Inefficient structures

In a 2014 study of steel-framed buildings in the UK, researchers found that steel was overused by about 40%. In a 2023 study of a hospital in Hong Kong, researchers identified significant excess material use in the original design. In surveys of typical villa buildings in Kuwait and Oman, researchers determined that engineers used up to twice as much reinforcement elements as necessary. Together, these findings suggest that inefficient structural design is not an isolated anomaly but a recurring problem across building types, structural elements, and regions—spanning decades.

Despite advancements in the field of structural engineering, overuse of concrete and steel remains deeply entrenched in construction practices worldwide. Embodied carbon, or the emissions associated with the entire life cycle of a building, from resource extraction to construction, accounts for roughly 14% of global carbon emissions—more than double the combined emissions from all domestic and international aviation and shipping.

A growing body of research suggests that a significant share of this material does not meaningfully contribute to structural performance or safety. Instead, it reflects the slow-to-innovate nature of the construction industry, where conservative design decisions, risk-averse engineering practices, and limited adoption of new technologies produce buildings that are inefficient and more carbon-intensive than necessary.

At the same time, optimization research demonstrates that many of these inefficiencies can be avoided through computer and AI-assisted design. In a 2022 study of the Cayan Tower in Dubai, researchers were able to reduce concrete volume by more than 35%, corresponding to an estimated reduction of roughly 4,400 tons of carbon dioxide. That amount is roughly equivalent to the annual emissions of 1,000 passenger cars. Other case studies of tall buildings similarly report cost and material reductions in the range of 10-30% through various optimization methods. Importantly, these improvements were achieved without compromising structural performance, safety standards, or code compliance.

The aforementioned examples come from a comprehensive literature review that the tBE team undertook to assess the prominence of structural inefficiency in our industry. Over the past year, we collated and reviewed more than 400 peer-reviewed articles from researchers across Europe, East and West Asia, Africa, Australia, and the Americas. This research allows us to situate individual examples of inefficient design within a global pattern and put researchers from disparate places in conversation. By examining optimization studies alongside analyses of real buildings, our review highlights how excess material use—and its associated carbon emissions—has persisted globally despite technical knowledge demonstrating that leaner, more efficient, and more sustainable structures are possible.

Summary of literature review

We began with a broad review of academic literature on structural inefficiency and optimization across disciplines including aerospace, computer science, and mechanical engineering before narrowing our focus to the built environment. From a corpus of roughly 17,000 articles, books, and conference proceedings published since 1970, we identified more than X papers on the optimization of buildings and construction-related projects—including bridges, high-rise towers, infrastructure, and structural components. Of these, 4,666 directly applied optimization techniques to buildings.

This research is global in scope, although North America, Asia, and Europe account for the majority of published case studies. A smaller but notable body of work comes from Africa, South America, Oceania, and West Asia. Interest in the topic has accelerated alongside advancements in computational tools, with more than two-thirds of the literature published between 2015 and 2023. Over 2,000 papers were published in 2022 alone.

For a deeper insight on our review, read our methodology article.

Why the inefficiency?

Much of the structural inefficiency in buildings is determined long before engineers begin calculating beam sizes or reinforcement. Basic choices like how far apart columns are, how regular the building’s layout is, or what kind of floor system is used largely determine how much steel and concrete a building will need. Research consistently shows that shrewder choices at this stage can significantly reduce both material use and emissions.

A 2021 Cambridge study found that “good early stage design decisions can halve [a building’s] embodied CO₂” while simultaneously lowering costs. Similarly, an analysis of hundreds of real buildings in the United States showed that simply redesigning floor systems more efficiently could theoretically reduce a structure’s weight by more than 50%, without changing how buildings function or appear. Once these early decisions are fixed, however, later design requirements are often accommodated by adding more material rather than rethinking the structure itself.

