For the current goal of limiting and in the long run eliminating CO2 emissions, the expansion of renewable energy systems is indispensable. However, even with sufficient green energy at hand, further challenges arise regarding its storage, transport and distribution.
Metals such as iron are promising candidates to serve as carbon neutral energy carriers due to the high availability and advantageous transport and storage characteristics. Iron could be used as an alternative to hydrogen, in a storage process in which green hydrogen only represents an intermediate product and is further used as reducing agent in a thermochemical reduction reaction. This way, energy is stored in form of iron resulting from the reduction of iron oxides with hydrogen.

This innovative concept requires the expansion and construction of infrastructure concerning renewable energy systems for green hydrogen production on the one hand, and reduction plants for iron oxide reduction on the other hand.
Finding the cost-optimal design of these energy systems and related process components represents an interesting task that can be tackled by optimization techniques.
To this end, a mathematical model describing the underlying physical processes is developed. In order to account for the location-specific potential of renewable energy sources and the inherent fluctuations of weather parameters, available weather time-series data is used within a robust optimization approach to obtain robust optimal system designs.
Numerical results will be presented to showcase the investigated approach using realistic data for various locations.

The substantial growth of distributed generation is driving an increasing need for active distribution networks to adapt to dynamic operating scenarios. This expansion poses important challenges to TSO-DSO coordination for the power grid's efficient operation, given the distribution network's energy management capacity at the connection point. This research presents a strategy based on robust optimization to determine the operational flexibility of active distribution networks. The methodology involves determining dynamic operation charts for both active and reactive power, considering the system's operational constraints and the characterization of the equipment associated with variable energy resources. The different modes of operation of these devices, based on power inverters, are also included in the analysis. The impact of control strategies based on droop, defined by the IEEE 1547 standard and equivalent policies documented in the literature, is also studied. The uncertainty sets necessary for the proposed robust approach are constructed directly from historical samples based on polyhedral approximation. Extensive computational experiments are performed to validate the proposal, using modified versions of IEEE test cases and considering out-of-sample evaluation to highlight the benefits of the proposed method compared to those documented in the literature.

In this work, we present a two-stage robust approach to tackle uncertainty of power loads in the alternating current optimal power flow (AC OPF) problem. Most of the work related to robust versions of the AC OPF problem makes use of convex relaxations or linear approximations to reduce the complexity of the non-convex problem. Moreover, many studies assume the existence of a feasible adaption of the network's state after the uncertainty realization. While the first approach might lead to operational points that are actually infeasible for the AC physics model, the second assumption is also known to be violated in many cases. Here, we model the AC OPF problem under uncertainty as a two-stage robust problem to (i) identify approximate robust AC-feasible operation points under (discrete or continuous) power load uncertainty and (ii) find a linear decision rule to model the adaption of the network's state once the uncertainty manifest itself. We solve the problem via a decomposition approach. Where, we discretize the uncertainty set and use spatial branching within an adversarial approach where we solve the subproblems to global optimality. Preliminary computational results for small test cases from the IEEE library show that the approach can detect robust infeasibility with a small number of iterations. However, solution times increase for computing approximate robust feasible points. From our results, with affine decision rules and box-type uncertainty sets, we notice that protection against the extreme points of the uncertainty set leads to a feasible solution that is robust for the whole uncertainty set.