Superalloy capillaries, critical for high-efficiency heat dissipation in aerospace high-end equipment, rely on multi-pass hollow sinking to achieve ultra-thin walls (tens of micrometers). However, due to processing errors, noise and other factors, with the increase in drawing passes and reduction in tube size, the effect of fluctuations in process parameters (including die inner diameter, die angle, friction coefficient, and sizing band length) on their forming performance is transmitted and cumulatively amplified, which hinders the precise forming of capillaries. In this study, with the three-pass hollow sinking of GH4169 superalloy capillaries with a diameter of D 3.0 mm × wall thickness of t 0.08 mm as the research object, the influence of parameter fluctuations on the shape and performance of capillaries was deeply investigated through hollow sinking experiments and finite element modeling, and the uncertainty optimization was conducted using a Long Short-Term Memory (LSTM)-based surrogate model combined with a genetic algorithm to improve process stability. The main results include: (1) Capillary tubes with sizes of D 2.7020 mm×t 0.0834 mm, D 2.5020 mm×t 0.0845 mm, and D 2.3060 mm×t 0.0862 mm were prepared through three passes of hollow sinking experiments. Based on this, a finite element model of the multi-pass hollow sinking of capillaries was established, and the reliability of the model was verified from the aspects of tube diameter, wall thickness, and drawing force, with an average relative error of less than 3%. (2) During the three-pass hollow sinking process, the influence of die angle fluctuation on diameter fluctuation exhibits the characteristics of gradual amplification with each pass, and the diameter range increases from 0.00126 mm to 0.0042 mm, representing an amplification of 234%; All four parameters lead to the transmission and amplification of wall thickness fluctuations, increasing the wall thickness range by approximately 80% to 1000%; the fluctuation of the sizing band length causes the transmission and amplification of drawing force fluctuations, and the drawing force range increases from 2.9 N to 9.1 N, showing an amplification of 214%. (3) Uncertainty optimization improved wall thickness accuracy from 0.0085 mm to 0.0057 mm and reduced standard deviation from 1 × 10⁻3 mm to 3.53 × 10⁻⁵ mm, significantly enhancing forming accuracy and process robustness.

Uncertainty optimization of GH4169 superalloy capillaries during the multi-pass hollow sinking process

期刊论文
发表日期：

2026年6月25日

查看全文

Wei Feng,
Heng Yang,
Yuanyang Zheng,
Duo Zhang,
Yuli Liu,
Heng Li

摘要

Superalloy capillaries, critical for high-efficiency heat dissipation in aerospace high-end equipment, rely on multi-pass hollow sinking to achieve ultra-thin walls (tens of micrometers). However, due to processing errors, noise and other factors, with the increase in drawing passes and reduction in tube size, the effect of fluctuations in process parameters (including die inner diameter, die angle, friction coefficient, and sizing band length) on their forming performance is transmitted and cumulatively amplified, which hinders the precise forming of capillaries. In this study, with the three-pass hollow sinking of GH4169 superalloy capillaries with a diameter of D 3.0 mm × wall thickness of t 0.08 mm as the research object, the influence of parameter fluctuations on the shape and performance of capillaries was deeply investigated through hollow sinking experiments and finite element modeling, and the uncertainty optimization was conducted using a Long Short-Term Memory (LSTM)-based surrogate model combined with a genetic algorithm to improve process stability. The main results include: (1) Capillary tubes with sizes of D 2.7020 mm×t 0.0834 mm, D 2.5020 mm×t 0.0845 mm, and D 2.3060 mm×t 0.0862 mm were prepared through three passes of hollow sinking experiments. Based on this, a finite element model of the multi-pass hollow sinking of capillaries was established, and the reliability of the model was verified from the aspects of tube diameter, wall thickness, and drawing force, with an average relative error of less than 3%. (2) During the three-pass hollow sinking process, the influence of die angle fluctuation on diameter fluctuation exhibits the characteristics of gradual amplification with each pass, and the diameter range increases from 0.00126 mm to 0.0042 mm, representing an amplification of 234%; All four parameters lead to the transmission and amplification of wall thickness fluctuations, increasing the wall thickness range by approximately 80% to 1000%; the fluctuation of the sizing band length causes the transmission and amplification of drawing force fluctuations, and the drawing force range increases from 2.9 N to 9.1 N, showing an amplification of 214%. (3) Uncertainty optimization improved wall thickness accuracy from 0.0085 mm to 0.0057 mm and reduced standard deviation from 1 × 10⁻³ mm to 3.53 × 10⁻⁵ mm, significantly enhancing forming accuracy and process robustness.