Standard Physics says its AI engineer proposed an energy-saving retrofit plan in a data center simulation

AT A GLANCE

Standard Physics says Archimedes used 17 hours to test 1,460 designs, and simulations show that a full retrofit of the data center could reduce total electricity consumption by 10.9%.

Article

AI engineer tests 1,460 designs in 17 hours; data center electricity bills could save 227,000 USD per year

Dongcha Beating AI News Flash, YC-incubated startup Standard Physics has launched AI engineer Archimedes. It can design engineering solutions on its own, run physics simulations, and then continuously revise based on the results to find better-performing designs.

At a data center in the United States, the cooling equipment was already running at full speed, yet the hottest server rack still reached 35.6°C. Through simulation, the AI found that some cold air was not being delivered to the correct locations, and that hot air exhausted by servers would flow back into the racks.

The AI first tried adjusting the air supply outlets, isolating the hot air, and then optimizing the operation of the cooling equipment. The energy-saving solution found in the first round looked very effective, but it ignored the actual capabilities of the old equipment. After review found it could not be implemented, the AI added equipment constraints and searched for solutions again.

The entire process lasted 17 hours, with 1,460 designs tried in total. In the end, two retrofit plans were selected. According to the company's simulation estimates, a full retrofit could reduce the computer room's total electricity consumption by 10.9%, bring the hottest rack down to 29.2°C, and save about 227,000 USD in electricity costs per year. Another low-cost plan requires less than a quarter of the retrofit cost of the full retrofit and can achieve 79% of the energy savings of the full retrofit.

The AI also tested 1,333 turbine blade designs through 2D simulation within 30 hours. The final design, while keeping the airflow turning requirements unchanged, reduced energy loss by 25.5% compared with the original blade, surpassing an optimized design published by a Belgian research team in 2018.

The entire system is completed through the cooperation of two parts: Archimedes is responsible for proposing designs, comparing results, and deciding how to revise next; Reynolds is responsible for running simulations. It can convert problems described by engineers in natural language into OpenFOAM simulation tasks, automatically build models, generate computational meshes, run calculations, and analyze results.

Standard Physics has open-sourced Reynolds, which is responsible for running physics simulations; anyone can use it and modify the code for free. The complete AI engineer Archimedes, which can autonomously design solutions, repeatedly test, and find optimal solutions, has not yet been open-sourced and requires contacting the official team to use.

Key points

Standard Physics has launched AI engineer Archimedes, which can design engineering solutions, run physics simulations, and revise designs based on results.
In the case, the data center's cooling equipment was already running at full speed, but the hottest server rack still reached 35.6°C.
The first round of energy-saving solutions ignored the actual capabilities of the old equipment and was judged unimplementable after review.
After adding equipment constraints, the system tried 1,460 designs within 17 hours and ultimately selected two retrofit plans.
According to the company's simulation estimates, a full retrofit could reduce the computer room's total electricity consumption by 10.9%, bring the hottest rack down to 29.2°C, and save about 227,000 USD in electricity costs per year.
The low-cost plan requires less than a quarter of the retrofit cost of the full retrofit and can achieve 79% of the energy savings of the full retrofit.
The AI also tested 1,333 turbine blade designs through 2D simulation within 30 hours; the final design reduced energy loss by 25.5% compared with the original blade.
Archimedes is responsible for proposing designs, comparing results, and deciding the direction of revisions, while Reynolds is responsible for running simulations.
Standard Physics has open-sourced Reynolds; Archimedes has not yet been open-sourced and requires contacting the official team to use.

AI-assisted interpretation

The following is analysis, separate from reported facts. Verify important claims independently.

This system turns engineering design into a continuous experimental process: Archimedes proposes plans and determines how to revise next, while Reynolds converts natural language problems into OpenFOAM simulation tasks and executes calculations. The data center case shows that it can identify problems with cold air supply locations and hot air backflow, but the first-round plan was also unimplementable because it ignored old equipment constraints.

Why it matters to readers

If the simulation results can be realized in actual retrofits, such tools may help data centers compare cooling and energy-saving plans more quickly and find trade-offs among cost, temperature, and electricity consumption.

Beginners should distinguish between "simulation estimates" and "actual deployment results": the 10.9% reduction in consumption, 29.2°C, and 227,000 USD annual savings in the text all come from the company's simulation estimates and are not equivalent to having been actually achieved.

Risks and unknowns

The report does not state whether the data center has completed an actual retrofit according to the final plan.
Whether actual operation can achieve the consumption reduction, temperature reduction, and cost savings in the simulation cannot yet be confirmed.
The specific retrofit cost and actual electricity savings of the low-cost plan were not disclosed.
Archimedes' pricing model, customer adoption, and commercialization results were not explained.

Standard Physics says its AI engineer proposed an energy-saving retrofit plan in a data center simulation

Article

AI-assisted interpretation

Why it matters to readers

Risks and unknowns

Related Developments

Related concepts

Keep following

Security Center