Energy, networks & IT for manufacturers

What a plant actually buys, end use by end use

EIA's most recent Manufacturing Energy Consumption Survey, for 2018, sorts every manufacturing end use as either a direct process use or a direct nonprocess use, and the two answer different questions. Process uses — machine drive, process heating and cooling, electrochemical work — sit inside production; nonprocess uses — facility HVAC, facility lighting, onsite transport — keep the building running around it. Across all fuels in 2018, direct process uses took about 63.5 per cent of manufacturing energy and direct nonprocess uses about 23.8 per cent, with the remaining 12.7 per cent classified as indirect, largely boiler fuel and cogeneration. Electricity splits differently again: about 69.2 per cent of purchased electricity went to direct process uses and about 30.8 per cent to direct nonprocess uses. That mix, and the shape of the production schedule around it, is what a supplier prices.

• Direct process uses were about 63.5 per cent of total manufacturing energy in 2018.
• Direct nonprocess uses (HVAC, lighting, transport) about 23.8 per cent.
• Indirect uses, mainly boiler fuel and cogeneration, about 12.7 per cent.
• The survey defines categories, not your site — your plant's mix will differ.

Machine drive is half the electricity bill

Electricity and fuel concentrate in different end uses. Machine drive — motors turning pumps, fans, compressors, conveyors and process drives — took about 52.5 per cent of manufacturing electricity in 2018, while process heating, which is the largest single use of total energy because it is largely gas-fired, was only about 9.8 per cent of electricity. Facility HVAC was about 8.6 per cent and facility lighting about 7.2 per cent. The practical consequence is that a plant's electricity strategy is mostly a motor strategy. The largest levers are the efficiency of the motors and drives themselves, whether a drive is matched to a varying load or runs at a fixed speed against it, and whether equipment is sized for the duty rather than bought at a nameplate. A motor running well below its rated output spends on energy what a controls change would recover, and motor decisions are engineering decisions long before they are procurement decisions.

• Machine drive is about 52.5 per cent of manufacturing electricity use (2018 MECS).
• Process heating is about 9.8 per cent of electricity but the largest use of total energy.
• Facility HVAC about 8.6 per cent and facility lighting about 7.2 per cent of electricity.
• Variable-speed control is the classic lever wherever load varies rather than runs flat.

The production schedule is a tariff decision

A demand charge bills the highest sustained draw recorded in the billing period, not the total energy used, so the billing determinant is a power figure measured over an interval — commonly fifteen minutes — and multiplied by a tariff rate. A short event can set it. Twenty-five kilowatt-hours consumed inside a fifteen-minute window averages to one hundred kilowatts, and if that is the month's highest interval it becomes the month's billed demand regardless of how lightly the rest of the month ran. That is why the sequence of motor starts, the compressor that cycles with production, and the simultaneous restart after a shift change matter as much as the annual volume. Shifting or sequencing large loads away from the peak interval lowers billed demand even when consumption is unchanged, and the load factor computed from the meter's own interval data is the number worth comparing against the shape a quote assumes.

• Demand is a rate of use over an interval (commonly 15 minutes), not a total.
• 25 kWh in a 15-minute interval averages to 100 kW — a short event can set the month.
• Sequencing start-ups and cycling loads flattens the peak without cutting output.
• Compute load factor from interval data; it is the shape a quote assumes.

Demand measured in kW and billed in kVA

Where a tariff bills apparent power rather than real power, power factor decides the bill. The relation is kVA equals kW divided by power factor, so a one-thousand-kilowatt real load draws about 1,250 kVA at a power factor of 0.80 but only about 1,053 kVA at 0.95. Tariffs commonly set a floor — often somewhere between 0.90 and 0.95 — below which the utility either bills the equivalent kVA demand or applies an adjustment to the measured demand. Induction motors, welding sets and older drives all pull reactive power, so a plant with a heavy motor base sits closer to that floor than its operators expect. The tariff, not a rule of thumb, states the threshold, the measurement interval and the adjustment formula, and those three are what to read before assuming power-factor correction will change a bill.

• kVA = kW ÷ power factor — a poor power factor raises billed demand.
• A 1,000 kW load is about 1,250 kVA at 0.80 and about 1,053 kVA at 0.95.
• Induction motors, welding and older drives are the usual reactive load.
• The tariff states the threshold, the interval and the formula — read them.

