Is steam trap monitoring worth it?
Work it out with your own numbers.

The short answer

The short answer. It depends on your plant. Mostly it depends on how fast you repair. Take the example on this page: 100 traps at US-average gas. The installed cost pays back in 24.6 months on fuel only and 11.4 months on all lines. All lines means fuel, the surveys you stop buying, the production stops you avoid and the hammer damage you stop repairing. The payback calculation on this page uses these assumptions, plus your own trap count and gas price. The two figures are not the same kind of number. The first counts fuel and surveys only. That is money you can check against your own bills. The second adds an assumed value for the production stops you avoid, starting from published reference costs for an hour of lost output. If those do not fit your plant, clear the box or type your own. Cheap gas means a longer payback. Repair slowly enough and the fuel saving alone never covers the cost. That is the condition. Monitoring does not save energy. Repairs do.

The variable that matters

A failed-open trap burns the boiler’s fuel every hour it runs. What it costs the plant comes down to one thing: how long it runs before anyone knows.

The Department of Energy is clear about how often traps should be tested: weekly to monthly on high-pressure lines, monthly to quarterly at medium pressure, and only once a year below 30 psig.^[1] It calls a system well kept when under 5 percent of traps are leaking. Almost no plant tests that often by hand. Most plants that check at all check once a year. Illinois has written down what that costs. Its technical reference manual is the rulebook its utility programs must calculate savings with. That manual assumes a failed trap goes unnoticed for half the year on average. And it puts the fraction of traps found leaking or blowing through at each audit at 16 percent, on industrial and process systems at 15 psig and above.^[2] Where no program exists, DOE puts the failed fraction at 15 to 30 percent after 3 to 5 years.^[1]

SteamIQ tests each trap 60 times an hour, 525,600 times a year, and reports three times a day; a survey tests once a year. The monitor confirms a failure across several readings before it tells the plant. So the value of monitoring is simple: how much sooner you find out, multiplied by what the failed traps are losing.

The loss rate, measured

SteamIQ puts a dollar figure on every monitored trap, every day. It starts with how much steam is passing through the trap’s opening at line pressure. That becomes therms of gas. Therms become dollars at the customer’s own state gas price. On top sits DOE’s 1.30× multiplier, which covers water treatment, blowdown, labor and maintenance. That is the cost of making steam beyond the fuel itself. The same therms convert to emissions at 11.7 pounds of CO₂ each.^[3] Here is how that loss is spread across the traps we have confirmed failed open:

The average is more than three times the median, the trap in the middle of the range. That is because most of the loss sits in a few big failures, at high pressure through a large opening. For planning across a whole plant, the average is the right number to use. And the traps at the top of that spread are exactly the ones monitoring finds first. At a 106-trap pharmaceutical site watched for six months, five traps accounted for three-quarters of 15,155 therms lost.^[4]

Monitoring does not save energy. Repairs do.

“If you knew better, you’d do better.” No plant wants to waste steam. Nobody fires a boiler to humidify the outdoors. Traps stay failed because nobody knows they have failed. And once somebody does know, they stay failed because nobody knows which of the hundred to fix first. Monitoring closes the first gap. What the plant does with that knowledge decides the second.

A monitor that spots a failed trap saves nothing until someone replaces the trap. What you save is the time you cut off: how much sooner you knew, less how long you take to act. That makes this a management decision as much as an engineering one. Trap repair has to stay on the priority list. Maintenance has to work that list. And the parts have to be on the shelf. SteamIQ tracks the two numbers that show whether it is happening: how long from detection to inspection, and from inspection to repair. Both should be falling.

