Mass timber is no longer a niche material. The 2021 International Building Code’s Type IV-A, IV-B, and IV-C construction categories opened the door to timber buildings up to 18 stories, and that change rested on years of full-scale fire testing by government labs, published and re-tested in peer-reviewed journals. The data behind it doesn’t reduce cleanly to “wood burns” or “timber is fireproof.” It’s messier than either slogan, and the mess is where the useful information lives.
For architects and engineers specifying cross-laminated timber (CLT) or glulam systems, knowing what that data shows, and where it still leaves questions open, feeds directly into both design decisions and liability exposure.
What the Full-Scale Tests Were Designed to Answer
A standard ASTM E119 furnace test evaluates a loaded assembly in isolation, against a controlled fire curve. Full-scale compartment tests do something different: they set entire furnished rooms on fire and watch what actually happens.
That’s the point of the NIST and National Research Council Canada CLT compartment fire test series, run under the Fire Protection Research Foundation’s Tall Wood Buildings program. Researchers built six full-scale CLT compartments, furnished them like real apartments, and burned them, specifically to measure how much the structure itself adds to the fire once it’s underway.
The USDA Forest Products Laboratory later ran a similar test on an actual two-story mass timber structure. Thermocouples, heat flux transducers, gas analyzers — the works. That data fed directly into the ICC’s tall wood buildings code review.
Char Rate: Predictable, But Not the Whole Story
Mass timber chars predictably. That’s one of the more consistent findings across this body of testing. Structural engineers already lean on this: a design char rate, drawn from Forest Products Laboratory testing on Douglas fir, spruce, and other species, lets them calculate the sacrificial layer protecting the remaining structural cross-section in a fire. It’s the basis for the fire-resistance calculations built into the National Design Specification for Wood Construction.
Real fire exposure complicates that number. A 2025 fire-performance review pulled together testing showing charring around 0.51 mm/min on ceiling surfaces and 0.65 mm/min on walls, rising toward roughly 1.0 mm/min once adhesive delamination exposed fresh wood underneath. So the design char rate and the observed char rate under real conditions aren’t quite the same number, and that gap is where the design and liability questions start.
Delamination: The Variable Design Teams Can’t Ignore
If there’s one finding from the NIST/NRC series that design teams should sit with, it’s delamination: charred lamella layers falling off once the glue line weakens under sustained heat. Fresh wood gets exposed mid-fire. Heat release rate jumps a second time. In NIST’s documented test scenarios, this sometimes meant the fire never self-extinguished, even after the furnishings had fully burned out.
That result reshaped how adhesives get tested. Fire-induced delamination can be reduced substantially by switching to adhesives formulated to resist heat-driven bond failure, and a standardized screening method now exists for exactly this — checking whether an adhesive is prone to delamination before it ever gets specified on a project.
Exposed Timber Surface Limits Are Data-Driven, Not Arbitrary
The IBC’s cap on exposed CLT within a compartment isn’t a guess pulled from thin air. It traces back to heat release rate data from these burns. European testing backs it up too. The 95 m² FRIC compartment fire experiments found the same pattern showing up in the North American tests: more exposed CLT surface area means higher peak heat release and a longer-burning fire. Current code pathways respond to exactly that relationship, capping exposed timber area and requiring encapsulation on the surfaces left over.
What This Means for Design and Risk Management
A few things follow from this for firms designing or reviewing mass timber projects.
Char rate tables based on species alone don’t capture what happens under real exposure. Duration and adhesive performance change the number, so calculations need to account for both.
Adhesive choice deserves the same attention as panel specification, maybe more. Specifying a delamination-resistant adhesive and documenting that decision is quickly becoming standard practice, and it’s cheap insurance against a question nobody wants asked after the fact.
And connections are where things go wrong. Across these full-scale burns, protected surfaces hold up the way they’re designed to. It’s the gaps, joints, and exposed transitions at connections and penetrations where testing keeps finding trouble.
Performance-based fire engineering for mass timber is still catching up to the material’s popularity. Reading the actual test data, rather than the shorthand version of it, is what lets a firm stand behind its design decisions later.
