A Tale of Two 100% Renewable Puerto Rico Studies

Intro

In April, the U.S. Department of Energy (DOE) published its new 843-page final report studying the viability of a 100% renewables electricity system for Puerto Rico—a detailed and interesting modeling analysis assembled by six different U.S. national laboratories.

In a remarkable coincidence of timing, one of the most vocal and well-known groups of academic researchers focusing on renewables-only net-zero energy pathways recently published their own modeling results for a Caribbean-wide shift to 100% renewable energy systems by 2050, including Puerto Rico. Professor Christian Breyer and LUT University colleagues present an optimal solution for Puerto Rico that deploys over three times the installed clean energy capacity as the DOE’s model—including 17 gigawatts of floating offshore solar power—yet claim this pathway cuts Puerto Rico’s cost of electricity by more than half.

Such striking differences between the DOE’s conclusions and the LUT University study highlight the stark contrast between a hard-eyed, rigorous analysis meant to provide useful insights to planners in an under-studied, disadvantaged region, and superficial, performative work professing to provide the same service.

As this serendipitous head-to-head comparison indicates, this latter category of 100% renewables studies really exists primarily to pad academic CVs while providing a thin veneer of ostensibly empirical justification for ideological tenets. Over the past 15 years, a small but prominent group of vocal academics have produced dozens of such studies claiming that a 100% renewables and storage-based energy strategy can not only power “all energy in all regions of the world at low cost” but accomplish this even more cheaply than the existing fossil-fueled energy system.

Whether leaps like floating offshore solar or the representation of Puerto Rico’s grid as a single point—lumped together with the British and U.S. Virgin Islands to boot—make the LUT study worthless to planners is irrelevant. Both its authors and a larger ecosystem of NGOs and sustainability scholars will likely eagerly cite it without a second thought.

In contrast, the DOE’s report clearly prioritizes meaningful insights for policymakers, utility officials, and communities, offering a far more frank and detailed discussion of the obstacles still ahead for challenging geographies like Puerto Rico. Yet the DOE’s own modeling effort falls short of being as helpful to Puerto Rico as it could be by overly deferring to Puerto Rico’s 2050 100% renewable policy. This policy reflects the same entrenched ideological bounds on acceptable decarbonization technologies and approaches. Neglecting to consider nuclear power or less aggressively stringent targets leaves potential for better pathways unexplored.

Ultimately, clean energy modeling advances knowledge further and serves the public interest better when it pursues efficient system design flexibly, for a variety of decarbonization scenarios and using all of the low-carbon technology options on the table. As the sharp contrast between the LUT University and DOE research illustrates, the more modeling efforts feel obligated to validate a prescribed vision like an all-renewables 2050 target for its own sake, the more they constrain their own usefulness to society.

How do the studies measure up to one another?

Many Puerto Ricans will need to see grand promises of a cleaner, better grid manifest on the ground to believe them. With the island dependent on imported fossil fuels, unable to directly receive electricity from the continental U.S. or neighboring countries, and long responsible for its own power generation, Puerto Rico’s residents understand intimately how unique geography produces unique energy system circumstances (Figure 1). With Jones Act regulations also forcing the island to use scarcer, higher-cost U.S.-built, U.S.-crewed, and U.S.-flagged vessels to import fuels from the U.S. mainland, some of Puerto Rico’s energy access and cost challenges are also conspicuously artificial.

High costs have in turn hampered Puerto Rico’s efforts to maintain electricity infrastructure, which has struggled to recover in the wake of hurricanes like Maria and Fiona, leaving many residents without power for weeks or months. The cost of resulting repairs and resiliency upgrades, on top of the high price of imported fuel, have saddled Puerto Ricans with electricity rates more than twice as high as rates on the U.S. mainland, igniting anger towards LUMA Energy, the island’s utility energy provider.

In this context, the recently-published study by Breyer and co-authors focusing on Puerto Rico and the Caribbean alleges that their modeled fully 100% RE energy system for Puerto Rico, with 17 GW of floating offshore solar arrays, cuts the territory’s electricity costs from around €120/MWh in 2020 to €47.4/MWh in 2050 (a change of $148/MWh to $58.3/MWh assuming an average long-term exchange rate of 1 EUR = 1.23 USD) (Figure 2).

