Betreuer/in: Fellerer
Warehouses are key nodes in modern supply chains, yet their performance is still primarily assessed using cost, throughput, inventory levels, and labour productivity. While these indicators remain important, they do not adequately capture sustainability dimensions such as resource efficiency, waste minimisation, and process reliability. As warehouse operations become increasingly digitalised through systems such as SAP Extended Warehouse Management (EWM), new opportunities arise to develop sustainability-oriented KPI systems based on existing enterprise data.
This thesis develops a Large Language Model (LLM)-supported KPI governance framework for sustainable warehouse operations. The objective is to identify, structure, prioritise, and operationalise sustainability-specific KPIs that are compatible with SAP EWM. The study follows a five-phase Design Science Research methodology, combining literature review, KPI engineering, prototype development, and evaluation.
A PRISMA-guided literature review across Scopus, IEEE Xplore, and Google Scholar identified 61 relevant publications. From these sources, 316 sustainability-relevant KPIs were extracted and organised into nine operational categories. Based on sustainability relevance and expected SAP EWM data availability, 158 indicators were prioritised as high-value KPIs covering process efficiency, inventory control, service responsiveness, and space utilisation.
The framework was implemented as a prototype consisting of a KPI repository, an interactive dashboard, and a controlled enrichment pipeline integrated within the Siemens LLM environment. Missing definitions or formulas are generated only when no literature-based values are available. All generated content is explicitly labelled and requires human approval, ensuring transparency and maintaining a literature-first governance principle.
The results show that many sustainability-oriented KPIs can be derived from existing enterprise data structures without requiring additional sensing infrastructure. This thesis contributes a consolidated KPI taxonomy, a transparent governance architecture, and a reproducible prototype for AI-supported KPI management.
Room 04.137, Martensstr. 3, Erlangen
or
Zoom:
https://fau.zoom-x.de/j/68350702053?pwd=UkF3aXY0QUdjeSsyR0tyRWtLQ0hYUT09
Meeting-ID: 683 5070 2053
Kenncode: 647333