What is chevron cracking in welds?
人字形裂紋（chevron cracking）是什麼？
How the cracks form, what they look like, which NDT method actually finds them and what stops them in the first place.
裂紋怎麼形成、長什麼樣子、哪一種 NDT 真的抓得到，以及一開始該怎麼避免。
Chevron cracking is hydrogen-induced cold cracking in the weld metal of multi-pass welds, most often submerged arc (SAW) welds in C-Mn and low alloy steels. The cracks run across the weld at roughly 45° to the plate surface. Cut the weld along its length and they stack into a herringbone, or chevron, pattern, which is where the name comes from.
It is not a new defect. Welding researchers described it in the 1970s and traced it back to hydrogen. It still turns up on thick SAW seams today, and it still gets missed, because a routine weld inspection is not set up to look for it.
What chevron cracks look like
- Orientation. Transverse to the welding direction and inclined at about 45° through the thickness. Neighbouring cracks can lean in opposite directions.
- Location. Inside the weld metal, usually in the middle of the cross section of a thick multi-pass weld. Most never reach the surface.
- Size and number. Individually small, often a few millimetres, but they come in rows. A single seam can hold dozens of them.
- Under the microscope. The crack path is step-like rather than straight, a typical sign of hydrogen cracking.
Because they are buried, tight and inclined, a visual check, MT or PT on the finished weld will normally show nothing.
Why chevron cracking happens
Hydrogen cracking needs three things at once: hydrogen in the weld, a microstructure or strength level that is susceptible, and tensile stress. Thick multi-pass SAW welds tend to supply all three.
- Hydrogen. In SAW the main source is moisture picked up by the flux. Early studies found that baking the flux at high temperature removed or sharply reduced the cracking, which is how hydrogen was confirmed as the controlling factor.
- Strong weld metal. When the weld metal is harder or stronger than the parent plate, cracking moves out of the heat affected zone and into the weld itself.
- Restraint. Thick sections and many passes build up high residual stress along the length of the weld. That stress pulls the cracks open across the weld, which is why they lie transverse.
Hydrogen cracks are also delayed. They can form hours or days after the weld has cooled, so a weld that is clean on the day it was made is not necessarily clean a week later.
How to detect chevron cracks
This is where most inspection plans fall short. The cracks are planar and oriented across the weld, so the method has to send sound or radiation in a direction that meets the crack face.
| Method | Will it find chevron cracks? |
|---|---|
| VT, MT, PT | Only if a crack breaks the surface, which is rare |
| RT | Unreliable. Tight cracks inclined to the beam leave little or no image |
| UT, probe beside the cap | Usually not. The beam runs almost parallel to the crack face |
| UT, probe on a dressed cap, beam along the weld | Yes, when run in both directions along the weld |
| PAUT | Yes, if the scan plan includes the transverse direction. A standard weld setup does not |
| TOFD | May show indications at mid-wall, but confirm with pulse-echo on the cap before accepting |
In ISO 17640 terms, that means the transverse scan positions on the weld (C, D, E, F), not just the usual positions on the parent metal. Timing matters as well. Specifications for thick or high-strength welds commonly hold final NDT until 24 to 48 hours after welding so delayed cracks have time to show up.
We wrote up a real case where two UT sign-offs missed 65 of these cracks in a 60 mm seam: The weld passed UT twice and still had chevron cracks.
How to prevent chevron cracking
- Bake and store SAW flux and electrodes exactly as the manufacturer specifies, and recycle flux only under control
- Use low-hydrogen consumables with a diffusible hydrogen class such as H5 for thick or restrained joints
- Hold the preheat and interpass temperatures in the WPS, and measure them rather than estimate
- Apply post-heating for hydrogen release on thick sections where the WPS calls for it, before the weld cools down
- Avoid more weld metal strength than the design needs, since over-matched weld metal is more susceptible
Can chevron cracks be accepted or repaired?
They cannot be accepted. Fabrication codes treat cracks differently from porosity or slag: ISO 5817 does not permit cracks at any quality level, and ASME Section VIII rejects any indication interpreted as a crack. The repair is to excavate, confirm removal with MT or UT, reweld under a qualified repair procedure and inspect again after the hold time.
Before repairing, find out where the hydrogen came from. If the flux handling has not changed, the repair weld is exposed to the same risk, and other seams welded with the same batch of flux deserve a transverse scan too.
Frequently asked questions
Is chevron cracking the same as transverse cracking?
Chevron cracks are one kind of transverse crack. The name describes the arrangement: many small hydrogen cracks inside the weld metal, inclined at about 45° and stacked in a chevron pattern in a longitudinal section.
Which welding process is most affected?
Submerged arc welding, mainly because of flux moisture and the heavy multi-pass welds it is used for. It has also been reported in manual metal arc welds.
Can radiography find chevron cracks?
Not reliably. UT on a dressed cap with the beam along the weld, in both directions, is the method most likely to find them.
How long after welding should NDT be done?
Commonly 24 to 48 hours for thick or high-strength welds. Follow the delay in your code and specification, and record the actual welding and inspection times so the hold can be verified.
Chevron cracks are a hydrogen problem at the welding station and a scan direction problem at the inspection stage. Controlling only one of the two leaves the seam exposed.
