construction · materials · family: it worked in the lab
when is wettimber dry enough?
No one can reliably predict — or cheaply verify — when a rain-soaked CLT floor is dry enough to seal
Problem statement
Cross-laminated timber (CLT) floors and roofs go up fast and get rained on; the end grain at panel edges and penetrations "absorbs water more quickly than its surface," and once a concrete topping, acoustic mat or low-permeance membrane goes on top, trapped water can take "up to a year or longer" to leave. WoodWorks' guidance sets a maximum of 16% moisture content before those layers are installed, and the trade treats readings above 19% as wet. The problem is that a contractor has no trustworthy way to know when the interior of a 5- to 7-ply panel has reached that threshold: earlier hygrothermal-simulation studies of CLT (Lepage; McClung et al., cited as refs 5–7 by Kordziel et al. 2018) could only reproduce measured drying after "the coefficient for liquid water redistribution had to be adjusted by over 10 orders of magnitude," a situation Kordziel et al. say "may leave designers lacking confidence in the ability of conventional models to predict drying behavior of mass timber" (their own alternative — a humidity-dependent vapour-diffusion resistance factor with a halved liquid-suction constant — avoided that step and agreed "reasonably well" with measurements, but still slightly under-predicted drying rates), and the embedded sensors meant to replace prediction with measurement mostly die: on the eight-story Portland building studied by USDA Forest Products Laboratory and Colorado School of Mines, "104 sensors were installed and only 38 remain functional after about 4 years," many lost to ponding water in the very sensor pockets cut to hold them.
Why this matters
Tall mass timber (6–20 stories) is becoming more common in North America and is being built as a low-carbon alternative to steel and concrete frames, yet "modern mass timber buildings have not been around long enough to provide validation data from real projects." Prolonged elevated moisture causes "dimensional instability, cracking, microbial attack, and fastener corrosion," and TallWood Design Institute researchers report that "lack of data and models" on how construction wetting affects long-term connection durability "is a significant hindrance to the understanding of the long term behavior of structures." The good news from Portland — every monitored location eventually reached a stable 10–15% MC, so buildings "can naturally recover from construction wetting provided that such buildings are properly enclosed and further moisture intrusion is prevented" — sharpens rather than resolves the problem: some locations were still above 19% a year after construction and took about 18 months to dry, so the risk sits entirely in the decision of when to enclose, and that decision is currently made on pin-meter spot checks and schedule pressure.
What’s been tried and why it hasn’t worked
Field monitoring programs (Portland eight-story; UBC Brock Commons; Oregon State's Peavy Hall; others in Ontario and the U.S. South) have deployed pin-type resistance sensors — calibrated in the lab to 1.6–2.1% MC RMS error — inside CLT panels, but installation methods are "currently inconsistent across building case studies," sensor pockets notched into panel tops collect water and destroy the sensors, and readings saturate at 40% MC and become insensitive below 10%, so the datasets are patchy exactly at the wettest and driest moments. Hygrothermal simulation (WUFI-class models) with lab-measured CLT properties matches laboratory wetting/drying only after parameter tuning that is either unphysical (the ten-orders-of-magnitude liquid-redistribution adjustment in earlier studies) or empirical (Kordziel et al.'s halved liquid-suction constant, which still slightly under-predicted drying rate), possibly because water moves through CLT along glue lines, board gaps and cross-lamination checks in ways a one-dimensional homogeneous-material model does not represent. Trade guidance therefore falls back on conservative thresholds and protection practices — cover panels, seal end grain, avoid ponding — which cost schedule and money and still cannot tell a superintendent that a specific bay is at 16% at mid-depth today. Existing research is well funded (a four-year USDA- and TallWood-funded program was due to finish in 2022) but is characterising behaviour, not delivering a field verification method.
