Overview
Big construction sites in cities are usually noisy and make traffic even worse. The researchers at Ruhr University Bochum (RUB), in Germany, aimed to make these construction sites for building tunnels better. The goal was to analyze and manage the tunneling processes by using process-focused simulation models.
Problem
In mechanized tunneling, there are always two alternating core processes: excavation and ring construction. The processes for building tunnels require a large number of machine components and logistics elements (tunnel boring machine, external logistics, etc.). Disruption of one element can lead to the disruption of the entire system.
Among the constraints, there were limited storage areas above and below ground. Also, transport distances in the tunnel steadily increased as the excavation of the tunnel progressed.
Achievable performance depended on the interaction of all construction-related processes. Therefore, conventional calculation of the tunneling time was only possible to a limited extent.
To reduce downtime, extensive analysis of the advanced support processes with the aid of process simulation was required. The aim was to achieve more robust planning of logistics and maintenance processes that could also consider uncertainties, especially in the input data.
Solution
Logistical challenges can be analyzed by using process-focused simulation. The main reason why process simulation was used, was the reduction of construction time. For this purpose, RUB developers analyzed various logistics processes of mechanized tunneling in different models.
Since optimization should always be carried out with as few variables as possible, different areas of tunneling were optimized in different AnyLogic models. The model developers needed to take into account city constraints such as noise regulation, traffic congestion, and environmental compatibility.
The goal was to decrease the tunnel building area and keep construction time as low as possible. In addition, RUB wanted to understand if process simulation could be employed in earlier planning phases of mechanized tunneling construction projects.
Two key processes, that significantly determine where these areas can be located, were the supply of lining segments and the disposal of the soil.
The model was divided into 4 main agents: 2 tunnels, the soil disposal of the muck system, and the local constructs inside. There was also an agent to track the simulation progress, but this was only for easy evaluation.
To try out the different logistics variations, a simulation model was created. From the analysis of all model variations, one optimized simulation model was developed. In this way, it was possible to optimize the logistic processes of mechanized tunneling considering a number of inner-city boundary conditions.
The agents marked in red were modified slightly for each model variation. AnyLogic simulation software enabled RUB to model the intrinsic behavior of individual agents and the interaction of different agents. Additionally, in mechanized tunneling, it was necessary to integrate fluid flows.
On top of that, 2D and 3D modeling representation in AnyLogic simplified verification and validation of the model for developers.
Results
Based on the analysis of model variations, an optimized model was created which was suitable for inner-city constraints and able to show possible construction time.
In this optimized model, the disposal of muck was implemented with trains instead of trucks. The size of the storage was increased minimally. Two histograms of initial and optimized models illustrated below, show the tunnel construction time achieved in a Monte Carlo simulation with a thousand repetitions per model.
The top two charts in the diagram below illustrate the results of the initial model. The lefthand chart shows the total tunnelling time, which includes excavating, ring building, and inoperability (downtime). The righthand chart shows the reasons for this downtime, including regular maintenance, equipment failures, and so on. The bottom charts give the same information, but for the optimized model.
Downtime is much less in the optimized model than in the initial model. The total construction time with the initial model was 194 days on average, while the total time for the optimized model was 169.
To sum up, the reduction of the construction time was 25 days (12,9 %). This was mainly achieved by reducing the tunneling machine downtime by 40,5 %, which then resulted in the average advance rate of 17,78 m/d (an increase of 15 %). Reduction of city traffic caused by trucks was achieved by changing the method of disposal of the soil material to train transport (30 trucks a day instead of 103 trucks).
With the help of process simulation for mechanized tunneling, specialists could already compare different variations of logistics in the early planning phase. AnyLogic enabled RUB researchers to consider uncertainties in process simulation when designing the logistics processes.
Since the AnyLogic model can be easily modified, action alternatives to varying conditions can be quickly examined. Therefore, it is beneficial to apply process simulation in the execution phase of mechanized tunneling as well.
The case study was presented by Judith Berns, of Ruhr University Bochum, at the AnyLogic Conference 2022.
The slides are available as a PDF.
