So no one in the (non-residential) industry cleans their modules, except if you're in the Southwest.

The phenomenon of panels getting dirty is called soiling loss, and its defined either by how many solid particulates (e.g. soil, dirt, sand, agricultural dusts from nearby harvests, chemical particulates from nearby factories, etc.) or snow accumulates. I make that distinction because there's different models that the industry uses for predicting these things: the Kimber model for non-snow, and the Townsend model for snow.

The reason the (non-residential) industry doesn't really care about washing their panels is because:

• 1. rain washes any solids away, and
• 2. panels generate heat if converting photons to electrons

The Kimber model assumes that solids accumulate as a line function which restarts once it rains more than a certain amount. Weather data provided by NOAA in the States for instance can be fed into the model to calculate what percentage losses your panels will experience over their lifespan. Usually solar engineers over design their systems so they reach the energy amount across the entire system's lifespan.

For snow its the same, except when it snows a crazy amount like in the Northeast US and so much accumulates, all of the panel's cells are blocked from direct sun and this delays the self-heating effect as the modules are essentially fully shaded. In those cases, depending on how bad the snowfall was or how frequently it falls, companies may elect to brush off the snow once or twice in a winter season.

You'll notice that I left out residential solar, which is what you most likely care about.

Since residential solar is so finely tuned to meet the greater degree of constraints with working with a smaller array than community or utility scale arrays, more attention does need to be paid to cleaning the modules.

Whoever is designing your system though should be able to build in a certain amount of soiling losses, and that will help dictate the final array design. If your developer or installer doesn't know what this number is (should range from 0.5-4% loss compared to perfect world conditions), then I'd try to dig more for that or switch developers/installers if they don't want to give that information up.

It's not worth it for homeowners or developers or installers to clean such small arrays unless it's their prerogative to do so I guess. But I guess I'm a lazy engineer making that call so who am I!