Optimized Reuse of Tunnel and Excavation Materials in Civil Engineering and Tunnel Construction using AI-supported Sensor Technologies

As part of large-scale civil engineering projects, such as the construction of the Future Circular Collider (FCC) and the Einstein Telescope research facility, or the expansion of the Trans-European Transport Network (TEN-T) for rail freight transport—for example, the extension of the Brenner Basis Tunnel or the Lyon-Turin connection—significant amounts of excavated materials are generated. These materials consist of both soft rocks (sedimentary) and hard rocks (igneous and metamorphic), making it difficult to achieve an optimal sustainable use and waste reduction [1, 2]. Excavated materials account for 37.5% of the annual waste generated in Europe. As a result, only 46% across Europe and just 8% in Austria are used for low-value reuse instead of disposal [3, 4]. In 2021, excavated materials accounted for around 59% of the total waste generated in Austria [4], amounting to 46 million tons. According to [5], 35 million tons are classified as uncontaminated excavated soil of the highest quality class according to Austrian guidelines. 62% of this uncontaminated excavated soil is landfilled. Out of a total of 1111 landfills in Austria, 932 are specifically used for the disposal of excavation as of 2020 [5]. Landfilling is not a solution in line with the European circular economy, which is why modern strategies focus on processing and reuse of these materials. In particular, AI-driven technologies offer promising approaches to maximize the efficiency and sustainability of resource recovery, especially for those excavation materials where no utilization possibilities exist yet [1, 2]. Building on the results of the HORIZON 2020 project DRAGON, which was completed in 2015, the currently running NNATT project designs an AI-supported holistic system for an excavation classification and sustainable use of excavated materials from deep and tunnel engineering. Spectral-sensor-based technologies are employed to enable a concurrent material analysis and characterization along with the tunnel boring process, aligned with the established schedule. Our scientific approach considers the heterogeneous nature of excavated materials, optimized material sorting, and economic aspects to yield a high-quality material utilization as a standard procedure in Austria and beyond in the EU.

The chemical-mineralogical and geotechnical classification of the excavated material is significantly affecting its recyclability. Therefore, the geological rock type as well as the excavation and tunneling methods in civil and tunnel engineering play a key role in the reuse of excavated materials. Additionally, anthropogenic contamination lowers the quality class of the excavated material [4]. Excavated and blasted materials can generally be divided into soft (sedimentary) and hard rocks (igneous and metamorphic), which, depending on their petrographic composition and grain size, may exhibit varying degrees of heterogeneity (Table 1; [11]).

The goal of a real-time analysis is to divide the excavated material into mass flows as homogeneous as possible, based on mineralogical-chemical composition and grain size. The material flows are then processed using a well-thought-out logistics concept to minimize the need for intermediate storage on the construction site (Fig. 1). Key parameters for classification include grain size and shape, mineralogical-chemical composition, and, in certain cases, geotechnical properties that determine the material’s further reuse path. Similar material stream characterizations are already used in the classification and sorting of municipal waste [14, 15] and in the processing of anthropogenic and geological resources [16], but have not yet been established in tunnel and excavation constructions due to the heterogeneity of the rock materials and environmental parameters. The analysis methods used for the excavation material are either directly installed above the main conveyor belt or on a separate bypass with reduced conveyor speed.

For the concurrent material analysis and characterization along with the tunnel boring process, established methods for determining the element distribution on material surfaces are already in use. These include LIBS (Laser-Induced Breakdown Spectroscopy) and PFTNA (Pulsed Fast Thermal Neutron Activation) as alternatives to X‑ray-based methods, which cannot be used on construction sites due to labor law regulations [17]. Photo-optical methods are used to analyze the grain shape and size. An additional technology is Hyperspectral Imaging (HSI) combined with visual (VIS) and near-infrared (NIR) hyperspectral cameras. The determination of the mineralogical composition can be effectively carried out using Raman spectroscopy. Hereby, the NNATT project concentrates its effort on LIBS as well as VIS/NIR and Raman HSI. An AI-supported decision matrix links the data from the online analysis with the thresholds of current guidelines and regulations, assigning the material to appropriate recovery based on its composition or potential contamination. This matrix is structured like a risk matrix and is based on the multiplication of two ordinal numbers: The first number assesses the degree of reuse, which is derived from the maximum recoverable amount and the effort required. The second number reflects the impact of use or landfilling on the environment, raw material market, and potential buyers [18].

