Plant Utility KPIs: Benchmarks for Compressed Air, Steam and Chilled Water

Plant utilities hide their losses in plain sight, so a handful of metered KPIs does most of the work. Use the ranges below as reference points for spotting a gap worth investigating, not as targets, and compare against your own metered trend first.

Every figure shows its basis. A superscript number links to a public source in the Sources list at the end, and the sentence states that source's scope. A figure marked no cited source is kept exactly as originally published because none of the public sources checked for this guide states it, so treat it as illustrative.

At a glance

Compressed air specific power

Specific power is the headline compressed air KPI: kilowatts drawn per 100 cubic feet per minute (cfm) delivered at the header. Measure it with a power meter on the compressors and a flow meter on the discharge, not from nameplate data.

World-class lubricated rotary screw systems hit 18 to 20 kW per 100 cfm at 100 psig no cited source. Typical plants sit at 22 to 26, and neglected systems exceed 28 no cited source.

For the compressor alone, DOE's Compressed Air Sourcebook (2016) lists lubricant-injected rotary screws at 16 to 19 kW per 100 cfm single-stage and 15 to 17 two-stage, at 100 psig¹. Those are full-load package figures that leave out dryers, part-load running and distribution losses, so a whole system reads higher.

For example, the gap between 20 and 26 kW per 100 cfm on a 500 cfm plant is 30 kW. Running around the clock for 8,760 hours a year at $0.10/kWh, that is near $26,000 a year.

Header pressure is the cheapest lever, so drop it to the lowest the process tolerates. In systems around 100 psig, DOE says every 2 psi of extra discharge pressure raises energy use by about 1 percent at full output¹. Counting the extra air that leaks and open blowing consume at the higher pressure, the combined penalty is about 1.6 to 2 percent per 2 psi¹.

How to measure

• Meter compressor input power with a true power meter, not nameplate current.
• Meter delivered flow at the header, after dryers and filters.
• Divide kW by delivered cfm and multiply by 100, always at a stated header pressure.
• Trend it weekly so a drift shows before the power bill does.

What moves it

• Header pressure, set to the lowest the process tolerates.
• Leaks, which load the compressors without doing any work.
• Compressor control and sequencing when demand falls below full load.

Put a cost on your own air demand with the Compressed Air Demand calculator.

Compressed air leakage

Leakage is the single highest-return utility KPI to chase first. Express it as a share of total compressed air production, measured while no end use is drawing air.

DOE's Sourcebook says the share lost to leakage should be less than 5 to 10 percent in a well-maintained system¹, and FEMP's O&M guide puts the same mark at less than 10 percent². Poorly maintained systems can lose 20 to 30 percent of air capacity, and DOE expects a typical plant that has not been well maintained to leak about 20 percent¹. Some plants exceed 40 percent no cited source.

Ultrasonic leak surveys twice a year, tagging and fixing the top 20 leaks, routinely recover 5 to 15 percent of total air no cited source. DOE recommends ultrasonic detection and fixing the biggest leaks first¹.

For example, if a compressor loads for 20 seconds and unloads for 60 with no production running, leakage is 20 / 80 = 25 percent of capacity.

How to measure

• Run the test in a no-production window, with every end use shut off.
• Log load and unload times over several cycles on a load/unload or start/stop compressor.
• Calculate leakage as on-load time divided by total cycle time, times 100, which is DOE's method¹.
• Repeat the test after each repair campaign to confirm the gain.

What moves it

• Ultrasonic surveys on a fixed schedule, with every leak tagged.
• Fixing the largest leaks first.
• Lower header pressure, since leaks pass more air at higher pressure.

Price your own leaks with the Compressed Air Leak Loss calculator.

Chilled water plant efficiency

Chiller efficiency is measured in kilowatts per ton of cooling, and the plant total matters more than the chiller alone. Track it monthly from metered kW divided by measured tons.

A world-class water-cooled central plant delivers 0.55 to 0.70 kW/ton all-in, including pumps and tower no cited source. Typical plants run 0.85 to 1.1 kW/ton total, and air-cooled sites 1.2 to 1.5 no cited source.

