Future of Construction


A central platform to exchange best practices and ideas guiding the infrastructure and urban development industry in its transformation, and helping it to address its key challenges.

Why Megaprojects Need A Tech Revolution: Lessons from Carillion

The collapse of Britain’s second-largest construction firm, Carillion, is symptomatic of a larger problem. Globally, construction performs poorly compared to other sectors. The underlying cause of the sector’s underperformance: construction lives in the 19th Century. Across any measure of process or technological advancement—be it software tools, big data, modularization, digitalisation, or the use of lean concepts—construction ranks among the worst performing sectors and remains deeply fragmented. To meet the demands of the 21st Century, construction requires a technology revolution.

When compared to others sectors—auto-manufacturing, computing, mobile telephony, or global shipping—construction’s performance lags alarmingly. Shipping a ton of cargo is 99% faster and over 90% cheaper today than it was in the 1950s. Similar order-of-magnitude leaps have been observed for both automobile manufacturing and computing. Yet construction remains slow, costly, and local: it is the only sector of the economy where labour productivity has in fact declined in the last thirty years. These problems compound as the scale of a project grows. Research at the Saïd Business School, University of Oxford has shown that the problems of cost and time overruns, benefit shortfalls, environmental damage, and negative social impacts plague the delivery of construction megaprojects in a systematic fashion. Such evidence is an indictment of the megaproject construction industry.

Despite its prominent position in the British construction industry, Carillion operated with antiquated technological systems. The low level of digitisation caused costly design information discrepancies and management control bottlenecks. Hidden surprises emerged because there was little or no measurement of real time problems deep in the frontline. Consider evidence from the Midland Metropolitan Hospital in Smethwick, one of Carillion’s largest live projects, whose losses contributed to the company’s demise. On 24th May 2017, the opening of the hospital was delayed. The proximate cause was reported to be issues with the mechanical and electrical design. The root cause, however, was a poorly integrated Building Information Model (BIM). There were large discrepancies between what was designed and what was being built; the construction job was compounding a manual patchwork of incremental fixes which added cost and time. Low levels of training in using tech tools like BIM exacerbated the poor fidelity between what should have been built and what was being built across the project.

Business wisdom suggests that what doesn’t get measured doesn’t get managed. The construction industry needs globally compatible tools that can accurately measure progress in a user-friendly management system to show all stakeholders, on one platform, the difference between the planned and the actual—as a percentage, unit cost, or unit time. Big data and advanced analytics—now integral to business success—are hard to find in the construction industry. Construction companies store fewer petabytes of data than their counterparts in industries as diverse as consumer services, communication, or discrete manufacturing. The lack of big data is a particular problem in improving cost and time outcomes.

Carillion, in line with its industry, used little empirical data. For example, Carillion managers used standard rules-of-thumb to determine the time, cost, or number of workers across projects. The actual outturn reality diverged markedly from these rules-of-thumb. However, since there were no systematic data collected on actual outcomes, the estimates were not corrected perpetuating errors across projects.

The poor use of data also pervades day-to-day delivery of construction projects. Carillion’s management control systems, particularly on the front line, suffered owing to poor or little application of ubiquitously available technologies, such as sensors. McKinsey estimates that there are over nine billion connected devices in the world—the Internet of Things (IoT)—powering better supply chains, asset utilization, or monitoring security. For the Midland Metropolitan Hospital, Carillion hardly used any IoT devices to track day-to-day progress. Live issues encountered on site were managed poorly via a manual database that was separate to the Primavera P6 programme the project managers were using to manage the time, cost, and scope. The complexity, incompleteness, low level of detail and incoherence of interfaces among the technology systems diminished the effectiveness of project managers. Carillion is a powerful reminder that project failures precipitate corporate failures, and with Interserve and Mitie wobbling, we should brace for more.

Unlike the highly automated manufacturing of automobiles or computer microprocessors, construction remains a handcrafted, artisanal industry. The manual and bespoke nature of construction exposes it to variance in process and outcomes. Learning from one task or project to another is hampered because of over-tinkering and improvisation. Off-site manufacturing of building components to obtain the benefits of standardization is gaining acceptance but its use remains undisciplined. For the Midland Metropolitan Hospital, the unitized curtain walling for the project was assembled offsite in Portugal. But it was hand built. It combined the disadvantages of offsite assembly, namely transport costs, with the variance and rework costs of manual building.

The inefficiencies of manual building and low labour productivity further illustrate that under the current model, there is simply too much labour being applied to construction. In government debates, construction projects are often seen positively in terms of job-creation and yet labour shortages in construction are now the norm. The notion that construction requires low-skill workers has to be banished. The Famer Review of the UK Construction Labour Model arrives at the same conclusions. Demanding high-tech jobs and new investment in training, the transition of construction from manual to automated will alleviate the labour crisis of the industry from which Carillion also suffered.

