Added electric load in Colorado can help create a more cost-effective electric system — and save money for all!
by Jeffrey Ackermann
Thanks for including in Big Pivots occasional commentaries and editorials such as the recent column by Morey Wolfson. I’ve known Morey for many years, and we have periodically enjoyed the spirited exchange of ideas and opinions. In that spirit I offer some comments in response to Morey’s essay on “big data centers.”
See: “Colorado should not offer breaks on data centers,” Big Pivots, Sept. 25, 2024
The broader question here is whether we want electric load growth. Since the passage of state legislation in 2007, energy utilities subject to Colorado Public Utilities Commission (PUC) regulation have been pursuing load reduction through demand-side management (DSM) programs. And we have been fairly successful, contributing to very low rates of load growth.
This is demonstrated in the most recent Public Service Co. of Colorado (PSCo, aka Xcel Energy) electric resource plan’s base case projections. These projections assume no new electrification efforts such as for EV charging or building electrification. Looking to 2030, Xcel projected a base-case growth of 0.3% per year in demand and 0.4% growth in energy sales.[1] (These are compounded annual growth rates.)
The emerging clean energy policy, however, can be summarized as “electrify everything.” This shift to carbon-free electricity yields emissions reductions in transportation, space heating, and some industrial processes.
Expanded electrification also can produce system benefits. Our electric systems were historically built to meet peak demand, no matter how infrequently that occurs. The analogy often used is building a church to hold Easter Sunday attendance. Both the electric system and church thus suffer from usage inefficiency much of the year. The more that supply matches demand, the more efficiently (and cost effectively) the system operates.
Pursuing load growth that is “flexible” offers an opportunity to increase system efficiency. Flexible load is electric demand that can be “adjusted” in real-time. It involves a more dynamic relationship between the utility and the customer, allowing for customer demand to be interrupted by the utility (under a compensation arrangement), or shifted to customer-sited generation or storage resources. It can also include a combination of other technology options, such as equipping residential, commercial or industrial appliances and equipment to be controlled remotely so that electric use can be adjusted to match system needs. Utilities can, by investing in such technology, better manage supply and demand, avoid delivering high-cost electricity during peak periods and instead shift use to when electricity is more available and affordable.
Maximizing this demand shifting requires recruiting new customers with flexible electric needs. Yet, all new load growth isn’t of equal value. This will require evaluating, and valuing, new load for its contribution (or impediment) to increased system efficiency.
While electric utilities are not precluded from recruiting desirable customers, such as those with flexible load, they tend to need financial encouragement (or a statutory mandate) to deviate from established practices.
One approach would be to have the PUC establish measurable performance targets for the utility, tied to financial incentives. This could be a combination of load growth, system efficiency (smaller peaks and valleys regarding changes in customer demand), and the resulting contribution to system decarbonization.
A carbon-free, growing and fully dynamic electric system should also produce savings to ratepayers because renewables cost less than the alternatives on a levelized cost-of-electricity basis and because of improved system use, once we stop paying for the fossil fuel resources. This transition requires significant up-front investments, such as advanced meters, transmission expansion, and EV infrastructure build-out.
Increased sales can protect consumers from future sizeable bill impacts of these investments by spreading the costs over more sales. Short of fundamentally rethinking how rates are set, (a topic for another time), increased sales can best shield consumers, especially the smaller consumers least able to defend themselves in PUC proceedings.
(And note: investment costs will hit customer bills before the benefits of growth occur. Yet, that could possibly be resolved through deferring cost recovery on growth-focused investments, synchronizing recovery with arrival of the growth.)
Regarding the merits and concerns of pursuing big data centers, let’s start with “economic development” rates. These EcoDevo rates are authorized by law and have multiple policy objectives, including new jobs and increased tax base. A less obvious objective is benefitting existing utility customers. Included in each customer’s bill is a portion of system costs — poles, wires, trucks, billing systems, etc. — factored into each unit of electricity sold. As more electricity is sold, the system costs can be more widely spread, reducing per unit costs (in a future rate case), as long as the new sales don’t require additional system costs.
