NYISO Climate Change Impact Studies

The New York Independent System Operator (NYISO) manages New York’s power grid and wholesale electric markets.   That responsibility not only includes the day-to-day management but also extends to long-term planning.  As part of the latter charge NYISO commissioned two studies of climate change impacts on power system reliability in New York.  While the studies provide valuable information, I think further work is needed before we can be assured that solar and wind resources will be sufficient to meet load requirements.

I have two degrees in meteorology, am a retired certified consulting meteorologist accredited by the American Meteorology Society, have over 45 years experience as a practicing meteorologist, and have been working in the electric utility business since 1981.  The contents of this post are based on that background and experience.  The opinions expressed in this post do not reflect the position of any of my previous employers or any other company I have been associated with, these comments are mine alone.

Background

In order to assess the potential impacts on power system reliability in 2040 associated with system changes due to climate change and policies to mitigate its effects, NYISO contracted with ITRON and the Analysis Group. In today’s New York it is necessary to address the political presumption that the effects of climate change are being felt today so a primary goal was to address that concern.  New York’s Climate Leadership and Community Protection Act (CLCPA) has targets for decreasing greenhouse gas emissions, increasing renewable electricity production, improving energy efficiency and an aggressive schedule as I have documented in CLCPA Summary Implementation Requirements.  Both studies also addressed the effects of this climate policy on the future electric system.

Itron developed long-term energy, peak, and hourly load projections that address the potential effect of climate change and the CLCPA. According to an Itron blog post that report identified “historical weather trends across more than 20 weather stations in New York State”. That information was used to drive system and planning area load models. They noted that “complicating factors include continued growth in behind-the-meter solar generation, increasing proliferation of electric vehicles and state policy to address climate change through electrification”.  The final report included two long-term hourly zonal-level load forecasts that reflect state policy goals and climate effects.

In the second phase the Analysis Group used the Itron load forecasts to evaluate system impacts and develop a climate resiliency plan.  According to the Executive Summary in the draft Climate Change Phase II Study, the “Phase II Study is designed to review the potential impacts on power system reliability of the (1) the electricity demand projections for 2040 developed in the preceding Climate Change Phase I Study, and (2) potential impacts on system load and resource availability associated with the impact of climate change on the power system in New York (“climate disruptions”). The NYISO Electric System Planning Working Group meeting on September 10, 2020 included a presentation by the Analysis Group that gives a good overview.

Climate Change

The original intent of these projects was to consider the effects of climate change on the electric system.  Iton claims that their forecasts “reflect the potential continuation of such weather trends during the next 30 years” corresponding to the implementation period of the CLCPA 2050 target.  Analysis Group considers potential impacts of “climate disruptions” on the electric system.  However, I think their projections actually represent something else.

Contrary to popular opinion, teasing out the effect of climate change presumed to be inextricably linked to GHG concentrations from natural climatic variation is a controversial topic in the meteorological community. The Analysis Group climate disruptions “include items that could potentially occur or intensify with a changing climate and that affect power system reliability, such as more frequent and severe storms, extended extreme temperature events (e.g., heat waves and cold snaps), and other meteorological events (e.g., wind lulls, droughts, and ice storms).”  Invariably in my experience a purported climate signal is, in reality, just a weather extreme.  All these “climate disruptions” fit that bill.

Bottom line is that while both studies provide valuable information the projections represent extreme weather more as a result of natural variability than any climate effect due to global warming.  The key point is that these weather impacts have to be considered to adequately represent future load.  The fact that I consider the climate change signal small compared to natural weather variability is irrelevant for the results.

Analysis Group Renewable Resource Approach

While I applaud the results provided by the Analysis Group, I don’t think it should represent the final word on the effect of weather on wind and solar resource availability.  I will explain my problems with what they did and offer my suggestion for what is needed below.

The Analysis Group estimated what electric generating resources will be necessary to meet the projected loads predicted by Itron. The primary goal was to estimate the generating and transmission infrastructure necessary to meet the CLCPA 2040 target to eliminate the use of fossil fuels for electricity generation.  Importantly, the emphasis was on the viability of a resource mix to meet this target and they repeatedly point out that their estimate is just one of many possible pathways to the goal.  Their electric system modeling is described in a recent presentation.

The draft report explains that there are three core elements to the modeling approach.  The first element is the load forecasts from the Phase I study.  The second element is the development of resource sets for two scenarios representing the climate change impacts and inputs from another NYISO study on the grid in transition.  The starting point for the resource allocations was earlier NYISO work based on New York’s announced procurement goals.  “This resource set alone is insufficient to meet demand; thus, the analysis adds renewable generating capacity, storage capacity, transmission capability, and Dispatchable Emission-free (DE) resource capacity in quantities sufficient to meet the seasonal peak demand.”  My primary interest is the third core element: “Climate Disruption Scenarios”.

