TODAY’S STUDY: WHAT UTILITIES ARE PLANNING

Survey of Western U.S. Electric Utility Resource Plans

Jordan Wilkersona, Peter Larsena, and Galen Barbose, January 2014 (Lawrence Berkeley National Laboratory)

Abstract

We review long-term electric utility plans representing ~90% of generation within the Western United States and Canadian provinces. We address: what utility planners assume about future growth of electricity demand and supply; what types of risk they consider in their long-term resource planning; and the consistency in which they report resource planning-related data. The region is anticipated to grow by 2% annually through 2020—before Demand Side Management. About two-thirds of the utilities that provided an annual energy forecast also reported energy efficiency savings projections; in aggregate, they anticipate an average 6.4% reduction in energy and 8.6% reduction in peak demand by 2020. New natural gas-fired and renewable generation will replace retiring coal plants. Although some utilities anticipate new coal-fired plants, most are planning for steady growth in renewable generation over the next two decades. Most planned solar capacity will come online before 2020, with most wind expansion after 2020. Fuel mix is expected to remain ~55% of total generation. Planners consider a wide range of risks but focus on future demand, fuel prices, and the possibility of GHG regulations. Data collection and reporting inconsistencies within and across electric utility resource plans lead to recommendations on policies to address this issue.

click to enlarge

Introduction

Electric utility resource planners’ decisions affect all residential, commercial, and industrial customers. Planners must decide how to meet future demand with limited information about future fuel prices, economic conditions, technology advancements, and governing policies.Assessing the risk of not meeting demand is essential to the planning process. Not surprisingly, load serving entities1 (LSEs) typically develop their plans for meeting future demand over the course of several years. The long-term planning process involves many stakeholders and can be computationally intensive. Many utilities are required to publicly-release and defend their integrated resource plans (IRPs) in front of consumer advocates, Public Utility Commissions (PUCs), and other stakeholders.

This study is a broad comparison of resource planning content and an aggregation of the collective forecasts of LSEs operating within the Western Electricity Coordinating Council (WECC) region. We review publicly-available planning information for nearly 40 utilities that, in aggregate, generate ~90% of the electricity in WECC. Since many of the resource plans are more than a year old, we also sent a supplemental survey to resource planning staff to give each an opportunity to update their load and resource projections. Most responded with updated information, including a few for which we could not locate plans. The results presented in the following sections are based on the best available information from LSEs as of August 2012. We conducted this analysis in order to gain insight into the following questions: (1) What are Western electric utility planners assuming about the future growth of electricity demand and mix of supply- and demand-side resources? (2) What types of risk do Western electric utilities consider and address in their long-term resource planning? (3) How does the collection and reporting of resource planning-related data differ across this region?

We report aggregate future demand and power plant fuel mix trends, identify the uncertainties LSEs focus on as they develop their IRPs, and report on emerging trends considered by planners. Reporting differences are a reflection of differing state reporting requirements, and these inconsistencies affect our ability to compare some planning assumptions. Accordingly, the availability and consistency of planning information is a focus of this analysis.

This paper is organized as follows. Section 0 provides a brief review of previous IRP surveys. Section 0 describes important steps in the planning process. Section 0 describes the data and methods we use to compare IRPs. In Sections 0‒0, we compare planning assumptions as we address the questions above. We conclude with suggestions that could improve inter-comparison and ultimately lead to more efficient long-term regional planning…

click to enlarge

How consistent is the collection and reporting of resource planning-related data across the Western United States?

One consistent trend we discovered in this review of resource plans is that critical planning assumptions are inconsistently collected and reported by the LSEs. In this analysis, we focus on a subset of long-term planning assumptions for the purpose of demonstrating a common set of issues associated with resource planning data. For example, we did not focus on supply-side assumptions related to capacity factors or capital costs. This is a reflection of differences in local, state, and regional reporting requirements—and these inconsistencies affect our ability to compare many planning assumptions. In this section, we highlight the availability (and consistency) of planning information to better frame the subsequent results of our study.

