Monitoring Kualitas Produk Ordinary Portland Cement (OPC) Menggunakan Maximum Half-Normal Multivariate Control Chart (Max-Half-Mchart) Berbasis Estimator Det-MCD

Pasadina, Garin Sinta (2025) Monitoring Kualitas Produk Ordinary Portland Cement (OPC) Menggunakan Maximum Half-Normal Multivariate Control Chart (Max-Half-Mchart) Berbasis Estimator Det-MCD. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003211141-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Ordinary Portland Cement (OPC) merupakan material strategis dalam pembangunan infrastruktur karena daya tahannya yang tinggi dan fleksibilitas penggunaannya. PT XYZ, sebagai produsen semen terbesar di Indonesia, menghadapi tantangan dalam menjaga konsistensi kualitas OPC di tengah tingginya permintaan, kompleksitas proses produksi, dan keberadaan outlier dalam data. Penelitian ini menggunakan metode Maximum Half-Normal Multivariate Control Chart (Max-Half-MChart) yang dirancang untuk mendeteksi pergeseran kecil maupun besar pada data multivariat secara simultan, termasuk saat terdapat outlier. Pada diagram kontrol Max-Half-MChart, outlier dapat menyebabkan fungsi distribusi kumulatif chi-square mendekati satu, sehingga menghasilkan invers distribusi kumulatif normal standar yang sangat besar. Efek masking dan swamping juga dapat muncul apabila data mengandung outlier. Oleh karena itu, penaksir robust diperlukan untuk mengestimasi rata-rata dan matriks kovarians agar analisis tetap valid. Penelitian ini menggunakan pendekatan Deterministic Minimum Covariance Determinant (Det-MCD) guna meningkatkan sensitivitas dalam mendeteksi penyimpangan proses. Analisis dilakukan dalam dua fase, yaitu fase I untuk mempelajari kondisi awal proses, dan fase II untuk mengevaluasi efektivitas metode setelah perbaikan kualitas. Pada fase I, diagram kontrol Max-Half-MChart berbasis Det-MCD mendeteksi 4 pengamatan out of control, yang mengindikasikan adanya pergeseran rata-rata dan variabilitas proses. Sementara pada fase II, tidak ditemukan pengamatan out of control, menunjukkan bahwa proses telah terkendali secara statistik. Dibandingkan dengan Max-Half-MChart, pendekatan Det-MCD terbukti lebih sensitif karena metode konvensional hanya mendeteksi 2 pengamatan. Berdasarkan identifikasi penyebab out of control, karakteristik kualitas C₃A dan LOI merupakan faktor dominan yang memengaruhi ketidakterkendalian proses. Seluruh karakteristik kualitas menunjukkan nilai indeks kapabilitas Pp dibawah 1,00 disertai nilai MPp sebesar 0,44 yang menandakan bahwa proses belum kapabel baik secara univariat maupun multivariat. Sehingga diperlukan pengendalian variasi proses guna meningkatkan kapabilitas serta menjaga kestabilan dan mutu produk secara berkelanjutan.
======================================================================================================================================
Ordinary Portland Cement (OPC) is a strategic material in infrastructure development due to its high durability and flexible application. PT XYZ, as the largest cement producer in Indonesia, faces challenges in maintaining the quality consistency of OPC amid high demand, complex production processes, and the presence of outliers in the data. This study employs the Maximum Half-Normal Multivariate Control Chart (Max-Half-MChart), designed to detect both small and large shifts in multivariate data simultaneously, including when outliers are present. In the Max-Half-MChart, outliers can cause the cumulative chi-square distribution function to approach one, resulting in an extremely large inverse standard normal cumulative distribution. Masking and swamping effects may also occur when the data contain outliers. Therefore, a robust estimator is required to accurately estimate the mean vector and covariance matrix for valid analysis. This study adopts the Deterministic Minimum Covariance Determinant (Det-MCD) approach to enhance sensitivity in detecting process deviations. The analysis is conducted in two phases: Phase I to examine the initial state of the process, and Phase II to evaluate the effectiveness of the method after quality improvements. In Phase I, the Max-Half-MChart with Det-MCD detected four out-of-control observations, indicating shifts in both the mean and process variability. Meanwhile, no out-of-control points were found in Phase II, suggesting that the process had achieved statistical control. Compared to the conventional Max-Half-MChart, the Det-MCD approach proved to be more sensitive, as the standard method detected only two out-of-control observations. Based on the identification of root causes, the quality characteristics C₃A and LOI were found to be the dominant factors contributing to process instability. All quality characteristics exhibited process capability index values (Pp) below 1,00, along with a multivariate capability index (MPp) of 0,44, indicating that the process is not capable, both univariately and multivariately. Therefore, process variation control is necessary to enhance capability and ensure long-term product stability and quality.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Max-Half-Mchart, Det-MCD, Pengendalian Kualitas, Diagram Kontrol, OPC, Max-Half-Mchart, Det-MCD, Quality Control, Control Chart, OPC.
Subjects:	H Social Sciences > HD Industries. Land use. Labor > HD9980.5 Service industries--Quality control. T Technology > TP Chemical technology > TP883 Portland cement.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Garin Sinta Pasadina
Date Deposited:	01 Aug 2025 07:07
Last Modified:	01 Aug 2025 07:07
URI:	http://repository.its.ac.id/id/eprint/125816

Actions (login required)

View Item