Metals is one of the few sectors where energy is not an overhead line, it is an input cost sitting close to the product. That changes the conversation. In most industries the question is how to pay less per unit. In metals the question is which part of the process generates the cost, because the three main stages behave nothing alike.

Melt, hold, finish

Melting is the spike. Induction and arc furnaces draw hard and fast, and that draw sets your demand charge for the month. A single badly timed melt can define a bill the rest of the month cannot undo.

Holding is the baseload. Unglamorous, continuous, and because it never spikes it rarely gets attention. On an annual view it is often the larger number.

Finishing, the rolling, drawing, annealing and coating end, is the most schedulable of the three and therefore the most useful lever you have.

A supplier quoting a single blended rate is pricing all three as one thing. They are not, and that is usually where money is left behind.

Demand charges are the part people underestimate

On most commercial and industrial tariffs you are billed on two things: the energy consumed, and the highest sustained draw during the billing period. The second is the demand charge, and in a melt operation it can rival the consumption charge outright.

What makes it painful is that a short window sets it. Fifteen or thirty minutes of coincident peak, depending on the utility, can price the entire month. Two furnaces starting together because that is how the shift happened to run is a decision nobody consciously made, showing up as real money.

• Know which interval actually set your peak last month. Most operators do not, and it is in the data.
• Stagger melt starts where the production schedule allows. This is free.
• Check whether your tariff carries a ratchet clause, where one peak sets a floor for months afterwards.
• Separate the demand question from the supply question. They are negotiated in different places.

Load profile is your negotiating position

Suppliers price risk. A predictable, flat load is cheap to serve and gets quoted accordingly. A spiky, unpredictable one carries a premium because the supplier has to cover the shape.

This is the practical argument for knowing your own interval data before anyone quotes you. A metals operation with a genuinely disciplined melt schedule has a better book than its neighbour and should not pay the same adder. That case can only be made with data.

Contract structure for a load like this

The instinct in a volatile market is to fix everything. For metals that is often the wrong shape, because a fully fixed contract prices peak exposure into every kilowatt hour you buy, including the flat holding load that carries no risk at all.

A block-and-index structure fits the profile better: fix the predictable baseload, let the variable portion float. We work through that tradeoff in fixed versus variable contracts and in block-and-index pricing.

Natural gas is a separate negotiation

Melting, heat treating and annealing often run on gas, and gas is bought on a different curve, in different units, on different terms than electricity. Bundling them because one salesperson is in the room is how the weaker of the two contracts gets ignored.

Run them as two processes with two sets of numbers. They can be timed together, and often should be, but they should not be evaluated as one.

Where to start

Pull twelve months of interval data for your largest facility and find the intervals that set your demand charge. That exercise usually reframes the conversation, because the peak is rarely where people assume it is.

From there: confirm which sites sit in markets with retail supply choice, aggregate those into one bid rather than negotiating site by site, and compare contract structures rather than headline rates.