This thesis investigates the potential of a large pumping station in IJmuiden, the Nether-lands, for participating in Demand Response. Due to climate change, renewable energy is onthe rise. The intermittency of energy, together with its unpredictable supply, are a big hurdlefor the energy transition. Two methods are promising solutions to this problem; large scaleenergy storage and demand response. Since large scale energy storage is not yet economi-cally feasible, demand response has an important role to play in the early days of the energytransition.Using energy when it is generated requires a data-stream from the generation facilities onproduction, which is not (yet) widely available. The market price, however, is an indicationof the scarcity of energy, since it is based on the ratio between supply and demand. Besidesthat, there is a correlation between a low energy price and sustainable energy productionsince marginal costs of sustainable energy production are lower than fossil energy produc-tion. This makes using sustainable energy cheaper that fossil energy, and gives DemandResponse a business case.In this thesis, a Model Predictive Control is created that uses energy market data to minimizeenergy costs. Multiple energy markets are analyzed with respect for their suitability for thepumping station in IJmuiden to act on them. The day ahead market is called the APX inthe Netherlands, and this is where energy is bought and sold the day before consumption.The intraday market, also called the flexibility market, is where energy can be bought andsold up to 5 minutes before consumption. A strategy combining these two markets will beevaluated. This is done by using a predicted day ahead price, generated by a SARIMA model,to create a plan. This plan will then be followed, but deviations from the plan are allowedagainst intraday market price.Due to imperfections of the market (mismatch between supply and demand), imbalances areoccurring. These imbalances result in frequency deviations of the grid, and voltage devia-tions. Tenner, the Dutch TSO (Transmission system operator), is responsible for minimizingthese imbalances. In order to minimize the imbalance, TenneT gives a real-time indication ofthe imbalance on the grid, and positive contributions are rewarded while negative contribu-tions are punished. This is done through the use of the imbalance price; a price per volumeof imbalance caused or solved. The imbalance price is based on the aFRR market, wherebids can be done on possible activation. Since the imbalance market is a fast-acting market,it is not suitable for a large pumping station like IJmuiden. However, the aFRR market willbe analyzed in this thesis.The effects of expected future development, like sea level rise and energy market changes,will be analyzed and simulated as well. A higher sea level would result in more pumping, andless discharging under gravity. Which causes the the pump schedule to become less flexible.The results show that it is possible to apply demand response to a pumping station, and theintraday market makes it possible for the MPC to adjust its energy use during the day.The aFRR market analysis shows a lot of potential for the pumping station, possibly makingup for all energy costs made through the spot markets.The conclusion of this thesis is that Rijkswaterstaat can possibly save energy costs on pump-ing, based on the fixed energy price, provided by Rijkswaterstaat, they pay now. Based ona reference scenario where the MPC only minimizes energy use, and a fixed ENDEX energyprice, the proposed MPC makes about 10% less costs in the German market scenario. TheDutch market scenario does not show cost savings. In the Netherlands there is not muchcorrelation between low energy prices and renewable energy yet, since renewable energy isnot a big part of the energy mix in the Netherlands. This correlation is expected to becomemore present when the Dutch energy mix becomes more sustainable. This is expected toresult in lower CO2emission through the energy use of the pumping station. However, moreresearch is needed to confirm this.

Nowcasting high-intensity precipitation is crucial for emergency services and municipalities when making weather-dependent decisions. This research implements and trains a deep generative model for nowcasting using a cleaned precipitation radar composite dataset spanning 15 years, with a 5-minute temporal and 1 km spatial resolution. We propose and apply a method for speckle-like clutter removal to enhance data quality, particularly for high-intensity precipitation rates. The deep generative model is trained with two data sampling strategies to balance the data and improve the accuracy of high-intensity precipitation forecasts. Model performance is evaluated using several standard metrics, and we propose an adaptation to one metric to quantify a score to peak anticipation time in precipitation nowcasting. We also compare our model's performance with a state-of-the-art deterministic Lagrangian extrapolation-based nowcasting system. Our results show that the proposed data quality improvement method effectively removes certain errors from historical radar data. Although the deep generative model currently scores low on the standard metrics, the model trained with a focus on high-intensity precipitation shows an improved score to peak anticipation time. Both deep generative models exhibit less blurring compared to the state-of-the-art model and, in some cases, perform similarly or outperform it.

Nowcasting high-intensity precipitation is crucial for emergency services and municipalities when making weather-dependent decisions. This research implements and trains a deep generative model for nowcasting using a cleaned precipitation radar composite dataset spanning 15 years, with a 5-minute temporal and 1 km spatial resolution. We propose and apply a method for speckle-like clutter removal to enhance data quality, particularly for high-intensity precipitation rates. The deep generative model is trained with two data sampling strategies to balance the data and improve the accuracy of high-intensity precipitation forecasts. Model performance is evaluated using several standard metrics, and we propose an adaptation to one metric to quantify a score to peak anticipation time in precipitation nowcasting. We also compare our model's performance with a state-of-the-art deterministic Lagrangian extrapolation-based nowcasting system. Our results show that the proposed data quality improvement method effectively removes certain errors from historical radar data. Although the deep generative model currently scores low on the standard metrics, the model trained with a focus on high-intensity precipitation shows an improved score to peak anticipation time. Both deep generative models exhibit less blurring compared to the state-of-the-art model and, in some cases, perform similarly or outperform it.