Despite the proven potential of such reductions, the methods needed to achieve them rarely become part of routine construction practice. Numerous studies show that computer-based optimization tools can redesign real buildings using substantially less steel or concrete while meeting the same safety requirements prescribed by building codes. Across multiple case studies, researchers report material savings of 10–60% by redistributing material more efficiently within a structure. In one notable example, optimizing the design of a tall tower reduced its concrete use by more than one-third, eliminating thousands of tons of embodied carbon. Yet our research shows that these gains are typically demonstrated only after a design is complete, as academic exercises rather than standard engineering practice. This gap suggests that waste persists not because engineers lack solutions, but because efficiency tools are rarely integrated into everyday practices.

One reason for this gap is time. Civil engineering researchers studying large building projects in Chile and Mexico argue that the design process is already so complex and time-consuming that generating highly efficient alternatives is often infeasible within project time constraints. As they note, tight deadlines leave little incentive to invest the additional time and specialist expertise required for optimization. Instead, engineers often default to standardized solutions—such as using uniform beam depths throughout a structure—rather than tailoring individual elements to the loads they actually carry.

Designs are also shaped by a preference for flexibility over efficiency. As John Orr, Professor of Structural Engineering at the University of Cambridge, concluded after surveying 129 engineers worldwide,

“the reality [is] that designers face no significant penalty if structures are over-designed, and that over-design is sometimes viewed as a positive attribute.”

These tendencies are reinforced by the fragmented nature of the construction industry. The lack of integration between disciplines, consultants, and project stages makes holistic optimization difficult. Structural elements within a single project often require input from multiple specialists, sometimes working with different software and assumptions. Aligning these parallel efforts introduces friction that discourages iterative refinement—a problem that has long plagued the industry. As Ove Arup observed in his 1968 lecture to the Institute of Structural Engineers:

“The need for specialization is obvious enough. But whereas invention and repetition must disregard boundaries and demolish barriers in order to be effective, specialization creates barriers.”

Finally, inefficiency persists because engineers rarely learn from how buildings perform once they are in use. Unlike other fields of engineering, structural design does not routinely monitor performance after construction, making it difficult to improve future designs based on empirical evidence. As Orr notes in a later study:

“Structural engineering remains the only engineering discipline that does not consistently measure in-service performance of its designs… The status quo, where structural material wastage in the order of 50% is common, cannot continue if we are to meet carbon emissions reduction targets.”

All in all, the literature shows that excess material use is a rational outcome of prevailing design cultures, risk structures, and economic incentives. Material is repeatedly overused due to liability, delays, and redesign, while optimization—despite its demonstrated benefits—remains unrewarded. As a result, large quantities of steel and concrete are embedded into buildings not because they are structurally necessary, but because over-design has become the global norm.

Structural optimization refers to the process of using computational methods to systematically adjust a building’s geometry, size, and material distribution to achieve specific objectives—most commonly minimizing material use, cost, or embodied carbon—while fully satisfying safety and performance requirements. Unlike conventional design, which often relies on standardized sections and conservative assumptions, optimization tools explore thousands of design alternatives to identify solutions that use material only where it is structurally necessary. A robust body of academic research demonstrates that these methods are not speculative or theoretical: when applied to real buildings, they repeatedly produce substantial reductions in material use without compromising code compliance or structural performance.

One of the clearest demonstrations of this potential is A. J. Unander’s MIT master’s thesis, which analyzed 640 buildings across more than 16 countries using data from the Database of Embodied Quantities Output. By comparing existing buildings to optimized designs reported in the literature, Unander found that optimized structures weighed, on average, 466–542 kg/m² less than conventional designs—equivalent to an average 38% reduction in total building weight. He concluded that:

“on average, recently constructed buildings contain nearly twice as much structural material as required by structural constraints.”

Because Unander’s dataset spans building types, materials, and regions worldwide, his findings suggest that optimization potential is not limited to specific markets or construction traditions, but is inherent to contemporary building practice globally.