According to the final report:

 “These climate disruptions are used to define seasonal ‘cases’, which are run through the energy balance model to identify any reliability risks associated with operations under those conditions. The results of the model identify the magnitude, frequency and duration of any periods where available generation was potentially insufficient to meet load over the duration of the seasonal modeling period, or where significant storage or DE resource output is needed to supplement renewable generation.”

The report developed these extreme-weather or physical disruption events to simulate conditions that “increase demand and/or reduce or eliminate the availability of renewable resources and transmission infrastructure.”  Table 12 Description of Physical Disruption Modeling Events from the draft Phase II study lists ten types of events that could physically disrupt the electric energy generation system in 2040 when it is strongly dependent upon wind and solar resources.  I will focus on the treatment of meteorological inputs on solar and wind output for these events below.

The biggest single weather factor on load is temperature.  Heat waves and cold snaps are the primary cause of peak loads.  In this analysis the meteorological conditions for these temperature extremes were adjusted as follows:

“Heat waves are modeled using the following model adjustments:

• • • • Load ‐ High temp 90° F or above for seven days, with daily zonal load increase of between 0 percent and percent 18.7 percent
      • Wind Generation ‐ 20 percent decrease for seven days
      • Solar Generation ‐ use solar profile from hottest day in Y2006 for seven days
      • Transmission ‐ five percent decrease for seven days

Cold waves are modeled using the following model adjustments:

• • • • Load ‐ Low temp of 0° F or below for seven days, with daily zonal load increase of between 2.3 percent and percent 25.6 percent.
      • Solar Generation ‐ Use solar profile from coldest day in Y2006 for seven days”

Three wind “lulls” physical disruption events were evaluated: just Upstate, just Off-shore and state-wide.  To evaluate potential variability, Analysis Group evaluated historical National Renewable Energy Laboratory (NREL) daily wind data from 2007 to 2012 to estimate the wind generation output.  Three sites representing upstate and offshore production were used: Niagara, Plattsburgh, and the offshore Empire Wind Zone.  The analysis found 19 wind lulls in the summer and only three in the winter.  In order to evaluate the effects on loads they adjusted the high load periods developed in Phase I as follows:

“Summer wind lulls are modeled using the following model adjustments:

• • • • • Wind Generation ‐ 15 percent Average Capacity Factor in all Zones for 12 days
        • Wind Lull overlaps the 12‐day period with highest load

Winter wind lulls are modeled using the following model adjustments:

• • • • • Wind Generation ‐ 25 percent Average Capacity Factor in all Zones for seven days
        • Wind Lull overlaps the seven‐day period with highest load”

I am not going to spend much time commenting on the remaining five disruptions considered.  The analysis considered four storm events: hurricane/coastal wind storm, severe wind storm upstate, severe wind storm offshore, and an icing event.  In all the cases they simply made assumptions about how the load, wind and solar resources would be affected and times for recovery.  The final disruption was a drought and that was assumed to reduce hydro output 50% for 30 days.

Critique

My primary concern as a meteorologist is the availability of renewable energy resources.  The question is just how much wind and solar energy is potentially available every hour.

According to the Analysis Group final report

“The generation profile, in terms of hourly capacity factors, assumed for the solar units are based on 2006 data from the NREL Solar Power database using 62 simulated solar farm sites across New York State. Two Zones did not have solar farm data. For Zone D BTM solar, a simple average of bordering Zones F and E was used. For Zone K utility solar, the BTM solar data from Zone K was uprated by the average ratio of utility to BTM solar NYCA‐wide.  The hourly capacity factors assumed for the wind units are based on 2009 data at simulated 100 meter turbine height from the NREL’s Wind Toolkit Database, using 721 weather sites in NY. A summary of renewable resource capacity factors by season is listed in Table 6. As shown, solar capacity factors are higher on average in the summer modeling period than in the winter, and wind capacity factors are higher on average in the winter than in the summer.”

The NREL Solar Power database consists of one year (2006) of 5-minute solar power and hourly day-ahead forecasts for approximately 6,000 simulated PV plants including 62 in New York.  NREL generated the 5-minute data set using the Sub-Hour Irradiance Algorithm that produces global horizontal irradiance (GHI) values.  The sub-hour algorithm produces “coherent sub-hour datasets that span distances ranging from 10 km to 4,000 km”. The algorithm “generates synthetic GHI values at an interval of one minute, for a specific location, using SUNY/Clean Power Research, satellite-derived, hourly irradiance values for the nearest grid cell to that location and grid cells within 40 km”.   Combining satellite cloud data and a probability distribution it estimates one of five cloud classifications which are used to generate the solar irradiance value.

