The most troublesome failure in laser cladding isn't the failure to clad, but rather the successful cladding and formation, only to crack upon cooling. This is especially true for thick-layer, high-hardness, and high-alloy systems, where cracking is almost inevitable.
For components like blade repair, mold reinforcement, and valve seals-parts whose lifespan depends on the cladding layer-cracking upon cooling not only requires rework but may also render the substrate unusable. Therefore, cracking isn't just an "aesthetic issue"; it's a direct signal that the process failed to manage stress.
Below, arranged from root cause to adjustable parameters, we list seven of the most common causes of cracking, along with which knob to adjust first. A supplementary section at the end explains how AI can help adjust cladding parameters.
1. Excessive Thermal Stress (Most Common)
Symptoms: Macroscopic longitudinal/transverse cracks, often appearing in the middle or ends of the cladding layer. Principle: Rapid heating and cooling of the laser, combined with the difference in expansion coefficients between the cladding layer and the substrate, generates enormous tensile stress upon cooling, exceeding the material's strength and causing cracking. First, adjust the substrate: Preheat the substrate (reduce the cooling rate and temperature difference), this is the most cost-effective step; simultaneously control the thickness of each layer, avoid being too thick. Common mistake: Only reducing laser power to be "gentler," without preheating. Reducing power worsens fusion, stress is not resolved, and it actually makes things worse. Empirically, preheating temperature for steel substrates is often in the range of 150–300℃ (depending on the system), and higher for high alloys.
2. Mismatch between powder and substrate thermal properties
Phenomenon: Cracks propagate along the cladding layer-substrate interface. Principle: A large difference in expansion coefficient/thermal conductivity leads to stress concentration at the interface, becoming the preferred location for crack initiation. First, adjust the powder: Select powders closely matching the substrate; if the difference is large, add a transition layer (compositional gradient transition, such as a Ni-based transition layer). Common mistake: Directly applying high-hardness powders (such as tungsten carbide cermet) for performance without adding a transition layer. A large difference in expansion coefficients between the hard layer and the steel substrate will inevitably lead to interface cracking.
3. Excessive Scanning Speed
Phenomenon: Rapid cooling results in fine but brittle grains with high cooling stress. Principle: Low linear energy (power/speed) → high cooling rate → high residual tensile stress; the strength gain from fine grains is offset by stress. Solution: Appropriately reduce scanning speed / increase linear energy, but be careful not to overdo it and cause an increase in dilution rate. Common Misconception: Insisting on increasing speed for efficiency. Higher speed means higher production capacity, but the risk of cracking increases simultaneously, leading to greater losses from rework later.
4. Stress Concentration in Multi-Pass Overlap Zones
Phenomenon: Transverse micro-cracks at the overlap seam. Principle: The overlap zone undergoes multiple thermal cycles, resulting in stress accumulation; improper overlap ratio can also cause localized tempering softening or missed coverage. Solution: Control the overlap ratio (usually within the empirical range of 30%~50%); slow cooling between layers to avoid heat accumulation; multi-pass sequences can use bidirectional or island-type methods to reduce accumulated deformation. Common Misconception: A larger overlap ratio is "safer". Too large an overlap ratio leads to repeated heat input and a wider softening zone; too small an overlap ratio results in missed coverage and uneven stress.
5. Excessive Dilution Rate
Phenomenon: Embrittlement of the cladding layer, cracks along the brittle phase. Principle: Excessive mixing of matrix elements (especially Fe), alloy composition deviating from design, resulting in the appearance of brittle phases such as martensite, and a sharp drop in toughness. Solution: Reduce laser power / increase powder feed / control defocusing amount to bring the dilution rate to a reasonable range (empirically, thick layers are often controlled at a lower percentage). Common Misconception: Over-increasing power for "stronger fusion." Fusion and dilution are two different things; excessive power leads to a soaring dilution rate, making the molten layer more brittle and prone to cracking.
6. Insufficient Shielding Gas
Phenomenon: Porosity and oxide inclusions become crack initiation points. Principle: Oxide scale and porosity become crack initiation points under service stress, especially if the molten pool is exposed to air, resulting in widespread inclusions. Solution: Ensure adequate shielding gas flow and coverage (Ar is commonly used; flow rate is set empirically based on the laser spot and nozzle), protecting both the laser path and the molten pool; standardize atmosphere operation. Common Misconception: Only providing gas to the laser path, forgetting to provide lateral protection to the molten pool. Crack initiation often lies on the unprotected side.
7. Residual Stress Accumulation (Thick/Multi-Layer)
Phenomenon: Overall cracking after multi-layer welding. Principle: Stress accumulation in each layer exceeds the critical total stress. First adjustment: Interlayer stress relief (hammering, vibration, or intermediate annealing); control total layer thickness and interlayer temperature to avoid excessively thick layers at once. Common misconception: Continuously piling thick layers to rush progress. The more layers there are without treatment, the closer the total stress gets to the critical level, eventually leading to overall collapse.
Systematic troubleshooting approach (locate the crack first)
Check crack location: Transverse cracks in the middle/end → often point to thermal stress or dilution (1, 5); cracks along the interface → matching problems (2); lap joint cracks → multiple cracks (4).
Check associated defects: Porosity and oxidation → protective gas (6); brittle phase/compositional deviation → dilution (5).
Check the number of layers: Single-layer cracks deviate from stress and matching; multi-layer cracks accumulate overall (7). After positioning, adjust in the following order. Don't turn all knobs at once, otherwise you won't know which step is effective.
Suggested Adjustment Order (Don't adjust randomly):
First, preheat (to treat thermal stress, most cost-effective).
Then, adjust the powder matching/transition layer (to treat the interface).
Next, adjust the scanning speed and line energy (to treat the cooling rate).
Then, adjust the overlap ratio (to treat multi-pass welding).
Finally, control the dilution rate, protective gas, and interlayer stress (to treat microscopic and cumulative stress).
Remember: Cracking is the result of "stress > strength". All adjustments are essentially about compressing stress or improving toughness. Adjusting according to this logic is much faster than blindly experimenting with parameters.
To what extent can AI help with cladding parameter tuning (Frontier Supplement):
The previous order is "human parameter tuning," while AI is turning trial and error into prediction. Several truly feasible directions:
Process-performance mapping modeling: Train the ML with historical "power/speed/powder → crack rate" data. For new systems, let the model predict risks first, then decide whether to run the machine, saving a lot of waste samples.
Online molten pool monitoring: High-speed cameras and computer vision are used to observe molten pool morphology/plume, providing real-time alarms for anomalies and intervening before cracking occurs-far more proactive than post-cracking detection.
Automatic parameter tuning (reinforcement learning): Under the constraint of "no cracking + meeting performance standards," the algorithm searches for the optimal parameter combination, particularly suitable for thick-layer processes with multiple coupled variables.
Intelligent defect identification: Automatically classifies porosity, lack of fusion, and cracks, resulting in quality inspection efficiency far exceeding that of manual methods.
Reminder: The upper limit of an AI model is your accumulated process data. If the first seven root causes are not clearly understood and experimental records are disorganized, the model will not learn anything. Standardize basic experiments first, then let AI help you improve efficiency-don't reverse the order.