We know that data driven design is an almost certain future, but how will this be realised? In this latest LSL thought piece we look at the potential for data driven design solutions, their impact in refining design, driving down construction costs and reducing embodied carbon within structural engineering. We question whether BIM is being fully utilised and what we must do to gain the benefit of many decades of existing building information. We outline the issues with the way data is managed today and identify potential steps toward true generative design.
What is Big Data?
Every two days, our modern culture generates an equivalent amount of data to that generated from the earliest recorded information up to the turn of the recent millennium. Our ability to store this information has doubled roughly every three years for the last forty years, with Moore’s and Kryder’s laws indicating data storage will continue to become cheaper and thus more accessible. We live in an age where almost every aspect of our lives, for better or for worse, generates data that is stored by some means.
Big data then, is the analysis of this data, and the extraction of useful information. These data sets can be unfathomably large and are typically beyond the ability of traditional approaches and software to manage and process within a usable time scale.
The information that is extracted can be used in many ways, but the principal benefits are better decision making, improved efficiency, driving innovation, and improved customer experience. An applied example is the Tesla model S car which collects data about the way it is driven and its environment. Data from each car is aggregated by the manufacturer, processed and then used to refine the autonomous systems within each and every car, creating a feedback and learning loop. In the oil and gas industry big data is creating improvements not just in the efficiency of drilling methods, but in the exploration and location of resources.
Within the construction industry there is already some movement toward big data in terms of transportation and construction logistics. In an industry with low margins and where between 10 and 30% of project costs account for material waste and remedial works, moves toward technology based efficiency are inevitable.
Information Challenges
Before we can start to consider how we might apply big data usage to design within the construction industry, it is important to understand a little about the properties of data, and how it is processed and managed.
In simple terms, data can be structured (good quality BIM models, databases, spreadsheets), semi-structured (poor quality BIM models, O&M manuals, calculation packages, Bills of Quantities), or unstructured (social media, emails, photos, paper documentation). Historically the industry has made some usage of structured data but tended to ignore all other sources as it was too difficult to work with, and too overwhelming to draw conclusions from.
Utilising artificial intelligence (AI) allows this data to be processed and for patterns to emerge. This is typically done via an application of AI called machine learning where software teaches itself to do identify what pieces of data represent, and then to extract useful information from vast sources of data far more reliably than a human ever could, and in a fraction of the time.
Challenges to the use of vast amounts of data are typically broken down into three main areas: the data itself, the processing of the data, and the management of the data. The data itself has become categorised by the V’s; originally volume, variety, veracity and velocity, with more recent versions of the list adding three further terms.
The processing of data ranges from the storage of it through to the analysis, extraction and interpretation of patterns. This section is where the AI and machine learning algorithms are utilised, and thus where technical skills are required outside that typically utilised in our industry.
The management of the data contains many challenges, these are not just technical in nature, but create legal and moral questions upon which individuals’ views may differ significantly. Management also has to contend with the skill shortage related to the data processing; governance and lack of skills are perceived as the two major challenges of big data application.
Data in the Construction Industry
There are huge amounts of information stored within our industry. These include decades of project information such as briefing documents, design reports, drawings, specifications and maintenance documents. There are three significant obstacles to this information being utilised in an effective way. Firstly, it is typically unstructured and physical (as opposed to digital) data; secondly it is siloed within the archives of designers and contractors; and thirdly it rarely contains much if any feedback from the end users.
In the last decade BIM has created a huge step forward in terms of data usage and storage. BIM data is typically structured and inherently digital, and where it is utilised post design and construction in facilities management, can begin to collect operational data and feedback. Unfortunately, despite leaps forward in terms of team collaboration and common data environments, the information typically remains siloed within organisations, destined to be archived after liabilities have passed. Subsequent alterations or refurbishments are often considered in isolation, and rarely update existing as built models.
Installation of sensors to provide real time feedback is fairly common in large infrastructure projects such as bridges and dams, these can be installed on existing as well as new build structures, for example bridges which are nearing the end of their design life but still appear to be in sound condition. At this point in time sensors are far less common in building structures, but the move toward smart buildings should have a positive impact here, which will in turn start to provide operational feedback data to supplement the design and construction data.
A further benefit of BIM is the power to visualise the data stored within a federated model, and the convergence of this with big data is allowing augmented reality technologies to be utilised whereby a user can walk around an area of a town or city and see development plans overlaid on the real world environment.
Stepping away from BIM for a moment, there are several very impressive applications which are big data related and could bring a number of benefits to the planning and construction process.
