This research develops a surrogate modeling framework to efficiently analyze and optimize a proposed pump and gate system designed to mitigate compound flooding in the Clear Lake region, Texas. Traditional numerical hydraulic models are often computationally expensive for large number of simulations. As probabilistic assessments can require $O(10^3)$ to O(10^5) runs, a cheaper alternative would be necessary for robust probabilistic assessment. This study addresses this limitation by developing a deep-learning surrogate to approximate the complex hydrodynamic behavior.

The methodology involved three main stages. First, a 1D HEC-RAS model of the Clear Lake system was adapted to serve as the physics-based "ground truth" generator. Second, this model was used to generate a training dataset of 2,400 simulations. This was achieved by systematically sampling key infrastructure design parameters (gate width $W_g$, number of pumps $n_p$, and activation levels $h_{on}$) alongside a wide range of synthetic compound flood forcings (inflow hydrographs and downstream storm surge boundaries).

Third, three distinct Long Short-Term Memory (LSTM) network architectures (Models A, B, and C) were developed to compare different data encoding strategies. Model A, a direct sequence-to-sequence (seq2seq) model, was provided with all dynamic inputs, including the known pump discharge time series ($Q_{pump}$). Model B tested the model's ability to infer dynamics by replacing the $Q_{pump}$ time series with static design parameters ($n_p$, $h_{on}$). Model C used an autoregressive structure, feeding its own past water level predictions back as inputs to dynamically infer the pump response.

The results demonstrate that the fully-informed LSTM (Model A) can successfully learn and reproduce the governing hydrodynamic processes with very high accuracy. However, models that attempted to infer dynamic behavior from static design parameters (Models B and C) show reduced performance. These models particularly struggled to capture the sharp, transient effects of pump (de)activation, leading to overly smoothed predictions. This study concludes that while LSTMs are capable of learning the physical patterns of the system. The main challenge lies in feature encoding, specifically, enabling the model to capture complex, dynamic responses from static inputs. The framework demonstrates the potential of LSTMs, but emphasizes that how the data is represented is the key factor in developing a surrogate model suitable for design optimization.

Dikes are a primary defence against flooding from the rivers and sea in the Netherlands, but many no longer meet current safety standards. To ensure the dikes keep meeting the safety standards in the future, climate change will need to be taken into account. Increase in river discharge and upstream water levels through rising temperatures, more frequent extreme precipitation, and increasing meltwater run-off, are climate uncertainties that pose additional challenges for dike reinforcement projects. One critical failure mechanism requiring attention is backward erosion piping (BEP), which has historically been underestimated. Over the past decade research has advanced the knowledge on BEP and led to innovative solutions to mitigate this risk, each based on different working principles and associated with distinct performance characteristics and life cycle costs. However, it is currently unclear how the design, performance, and cost-effectiveness of these solutions compare under climate change uncertainties.

This study evaluates the design performance and life cycle of three piping solutions, both innovative and conventional, under climate change uncertainties. Using a numerical model in COMSOL Multiphysics, the performance of each solution is assessed through fragility curves based on the relevant failure criterion for each solution. Due to the different working principles, the fault trees differ between the solutions. The failure criterion for each solution is based on the critical step they tackle in the fault tree that leads to piping failure. Climate impact uncertainties are represented by considering different future climate scenarios. Using Hydra-NL, water levels for different return periods and future scenarios, including both moderate and high-emission pathways, are calculated. The resulting probabilities of the water levels and the conditional failure probabilities from the fragility curves are used to determine and quantify the timespan of the life cycles. The life cycles are evaluated in terms of cost and their sensitivity
to design and climate uncertainties is assessed. This assessment is done with a life cycle cost analysis using the Equivalent Annual Cost (EAC) method.

The results show that while the sheet pile and plastic filter screen exhibit similar performance in preventing hydraulic heave, their sensitivity to rising water levels differs, with the filter screen performing slightly better under extreme conditions. The SoSEAL barrier demonstrates lower sensitivity to high water levels but exhibits higher failure probabilities at lower water levels due to heterogeneity in the barrier’s hydraulic conductivity and contact zone effects. It is emphasised that the plastic filter screen and SoSEAL barrier have inherent uncertainties due to their innovative designs. Reducing these uncertainties in the future can significantly increase the performance and make them competitive alternatives to conventional solutions like the sheet pile. The life cycle cost analysis indicates that the sheet pile is currently the most cost-effective solution and least sensitive to the uncertainties across the different scenarios, followed by the filter screen and SoSEAL barrier. Cost uncertainties, particularly in annual maintenance & monitoring, were found to influence the performance more than climate uncertainties. Constant yearly costs, resulted in lower values of equivalent annual costs for longer lifespans, while annually increasing costs produce higher values for longer lifespans, stressing the importance of accurate cost forecasting. Furthermore, the results highlight the trade-offs between the technical and functional lifespan, suggesting that increasing annual costs can offset the benefits of extended design life. In the future, the uncertainties in the design-related expenses for the innovative solutions will decrease. The amount and range of the yearly costs will then decrease with time, increasing the cost-effectiveness of the SoSEAL barrier and plastic filter and decreasing their sensitivity to future uncertainties.

This study concludes that the performance of the different piping solutions is influenced by both design and climate change uncertainties. Varying lifespans and annual costs significantly affect the performance of the solutions across different scenarios and have a even greater impact than the uncertainties of climate change. Life cycle cost variability underscores the need for considering economic uncertainties alongside the uncertainties in technical performance into current decision-making processes. Decreasing these uncertainties in innovative solutions could improve their performance and make them suitable alternatives in the evaluation and comparison of different solutions.