The current project “OffshorePlan - Complimentary Use of Mathematical and Discrete Event Models for Solving Complex Planning and Control Problems in Offshore Construction Site Logistics” founded by German Research Foundation (DFG) primarily dealt with the development of the basics for a combination of simulation-based approaches with mathematical optimization as part of decision support for the installation of offshore wind farms.
For this purpose, new methods are developed by combining model predictive control with classic mixed-integer optimization, which can be used on the mathematical side both for online scheduling and, using suitable access mechanisms, for the offline design of transport routes and capacities at the base port. Besides, a simulation model based on Petri Nets is developed, which can be used for the above-mentioned applications. To this end, different approaches to simulation-based optimization are evaluated.
At the same time, a common meta-model is developed, which can generate executable models using the model-driven architecture and enables complementary use.
Offshore logistics with a focus on wind energy defines a complex planning and control problem. Depending on the dynamically changing offshore weather, cost-intensive ships and personnel must be deployed in such a way that wind farms are built on schedule. Due to the novelty of offshore wind energy technology, there are no established planning and control methods for the construction planning of wind energy plants. In principle, event-discrete simulation methods or mathematical approaches in combination with stochastic optimization are used for this purpose. Both methods have advantages and disadvantages in terms of runtime, level of detail and optimality conditions.
In this project, the complementary use is therefore to be examined. Starting from a uniform basic model, discrete event simulation models as well as models of stochastic optimization for different abstraction / aggregation levels are derived and linked. As a result, the respective advantages of the two methods should be combined into a complementary approach for an improved computer-aided planning and control.