AI predicts wind-uplift resistance of solar PV piles

Researchers in Iran have built an AI tool that predicts how well steel piles will hold down large solar panel arrays against wind. This is a research development for utility-scale solar farms, not home rooftop systems, so it does not directly change anything for a homeowner's own property.

The problem the researchers are tackling: ground-mounted solar panels catch a lot of wind but don't weigh much, so strong wind can actually try to lift or tip the whole structure. The steel piles driven into the ground have to resist that uplift. In soft, loose, or wet soil, figuring out whether a pile will hold is tricky, and standard engineering formulas often get it wrong, leading to designs that are either overbuilt (more expensive) or risky.

To improve on this, the team tested 130 steel piles at a solar site in Rafsanjan, Iran, measuring how fast each pile was driven into the ground, the soil's friction properties, and how much pulling force it took to shift the pile. They fed this data into a neural network (a type of AI model) that could predict a pile's holding strength from installation data alone. The model performed reasonably well and found that how fast a pile is driven and the soil's friction angle matter most, more than the pile's length or surface area.

The researchers caution that their model was trained on one sandy site and may not work for different soil types like clay or gravel elsewhere. For homeowners, this is background engineering research aimed at making large solar installations more reliable and cost-effective, with no direct bearing on home energy upgrades or rebates.