All 34 IRPs we reviewed included some level of detail about current resources and most identified a preferred or recommended portfolio of future resources. Six of the 34 LSEs we studied did not respond to the supplemental survey (PSCo, SDG&E, WMPA, Grant, PRPA, and EWEB). However, the survey allowed us to collect generation information from four additional LSEs for which we did not find publicly-released plans (Deseret, Basin, CAZ, and Alcoa)…

What are Western U.S. electric utility planners assuming about the future growth of electricity demand and supply-side resources?

click to enlarge

Demand-side assumptions Energy and Load forecasts

Three of the 34 IRPs we evaluated, did not provide annual energy (GWh) forecasts and one did not provide a peak demand (MW) forecast. The remaining 31 LSEs reporting this information represented 83.2% of WECC in 2011. If we assume these 31 LSEs will continue to represent the same share of WECC, then the region is expected grow from 857 TWh (2011) to 1,011 TWh by 2020.

Peak power represents a single hour each year when an LSE anticipates the highest demand. However, this event does not occur at the same time of day or season for all LSEs. Annual peak load occurs in the afternoon during the hottest summer day for many western utilities; others report peak loads during the winter. Consequently, we compare load shapes using load factor (LF), which is the ratio of average demand (GWh/8760h) and peak demand (GW). DSM forecasts DSM programs, often considered alternatives to supply-side resources, reduce the amount of energy and peak load an LSE will need to meet in future years. Energy efficiency (EE) programs seek to reduce overall consumer energy consumption replacing inefficient components. If the saved energy is coincident with peak demand, then the utility will also see a reduction in peak power from EE programs. Demand Response (DR) programs seek to shift demand away from the system peak demand, but will not likely reduce energy consumption and will only reduce demand if the DR contract is activated…

click to enlarge

What types of risk do Western U.S. utilities consider and focus on in their long-term resource planning?

LSEs are confronted with many uncertainties and risks in their attempt to meet consumer demand in a cost-effective and reliable manner. Future demand may be higher than forecast due to changes in weather patterns, higher population growth, or lower DSM participation. Spot market energy prices may be higher than anticipated or unexpectedly fall. New technologies could dramatically transform the market and affect both the quantity and shape of a future load profile. The construction of a power plant could take longer than projected, increasing the construction costs and forcing the LSE to enter into a temporary supply contract until the facility is operational.

Section 3.3 introduced several analytical techniques commonly used to evaluate long-term planning uncertainty. Given this framework, we compiled LSE risk evaluation techniques and organized them into a number of risk categories. Many LSEs developed scenarios to explore portfolio performance given alternative visions of the future. Some evaluated the sensitivity of portfolio costs to a wide-range of input values. A few used the results of the sensitivity analysis to determine which uncertain inputs to evaluate using advanced statistical techniques.

click to enlarge

Figure 10 is a comparison of utility methods used to assess uncertainty organized by a number of quantitative risk categories. A yellow box in the matrix indicates an LSE only conducted a scenario or sensitivity analysis; a blue box represents that only a probabilistic or stochastic analysis was conducted; a red box indicates that both scenario/sensitivity and probabilistic analyses were employed. Unfilled boxes represent risk categories that LSEs did not evaluate quantitatively. For example, AESO considered the price of coal in their scenarios, but did not report alternative prices beyond the base assumptions. There are very few examples of probabilistic-only analyses (blue boxes), such as PSCo which hired a consultant to conduct a stochastic analysis to provide the most likely load forecast. PSCo used this load forecast in their scenario analyses. In general, larger LSEs undertook more comprehensive risk assessments than smaller LSEs.

We found that a number of risk categories were evaluated by LSEs consistently. For example, almost all conducted scenario/sensitivity analyses for natural gas and electricity prices. This finding is not surprising since these two inputs have significant and direct impacts on LSE operational costs. In addition, uncertainty about future load and the availability of DSM also received considerable attention, as did the availability of lower cost supply-side resources. The data bars on the right of the figure represent an aggregation of perceived exposure to a particular risk category. The relative length of the data bar is computed by the summing how many LSEs addressed that particular risk type in their planning (i.e. we assigned one point for the use of scenario or probabilistic analysis and two points when both were used). Using this method, future demand, natural gas prices, and GHG compliance uncertainties dominate LSE risk assessments, followed closely by uncertainty about the cost and performance of future wind resources and DSM participation.

