How Good is the EIA at Predicting Henry Hub?

Natural gas power plants are a key component of bulk electrical systems in North America. In the U.S., natural gas power plants made up the largest portion of installed capacity, 42%, as of December 2016 and contributed more to generation than any other source. In Mexico, natural gas power plants supplied 54% of the required electricity in 2015 and are a key component of the capacity additions in development of the national electrical system. Natural gas is also likely to be the primary energy source in the U.S. due to increased regulation on coal units, uncertainty around the future of nuclear generation, and low natural gas prices.

Natural gas prices are a critical driver of electricity prices and a key input variable in electric power models. Due to the large amount of natural gas power plants in North America, and because fuel costs are the largest cost component of a thermal power plant, wholesale electricity prices are tightly coupled with natural gas prices. There is also an important feedback loop, in that natural gas demand, and price, is tightly coupled to the operation of natural gas power plants. Understanding the interplay between gas and power markets, and uncertainties in forecasts, is critical for forecasting either.

The U.S. Energy Information Administration (EIA) provides natural gas price short-term forecasts through the Short-Term Energy Outlook (STEO) and long-term forecasts through the Annual Energy Outlook (AEO). For the purposes of this article, we will focus on the STEO. The STEO is a monthly report with, among other items, a natural gas consumption and price forecast for 13 to 24 months in the future depending on the month published. The model predicts consumption and prices for three sectors (commercial, industrial, and residential) in the nine U.S. census districts. To do this, the model calculates natural gas consumption and supply levels to build an inventory. Prices are derived from a regression equation using the inventory and heating and cooling degree days, and analysts then make adjustments for final prices. Detailed information on each equation and method is provided by EIA Natural Gas Consumption and Prices document.

How good is the EIA at forecasting natural gas prices from a month to a year out?

To evaluate the STEO forecasts of natural gas prices, we downloaded each monthly STEO report from January 2012 to December 2016 to allow for at least a full year of analysis with historical prices. This period was selected because it is representative of the current trend of low natural gas prices (relative to historical). The mean absolute error (MAE) and mean absolute percent error (MAPE) were calculated for each forecasted value. Prices were then evaluated for the first forecast in each year and a subset of forecasts from consecutive months during a price spike. The mean absolute percent error was also evaluated for each report year and across all reports.

For the period analyzed (2012 to 2016, shown in orange below), the wholesale Henry Hub gas price averaged $3.30/mmbtu with a high price of $6.19/mmbtu in early 2014 due to the extreme Northeast weather (i.e., the polar vortex) and a low price of $1.78/mmbtu due to warm weather conditions and large amount of storage late in 2016. This period is representative of relatively low natural gas prices as compared to the previous five-year period with high prices exceeding $10/mmbtu driven by high oil prices and an average of $5.63/mmbtu despite the sharp decline due to the financial crisis in 2008-2009.


Figure 1. Historical Henry Hub natural gas prices. The yellow period denotes the study period used for this analysis. Source: EIA.

We started by looking at the longest-term forecasts (24 months) that are delivered in January of each year, and saw an inability to capture rapid fluctuations in prices in the study period:


Figure 2. Historical Henry Hub gas prices with 24 month forecasts from the January STEO of each year starting in 2012 and ending in 2015 using the base case data. Source: EIA STEO.

The January 2012 forecast missed the sharp reduction in prices from the winter to summer that were driven by high storage volumes. Less volatility occurred over the first part of the January 2013 forecast, however this forecast missed the large increase in prices to over $6/mmbtu which were driven by extreme weather conditions. The January 2014 forecast also missed the weather-driven high price for this period and then was high-biased in the later months of the forecast. The January 2015 forecast was high-biased the entire forecast period and missed the lower prices which were driven by a combination of mild weather and high storage volumes.
The STEO forecast is very sensitive to the initial conditions or starting month’s price. For example, plotting each month’s forecast during the increase from $3.74/mmbtu in November 2013 to $6.19/mmbtu in February of 2014 shows the impact of the rapid change in initial condition (last known price) on the first month forecasted value:


Figure 3. Historical Henry Hub gas prices with forecasted values from the months leading up to the rapid price spike in February 2014.