OT and IT are two networks with different requirements

The plant floor and the office are now attached, and they are still governed by different standards. The Purdue Model and ISA-95 describe a layered reference architecture — physical process, sensing and actuation, control, manufacturing operations, then business planning — and most deployments put a demilitarised zone between the control layers and the enterprise, though that level is a design convention rather than a level in the standard itself. IEC 62443 is the more prescriptive frame: it requires the system to be partitioned into zones of assets that share security requirements and conduits that carry traffic between them, each with a target security level. NIST SP 800-82 reaches the same place from the guidance side, recommending security zones and controlled communication paths with tightly restricted protocols across the IT/OT boundary. For a manufacturer buying connectivity, that means the controls network is a requirement to be specified once, not a wing of the office LAN.

• ISA-95 and the Purdue layers separate process, control, operations and business.
• The DMZ between the layers is a deployment convention, not a standard level.
• IEC 62443 requires zones and conduits, each with a target security level.
• NIST SP 800-82 endorses segregated control networks and controlled paths.

Power quality is a boundary condition you pay for either way

Harmonics, sags and transients either shorten equipment life and cause trips, or are bought off with conditioning — there is rarely a third option. IEEE 519 sets steady-state harmonic current limits at the point of common coupling, keyed to the ratio of available short-circuit current to load current and expressed as total demand distortion, with the allowed distortion tightening as that ratio falls; a plant with a large drive population on a relatively weak service sits in the strictest band. IEEE 1159 is the vocabulary standard, defining and classifying the events themselves — interruptions, sags, swells, transients — which is what makes an event log interpretable and comparable between sites. SEMI F47 is a ride-through curve some semiconductor and process equipment is specified to, stating how deep and how long a voltage sag a device must tolerate. It is the clearest example of buying immunity rather than buying a fix after the fact.

• IEEE 519 limits harmonic current at the point of common coupling as total demand distortion.
• The limit tightens as the short-circuit ratio falls — weak services are stricter.
• IEEE 1159 defines the event vocabulary that makes logs comparable.
• SEMI F47 states a sag ride-through curve equipment can be specified to.

Interruptible service, demand response and self-generation

Where a utility needs the ability to shed load, it offers a lower rate in exchange for the right to interrupt. Interruptible and curtailable tariffs typically carry a notice period, a maximum number of events or interruption hours, a required reduction, and a penalty if the customer fails to deliver the reduction it agreed to; the terms are set by the tariff and the words are not interchangeable between programmes. Demand response runs along the same idea under a market or utility programme rather than under a rate. Onsite generation — combined heat and power, or cogeneration — is a different arrangement again: the survey classifies it as an indirect use, and the utility side of it is usually standby or supplementary service, priced under a separate tariff that also governs parallel operation, protection and metering. CHP does not remove demand charges by itself, because a plant that imports heavily during a generator outage can still set its billed demand.

• Interruptible and curtailable riders trade a lower rate for a call option on your load.
• Their terms include notice, event caps, a required reduction and a failure penalty.
• CHP is classified as an indirect energy use; the utility side is standby service.
• Standby tariffs govern parallel operation, protection and metering separately.

What a supplier can price, and what it cannot

A competitive supplier prices the commodity and the risk it carries on your behalf — volume, shape and term. It does not price the delivery system, the meter, the utility's reliability obligation or the taxes attached to the address. That boundary decides what an energy quote can and cannot cover: a supplier can shape a contract around the load's measured shape and the term you want, but it cannot change the wires charge, and it cannot remove a demand charge the utility tariff imposes. The same logic runs through the non-energy purchases. Connectivity, network security and power-quality mitigation are separate scopes with separate suppliers, and the plant's own operating decisions — scheduling, maintenance, setpoints — sit inside the fence, where no supplier can reach them.

• Suppliers price commodity, volume risk and term — not delivery or taxes.
• A demand charge the utility tariff imposes cannot be negotiated away through supply.
• Connectivity, security and power quality are separate scopes and suppliers.
• Operating decisions inside the fence are yours, not a supplier's, to change.

What to send first: the schedule, the meter and the boundary

Three things let any supplier or adviser size the opportunity honestly: the meter's own interval data, the production schedule that explains it, and a clear statement of where the IT/OT boundary sits. Twelve months of interval data shows the peak, the load factor and the seasonal shape rather than the annual total alone, and it shows whether an apparently high bill is a volume problem or a peak problem. The schedule explains why the peak falls where it does — a shift start, a batch process, a coinciding compressor cycle — which is what turns a flat quote into a shape that fits. The boundary drawing tells a connectivity supplier which systems may be reached from the office network, so that requirement is specified once instead of rediscovered later. Send the three together and the questions that follow are about terms rather than about baseline.

• Twelve months of interval data: peak, load factor and seasonal shape.
• The schedule that explains the peak, not just the annual consumption.
• A one-line sketch of the IT, DMZ and control zones.
• Planned output or equipment changes, so the quote is built on the right baseline.

PUT THE OPTIONS SIDE BY SIDE

Your comparison checklist

Common questions