In a world-class plant the chiller by itself hits 0.50 to 0.60 kW/ton no cited source. For comparison, FEMP's federal purchasing requirement for full-load optimized water-cooled centrifugal chillers (updated October 2024, paired to ASHRAE 90.1-2022) runs from 0.610 kW/ton under 150 tons to 0.501 kW/ton at 600 tons and above³.

Reset chilled water supply temperature upward and condenser water temperature down toward design. FEMP's O&M guide says lowering condenser water temperature 2 F to 3 F can increase system efficiency by as much as 2 to 3 percent². The same guide says raising chilled water temperature by 1 F cuts chiller energy use by 1.7 percent for centrifugal and 1.2 percent for reciprocating machines².

How to measure

• Meter chiller, chilled water pump, condenser pump and tower fan kW together for the plant figure.
• Measure tons from chilled water flow and the supply and return temperatures, not from the chiller nameplate.
• Report the chiller alone and the plant total side by side, every month.

What moves it

• Chilled water supply temperature, reset upward when loads allow.
• Condenser water temperature, driven down toward design.
• Fouled coils and heat exchangers, which make the whole plant work harder.

Compare measured load with installed capacity in the Refrigeration Tonnage calculator, and size the duty behind it with the Chiller Load calculator.

Chilled water delta T

Chilled water delta T, the temperature rise across the coils, is the KPI that reveals hidden waste. Design is usually 10 to 12 F across the coils no cited source.

In low delta T syndrome the actual spread falls to 5 to 7 F no cited source. For a fixed cooling load, flow rises as delta T falls, so the pumps move nearly double the flow and the plant starves for capacity.

For example, restoring delta T from 6 F to 11 F cuts the flow needed for the same load to 6/11 of what it was, or about 55 percent. That can cut pumping energy by half and let one fewer pump run.

How to measure

• Log supply and return temperatures at the plant.
• Log them again at the worst coil, where the shortfall usually starts.
• Trend the spread alongside chilled water flow, not on its own.

What moves it

• Fouled coils that cannot pull heat out of the air.
• Stuck three-way valves that pass chilled water straight to return.
• Open bypasses between supply and return.

Chase all three before adding chiller capacity, and check the duty you actually serve with the Chiller Load calculator.

Boiler combustion efficiency

Boiler combustion efficiency is the steam KPI that pays first. Modern firetube boilers reach 82 to 85 percent combustion efficiency, and 87 to 90 percent with economizers and tight excess air no cited source.

For reference, the natural gas table in DOE's Steam Tip Sheet 4 (2012) gives combustion efficiency of 82.8 to 85.4 percent at 2 to 3 percent flue gas oxygen and a flue gas temperature 200 to 300 F above combustion air⁴.

Measure stack temperature and oxygen together: target 3 to 5 percent excess O2 and stack temperature within 100 to 150 F of steam temperature no cited source. DOE's natural gas table equates 3 to 5 percent flue gas oxygen with 15 to 28.1 percent excess air⁴. It also says an excess air level of 10 percent, about 2 percent oxygen, is attainable on well-designed natural gas-fired systems⁴.

DOE's rule of thumb is that boiler efficiency rises 1 percent for each 40 F reduction in stack gas temperature, or for each 15 percent reduction in excess air⁴. Dividing that rule out, each 1 percent of excess air above target costs a fraction of a point, about 0.07⁴.

How to measure

• Read flue gas oxygen and stack temperature with a calibrated combustion analyzer.
• Test across the firing range, not only at high fire.
• Log the readings monthly next to fuel use and steam output.

What moves it

• Burner tuning and oxygen trim to hold excess air down.
• Clean fireside and waterside surfaces, which keep stack temperature down.
• An economizer that recovers stack heat into the feedwater.

Check the steaming duty you fire for with the Boiler Load calculator.

Boiler blowdown

Blowdown removes dissolved solids from the boiler, and every pound of it sends heated, treated water to drain. DOE's Steam Tip Sheet 9 (2012) says blowdown rates typically range from 4 to 8 percent of boiler feedwater flow, and can be as high as 10 percent when makeup water has a high solids content⁵.