Presumably the long-term fundamental drivers of the STEO do not change as much as the initial conditions, and thus the longer-term forecast is much less sensitive to initial conditions.
Despite missing the fluctuation events, on average across the years analyzed the STEO is within 8% of the price in the first month of the forecast, 25% of the price out to eight months and 33% of the price out to 13 months:


Figure 4. Mean absolute percent error calculated for each forecasted month of STEO reports. Data are averaged over a report year, as well as over all of the report years. Maximum and minimum percent error is calculated over all STEO reports.

On average, the trend has increasing error with forecast length, however, this does not occur in the 13-month 2012 or 2013 STEOs. The expected error growth with time does appear in the 2014 and 2015 STEOs, reaching nearly 60% in the 2014 STEO. The maximum percent error in any given forecast grows rapidly from 26% in the first forecasted month to 75% in the fourth forecasted month, and reaches a high of over 100% 12 and 13 months out.
In absolute terms, the error ranges on average from $0.25/mmbtu in the first forecasted month to $0.88/mmbtu 13 months out. Maximum and minimum errors range from less than a penny up to $2.45/mmbtu.


Figure 5. Mean absolute error calculated for each forecasted month of STEO reports. Data are averaged over a report year, as well as over all of the report years. Maximum and minimum absolute error is calculated over all STEO reports.

Is the STEO forecast good enough? Unfortunately, as with many answers, it depends. More importantly, however, is understanding the limitations and uncertainties in their gas forecasts. If relying on EIA forecasts, you must realize the sensitivity to initial conditions and the typical error growth in the first months to year of the forecast. With this information, sensitivity studies can be formulated to capture possible fluctuations in gas prices. Taken together with other uncertainties such as demand, transmission outages, and plant outages, you can begin to form an ensemble of forecasts.

Filed under: Natural Gas, UncategorizedTagged with: , , ,

Integrated Gas-Power Modeling

Quantifying the Impacts of the EPA’s Clean Power Plan

Notwithstanding the recent legal stay from the U.S. Supreme Court, it is still important to understand the U.S. EPA’s Clean Power Plan (CPP) and its impact in the larger context of natural gas markets and its role in electric power generation. Because these two markets are becoming even more highly interrelated, integrated gas-power modeling is the most realistic approach for such analyses. EPIS has tested interfacing AURORAxmp® with GPCM®, a calibrated NG model developed by RBAC, Inc. The following is a brief discussion of our experimental setup as well as some of our findings.

Integration Approach

Monthly prices for 39 major natural gas hubs for the next 20 years are represented in AURORAxmp (as an input). They were developed utilizing GPCM’s market model (as an output) in pipeline capacity expansion mode. AURORAxmp then simulates a long-term capacity expansion that utilizes the GPCM-generated gas prices, and produces many results: power prices, transmission flows, generation by each resource/resource type including gas-consumption data. This gas-consumption (output from AURORAxmp) is fed back into GPCM as gas demand by the electricity sector (input to GPCM) for a subsequent market balancing and pipeline capacity expansion simulation which generates a new set of monthly gas hub prices. The iterative process begins at some arbitrary, but plausible, starting point and continues until the solution has converged. Convergence is measured in terms of changes in the gas-burn figures and monthly gas-hub prices between subsequent iterations.

This two-model feedback loop can be utilized as a tool to evaluate energy policies and regulations. To quantify the impact of an energy policy, we need two sets of integrated gas-power runs which are identical in all respects except the specific policy being evaluated. For example, to understand the likely impacts of emission regulation such as CPP, we need two integrated gas-power models with the identical setup, except the implementation of CPP.

Before presenting our findings on the impact of “CPP vs No CPP”, we first provide some further details on the setup of the GPCM and AURORAxmp models.

GPCM Setup Details

• Footprint: All of North America (Alaska, Canada, contiguous USA, and Mexico), including liquefied natural gas terminals for imports, and exports to rest-of-world.
• Time Period: 2016-2036 (monthly)
• CPP Program: All the effects of CPP on the gas market derived from changes to gas demand in the power generation sector.
• Economics: Competitive market produces economically efficient levels of gas production, transmission, storage and consumption, as well as pipeline capacity expansion where needed.

