energy · ocean · family: designed for a world that doesn’t exist here
deep waterwind, deeper costs
Floating offshore wind turbines are too massive and expensive for deep-water deployment
Problem statement
Nearly 60% of the potential energy of U.S. offshore wind blows across waters more than 200 feet deep, an area that ARPA-E says cannot be economically accessed today. Floating offshore wind turbines (FOWTs) could unlock this resource, but ARPA-E describes the current state of the art as too unwieldy and expensive for deployment. FOWTs are currently designed to replicate the dynamics of more familiar onshore wind turbines, maintain stability, and survive storms; the program says radically new designs are needed that do not require a massive floating platform and that substitute control systems for mass.
Why this matters
The ATLANTIS program page says the potential energy of U.S. offshore wind is estimated at more than 25 quads per year, and that the nearly 60% of it that blows over deep water is equivalent to the entire annual electricity consumption of the United States. Without floating turbines that are economical to deploy, that majority of the offshore wind potential remains out of reach. The program page states that developments in FOWTs could reduce the cost of wind energy production.
What’s been tried and why it hasn’t worked
The ATLANTIS program page does not survey earlier platform designs. What it says about current practice is that FOWTs are designed to replicate more familiar onshore wind turbine dynamics, maintain stability, and survive storms, and that radically new designs that do not require a massive floating platform are needed. It names three technical barriers to designing economically competitive FOWTs: insufficient knowledge of how FOWT sub-system dynamics interact; insufficient computer tools for dynamic simulation; and a dearth of experimental data.
What would unlock progress
Control co-design — integrating all relevant engineering disciplines at the start of the design process, with feedback control and dynamic interaction as the primary drivers of the design — is the approach the program encourages; the page describes it as substituting mass by control systems. It lets designers analyze the interactions of a FOWT's aero-, hydro-, elastic-, electric-, economic-, and servo-system dynamics together. ARPA-E's ATLANTIS program ($26M for 13 projects at announcement) aims to maximize the rotor-area-to-total-weight ratio while maintaining or increasing turbine generation efficiency, to build a new generation of computer tools for FOWT design, and to collect experimental data to validate the designs and the tools.
Entry points for student teams
A team could develop a coupled aero-hydro-servo-elastic simulation of a simplified FOWT concept and use optimization algorithms to minimize platform mass while maintaining stability and power output. Alternatively, a team could design a scaled physical model for wave-tank testing of a novel lightweight platform geometry. Aerospace, naval architecture, and controls engineering skills would be most relevant.
Genes
Structural cousins: stuck the same way, in other fields
Sources
ARPA-E ATLANTIS (Aerodynamic Turbines Lighter and Afloat with Nautical Technologies and Integrated Servo-control) program description, U.S. Department of Energy, accessed 2026-02-16 go to source ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Related to energy-wind-blade-composite-recycling-no-standard (wind energy materials challenges). The ATLANTIS program page gives a release date of January 31, 2019 and lists the program's status as Active.
Reconciliation 2026-10-01: Fact repair after reading the whole write-up against its one source, the ARPA-E ATLANTIS program page (https://arpa-e.energy.gov/programs-and-initiatives/view-all-programs/atlantis, fetched 2026-10-01). The page draws its text with a script, so the text was read in full from the site's own data record for the page (https://arpa-e.energy.gov/jsonapi/custom/node?filter[content_type]=program&filter[alias]=atlantis): the program description, the "innovation need" paragraph, the impact lines, the project list, and the two news-release summaries shown on the page. The page is a short program description, and most of the write-up's detail was not in it. (1) The text said nearly 60% of "accessible" U.S. offshore wind energy lies over water deeper than 200 feet, "where conventional fixed-bottom foundations cannot be economically installed." The page now says nearly 60% of the potential energy of U.S. offshore wind blows across waters more than 200 feet deep, "an area that cannot be economically accessed today," and no version of it mentions fixed-bottom foundations; the sentence now follows the live page. (The word "accessible" was the page's own wording when the write-up was made: Internet Archive captures of the page from 2020 through February 2026 open "Accessible U.S. offshore wind is estimated at more than 25 quads per year"; the live page reads "The potential energy of U.S. offshore wind is estimated at more than 25 quads per year.") (2) Removed because the page does not contain it: that platforms "weigh thousands of tons and cost 2–3× more than fixed-bottom equivalents." (3) The text gave the cause as conservative design that treats turbine and platform as independent systems. The page says FOWTs are designed to replicate onshore turbine dynamics, maintain stability and survive storms, and that new designs are needed that do not require a massive floating platform; the sentence now says that. (4) Removed because the page does not contain them: a deep-water resource of more than 2,000 GW, "enough to power the country several times over"; the West Coast, Gulf of Maine and Hawaii; Japan, South Korea and Europe; grid decarbonization; and the Hywind Scotland and WindFloat Atlantic costs of $150–200/MWh against a $50–70/MWh target. The paragraph now gives the page's own figures, attributed to the page rather than to an ARPA-E estimate: more than 25 quads per year, with the deep-water share equal to the entire U.S. annual electricity consumption. (5) The whole account of what has been tried (three platform types, methods inherited from oil and gas, blade-pitch control held back by misaligned incentives between separate companies, multi-rotor and downwind concepts) is not on the page. The section now says so and gives what the page does say: the current design approach and three named technical barriers. (6) The text said control co-design optimizes the "mooring system" among other things and "could dramatically reduce structural mass," and listed composite materials, tensioned moorings and AI-driven control as contributors. None of these is on the page. The section now gives the page's description of control co-design and the program's three stated aims. (7) In these notes, "ran from 2020 to 2023" is not on the page, which gives a release date of January 31, 2019 and shows the program's status as Active (a funding year of 2019 sits in the page's data record but the page script does not display it, so it is not stated); the sentences on Hywind Scotland, WindFloat Atlantic and IEA Wind Task 30 are not on the page either and were removed. Confirmed in the source and left as written: the program's full name; the nearly-60% share and the 200-foot depth; the aim of maximizing the rotor-area-to-total-weight ratio; and $26 million for 13 projects, which appears in the summary of the Department of Energy news release shown on the page. The entry points are suggestions and were left as written. No title changed; the title's "too massive and expensive" is the page's wording in the archived captures through February 2026, and the live page says "too unwieldy and expensive for deployment." Genome tags untouched. A second reader went over these repairs on the same day, re-fetched the page and its data record, and read the archived server-rendered versions of the same page at its earlier address (arpa-e.energy.gov/technologies/programs/atlantis, captures from October 2020 to December 2024, which now redirects to the address above); the second reader kept all seven repairs and reworded two of them as described in (4) and (7).