click to enlarge

Demand Risks

Predicting future demand is a critical step in the creation of an IRP. If demand grows faster than predicted, a number of consequences could occur including rolling blackouts during peak demand periods and high market spot prices for wholesale electricity. Conversely, if demand is lower than anticipated, it is possible the utility will over-invest in the construction of new resources—which could potentially sit idle. Effective DSM reduces demand and can alleviate consequences of over-prediction, but these activities cannot eliminate the risk altogether. Yet, even the successful adoption of DSM is another type of uncertainty considered by LSEs in their long-term planning. Most LSEs run multiple load forecast sensitivity analyses to see how robustly their preferred portfolio responds to a range of consumer demand assumptions.

Technological advancements can have dramatic impacts on the quantity and temporal profile of both demand and supply. For example, Smart Grid10 and plug-in hybrid electric vehicles (PHEV) are two technologies that have the potential to fundamentally affect future load and the resources that may be selected to meet this demand. Less than one third of the IRPs we reviewed evaluated impacts of Smart Grid technologies in their future demand scenarios, and most of those only alluded to efficiency gains attributed to generic Smart Grid technologies in the future. PHEV and electric vehicles (EV) have long been considered the future of ground transportation (NIST, 1993; Wilkerson et al., 1994). There are a number of studies on the integration of PHEVs/PVs for off-peak charging, load balancing, and general market impacts (e.g., Brown et al., 2010; Hadley and Tsvetkova, 2009; Kim et al., 2012; Pleat, 2012), and most agree that vehicle grid integration will have significant impacts on the electric utility industry. This potential impact has led the PUC of Oregon to require LSEs to address the integration of PHEV/EVs in their RPs (ORPUC, 2012). Most of the LSEs who considered smart grid technology impacts also considered the adoption of EVs in some of their portfolio analyses.

click to enlarge

Natural gas price risk

Over the past decade, the U.S. has seen a substantial increase in estimates of conventional and unconventional natural gas supply. Unconventional gas resources, such as coalbed methane and shale gas, have shown substantial growth in the last few years and are rapidly becoming a significant part of the U.S. energy portfolio (Jacoby et al., 2012). There is a considerable amount of uncertainty about how the current natural gas “boom” will affect long-run prices and the future mix of electricity generation capacity across the Western U.S., especially when combined with greenhouse gases (GHG) policies. For example, the Energy Information Administration (EIA) and Wood Mackenzie project that gas consumption will increase if GHGs are regulated; yet, Resources for the Future and MIT project that U.S. gas consumption will decrease if GHGs are regulated (Huntington, 2011).

Figure 11 shows the range of base-case Henry Hub forecasts from the 13 IRPs reporting price forecasts for this hub. The range represents an aggregate of resource planners’ mid-value, and not any high or low price sensitivity values. The figure compares the IRP forecast range to the spot price, futures market price, and range of EIA price forecasts. Despite the recent boom in production—which has lowered prices—all forecasts indicate the prices are expected to increase.

Surprisingly, the year the IRP was released played very little role in the range of anticipated prices, since base values from more recent plans span the entire range. This is also a good indication of the increasing level of uncertainty underlying planning decisions.

Surprisingly, the year the IRP was released played very little role in the range of anticipated prices, since base values from more recent plans span the entire range. This is also a good indication of the increasing level of uncertainty underlying planning decisions Surprisingly, the year the IRP was released played very little role in the range of anticipated prices, since base values from more recent plans span the entire range. This is also a good indication of the increasing level of uncertainty underlying planning decisions.

click to enlarge

Conclusion

In this study, we reviewed publicly-available planning information for nearly 40 utilities that, in aggregate, are responsible for generating ~90% of the energy in the WECC. We also sent a supplemental survey to resource planning staff to give each LSE an opportunity to update their L&R projections. We conducted this analysis to gain insight into a number of important topics:

(1) what Western electric utility planners are assuming about the future growth of electricity demand and supply-side resources; (2) what types of risk Western utilities consider and focus on in their long-term resource planning; and (3) how consistent is the collection and reporting of resource planning-related data across this region.