AURORAxmp Setup Details

  • Footprint: All three major interconnections in North America (WECC, ERCOT, and the East Interconnect; which includes the contiguous U.S., most Canadian provinces and Baja California).
  • Time Period: 2016 – 2036 (CPP regulatory period + 6 years to account for economic evaluation)
  • CPP Program: mass-based with new source complement for all U.S. states
    • Mass limits for the CPP were applied using the Constraint table
    • Mass limits were set to arbitrarily high values in the Constraint table for the “No CPP” case.
  • RPS targets were not explicitly enforced in this particular experiment. Future studies will account for these.
  • LT Logic: MIP Maximize Value objective function


  1. “CPP” – Convergent result from integrated gas-power model with CPP mass limits.
  2. “No CPP” – Convergent result from integrated gas-power model with arbitrarily high mass limits.
  3. “Starting Point” – Gas prices used in the first iteration of integrated gas-power modeling.
    • This is the same for both “CPP” and “No CPP” case.

Quantifying the CPP vs. No CPP

Impact on Gas and Electricity Prices

  1. Both No CPP and CPP cases have generally lower prices than the Starting Point case in our experiment. However, post-2030, CPP prices are higher than the Starting Point.
    • This happens due to capacity expansion in both markets.
    • We stress that the final convergent solutions are independent of the Starting Point case. The lower prices in CPP and No CPP cases compared to the Starting Point case are a feature of our particular setup. If we had selected any other starting price trajectories, the integrated NG-power feedback model would have converged on the same CPP and No CPP price trajectories.
  2. CPP prices are always higher than the No CPP case.
    • This is likely driven by increased NG consumption in CPP over No CPP case.

This behavior was observed in all major gas hubs. Figure 1 shows the average monthly Henry Hub price (in $/mmBTU) for the three cases.

Impact of CPP on Henry Hub PricesFigure 1: Monthly gas prices at Henry Hub for all three cases.

Figure 2 presents the monthly average power prices in a representative AURORAxmp zone.

Comparison of Power Prices in PJM Dominion VPFigure 2: Average monthly price in AURORAxmp zone PJM_Dominion_VP with and without CPP.

Figure 3 shows the impact of CPP as a ratio of average monthly prices in AURORAxmp’s zones for the CPP case over No CPP case. As expected, power prices with the additional CPP constraints are at the same level or higher than those in the No CPP case. However, it is interesting to note that the increase in power prices happens largely in the second half of CPP regulatory period (2026 onwards). It appears that while gas prices go up as soon as the CPP regulation is effective, there is latency in the increase in power prices.

Impact of CPP on Zone Price (CPP/No CPP)Figure 3: Impact of CPP on electricity prices expressed as a ratio of CPP prices over No CPP prices.

Figure 4 presents a comparison of total annual production cost (in $billions) for each of the three regions.

Annual Production Cost (In $billions) for each of the three regions.Figure 4: Total annual production costs by region for CPP and No CPP case.

Figure 5 presents the same comparison as a percentage increase in production cost for the CPP case. The results show that while the CPP drives up the cost of production in all regions, the most dramatic increase is likely to occur in the Eastern Interconnect.

Percentage increase in production cost total for CPP over No CPP CaseFigure 5: Percent increase in production cost for CPP case.

Electricity Capacity Expansions

Comparing the power capacity expansions in Figure 6 and Figure 7, we see that AURORAxmp projected building more SCCTs in the CPP case vs. the No CPP case in the Eastern Interconnect. We believe this is primarily driven by the higher gas prices in the CPP case over No CPP case. SCCTs typically have slightly higher fuel prices compared to CCCTs, which get their fuel directly from the gas hub for the most part. In this long-term analysis, AURORAxmp is seeking to create the mix of new resources that are most profitable while adhering to all of the constraints. The higher gas prices in the CPP case are just high enough to make the SCCTs return on investment whole.

Eastern Interconnect Build Out - No CPPFigure 6: Capacity expansion for Eastern Interconnect – No CPP Case.

Eastern Interconnect Build Out - CPPFigure 7: Capacity expansion for Eastern Interconnect – CPP Case.

Table 1: Capacity expansion by fuel type in total MW.



