A web-based interactive tool for benchmarking quay crane efficiency using Multi-directional Efficiency Analysis (MEA), visual analytics, and a conversational AI assistant. It helps port managers compare equipment performance, identify variable-level inefficiency drivers, explore operational trends, and obtain natural language decision support without requiring expertise in efficiency analysis methods. This tool can be used to measure efficient of other types of port equipment.
-
Main Gain(s): Improved equipment efficiency diagnosis, Support for operational benchmarking, Faster identification of inefficiency drivers, Natural language decision support for port managers
-
Use Case(s): Operational data from quay cranes at the Port of Sines
-
Start TRL: 0 - (evidences)
-
Final TRL: 0 - (evidences)
-
Main Contributions: (theoretics) S. Kazmi, E. Rocha, A. Brochado; (implementation) L. Ribeiro; (integration) E. Rocha
Main Feature
- Interactive dashboard for quay crane efficiency benchmarking
- Multi-directional Efficiency Analysis (MEA) for equipment performance assessment
- Variable-level diagnosis of inefficiency drivers
- Colour-coded efficiency rankings and temporal trend analysis
- Side-by-side comparison of algorithmic results (linear vs non-linear versions)
- Conversational AI assistant for natural language decision support
- Backend using pre-computed MEA scores from operational data
Videos
| Product Interface Demonstration |
Product Development Context
Seaports generate large volumes of operational data from equipment such as quay cranes. However, existing business intelligence tools typically report aggregated KPIs without supporting efficiency benchmarking or variable-level diagnosis. This product was developed to address that gap by providing port managers with a structured and interactive means of analyzing equipment efficiency in the context of competing operational objectives.
Product Definition and Benefits for Users
The product is a web-based interactive dashboard integrating Multi-directional Efficiency Analysis (MEA), visual analytics, and a conversational AI assistant. It enables port managers to benchmark equipment performance, identify specific inefficiency drivers, and obtain natural language decision support without requiring expertise in efficiency analysis methods.
Product Characterization – Technical Specs
1. Frontend
The interface is organised into five main pages:
- Configuration: variable mapping and algorithm selection
- Statistics: exploratory data analysis
- Results: efficiency benchmark and temporal trends
- Comparison: side-by-side algorithm comparison
- Assistant: conversational interface For example, to identify the least efficient equipment unit, the user navigates to the Results page, where a colour-coded bar chart ranks all units by comprehensive efficiency score.
2. Backend
MEA scores are pre-computed offline from raw operational data in Excel format using dedicated MEA processing scripts. These scripts generate CSV files that are ingested by the dashboard at runtime. The conversational assistant uses GPT-4o-mini via the OpenAI API and is augmented with 14 deterministic analytical tools that retrieve verified data values before generating responses.
Product Testing, Validation and Evaluation
The product was validated through a case study using operational data from quay cranes at the Port of Sines (2022). A formal user evaluation with target-audience participants is currently underway. The evaluation measures task success rate, completion time, and perceived usability through the System Usability Scale (SUS) questionnaire.
Product´s Internal and External Limitations/Restrictions
Internally, the conversational assistant does not enforce tool usage for all query types, which introduces residual hallucination risk. Conversation history is also not persisted across sessions. Externally, the framework requires pre-computed MEA scores as input. In addition, raw operational data must follow the expected schema and variable definitions for the results to be meaningful.