The vast majority of executives in the chemical industry recognise the need for heightened investment in AI, so why isn’t it happening? Benson Wang, product director for Profet AI, explains.
Eighty per cent of chemicals executives surveyed for a recent IBM report say that AI will be important for the success of their business over the next three years.
Yet only four in 10 said their company has already implemented an enterprise-wide AI strategy.
Clearly, while it’s widely understood that AI is critical for survival, there are currently obstructions to mass adoption, which can be narrowed down to three areas:
1. Lack of expertise
Finding experts who possess both AI and domain-specific knowledge is a major challenge. Developing and implementing AI in chemistry requires a unique blend of expertise in chemistry, data science, and computer science, thus a shortage of professionals with these interdisciplinary skills is proving a major hindrance.
2. Lack of organisational buy-in
Gaining organisation-wide buy-in for AI initiatives in the chemical manufacturing sector is also a major challenge. Many AI initiatives are led by data or tech teams, which ends up creating resistance from other employees who view them as a hindrance to their work. This is especially true if they don’t have vocal championing from top management.
3. The perception that AI initiatives are time-consuming and costly
Implementing AI solutions in the sector can be time-consuming, often taking over three months to see tangible results, which makes it difficult to prioritise teams that are already overloaded. Moreover, there is a general perception that AI will be expensive and use up a lot of resources.
While these issues are pertinent, if CEOs get behind AI and bring in appropriate training, companies are often quite surprised to find that they can quickly see results in terms of streamlining operations, improving product quality, reducing costs, and driving new product development.
We outline some examples where benefits are commonly identified below.
Major Use Cases of AI in Chemical Manufacturing
Raw materials costs account for 50-70% of chemical companies’ sales revenue, meaning it’s critical to a company’s bottom line for procurement managers to secure the best possible pricing. This is challenging in a complex chemical supply chain that involves manufacturers, distributors and retailers, and where pricing is always in flux. Procurement managers typically require more holistic visibility of the whole situation to guide pricing decisions.
Here in Taiwan, we find that many manufacturers tend to make procurement decisions based on key staff’s personal judgment or – if they do utilise data – it tends to be centered on historical trend averages for material price. The weakness of this method is that it fails to consider wider macroeconomic conditions and changes.
Predictive analytics can make a difference and gift procurement with a much clearer picture through models utilising various factors such as crude oil futures data, plastic futures data, exchange rates, market volatility, behavioral science, and geopolitical trends. This enables managers to be proactive rather than reactive in their decision-making process.
Research and Development
Chemical industries have begun to use generative AI models in their R&D process to identify new molecules, recipes, or compounds.
In a recent webinar, McKinsey & Company demonstrated the pace at which an AI model trained on an extensive database of chemical compounds could identify new compounds. AI can expedite the discovery process by two or three times and find molecules that are much better in specific properties that companies are interested in.
In the chemical industry, it is crucial to take immediate action in the event of a defect in the production line, as it can lead to the contamination and spoilage of an entire batch. Traditional manufacturers often rely on personnel experience and trial-and-error experiments to address quality issues, which is not ideal in terms of effectiveness and means they still struggle to find the root of problems.
The combination of AI-based tools, sensors, and computer vision technologies is perfect for helping companies not only quickly identify and resolve issues, but also learn from them and prevent similar problems in the future.
We worked with one client specialising in cellulose nanofiber coatings, which uses Roll-to-Roll processing in which films or soft boards are unrolled from cylindrical rolls. This is a complex process with a relatively high defect rate of 12%. AI-powered quality factor analysis allowed them to quickly resolve defect issues, reducing the defect rate to only 5%.
Meeting ESG targets
Demands from both governments and consumers have made sustainability a critical concern for chemical manufacturers, especially in light of recent research showing that the chemical industry accounts for approximately 10% of global total final energy consumption and 7% of greenhouse gas emissions. Accordingly, 82% of chemical executives now prioritise environmental, social, and governance (ESG) and sustainability as much as revenue growth.
Artificial intelligence (AI) has proven to be an effective tool in optimising operations for chemical factories to meet sustainability targets, particularly in reducing energy consumption, including electricity usage.
This is done by using data such as historical electricity consumption, environmental conditions (temperature and humidity), production information, and equipment operation data. Commonly we have observed that AI adoption can help chemical companies in Taiwan reduce their electricity consumption by around 3%.
Benson Wang is Product Director for Profet AI, an AutoML enterprise solution for the manufacturing sector.