Sonae Arauco manufactures wood-based panels through a continuous process involving the feeding of raw materials (wood and resins from external suppliers), their processing (through heat and pressure) and finishing of the panels (sanding and cutting) and/or further processing (such as surfacing with decorative papers from external suppliers). Panels are then supplied to industrial clients from every size to satisfy a wide variety of end uses.

The Sonae Arauco’s use case will focus precisely on the process of decorative surfaced wood panels at the Linares, Spain production site towards a zero-defect manufacturing approach.

In a nutshell, the process consists in coating a raw board (particle board or medium density fiberboard) with impregnated papers that are then pressed to the raw board surface using a combination of pressure and temperature.

Detailing the process, the production flow process starts with the production of unsanded particle boards in the press. The unsanded boards are then, in the most normal flow, sanded in the sanding line. The sanded boards are then stored in the crane warehouse to be either dispatched as finished goods, to be cutted or to be impregnated.

The production is done in 2 lines which cover the clients’ demand. The surfaced panels production plan triggers the requirements for the impregnation line and for the sanded particle boards (or MDF boards purchased to Valladolid, Spain and Mangualde, Portugal). The impregnation production transforms the dry paper rolls in pallets of impregnated paper. The paper is either transported immediately to the decorative surfaced presses or stored in the systraplan warehouse.

The produced decorative surfaced boards, combining the paper and boards and finishing type – given by the finishing plates – are then stored in the finished goods warehouse. The melamine boards can be cutted either in the decorative surface line or in next operation in the Giben saw. 

Expected goals

Industries around the world require high level quality. As client’s needs and product combinations and its complexity expand, industries face real challenges in detecting production flaws. The wood-based panels’ industry isn’t different. Traditionally industries invest in equipment to detect quality issues as they happen in the production line. Sonae Arauco wants to take a different and innovate approach – to use data and artificial intelligence to predict quality towards a zero-defect approach.

Non-compliant wood surfaced panels that present some type of defect must be downgraded to 2nd quality, sold for a far lower price, or 3rd quality where they are sent to recycling. This can amount to a loss of around 900k€ per year in a single factory.

Additionally, resources (such as wood, energy and gas or chemicals) are used to manufacture panels of lower-than-expected quality, representing a waste that manufacturing companies want to avoid for both economic and sustainability reasons.

Identified challenges

• Improve decision-making at the plant level to reduce defects towards a zero defects paradigm through a data-driven defect prediction based on the current process data and historical information: the main goal is to predict defects beforehand such as broken paper, dust or glued paper so that changes in machinery/parameters can be performed and the defects avoided
• Improve product development by making it faster and with less waste generated through a recommendation process of recipes for new products being developed: one key challenge when using data-driven approaches is the need for a data collection period that is fully representative of the real-world conditions. This normally involves high costs, not only from data collection but from the wasted material that is required in the ramp-up phase of a new process.
• Improve evaluation of machinery components through an estimation of the machine components’ degradation through drifts in data distribution used to detect defects: normally, data collected once is only representative for a particular moment in time, meaning for a particular state of the machine. As soon as the machine starts operating, slight and smooth changes in the process start occurring due to component wear-out and, if the data is representative enough, it will lead to a different data distribution in time.

Impact from openZDM technologies

Help to increase the TRL of the previous results and further develop them, through the digital twin enabled ML approaches for data analytics towards ZDM.

Improvement of data collection and in-situ monitoring namely in what concerns with time-related annotation.

Improvement of inspection of quality issues through artificial intelligence and digital-twin technology.

Further advances in predictive quality based on data-driven and model-driven approaches for defect prediction and quality assessment.

Improvement in manufacturing decision-making for ZDM strategies and process adaptation.

Overall, these points will impact Sonae Arauco on the following:

• Improving product quality (through the reduction of defects);
• Promoting better usage of resources (wood, chemicals, energy, time…);
• Helping shopfloor staff (through better decision making).