This analysis found that aggregate energy generation could increase from 857 TWh in 2011 to 1,011 TWh by 2020, which represents an average annual growth rate of ~2%. Unfortunately, not all utilities provided information about expected DSM activities. For those utilities that did report, we anticipate their energy demand and peak power will decrease by 39 TWh (6.4%) and 12,414 GW (8.6%), respectively, from EE activities—and an additional 7,188 MW from DR programs.

We also collected information related to existing and planned power capacity. We found that ~90% of anticipated plant retirements are split between natural gas and coal-fired generators. The most common type of new capacity is natural-gas fired units. A few LSEs anticipate building new (or upgrading) coal-fired generation, but many anticipate growth in wind, solar (mostly PV), biomass, and hydropower over the next few decades. Most of the planned solar capacity is anticipated before 2020, while most of the new wind is expected to come online after 2020. Despite these anticipated changes, thermal generation is expected to remain ~55% of total generation through 2030. However, it is important to note that a significant share of planned generation (~12%) is of an unknown type.

We also evaluated how LSEs are assessing risk to the performance of their resource portfolios. We considered different risk categories and determined which categories receive the most attention from resource planners. Almost all LSEs conducted scenario/sensitivity analyses for natural gas and electricity prices. In addition, uncertainty about future load and the availability of DSM also received considerable attention from most LSEs, as did the availability of lower cost supply-side resources. When we quantify the attention each risk type receives, future load, natural gas prices, and GHG compliance uncertainties dominate LSE risk assessments, followed closely by risk analyses about cost and availability of future wind resources and DSM participation.

click to enlarge

One consistent trend we discovered in our review of resource plans is that critical planning assumptions are inconsistently collected and reported by the LSEs. Despite a clear need for new supply and demand-side resources, we found that LSEs rarely reported additions (or improvements) to transmission interconnections, fuel delivery systems, and energy storage facilities. Furthermore, there are numerous examples of resource plans that do not provide sufficient clarity on units of measurement related to important risks (e.g., short tons vs. long tons of GHGs; carbon price vs. CO2e price; hub vs. delivered natural gas price).

FERC Order 1000 requires that regions begin to coordinate their long-term planning activities (FERC, 2011). We believe that a pair of enabling policies could facilitate regional inter- comparisons and, ultimately, lead to more efficient long-term planning processes across the Western United States and Canada. For example, local/state/regional policymakers should consider: (1) promoting inter-regional electric-gas-transmission planning data collection standards; and (2) supporting additional development of publicly-accessible databases of long- term gas-electric-transmission industry planning assumptions.12

Unfortunately, there are no inter-regional planning data standards in place to collect information in a consistent manner despite a clear need. The state of Washington requires that LSEs serving at least 25,000 customers must complete a standardized resource planning data template every year (WA, 2008). Data standardization experiences in states like Washington could serve as a useful model for determining inter-regional data collection standards. Encouraging gas-electric- transmission planning entities to identify their most important assumptions—and standardizing the collection of this information (e.g., units of measurement, planning horizons) will clearly improve the long-term coordination of regional planning organizations.

Furthermore, there is a general lack of access to publicly-available and consolidated long-term planning assumptions from the natural gas-electric-transmission industries. This lack of consolidation likely leads to inefficient inter-regional planning and increased costs to LSEs and their customers. In addition to requiring gas-electric-transmission entities to report all planning information in a standardized format, we recommend that policymakers continue to support the development of publicly-available systems to collect and distribute a variety of planning assumptions (e.g., load forecasts, supply-side resources, policy assumptions, fuel prices, other cost adders) to stakeholders in a user-friendly format.

Resource planning activities are typically subject to state-level jurisdiction, so the responsibility to standardize processes and improve transparency lies with each public utility commission. However, WECC (or other regional entities, such as the Western Interstate Energy Board) could play an important role coordinating (or convening stakeholders) in the development of common standards and reporting formats. Despite these shortcomings, the general quality and level of detail included in resource plans has increased over time. We anticipate that our recommendations will eventually have direct implications for the ability to explore additional policy-relevant questions in the future (e.g., among comparable LSEs are power plant costs and capacity factors significantly different?). We suspect that emerging policies, like FERC Order 1000, will lead to additional improvements in how effectively utilities—and their customers-- plan for the future.

Tweet