Your House Frame Has 30% More Wood Than It Needs. The Code Says Use It Anyway.
Construction materials produce over 7% of global CO2 emissions. MIT's topology optimization framework just cut structural material use by up to 90% on trusses and 33% on reinforced concrete beams. It runs on a MacBook Pro. But prescriptive building codes guarantee every American house frame carries wood the structure never asked for.
Pull a 2x10 floor joist out of any tract home built in the last twenty years. It probably spans 14 feet, sits 16 inches from the next one, and carries a live load well below the IRC-prescribed 40 pounds per square foot that the code table assumed it would need to resist. An engineer calculating the actual demand on that specific joist in that specific span with that specific load path would almost certainly specify a smaller member. Maybe a 2x8, maybe an engineered I-joist at 24 inches on center. The prescriptive table does not care. It says 2x10 at 16 inches, and the framer installs 2x10 at 16 inches, because that is the row in the table that matches his span and nobody on the job site is paid to question whether the table is conservative.
It is. Deliberately, by design, and at enormous carbon cost.
Prescriptive code tables in the International Residential Code exist to let any licensed framer build a structurally adequate house without hiring a structural engineer. They standardize lumber sizes, spacing, and connection requirements for common configurations, and they work brilliantly at their intended purpose: preventing structural failures in residential construction. What they do not do, and were never designed to do, is minimize material consumption. Every entry in every table carries a safety factor that assumes worst-case loading, worst-case material grade, and worst-case span geometry, because the table cannot know what your specific house looks like and the framers reading it should not need to.
That safety margin has a weight, and a carbon footprint that compounds with every house built to the same tables.
7% of Everything We Emit
In 2022, producing construction materials generated more than 7% of all global carbon emissions, according to data compiled by MIT researchers studying structural optimization. Concrete alone accounts for 91% of a typical building structure's mass and 74% of its lifecycle global warming potential, per MDPI lifecycle assessment studies. Steel contributes just 9% of mass but 26% of warming potential, because producing a kilogram of structural steel releases 1.55 to 2.46 kg of CO2 equivalent, roughly 10 to 70 times the carbon intensity of concrete by weight.
For a standard US residential build, embodied carbon in materials and construction (lifecycle stages A1 through A5) represents approximately 21% of total lifecycle emissions over a 60-year lifespan, per Pinewood Structures analysis. Operational energy dominates at 67%, which is why so much of the green building conversation focuses on insulation, HVAC efficiency, and solar panels. But as operational energy gets cleaner through grid decarbonization and heat pump adoption, embodied carbon's share of the total rises. In new net-zero homes, embodied carbon can represent 50% or more of lifetime emissions, because there is very little operational carbon left to dilute it.
Which raises a question that almost nobody in residential construction is asking: how much of that embodied carbon is structurally necessary, and how much exists because the framing table said so?
MIT Found the Gap
In June 2026, a team led by Josephine Carstensen, the Gilbert W. Winslow Career Development Professor in Civil Engineering at MIT, published a framework in Automation in Construction that bridges the gap between theoretical topology optimization and practical, buildable structural design. Topology optimization itself is not new, and computers have been able to calculate the mathematically optimal distribution of material within a structure for decades, generating web-like, organic forms that use the absolute minimum material to resist a given set of loads. Aerospace and automotive industries use it routinely to design lightweight brackets, engine mounts, and chassis components.
Construction has not adopted it, for a reason that is not about computing power or engineering knowledge but about a much more fundamental constraint: you cannot hand a framing crew a topology-optimized design that looks like a dried coral branch and expect them to build it with standard lumber and a nail gun.
Carstensen's framework solves this by letting engineers impose constructability constraints from the start. Maximum number of members meeting at any single joint. Minimum component sizes that correspond to available lumber dimensions. Minimum angles between connected members so that real connectors can join them. And critically, multi-material support that assigns steel or timber to each structural component based on strength requirements and carbon impact, not as a blended fraction but as a discrete material choice for each element.