The parts on the shelf. One of our pharmaceutical customers built its trap program around a single decision: buy the replacement parts before they are needed. The team aims to repair within three to seven days, and sometimes swaps a trap the same day. The parts are there, and the work stays in-house. Where a trap cannot be isolated, the repair waits for a scheduled shutdown. The commitment is to remove the delays that can be removed, and to be straight about the ones that cannot.^[5]

That customer has stopped paying for contracted annual surveys. The Department of Energy calls a system well kept when under 5 percent of its traps are leaking; fewer than 5 percent of this customer’s traps are failed at any one time. That counts failed open and failed closed, which is the stricter test. That is the backlog prompt repair leaves behind, not a claim that only 5 percent fail in a year. A utility funded the first installation in 2017 and can see the account to check the records, and at eligible sites it pays for the therms saved, one more reason to repair quickly. Most of this customer’s sites get no utility money at all. It has since taken SteamIQ international and written it into its specification for new builds.^[5]

How trap repair compares to other energy projects

Since 1981, university engineering teams funded by the Department of Energy have assessed plants and written up what to fix. The program has recorded 170,055 recommendations across 22,916 assessments. Each one carries an estimated cost, an estimated annual saving, and a follow-up on whether the plant did the work. Ranked by how fast those estimates pay back, steam trap repair is 10th of 126 measures at 3.1 months. It is 4th on how often plants actually do the work.^[6] For comparison, variable-speed motor drives, lighting upgrades and heat recovery on the boiler flue all take about a year and a half. Of the nine measures ahead of trap repair, three are changes to how you buy gas or electricity, four are turning a dial, and two are fixing leaks. Not one is equipment you have to buy.

One catch: repairs do not last. Illinois credits a trap repair with six years of savings. It credits lighting or drives with fifteen.^[7] Traps fail again, so the savings from a one-off repair campaign fade away. Monitoring is what stops the fade. Berkeley Lab lists “automatic steam trap monitoring” as a measure in its own right, at a one-year payback.^[8] The federal energy office measured the same effect. In buildings with no trap maintenance program, close to 20 percent of steam was lost through traps. With an average program, about 6 percent. With continuous monitoring, it concluded, “the loss rate should approach 0%.” That same report says continuous monitoring is probably not worth it in most federal buildings, because their heating traps are small. A process plant is the opposite case.^[9] The trap programs it measured paid back in three weeks to three months.

From readings to a job list

The harder problem is turning a signal into a work order. A trace on a screen tells an expert a great deal. But reading a trap’s sound at a given pressure and load, spotting when the trouble started, and deciding who to send where first: that is specialist knowledge. The specialist is rarely the person at the screen. That gap, between knowing and doing, is where most of the energy stays lost. SteamIQ is built to close it. The system does the interpreting and hands the plant the job.

Each monitor records sound, cycle count and temperature every minute, and shock continuously, for the life of the trap. Over months that builds a full history of every trap: when the trouble started, how the load moves through the year, and what each repair actually changed. The system works through that history before a word of the Fleet Report is written. Here is what it works out for you, and what a screen full of traces would leave you to work out yourself:

• When it started, and what it has cost. Not just that a trap is failed open, but since when, what it has cost so far, and how fast the bill is growing. A trap seven days into failure is a different job from one at thirty-one days. The Fleet Report says so, and explains why acting early caps the loss.^[10]
• A real slide, or just catching up. When a fleet health score drops, the system separates new damage from failures it had been tracking all along and has now confirmed. A part month is marked as a part month, not read as a trend.
• Traps that fail together. When several traps show the same trouble in the same window, the system names the likely shared cause: a common condensate return, a pressure zone, an installation crew, dirty steam. It warns that replacing those traps without fixing the shared cause invites the same failure again. On one campus system, seven traps that failed together were flagged as a header problem, not seven separate faults.^[10]
• Jobs that belong on one visit. A failed-closed trap in the zone next to a water hammer event is put on the same inspection visit. The Fleet Report makes the connection.
• When to act, and what waiting costs. Gas is priced by the customer’s own state and month. Massachusetts industrial gas swings more than 100 percent between its April peak and October trough. Where the swing is that steep, the Fleet Report projects what leaving today’s failures unrepaired will cost through the coming months, and says when acting is cheapest.^[3]
• Whether the repair took. Every repair is checked against the trap’s own readings afterwards. A repair that did not take is called out, with another look recommended. That looks like $39 a day before and $37 after.^[10]
• Where it is not sure. A trap showing a big loss signal but no daily loss is flagged as a disagreement to settle before anyone is sent out. Where the evidence is thin, the Fleet Report says so.^[10]

The Fleet Report is built from figures already checked against the database, and every figure is checked again after the words are written. A report whose money figures disagree with each other by more than 15 percent is not released. Traps are classified by how much of the opening is leaking, not by computed flow: a ¾-inch opening at 15 percent passes more steam than a 3/32-inch opening at 40 percent. And a failure is confirmed across several readings before the plant is told.

