Industries / Food, farming and forestry / Forestry and logging / Timber and sawmills / Lumber treating
Pressure treatment
Forcing preservative deep into wood by sealing it in a steel cylinder, drawing air out with a vacuum or packing it with compressed air, and pushing the liquid in under pressure.
6
Milestones achieved, from 1838 to 2023.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1838
Creosote forced in by pressure
John Bethell, a London solicitor, "took out a patent for his process of preserving timber from decay by impregnating it with creosote oil" on 11 July 1838, by the account of his 1868 obituary for the Institution of Civil Engineers. As the US Forest Service's William Sherfesee described the method in 1909, timber on rail trucks went into horizontal cylinders up to 9 ft (2.7 m) across and more than 150 ft (46 m) long, was steamed, held under a vacuum, and then flooded with creosote at about 160 °F (71 °C) that pumps forced in until the wood had taken up the amount wanted. This full-cell process leaves the wood's cells full of oil. Sherfesee called it, with its zinc chloride twin Burnettizing, the standard "both in this country and in Europe" against which other processes were compared.
Credit
John Bethell
1865
Shallow creosote, cut piles
By the account of Bernhard Fernow's 1887 report for the US Department of Agriculture, creosoting "was first applied" in the United States in 1865, by an engineer named Hinckley on piles for the Old Colony Railroad in Massachusetts. The oil reached only 1/4 to 1/2 inch (6 to 13 mm) below the surface, and the piles were trimmed and cut after treatment, exposing untreated wood. When they were examined in 1878, 200 piles still stood but were not sound, while hemlock ties creosoted at the same time were sound. Fernow wrote that every US failure "can be shown to have resulted from imperfect treatment".
LessonCutting or boring wood after treatment opens wood the preservative never reached, so the US Forest Products Laboratory still advises doing all cutting and boring before treatment.
Credit
Hinckley, for the Old Colony Railroad; account by B. E. Fernow, USDA Forestry Division
1885
Engineers rank the woods
A committee of the American Society of Civil Engineers reported to the society's convention in June 1885 that the method chosen to preserve wood "depends almost wholly upon its proposed subsequent exposure": creosote against marine borers and in very wet places, and Burnettizing, forcing in a zinc chloride solution, as "the advisable process to use for ties at present in this country" because creosote cost too much away from the coast. It told railroads to "select the more open-grained, porous, and sappy varieties of wood", since preservatives "penetrate but little" into white oak, bur oak and heart pine and are "of doubtful utility" for white pine, chestnut and spruce. At Las Vegas, New Mexico, the Atchison, Topeka and Santa Fe Railroad's new tie-treating works, built on plans partly supervised by Octave Chanute, kept a record of every operation and cost that Fernow called more complete than any he knew of.
LessonWhich wood goes into the cylinder decides the result as much as the process does, because some species take almost no preservative.
Credit
the ASCE committee on the preservation of timber; Octave Chanute
1902
Less oil, same depth
Max Rueping of Charlottenburg, Germany, patented in the United States (US 709,799, filed 31 March 1902, granted 23 September 1902) a method that first fills the wood with compressed air, then forces the preservative in at a higher pressure. When the pressure is released, the trapped air expands and pushes out the liquid "which does not adhere to the walls", so that, in the patent's words, it uses "less impregnating liquid than has previously been possible". Cuthbert B. Lowry of Lexington, Kentucky, patented a version without the compressed air (US 831,450, granted 18 September 1906): ties are soaked in creosote under pressure, the free oil is drained, and a quick vacuum pulls much of the rest back out. These empty-cell processes coat the cell walls instead of filling the cells. The US Forest Products Laboratory (FPL) puts the preservative recovered at 20% to 60% of the gross amount injected.
Credit
Max Rueping; Cuthbert B. Lowry
2018
Incising tested for 40 years
Incising runs lumber through rollers whose teeth cut rows of slits, usually 13 to 19 mm deep, so that preservative can travel along the grain from each cut; it is used mainly on Douglas fir, western hemlock and western larch, and it can reduce strength (FPL, 2010). Mark Mankowski, Stan Lebow and Grant Kirker of FPL reported in 2018 on heartwood stakes of Douglas fir, Engelmann spruce and southern pine, incised or not, treated with two copper-arsenic preservatives (CCA and ACA) at three strengths and set in the ground in Mississippi in 1975 and Wisconsin in 1976. After 40 years, incising had improved southern pine and Engelmann spruce heartwood, mostly at medium and high retentions, but "had little effect on Douglas-fir heartwood". Some low-retention stakes did worse in Wisconsin, rated a lower decay hazard, than in Mississippi, which the authors put down to soil, microclimate and different decay fungi.
LessonIncising helps some hard-to-treat heartwoods and not others, and a test site's soil can matter as much as its climate rating.
Credit
Mark E. Mankowski, Stan Lebow and Grant Kirker, FPL
2023
Carbon dioxide carries the biocide
Superwood A/S in Denmark treats Norway spruce cladding and decking in a commercial plant that uses supercritical carbon dioxide, CO2 held above its critical temperature and pressure so that it dissolves organic biocides like a liquid but flows into wood like a gas, and leaves no water behind. Kyra Wood of the University of Tasmania, with Anders Kjellow of the Danish Technological Institute and researchers at Oregon State University and the University of the Sunshine Coast, sent boards of plantation Eucalyptus nitens through one of Superwood's commercial spruce charges, at 45 °C and up to 15 MPa. Conventional vacuum-pressure treatment barely enters this species' heartwood. The azole fungicides reached every zone of boards 19 to 35 mm thick, but combined retentions of 0.031% to 0.035% by mass were below the 0.06% that Australian standards set for exterior use above ground, and because the azoles carry no color, penetration could not be checked by eye as those standards require. Superwood says on its website that it is still the only company in the world using the technology.
