When bidding to deliver the $US2.6 billion Valley Line – West project (pictured above) to extend the light-rail line in the City of Edmonton (Canada) by 8.7 miles (14 kilometres) with fourteen new street level stations and two elevated stations, US engineering firm Parsons faced several challenges.

These included the complexity of the project, the congested nature of the area and uncertainty surrounding construction during the city’s bitterly cold winters. 

Despite this, the company needed to be confident about delivery timeframes.

Using a platform from California based AI-powered construction simulation software firm ALICE Technologies, it ran thousands of simulations for different options regarding sequencing and resource deployment – such as the number of crews being used.

From this, it determined the effect of each option on the overall cost and time of construction. 

Moreover, Parsons was able to use this information to create schedules which required fewer crews and less equipment such as concrete pumps. 

It also determined the optimal amount of formwork which was needed to accelerate the development of the elevated rail portion of the project.

As a result, Parsons delivered a competitive bid upon which it felt comfortable that it could deliver within time and budget. It won. Construction will start this year. 

This is not the only case where artificial intelligence (AI) has been applied to deliver better outcomes on major construction projects.

When planning its luxury Elio Del Nest high-rise residential project in Bangkok, Thai property developer Ananda Development PLC used ALICE to experiment with various construction strategies. 

This included probing for potential bottlenecks on formwork, testing what might happen if it ran varying numbers of crews – and if crews worked overtime, and testing for a tightly sequenced plan against a more ‘relaxed’ program. 

Perhaps surprisingly, it found that by paying overtime to its crew starting one month after project initiation, it was able to reduce overall costs by 32 percent. This was the case as a 208-day reduction in project duration more than offset the additional cost of paying overtime.  

(Elio Del Nest high-rise residential project in Bangkok. Image source: ALICE Technologies)

In another example, a major developer in Kazakhstan wanted to understand the impact on construction time and cost on a 13-storey apartment building in the city of Astana of using quick-set concrete as opposed to traditional concrete.

To do this, the company entered plans for its building, along with costs and curing times for both quick-set and traditional concrete, into ALICE. It then analysed and explored results.

It determined that quick-set concrete would slash build time from 176 days to 126 days and would reduce overall project costs by 3.5 percent notwithstanding that this type of concrete was pricier.  

Such examples highlight the use of artificial intelligence in construction – a phenomenon ALICE Technologies CEO Rene Morkos describes as ‘generative construction’.

In a recent interview, Morkos spoke with Sourceable about how this can be used to deliver better outcomes.

To understand generative construction, it is useful to contrast this with generative design.

Generative design is a process used to identify potential new ways of designing buildings, products, goods or machinery. When using this, designers or engineers input objectives for whatever they are designing into generative design software. They also specify parameters such as performance or spatial requirements, materials, manufacturing methods, and cost constraints. The software then explores all possible permutations of a solution and generates design alternatives. This process uncovers a far greater number of options compared with those which the designers themselves would have considered without the computational power of the AI software.

By contrast, generative construction enables project managers and construction contractors to leverage computational power to simulate millions of different options for the best way to build what has been designed. 

It enables a clear understanding of the impact which decisions such as the number of cranes being used or the use/non-use of overtime will have on construction schedules and costs.  

Speaking of ALICE, Morkos says the software complements the knowledge of people with the number crunching power of computers. 

This works as follows.

First, users upload a model of what they want to build – often a 3D model created in programs such as Revit. 

They then input the ‘rules’ which govern how construction needs to happen. 

This includes matters such as:

  • The order in which different building elements should be put in place (foundations, for instance, need to be laid before the frame and roof are installed).
  • How parts of the process should be grouped and split in order to be managed; and
  • The specific processes which are required for each task and the resources which are needed for these processes. When installing a pier cap, for example, tasks could include putting in the formwork, adding the steel, pouring the concrete, curing the concrete and removing the formwork. Resources needed could include a crane, a carpentry crew and some formwork. Rules regarding these processes and resources are known as ‘recipes’ (see image below).

Once the rules are established, the system calculates the schedule and shows the time and cost which will be needed to deliver the project. 

It then enables you to analyse the effect of different options.

Should you wish to add or remove cranes, introduce overtime or add production constraints, you alter the relevant parameter, hit a button and the system restimulates the entire construction. 

(A screenshot of the Explore Scenarios page in ALICE. Image: supplied)

Asked why this is needed, Morkos says determining the best way to deliver construction on complex projects can be difficult.  

He says traditional ways to do this involve running through options in the heads of project managers or using tools such as critical path method (CPM). 

With regard to the former method, this is not feasible on major works such as WestConnex, the Sydney Metro or Melbourne Metro Tunnel project (major construction projects in Sydney and Melbourne) as these can involve up to ten, fifteen or twenty thousand tasks across the project.

Whilst CPM has its uses, meanwhile, it does not readily enable simulation of different construction approaches such as adding cranes, changing the radius of cranes or altering the design. 

Where you want to change the design and simulate through ALICE, by contrast, you can simply change the design, upload the new design, copy the rule set and the system will recalculate.

Over the course of an afternoon, the system can analyse and calculate around 600 million different solutions.

From there, users can identify methodologies which will deliver the fastest timeframes and/or the lowest possible cost.  

(A screenshot of the Recipes page in Alice. Image: supplied)


Morkos’ comments come as governments both worldwide and in Australia are ramping up investment on complex infrastructure projects.

In the US, for example, the Senate last month struck a deal to spend $US1.2 trillion on an eight-year infrastructure plan to fund roads, bridges, public transport, power and the internet. 

The deal is supported by President Joe Biden, although Democrats have indicated that it will only be passed if another $US6 trillion spending bill is also passed.

In Australia, A336 billion ($US249 billion) is expected to be spent in infrastructure (public and private – excludes mining) over the five years to June 2025, according to the most recent forecast by the Australian Construction Industry Forum.

In some specific sectors, work is expected to ramp up. 

In rail, the value of work is expected to increase  from $40.1 billion over the five years to June 2020 to $58.9 billion over the five years to June 2025 thanks to work on large-scale mass-transit projects in major cities. 

On energy, the value of work is expected to increase from $65.3 billion to $80.1 billion over that same period as work ramps up on transmission projects such as interconnectors which are needed to improve grid stability in light of an influx of renewable energy generation.

Morkos says ALICE has been deployed on major road, rail and bridge projects in countries such as the US, Canada, the UK, Hong Kong, Malaysia, Singapore and India. 

It has also been deployed on construction of solar and wind facilities along with some commercial buildings. 

Speaking of Australia, Morkos says the software is yet to be deployed on any local projects but adds that the country is expected to be a prime market for ALICE going forward.

Across all projects thus far, the company says it has helped to deliver average reductions 11 percent in cost and 17 percent in build times.

Speaking particularly of ALICE, Morkos says implementation typically takes around thirty days. This includes a five-day period (three to four hours per day) during which participants learn theoretical material such as how to move beyond the limitations of the critical path method along with how to translate complex construction constraints into algorithmic materials. 

After that, a couple of weeks is taken to load the project into the system and translate project constraints. 

As well as during pre-construction planning, Markos says ALICE has recently begun to be brought into existing projects which are in-flight and are facing challenges or delays. 

This has been the case with a number of commercial building developments. 

In such projects, he says the software has helped project teams to plan effective strategies to get projects back on track.