What lands on the plant manager’s desk is short. A summary they can read in two minutes. A ranked repair list with location, estimated cost and payback in days. Jobs sorted into this week, this month and this quarter. And a running cost of doing nothing on every open item. A daily note and a weekly digest carry the same findings at shorter length. None of it asks you to read a trace.

You can also just ask. An operator can ask in plain language which traps failed this quarter, what a particular header cost last month, whether a repair took, or what a classification means. The answer is drawn from the same checked database the Fleet Report is built on, not from a general model’s memory. So using the system takes no more skill than asking a question. The expert knowledge about how a trap behaves at a given pressure and load lives in the system, not in the person at the keyboard. And the plant never has to call us to understand its own data.

The data belongs to the plant. Everything we compute is available through our interface and in standard file formats: trap states, loss figures, signals, repair checks. Findings can flow straight into the plant’s own maintenance system as work orders, and into its energy or sustainability records as verified savings. For a plant that documents maintenance for compliance, the time-stamped record of detection, inspection and repair is an audit trail it did not have to build.

The payback, worked

Take 100 traps at 145 psig. Assume 20 percent are found failed at the annual survey, that survey costs $75 a trap, repairs happen 14 days after detection, and US-average gas at the worked $0.5139 per therm. For production, the boxes start at $10,000 an hour and 4 hours per stop, with annual downtime and hammer repair spend both left blank. Installed cost is $100,000. The annual fee is $5,000, which is $2,500 less than the survey it replaces. The payback calculation on this page uses these assumptions, plus your own trap count and gas price.

At those inputs, one failed-open trap loses $13.82 a day all-in. Loss today, under the annual survey, is $50,000. With SteamIQ it is $4,000. Acting sooner saves $46,000, rounded to the nearest thousand. The surveys you stop paying for cover the fee with $2,500 to spare. So the exact annual net is $48,838 on fuel only and $104,838 on all lines. The difference is $56,000 of avoided production stops, and that is an assumption, not a measurement. Payback is 24.6 months on fuel only and 11.4 months on all lines.

Two things are deliberately left out: the traps already failed on day one, and the cost of the replacement traps themselves. The day-one backlog is extra upside in the first year, and the same traps get repaired either way. That is the “Fuel only” floor. The “All lines” case starts from the published $10,000-per-hour reference, and the reader can clear it or replace it with their own value. “Beyond the fuel bill” explains it.

Repair speed is the whole story. Stretch the repair to 90 days and fuel saved drops to $28,000. That is 39.0 months on fuel only and 13.8 months on all lines. At 180 days it is $7,000: 10.7 years on fuel only, 18.4 months on all lines. At 209 days the SteamIQ case and the annual survey both carry 212 days of leaking, so fuel saved is exactly $0. Only the $2,500 survey saving is left: payback reads “not on fuel alone” on fuel only and 20.5 months on all lines. The repair lag on this table is the wait for a leaking trap. A leaking trap wastes fuel and can drag throughput, but it rarely stops a line, so it waits its turn. A trap that fails closed on equipment that matters stops or slows the line, and a plant answers that on its own urgency rather than from the trap-repair queue. That is our assumption, not a measured response time. It is why the interruption line does not move with repair lag: its planning proxy is priced on failures, not leak-days. The half factor already carries the stops that are not caught in time. If your plant is slow to answer a stopped line as well, clear the $/hour box and read the fuel-only column. Gas price matters too. Same plant, different states. Massachusetts at $1.5084 a therm is 8.7 months on fuel only and 6.2 months on all lines. Illinois at $0.8157 is 15.8 months on fuel only and 9.1 months on all lines. Texas at $0.3657 is 33.8 months on fuel only and 13.1 months on all lines.