If the LUT University team’s body of 100% RE work was robust, one might expect the Department of Energy’s Puerto Rico study to produce corroborating findings, producing a relatively similar renewable energy system in scale, composition, executability, and affordability.

But the DOE’s research gives a dramatically different result: a fully-renewable future for Puerto Rico is likely feasible, but will maintain or increase the cost of electricity relative to Puerto Rico’s current, already pricey electricity system (from perhaps roughly ~$169/MWh in 2021 to $154-$179/MWh in 2050, based on 2021 and 2050 utility revenue requirements and total generation, see Section 12.1.3.1). Moreover, the DOE’s proposed renewable electricity system for Puerto Rico differs wildly from the LUT University group’s solution.

A first key difference involves the scale of renewable electricity production. U.S. National Laboratory modelers assessed two scenarios for Puerto Rico’s future energy demand: a baseline scenario where electricity consumption continues to slowly decline to 24-25 TWh/yr by 2050, and a more aggressive scenario where annual power demand surges to 32-35 TWh/yr (Figure 3). This contrasts with 73 TWh/yr of electricity generation by 2050 in the LUT analysis. Importantly, the LUT University study targets full economy-wide decarbonization for Puerto Rico and the Caribbean by 2050—a much more ambitious goal than the DOE’s 2050 100% renewable electricity target. This partially explains the difference in modeled system scales, but only intensifies scrutiny of the LUT team’s claimed lower costs.

The second major category of differences centers around the composition of the island’s future renewable power system. The LUT University Caribbean study aggressively slashes Puerto Rico’s energy costs relative to the DOE’s study despite ballooning Puerto Rico’s generation assets from the current total of 6 GW in 2020 to ~44 GW of installed capacity in 2050 (plus ~48 GWh of distributed and grid battery storage). This is well over triple the ~13.5 GW of generation capacity and ~1.9x the grid battery storage capacity the DOE modeled for the territory in 2050 (Figure 4).

On its own, the LUT study’s 17 GW of Puerto Rican floating offshore solar capacity already exceeds three-fold the DOE’s modeled total solar deployment (at most ~5.5 GW). In the DOE’s study, the island deploys 1.9 GW to 5.6 GW of distributed PV across all scenarios, relying extensively on distributed PV deployment to achieve a 100%-renewable grid. Meanwhile, planned utility solar PV reaches 172 MW to 512 MW in the DOE’s land-limited cases, growing to 362 MW to 1007 MW of planned utility solar in cases with greater land availability. This utility solar PV occupies 2.3% of Puerto Rico’s available land in the land-limited cases, and about 7.3% in the cases with greater land availability. In the LUT studies, utility-scale solar deployment is capped at 1% and 6% of land area in two different scenarios, contributing partly to the heavy reliance on floating solar.

Wind deployment remained at a relatively modest level in the DOE analysis, with 930 MW of onshore wind deployed by 2050. The DOE ultimately assesses offshore wind as cost-prohibitive for Puerto Rico to build. In comparison, the LUT modeling result produces ~5GW of onshore wind for Puerto Rico.

The differences between the two studies when it comes to technologies for storage and flexibility are more moderate, and stem largely from differences in methodology. To meet evening, night, and morning needs in 2050, the DOE’s system relies upon ~1 GW of short-duration battery storage, backed by longer-duration, 8-hour and 10-hour battery storage (~1.5 GW and 0.37 to 0.89 GW respectively, totaling 24.9 GWh) (Figure 4). Up to 1.6 to 4 GW of biodiesel-burning power plants provide reserve and peaking generation, representing up to 7-8% of annual power production (up to 2.8 TWh). Puerto Rico also possesses 700 MW of generation using landfill gas, although this capacity operates only very rarely. Apart from imported biodiesel, the DOE’s approach did not consider imports of hydrogen or other renewable fuels to operate thermal power on the island. The DOE’s most important finding was that local hydrogen production proved too expensive for deployment in any of their considered scenarios.

