Delving Deeper into AI For Coatings Applications

An early aspect of the paint and coatings industry that I learnt when I started writing about it is that it is a fairly “conservative” one in terms of embodying traditional techniques and technologies.

That should not detract from the fact that huge advances in the industry have been achieved over the last 50-60 years, and certainly not with the flurry of more contemporary technological offerings such as nanotechnology, graphenes and bio-based design etc.

Various current reports indicate a degree of caution on the part of the paint industry to boldly grasp greater digitalization as part of the Industry 4.0 approach, especially the AI aspects of it; yet the potential that AI offers to the industry is huge.

Understandably, one of the barriers to entry for SMEs is the cost of implementing AI technology, since businesses will want to be confident of a return on their investment: proven benefits. But for any major company, integration of an AI system is a way of advancing with greater efficiency. The sheer scope of what can be achieved is huge.

Predictive vs. Iterative Approaches in the Laboratory

Viewed purely from the angle of computational input, early potential uses for AI in the laboratory stage divide between those that are predictive and those that are iterative. Predictive applications are envisaged as being ideally suited to resin and additive design, and much of this predictive computational work can be accomplished before any synthesis takes place; one of the advantages is a more targeted route to synthesis and development that reduces cost and laboratory time.

Relationships between chemical structures and performance are as fundamental to the coatings sector as any other sector of the chemical industry, but being able to optimize resin chemistry as the essential skeletal strength of a coating can be a key factor in its future development.

Achieving these kinds of goals requires solid, reliable data sets to be able to work confidently with predictive algorithms and models in what moves from a modelling process and into simulation process. This is likely to be a highly significant area in the future because many formulators and their clients are interested in bespoke formulations that can be tailored to specific uses.

Inevitably, the design and prediction of coatings and coatings performance is something that requires industrial co-operation between raw materials companies and paint makers alike. This has already been exemplified in Europe by an approach to develop a mass-market, low-carbon footprint decorative paint formulation through co-operation between AkzoNobel, Omya and Arkema.

Iterative applications come into play as a matter of rapidly refining formulation so that performance moves closer to a target. High-throughput experimental design (HTE) represents a relatively rapid route to formula optimization through accelerating many of the iterative cycles of work that take place in research and development.

The linear mathematical approach of using least squares in a chemical/computing laboratory environment represents an intuitive, easily understood way of identifying correlations when analyzing fundamental performance outcomes according to different raw material concentrations.

AI’s ability to identify patterns and correlations in data is one of its strengths and its ability to move forward and zero in with more targeted accuracy at the laboratory stage is another. In some cases, this may be a balancing act that becomes a trade-off between different performance attributes in an optimized formula.

Defect Detection, Quality Control and Maintenance

The integrity and ideally flawlessness of substrates and the coatings applied to finish and protect them are crucial to coatings performance and AI can be incorporated both pre- and post-finishing. The risk of corrosion is the obvious enemy. Traditionally, many substrates are inspected visually before the finishing stage, which may not be powerful enough to detect surface defects early-on at a microscopic level. AI systems are now more likely to play greater roles in defect scanning either for the substrates or for the quality of the finished product.

AI systems that have been primed on, or machine-learnt with, a system that scans for quality of finish (through camera imaging) is a route to greater quality control in the future. Irregular film thicknesses, cracks, bubbles and colour anomalies are typical examples of imperfections that can be monitored and scanned for; early identification of such technical flaws allows for rapid intervention on the part of the company, which minimizes losses and downtime.

Manufacturing Maintenance and Efficiency

For any industry, but particularly the paint and coatings industry, where process outcomes may be very dependent upon production (or curing) conditions, the operational efficiency of manufacturing is vital to minimizing economic losses through equipment downtime. Predictive maintenance can be accomplished through monitoring for temperature, vibration and energy use and allows companies early opportunities to troubleshoot any faulty equipment.

Any aspect of enhanced equipment safety readily translates into greater employee safety, and where maintenance may be deemed as a necessary intervention, early identification allows for it to be planned in order to minimize disruption.

Color Matching and Development

At the consumer end of the chain, color matching in any sector is a key application for AI in paint and coatings development, and perfect results here lead to client satisfaction.

Many attractions and benefits are to be found in the area of color matching – two of them are greater accuracy in matching and superior efficiency in reaching that match. Each relies on the advantages of automation over human judgment (again), thereby reducing the time and equipment necessary to achieve a perfect match.

In addition, paint and coatings companies that work with AI may find themselves equipped with options for color development where certain shades have been elusive to achieve. A few top paint companies and certain of the major spectrophotometry specialists are already well-entrenched in using AI for color matching and color development, and such systems have been built with decades’ worth of formulation data.

Application Sectors for AI Use

The approach to mass-market decorative paints was highlighted earlier. One of the key aspects of it was that the companies pursued the optimization of just three key performance attributes. 

Since coatings demonstrate many different performance attributes, any mass-market (e.g., decorative) or major performance application (e.g., construction) may lend itself to some coatings formulae being tweaked or refined with other performance strengths in mind.

In the industrial coatings sector, areas that are rich with AI development potential include packaging, automotive, military/defense, protective and aerospace sectors. Of these, protective, aerospace and defense equipment are some of the key areas where the European industry may benefit from AI integration.

High-performing industrial coatings have already been with us for decades, but the coatings of the future are going to become even more functional in terms of the way they are designed. As end-use sectors become more advanced in their demands, so too will the demands being made on the coatings they will require.

Once again, the traditional limitations of the slow laboratory R&D will be overcome as AI furnishes routes to more targeted, more specific end-use applications. Some of this will lie outside mainstream sectors and in more specialist applications, such as electronics, though, and it heralds a greater period in coatings development where things move from coatings formulation to coatings engineering. That is, if we’re not on the verge of it now!