· The London Infrastructure Mapping Application https://maps.london.gov.uk/ima/ is an online tool that allows anyone to access a broad range of data from numerous sources including statutory authorities. Longer term this application aims to link directly with many of the data sources, allowing real time updates to the information provided.
· Underground Asset Register https://www.gov.uk/government/news/map-of-underground-pipes-and-cables-designed-to-save-lives-and-prevent-major-disruption schemes are being piloted in Sunderland and a number of London Boroughs.
· The VU City platform https://vu.city/about/ provides a digital twin of several major UK cities. Following its adoption by over 80% of London boroughs it is becoming a principal tool in the planning process.
· The Buildings As Material Banks project https://www.bamb2020.eu/about-bamb/ is developing tools that will allow material data to be stored facilitating a circular economy with reduced waste and less reliance on raw materials.
Applying big data to design
A system whereby data is collected during design, construction and operation, and then stored and processed would allow valuable information to be extracted to enhance the brief of the subsequent projects. This is fundamentally the ‘lessons learned’ process that is the cornerstone of quality management systems. Imagine then if the data was shared and available, turning all those siloed feedback loops into a big data feedback loop.
This would provide benefits at every stage of the process – briefing documents could more accurately define end users needs and design criteria could be optimised to meet those needs. Optioneering could find the definitive most economical, or the lowest carbon solution rather than a good fit based on the previous experience of team members. Public realm design could benefit from usage, weather, operational and transport data, allowing masterplans to be developed with fully integrated public spaces and interactions.
The logistics of the construction stage could optimise in terms of time, cost and carbon. Every aspect could be considered, from the extraction of raw materials through to processing and installation could be considered. Design could be optimised right up to the point of construction to allow for supplies of locally available materials and perhaps even accounting for ever changing market forces.
Likewise, the operation stage could also become optimised, and anything that functioned in a sub-optimal way would feed back into the system to further optimise the brief, optioneering and perhaps even aspects of workmanship in future projects. The term Generative Design has been coined to describe this system of perfect feedback.
Admittedly this does all sound a little utopian. The design world is still a long way from being able to access these data sources and the issues of sharing and management of data are unlikely to allow a perfect system. However, some degree of generative design is required if we wish to gain the full benefit from smart buildings and smart cities, and truly seek the optimum between comfort maximisation and energy minimisation.
Unlocking Potential
So how do we begin to unlock this potential? The RIBA has identified four ways for designers and planners to utilise data in ways that will improve both workflow and end product.
· Designing for citizens – using data to better match the needs of users
· Experimentation – rapid modelling and simulation during optioneering
· City analytics – using big data feedback to improve policy implementation and planning
· Transparency – the more data you have available, the more reliably you can gain new insights
The first three of these are all aspects of generative design as outlined earlier, the fourth, transparency, is one of the key barriers. Assuming all the issues relating to the processing and management of data could somehow be resolved, the availability of data will always limit the power of generative design unless there are incentives to share. The government could play a role here, both in making data available and somehow rewarding industry for contributing to the mass of data. The incentive has to be greater than the perceived commercial value of hoarding such data. Alternatively, it is something that could be enforced via the planning process; however, most will take the view that the stick is rarely as effective as the carrot.
Another interesting point here is the action on climate emergency which most design practices and industry organisations have signed up to. These typically include a commitment to share data to allow material and energy efficiency to be improved. If this commitment is realised, then it could make a valuable contribution to the pool of industry data.
Another key barrier as noted previously is the technical skill required to utilise big data. Can the construction industry which is already suffering a skills shortage attract data scientists to organise and manage systems? Or is it more likely that tech companies will harness the available project data along with feedback from smart buildings and provide us with a black box software solution. If at any point during this article you were wondering what the role of existing designers might be in the big data future, the possibility of tech companies managing the generative design loop is something the industry needs to think about. Generative design loops will provide a paradigm shift in efficiency to all aspects of our industry, but loops need to remain open enough to allow new ideas and industry innovators to feed into the system.
Our view at London Structures Lab is that there will be a move from tech companies to invest in this field, and much of the number crunching and repetitive tasks that engineers currently undertake will be consumed by automated systems. However, we also believe that the future of structural engineering practice remains within the understanding and assessment of the complexities of development within the urban fabric. As our role becomes more nuanced in terms of production, Engineers will become the trim tab of design, making small but significant movements that provide the new ideas and innovation that the generative design loops require, and focusing on unique and bespoke problems outside of the data pool.