The LUT analysis deploys nearly double the total capacity of utility and distributed batteries by 2050 (48 GWh), and opts to import 10 TWh/yr of renewable fuels including synthetic methane from overseas to supply a backup fleet of 5 GW of gas turbines, plus local low-carbon transportation needs. LUT’s Puerto Rico synthesizes 10 TWh/yr of green hydrogen locally, plus a limited quantity of renewable methane.

Now, different energy system modeling groups produce markedly different results for the same region of study all the time. Diverging numbers on their own hardly incriminate one study versus the other. Assuming rigorous assumptions by the LUT team, it could just as easily be the case that the Department of Energy’s modelers had neglected to consider the full potential of offshore floating solar on Puerto Rico, for instance.

Upon closer examination however, the LUT group’s choice of inexplicably optimistic cost and modeling assumptions seems to far better explain the considerable divergence between the two studies’ results. Overall, the DOE is considering more realistic assumptions, more diverse scenarios, and more complex operational factors that pro-100% RE academics have consistently handwaved as minor cost drivers and dismissed the need to consider at all.

Rigorous modeling is useful modeling

A detailed read of both studies highlights stark dissimilarities in each group’s research approach. The DOE takes extensive care to tailor energy technology cost assumptions specifically to Puerto Rico’s unique environment, while modeling Puerto Rico’s energy system in high-resolution geographic detail to better capture total costs associated with electricity network upgrades. In comparison, the LUT team either adopts the same cost assumptions they use in their other global modeling analyses for Puerto Rico and the Caribbean—or dials technology costs down even further. Meanwhile, their decision to model Puerto Rico’s energy system as a single point or copper plate, even grouping the territory bizarrely with the British and U.S. Virgin Islands, certainly simplifies their study’s wholehearted embrace of floating offshore solar.

Extensive benchmarking, location-specific analysis, and stakeholder consultation by the DOE unsurprisingly reveals that energy project costs in Puerto Rico would likely exceed global average project costs, often by a considerable margin. For their modeling, the DOE assesses utility-scale solar PV project levelized cost of electricity (LCOE) in 2035 to range between $58 to $80/MWh, within but on the higher end of the current range of U.S. utility solar PV LCOE ($29 to $92/MWh) as presented by the popularly-cited Lazard Levelized Cost of Energy Analysis. Onshore and offshore wind in 2035 in Puerto Rico ($83 to $480/MWh for onshore and $179 to $349/MWh for offshore) remains at least 3.1 and 2.4 times more expensive than U.S. wind costs today in the latest Lazard report ($27 to $73/MWh and $74 to $139/MWh). Essentially, the assumed future Puerto Rico project costs for both onshore and offshore wind stay well beyond the upper end of current U.S. costs. Such considerations manifest in the final modeling results, with the DOE judging offshore wind completely uneconomical for Puerto Rico.

Meanwhile, fully flexible renewable electricity generation comes at an even higher premium. National Laboratory modeling settles on biodiesel generators despite their eye-wateringly high levelized electricity cost of $401/MWh (Figure 5). For context, the recent construction of the Vogtle Unit 3 and Unit 4 nuclear reactors—widely acknowledged as an expensive and mismanaged project—carries a price tag of approximately $178/MWh. The Puerto Rico analysis furthermore deemed other options, like hydrogen electrolysis, offshore wind, and ocean thermal energy conversion, expensive to the point of non-viability. Overall, such determinations highlight the critical importance of realistic modeled technology cost assumptions, indexed properly to the region of interest.

In contrast, the Breyer study assumptions imply far cheaper levelized costs of around $21/MWh and $57/MWh for land-based utility-scale solar and onshore wind in Puerto Rico respectively. This allows the LUT model to build onshore solar and wind on Puerto Rico, by 2030, at less than half the costs represented in the DOE’s study for the model year 2035 (Figure 6). Meanwhile, Puerto Rican offshore floating solar power costs fall by 2030 to below the Department of Energy’s lowest-cost 2035 projection for conventional fixed-mount solar projects on land.

