Design and Performance Evaluation of an Automated Oil Palm Fruit Processing System for Improved Production Efficiency

Abstract

Automation in palm oil processing has advanced most rapidly at the point where it is easiest to demonstrate, namely image-based fruit grading, while the mechanical and control subsystems that would allow a grading decision to change what the plant actually does remain largely undeveloped. This paper presents the design of an automated oil palm fruit processing system and a framework for evaluating its performance. The study followed a design research approach in which subsystem requirements were derived from published automation and grading work indexed in Scopus and SINTA, a system design was synthesised, and a performance evaluation framework was constructed and applied analytically. The proposed system comprises four subsystems: automated reception grading, batch identity tracking, adaptive sterilization control, and press and clarification monitoring, linked by a supervisory layer that carries grading information forward to downstream setpoints. The review of published grading work indicates that classification performance is now adequate for operational use, with reported accuracies in the region of eighty per cent for six-grade mobile classification and higher for coarser category schemes, and that annotated datasets collected from operating mill grading stations are publicly available. The evaluation framework proposed here specifies six measures spanning classification agreement, batch traceability, control response, oil recovery, energy consumption, and availability. The paper argues that automation value in this application arises from closing the loop between grading and process control rather than from grading accuracy alone.