Cost optimisation in chemicals
Chemical companies can optimise costs by analysing yield drift and feedstock variability, then implementing changes to mitigate their impact, such as caching and batching, to reduce uncertainty and make informed decisions about resource allocation, thereby improving their bottom line.
Despite the promise of AI cost optimisation in chemicals, many initiatives fail to deliver lasting reductions, often because they overlook the intricacies of continuous processes with tight yield and safety envelopes, where small changes can have significant consequences. The business consequence of such failures is a lack of trust in the optimisation process, leading to a failure to implement changes that could have a real impact on the bottom line.
Chemicals industry challenges
The chemicals industry presents unique challenges for cost optimisation, primarily due to yield drift, feedstock variability, and plant-to-plant recipe differences. Yield drift, which refers to the gradual change in production yields over time, can significantly impact cost calculations, as small changes in yield can result in substantial changes in costs. Feedstock variability, on the other hand, introduces uncertainty into the production process, making it difficult to predict costs accurately. Furthermore, differences in recipes between plants can lead to inconsistencies in cost calculations, making it challenging to compare costs across different facilities.
A shift supervisor checks the maintenance log to identify areas where yield drift is most pronounced, and then works with the production team to implement changes to mitigate its impact. By understanding the sources of yield drift and feedstock variability, and by implementing strategies to address these issues, companies can reduce the uncertainty associated with cost calculations and make more informed decisions about where to allocate resources.
Unit cost vs total cost
A common mistake in cost optimisation is focusing solely on unit cost, without considering the total cost of an AI workflow. Unit cost refers to the cost of producing a single unit of a product, whereas total cost takes into account all the costs associated with producing a batch of products, including overheads, maintenance, and other expenses. By ignoring the total cost, companies may overlook opportunities to reduce costs that are not immediately apparent, such as reducing energy consumption or minimizing waste.
The difference between unit cost and total cost can be significant, and companies that fail to consider both may find that their cost optimisation efforts are less effective than they could be. For example, a company may reduce the unit cost of a product by optimizing the production process, but if the total cost of producing the product remains high due to other factors, such as energy consumption or waste disposal, the overall cost savings may be limited.
Real levers for cost optimisation
There are three real levers for cost optimisation in the chemicals industry: caching, batching, and model routing. Caching involves storing frequently used data or results to reduce the need for redundant calculations, which can help reduce computational costs. Batching involves grouping similar tasks or processes together to reduce overheads and improve efficiency. Model routing, on the other hand, involves selecting the most appropriate model or algorithm for a given task, based on factors such as accuracy, speed, and cost.
By using these levers, companies can reduce costs and improve efficiency, without compromising on yield or safety. For example, a company may use caching to store the results of complex calculations, reducing the need for redundant calculations and improving the speed of the production process. Similarly, batching can be used to group similar tasks together, reducing overheads and improving efficiency.
For more information on how to implement these levers, companies can visit our website via this link to learn more about our cost optimisation solutions.
Token cost is rarely the dominant line
Token cost, which refers to the cost of using a particular AI model or algorithm, is often not the dominant line in the cost calculation. While token cost can be significant, other costs, such as data preparation, model training, and deployment, can be much higher. Furthermore, the cost of maintaining and updating AI models over time can also be substantial, and companies that fail to consider these costs may find that their cost optimisation efforts are less effective than they could be.
Companies that focus solely on reducing token cost may overlook other opportunities to reduce costs, such as improving data quality, reducing energy consumption, or minimizing waste. By taking a more holistic approach to cost optimisation, companies can identify areas where costs can be reduced, and implement strategies to achieve these reductions.
Limitations of AI cost optimisation
AI cost optimisation is not a panacea for all cost reduction challenges, and there are situations where it may not be the best solution. For example, in situations where the production process is highly variable, or where the costs are highly uncertain, AI cost optimisation may not be effective. In such cases, other approaches, such as traditional cost accounting or process optimization, may be more effective.
Companies that are considering implementing AI cost optimisation should carefully evaluate their needs and circumstances before deciding whether to proceed. They should also consider seeking advice from experts, such as our team, via this link, to determine whether AI cost optimisation is the right solution for their specific challenges.
Frequently asked questions
What are the unique challenges of cost optimisation in the chemicals industry?
Yield drift, feedstock variability, and plant-to-plant recipe differences pose significant challenges for cost optimisation in the chemicals industry.
Why is it important to consider total cost, not just unit cost, in cost optimisation?
Focusing solely on unit cost overlooks opportunities to reduce costs associated with overheads, maintenance, and other expenses, which can limit overall cost savings.
What are the real levers for cost optimisation in the chemicals industry?
Caching, batching, and model routing are effective levers for cost optimisation, helping to reduce computational costs, improve efficiency, and select the most appropriate models or algorithms.
When may AI cost optimisation not be the best solution for cost reduction challenges?
AI cost optimisation may not be effective in situations with highly variable production processes or uncertain costs, where traditional cost accounting or process optimization may be more suitable.