"There's an interplay between the materials you're using, the constructability of designs, and the optimization of the structure. You need to be able to address all three at the same time. That's what we tried to do here." — Josephine Carstensen, MIT
Zane Schemmer, the PhD student who led the implementation, tested the framework on the Lockport "Upside-Down Bridge" near Buffalo, New York, generating timber-only, steel-only, and hybrid timber-steel truss designs under varying constructability constraints. Material reduction on truss structures reached up to 90% compared to conventional designs while meeting identical load requirements. Hybrid designs placed timber where carbon savings mattered most and steel only where additional strength was structurally unavoidable.
It ran on a MacBook Pro.
From Trusses to Foundations
A companion study from the same lab, published as a preprint in April 2026 by Jackson Jewett and Carstensen, moved closer to residential construction by applying topology optimization specifically to reinforced concrete beam design. Unlike the truss study, this one fabricated and tested physical prototypes. Real beams, real loads, real failure modes.
Results: optimized beams reached loads 36% to 42% higher than conventionally designed beams using the same amount of material. Inverting that comparison, the researchers demonstrated approximately 33% material reduction potential while maintaining identical performance to current code-compliant designs, without adding structural depth. This is not a theoretical projection. These beams were built, loaded, and broken in a lab, and the optimized geometries exhibited ductile failure modes consistent with safe structural behavior.
Residential foundations use reinforced concrete beams. So do grade beams, stem walls, retaining walls, and garage headers. A 33% reduction in concrete volume across the foundation system of a typical 2,400-square-foot home translates to roughly 8 to 12 cubic yards of concrete not poured, not trucked, not manufactured. At an embodied carbon intensity of 0.034 to 0.172 kg CO2e per kilogram of concrete, and at approximately 3,900 pounds per cubic yard, the carbon savings on the foundation alone range from 1,000 to 8,000 kg CO2e per house, depending on the mix design and accounting methodology.
Scale that across the 1.4 million housing starts per year in the United States, and the aggregate number is large enough to matter at the national emissions level, even before touching the framing lumber above the foundation.
What Advanced Framing Already Proves
For anyone inclined to dismiss topology optimization as academic, consider that the US Department of Energy has been documenting a much simpler version of structural optimization for residential construction since the 1990s: advanced framing, also called optimum value engineering.
DOE's published data shows that spacing wall studs at 24 inches on center instead of 16, using two-stud corners instead of three-stud assemblies, eliminating unnecessary headers in non-load-bearing walls, and aligning floor, wall, and roof framing for direct load transfer delivers measurable savings that most production builders still ignore.
Materials cost savings: $500 to $1,000 per home for a 1,200 to 2,400 square foot house. Labor savings: 3% to 5%. Annual heating and cooling cost reductions of up to 5%, because fewer studs mean more room for cavity insulation and fewer thermal bridges conducting heat through framing members.
Advanced framing is the simplest possible form of structural optimization, achieved entirely within the prescriptive code framework by choosing different rows in the same IRC tables. No engineer required, no computational optimization, just fewer sticks in the wall. If the lowest-effort, zero-technology version of material optimization saves a thousand dollars and several percent on energy costs, the ceiling for computationally optimized structural design is substantially higher, probably by an order of magnitude for custom homes where an engineer is already on the project.
Why Builders Will Not Do This
Topology optimization produces non-standard structural elements. Every custom truss, non-rectangular beam, or hybrid timber-steel assembly requires a licensed structural engineer's stamp. Engineering a residential structure from scratch costs $3,000 to $10,000 per project, compared to zero dollars for pulling prescriptive tables from the IRC.
That cost gap is the entire story.
