Following up on our report from two weeks ago, which highlighted how Vulcan improves the accuracy of the EIA-860M's online schedule for power plants by over three times, we wanted to delve deeper into that analysis. All data from this report is easily accessible to Vulcan clients via the user interface, API, and Snowflake. The figures below were generated using a Snowflake connection in Tableau.
Vs.
For the six-month outlook, we observed a delta of approximately 10 GW between the EIA-860M forecast and our own. However, the net impact on natural gas demand for the summer was negligible. This was due to a wash effect, as delays in gas plant builds and the addition of solar and wind capacity effectively canceled each other out. This analysis used basic assumptions, including a 27% capacity factor for wind and solar, a 7 heat rate for combined cycle, and a 10 heat rate for combustion turbines and steam turbines.
When we extend the outlook to next summer, a nine-month forecast, the error grows.
Vs.
There are two major variables in forecasting future infrastructure builds:
Vulcan addresses both of these variables. Our platform not only provides more accurate online dates but also helps clients identify projects at high risk of cancellation.
As noted in other reports, Vulcan ranks all projects to enable our clients to create more realistic cancellation scenarios. Historical data from the EIA-860M shows that 20-40% of proposed projects in any given month are eventually canceled. Our ranking system allows users to create a custom cancellation scenario by simply sorting and removing a desired percentage. Currently, removing projects with a rank below 2.55 eliminates approximately 20% of the projects in the proposed set. The farther out a project's online date, the lower its ranking, as its risk of cancellation increases. This contributes to the greater difference we see between our nine-month outlook and the EIA-860M forecast.
Unlike the summer analysis, the impact on natural gas demand is not a wash. We are seeing a net impact of 0.3 Bcf/d more gas demand compared to the EIA-860M outlook. In the EIA-860M forecast, the net impact of new builds resulted in a 1.5 Bcf/d decrease in gas demand, based on the addition of solar and wind with our generic assumptions. In contrast, the Vulcan build-out forecast, using the same methodology, shows a net decrease of only 1.2 Bcf/d.
Vulcan V2 Table - REDACTED TABLE FOR CLIENT ONLY
This is a high-level analysis. You could also use more precise capacity factors by calculating state-by-state averages instead of a national average. An even more precise forecast would involve combining a load forecast (using Vulcan's data center intelligence) with our online dates in a dispatch model. Additionally, including storage will likely change the shape of the demand curve, potentially reducing the net impact on gas demand as load increases. Power traders would also gain significant value from power modeling, including insights into new build risks on congestion at various locations.
Vulcan was initially designed for power modelers, and its extensive historical dataset for power projects allows for a much more in-depth analysis compared to other infrastructure types available on the platform.
For more information on Vulcan or to schedule a demo, please reach out to David Bellman, Lead Researcher / Vulcan Product Lead, at dbellman@synmax.com, or visit our website at synmax.com or email us at info@synmax.com.