Introduction #

When this guide fits: Plant managers and electrical leads want a prioritized playbook to cut kWh, peak kW, and reactive charges without jumping straight to capital projects.

When it is not suitable: You are executing a utility-scale renewable PPA or behind-the-meter storage arbitrage—those need financial models and interconnection studies beyond this operations-focused overview.

Industrial bills usually blend energy (kWh), demand (kW or kVA interval peaks), power factor penalties, and sometimes time-of-use windows. Optimization starts by reading the tariff line items, not only the total.

Illustrative month (numbers for pattern only) #

Assume a simplified large-industrial stub (rates are examples, not a quote):

Component Quantity Example rate Example charge
kWh energy 2,000,000 kWh $0.045 / kWh $90,000
kW demand peak 18,000 kW $12 / kW-mo $216,000
Reactive excess 4,000 kVARh $0.80 / kVARh $3,200

In this pattern, demand dominates—your best ROI may be 15-minute peak shaving and start sequencing, not another lighting rebate. Always rebuild the table with your tariff PDF.

Illustrative cost stack: demand vs energy vs reactiveEnergyDemandReactiveBar heights =example $only

Read the bill like an engineer #

Line item What moves it First lever
kWh energy Runtime hours, efficiency Scheduling, VFDs, setpoints
kW demand Simultaneous peaks Staggered starts, load shedding
PF / reactive Displacement VAR Cap banks, motor loading, detuned stages
TOU Clock vs production Shift loads, thermal storage where viable

Quick wins (usually low Capex) #

  1. Eliminate simultaneous starts after outages—stagger large motors by seconds to minutes per stability testing.
  2. Fix compressed air leaks before buying new compressors; leaks inflate both kWh and demand.
  3. HVAC setpoint and deadband hygiene—document why each setpoint exists; drift creates hidden demand.
  4. Lighting controls with verified occupancy patterns (not default 24 h schedules).

Try our Energy Estimator for coarse what-if kWh cost sensitivity and Factory Load Calculator when motor or line additions shift demand.

Savings ideas mapped to bill lines #

If this line hurts… First engineering actions KPI to watch
kWh VFDs on fans/pumps, compressed air leak program kWh per unit produced
kW demand Stagger starts after outages, shed non-critical banks Monthly 15-min peak kW
Reactive Stage PF correction, fix oversized magnetics Displacement PF at PCC
TOU energy Shift melting/curing/charging into off-peak Clock vs production Gantt

Motors and drives #

Action Typical effect Caveat
VFD on variable torque loads Large kWh reduction Harmonics—plan detuning/filters
Resize oversized motors Better efficiency band Mechanical margin still required
Maintain belts and sheaves Fan kW drops Measure before/after with submeters

HVAC and process cooling #

Rightsizing and sequencing chillers or packaged units often beats adding capacity. Pair HVAC projects with HVAC Load vs Capacity thinking so new tonnage does not erase savings with short cycling.

Power factor and harmonics #

Improving displacement PF helps VAR charges and losses; harmonic distortion is separate. See Power Factor Correction: Best Practices before expanding capacitor banks on VFD-rich buses.

Measurement you should fund early #

Permanent submetering at incomers and top feeders pays for itself when tariffs change or production lines move. Align meter definitions with the utility settlement meter to avoid arguing with two different PF numbers.

Browse Power calculator hub for conversion and load tools.

Next steps you should take #

  1. Export 15-minute demand data for one month and circle the top five peak events—what equipment was running?
  2. Run a no-cost control review: schedules, VFD minimum speeds, and stuck dampers.
  3. Build a one-page business case per major Capex item with metered baseline first.
Will VFDs always lower my bill?

Usually on variable torque fans and pumps, but verify harmonic and cable heating impacts. Sometimes line reactors and filter maintenance become new OpEx lines.

Why does our demand charge stay high after LED retrofit?

LEDs cut kWh but not necessarily the 15-minute kW peak if motors or chillers still coincide—attack simultaneity, not only efficiency class.

Is power factor correction always safe?

No—resonance and leading PF near generators can trip equipment. Size and stage banks with engineering oversight, not catalog kVAR alone.

Should we buy a battery peak-shaver before fixing starts?

Usually no—sequence and controls fixes are cheaper and de-risk the battery business case. Model **15-minute** peaks before Capex.

Who owns the tariff interpretation?

Finance + energy lead should own a controlled register of rider changes; engineering feeds meter definitions so both sides use the same interval data.

Two demand-math drills (illustrative, not your tariff) #

Use these only to train the team on how demand arithmetic behaves; drop in your real rates afterward.

Drill A — stagger two identical starts: Baseline 15-minute peak 12.0 MW. Each of two 2.0 MW chunks is currently starting inside the same interval. If operations moves one chunk so peaks do not coincide, the interval drops toward 12.0 MW (unchanged) only if nothing else rises; if the peaks were additive at 14.0 MW, a clean stagger back to 12.0 MW saves 2.0 MW × your $/kW-mo demand rate. Multiply by 12 for an annual order-of-magnitude conversation.

Drill B — power factor on kVA-based demand: Suppose tariff language bills a kVA interval peak of 9.5 MVA while your true kW peak is 8.2 MW because reactive support is thin. A PF improvement that pulls apparent demand down to 8.8 MVA without changing productive kW is worth 0.7 MVA × the kVA demand charge—model it explicitly instead of assuming PF correction always pays.

Lever What you measure Typical first plot
Start sequencing 15-minute kW or kVA Heatmap by shift
Chiller reset Supply °F vs kW/ton Scatter by OAT bin
Compressed air leaks kW vs line pressure Overnight baseline drift

Try our kW to kVA converter when demand line items mix kW and kVA vocabulary on the same bill.

Submeter placement that actually changes behavior #

Put submeters where operators can see them: chiller plant, compressed air header, largest motor bus, and the paint or curing line that sneaks back into peaks when schedules slip. A meter buried in a report that only finance reads rarely changes setpoints. Pair each meter with one KPI on the shift board (for example kW per unit output) so savings stories survive staff turnover. When a pilot project claims 12% savings, require two weeks of before/after interval files—not only a vendor slide deck.

Conclusion #

Energy optimization is iterative measurement: tariff-aware KPIs, operator-visible setpoints, and submetered proof of savings. Start with schedule and simultaneity fixes—then spend Capex where data proves the bottleneck.