This is a back-of-envelope check, not a full plant study. It assumes you survey at least once a year and return condensate. The failure rate is the fraction found failed open or closed at an annual survey. The model charges only the 86 percent open share a full leak-period, so the 20 percent default prices 17.2 open failures per 100 traps. A 20 percent failure rate, open and closed, sits inside DOE's 15–30 percent no-program range and below the 25.5 per 100 the SteamIQ fleet measured (22.0 open + 3.5 closed). It prices 17.2 open failures per 100 traps, above the Illinois technical reference manual's open-only 16 and below the fleet's 22.0 failed open.

The 20.68 measured therms are scaled to your pressure first, then priced at your own 12-month gas price, then multiplied by 1.30. Pressure scaling is approximate because the measured blend mixes pressures and the larger losses sit at higher pressure. The system does not create extra repairs. It moves them earlier and points them at the traps costing the most. Where utility programs exist, the $1,000-per-trap capital figure may come down.

Regulators reached the same place by a deliberately cautious route. Illinois credits continuous monitoring as a savings measure in its own right: 395 therms per monitored trap per year at 75–125 psig, after cutting its own estimate in half twice. There is money on the table, too. Ameren Illinois’s 2026 metering and monitoring program may cover the full cost, up to $30,000. Nicor Gas’s 2026 commercial rebate pays $300 per industrial or process steam trap at 15 psig and above, once the required survey is done.^[11] Check your own utility before you budget.

Beyond the fuel bill

Failed-closed traps. A trap that fails closed wastes no steam, so a leak-only monitor never sees it. But the heat exchanger or jacketed vessel behind it floods and loses capacity. And that backed-up condensate is the charge that fires water hammer. Temperature alone cannot tell a trap that is quietly working from one that has stopped. SteamIQ’s patented cycle counting picks out each individual discharge, so a trap that has gone silent is flagged, usually before its temperature has moved at all.

Production interruptions, counted once. One failed-open trap does not stop a plant; a trap that fails closed on equipment that matters can. Across the SteamIQ network, the measured counts were 22.0 failed open and 3.5 failed closed per 100 traps, a combined failure rate of 25.5 percent, through 2 September 2026. The 14 percent closed share is the closed count divided by the combined open-and-closed count, rounded.^[3] The interruption estimate (failures × 0.14 × hours per stop, starting at 4 hours) is a planning proxy for all steam-related stops, including failed-closed traps and water hammer; it is not a measured count of production stops or a measured water-hammer frequency. Your own total annual steam-related downtime replaces the proxy; it is never added to it. Failed-closed traps and water hammer share one interruption line because the same stop can be both. The estimate prices those hours at the value per hour in the box and applies 50 percent once. The production half factor is SteamIQ’s own judgment, not a published factor: it allows for one production line stopping rather than the whole plant, for capacity loss short of a full stop, and for closed traps not caught in time. The value box begins at $10,000 per hour, sourced to ABB’s 2025 study; the reader can clear it or replace it with their own value.

Where four hours comes from. The hours-per-stop input begins at 4, SteamIQ’s own starting point. A steam stop has to drain and reheat. It sits between a published one-hour all-cause average and a published 6-hour hammer outage. Neither source establishes a typical steam stop: the Siemens figure covers all causes, and the Emerson figure is one incident at one plant. The box is yours to change.

What an hour is worth. Published figures are reference points for that question, not a claim about any reader’s plant. For a manufacturer, a stopped pharmaceutical or food line can mean a lost batch and real production money. A campus on low-pressure heating steam has no production stop. For that reader, the “Fuel only” case is the whole answer unless a hammer repair bill applies.

Water hammer damage repair. The reader’s annual spend repairing hammer damage has its own line, separate from lost production time. It starts blank and counts as $0 until the reader enters their own spend; half is counted as savings, our judgment rather than a published factor. SteamIQ locates hammer; it does not prevent it. The saving comes from what location makes possible: an event is placed and the trap most likely feeding it is named, so the damage stops compounding instead of being found at the next bang. If the value per hour is left empty, repair savings still count toward the all-lines total. We claim no hammer frequency, fleet rate, probability or average cost from our fleet.