Also, the Breyer group’s cost assumptions are either borrowed wholesale from the team’s other papers focusing on entirely different global regions—or revised downward from such values. Essentially, the LUT team assumes that energy project costs for Puerto Rico and the Caribbean are less than or equal to the global average costs they adopt in their other work (Figure 7). The researchers borrow their floating offshore solar PV cost assumptions verbatim from one single previously-published paper focusing on the Maldives, on which Christian Breyer enjoys credit as a co-author. As implemented, the model’s assumed capital expenditure costs for new offshore floating solar projects in 2030 (~$855/kW) actually sit on par with its residential rooftop solar cost assumptions. Such optimistic methodological choices clearly stretch the credibility of the Breyer analysis.

As such, it is neither commendable nor particularly surprising that the Breyer paper purports to achieve a reduction in Puerto Rico’s levelized cost of electricity from $148/MWh to $58.3/MWh, while the DOE concludes electricity costs remain high, shifting from $169/MWh in 2021 to $154-$179/MWh in 2050.

Aside from higher-cost power plant projects, the DOE study also attributes significant costs to system upgrades required to facilitate operation of a fully-renewable grid. These improvements include grid-enhancing equipment to maintain grid strength and compensate for retiring fossil generators. Meanwhile, upgraded local power networks and distributed storage systems help handle reverse power flow as rooftop solar arrays export electricity back to the system at large.

In contrast, in the LUT team’s single-point grid model, deployments of offshore floating solar, rooftop solar, and utility solar pose no differences in transmission and distribution costs.

To be clear, even a true “status quo” counterfactual scenario for Puerto Rico with continuing fossil fuel use to 2050 and beyond—a case the DOE did not model—would also likely see rising energy costs. Almost any imaginable scenario for strengthening the island’s grid to acceptable standards of reliability would likely involve near-term cost increases. That said, the study’s results indicate clearly that the last 10% of grid decarbonization is the hardest and that all else equal, a less stringent policy vision would be more affordable. Meanwhile, the DOE’s modeling clearly does not produce the same dramatic improvements in electricity costs that simpler global 100% renewables modeling studies claim to achieve.

Open questions and the case for open-mindedness

All models are ultimately wrong, but the DOE’s analysis is most certainly useful. Overall, the DOE’s 843-page report seeks to provide actual planning value to the region and people it seeks to investigate, as opposed to simply bestowing a stamp of certified 100% renewable viability and moving on.

The DOE report models multiple scenarios for land availability, energy demand, and rooftop solar potential. Researchers derived technology costs reflecting expected circumstances in the region of study. Multiple years of weather data help verify grid resilience under extreme circumstances. Assessment of future grid strength changes and backfeeding and reverse power flow from large quantities of rooftop solar ensure infrastructure meets requirements for operability. Unsurprisingly, such considerations turn out to be extremely important, producing a far more informative solution space than simple nodal 100% RE studies using generalized cost assumptions and considering only four, two, or even just a single scenario.

Certainly, none of the challenges of 100% RE system operation are insurmountable. While commentators should not dismiss or handwave operational or technical limitations casually, neither should they assume technologies have only fixed, inherent capabilities. Indeed, the DOE's finding that Puerto Rico could operate primarily on solar and batteries backed by long-duration storage and biodiesel highlights the considerable progress in renewables and related technologies. New capabilities, such as grid-forming battery storage systems that can support black-start resuscitation of the wider grid following storm damage, could improve Puerto Rican energy resilience. Grid-forming inverters, distributed storage, and supporting technologies like synchronous condensers can facilitate reliable operation of massed solar assets.

The Puerto Rico report also concludes the territory will absolutely need firm, flexible, clean generation. Hard upper limits to cost-effective wind energy on island grids like Puerto Rico’s mean the system must meet nighttime loads with some combination of overbuilt solar and storage paired with flexible generation. Whether biodiesel, hydrogen, synthetic renewable gas, long-duration storage, geothermal, carbon capture, or nuclear is the best option for optimizing solar, storage, and flexible generation in various different geographies is far from a settled matter, but the need for firm clean capacity is clear. Such technologies fill an entirely different supporting role and grid needs compared to solar and wind (Figure 8), justifying their higher costs and making direct cost comparisons between dispatchable and variable clean energy options nonsensical at the end of the day.

