Development is hard, and delivering complex architectural projectsbrings real risks: the quality of the finished product, the time taken todeliver, and the total cost all impact an owner’s profitability.
A 2016 McKinsey analysis found that construction projects typically take 20 percent longer to finish than scheduledand are up to 80 percent over budget, frequently resulting in litigation.
(RajatAgarwal, Shankar Chandrasekaran, and Mukund Sridhar, “Imagining construction’s digital future,” June 2016, McKinsey.com)
The Darling Square Library isa great example of how an integrated process can eliminate risk and ensure youdeliver your project on time and within budget.
Developed by Lendlease anddesigned by Kengo Kuma and Associates, the building is a circularpost-tensioned concrete structure with a striking façade composed of a doubleskin of timber strips that wrap and encircle the building.
Darling Square had all the right people: an amazing design team, superb engineering, and a stellar subcontracting team. Even with all the right people,the intersection between teams reduces risk only if you can integrate their expertise into a knowledge model. The problem was that there was no comprehensive understanding of the complete design. Rather, a disparatecollection of incomplete, single-domain models and drawings were being markedup as PDFs and circulated for review.
The architecture team maintained a design intent model that specified the look, feel and performance.The engineering team maintained a structural analysis model. The subcontracting team provided feedback on cost, time, and constructability. However, there was no integrated knowledge model of the team’s combined intelligence.
Without a comprehensive, fabrication-level model, projects are exposed to severe delivery risk due to cost escalation, time overruns, and safety on-site. For Darling Square Library, there was too much at stake. Fortunately, the project manager had previously worked for Frank Gehry’s Gehry Technologies and deeply understood the importance of designing and building from a single source of truth. Without this, it would be impossible to deliver the kind of complex buildings for which Frank Gehry and Kengo Kuma are famous.
The development team engaged AR-MA as the project integration and DfMA consultant. Our role was to integrate the knowledge from each stakeholder (development, design, structure, fabrication, and construction) into a single comprehensive model that was designed for manufacture and installation. This case study explores how we created a comprehensive and integrated DfMA model and the value this brought to the development team.
The development team were working to a budget and a delivery schedule, and the façade introduced a high level of risk to the project’s cost, timeframe and overall quality. Getting it correct was crucial. An early challenge the team faced was ordering the material for the façade and managing the schedule risk this brought to the project.
The team proposed Accoya as the material for the timber façade. Accoya is a modified wood product that is dimensionally stable with a 50-year warranty. Our initial material take-off indicated that to produce the design, we would require 27km of Accoya! This was an extraordinary amount of material and exceeded the budget for the façade scope.
Problem 1: the project budget was at risk.
Secondly, the estimated lead time for the procurement, fabrication, and installation of the material was 12 months.
Problem 2: our date for practical completion was at risk.
We only had one chance to order the correct amount of timber due to the long lead time in acquiring and fabricating it. Our goal was to optimise the façade to reduce cost. We had to reduce the amount of timber while also maintaining the design intent of the façade. Finally, we had to submit a detailed bill-of-materials (BOM) for the accurate order of timber.
This was an exercise in value-creation as opposed to value-engineering. Value-engineering strips design value out of a project to bring costs in line with budget – it does this at thecost of the delivered product. The product itself becomes less valuable. In contrast, we created value in the project by integrating the team’s intelligence and collaboratively innovating to enhance the design while preserving the budget.
Our first challenge was to reduce the amount of timber in the façade, but equally important, we needed to order the correct amount of timber. There was no budget to waste by over-ordering, and if we ordered too little, we would have an unfinished façade for up to 12 months. With no room for error, we couldn’t estimate the quantity with a low-resolution material-take-off. Instead, we needed to produce an accurate BOM for the timber. A BOM is usually delivered at the end of the project as part of the fabrication set. It is the result of all the design development efforts. However, to achieve the precision required, we needed to produce the BOM up-front.
Let’s look closely at the model to understand the complexity.
The timber strips are divided into prefabricated modules that are held together by steel posts at each end. The posts are connected to the building by steel outriggers. An early design task was to lay out the timber façade modules and steel posts. In doing so, we faced several competing objectives that included design, structural, and installation constraints.
To optimise the post spacing and solve this problem, AR-MA’s computational design team wrote a series of algorithms that leveraged a multi-objective evolutionary solver.
Now that the posts were accurately located, we needed to split the timber strips into segments and nest them onto stock material lengths for ordering. The stock material lengths are the available lengths of timber from the supplier. Nesting the fabricated strips onto stock lengths would provide the most accurate estimate of the timber required. In order to get there, though, there were many challenges.
Firstly, we had to account for 50mm docked from each end of the stock material to provide a clean-cut edge. Secondly, we had to distribute two different timber joints across thefaçade based on a design logic: butt joints and overlapping joints. The overlapping joints were 400mm long. Third, the joint distribution rules were further refined to include rules for pre-assembled timbers vs site-fixed timbers. The installers would site-fix timbers after the façade modules, which would span two modules. These site-fixed timbers would disguise the modules and make the façade look like a continuous spiral. Finally, we had to run the nesting multiple time on different stock lengths to find the optimal stock with the highest material utilisation.
