A re-rolling mill's roof cannot carry a plant big enough to matter against its demand. The demand-charge argument that can still work, how to test it from twelve months of meter data, and the power factor benefit nobody tenders for.
A re-rolling mill will ask for a solar proposal and then be quietly disappointed by the answer, so it is better to give the disappointing part first. The roof of a rolling mill shed, however large it looks, cannot carry a plant big enough to be a serious fraction of what the mill draws. Anyone quoting a percentage of consumption without first looking at the mill's load factor has not done the sum.
That does not mean the answer is no. It means the case has to be built on the right line of the bill, and there are two lines that most proposals ignore.
A re-rolling mill has a very high connected load and a low load factor. The reheating furnace is usually fired on gas or furnace oil, so the largest energy input is not electrical. What is electrical is the rolling train: main drive motors that see a heavy transient every time a billet enters a stand, plus the roughing and finishing stands, the cooling bed, the shear, the descaler pumps and the water treatment. The demand swings hard and it swings fast.
An induction furnace based billet plant is a different animal again. There the melting itself is electrical and the load is enormous, and a rooftop array will not register against it in any meaningful way. Say so early and the conversation moves to something useful.
So the arithmetic goes in this order. Take twelve months of consumption. Take the roof area actually available after obstructions and setbacks. Work out the plant that area supports. Express it as a share of annual consumption, and put that number in front of the owner before anyone talks about payback. If the honest share is small, that is the starting position, and every subsequent claim has to live inside it.
There is a second consideration on the energy side that cuts the wrong way for heavy industry. Industrial tariffs in Bangladesh charge a higher rate in the evening peak hours. A solar plant generates in the off-peak daytime block, so it displaces the cheaper units and leaves the expensive ones untouched. For a business whose demand peaks in the evening, that weakens the case relative to a business that runs hardest at noon. It is worth stating plainly rather than being discovered in the first year's bills.
Bangladeshi industrial tariffs bill maximum demand as well as energy, and for a plant with a low load factor the demand charge is a disproportionate share of the bill. That is the line where a small solar plant can matter more than its size suggests, because the demand charge is set by a single moment in the month rather than by total consumption.
The argument is straightforward. The utility meters demand at the point of supply, which is a net figure. If the array is generating strongly at the instant the mill sets its peak, the recorded demand is lower by roughly the amount the array was producing. On a plant where the demand charge is a large slice of the bill, that is real money against a modest capital outlay.
The argument has one condition and it is not negotiable: the peak has to fall in daylight, and it has to fall in daylight reliably. A single cloudy afternoon peak sets the demand for the whole month and undoes the benefit. Testing the condition takes an afternoon, not a study.
Where the peaks do sit in daylight but the mill wants certainty rather than probability, the honest recommendation is not a bigger array. It is a demand controller that sheds a defined non-critical load as the integrating window approaches the threshold, with the solar plant reducing energy underneath it. Photovoltaics dispatch on sunlight; a controller dispatches on command, and only one of those can be relied on at the minute that matters.
Rolling mills run large induction motors with heavy transient loading, and the power factor is frequently poor enough to attract a penalty. Most sites deal with it through a capacitor bank that was sized for a load pattern the mill has since outgrown.
Modern string inverters can operate at a settable power factor and can supply or absorb reactive power within their apparent power rating. An inverter fleet on the site is therefore capable of contributing to power factor correction through every daylight hour, at no additional equipment cost, simply by being commissioned with the right parameter set instead of the factory default of unity.
Three things to establish before counting on it. Confirm the inverter model's reactive power capability and the range it will hold at, since supplying reactive power consumes part of the inverter's rating and reduces the active power it can deliver at the same time. Confirm with the distribution utility that operating at a non-unity power factor is permitted under your interconnection agreement, because the setting is one they approve. And check the interaction with the existing capacitor bank so that the two are not fighting each other, which is a real risk where the bank is switched on a controller with its own target.
Some manufacturers also offer reactive power support at night, with the inverter drawing a small auxiliary supply to provide correction outside generating hours. Whether it is worth enabling depends on the tariff and on what the bank already does, but it is worth asking the question at tender stage rather than discovering the capability three years in.
Rolling mill sheds are tall, wide and usually uninsulated, and the roof runs hot. That costs yield, so mount with a real air gap rather than flush to the sheet. The atmosphere inside carries mill scale and, near the furnace, oily particulate, which makes for a soiling deposit that is greasy rather than dusty and needs a detergent wash rather than water alone.
Light engineering workshops, incidentally, are much better candidates than the mills they serve. A workshop with CNC machines, welding sets, compressors, a paint booth and a single day shift has a daytime load with good coincidence to generation and a roof that stays comparatively clean. Their difficulty is the opposite one: the roof is often too small relative to what they would happily consume.