Beyond large datasets, numerous case studies illustrate how different optimization tools operate in practice. Genetic algorithms, for example, are frequently used to iteratively refine structural layouts by evaluating thousands of possible configurations. Alkhatib et al. applied a genetic algorithm combined with computational fluid dynamics to optimize a tall twisting tower in Dubai, achieving a 35.7% reduction in concrete volume and eliminating approximately 4,400 tons of embodied carbon while maintaining wind performance. Similarly, Sherif et al. embedded a genetic algorithm directly within a BIM environment, enabling automated optimization of reinforced concrete floor systems and achieving up to 15% material cost savings per floor across multiple real buildings.

Other studies show that optimization is not limited to component sizing, but can reshape entire structural systems. Dillen et al. developed an algorithm that simultaneously optimized the size, shape, and topology of steel structures, reducing material use by up to 18.6% in real multi-story buildings compared to original designs. In bridge design, Zordan et al. employed structural optimization to redesign a five-span arch bridge in Venice, achieving a 40% reduction in structural weight while satisfying aesthetic and budget constraints. These examples demonstrate that optimization tools are flexible and adaptable across building types, scales, and materials.

Optimization exposes inefficiencies that conventional design practices systematically overlook. Studies of real buildings repeatedly show that standard designs operate far below their allowable capacity, embedding large amounts of unused material. Moynihan and Allwood found that steel-framed buildings in the UK used, on average, 46% more beam mass than necessary, while Dunant et al. showed that 36% of total structural mass in real buildings could be safely removed. Optimization techniques make these hidden inefficiencies visible and quantifiable, transforming excess material from an accepted norm into a design variable that can be actively reduced.

It is worth noting that, in practice, many powerful optimization techniques can be hard to implement on real construction projects. Methods like topological optimization often produce highly irregular or organic shapes that are difficult to build with standard materials and construction methods, making them expensive or impractical to realize on site. As a result, engineers may have to simplify or redesign optimized results before they can be built, which reduces some of the material savings. Researchers have also noted that optimization tools frequently struggle to account for real-world constraints such as standardized components, construction tolerances, and existing building codes, all of which slow adoption in practice.

Despite its potential, structural optimization remains under-utilized in the construction industry globally

The application of structural optimization in real-world construction remains limited. Of the 600+ papers reviewed for this study, only 125 applied optimization to real case studies, and just 29 reported actual implementation of the optimized designs. Researchers from the Universities of Cambridge and Leeds describe this gap plainly:

“Optimization techniques are rarely used in practice, largely for prestige projects pushing the boundaries of engineering and not for the bulk of constructions where the potential for greenhouse gas emissions mitigation is the highest.”

Echoing this assessment, Rory Clune, a partner at McKinsey & Company, observed in his MIT PhD dissertation:

“There are strikingly few examples of buildings, bridges, and other infrastructure designs using optimization, even though efficient and cost-effective design of such resource-intensive and expensive projects is of great importance to society.”

These limitations are both structural and institutional. Despite being one of the world’s largest industries, the construction industry has historically been slow to innovate and remains among the least digitized industries globally. A major barrier is workflow fragmentation: projects involve numerous stakeholders—designers, contractors, subcontractors—often operating under short-term contracts that discourage coordination and long-term investment in new processes. Teams frequently change from one project to the next, further inhibiting the adoption of unfamiliar design methods.

Compounding this challenge, each construction project is shaped by unique site conditions, regulatory environments, and client requirements, making it difficult to replicate optimized solutions at scale. This stands in sharp contrast to industries such as automotive or aerospace, where a single optimized design can be applied repeatedly across thousands of products. In construction, design timelines are short and situational demands are high. Optimizing a reinforced concrete system, for example, may require different assumptions depending on regional codes, material availability, or construction practices, adding challenges to implementation.

Additionally, optimization increasingly relies on data-driven and AI-assisted methods—an area that the AEC industry lags behind others in adopting. This is due both to limited availability of large, standardized project datasets and to the under-use of existing data that could otherwise inform optimization algorithms. As a result, the industry continues to rely on conservative, material-intensive design practices, even as decades of research demonstrate that far leaner structures are not only possible, but common within the realm of academic studies.