Entwicklung von simulations- und Kl-basierten Methoden zur Erhöhung der Ladungssicherheit im Containertransport


Project Acronym
EsiKIEL
 
Project Title
RetroLadung – Development of simulation and AI-based Methods for increased Loading Safety of Container Transportation
 
Funding Code
03SX484C
 
 
Principal Investigator
 
Status
Laufend
 
Duration
01-09-2019
-
31-08-2022
 
 
Abstract
Background and Objectives

Every year, several hundred containers are lost at sea, causing environmental pollution and endangers for shipping traffic. Most containers go overbord because of insufficient lashing or ship operation in severe environmental conditions leading to exceedance of the dimensioning loads of the lashing equipment. Therefore, technical solutions for safe and efficient container transport are needed.
In the joint research project RetroLadung, a modular cell guide system with an integrated sensor network is developed. As the weight of the installed cell guide system reduced the vessel’s payload, the additional cell guides must be as light as possible. For the dimensioning and weight optimization of the cell guide structures, the occurring loads must be known exactly. During ship operations, an artificial intelligence-based monitoring and decision support system coupled to the integrated sensor network guarantee that the loads stay within acceptable limits.

Approach

Within the research project RetroLadung-EsiKIEL, simulation and artificial intelligence-based methods for the improvement of the operational safety of container transportation are developed.
In the first stage, an efficient non-linear time domain motion simulation method is used to determine the wave induced motions and accelerations at container stowage positions needed for structural dimension and optimization of the cell guide system. Extensive model tests will be carried out at the towing tank to validate the simulation results. Main objective of RetroLadung-EsiKIEL is the development of an artificial intelligence-based monitoring and assistance system to ensure that the operational limits of the cell guide system are not exceeded during ship operation. Based on acceleration measurements at the cell guides, the operational status is evaluated and recommendations for safe ship operation are given if necessary, e.g. regarding forward speed and course of the vessel.


Application and Examples

In a first step, the ship models and environmental and operating conditions to be investigated are specified. The project focusses on two relevant industrial application cases: Container transportation at the American east coast using towed barges and midsized open top container vessels.

For the simulation-based assessment of wave induced loads, a sequential process is developed. Since the dimensioning of the cell guide structures requires reliable statements about the occurring loads at a very early stage of the project, an efficient linear frequency domain method is applied first.
In the next step, a non-linear time domain simulation method is used for a more detailed motions analysis. Additionally, finite volume methods are used to determine direct loads, e.g. from wave impacts in extreme situations.

As only very little validation data is available for the investigated cases, model tests are carried out in the institute’s wave basin to evaluate the accuracy of the applied numerical methods.

In the second phase of RetroLadung-EsiKIEL the monitoring and assistance system is developed. For the modelling of the complex dynamic behavior of the vessels, artificial neural networks (ANN) are used. Based on the measurement data from the integrated sensor network of the cell guides, e.g. accelerations at the container stowage positions and stack weights, the operational condition is monitored and evaluated. If potential exceedance of the operational limits is detected, recommendations for safe ship operation are provided. Therefore, separate ANN-based methods for short time forecasting and decision support are implemented. For the initial training and design of the ANNs the motion simulation results are used. Because of the high variability of the dynamic behavior of the vessel, machine learning methods for run-time adjustment using measurement data and online co-simulation must be developed and tested.