Not counted. The estimate leaves out throughput drag from failed-open traps raising return-line pressure and the 9.7 traps per 100 that ran cold (cold with no cycling, but not confirmed failed closed because the trap may be legitimately idle). Both remain upside beyond the priced case. The production estimate can also run high. It will if the network’s closed share does not hold in the reader’s fleet, if a failed-closed trap reduces capacity without stopping the line, or if the real stop is shorter than the hours entered. The production half factor is meant to absorb the last two. Safety is never priced: people hurt by a failed trap or by hammer are never given a dollar figure, on any line, in any total.

Water hammer. Closed or clogged traps let condensate accumulate in a live steam line, and that is what makes hammer the failure mode that connects a trap program to plant safety. Water is roughly 1,200 times denser than steam. Condensation-induced hammer is the dangerous form. In it, a pocket of steam is trapped against cold condensate and collapses in milliseconds. The slug of water that fills the void hits the pipe wall at pressures well above design.^[13] The damage is cumulative: cracked welds, eroded seats, bent stems, broken supports and, in the worst cases, ruptured pipe. It is also self-reinforcing. Hammer damages downstream traps, which back up more condensate, which produces more hammer. A plant that learns of it when someone hears the bang has usually been accumulating damage for some time.

Two fatal ruptures. On 18 July 2007 a Con Edison steam main ruptured at 41st Street and Lexington Avenue. New York’s Department of Public Service (NY DPS) found the cause was condensation-induced water hammer: rain and groundwater reaching the buried main had condensed the steam inside it until the section filled, one primary causal factor, and traps “nearly completely clogged with debris” were the second. One person died after suffering a heart attack while fleeing; two people were seriously burned and many others were hurt. The utility’s own costs ran to tens of millions of dollars; source 14 records them. Those are its costs, not a price on a life or a typical plant repair bill.^[14] On 22 March 2024, two crew members aboard the cruise ship Nieuw Amsterdam, calling at Half Moon Cay in the Bahamas, died after steam hammer (the Dutch report’s term for the same condensate-driven hammer) ruptured an expansion joint. The Dutch Safety Board’s October 2025 report found that the steam traps on that line had been closed (shut, not failed; the Board could not find out why), so condensate built up in the pipe, and that a valve normally left slightly open to keep the pipe warm had been fully closed, turning that pipe into a dead end. These incidents establish consequences and causes, not how often hammer will occur in your plant.^[15]

It is also intermittent, concentrated at startup and off-hours, and rarely witnessed, which is why it is so hard to manage by inspection. SteamIQ monitors carry a three-axis accelerometer sampling continuously. An impact is time-stamped, scored for severity and correlated across neighboring monitors to locate it, so the plant knows where to look. Because every event is recorded, a zone’s history of recurrence is visible. The Fleet Report for a hospital system flagged a single sub-basement event precisely because that zone had a record of them. The failed-closed trap most likely to be feeding it is named for the same visit. The method is patented and, to our knowledge, offered by no other trap monitor.

The instrument

A SteamIQ monitor mounts on the trap. It measures sound carried through the metal, temperature, and shock in three directions, on a battery designed to last 12 to 15 years. Trap health goes out three times a day over LoRaWAN, a long-range radio: no wiring, no site Wi-Fi, no pairing each device to a gateway, and one gateway serves dozens of monitors. It is sealed to IP67, rated from −40 °C to +80 °C, made in Hingham, Massachusetts, and protected by three US patents.

Is it worth it?

At the US-average inputs, yes: 24.6 months on fuel only and 11.4 months on all lines, with annual nets of $48,838 and $104,838. Take the fuel-only figure as the one you can check against your own bills, and the all-lines figure as what it becomes if the production assumptions hold at your plant. The payback calculation on this page uses these assumptions, plus your own trap count and gas price. That answer is not universal. At a 209-day repair lag, fuel saved is $0, so payback reads “not on fuel alone” on fuel only and 20.5 months on all lines. The interruption line does not move with repair lag. The condition is your own plant’s.

Trap repair is one of the fastest-paying maintenance jobs a plant can do. Forty-five years of federal assessment data say so. What eats that return is the time before anyone knows. And what wastes the knowing is the distance between a data point and a work order. Monitoring closes the first gap. A system that reads the data and writes the job closes the second. The plants that save the most are not the ones with the best data. They are the ones that act fastest on it, and this is built to make acting the easy path. “If you knew better, you’d do better.” Our job is the knowing.