Advancing AI Assisted Conceptual Design of Subsea Inline Structures: Mechanical Behaviour of Pipeline-Mounted Structural Assemblies during S-Lay Installation

Abstract

The first paper in this series introduced a dual classification framework for subsea pipeline inline hardware — a physical taxonomy describing how hardware joins the pipeline string (IW/EA) and a mechanical taxonomy describing how it amplifies strain during stinger passage (Types A, B and C) — and characterized the behaviour of inline welded components, which are cut into the string and act over a short length. This second paper turns to a different and far less studied class of hardware: complete structural assemblies mounted onto the pipeline. Top Structures (EA-ST) and Base Structures (EA-SB) are not welded into the string; they are built around the pipeline, span twenty pipe diameters or more, and transfer load into it at a small number of discrete attachment points. Their behaviour during stinger overbend passage has not previously been characterized in the open literature, and it cannot be inferred from inline component results, because the governing parameters are different. Industry practice for top structures, base structures, mudmats and branch piping is first surveyed and reduced to a taxonomy suitable for automated design, including a unified classification of the connection systems through which a structure is attached to the pipeline: fixed at one point, fixed at two points, pin-slot, and variants of each incorporating a deadband gap (F1, F2, PS and their D variants). Parametric Abaqus finite element analyses incorporating stinger–pipeline contact are then used to characterize the resulting strain behaviour, extending the mechanical taxonomy of the first paper from effects distributed along a component to effects concentrated at discrete attachment points. Added mass is treated as an independent design variable, decoupled from stiffness, and branch piping support layout is examined as the canonical worked example of concentrated stiffness transfer. Key findings include: the attachment layout, not the size or weight of the structure, governs the strain imposed on the pipeline; a top structure attached at a single point, or through a pin-slot pair, leaves the pipeline close to its plain-pipe strain, while a rigid two-point attachment concentrates high strain at those points and creates a low-strain zone between them suitable for housing strain-sensitive equipment; the same pin-slot arrangement is markedly less effective on a base structure, showing that connection behaviour does not transfer directly between structure types and must be assessed against the structure geometry; deadband connectors redistribute strain along the pipeline rather than reducing it, so gap control governs the outcome; branch piping strain and header pipeline strain are governed by two largely independent levers that can be optimized separately; and added mass follows the same position-governs principle as stiffness and elevation. Together with the first paper, these results complete the design intuition layer of the proposed AI-assisted conceptual engineering programme for the full inline hardware set.

Introduction

This paper is the second study in a seven-step AI-assisted conceptual engineering framework for structural design, using the S-Lay Inline Tee (ILT) as the main case study. It builds on the first paper, which established the classification of inline hardware and studied distributed mechanical behaviours. This paper focuses on Top Structures, Base Structures, and branch-piping support arrangements, whose effects on pipeline strain are concentrated at discrete attachment points.

Main Problem

In S-Lay installation, an Inline Tee is a complete structural assembly rather than simply a thick pipe component. It may include:

• Top Structure supporting branch piping and connectors.
• Base Structure supporting the pipeline on vessel rollers and later transferring loads to the seabed.
• Mudmat and other structural elements.
• Branch piping connected to the Tee.

Unlike inline-welded components, these structures are not continuously integrated into the pipeline. They are connected at a limited number of points. Therefore, their effect on pipeline strain depends mainly on attachment location, number of attachment points, connection stiffness, and structural configuration, rather than simply on component thickness or length.

Main Contributions

The paper makes four major contributions:

1. Structural survey:
   It identifies the major structural modules used in subsea ILT assemblies and examines how they are connected to each other and to the pipeline.
2. New taxonomy:
   It develops a classification system for structural components and their pipeline connection systems, including F1, F2, PS, and Deadband variants. The taxonomy is designed so assemblies can eventually be represented and generated automatically for AI-assisted engineering.
3. Finite Element Analysis:
   Parametric finite element simulations are used to study the mechanical behaviour of Top and Base Structures. Their behaviour is classified into the concentrated-parameter types A2, B2, and C2.
4. Mass and branch-piping effects:
   The study treats added mass independently from stiffness and investigates branch-piping support arrangements as a representative example of Type A2 behaviour, where stiffness and mass are transferred to the pipeline through discrete supports.

Mechanical Classification

The framework uses two independent classification systems.

Physical classification:

• IW – Inline Welded: Components welded into the pipeline string.
• EA – Externally Attached: Components attached externally without becoming part of the pressure boundary.

The structures in this paper mainly belong to EA-S, which includes Top and Base Structures.

Mechanical classification:

• Type A: Stiffness-dominant behaviour.
• Type B: Elevation-dominant behaviour.
• Type C: Combined stiffness and elevation effects.

Each type is further divided into distributed and concentrated forms:

• A1, B1, C1: Distributed effects, studied in the first paper.
• A2, B2, C2: Concentrated effects, studied in this paper.

For example, A2 represents stiffness concentrated at discrete connection points, such as a Top Structure or rigidly supported branch piping.

Scope Boundary

The paper deliberately excludes IW-A anchoring components from its structural models. These are short inline-welded components whose local strain effects were already characterized as Type A1 in the first paper. In a complete design assessment, their known contribution would be added separately at each attachment point.

Therefore, the strains presented in this study are mainly intended for comparing different structural arrangements, rather than being directly treated as final code-compliance values.

Overall Significance

Conclusion

This paper is the second in a series building an AI-assisted conceptual design framework for subsea inline structures. Together with [1], it completes Steps 1 and 2 of the seven-step programme — Component Classification and Design Intuition — for the full set of IW/EA components, by covering the concentrated-parameter behaviour types (A2, B2, C2) left open in [1], plus added mass and branch piping. Three results stand out. First, connector type is a powerful and simple design lever: F1 and PS connections leave the pipeline close to its plain-pipe strain, while F2 connections concentrate strain at the connection points in exchange for a calmer, lower-strain zone inside the structure. Second, a deadband connector redistributes strain rather than removing it, and the outcome depends on tight control of the gap — this should be treated as a tuning parameter, not a free strain reduction. Third, branch piping strain and header pipeline strain are governed by two largely independent connection choices — the branch-to-support connector and the header structure\'s own connector layout — which means the two can, to a first approximation, be optimised separately during conceptual design. These findings extend the quantitative design intuition available for Step 3 of the framework (Intuitive Assembly), in which individual component behaviours — both the distributed types from [1] and the concentrated types from this paper — will be combined into complete, worked ILT structural assemblies.

References

[1] Sivaraman,S.M., Reddy J.V.,. “Toward AI-Assisted Conceptual Design of Subsea Inline Structures: Classification and Strain Behaviour of Inline Components During S-Lay Installation,” IJRASET, 2026

Copyright

Copyright © 2026 Sreekanth Manakkattil Sivaraman, Jagannatha Venakataramana Reddy. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.