Welding Wood-Plastic Composites

Researchers are using machine learning (ML) to unlock the potential of an alternative welding technique for joining wood-plastic composites (WPCs).

WPCs use a combination of wood fiber or flour and polyethylene, polyvinyl chloride, or polypropylene. This material is common in construction applications, outdoor furniture, and automotive interiors. It provides superior degradation resistance compared to traditional wood. Typical joining mechanisms for WPCs include screws and bolts, as well as adhesive bonding. Thermal welding techniques are another possible joining mechanism but can lead to degradation of the polymers at high temperatures. Strengthening bonds without damaging the aesthetic or functional properties of WPCs remains an obstacle for this material.

You can also read: HDPE: Superior Performance and Durability in Construction.

Frictional stir welding (FSW), a process that softens the polymer without causing it to melt, shows potential as an alternative. This process is a solid-state joining method that involves stirring softened composites along their joint line. Once the composite cools, FSW creates a solid bond between two joined parts. This approach has several advantages, including:

• Energy efficiency
• Minimal waste production
• High strength joints
• Lack of joint defects

Prior FSW research mainly pertains to metal welding; applying FSW to WPCs is a novel area of research. As WPCs continue to gain popularity in the construction and furniture industry, FSW could lead to increased performance and aesthetics.

Challenges of FSW in WPCs

Using FSW for the joining of WPCs presents unique challenges compared to metal. The thermoplastic and wood fibers of WPCs have a non-uniform heat distribution. Thus, issues such as inconsistent welds and the burning of wood fibers can weaken the joint. Additionally, depending on the type of polymer used, elevated temperatures can lead to polymer degradation. The wood fibers in WPCs contain moisture, which can vaporize during FSW, leading to bubbles or voids in the composite. Achieving uniform welds can also be challenging due to the reduced flowability of the composite.

Optimizing Performance with Machine Learning

To overcome these challenges and unlock the potential of FSW, researchers have employed ML. They modeled how composite parameters effect performance using three ML algorithms. These included:

• Multilayer perceptron (MLP): An artificial neural network that uses multiple layers that is effective for modeling non-linear relationships.
• Decision tree (DT): A method of modeling decisions and consequences, producing a simple-to-understand model.
• Adaptive neuro-fuzzy inference system (ANFIS): A system that can model complex relationships using expert knowledge by adjusting fuzzy rules using a neural network framework.

Using these algorithms, researchers sought to optimize rotational speed and welding speed during FSW. The optimized outputs included flexural modulus and flexural strength at entry and exit.

Table – The ML optimization output parameters for three different objectives, which manufacturers can employ to achieve better FSW results.

With these optimized parameters, manufacturers can ensure joints welded by FSW meet performance and reliability standards. Refining approaches to this welding technique can lead to better performance and aesthetics of WPCs.