Industries / Industrial equipment and services / Industrial services / Heat exchanger services
Cleaning
Cleaning removes scale, slime, coke and other deposits from exchanger tubes and plates, either with the exchanger taken apart or while it keeps running.
4
Milestones achieved, from 1953 to 2019.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1953
Balls clean tubes while running
Before this, condenser tubes were brushed, drilled out or dissolved clean only when the condenser was shut down or running at reduced load, as the patent itself describes. On 2 February 1953 Josef Taprogge of Essen, Germany, applied for a patent, granted in the US on 6 August 1957, in which slightly oversized rubbing bodies ride the cooling water through the tubes, are caught by a screen at the outlet and are pumped back to the inlet, over and over, while the condenser runs. The company dates its first installation to 1949 and says more than 18,000 of its systems are now in use.
Credit
Josef Taprogge; Taprogge, the company Ludwig Taprogge founded in 1953
2015
Sponge balls cut chiller power
Consultants at Pangolin Associates in South Australia compared a year of electricity use before and after sponge-ball tube cleaning systems were fitted to the water-cooled chillers of four office buildings in central Sydney. The systems send foam balls through the condenser tubes every 30 minutes. After correcting for weather with cooling degree days, use fell by 24.5 to 26.5%, in line with a 2002 Korean study that predicted about 24% with the Kern and Seaton fouling model and measured 26%.
Lessonthe balls need full water flow; when one building fitted a variable-speed drive that cut the speed of water in the tubes by almost half, the balls stuck in the tubes, and in another, 12 misaligned tubes trapped all 20 balls.
Credit
D. P. Ross, P. A. Cirtog and A. Swanson, Pangolin Associates
2019
Sound cleans whole bundles
Braskem, a petrochemical maker, bought one of the largest ultrasonic cleaning tanks then available, 7.7 m long, 1.9 m wide and 1.9 m deep, for exchanger bundles from its plants in Bahia, Brazil. In an ultrasonic bath, sound waves in water make tiny bubbles that collapse against the tubes and knock deposits loose, including on the shell side that water jets reach poorly. Braskem reported verified gains of about US$1.6 million a year from one exchanger, 90% less cleaning water and cleaning times cut by up to half.
Lessonthe authors' experience is that hydroblasted bundles often go back into service below 80% of design performance, and the fouling left in place speeds the next round of fouling.
Credit
C. H. Schulz of Braskem, with B. Kieser of Tech Sonic, the method's supplier and co-author
2019
Inserts keep fouling flat
Tube inserts are devices fitted inside each tube that turn or vibrate in the flow, which helps keep deposits from settling. At TOTAL's Grandpuits refinery in France, monitoring software tracked two exchangers of similar build and flow on the crude oil preheat train, one with plain tubes and one with tube inserts; the refinery uses Petroval's Turbotal and Spirelf inserts. Fouling barely rose in the unit with inserts, even though its tube walls ran hotter.
Credit
Petroval (inserts); TOTAL and Heat Transfer Research, Inc. (HTRI) (measurement)
Watch for
Balls stop river ice
A large heat pump that takes heat from cold river water risks freezing the water inside its exchanger, which normally limits use to river water above about 4 °C. At the Überseeinsel district in Bremen, Germany, four heat pumps of 5 MW in all went into service in May 2025 drawing heat from the Weser. The developer says it uses river water as cold as 0.1 °C, partly by turning some of it into ice slurry in a vacuum, and that small moving metal balls keep the heat exchangers free of ice. Taprogge, whose Klaren exchangers keep a bed of moving particles in the tubes, says they allow heat recovery there at very low water temperatures and calls it a world first.
Achieved whena full winter of operating records, with river temperature, heat delivered and hours lost to icing or cleaning, is published and checked by someone outside the two companies.