To identify any hidden risks and remove them, we needed to get to a fabrication level of detail early. We coded a series of algorithms to automatically create the detailed connections between the steel posts and timber and then iteratively identify any clashes and solve them. Achieving this level of detail early in the process identified any hidden risks and provided the team with certainty moving into fabrication.
Our DfMA process allowed us to rapidly calculate the number of required modules, steel posts and outriggers, and the length of the timber façade and develop the detail for fabrication. This precise understanding of quantities and costs early in the design development process allowed the development team to tune the design intent while adhering to their project budget.
The outcome of this exercise is that we could radically reduce the quantity of material and the cost. Initially, the project required 27km of timber. After optimising, we were able to achieve the design intent and reduce the timber from 27km down to 17km of timber – an almost 30% reduction in material, handling and fabrication costs. Further, we now had a solid strategy to manufacture, pre-assemble, and install the facade. The development team could proceed confidently with ordering material.
When it comes to prototypes,we have three principles: early, often, and everything. We prototype early, we prototype often, and we prototype everything.
On the Darling SquareLibrary, we started with digital prototypes of the façade model. We collaboratively built and reviewed these with design, engineering, and fabrication. These prototypes ensured alignment across the team and that we met all constraints and objectives.
Digitally prototyping the design and construction is essential with modern complex projects; however, building and reviewing physical prototypes is no less essential. Without the detailed resolution a prototype brings, you are leaving problems to be discovered on site. So many problems in design and construction are created and exacerbated by incomplete documentation and latent physical risks that are only understood once fabrication starts. We always advise building physical prototypes to understand the finished product and identify hidden risks.
Our digital prototypes progressed to physical prototypes.
Physical prototypes are essential to not only test the fabrication process and final product but alsoto test the design-to-production pipeline. Our DfMA pipeline includes documentation standards, file output types, naming, and fabrication processes.
Based on the feedback from the physical prototype, we made several changes to our documentation and the design of the modules. The physical prototype allowed the team to rehearse the manufacture and pre-assembly process and ensure smooth execution on site.
In the Darling Square Library, we maintained control over the design all the way through fabrication. The manufacture and installation stage required shop drawings in the form of laser-cut files, dimensioned part drawings for quality control, and assembly drawings. Each assembly contained detailed BOMs, measurements for assembly and checking, and weight and centre of gravity for lifting.
To accomplish this, we build a Stage 4 - Fabrication model – this is the culmination of our DfMA process. Our fabrication model is comprehensive and prescriptive. It contains all the necessary information to procure material, fabricate and manufacture components, pre-assembly into modules and final assembly onto the building on-site.
We control and systematise our fabrication model by coding parametric templates. A parametric template is a component that will adjust its geometry, material, and other parameters when placed into the digital model. Our parametric model for the Darling Square Library allowed us to automatically adjust the size, shape, and placement of thousands of components that would have otherwise been modelled by hand – a costly and time-consuming task prone to error.
For example, the steel outriggers grow or shrink based on the distance between the slab and thefaçade, and the number of connection points changes based on the position ofthe timber strips. We completed this computationally and automatically to ensure accuracy and remove the risk of modelling errors.
The design-to-production pipeline creates a digital bridge from the design intent to fabrication while maintaining all structural and installation constraints. Each component in the model contains its unique data ready forextraction into the BOM.
As we progressed with the design and detail of the timber façade and glass balustrade model, it became apparent that a new problem was emerging – the accuracy of the concrete slab. We were not responsible for the concrete but were about to inherit its problems.
We had effectively designed afaçade and balustrade system with millimetre accuracy. However, this precision enclosure needed to connect to a relatively sloppy cast-in-situ concrete. The post-tensioned concrete slabs could deviate by a tolerance of +- 50mm. Compounding the extreme deviation, the slab profiles were all curved. The setout of the slab edge would be complex and challenging to achieve the designed shape, and the post-tensioned concrete threatened to throw it out even further. This late-stage information threatened the entire façade. The risk was that it might just not fit.
Our general solution to riskis to own it. We solved the interface challenge by taking over the design and management of the concrete edge formwork system. We needed a setout and edge-form strategy to precisely construct the concrete to marry our balustrade and timber façade. Further, we had to accommodate cast-in anchor points for the façade brackets and ferrules for the balustrade. Finally, we also had to setout and form the cast-in post-tension embed locations.
To accomplish this, we designed a proprietary concrete formwork module for slab edges that was easy toinstall and increased accuracy.
Our formwork picks up all the required elements from our driver model: the channels, embeds, and ferrules, and automatically models and documents the modular formwork system. The system is CNC fabricated and designed to be simple to produce, and easy to fabricate, with a self-set-out system for fast, simple, and safe setout on site.
Our formwork system saved the project over $60k per level or $300k in total by reducing site labour and preliminary costs by reducing days on site.
Development is hard, and developing complex projects is fraught with risk. Risk is always associated with the unknown, the ambiguous, and the unprepared. A comprehensive model designed for manufacture, where the assembly and installation are simulated and rehearsed before construction, creates transparency, surfaces risk, and increases confidence for the entire project team. However, as shown on the Darling Square Library, an integrated DfMA process can eliminate product,timing, and cost risk. Rehearsing the manufacture and installation through digital and physical models creates a smooth design and construction experience for the development team and maximises project profitability. Through our integrated and comprehensive model designed for manufacture and assembly, AR-MA ensured the project came in on time and on budget.