Carbon-Free Resource Integration Study

The Pacific Northwest National Laboratory study estimated how much solar the Duke Energy Carolinas and Duke Energy Progress systems could support in total based on assumptions that solar could not be curtailed (except for emergencies, which was the case at the time of that study). The PV generation evaluated ranged from 673 MW to 6800 MW (2% to 20% of peak load).

In contrast, the NREL study estimates Duke Energy Carolinas and Duke Energy Progress’ total percent of carbon-free energy produced, when accounting for escalating curtailment levels, by modeling PV at levels up to 35% and scenarios that include wind, storage, and nuclear. In Phase 2 of the analysis, NREL will incorporate capacity expansion results as well as production cost modeling, which the Pacific Northwest National Laboratory study did not do. It will also expand on other scenarios of carbon-free resource generation.

In Phase 1, NREL used in-house research and analysis tools as well as commercially available tools. For Phase 2, the intent is to use NREL's Regional Energy Deployment System Model (a national capacity expansion and dispatch model) and PLEXOS, a commercially available production cost model.

This report uses 2019 forecasted annual load and solar PV time-series profiles supplied by Duke Energy and based on the same weather period to ensure the solar profiles are synchronized with the weather assumed in the load. Duke also provided capacity rating for nuclear, pumped storage, and other thermal generation. 

The initial focus on the Carolinas is due to the significant amount of solar PV Duke Energy has connected to its Carolinas system. This amount is projected to double by 2030. As of July 2019, Duke had approximately 3,700 MW of solar connected across its jurisdictions, and more than 90% (approximately 3,300 MW) was concentrated in the Carolinas. 

The report does not consider an economic dispatch model. Rather, it evaluates the quantities of overgeneration if renewables are must-take and the impact on total carbon-free generation if energy storage is added.

NREL will provide the analysis for production costs in Phase 2. It will be conducted in three steps, or modeling approaches, which are interdependent.

• First, we will run a locational resource potential model (also known as reV or renewable energy model) to conduct a detailed assessment of renewable energy resources and their geospatial intersection with grid infrastructure and land use characteristics.
  
  
• Second, using the resource potential and costs developed with reV, ReEDS will be used to propose generation and transmission investments under a reference case and low-carbon case that maximizes carbon-free resource integration.
  
  
• Then we can run a production cost and operational feasibility model to simulate and analyze the resulting operation of the generation build-out and transmission upgrades identified in ReEDS. Examples of outputs include things like fuel costs, O&M, capacity factors and transmission utilization.

• The study does not consider cost, hourly reliability, or transmission capability. It is not intended to replace Duke's integrated resource plan, provide a roadmap for how to meet carbon goals, or serve as a financial forecast.
  
  
• The report uses 2019 forecasted annual load and solar PV time-series profiles supplied by Duke Energy and based on the same weather period to ensure that the solar profiles are synchronized with the weather assumed in the load.
  
  
• The study evaluates a variety of solar power penetration levels in Duke Energy’s Carolinas service territory, compared to load and systemwide minimum generation levels, that best represent potential challenges and opportunities for renewable generation integration (e.g., balancing solar and load for typical days during different seasons and extreme days, such as minimum and peak net load days).
  
  
• The study evaluates North Carolina and South Carolina and both balancing authorities (Duke Energy Carolinas and Duke Energy Progress); however, for scenarios 1 to 11, no restrictions between the balancing authorities are modeled and in Scenario 12, they are modeled separately with interconnection limits between them. All scenarios assume no interaction with outside markets.
  
  
• Net load is defined as the customer load less solar power and wind power generation. The analysis compares estimated hourly solar, wind, net load, and system minimum generation time series for the different scenarios.
  
  
• Solar is assumed to be nondispatchable, though utility solar can be curtailed down. 
  
  
• Generation flexibility limit consists of nuclear, hydropower, and must-run units offset by the hydropower pumped storage capacity. Nuclear is assumed to run at 100% capacity for this analysis.
  
  
• Must-run units have hourly triggers and therefore could change intra-daily, whereas hydro schedules vary monthly. This explains why the flexibility limit line could change seasonally and possibly daily.

The Phase 1 study evaluated 12 scenarios, with 7 focused on larger-scale solar at different penetration levels (from 5% to 35% by energy with an increase of 5% percent in each scenario) and another scenario focused on increasing the proportion of behind-the-meter solar. The other scenarios included additional changes to the carbon-free generation mix (along with high PV penetration), such as adding more storage, adding wind, or retiring a single nuclear unit.  While the first 11 scenarios assumed a single balancing area, the final scenario assumed separate balancing areas at existing transmission constraint levels and various levels of PV penetration (5%, 10%, and 15%). Phase 1 does not consider system cost implications and does not put a specific timeline on reaching scenario levels of penetration.

