Ester mempunyai pelbagai aplikasi dalam industri makanan, kosmetik dan farmaseutikal kerana ciri-ciri perisa dan wangian mereka. Proses pengesteran kelompok bermangkin lipase telah muncul sebagai teknik yang berkesan untuk penghasilan ester . Suhu telah dikenalpasti sebagai satu parameter penting yang memberi kesan kepada produktiviti ester kerana saling tindak di antara suhu dan pemangkin aktiviti kinetik mewujudkan variasi dalam kadar tindak balas. Oleh itu, pengawalan keadaan operasi reaktor pengesteran kelompok adalah sangat penting bagi meningkatkan penghasilan ester.
Secara umum, pengesteran kelompok adalah satu proses yang rumit kerana ia melibatkan kelakuan dinamik yang tak lelurus. Ini adalah disebabkan faktor pencirian kinetik dan proses yang tidak stabil. Pengawal PID konvensional tidak dapat mencapai prestasi kawalan yang memuaskan untuk dinamik sistem sebegini. Sehubungan dengan itu, strategi kawalan berdasarkan model telah diperkenalkan. Strategi kawalan ini memerlukan model yang tepat dan kerumitan yang rendah. Bagi mengatasi kesukaran dalam membina model prinsip pertama untuk proses kelompok, model masukan-keluaran empirik yang lebih mudah, dibina berdasarkan data eksperimen dalam kajian ini. Untuk mengawal suhu reaktor, model masukan-keluaran untuk sistem penyejukan proses pengesteran kelompok telah dikenalpasti kerana tindak balas pengesteran adalah eksotermik dan suhu tindak balas tidak boleh melebihi suhu maksimum 47℃, kerana kepekaan enzim terhadap suhu. Berdasarkan kajian literatur, suhu reaktor telah dipilih sebagai pembolehubah keluaran manakala kadar aliran air di dalam jaket dan suhu air di dalam jacket sebagai pembolehubah masukan. Model lelurus dan model tak lelurus telah dibina untuk memilih model yang terbaik bagi mewakili sistem penyejukan kerana ketaklelurusan wujud di antara pemindahan haba daripada reaktor ke jaket dan kadar aliran medium penyejukan dalam jaket. Teknik sistem pengenalpastian telah digunakan dengan bantuan kotak perkakasan bagi sistem pengenalpastian dalam MATLAB untuk pengenalan dan pengesahan model. Empat model lelurus, iaitu model keadaan-ruang, model ARX, model ARMAX dan model BJ dan dua model tak lelurus, iaitu model NARX dan model Hammerstein-Wiener telah dianggarkan dan nilai peratusan padan terbaik di antara semua model telah dibandingkan. Keputusan selaku menunjukkan bahawa model BJ lelurus dapat menghasilkan anggaran dan pengesahan yang baik. Setelah pengesahan, model BJ mempunyai nilai peratusan padan sebanyak 90.26% dan pada masa yang sama lulus analisis sisa.
_______________________________________________________________________________________________________
Esters have wide range of applications in food, cosmetic and pharmaceutical industries due to their characteristic flavour and fragrance. Lipase-catalysed batch esterification process has emerged as an effective technique for the production of esters. It is proven that temperature is an important parameter that affect the productivity of esters as the interaction between temperature and catalyst kinetic activity creates a variation in reaction rate. Hence, controlling the operating condition of batch esterification reactor is very important in order to improve the production of esters. In general, batch esterification is a complex process since the system involves a nonlinear dynamic behaviour due to kinetic characteristics and the unsteady state process. Conventional PID controller cannot achieve satisfactory control performance for such system dynamics. Therefore, a model-based control strategy has been introduced. This type of control strategy requires a model that is accurate and low complexity. To overcome the difficulty in developing first principle model for batch processes, empirical input-output model, which is easier to develop based on experimental data is studied in this work. To control the temperature of reactor, the input-output model is identified on the cooling system of batch esterification process as esterification reaction is exothermic and the reaction temperature should not exceed the maximum temperature of 47℃, due to temperature sensitivity of the enzymes. Based on literature, reactor temperature was selected as the output variable while jacket flow rate and jacket temperature as the input variables. As nonlinearity exists between the heat transferred from the reactor to the jacket and the flow rate of the cooling medium in the jacket, both linear and nonlinear models were developed to determine the best possible model to represent the cooling system. System identification technique have been applied with the aid of MATLAB System Identification Toolbox for the identification and validation of models. Four linear models, i.e. state-space model, ARX model, ARMAX model and BJ model and two nonlinear models, i.e. NARX model and Hammerstein-Wiener model have been estimated and compared in terms of their best fit percentage. Simulated results have shown that linear BJ model is able to provide good estimation and validation. Upon validation, it is capable to reproduce new measured data with best fit of 90.26% and at the same time not falsified by residual analysis.
Identification of input-output models for cooling system in batch esterification reactor / Ng Sin Yee