Speaker
Description
Uncertainty in harvesting output and road accessibility significantly affects supply chain planning and the ability to meet customers’ demand for high-quality wood products in a timely manner. Despite a long-term harvest scheduling, variations in product volumes can reduce delivery precision and increase the risk of product backorders. Weather conditions also influence landing accessibility; for example, schedules based on expected road conditions may suddenly change due to warmer weather, thawing, or heavy rainfall.
In response to these challenges, there is interest in modeling variations, such as predicting road accessibility, which is further stressed by climate change. Predicting product variations and road accessibility can support forest managers in operational planning; however, decision-making on transport prioritization must remain aligned with planning objectives to ensure that established delivery performance targets are met. Evaluating road accessibility is an important tool, but it should be complemented by decisions on transport prioritization to meet delivery goals, such as minimizing backorders and delivery variations.
To address this problem, we present a methodology that uses an AI agent for decision-making on transport prioritization, fed with data from a digital twin to monitor landing inventory levels and road accessibility. Despite operational variability, the agent’s goal is to prioritize transports to maintain adherence to multiple long-term delivery targets.
The methodology is implemented in a decision-support prototype that enables managers to track long-term planning goals on a dashboard and to evaluate how operational decisions based on predicted road accessibility and real-time landing inventories align with delivery targets. The dashboard visually helps identify landings at risk of reduced road accessibility and provides guidance on prioritizing transport. It also supports the comparison of alternative decision scenarios, helping managers determine when a new long-term harvest schedule may be needed.
The prototype is expected to be further developed and integrated into operational environments in future projects in collaboration with forestry stakeholders.
| Keywords | digital twin; decision support |
|---|
