Over the next ten years Norway will build many new roads and railway lines, many of them in areas with challenging ground conditions. Such projects are extensive, complex and expensive, requiring great resource efforts and skilled engineers. Traditionally, site investigations of the sub-surface and depth to bedrock has been mapped by boring and sampling conducted by crawler drilling rigs, followed by geotechnical laboratory testing to determine soil properties.

"These methods are extensively used and proven, but both time consuming and costly. Therefore, we seek innovative solutions and new technology that can make site investigations more effective by saving time and reducing cost", says Lars Andresen, CEO of the NGI (Norwegian Geotechnical Institute).

Faster and cheaper

In recent years NGI has employed two methods that together make up a unique tool, which has the potential to map sub-surface properties and depth to bedrock quicker and less expensively than traditional methods. The first method, AEM (Airborne Electromagnetic Measurements), involves measuring the electrical conductivity in the ground from an antenna suspended below a helicopter. The second method is to use artificial intelligence to interpret and transform the AEM-signals to show depth to bedrock, distinguish between changes in soil layers (ie. sand and clay), and also to indicate possible layers of quick clay.

AEM is a relatively new method in Norway. The method has been applied to projects conducted by the Norwegian Public Roads Administration and Bane NOR (the Norwegian state-owned company responsible for the Norwegian national railway infrastructure) to map the properties of the sub-surface for road and rail projects. Examples are the new E16 between Nybakk and Slomarka, a number of InterCity routes and the Ringeriksbanen from Sandvika to Hønefoss north-west of Oslo.

"The advantage is that several alternative options for pathways can be mapped over long stretches by helicopter in a matter of a few days, says Lars Andresen.

"The challenge is the interpretation of AEM-signals and their conversion from electrical conductivity to geotechnical information, ie. depth to bedrock. The method is not as reliable as traditional soil borings and soundings, so some soil borings are needed to verify and calibrate the AEM measurements. However, AEM investigations drastically reduces the number of such traditional borings", Andresen concludes.

Expert group for artificial intelligence

In 2015 NGI established an internal expert group for research on artificial intelligence for geotechnical applications. The group includes members from all disciplines and departments at NGI. Software developed by this group is based on machine learning, and interprets data generated from the AEM measurements.

The method uses so-called Artificial Neural Networks (ANN), which is inspired by the way the human brain is built up. Similar technology is presently used in computers to identify breast cancer from mammography images, and to identify people by using face detection technology.

"Helicopter measurements produce large amounts of data that must be interpreted by an expert. Previously this job was done manually and was very extensive and time consuming. Now we use artificial intelligence, so that computers can learn to interpret the signals", explains Asgeir O. K. Lysdahl, project engineer in the Department of Geomapping at NGI. He is also a member of the artificial intelligence research group at NGI.

"Your computer learns how the geotechnical engineer performs the interpretation. Based on this, the computer completes the interpretations in the remaining areas with AEM-measurements. The method is still under development, but we have already used it to survey the depth to bedrock on several projects, such as the Ringeriksbanen.

AEM results Inter city 2015


Great need for innovation

The Norwegian building and construction industry can greatly benefit from research, development and innovation. New and improved ways are needed to build cheaper, safer, and greener. In many areas of engineering, the use of artificial intelligence has great potential to improve methods, and also help with decision support in the actual design work.

"With Norway's skilled environments within the geosciences, engineering and ICT, I believe we have great opportunities to take on a leading international role", says Lars Andresen. "This will have positive socio-economic spin-offs in the form of efficient project execution, risk reduction and cost reduction. This will, however, require that we have developers and contractors who embrace innovative solutions and that multidisciplinary applied research and development is given priority.


[1] Meld. St. 33 (2016–2017), Nasjonal transportplan 2018–2029 (Norwegian only)

[2] http://onlinelibrary.wiley.com/doi/10.1002/cncr.30245/abstract

[3] http://www.banenor.no/Prosjekter/prosjekter/ringeriksbanenoge16/ (Norwegian only)