At the end of the day, the choice to pursue an electricity system running solely on wind, solar, storage, and renewable fuels by 2050 reflects an aesthetic preference articulated in policy by Puerto Rico’s government. The Department of Energy and the Puerto Rican government possess a responsibility to consider cheaper pathways that could equally promote energy security and economic growth while minimizing energy costs for Puerto Ricans.

To challenge the very premise of a net-zero grid by 2050 for Puerto Rico itself, wouldn’t a goal of 90% clean electricity by 2050—consistent with a more gradual transition to a fully clean grid by, say, 2060 for Puerto Rico—still drive remarkable decarbonization progress while softening the impact of rising electricity costs for residents? The fate of the planet’s climate hardly hinges upon zeroing out Puerto Rico’s power sector emissions. Even if it did, the mainland United States arguably bears some responsibility for putting in a little extra effort to compensate, rather than placing a heavier expectation upon a disadvantaged U.S. territory.

Alternatively, what if Puerto Rico’s government sought to achieve a fully clean grid as opposed to a fully renewable one? This subtle distinction would open up an opportunity for the island to perhaps consider low-carbon nuclear energy as an affordable substitution option, at least for perhaps 1.6 to 4 GW of eye-wateringly expensive, $401/MWh biodiesel generation, mitigating both biodiesel import reliance and local biodiesel-related air pollution impacts in the process. While this question demands more intensive study, one recent report by the Nuclear Alternative Project (NAP) suggested new small advanced reactors for Puerto Rico might come at a future price tag of $98 to $108/MWh, around a quarter of the cost of the DOE’s modeling assumption for biodiesel power. Meanwhile, a growing number of life-cycle analyses suggest that many biodiesel products offer only a 20% to 50% CO2 emissions cut relative to fossil diesel, raising the question of whether biodiesel even qualifies as low-carbon to begin with.

Puerto Ricans may very well wish to preserve more of the island’s naturally beautiful forests and coastlines from solar and wind farms at the cost of more expensive centralized clean generation. Or the territory might weigh the guaranteed higher environmental, climate, and air pollution impacts of biodiesel generation against the minimal and ever-improving risks of nuclear power plants and elect for a future system that includes new nuclear. One can easily imagine such choices leaving Puerto Ricans better off. Conversely, Puerto Ricans have every right to elect to pursue a truly 100% renewable grid, accepting that the last few percentage points needed to achieve that distinction would be the most expensive.

In the end, the PR100 report is a valuable deliverable, leveraging the vast expertise of the U.S. National Laboratory system to provide key technical and economic insights to Puerto Rican leaders. Chief among those important insights is the understanding that the path to a 100% renewable future for the island will be challenging, should the territory choose to pursue it fully and in earnest. In contrast, the LUT University 100% RE study for Puerto Rico and the Caribbean is anything but useful.

As our commentary illustrates, claims of universally viable and cheap 100% RE systems like those of the LUT group typically fail to survive rigorous inspection. Rather than engage in wishful modeling that papers over the inconvenient obstacles, researchers and commentators should grapple directly with global disparities when it comes to energy transition obstacles and potential solutions. Techno-economic factors that make certain things possible in Europe, Australia, or China are far from guaranteed to apply to Sri Lanka, Ethiopia, or Haiti. For one thing, not every Caribbean island has the same political and economic relationship, however flawed, with a wealthy United States mainland as Puerto Rico does.

Indeed, we have perhaps neglected to consider energy challenges of the Caribbean for far too long, and the lessons that energy transitions for small island states may hold for the rest of the world. Being able to solve decarbonization affordably and optimally with honest analysis for islands like Puerto Rico, Jamaica, and Barbados, in the long run, truly progress our ability to advance the energy transition globally.