人字形裂紋（chevron cracking）是多道焊焊道金屬裡的氫致冷裂，最常出現在碳錳鋼與低合金鋼的潛弧焊（SAW）焊道。裂紋橫跨焊道，跟板面約成 45°。把焊道沿長度方向剖開，會看到一排排像魚骨的人字形排列，名字就是這樣來的。
這不是新的缺陷，1970 年代焊接研究就已經描述過，並追查到氫是元兇。可是到今天，厚板潛弧焊道還是會出現，也還是常被漏掉，因為一般的焊道檢驗本來就不是為了找它而設計的。
人字形裂紋長什麼樣子
- 方向。橫向於焊接方向，沿厚度方向傾斜約 45°。相鄰的裂紋可能往相反方向傾斜。
- 位置。在焊道金屬內部，通常落在厚板多道焊斷面的中間，大多不會延伸到表面。
- 大小與數量。單一條很小，常常只有幾 mm，但會成排出現，一條焊道裡可能有幾十處。
- 顯微組織。裂紋路徑呈階梯狀而不是一直線，這是氫致裂紋的典型特徵。
因為埋在裡面、又緊又斜，完工後做目視、MT 或 PT，通常什麼都看不到。
為什麼會產生人字形裂紋
氫致裂紋要三個條件同時成立：焊道裡有氫、組織或強度容易裂、有拉應力。厚板多道潛弧焊三樣通常都齊了。
- 氫。潛弧焊的氫主要來自焊劑吸收的水氣。早期研究發現，把焊劑高溫烘乾後，裂紋消失或大幅減少，氫是主因就是這樣確認的。
- 焊道金屬強度高。焊道金屬比母材硬、比母材強時，裂紋會從熱影響區轉移到焊道本身。
- 拘束。厚板加上很多道次，沿焊道長度方向累積很高的殘留應力，把裂紋往橫跨焊道的方向拉開，所以裂紋是橫向的。
氫致裂紋還會延遲出現，焊完冷卻後幾小時甚至幾天才裂。焊完當天乾淨的焊道，一週後不一定還是乾淨的。
怎麼檢出人字形裂紋
多數檢驗計畫就是在這裡出問題。裂紋是平面型、又橫跨焊道，檢測方法必須讓音波或射線朝著能打到裂紋面的方向去。
| 方法 | 抓得到人字形裂紋嗎？ |
|---|---|
| VT、MT、PT | 裂紋要延伸到表面才看得到，很少見 |
| RT | 不可靠。又緊又斜的裂紋在底片上幾乎沒有影像 |
| UT，探頭在蓋面兩側 | 通常抓不到，音束幾乎跟裂紋面平行 |
| UT，探頭在磨平的蓋面上、沿焊道打 | 抓得到，前提是沿焊道正反兩個方向都掃 |
| PAUT | 掃查計畫有含橫向才抓得到，標準焊道設定沒有 |
| TOFD | 可能在壁厚中間看到訊號，允收前要用蓋面上的脈衝回波確認 |
用 ISO 17640 的說法，就是要做焊道上的橫向掃查位置（C、D、E、F），不能只做母材上的那幾個位置。時間點也很重要，厚板或高強度焊道的規範常要求焊後 24 到 48 小時才做最終 NDT，讓延遲裂紋有時間現形。
我們寫過一個真實案例，兩次 UT 簽核都漏掉一條 60 mm 焊道裡的 65 處這種裂紋：兩次 UT 都判合格，焊道裡還有裂紋。
怎麼預防人字形裂紋
- 潛弧焊劑與焊條完全照製造商規定烘乾與保存，焊劑回收再用要有管制
- 厚板或高拘束接頭使用低氫焊材，擴散氫等級選 H5 這一類
- 預熱與層間溫度照 WPS 做到，用量的，不要用估的
- 厚板照 WPS 要求在焊道冷卻前做後熱除氫
- 焊道金屬強度不要超過設計需要，過度匹配的焊道金屬比較容易裂
人字形裂紋可以允收或修補嗎？
不能允收。規範對裂紋的處理跟氣孔、夾渣不一樣：ISO 5817 在任何品質等級都不允許裂紋，ASME Section VIII 只要判定為裂紋的指示就判退。修補做法是挖除、用 MT 或 UT 確認清除乾淨、依合格的修補程序重焊，並在等待時間過後再檢測一次。
修補前先查出氫從哪裡來。焊劑管理沒改，修補焊道一樣有風險，同一批焊劑焊的其他焊道也值得補做一次橫向掃查。
常見問答
人字形裂紋跟橫向裂紋一樣嗎？
人字形裂紋是橫向裂紋的一種。這個名字描述的是排列方式：焊道金屬內很多小的氫致裂紋，傾斜約 45°，縱向剖面呈人字形排列。
哪一種焊接方法最容易出現？
潛弧焊，主要因為焊劑會吸濕，而且多用在厚板多道焊。手工電弧焊也有案例。
RT 抓得到人字形裂紋嗎？
不可靠。最有機會抓到的是在磨平的蓋面上做 UT，音束沿焊道正反兩個方向打。
焊後多久做 NDT？
厚板或高強度焊道常見是 24 到 48 小時。照你的規範與規格書要求的等待時間，並記錄實際的焊接與檢測時間，等待時間才查得到。
人字形裂紋在焊接站是氫的問題，在檢驗端是掃查方向的問題。只管其中一邊，焊道就還有風險。