Fibre fly mats on glass in a way rain will not clear, and many mill roofs are older than the business case. What to survey, where the inverter room goes, and what a 575 KWp plant asks of a site.
The first thing worth measuring on a jute mill roof is not the area. It is the purlin. Mills in Bangladesh have been running for decades, many of them in sheds put up long before anyone thought about hanging half a kilogram per square metre of glass and aluminium off the trusses, and the structure is what decides whether the project is a rooftop plant, a partial rooftop plant or a ground-mounted one next to the shed.
Vvon engineered and commissioned 575 KWp at Ahad Jute Mills. A plant at that scale is not a scaled-up domestic system. It is a multi-inverter installation with string-level monitoring, its own cable containment, its own earthing and a survey that came before the quotation rather than after it.
Softening, carding, drawing and spinning all throw fine jute fibre into the air. It leaves through ridge ventilators, turbine vents and extraction discharge, and it settles on every horizontal surface within reach, including glass.
Ordinary urban dust is a nuisance that rain mostly deals with. Fibre fly is different in a way that matters to the array. It is filamentous, so it catches on frame edges, clamps and cable ties rather than sliding off, and it mats. Once matted it traps ordinary dust and forms a felt that sits along the lower edge of a tilted module where water runs off. Rain will not clear it. In some conditions rain makes it worse, because a wet mat compacts and then dries into a crust.
The failure mode is not a general loss of yield across the array. It is a strip of shaded cells at the bottom of the affected modules, which forces bypass diodes into conduction and puts heat where the module was not designed to have it. String monitoring shows it as one or two strings drifting below their neighbours. Module-level thermography during an O&M visit finds the specific panels.
None of this is a reason to avoid a jute mill roof. It is a reason to price the maintenance honestly. An AMC written for a clean warehouse roof and then applied to a spinning shed will be abandoned within a year, and the array will quietly lose output that nobody is measuring.
Survey before you quote, and open up a bay to do it. What the drawings say a 1960s shed has and what is actually up there are frequently different things.
A good number of older mill sheds are roofed in asbestos cement. It is fragile, it cannot be walked on, it should not be drilled, and clamping an array to it is not an option any competent contractor will offer. There are two honest answers: replace the sheeting under the array, or build an independent structure that spans to the trusses or to the columns and never touches the sheet. Both cost money, and both are cheaper than the alternative, which is a man through a roof.
On a light sheet roof the load case that usually governs is not the weight of the array. It is wind uplift. Modules on a shed roof form a surface that the wind gets under, and the uplift at edges and corners is far higher than in the middle of the roof. Get the wind speed for the site's zone from the national building code and have a structural engineer check dead load plus uplift together. Where the truss will not take it, an independent structure standing on the columns, or a ground mount alongside the shed, is a legitimate design decision rather than a defeat.
String inverters breathe. They pull air across heatsinks and, on larger units, through filtered intakes. Put them on a wall inside a carding or spinning hall and the fans load up with fibre, the filters block, the unit derates on temperature and then trips, and the plant loses generation on the hottest and sunniest days of the year, which are precisely the days it should be earning most.
Put the inverter room outside the fibre-laden areas. A dedicated room on an external wall, or in the utility block, is the right answer. Where it has to be inside the mill, build it as a sealed room with a filtered, positively pressurised supply so that air leaks outward rather than fibre leaking inward, and add filter cleaning to the maintenance schedule at an interval set by the mill's dust load rather than by the inverter manual.
Keep DC cable runs short and the AC run long where the layout allows. Long DC runs through a mill mean more joints, more containment and more places for a fibre-packed cable tray to become a problem.
Jute dust burns. A rooftop array puts live DC cabling above a building with a real combustible dust load inside it, and DC arcs do not self-extinguish at a zero crossing the way AC arcs do. This is not an argument against the project. It is an argument for the details being right.
Jute arrives from the fields after the monsoon harvest, but a mill's electrical load follows its order book and its shift pattern far more closely than it follows the fibre calendar. A mill running three shifts on hessian and sacking has a load curve that barely dips, and it will self-consume nearly everything the array produces. A single-shift mill exports at both ends of the generating day.
Two things to insist on before sizing. First, twelve months of bills, not one, because a single month taken in a slack period will undersize the plant and a single month taken at peak will oversize it. Second, an actual load profile logged at the incomer over at least a full working week, so the design is set against a measured curve rather than an average derived from a kWh total.
The monsoon is worth calling out separately. June to September is the lowest-irradiance stretch of the Bangladeshi year and it is also when incoming fibre is wettest and the drying and conditioning load shifts. A design signed off on a March meter reading has been measured in the best month and will be judged in the worst one.
At 575 KWp the project stops being a roof job and becomes a small power plant on an existing electrical installation. The questions change accordingly.