Pengendalian Konsentrasi Hidrogen Peroksida Pada Proses Dekontaminasi Chamber Isolator Menggunakan Model Predictive Control

Isnen, Maizal (2026) Pengendalian Konsentrasi Hidrogen Peroksida Pada Proses Dekontaminasi Chamber Isolator Menggunakan Model Predictive Control. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6009241001-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Proses dekontaminasi pada isolator farmasi memerlukan pengaturan konsentrasi H₂O₂, kelembaban relatif, temperatur, serta kondisi kondensasi agar proses sterilisasi dapat berlangsung efektif tanpa menghasilkan pengembunan yang berlebih. Pada penelitian ini, model matematik proses dekontaminasi berhasil dirumuskan dalam bentuk sistem persamaan diferensial nonlinier dengan enam variabel keadaan, yaitu konsentrasi uap H₂O₂ (C), konsentrasi H₂O₂ terserap pada permukaan (Cs), kelembaban relatif (RH), temperatur (T), indeks kondensasi (CI), dan paparan kumulatif (L). Model tersebut merepresentasikan dinamika injeksi, adsorpsi–desorpsi, peluruhan, serta kondensasi secara fisis. Parameter proses diidentifikasi dan dikalibrasi dengan mengacu pada data hasil pengukuran sistem aktual dengan beberapa penyesuaian. Nilai parameter operasi dan kendali ditentukan melalui simulasi open-loop yang tervalidasi, dengan nilai NRMSE 3,1% pada kanal konsentrasi H₂O₂, MAPE 16,1%, dan R² 0,98. Sementara itu kanal kelembaban relatif NRMSE 7,6%, MAPE 5,7%, dan R² 0,92. Adapun kanal temperatur NRMSE 7,4%, MAPE 0,9%, dan R² 0,96, sehingga lulus kriteria penerimaan. Uji sensitivitas lokal menyaring tujuh parameter dominan (kv, kdec, UA, Cth, Wdry, kw,inj, dan kdryer) sebagai prioritas kalibrasi. Hasil simulasi menunjukkan bahwa model mampu menjaga keseimbangan termal–higrometrik, diperoleh pada Np = 40 dan Nc = 3, dimana kenaikan temperatur berperan dalam menekan indeks kondensasi sebagai syarat tercapainya dekontaminasi tanpa pengembunan. Pendekatan kontrol prediktif NMPC juga berhasil disimulasikan secara closed-loop dan mampu menjadwalkan fase conditioning, dwell, dan aerasi secara otomatis pada variasi setpoint 600, 700, dan 800 ppm dengan galat tunak kurang dari 1%. Potensi penghematan waktu siklus sebesar 26 hingga 34% lebih cepat, namun estimasi konsumsi hidrogen peroksida model justru 40,8% hingga 75,1% lebih tinggi. Dengan demikian, pendekatan berbasis model ini berpotensi mereduksi ketergantungan terhadap trial and error, meningkatkan kestabilan proses, serta mendukung estimasi optimasi konsumsi hidrogen peroksida dan waktu siklus.
==================================================================================================================================
The decontamination process in pharmaceutical isolators requires the control of H₂O₂ concentration, relative humidity, temperature, and condensation conditions to ensure effective sterilization without excessive condensation. In this study, the mathematical model of the decontamination process was successfully formulated as a system of nonlinear differential equations with six state variables, namely H₂O₂ vapor concentration (C), H₂O₂ concentration absorbed on the surface (Cs), relative humidity (RH), temperature (T), condensation index (CI), and cumulative exposure (L). The model physically represents the dynamics of injection, adsorption–desorption, decay, and condensation. Process parameters were identified and calibrated based on experimental measurements obtained from the actual system with several adjustments. The operating and control parameters were determined through validated open-loop simulations, yielding an NRMSE of 3.1%, MAPE of 16.1%, and an R² of 0.98 for the H₂O₂ concentration channel. For the relative humidity channel, the model achieved an NRMSE of 7.6%, MAPE of 5.7%, and an R² of 0.92. For the temperature channel, the NRMSE was 7.4%, with a MAPE of 0.9% and an R² of 0.96. These results satisfy the acceptance criteria. Local sensitivity testing screened seven dominant parameters (kv, kdec, UA, Cth, Wdry, kw,inj, and kdryer) as calibration priorities. Simulation results demonstrated that the model was capable of maintaining thermo-hygrometric equilibrium, obtained at Np = 40 and Nc = 3, where the increase in temperature contributed to suppressing the condensation index as a prerequisite for achieving decontamination without condensation. The NMPC predictive control approach was also successfully simulated in closed loop and was able to automatically schedule the conditioning, dwell, and aeration phases at setpoint variations of 600, 700, and 800 ppm with steady-state errors of less than 1%. The potential cycle time saving is 26% to 34% faster; however, the model estimated hydrogen peroxide consumption is instead 40.8% to 75.1% higher. Therefore, this model-based approach has the potential to reduce reliance on trial and error, improve process stability, and support the estimation and optimization of hydrogen peroxide consumption and cycle time.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Dekontaminasi, Isolator, Model Predictive Control, Simulink. Decontamination, Isolator, Model Predictive Control, Simulink.
Subjects:	Q Science > QC Physics Q Science > QC Physics > QC151 Fluid dynamics Q Science > QC Physics > QC162 Adsorption and absorption R Medicine > R Medicine (General) > R856.2 Medical instruments and apparatus. R Medicine > RS Pharmacy and materia medica T Technology > TP Chemical technology > TP994 Surface active agents.
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30101-(S2) Master Thesis
Depositing User:	Maizal Isnen
Date Deposited:	06 Aug 2026 09:19
Last Modified:	06 Aug 2026 09:19
URI:	http://repository.its.ac.id/id/eprint/144175

Actions (login required)

View Item