The choice of processing methods should generally be based on the extraction method as it influences the fine fraction as well as potential anthropogenic contaminants. This also includes the possible use of conditioning agents, with wet processing in a closed water cycle being preferred. After the excavation material is fed into a hopper, it is transported via a conveyor belt on which the analysis unit for material characterization is installed. By this online analysis, contaminated material can be immediately separated and directed to thermal recovery. Initial binders are then removed in a washing drum, with the sludge-containing wastewater being directed through perforated impact plates. The coarse fraction is separated and sent for crushing. In a second washing process, the material is conveyed through a turbowasher, gradually washing out fine components. The resulting slurry promotes the dissolution and floating of light materials. Using vibratory sieves and decanters, the cleaned coarse fractions are separated from the slurry. This slurry is collected and passed through several hydrocyclone stages, an attrition cell, and a cyclone separator to separate the contained fine fractions. The remaining slurry is treated in a thickener with flocculants to precipitate the solids. The thickened sludge is discharged via a rake, while the clear liquid is pumped back into the process. Potential contaminants accumulate in the recirculated water. Therefore, a supplementary water treatment should be considered. The dewatered sludge is stored in a silo before being further dewatered via chamber filter presses. The goal is to produce seven fractions ranging from 10 μm to 10 mm, which can be applied as concrete sand. Clay and silt particles are suitable for producing low-carbon cement or raw concrete. Figure 2 shows a flowchart of potential processing, which could be used to process excavation material [19].

In addition to the traditional use of excavation materials for earthworks and technical fill materials such as rollings, frost protection, or drainage layers, there is the possibility for material utilization on-site and off-site of the excavation site as recycled construction material. After one or more processing stages, the material can be used as an aggregate for concrete and shotcrete, in road and path construction, in railway works, as backfill material, and water construction stones, if the required geotechnical parameters are met. [11]. Another potential use, especially for fine-grained, cohesive materials, lies in the substitution of secondary raw materials such as low-carbon cement, thermally activated clays, raw concrete, raw bricks, or geopolymers [20, 21]. The use of these alternatives allows for a reduction in the demand for conventional raw materials, the production of which is associated with high energy consumption and CO₂ emissions (Fig. 3). However, it must be ensured that the substitution material exhibits a comparable quality [9], which is largely determined by the mineralogical and chemical composition as well as the geotechnical properties of the excavation material. An alternative use for excavated lithologies with low contamination is soil engineering, which is used to produce fertile topsoil. By adjusting recipes with excavation materials, the end product can meet specific requirements and be used in agriculture, for greening measures, or for the restoration of various ecosystems. Additionally, each kilogram of newly formed soil stores 14 g of stable carbon, equivalent to approximately 54.38 g of CO₂ [22].

The high-quality utilization of excavation material is an essential component of the European circular economy. It offers significant ecological, economic, and social benefits. Our approach aims to use (raw) mineral resources more efficiently, reduce the burden on landfills, shorten transportation routes, and promote sustainability in the construction industry and agriculture. This is made possible by combining chemical-mineralogical and geotechnical analyses, the development of a database, the use of AI for material classification, spectral data collection in a pilot plant, the application of sustainable material processes to produce alternative building materials as well as soil additives for agriculture [24] and the planning of material sorting methods.

Studies show that, despite the additional effort required for material characterization and processing, the overall costs for material management on construction sites can be reduced by up to 85% [25]. The associated savings enable the optimized investment planning for clients and a more efficient implementation of large construction projects.