During tunnel excavation, support measures may be required to seal or stabilize the ground. Since these sometimes significantly reduce the advance rate, their efficient design is of great importance. Injections are often chosen for this purpose because they can be flexible with comparatively simple equipment technology. Along with this flexibility, it is possible to continuously adjust the grouting parameters during execution. This can mean regular modifications of the drilling grid and length as well as the grouting parameters, grouting materials and stop criteria.
Digital data recording systems have increasingly found their way into drilling and grouting technology. Currently these systems are used almost exclusively for documentation in the framework of quality management. The potential to consider this data in the further construction process and to optimize it has remained practically unused until now. The reason for this is not least the complexity of grouting work and the interactions with the subsoil, so that grouting measures are carried out using the observation method
The project "AVANT - Adaptive Planning of grouting measures in tunnel construction using Artificial Intelligence" has set itself the goal of using grouting data to set up a dynamic real-time process in which artificial intelligence methods are used at several levels.
Since grouting data is available in digital form for all major construction projects, the foundations have already been laid for combining these data with AI approaches. As this paper showed it is possible to train artificial neural networks (ANN) in such a way that they can reliably predict the evolution of pressure-volume records and the volume of grout injected at the end. The case study shows that observed deviations are less than 20 % after only a quarter of the grouting has been completed. In terms of construction site efficiency, this is a very satisfactory amount.