Utilities Tech Outlook Weekly Brief
Be first to read the latest tech news, Industry Leader's Insights, and CIO interviews of medium and large enterprises exclusively from Utilities Tech Outlook
Erick Marquardt de Araujo, Head of Renewable Portfolio Management, ENGIE North AmericaThe Renewable Energy Certificates (RECs) market transacts more than 500 million RECs yearly, with a total value of $11.5 billion and continuously growing, in sync with renewables. There is little doubt that the Inflation Reduction Act (IRA) is a powerful tailwind, and some analysts go as far as to estimate that the US will more than triple its renewable energy capacity by 2030, reaching 725-750GW.
This market has seen increasing interest in tracking RECs’ time of production, in the pursuit of 24/7 carbon-free energy (CFE). The central idea is to power facilities with clean energy around the clock, avoiding (or reducing) the consumption of brown power when there is low renewable energy production (i.e., nights with no wind). With hourly tracked RECs, a given organization could assure that enough RECs are being produced every hour, from different sources, to fulfill their consumption. This ambition has famously made to the climate pledges of many tech companies (i.e., Google and Microsoft), but also became the cornerstone for another very promising (albeit nascent) industry – green hydrogen.
The Inflation Reduction Act (IRA) brings incentives of up to $3/kg for “carbon-free” hydrogen production - roughly double the price of hydrogen produced from NG. Some guidance and clear criteria still have to be published by IRS to define what qualifies as “carbon-free” hydrogen, but the expectation is that systems using electrolyzers will be required to show that the energy used comes from clean sources, at least within a tolerance band.
In both applications, data centers and green hydrogen, the question is fairly similar: which portfolio of resources would be required to supply 24/7 carbon-free energy (CFE)? Different energy markets will require different solutions, and for this article, the focus is on ERCOT, given its plurality of renewable resources (wind and solar) and recent developments in battery storage
“There is little doubt that the Inflation Reduction Act (IRA) is a powerful tailwind, and some analysts go as far as to estimate that the US will more than triple its renewable energy capacity by 2030, reaching 725-750GW”
The load considered for this exercise is 100MW aroundthe-clock (or 876GWh/year). The supply will come from a portfolio of assets, including wind farms in different regions (West, North, and South), a solar farm, and (possibly) battery storage. The selection of regions for the wind farms was made aiming for a diversity of production shapes (hourly and monthly), which can be seen in Figure 1. Solar production has less variability across regions, so it made less sense to be broken down.
The shapes displayed in Figure 1 and used in the optimization are derived from the actual production during the year 2022. Similar studies could be done using the curves from production estimates (i.e., P50 or P99 curves), simulated time series, or even other years. Of course, the robustness of the study will increase with the diversity of curves used, but so will the computing power and time required to run the optimization algorithm. Here, given the objective to test the concept, one year of observed data seemed a good compromise.
The final element needed to put the optimization algorithm to work is the cost of each resource. I assumed the prices shown in Table 1, roughly based on market trends for AsGen PPAs in these different regions. For the battery, the rationale is somewhat different. There are some tolling agreements in ERCOT being negotiated in the high single digits per kilowatt-month. However, the battery is only needed here when all the other possible resources can’t provide energy. In all other circumstances, the battery could sell ancillaries services or execute some peak-shifting, capturing extra revenues. So, I assumed that only a portion of the tolling cost (~50%) would be paid by this portfolio. Moreover, I disentangled the power and energy costs of the battery, to allow the algorithm to select the ideal battery duration.
The optimization was performed considering different percentages of guaranteed clean power and the results can be seen in Figure 2. As expected, hawkish requirements demand exponentially more power to be purchased and some levels are only achievable with sizeable battery storage.
An additional assumption is needed regarding the value of the power surplus - to compute the effective cost of powering the 100MW load, all energy not used is liquidated in the market producing some revenue. Observing the prices of 2022, we can see the value of this surplus is in the range of 25-45/MWh, depending on the scenario. However, going forward, we can expect this value to be considerably lower with reduced gas prices and higher renewable penetration intensifying the negative gamma effect. On the other extreme, we could consider that this surplus is worth zero. As usual, the reality is likely to lie somewhere in the middle. Figure 3 depicts this interval alongside the total amount of power acquired.
The results don’t look obscenely high as one could expect, even considering the 99% scenario. In fact, 10 years ago the LCOE for solar PV was around 100/MWh, and two years before that, onshore wind power was in that range, so not surprisingly expensive given the novelty of this product. Secondly, it is worthy of note that around 25% of the total costs in the high percentage scenarios is coming from battery storage, a technology that has evolved and where developments in materials and manufacturing processes can materially reduce the costs (of course, if we also solve the current supply-chain bottlenecks).
Finally, another possible interpretation here is that it might be advisable to walk before we run. The scenarios of 80% or 75% don’t represent material additional costs and could be great starting points. Moreover, the diversification of the portfolio, combining resources from different regions, brings other advantages to the off-taker, such as resilience against operational unavailability (e.g., maintenance, forced outages), congestion, and even some extreme weather events. Thus, an organization could start with a portfolio aimed to achieve a medium level of CFE, benefit from these effects, and add optionality over time (i.e., tolling a battery pack, or adding new resources) allowing it to achieve higher commitments percentages of assured CFE.
