Innovations in Marine Construction: The Rise of Seawater Sea Sand Concrete

With the increasing global demand for marine infrastructure and the concurrent depletion of freshwater and river sand resources, the construction industry is at a crossroads. Two advanced materials, seawater sea sand concrete (SWSSC) and ultra-high-performance concrete (UHPC), have emerged as pivotal solutions for marine construction needs (Ahmed et al., 2020; Du et al., 2021). These materials not only meet sustainability goals but also exhibit exceptional performance characteristics tailored to marine environments.

The Promise of Seawater Sea Sand Concrete

SWSSC has garnered attention primarily due to its innovative use of seawater and sea sand, making it a sustainable alternative in regions where traditional resources are becoming scarce. This material uses the natural chemical properties of seawater and integrates sea sand, which is often viewed as unsuitable for conventional concrete due to concerns about corrosion and durability. However, research has shown that SWSSC can provide comparable, if not superior, strength and resistance to marine conditions (Saleh et al., 2024).

Ultra-High-Performance Concrete: Setting New Standards

On the other hand, ultra-high-performance concrete (UHPC) represents a technological leap forward in material science. Its composition typically includes superior cementitious materials and fibers, leading to remarkable compressive strengths and durability (Du et al., 2021). UHPC is vital for structures subjected to extreme loads and environmental stresses, offering resilience that traditional concretes simply cannot match. The development of ultra-high-performance seawater sea sand concrete (UHPSSC) represents the synthesis of these two groundbreaking materials, combining the advantages of SWSSC’s sustainability and UHPC’s high strength to withstand harsh marine environments (Huang et al., 2021).

Addressing Corrosion with Fiber-Reinforced Polymer Bars

One of the main challenges in marine construction involves the corrosion of steel reinforcement, which can severely affect the longevity of structures. To combat this issue, fiber-reinforced polymer (FRP) bars have been increasingly adopted. FRP bars are not only resistant to saline environments but also offer excellent mechanical performance and durability (Zhao et al., 2021; Zeng et al., 2022a). This makes them ideal candidates for reinforcing modern concrete formulations like UHPSSC, where the bond strength between FRP bars and the concrete matrix is critical for structural integrity and service life.

The Importance of Bond Strength

The bond strength between FRP bars and advanced concrete types like SWSSC, UHPC, and UHPSSC plays a crucial role in the performance of marine structures. High bond strength ensures that the reinforcement effectively transfers loads, thereby enhancing the overall stability of the structure (Yan et al., 2016). In the context of UHPSSC, research on FRP-UHPSSC bonding is notably limited, with fewer than 50 available pullout test data points (Zhang et al., 2022a). This scarcity restricts conventional modeling approaches and emphasizes the need for innovative methodologies.

Leveraging Machine Learning and Transfer Learning

In light of the data limitations surrounding FRP-UHPSSC bond strength, advanced computational techniques such as machine learning (ML) and transfer learning (TL) offer promising pathways. TL is particularly beneficial as it allows for the utilization of knowledge from related, data-rich domains to enhance predictive accuracy in less-researched areas (Pan and Yang, 2010). By merging FRP-SWSSC and FRP-UHPC bond data, researchers can pre-train models that can then be fine-tuned for UHPSSC applications, overcoming challenges imposed by data scarcity.

The Role of Natural Language Processing

Further enhancing this approach is the integration of natural language processing (NLP) with deep learning (DL) techniques. These combined technologies foster semantic understanding and improve the representation of heterogeneous or missing variables within datasets. This can be particularly advantageous when dealing with diverse experimental conditions and input formats (Devlin et al., 2019; Wang and Sun, 2022).

Proposed Methodology for Predicting Bond Strength

The proposed methodology involves creating an NLP-based TL framework designed to harness the wealth of data available from FRP-SWSSC and FRP-UHPC systems. This pre-training phase lays the groundwork for refining models specifically for FRP-UHPSSC bond strength predictions. By focusing on robust data processing and model training, researchers aim to develop predictive frameworks that are not only effective but also generalizable to emerging composite materials in marine contexts.

Structuring Research for Future Insights

The continuation of this research is meticulously structured to cover several key areas. The initial sections will review existing literature on FRP bond behavior as well as the application of ML, TL, and NLP in engineering contexts. Subsequent sections will detail the methodology of the proposed framework, including data development and evaluation metrics, ensuring that every angle is explored to validate and refine the predictive models.

By paving the path for advanced predictive modeling in marine construction, this study anticipates not only offering significant insights but also establishing a foundation for future research that could reshape practices in the industry.