we can slow down the response of the plant in order to achieve that rate.
  • Of course this control depends on the solar resources available. In other words, if the resource disappears suddenly, there's not much we can do. But as long as the solar resource is available, then we can certainly ramp up the plant really fast, much faster than any conventional generation, or we can slow it down to ensure that the grid does not see a huge impact due to the change. 

Have you used droop control for voltage regulation as well?

  • Yes. The voltage droop capability has already been demonstrated during previous years.
  • How can you justify PV curtailment with the AEP loss when providing AGC?
  • That is a tradeoff that the plant owners can make. There’s certainly a downside of leaving certain available power that’s not being utilized at any particular period of time. 
  • However, if the plant is already being curtailed for other reasons, then you could gainfully use that plant to provide these services while it is being curtailed. 

Do the inverters have any special feature to be able to do these controls?

  • In general, most utility-scale inverters are capable of providing P&Q (Power and Reactive Power) controls.  In general the switching frequency used by the inverters operates at a much faster rate, on the order of several kilohertz, so they do have the technical capability for speedy response. 

Have you tested the system in islanded mode?

  • In order to apply in islanded mode, you have to have grid-forming inverters and local load. That wasn't the purpose of this plant/project, which focused on a commercial utility-scale plant.
  • There are other projects at the smaller scale that look at PV as an energy supplier in a microgrid where this is done in many other places, but that's a different area of work.  

Which markets demonstrate the best approaches to making best use of the capabilities described here?

  • That’s difficult; right now we think the regulation market might be someplace to participate.
  • We also think that if the plants already being curtailed, you could participate in the market. If you know you're most likely going to be curtailed, say during certain hours, then you could participate to provide these services in certain markets. 

To provide frequency support services, do you have any idea how much money is lost for curtailment and how much is gained from ancillary service market?

  • We have not done that type of economic evaluation. On the CAISO website, we have prices for ancillary services. I think a developer could look at the prices, see if it makes sense, or what market makes sense to participate in. 

Can the power electronics and inverters be used for grid support at night? Or must there be power from the PV available to make those assets useful?

  • You need power electronics, inverters, and the controller, to provide grid support at night or more specifically provide reactive support or voltage control. It doesn't need any PV DC power for that. The reason is the reactive power is only relevant between the grid and the AC side of the inverter. The inverter can basically use grid power, and change its phase angle and send it back to the grid as reactive power minus whatever losses incurred in the inverter etc.  It works like a STATCOM device in this case.
  • This  capability will be a useful if a grid operator or the asset owner has need for supporting voltage or providing reactive power even when the PV plant is not generating any active power (such as at night time). 

How well designed are California's current utility PV grid connected systems at providing these reliability services? 

  • I think the grid is pretty well designed but we would have to look at places where there are operating challenges. In areas where we have high voltages at night, for instance, it would be useful to see if we have PV plants that could help control the voltage at night.
  • The first phase of this test was to really look at the plant's capability. The second phase now is to go back and evaluate the existing solar fleet to see, which ones have the capability to provide these reliability services that we tested. Later on this year we're going to have an actual PV plant participate in regulation service as a test. 

Does this 300-MW test plant have battery storage installed? 

  • No. The purpose of this test was to demonstrate the ability of PV technology itself.   

Considering the interaction of the inverters within the plant, what is the approximate time response of a 300-MW PV plant to an active step command from ISO?

  • If the question is about reactive power it can be changed in a couple of seconds. If the question is about active power, that is also pretty fast. See Morjaria, Mahesh, et al. "A grid-friendly plant: The role of utility-scale photovoltaic plants in grid stability and reliability." IEEE Power and Energy Magazine 12.3 (2014): 87-95. 

What was the maximum power rating for the inverter that has been used?

  • 4 MVA in the nominal rating. If it puts out zero reactive power then it will be 4 MW in general. If it put outs zero active power it can produce up to essentially 4 MVAR.  

It looks like the smart inverter is a critical part of these features to be utilized. Do you have any updates regarding smart inverter deployment? Any concerns about using smart inverters such as cybersecurity? 