What would unlock progress
Two complementary advances would close the loop: (1) a rugged, low-cost, installable-from-below or non-contact way to read moisture profile through a thick CLT panel — capacitance/impedance tomography, time-domain reflectometry, or microwave/NIR reflectance calibrated on real panels — that survives construction traffic; and (2) a drying-time predictor for CLT that accounts for edge and glue-line pathways and can be run from a few readings plus weather. Adjacent precedents: grain-storage and lumber-kiln moisture sensing already read bulk wood moisture non-invasively at scale, and geotechnical/concrete practice uses embedded RH probes with far better survival rates than notched pin sensors.
Entry points for student teams
A team with a few CLT offcuts could benchmark non-invasive moisture-reading approaches (handheld capacitance meters at multiple depths, low-cost impedance electrodes, or a microwave sensor) against gravimetric moisture during controlled wetting and drying, focusing on end-grain edges where the risk concentrates. A modelling team could fit a simple two-pathway (bulk plus edge/glue-line) drying model to the published Portland time series (printed as charts in the papers; the underlying data are available from the corresponding author on request) and test whether it predicts hold-out sensor traces without the ten-orders-of-magnitude fudge. A construction-management team could design and cost a "ready-to-enclose" verification protocol (sampling locations, depths, thresholds) for a real jobsite. Relevant skills: wood science/building physics, sensor design, hygrothermal modelling, construction management.
Genes
Genes marked “+” were added in a later review.
Structural cousins: stuck the same way, in other fields
Sources
Kordziel, S., S. V. Glass, S. Pei, S. L. Zelinka, P. C. Tabares-Velasco, "Moisture Monitoring and Modeling of Mass Timber Building Systems," WCTE 2018 (World Conference on Timber Engineering), USDA Forest Products Laboratory, accessed 2026-08-17; Pei, S., J. Stogdill, S. V. Glass, S. Zelinka, S. Kordziel, P. C. Tabares-Velasco, "Long-Term Moisture Monitoring Results of an Eight-Story Mass Timber Building in the Pacific Northwest," Journal of Architectural Engineering 27(4), 2021, DOI 10.1061/(ASCE)AE.1943-5568.0000497, accessed 2026-08-17; "Mass Timber Moisture Management for Construction," WoodWorks – Wood Products Council, accessed 2026-08-17; Schmidt, Evan, "Moisture and Mass Timber," STRUCTURE magazine, accessed 2026-08-17 go to source 1 ↗ go to source 2 ↗ go to source 3 ↗ go to source 4 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Primary sources are peer-reviewed/proceedings research from USDA Forest Products Laboratory and Colorado School of Mines (tier 1); WoodWorks (industry technical guidance) supplies the 16%/19% thresholds and the "up to a year or longer" drying statement; STRUCTURE magazine (NCSEA/CASE/SEI professional publication) supplies the TallWood Design Institute quotations. The "installation methods are currently inconsistent across building case studies" phrase is from the abstract of an ASCE Journal of Architectural Engineering review (2024, DOI 10.1061/JAEIED.AEENG-1638) seen via search results only — full text not read. Sensor saturation at 40% and insensitivity below 10% are from the Pei et al. 2021 technical note. The claim that no field verification method exists (as opposed to threshold guidance) is the author's synthesis; the buildings-recover finding means this brief must not be read as "mass timber has a moisture problem" but as "the enclosure-timing decision lacks a measurement." `temporal:newly-created` because the problem arrived with code-permitted tall mass timber (recent); `failure:tech-limitation-now-resolved` was considered and rejected — nothing has been resolved. Related collection brief: `construction-mass-timber-tall-building-fire-gap` (fire testing) — same material class, distinct problem.
