Multi-Objective Optimization of Steel Pipe Drawing Forming Process Design

Literature Overview

This paper by Hu Longfei and colleagues from Hefei University of Technology (2007, Transactions of the Chinese Society for Agricultural Machinery, Vol. 38, No. 10) addresses a critical practical challenge in steel pipe manufacturing: the excessive drawing force and high residual stress encountered during the cold drawing (pulling) forming process. The authors propose a comprehensive optimization framework based on the "FEM-ANN-MOGA" methodology, which integrates finite element method (FEM), artificial neural networks (ANN), and multi-objective genetic algorithm (MOGA) to systematically reduce both drawing force and residual stress simultaneously.

Core Methodology and Technical Framework

The optimization approach follows a well-structured three-stage workflow that is highly relevant to modern process engineering:

1. Orthogonal experimental design combined with FEM simulation — A set of representative forming parameters is selected using orthogonal arrays, and each combination is evaluated through finite element analysis to capture the nonlinear relationship between process inputs and quality outputs.
2. BP neural network mapping — The FEM results are used as training data to build a back-propagation neural network that serves as a surrogate model for the complex nonlinear relationships between drawing parameters (die angle, reduction ratio, lubrication conditions, strain rate) and the dual objectives (drawing stress and residual stress).
3. Multi-objective genetic algorithm with vector evaluation — A Pareto-optimal solution set is obtained using a multi-objective genetic algorithm enhanced with niche technology to maintain solution diversity, followed by a satisfaction function to select the most engineering-appropriate compromise solution.

Key Technical Parameters and Process Insights

The residual stress distribution in cold-drawn steel pipes is of particular concern for downstream applications. High tensile residual stress on the outer surface can initiate fatigue cracks, promote stress corrosion cracking (SCC), and reduce the effective pressure-bearing capacity. From a quality control perspective, the typical target for surface residual stress in drawn pipes for pressure service should be controlled below 50 MPa tensile, ideally achieving compressive residual stress through subsequent stress relief or controlled drawing parameters.

Engineering Practice Implications

The FEM-ANN-MOGA approach described in this paper represents a significant advancement over traditional trial-and-error or single-objective optimization methods. In my experience with steel pipe production lines, the drawing process is often optimized for minimum force alone, neglecting the residual stress implications. This dual-objective framework provides a more holistic approach that directly addresses the root cause of many post-drawing quality issues, including:

• Cracking during stress relief annealing — caused by excessive tensile residual stress exceeding the material's ductility at elevated temperatures
• Dimensional instability — residual stress-driven distortion during subsequent heat treatment or welding operations
• Reduced fatigue life — particularly critical for drawn pipe used in high-cycle applications such as hydraulic cylinders and automotive suspension components

The satisfaction function concept introduced in the paper is particularly practical for production environments where process engineers must make trade-off decisions under constraints such as maximum allowable drawing force (limited by press capacity) and minimum acceptable residual stress levels (dictated by end-use specifications).

Connection to Standards and Quality Requirements

For seamless steel pipes subject to drawing operations, relevant standards include GB/T 8162 (hot-rolled seamless steel pipes) and GB/T 8163 (fluid transport seamless steel pipes), which specify mechanical properties but do not directly address residual stress. However, ASME B31.3 and API 5L both implicitly require controlled residual stress through their requirements for stress relief heat treatment. The optimization methodology presented here provides a pathway to achieving compliant products without relying solely on post-drawing stress relief, which would be more energy-efficient and could reduce scale formation on the pipe surface.

Key Questions and Reflections

The paper raises several questions worth further investigation: First, how sensitive are the optimization results to the accuracy of the FEM material model, particularly the flow stress curve at elevated temperatures and strain rates? Second, the BP neural network serves as a surrogate model, but what is the generalization capability when the process parameters deviate significantly from the training domain? Third, the study focuses on axisymmetric drawing conditions, but practical drawing often involves non-uniform deformation due to die wear and material inhomogeneity. These limitations suggest that while the methodology is sound, its implementation in production requires careful validation through physical experiments and continuous monitoring.

Study Insights and Conclusion

This paper exemplifies the power of integrating computational tools in a structured optimization framework for metal forming processes. The FEM-ANN-MOGA approach is not limited to pipe drawing but can be extended to other cold forming operations such as rolling, bending, and flanging. For steel pipe manufacturers seeking to reduce production costs while improving product quality, this methodology offers a systematic alternative to empirical parameter tuning. The key insight is that multi-objective optimization, when properly implemented, can reveal Pareto-optimal process windows that single-objective approaches would entirely miss, leading to products with superior mechanical performance and longer service life.