Buildings Are Not Monoliths: Rethinking Renovation Models Could Slash Material Demand Estimates

Every façade replaced, every roof renewed and every interior wall reconfigured sends ripples through the global flows of concrete, timber, gypsum and steel. Yet the models that researchers and policymakers rely on to forecast these renovation-driven material demands may be quietly, and substantially, overstating them. A new study published in the Journal of Industrial Ecology introduces a layered dynamic material flow analysis framework that treats buildings not as single, homogeneous units with one lifetime, but as assemblies of components with distinct and interrelated lifespans. The result is a correction of roughly ten percent in estimated renovated floor area, with even larger discrepancies for older buildings approaching the end of their service lives.

The research team, led by Qiyu Liu of Chalmers University of Technology together with Maud Lanau, Johan Rootzén, Zhi Cao and Filip Johnsson, set out to address a long-standing blind spot in the field of dynamic material flow analysis, or dMFA. This modeling approach tracks the accumulation, use and retirement of materials over time, enabling consistent estimates of construction demand, demolition waste and in-use stocks across building cohorts. It has become a cornerstone technique for quantifying the material metabolism of the built environment, a sector that accounts for more than one-third of global final energy demand and a comparable share of energy-related carbon dioxide emissions.

The problem, the authors argue, lies in a simplifying assumption that pervades most existing studies. A review cited in the paper found that around eighty percent of published building stock analyses did not distinguish between materials held in structural and non-structural components. Buildings are typically modeled as monolithic entities governed by a single lifetime distribution, with renovation represented either as a fixed annual rate or as statistically independent renovation cycles decoupled from demolition probabilities. But real buildings are not monoliths. Structural frames can stand for a century or more, while façades, roofs and interior partitions are replaced repeatedly over the same period without ever triggering demolition.

To capture this heterogeneity, the researchers turned to an influential idea from architectural theory: the shearing layer concept, popularized by Stewart Brand in his 1995 book How Buildings Learn. Brand described buildings as several layers of longevity, from the site and structure to the skin, services and movable contents. The new framework focuses on three layers most relevant to material stock modeling: the structure, which determines when a building is demolished; the skin, meaning the building envelope; and the space layer, comprising non-structural interior elements such as partition walls, ceilings and flooring. Each layer ages on its own schedule, and each renovation targets a specific layer, so capturing their different longevities is essential for realistic modeling.

The methodological core of the study is a cohort-based, stock-driven dMFA framework implemented in Python using the Open Dynamic Material Systems Model. The first step calculates the gross floor area dynamics of the building stock from construction and demolition, applying a Weibull survival function calibrated to Swedish demolition statistics, which show that only about 0.05 percent of multi-family buildings are demolished each year. The second step runs two parallel stock-driven models for the skin and space layers, each with its own lifetime distribution, drawn from Sweden’s National Renovation Strategy: a mean of 35 years for the skin and 25 years for interior elements.

The key innovation is a lifetime convolution approach that ensures renovation probabilities are conditional on building survival. In plain terms, a demolished building cannot be renovated, and a building unlikely to survive long enough to recoup the value of a new façade is also unlikely to receive one. The framework shifts each layer’s lifetime profile by the mean renovation cycle so that a building’s expected remaining life after renovation is at least as long as the renovation interval itself. Material inflows are then calculated by multiplying renovated floor area by layer-specific and cohort-specific material intensities, reflecting the fact that buildings of different construction eras embed different material recipes per square meter.

The framework was applied to Sweden’s residential stock of 2.8 million buildings, using property registry data from the Swedish Land Survey, machine-learning-derived estimates for missing building attributes, and a refined set of material intensities disaggregated across building layers and ten construction materials. After recalculating these intensities with detailed density data and expert consultation, the team arrived at average total material intensities of 557 kilograms per square meter for multi-family buildings and 1,054 kilograms per square meter for single-family homes, figures notably lower than those of a widely used earlier Swedish database.

The results paint a striking picture of the materials locked in Swedish homes. The residential building stock held 354 million tonnes of materials in 2024, equivalent to 33 tonnes per person, with multi-family buildings accounting for 59 percent of the total. Structures dominate, comprising 81 percent of the stock, while skin layers contribute 22 percent of single-family stocks but only 4 percent of multi-family stocks. Between 2025 and 2050, single-family buildings are projected to require between 1.4 and 1.6 million tonnes of renovation materials annually, roughly triple the roughly 590 kilotonnes needed for multi-family buildings each year.

The comparison against conventional monolithic models confirmed the team’s hypothesis. Because the traditional approach lets renovation cycles run independently of demolition risk, it keeps renovating buildings that are in reality near the wrecking ball. Over the modeled period, the monolithic model overestimated renovated floor area by 12.6 percent for the skin layer and 8.8 percent for the space layer of single-family buildings, and by 7.4 and 9.5 percent respectively for multi-family buildings. Cumulative discrepancies by 2050 reached 44 million square meters for single-family and 30 million for multi-family dwellings, equivalent to 14 and 15 percent of their total floor areas. Sensitivity analyses showed the result holds under faster and slower renovation cycles, and that structural lifetime assumptions have only a minimal effect on the layered model’s output.

The implications reach well beyond Sweden. Accurate renovation forecasts underpin climate mitigation strategies, energy-efficiency retrofit programs and circular economy policies, all of which depend on knowing when and where materials will be demanded and discarded. Because different building layers carry very different circularity potentials, from as little as one to five percent for structural materials to ten to twenty-five percent for interior elements, a layered model enables far more targeted policy design. It also allows energy retrofit schemes to be mapped precisely to the layers where they occur, such as wall and window upgrades in the skin. The authors acknowledge that renovation decisions in the real world respond to economics as much as to physical decay, and that detailed renovation histories are still lacking, but they present the layer renovation model as a robust alternative wherever material intensity data permits. As mature economies shift from expanding their building stocks to transforming them, the difference between a monolith and a system of layers may determine whether material demand estimates, and the climate strategies built upon them, stand on solid ground.

Subject of Research: Layered dynamic material flow analysis of residential building renovation material demand in Sweden

Article Title: A layered dynamic material flow framework for modeling building renovations

Article References: Liu, Q., Lanau, M., Rootzén, J., Cao, Z., & Johnsson, F. (2026). A layered dynamic material flow framework for modeling building renovations. Journal of Industrial Ecology. https://doi.org/10.1007/s44498-026-00164-3

Image Credits: AI Generated

DOI: 10.1007/s44498-026-00164-3

Keywords: dynamic material flow analysis, building renovation, material stocks, shearing layers, Sweden, residential buildings, circular economy, material intensity, building lifetimes, construction materials, sustainability modeling, layered