LessonA process can carry chemicals into wood that liquids cannot reach, and still fail inspection rules written around a visible stain.
Credit
Superwood A/S; Kyra Wood, Anders Kjellow, Matthew Konkler, Gerald Presley and Jeffrey Morrell
Watch for
Treating CLT after gluing
Cross-laminated timber (CLT) is made of layers of boards glued crosswise into large panels, and US building codes do not accept it for wet or exposed service. Researchers at Mississippi State University, the preservative maker Arxada and FPL glued 3-ply and 5-ply southern pine panels and then pressure-treated them with copper azole (CA-C) or micronized copper azole (MCA). In laboratory tests the treated panels resisted termites and, as 3-ply panels, two decay fungi, but treated 5-ply panels lost about as much mass to decay as untreated ones. WoodWorks, a US mass-timber advisory group, wrote in summer 2026 that there are "no known commercial manufacturers" of standard CLT made from naturally durable species.
Achieved whenpanels treated after gluing pass field exposure tests run by a party outside the study, and a code evaluation accepts them for exterior use.
Claimed by
Bradia Henfield, Tamara França and co-authors, Wood and Fiber Science, July 2026
Open
01
Penetration read by camera
An open set of photographs of treated wood borings and cut ends sprayed with the standard copper indicator, each labeled with the penetration a trained inspector recorded, and an image model that measures the depth reached and the share of sapwood penetrated, tested on photographs from plants it never saw in training.
The briefCredit · unclaimed02
Treating charges, made public
An openly licensed set of treating charges from working plants, each with species, sizes, moisture before treatment, incising, solution strength, the vacuum, pressure and time record of the cycle, the gauge retention, and the retention and penetration later found in assay borings, with a baseline model that flags charges likely to fail inspection from the cycle record alone.
The briefCredit · unclaimed
Open 01 · the brief
Penetration read by camera
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An open set of photographs of treated wood borings and cut ends sprayed with the standard copper indicator, each labeled with the penetration a trained inspector recorded, and an image model that measures the depth reached and the share of sapwood penetrated, tested on photographs from plants it never saw in training.
Why it is still open
Penetration under the American Wood Protection Association's A3 methods is judged by eye: chrome azurol S sprayed on split borings or fresh-cut surfaces turns treated wood deep blue (AWPA A3, 1991 edition, as filed with the US Code of Federal Regulations). The one automated reader found, a 2014 US patent application by Osmose Utilities Services for an image sensor that reads the indicator's color (US 2015/0226663), was abandoned in December 2015 for failure to respond to an office action, by Google Patents' record. The 2023 Eucalyptus study above found that colorless azole treatments cannot be checked by eye at all. Searches of GitHub, Zenodo and Hugging Face in November 2026 found no public image set.
Built with what exists
A phone camera and a printed color reference card; chrome azurol S indicator mixed as AWPA A3 Method 2 describes; offcuts of copper-treated lumber, cut with gloves and a dust mask; calipers for measured depth; Label Studio for labeling; Python with OpenCV and PyTorch, or a small open-weight vision model fine-tuned on a single graphics card (the Note "One GPU to fine-tune" covers Liquid AI's LFM3 family).
First step
Cut 50 cross-sections from store-bought treated southern pine and Douglas fir, spray them, photograph each beside a color card, and measure penetration with calipers.
Done looks like
At least 1,000 labeled photographs from at least five species and three treating plants, released openly, and a model whose depth reading is within 2 mm of the inspector's on 90% of held-out photographs and whose pass or fail call agrees with the inspector at least as often as two inspectors agree with each other.
Open 02 · the brief
Treating charges, made public
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An openly licensed set of treating charges from working plants, each with species, sizes, moisture before treatment, incising, solution strength, the vacuum, pressure and time record of the cycle, the gauge retention, and the retention and penetration later found in assay borings, with a baseline model that flags charges likely to fail inspection from the cycle record alone.
Why it is still open
Retention and penetration are confirmed only after treatment, by taking borings from the charge and analyzing a zone the standard sets, such as 13 to 51 mm from the surface for southern pine poles (FPL, 2010). Plants already keep a permanent treating record for each charge under the industry's best management practices. Searches of GitHub, Zenodo and Hugging Face in November 2026 found no public set of charge records; the nearest, a 2026 EU project report by Nasko Terziev and colleagues, models laboratory impregnation of wood with a bio-based heat-storage material, not plant charges.
Built with what exists
Exports from plant control systems or scanned charge sheets; an AI document reader to transcribe paper sheets, every figure checked by hand (the Note "Seven more flagged" covers an AI reviewer of long documents); Python with pandas and XGBoost; Zenodo to host the data.
First step
Write the column list and a plan for removing plant names, then ask one treating plant or a university treating laboratory for a year of charge sheets and assay results.
Done looks like
At least 2,000 charges from at least three plants with matched assay results, published with plant names removed, and a model tested on a held-out year that catches at least 80% of the charges that failed retention while flagging no more than 10% of those that passed.