Verifier note (2026-08-17): Kordziel et al. WCTE 2018 (6 pp) and Pei et al. 2021 (JAE technical note) fetched from FPL and read; "10 orders of magnitude", "lacking confidence", "104 sensors ... 38 remain functional", 40% sensor ceiling, insensitivity below 10% MC, 1.6/2.1% RMS calibration, ponding in notched sensor pockets, 18-month drying and "naturally recover" all located verbatim; WoodWorks 16% threshold / "up to a year or longer" / end-grain statement and STRUCTURE (Sinha) "significant hindrance" quotation confirmed on the live pages. The 10-orders-of-magnitude claim was hedged in place: it applies to the earlier studies Kordziel et al. cite, not to Kordziel et al.'s own model, which used a different (still empirical) adjustment.
Verified at intake 2026-08-17: gate (net) + adversarial source check + contested-tag second coding.
Reconciliation 2026-09-29: All four sources in the Source line were read in full on 2026-09-29. The two Forest Products Laboratory addresses no longer deliver the papers; they now redirect to a Forest Service landing page. The same two PDFs were read from the Internet Archive copies of those addresses (web.archive.org, captures of April 2026). The WoodWorks and STRUCTURE pages were read at the cited addresses. The abstract of the 2024 review named above was read through OpenAlex; its full text was not read. Changes: (1) The text said Kordziel et al.'s own approach was "adjusting the vapour-diffusion resistance factor instead." The paper says the vapour-diffusion resistance factor was set from measurements as a function of relative humidity, and the value that was adjusted was the constant in the liquid-suction equation, changed from 3.8 to 1.9. The paper also says the calibrated simulations agreed "reasonably well" with measurements and that drying rates were "slightly under-predicted." Both places that describe this approach now name the liquid-suction constant and say "slightly"; the Problem Statement also says "reasonably well." (2) The text said water movement along glue lines, board gaps and checks was the reason the models need tuning. The paper says only that water filling gaps between boards was the most likely reason for one higher uptake value. "Because" is now "possibly because." (3) The text said tall mass timber of 6–20 stories "is now permitted under U.S. codes." None of the sources says this. The 6–20 figure is from Kordziel et al., who say such buildings are "becoming more common around the world including North America." The sentence now says that. (4) The text said the decision to enclose is made on "surface pin-meter spot checks." WoodWorks describes readings with pin probes "or extension probes for deeper measurements." The word "surface" was removed. (5) The text said a USDA/TallWood multi-year program "is under way." STRUCTURE describes two grants, $500,000 from USDA and $250,000 from the TallWood Design Institute, for a four-year project with "expected completion in 2022." The sentence now says the program was due to finish in 2022. (6) The entry point that uses the Portland time series now says the series are printed as charts and that Pei et al. offer the underlying data from the corresponding author on request. Confirmed against the sources and left as written: both Source line citations for the Forest Products Laboratory papers; the WoodWorks and STRUCTURE titles and author; the quotations on end grain, "up to a year or longer," ten orders of magnitude, designer confidence, 104 and 38 sensors, validation data, the four kinds of moisture damage, natural recovery, and the "significant hindrance" statement by Arijit Sinha of Oregon State University; the 16% threshold (WoodWorks) and the 19% threshold (Pei et al., who call readings above 19% "considered wet for wood engineering"; the WoodWorks page as read today does not give a 19% figure); the 10–15% stable range; about 18 months to dry; the 1.6% and 2.1% calibration errors; the 40% sensor ceiling and low sensitivity below 10%; ponding in the notched sensor pockets; Brock Commons and Peavy Hall as monitored buildings; and the phrase "currently inconsistent across building case studies," which in the review's abstract follows the words "In situ moisture monitoring deployment techniques are," so the words "installation methods are" in front of it in the first paragraph of these notes are not the abstract's. NOT checked, because they are not in the four sources and may rest on the full text of the 2024 review: monitoring programs "in Ontario and the U.S. South"; panels of up to 7 plies (the Portland panels were 5-ply); and the comparison of sensor survival with grain-storage, lumber-kiln, geotechnical and concrete practice. These are left as written. The headline claim that no one can reliably predict or cheaply verify dryness is the author's reading, as the notes above already say. Genes untouched.