For a production builder constructing 200 identical units in a subdivision, the engineering fee amortizes to $15 to $50 per house, which is trivial relative to the material savings. For a custom builder doing 5 to 10 unique homes per year, the engineering cost may exceed the material savings on any individual project unless the home is large enough or expensive enough that material costs dominate the budget. For a spec builder doing one or two houses at a time, the prescriptive path is cheaper, faster, and carries less professional liability because the code tables absorb the engineering risk that would otherwise fall on the builder's insurance.
And even when the math works, there is a workforce problem. Framers know how to build with standard lumber in standard dimensions at standard spacing. An optimized truss with non-standard angles, hybrid steel-timber connections, and computationally derived member sizes requires skilled carpenters who can read engineered drawings, fabricators who can produce custom components to tight tolerances, and inspectors who understand what they are looking at when the framing does not match the prescriptive tables they memorized. None of these people are impossible to find. All of them are more expensive and less available than the crew that shows up knowing how to frame a conventional wall.
Where This Actually Lands
ASCE's Structural Engineering Institute Sustainability Committee published a white paper explicitly recommending "increased use of topology optimization as a way to avoid over-design and reduce construction material consumption." When the American Society of Civil Engineers tells its members that the structures they are designing contain more material than structurally necessary, and that computational tools exist to close the gap, the professional consensus has shifted even if the market has not.
Realistic deployment for residential construction follows the same pattern as every other technology that is cheaper for production builders than custom builders. Truss manufacturers already use computational optimization for roof trusses and floor systems. Engineered lumber products like LVLs, I-joists, and glulam beams already represent material-optimized alternatives to solid-sawn lumber. Precast concrete foundation systems already use less material per linear foot than poured-in-place walls. Each of these technologies penetrated the residential market through manufacturing economies, not through individual builders running optimization software on their laptops.
Carstensen's framework running on a MacBook Pro matters not because individual framers will use it, but because truss fabricators, engineered lumber manufacturers, and precast concrete suppliers will. When a truss plant can generate optimized roof designs that use 20% to 40% less lumber per unit area while meeting identical load tables, and when those trusses arrive on site looking like normal trusses that install with normal equipment and normal labor, the builder does not need to know or care that topology optimization happened. They order trusses, install them, and move on. The carbon savings are invisible and embedded.
For high-performance custom homes where whole-building carbon budgets are increasingly part of the program, direct application of topology optimization by the project engineer creates immediate value. A 33% reduction in foundation concrete, combined with optimized framing through engineered lumber and advanced framing techniques, combined with timber-steel hybrid roof structures that replace all-steel beam configurations in open-plan living areas, can reduce total embodied carbon by 15% to 25% on a whole-building basis. On a 3,000-square-foot home with a typical embodied carbon budget of 50,000 to 80,000 kg CO2e, that is 7,500 to 20,000 kg of emissions that simply do not happen, because the structure was designed to carry the loads it actually faces rather than the loads a worst-case table assumed.
Limitations
MIT's constructability-constrained framework has been tested on truss structures and bridges. Physical prototypes of the truss designs have not been fabricated; those are planned as a next step. The reinforced concrete beam study from the same lab did fabricate and test prototypes, but beam optimization is a simpler problem than whole-building structural optimization, and extrapolating beam-level savings to complete foundation systems introduces uncertainty.
Residential framing material reduction through topology optimization would likely fall in the 20% to 35% range, not the 90% demonstrated on idealized truss geometries. We could not find published data on topology-optimized residential framing in the field, which means the cost-benefit analysis for custom homes is based on projected material savings rather than documented project outcomes.
Advanced framing savings data from DOE dates to the 2000s and may not fully reflect current lumber pricing or current code editions, though the fundamental engineering logic of spacing studs at 24 inches instead of 16 has not changed and the IRC still permits it.
Carbon accounting for building materials varies significantly by methodology, boundary conditions, and data sources. We used multiple published datasets with different geographic scopes and accounting frameworks. This means the carbon savings ranges presented here are approximate bounds rather than precise calculations for any specific project.