In my comments on the resource adequacy hearing and elsewhere I have argued that actual short-term meteorological data must be used to correctly characterize the renewable resource availability for New York in general and in areas downwind of the Great Lakes in particular. This is because the lakes create meso-scale features, most notably lake-effect precipitation and clouds, that can affect solar resources many miles from the lake shore. It is important that the solar resources be evaluated based on geographically representative short-term data and I do not believe that the NREL approach adequately addresses this concern.

On the other hand, their approach for wind data is acceptable.  They have more stations included and wind speed fields are generally well connected as opposed to discontinuous lake-effect clouds.  As a result, the data used are adequately representative.  However, there is a problem with the Analysis Group physical disruptions analysis.  They only looked at light wind disruption of wind energy output.  Because wind turbines have a high wind speed cutoff there could also be reductions if the winds are too fast.

Finally, there is a major flaw in the approach.  Analysis Group makes assumptions about the effects on wind and solar output for each physical disruption on its own.  In reality a study that considers the joint distribution of wind and solar energy impacts from weather events is needed.  This isn’t even possible using the NREL data sets they used because they are for different years.

I did my own analyses of the renewable resource availability for two short periods using observed data for summer peak energy storage requirements and winter peak energy storage requirements. My guesses for the generating resources were extremely crude but I think the approach should be the next step check on the feasibility of renewable resource dependency.  In particular, I used historical meteorological data and estimated wind and solar output relative to observed load for the same time periods.

When I started my analysis, I expected that the winter observed peak load would occur during very cold weather associated with a slowly moving high pressure system that originated in the cold northern plains large enough to cover the entire northeastern US.  The resulting multi-day period of clear skies, light winds, and inherent cold temperatures would result in very high energy demand for heating at the same time the wind resource was weak.  In my example high load period in early January 2018 conditions were very different.  Weather maps for this period show (January 2018 Weather Maps) a relatively small high-pressure system in the central US on January 2 that moved east ahead of a storm system on January 3.  The high pressure was strong enough over the New York offshore wind region that winds were less than 3.5 m/s for five hours on January 3.  However, the storm system moved eastward and re-developed into a strong storm just off the coast on January 4 with an eleven-hour period of greater than 25 m/s wind speed 13 hours after the light wind period ended.  By January 5 the storm had raced northeast to the Canadian Maritimes but was pumping cold air back across New York State.

This period shows why actual data must be analyzed in more detail by New York State to determine whether the CLCPA requirements endanger fuel and energy security.  The actual solar irradiance irrespective of cloudiness was low in this period because it was near the winter solstice.  I assumed that the wind turbine low speed cutoff was 3.5 m/s and the high speed cutoff was 25 m/s.  If the assumptions I used for no wind power due to light winds and strong winds are correct then there will be 16 hours of no wind power in a 29-hour period during the coldest extended duration cold weather event that the Analysis Group identified after analyzing 25 years of data.  Furthermore, this period also overlaps fourth worst 3-day cold snap.

 Conclusion

The Itron Phase I and Analysis Group Phase II climate change studies provide valuable results and address my worries about the meteorological impacts on renewable energy resources.  However, I don’t think they go far enough to answer my fundamental concern that wind and solar energy might not be sufficient to power the state during the winter peak.

In my comments on the resource adequacy hearing and elsewhere I have argued that actual short-term meteorological data must be used to correctly characterize the renewable resource availability for New York in general and in areas downwind of the Great Lakes in particular. This is because the lakes create meso-scale features, most notably lake-effect snow and clouds, that can affect solar resources many miles from the lake shore.  In my opinion as a meteorologist living downwind of Lake Ontario, I don’t think the output from any cloud modeling approach has enough resolution to adequately simulate lake-effect clouds.  Therefore, the solar and wind resources should be evaluated using geographically representative short-term data so that site-specific temporal effects can be included.

I strongly recommend that meteorological data available from the NYS Mesonet meteorological system be used to determine the availability of wind and solar energy over as long a period as is available. The NYS Mesonet is a network of 126 weather observing sites across New York State so it can provide representative data for this kind of analysis.  If historical meteorological data are used to estimate solar and wind output against the observed load, suitably adjusted for climate and climate policy, then it will be a much better test than using the assumptions made by the Analysis Group to estimate how the meteorology might affect renewable output.