Akurasi OFA terhadap Loading Time Excavator Tambang Batubara PT Antareja Mahada Makmur (AMM) Jobsite PIK

Abstract. Accurate equipment time monitoring is essential for productivity control and operational efficiency in open-pit coal mining. PT Antareja Mahada Makmur (AMM) at the PT Perkasa Inakakerta (PIK) jobsite uses Onboard FleetSafe Assist (OFA) to record fleet activities, including excavator loading time. This study evaluates OFA loading-time accuracy against field time-study measurements at Pit 81 and proposes an operational deviation threshold for data quality control. Data from October–November 2025 were matched per haul cycle (n=312 paired records; 81.04% coverage of 385 observed cycles). Because paired differences were non-normal, the Wilcoxon Signed-Rank test was applied. Accuracy metrics (bias, MAE, RMSE) and agreement via Bland–Altman Limits of Agreement (LoA) were computed, and a deviation threshold was derived using Tukey’s rule. The mean bias was −0.06 min, MAE 0.60 min, and RMSE 1.206 min, indicating small average deviation with several outliers. The 95% LoA ranged from −2.42 to 2.30 min, with 92.3% of records within the LoA band. The Tukey threshold was 0.706 min (18.43%), classifying 79.2% of cycles as within-threshold and 20.8% as deviations. Overall, OFA loading time is suitable for routine monitoring, but high-deviation cycles should be flagged and investigated through event-definition alignment, time synchronization, and system configuration review. Abstrak. Pemantauan waktu kerja alat yang akurat diperlukan untuk mengendalikan produktivitas dan efisiensi operasi tambang batubara terbuka. PT Antareja Mahada Makmur (AMM) pada jobsite PT Perkasa Inakakerta (PIK) menggunakan Onboard FleetSafe Assist (OFA) untuk merekam aktivitas armada, termasuk loading time excavator. Penelitian ini mengevaluasi akurasi data loading time OFA terhadap data aktual hasil time study lapangan di Pit 81 serta menetapkan ambang operasional deviasi untuk kontrol mutu data. Data Oktober–November 2025 dipadankan per ritase antara log OFA dan observasi lapangan (n=312 pasangan; coverage 81,04% dari 385 ritase). Karena selisih tidak berdistribusi normal, digunakan uji Wilcoxon Signed-Rank. Ukuran akurasi dihitung (bias, MAE, RMSE), kesepakatan dinilai dengan Bland–Altman Limits of Agreement (LoA), dan ambang deviasi ditetapkan menggunakan aturan Tukey. Hasil menunjukkan bias rata-rata −0,06 menit, MAE 0,60 menit, dan RMSE 1,206 menit, mengindikasikan penyimpangan rata-rata kecil namun terdapat outlier. LoA 95% berada pada −2,42 hingga 2,30 menit dengan 92,3% data berada dalam rentang LoA. Ambang Tukey 0,706 menit (18,43%) mengklasifikasikan 79,2% ritase sesuai dan 20,8% deviasi. Secara umum, data OFA cukup representatif untuk pemantauan rutin, namun ritase deviasi besar perlu ditandai dan ditinjau; penyelarasan definisi event, sinkronisasi waktu, dan perbaikan konfigurasi sistem direkomendasikan.