BUC-5 — AI process control for coffee & utilities (TAF).

AI predicts steam demand and controls extraction cycle, drying and process settings within approved limits. Today: • In coffee process, the extraction process needs to be controlled within a defined cycle time to achieve the best yield of caffeine. Operators need to adjust flow, pressure, pump speed and other settings. Variation is hard to control or predict in time, so the cycle time target is not always met, which could affect production capacity and utilization. • In operation, the process maintains a fixed level of steam to ensure supply to process areas, kept higher than requirement to buffer complex operation scenarios that can consume more/less steam at a time. • In coffee process, the drying process has many coffee strains as input materials which differ in size, moisture, weight. They undergo multiple heaters, fans, decks with different sets of parameters, controlled by operators to achieve a consistent output (total solid, moistures) of products. • Production processes require frequent manual adjustments to maintain quality, yield and energy performance. Process variability and changing operating conditions make optimization highly dependent on operator experience.

• Help TAF operators keep every coffee extraction cycle on target time with the best yield, without manual adjustment – measured by the share of cycles on target time and extraction yield. • Improve extraction efficiency, production capacity, and process consistency. Reduce manual adjustment delay, minimize yield loss, lower conversion cost per ton while supporting more stable product quality. • Reduce energy consumption and operating costs, save working hours through automated boiler adjustment, and ensure a stable steam supply to minimize process delays and production losses. • Improve product consistency, reduce energy and material loss, increase process efficiency, and reduce workload and dependency on operator experience. • Improve yield and quality consistency, reduce energy consumption, minimize manual adjustments and increase production efficiency.

• TAF operators • Technical / factory process teams

Note: Synthetic data could be potentially provided during the process, upon approval of the Nestlé team and in compliance with data law.

To be determined Note: Nestlé's current environment includes Microsoft Azure and Microsoft Copilot. Solutions that can work with this environment are welcome, and final integration and deployment requirements will be confirmed with the Nestlé team.