These additional costs associated with serving an EcoDevo customer are referred to as the “marginal costs.”
When the PUC is reviewing and approving an EcoDevo rate, it is using this definition of marginal cost. And per the statute, “the rate must not be lower than the utility’s marginal cost of providing service to the qualifying …customer[2]”This is to assure that the EcoDevo customer covers these new costs versus spreading them among existing customers.
And as Morey notes in the QTS example from Aurora, they are covering specific direct costs attributable to the new service, such as line extensions. Those costs are not shared among all customers. And these customers, by statute, will receive service through an EcoDevo rate for a maximum of 10 years. Then the customer will be allocated a full share of the system costs, like all other customers.
When I presided over the Colorado PUC as chairman, we entertained a number of applications of this rate concept, as authorized by statute. Some versions worked out rather positively, such as the Xcel tariff for EVRAZ Steel in Pueblo (related to the 2016 ERP). Others, such as the pursuit by Black Hills of a “data mining/cryptocurrency” facility in Pueblo left much to be desired and fortunately did not come into being.
I agree that the 2023 and 2024 legislative proposals regarding data centers were not well thought-out. Yet, that doesn’t mean that encouraging load growth onto a cleaner (soon to be all or nearly all clean) system is a bad policy. Rather, if new load can be recruited to the system and ultimately share some system costs, that will help existing customers, especially residential and income-qualified customers.
If the new customer recruitment focuses on those with the most flexible load, (and if part of the proposed rate compensates the customer for its flexibility), then all may benefit: the new customer, existing customers, the utility and the climate.
Now, let’s apply this to big data centers. As Morey points out, this type of customer does not generally offer a flexible load, i.e., one that can be interrupted or time-shifted by the utility to maximize system efficiency. But what if data centers were encouraged to include on-site storage (or other system-beneficial clean energy assets) on their premises, assets that would support the customer for periods of service interruption by the system operator? Might not that be the type of new load that meets the broader objectives discussed above?
As to water use by big data centers, it should be viewed through the filter of water scarcity and the Colorado Water Plan. But if the electric use of data centers must be evaluated in this way, should we assess all new customer’s operational processes and impacts upon all resources – water, air, land, minerals, etc.?
For example, would this include reviewing electric service to new agricultural customers, to determine if their proposed crop is a low-water crop? That’s a challenging task to undertake and a potentially dangerous precedent. What other policy objectives should be overlaid upon recruiting new electric load?
Such matters, whether it is water use or other social or economic impacts of the new customer, are best left to state legislators.
Finally, to the issue raised about Xcel (or other utilities) asking to extend the life of coal plants slated for retirement: posing this as a binary option (electric growth vs. plant retirement) oversimplifies the situation.
For example, Colorado’s investor-owned utilities are required by state law to join an “organized wholesale market,” pursuant to SB21-072[3]. Market participation by Colorado utilities could provide access to regional resources, especially carbon-free resources, that can handle, at least in part, sizeable new electric demand. Also, other resources, such as wind, solar and storage, can come on-line relatively soon as needed. The PUC’s electric resource planning process can assess and respond effectively to load growth that could otherwise risk the coal plant retirement plans.
We need to equip and motivate the utilities to use more creative rate policy by directing them to pursue new electric load that supports our public policy objectives of decarbonization, electrification, and system dynamism. Such an approach would clearly support the public interest. If big data centers can fit these criteria, then they should be pursued, (leaving water use to the legislature and the state agencies overseeing the state’s water policy).
Jeffrey Ackermann was chair of the Colorado Public Utilities Commission from 2017 to 2021. He is now senior policy advisor to the Center for the New Energy Economy, senior fellow at Gridworks, and board chair of The Alliance for Collective Action.
[1] Proceeding 21A-0141E, Phase I Decision Order, C22-0459, paragraph 168.
[2] 40-3-104.3(6)(b)(2)
[3] SB 21-072 is codified as 40-8.5-108, C.R.S.
- Big data centers can have upsides - October 7, 2024