In scoping Phase 1, NREL, in consultation with Duke Energy, designed the scenarios to be considered. Collectively, the team believed it was important to be comprehensive, so the scenarios considered how different types of generation variables contributed to carbon-free generation during different times of year. Because one of the primary goals is to understand how high levels of solar PV contribute to a carbon-free future, 8 of the 12 scenarios evaluate how these different “solar only” levels perform. The other scenarios introduce other variables, such as additional storage, nuclear retirement, additional wind, and a difference in balancing authorities.

North Carolina and South Carolina have a large solar resource, which is why we looked to that resource first to achieve carbon reduction. Based on cost-optimal capacity expansion results from ReEDS, we will incorporate a scenario of other combinations of renewables and storage resources in Phase 2.

All scenarios reflect the existing energy storage capability of Duke Energy Carolina’s pumped hydro storage.

Scenario 9 captures the effect of an increase in storage with 25% solar energy penetration and demonstrates how this additional technology resource might reduce the curtailment required in a high-solar-penetration scenario. The scenario modeled is 1,000 MW of 4-hour storage, 1,000 MW of 6-hour storage, and 2,000 MW of 8-hour storage. These different hourly ranges reflect observations from another recent NREL study that increasing durations of storage are needed as more storage is deployed to serve increasingly wider portions of the demand curve, which aligns with Duke’s observations from prior energy storage studies. Based on EIA-860, the storage volume studied is roughly four times the amount that will be installed nationwide by the end of 2019. This level was selected to be large enough, relative to the 25% PV penetration case, to show significant impact when shifting PV out of curtailments hours, though the remaining curtailment is still high.

The net load analysis methodology used here does not discern other battery storage value streams, since it is an energy-only evaluation.

NREL and Duke used one balancing authority to simplify analysis for most scenarios in Phase 1. This allowed us to focus on the impact of resource mixes separate from power delivery constraints across authorities.

Another simplification is that the model does not account for possible imports or exports to the Carolinas. The variations of Scenario 12 investigate the power delivery constraint more, though, and examine how a limit between the balancing authority can impact carbon-free generation and how siting of resources in one balancing authority versus another also influences total carbon-free generation.

None of these scenarios are necessarily indicative of future planning for Duke.

As the amount of solar increases, the flexibility limit of the system is reached. This limit is the point at which additional PV production must be curtailed because of the amount of carbon-free energy on the system relative to load, ultimately diminishing the carbon benefit of additional solar. The benefit diminishes more rapidly at 15% through 35% penetration.

• In these scenarios, the average annual percentage of load met by carbon-free generation ranges from 60% to 79%.
  
  
• The scenario with the highest share of carbon-free generation is one with resource diversity; 30% solar and 5% wind results in 79% carbon-free generation.
  
  
• Due to Duke Energy’s nuclear energy fleet, nuclear energy is the greatest contributor to carbon-free energy, even with increasing levels of solar PV.
  
  
• Most of the solar curtailment occurs in the spring and fall seasons, and the days with the highest percentage of solar curtailment occur in the spring for both Duke Energy Carolinas and Duke Energy Progress.
  
  
• Additional storage can reduce curtailment in a high-solar-penetration scenario; however, another recent NREL study showed that as more storage is deployed, the minimum load and peaking events it serves become longer and flatter. As storage is added, the economics for additional storage become more challenging as progressively longer storage durations are required to serve wider parts of the demand curve.
  
  
• Solar power curtailment is greater when Duke Energy Carolinas and Duke Energy Progress are studied as separate balancing authorities.

They are different metrics with distinct but related scopes. Duke Energy’s climate goals are enterprise-wide goals for Duke Energy, while this report was designed to only analyze the Carolinas.

The NREL study does not analyze resources that are not carbon-free (and assumes they will backfill whatever deficits the system has after the carbon-free resources) and measures the total amount of carbon-free generation, as opposed to a carbon-reduction target. In Phase 2, NREL will evaluate the potential investments and operational decisions required to reach carbon-free energy resource integration at higher levels.

No. It is evident from the study that even with a balanced penetration of PV and storage, there is a point at which adding more PV + storage has diminishing carbon benefit. The cost-optimal expansion plan in Phase 2 will identify the estimated resource mix and location to maximize the value from that investment. To achieve further increases of carbon-free energy beyond that point will likely require new technologies.

Both planned and forced outage rates and their impact on unit commitment and dispatch will be studied extensively in Phase 2. However, in this preliminary net load analysis, we have assumed nuclear runs consistently at full capacity and has no outages. To get a glimpse of what the results would look like with nuclear outages, Scenario 10 in Phase 1 considers 10% of nuclear power is retired and assumed replaced with flexible thermal generation.

Although this is a high-level preliminary analysis, it does give some useful insights for Phase 2. For instance, it shows the impact of resource diversity on variable generation curtailment and at what percentage of variable generation penetration Duke’s system might begin to experience significant curtailment. This will allow the team to quickly home in on interesting scenarios. While Phase 1 has also established a data foundation for Phase 2, the assumptions used will not necessarily translate over into Phase 2.

Demand-side management programs will be incorporated into Phase 2 modeling and analysis. Phase 1 did not quantify the effect of such programs.