East Int.












































Table 1 shows the details of power capacity expansion in the three regions with and without CPP emission constraints. In addition to increasing the expansion of SCCTs, we can see that CPP implementation incentivizes growth of wind generation, as well as accelerates retirements. Coal and Peaking Fuel Oil units form the majority of economic retirements in the CPP case.

Fuel Share Displacement

Figure 8 shows the percent share of the three dominant fuels used for power generation: coal, gas, and nuclear. Figure 9 shows the same data as the change in the fuel percentage share between the CPP and No CPP case. Looking at North American as a whole, we see that coal-fired generation is essentially being replaced by gas-fired generation. Our regional data shows that this is most prominent in the Eastern Interconnect and ERCOT regions.

Percentage Share of Dominate Fuel TypeFigure 8: Percentage share of dominant fuel type.

Change in fuel share for power generation (cpp - no cpp)Figure 9: Change in fuel share for power generation (CPP – No CPP).

Natural Gas Pipeline Expansions
The following chart presents a measure of needed additional capacity for the two cases. The needed capacity is highly seasonal, so the real expansion need would follow the upper boundary for both cases.


Additional NG Pipeline Capacity RequiredFigure 10: Pipeline capacity needed for the CPP and No CPP cases.

Our analysis shows that the CPP will drive an increase in natural gas consumption for electricity generation. The following chart quantifies the additional capacity required to meet CPP demand for NG.
Additional NG Capacity Required CPP vs No-CPP (bcf/day)

While the analysis presented here assumes a very specific CPP scenario, we stress that the integrated gas-power modeling is an apt tool for obtaining key insights into the potential impacts of CPP on both electricity and gas markets. We are continuously refining the AURORAxmp®-GPCM® integration process as well as performing impact studies for different CPP scenarios. We plan to publish additional findings as they become available.

Filed under: Clean Power Plan, Natural Gas, Power Market Insights, UncategorizedTagged with: , , , ,

Integrated Modeling of Natural Gas & Power

Natural gas (NG) and electric power markets are becoming increasingly intertwined. The clean burning nature of NG, not to mention its low cost due to increases in discovery and extraction technologies over the past several years, has made it a very popular fuel for the generation of electricity. As a result, the power sector is consistently the largest NG consumer. For example, in 2014, 30.5% of the total NG consumption in the United States was used for the generation of electricity (Figure 1).


Figure 1: U.S. Natural Gas Consumption by Sector, 2014. Source

According to EIA’s Annual Energy Outlook (AEO) 2015 projections,

“…natural gas fuels more than 60% of the new generation needed from 2025 to 2040, and growth in generation from renewable energy supplies most of the remainder. Generation from coal and nuclear energy remains fairly flat, as high utilization rates at existing units and high capital costs and long lead times for new units mitigate growth in nuclear and coal-fired generation.”

Economic, environmental and technological changes have helped NG begin to displace coal from its dominant position in power production. Although it was just for a single month, NG surpassed coal for the first time as the most used fuel for electricity generation in April 2015. The EIA also notes that considerable variation in the fuel mix can occur when fuel prices or economic conditions differ from those in the AEO 2015 reference case. The AEO reference case assumes adoption of the Environmental Protection Agency’s (EPA) implementation of Mercury and Air Toxics Standard (MATS) in 2016, but not the Clean Power Plan (CPP). Adoption of CPP, along with favorable market forces, could change the projections of the AEO 2015 reference case significantly. There is a consensus within both NG and power industry that NG-fired power generation will likely increase with the adoption of CPP.

Quantifying such a trend is non-trivial, but is crucial for stakeholders and regulators in both gas and power markets to fully understand what the future holds. Proper accounting of the interdependencies between NG and power markets is integral to the quality of any long-term predictions. Approaches for modelling an integrated NG-power capacity expansion that account for economics and market operations is the key to the most effective analysis.

The issue of gas-power integration has been a topic of active interest in the industry, and that interest is increasing. For example, the East Interconnect Planning Collaborative coordinated a major study in 2013 – 2014 to evaluate the capability of NG infrastructure to: satisfy the needs of electric generation, identify contingencies that could impact reliability in both directions and review dual-fuel capability. Likewise, the notorious “polar vortex” during the winter of 2013-2014 caused unusually cold weather in the New England region, which “tested the ability of gas-fired generators to access fuel supplies,” and caused ISO-NE and others to acknowledge the need to further investigate the issues affecting synchronization between gas and electric systems. More recently, companies like PIRA Energy are sharpening their focus on the interdependencies between gas and electric power.