Hold cycles of concentration high enough to keep blowdown under 4 to 8 percent of feedwater no cited source. DOE notes that the right rate depends on boiler type, operating pressure, water treatment and makeup water quality⁵.

How to measure

• Work out cycles of concentration from boiler water and feedwater conductivity.
• Express blowdown as a share of feedwater flow.
• Log it monthly alongside the water treatment results.

What moves it

• Makeup water quality and pretreatment.
• Automatic blowdown control on conductivity.
• Condensate return, which cuts makeup water and the solids it brings in.

Size the steam load that blowdown is part of with the Boiler Load calculator.

Steam trap failure rate

Steam trap failure rate is the maintenance KPI that quietly bleeds a plant. It is the share of installed traps that have failed, and the costly ones are those that fail open and pass live steam.

With a regularly scheduled maintenance program, DOE says leaking traps should account for less than 5 percent of the trap population⁶. FEMP's O&M guide says giving repairs high priority can bring failures to 3 to 5 percent or less².

In systems that have not been maintained for 3 to 5 years, DOE says 15 to 30 percent of installed traps may have failed⁶. FEMP gives a wider 15 to 60 percent failed open without a good trap maintenance program².

Test traps on an interval set by their pressure. DOE recommends weekly to monthly testing for high-pressure traps (150 psig and above), monthly to quarterly for medium pressure (30 to 150 psig), and annually for low pressure (below 30 psig)⁶.

For example, a plant with 500 traps at 20 percent failure and an average 30 lb/hr loss per failed trap wastes 3,000 lb/hr, over 26 million lb of steam a year at 8,760 hours. Bringing failure to 5 percent recovers three quarters of that.

For scale, DOE's table puts the loss from a leaking trap at 100 psig at 13.2 lb/hr through a 1/16 inch orifice and 52.8 lb/hr through a 1/8 inch orifice⁶.

How to measure

• Tag and number every trap so each survey covers the whole population.
• Test with ultrasonic and temperature tools.
• Report failed traps as a share of the population, split into failed open and failed closed.

What moves it

• A testing interval matched to trap pressure.
• Ranking each survey by loss rate so the biggest leaks get fixed first.
• Repairing failed traps as they are found rather than at the next shutdown.

Price the steam your failed traps waste with the Steam Trap Loss calculator.

Utility energy intensity

System-wide utility intensity ties the KPIs together. Track compressed air as kWh per unit of production, cooling as ton-hours per unit, and steam as pounds per unit, then trend each month over month against output.

A ratio that drifts up while production is flat signals leaks, fouling or failed traps before the energy bill confirms it. Benchmark internally first, since cross-plant comparisons hide process differences.

The goal is a downward trend of 2 to 5 percent a year no cited source, driven by the specific levers above rather than one-time capital projects. For comparison, DOE's Better Plants partners pledge to reduce energy intensity throughout their U.S. manufacturing operations by 25 percent over ten years⁷, which works out to roughly 2.5 to 3 percent a year.

How to measure

• Normalize each utility by the same production unit every month.
• Plot each ratio against output so volume swings do not hide waste.

What moves it

• Leak, trap and fouling fixes.
• Setpoint discipline on header pressure, chilled water temperature and excess air.

The Compressed Air Leak Loss calculator takes production units as an input, so it can express leak cost per unit made.

How to build your own baseline

Turn the KPIs into a scorecard with owners and a review cadence. Plants that hold the top quartile on these numbers spend 20 to 35 percent less on utilities than median plants running identical equipment, purely through measurement and control discipline no cited source.

• Meter the big three continuously: compressor kW and flow, chiller kW and tonnage, boiler fuel and steam.
• Review specific power, kW per ton and boiler efficiency every week.
• Run leak and trap surveys on a fixed schedule.
• Give header pressure, chilled water supply temperature and excess air each a setpoint someone defends.
• Compare against your own trend first and published ranges second.