  • We are talking about utility-scale plants, consisting of plant controls and inverters and not simply an individual inverter, which is but one element of the plant. Regarding the plant itself, we do have systems that provide an electronic fence around the plant, so there is, NERC CIP compliance capability as required especially for large plants. Cybersecurity is maintained in that context.
  • Perhaps a bigger question: if we start going to a smaller size/a smaller inverter, distributed PV, etc., how does it all work out from a cybersecurity perspective? I do not know. It's a significant challenge, perhaps not simply a technical challenge but an economic one as well.  

How does First Solar estimate the power available from the sun at any given time?

  • That needs some thought. Generally, we do know from the inverters themselves as to what power each one can produce at any time. They perform MPPT (Maximum Power Point Tracking) tracking. When we curtail the plant, we have a good idea of where and how much to curtail so we can use that information. There are other ways as well. We are currently working through those means to figure out what is the most accurate way. This refers only to instantaneous available power and not to what the power will be an hour from now. 

Is there a constraint on the number of inverters within a plant due to communication delays? What kind of communication is used between controller and inverters?

  • In general, we have not found that to be the limiting factor. The fiber communications is pretty fast and the 100-milisecond loops are fast, the communications piece is the least of our concerns. No limit found yet.
  • Is there a protocol?
  • We have used quite a few different ones but all are generally based on TCP IP type calls, and I think the sun specs establish a good standard. But before that we used a specific protocol that was pretty fast and has been used in industrial applications by companies that we found to be quite adequate.  

Do older plants need new inverters to do this kind of control?

  • I would say it's a function of what kind of inverters they are. That decision will depend on the features of the inverter. But generally, most modern inverters should be capable. 

Are you using PMUs to synchronize the two feeders 230 KV system?

  • No, we are not using PMUs to synchronize. Just referring to them as a way of collecting the data and that's all. Not used in any controls in the plant.  

Is there a study to understand impact of the modules themselves due to the curtailments?

  • We do worry about that in the sense that we need to understand the implication of operating modules under non-MPPT operating points. First of all even in normal operating conditions only a small fraction of energy falling on the modules is being converted into usable power. The rest heats up the module. If the inverters are completely turned off, for example, then essentially the modules are not producing any power and are in open circuit. They would typically experience elevated temperature which is the most stressful case. So partial curtailment is somewhere in between. In any case, we test for all of those aspects as a module manufacturer, and in general, we do not see any issues associated with curtailment or being open circuit conditions. 

How is regulation accuracy determined, and what accuracy determined a provision of fast ramping or other services, or how would that be determined? 

  • We do have documentation on CAISO website as to how we do this calculation, but essentially we issued a 4-second dispatch instruction and then we go back and look and see how well that user was able to follow that instruction. So we accumulate that across. I think it's a 15-minute time frame and then across an hourly time frame and then looking at the accuracy following that dispatch signal. There's a document out on the CAISO’s website that goes into how we do this accuracy calculation.

While active/reactive powers are independently controllable at an inverter level, is it correct that active and reactive control is also at the plant level? 

  • The objective of the plant controls is to control the output of plant i.e., active (P) and reactive power (Q) at the point of interconnection of the plant. We are basically controlling each of the inverters so that both active and reactive power aggregated over the whole plant minus all the losses in the plant give us the desired quantities. So, yes both active and reactive powers are being controlled independently at the plant level and at the inverter level as well. However, there are limits to the extent to which we can control P&Q independently since P&Q cannot exceed the aggregated rating of the equipment (e.g., MVA rating).                                                                                                                                              

If a PV plant is providing controlled curtailment, doesn't that need to be included in the PPA provisions and do you see a trend in this happening?

  •  In the PPA, we do have some curtailment provisions, but there are times that you have oversupply for example. You will have to curtail. There are times when, if you have congestion on the system, or when you have any real time problems, you're going to have to curtail resources and that may be any type of resource including PV plants. So, not necessarily during the curtailment provisions that we have.  

What is the difference of AGC and fast frequency response?

  • AGC (Automated Generation Control) are centralized commands sent by ISO. Fast frequency response is a local service like primary frequency control that does not depend upon the signal from ISO. 

How is the array voltage controlled when the output has to be curtailed?

  • The inverter under curtailment moves along the IV curve of the DC Array to reduce the DC power extracted from the array (away from MPPT point). Typically it increases the DC voltage while reducing the DC current resulting in lowered output. At all times it ensures that the DC voltage is within the design limits and operating limits. 