There is a need for new and improved modeling approaches that realistically consider this growing gas-power market integration. An even greater need is to integrate the modeling of these markets in a way that is both efficient and practical for the end user, and still able to produce commercially viable results. EPIS has extensively tested interfacing AURORAxmp with GPCM, a calibrated NG model developed by RBAC, Inc. Several organizations and agencies have found this approach successful. Utilizing the two models allows us to develop projections for endogenously derived capacity additions (in both electric generation expansion and gas-pipeline expansion), electricity pricing, gas usage and pricing, etc. which are consistent between the two markets. This consistency leads to greater insight and confidence to aid decision-makers.

Figure 2: Abstract representation of integrated NG-power modeling using AURORAxmp and GPCM..

Although the industry is now anxiously waiting for the judiciary to weigh in on the legality of CPP regulations, there is a consensus that some form of carbon emission regulation will likely be in effect in the near future. Some states, such as Colorado, have already undertaken several regulatory initiatives and may implement a state-level CPP-like emissions regulation even if the federal plan is vacated by the courts.

As part of our ongoing research on the topic of gas-power modeling, we have designed and executed a series of test scenarios comparing the standard calibrated cases of AURORAxmp and GPCM against a potential implementation of CPP. If the proposed form of CPP is upheld in the courts, states have a number of implementation options. At this early stage, there has been no good evidence to indicate that one option would be more popular over another. This necessitated we make some broad assumptions in our experimental gas-power integration process. In our test scenarios, we assumed that all states would adopt the mass-based goal with new resource complement option.

An integrated gas-power framework allows us to better understand the most probable direction for the two markets. Our integrated GPCM-AURORAxmp CPP test scenario for the Eastern Interconnect took 7 iterations to converge to a common solution that satisfied both markets. By comparing resulting capacity expansions, fuel share changes, and gas prices between the starting point (Iteration 0) and ending point (Iteration 6) we get a sense of how the markets will coevolve.

Starting capacity expansion in the Eastern Interconnect for GPCM-AURORAxmp model.

Figure 3: Starting capacity expansion in the Eastern Interconnect for GPCM-AURORAxmp model.

Figure 3 shows the capacity expansion resulting from Iteration 0, the starting point of the integrated iterations. Iteration 0 is essentially a standalone power model with no regard for the impact the capacity expansion would have on the gas market. Figure 4 shows the capacity expansion after Iteration 6.

Resulting capacity expansion in the Eastern Interconnect for GPCM-AURORAxmp model.

Figure 4: Resulting capacity expansion in the Eastern Interconnect for GPCM-AURORAxmp model.

The convergent prices of NG were lower for Iteration 6 than Iteration 0 at all major gas hubs. Figure 5 shows the monthly prices at Henry Hub for both the iterations. The lower gas prices are unintuitive, but plausible. The combined gas-power sector has several market forces that are interdependent. We are currently working with gas experts to understand some of the mechanisms that could lead to lower gas prices. We hypothesize that our accounting for capacity expansion in both the markets is one of the drivers for this behaviors and our findings will be reported in a future publication.

Comparison of starting and ending price trajectories with integrated GPCM-AURORAxmp model.

Figure 5: Comparison of starting and ending price trajectories with integrated GPCM-AURORAxmp model.

The lower gas prices highlight one of the key benefits of integrated gas-power models. Standalone modeling frameworks are likely to misrepresent the impact of the complex cross-market mechanisms. Integrated models avoid this particular pitfall by explicitly modeling each market and is a more apt tool for evaluating policies such the CPP. AURORAxmp provides the capability to model any of the implementation plans that states might adopt in the future – rate-based, mass-based, emission trading schemes and so forth. The ability to interface with widely used NG models, such as GPCM, provides a convenient option for analysts to confidently navigate the highly uncertain future of intertwined NG and power markets.

Filed under: Clean Power Plan, Natural GasTagged with: , , ,