Claimed by
Taprogge and the developer Stadt.Energie.Speicher, May 2025
Open
01
Measure what cleaning restores
A public table of how much heat-transfer performance exchangers get back from each cleaning method (hydroblasting, chemical cleaning in place, ultrasonic baths, thermal cleaning), worked out from the temperatures and flows plants already log. For each job it gives the fouling resistance just before shutdown, just after restart and 30 days later, with exchanger type, fluid and deposit.
The briefCredit · unclaimed02
Track where the balls go
A clear-plastic bench model of a condenser's inlet box and tube bundle, run with water at low pressure, in which sponge cleaning balls circulate and cameras at the outlet count every ball leaving every tube. It publishes the share of balls each tube receives at several flows and ball loads, and tests simple fixes such as balls of different densities.
The briefCredit · unclaimed
Open 01 · the brief
Measure what cleaning restores
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of how much heat-transfer performance exchangers get back from each cleaning method (hydroblasting, chemical cleaning in place, ultrasonic baths, thermal cleaning), worked out from the temperatures and flows plants already log. For each job it gives the fouling resistance just before shutdown, just after restart and 30 days later, with exchanger type, fluid and deposit.
Why it is still open
The figures in print come from firms that sell one method. Clean As New, an ultrasonic cleaning firm, wrote in 2024 that with hydroblasting "in many cases, it is not possible to return the bundle to a zero-fouling condition" and that there is "no adequate method to determine the progress of the cleaning". A 2022 paper by Tech Sonic, Clean As New and Hexxcell found ultrasonic cleaning recovered about 20% more performance than hydroblasting, on trial data from US operators that it does not release. A performance-based way of "defining cleaning completion" was first presented by Clean Harbors and Dow in June 2026, as a talk title without data. Hours and water per job are a different question, left to industrial cleaning.
heatexchanger-fouling.comheatexchanger-fouling.comheatexchanger-fouling.com
Built with what exists
Historian exports (flows and the four inlet and outlet temperatures) from cooperating plants; the open Python libraries ht and CoolProp for exchanger and fluid-property calculations; pandas; WebPlotDigitizer to read before-and-after values off published graphs; a document-reading AI model to pull job details from cleaning reports, with a person checking every row (Notes: "Seven more flagged").
First step
Digitize every before-and-after fouling figure in the free Heat Exchanger Fouling and Cleaning conference papers from 2015 to 2024 into the first rows of the table.
Done looks like
At least 100 cleaning jobs from at least three plants and three methods, all computed with the same published code; a second analyst recomputes 20 jobs from the raw data and agrees within 10%; medians and ranges of recovered performance by method published.
Open 02 · the brief
Track where the balls go
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A clear-plastic bench model of a condenser's inlet box and tube bundle, run with water at low pressure, in which sponge cleaning balls circulate and cameras at the outlet count every ball leaving every tube. It publishes the share of balls each tube receives at several flows and ball loads, and tests simple fixes such as balls of different densities.
Why it is still open
Makers and users disagree, and no count was found. Taprogge wrote in 1998 that "almost identical hydraulic conditions within the tubes guarantee uniform distribution". Hitachi's 1984 patent says some tubes get enough balls "while the others can not, mainly due to the influence of state of flow of water", so "it is impossible to clean all cooling tubes equally". A 2017 CQUniversity study modeled water speeds in the inlet box at Callide Power Station, Queensland, to predict where balls would go, by simulation only. Searches found patents and simulations, not measured counts per tube.
prozesstechnik.industrie.deimage-ppubs.uspto.govacquire.cqu.edu.au
Built with what exists
Clear acrylic tubes and a laser-cut tubesheet; a pool pump with a flow meter; commercial sponge-rubber cleaning balls 1 to 2 mm wider than the tube bore; two phone cameras at 240 frames a second; OpenCV and an Ultralytics YOLO model to detect balls; Python for the counts.
First step
Build a five-by-five tube test section and count 1,000 ball passages by hand from video to settle the counting rules.
Done looks like
At least 10,000 ball passages counted at three flows; the share per tube and its spread published; the automatic count checked by hand on 500 passages and at least 99% right; video, counts and drawings released.