Although the construction sector’s problems are towering, they are not insurmountable. Global shipping was a similarly manual and bespoke industry in the 1950s when cargo was shipped loose—freight arrived in receptacles of all shapes and sizes: paperboard cartons, sacks, bags, bales, barrels, boxes, open crates, or casks. The advent of the container in the 1960s changes all that. The S.S. Warrior, a typical break-bulk cargo ship in the 1950s, sailed from Brooklyn to Bremerhaven in March 1954 and carried roughly 5,000 tonnes of goods, loaded and unloaded over a ten-day-long manual, slow, and costly process. Emma Maersk, typical of ultra-large container ships today, carries over 50,000 tonnes of goods and takes a mere two hours to load and unload in a highly automated process.

Containerised shipping became faster, better, and cheaper by achieving three transformations: bespoke became standardized and modularized; analog became digital; and manual became automatic. As a consequence of these three transformations, a fragmented shipping industry became a global platform. The tech revolution of the construction sector similarly rests on these three pillars: modularization; digitization; and automation.

Modularization is the process of decomposing interdependent, bespoke, and interlinked components into independent, standardized, and interoperable components. Before modularization, a system is fragile. Remove a keystone from a bridge and the bridge falls down. After modularization, a system is resilient: individual components can be added, subtracted, and swapped without disrupting the whole. The process of going from bespoke to modularized requires discipline and adherence to strict design rules. The computing hardware industry went through a two-decade process of modularization from the late 1950s-70s. The pain was worth it. Computing devices—computers, phones, or IoT devices—are the only man-made devices to have scaled to more than 100% of the human population. People in some developing countries without regular access to clean drinking water often have easier access to a telephone. If the benefits of construction are to flow to 7.6 billion people, the industry must modularize.

Digitization is the process of converting continuous waves of information into discrete packets. In the early history of computing this entailed converting natural language (analog waves) into binary code (digital packets of 0s and 1s). Digitization today has come to encompass a large number of tools enabled by computing. The most salient element of construction digitization is to convert 2D analog design information (e.g., architects drawings) into 3D digital models. A fourth dimension (time) and a fifth dimension (cost) can be layered on a 3D digital depiction of the construction object. The rise of 4D and 5D Building Information Modelling (BIM) is a step in the right direction but requires disciplined adoption and compatibiity across software tools, players, and markets. Automobile or aircraft manufacturing heavily rely on such tools to minimise variance between the designed and the produced object. Similarly, by using high-fidelity BIM tools, starchitect Frank Gehry is able to deliver complex and beautiful wonders like the Guggenheim Bilbao on time and on budget.

Automation turns repetitive human tasks into tasks performed by a machine. Between 1990 and 2000, the automobile industry more than doubled the use of robotics on the assembly line: the robotic index—robots per vehicle per hour—grew from 2.2 to 5.4. Commensurate labour productivity gains resulted. Labour hours per vehicle nearly halved from 35.5 to 20.01.  Automation is not only limited to robots in the physical world—it can be applied equally to processes. Closing books is a substantial bottleneck in many large companies, with the process to close month-end books often taking a week or more. The process is mostly manual and bespoke, with audit trails stretching across emails, spreadsheets, meeting minutes and so on. In partnership with Oracle, Amazon has leveraged the cloud automation to dramatically speed up this process and can close their books in one hour. This is an orders of magnitude improvement over industry standard. Amazon’s system is nearly 90% automated—a rules-based system that works by exception and only flags unusual cases for human attention. This simple technology—this does not even use Artificial Intelligence (AI) yet—has been critical to Amazon’s ability to scale a business with a sevenfold revenue increasefrom 2008-2016. For the construction sector to enjoy similar benefits, its internal processes must become similarly effortless.

Experiments in modular, digital, and automated construction are taking hold: Google’s modular server farms or its new headquarters in California being built using robot-crane hybrids (crabots), Bechtel’s innovations in offshore oil platforms, Arcadis’ 100% BIM pledge, Laing O’Rourke’s investment in design for manufacture and assembly (DFMA), Skanska’s wireless monitoring of buildings with sensors, Con-X-Tech’s steel manufacturing, the City of Rotterdam’s recycled plastic road construction, Natel Energy’s modular minihydro units, Arup’s big data initiative, Wowjoint Machinery Company’s Segmental Bridge Launching Machine, Outotec’s modular solutions for small mines, or WinSin Construction’s 3D-printed apartment building. The construction sector must now turn to establishing conditions that enable these radical experiments into a technological revolution at scale.