How do the ancillary services provided by this configuration compare to on-site battery storage systems?

  • It is similar since the battery storage systems also have inverters that essentially function the same way as a PV inverter. 

Which software would the presenters recommend to be used for modeling reactive power control for PV integration in the generation end?

  •  PSLF, PSS/E are typically what we use. 

What is the fault response behavior of the plant? Adhering to PRC-024?

  • Yes the fault response adheres to PRC-024 (at least where it is applicable) 

Have you tried angle droop control also?

  • No 

Could you explain with more detail the control of reactive power in PV power plants?

  • The inverters can produce both active and reactive power within its capability. Once the controller determines the amount of reactive power required at POI, it figures out how much reactive power required from each inverter and in a closed loop system manage the total reactive power at POI. See Morjaria, Mahesh, et al. "A grid-friendly plant: The role of utility-scale photovoltaic plants in grid stability and reliability." IEEE Power and Energy Magazine 12.3 (2014): 87-95

What type of control mechanism has been implemented, and did they use the central inverter? What was the single string power rating?

  • Central inverter of 4 MVA nominal size were used in this project. Please refer to Morjaria, Mahesh, et al. "A grid-friendly plant: The role of utility-scale photovoltaic plants in grid stability and reliability." See IEEE Power and Energy Magazine 12.3 (2014): 87-95 for an explanation of the control mechanism. 

Are there any reference projects regarding ramp down restrictions and storage integration?

  • We are not aware of such restrictions.

 In the NREL speaker's presentation, the "inverter capability" showed a unit circle VAR capability, however, it appeared that the actual outputs are limited to a "V" shape output dependent on plant MW output. Can no load full VAR output be achieved?  Can you comment on this?

  • The PV inverter is theoretically capable to achieve close to circular reactive power capability. The V shape is actually a draft reactive power requirement by CAISO. The PV inverters can obviously deliver more reactive power.  

 Is the forecasting capability there for what you need to bid on to markets or operate a fleet without needing as many MW in reserve?

  • The forecasting capability gives the developer an idea of how much capacity can potentially bid into the market.

 Any observations on total harmonic distortion of the plant under different conditions?

  •  The voltage THD was not measured during the testing since power quality issues were outside of the scope. 

 Does the model for BTM residential ReE take into account the ZNE strategy for much smaller PV kW arrays? If so, and if the current rooftop PV average kW is around 6 - 7, what do you think the average size kW array will be?

  • This question is outside the scope of the test.

 Any future plans to demonstrate synthetic inertia capability of a large PV plant?

  • Yes, there are future plans to demonstrate synthetic inertia capability of utility-scale PV plants in near future. NREL and First Solar will be conducting such testing under the on-going research collaboration

 Shouldn't the time to Point B be around 52 seconds after the disturbance after the nadir, with regard to primary frequency response? The graph showed the response over a period of 500 seconds.

  • The point B or settling frequency is calculated as an average frequency for the 20-52 sec time interval after the disturbance.  The primary reserves are expected to be fully deployed by this time to stabilize the system frequency at some steady state value.  In this case, we used about a 10-minute time series of actual grid frequency measured in the Western Interconnection to demonstrate the droop response by the PV plant immediately after the frequency event and during frequency recovery.  As frequency continues to recover after t=52s, the output of the plant decreases according to the droop, until it comes back to the pre-fault set point. 

 What if any effect does a PV Utility Scale Plant have on rate of change of frequency following a disturbance? 

  • The impact on initial rate of change of frequency during contingency is not expected to be significant with a PV plant providing only droop response (response to frequency itself but not rate of change). However, with special controls that are responsive to rate of change of frequency (ROCOF) the PV plants can also have impacts on ROCOFs if necessary.  

 I am wondering when NREL plans to test the ability of PV to provide voltage support at night, and whether there are plans to test the ability of rooftop PV to provide any essential grid reliability services.

  • Yes, NREL is planning to test the ability of utility-scale inverters to provide voltage support at night. For rooftop PV, there are several ongoing NREL projects including one in Hawaii where rooftop PV inverters will be providing various grid support functions.