Monitoring and control of droplet size in a micro-channel reactor for biodiesel production

A micro-channel reactor for a biodiesel production of palm oil by transesterification is a technology that has received much attention in recent years due to the high efficiency with short reaction time compared to those of a batch reactor typically used for an hour to obtain a standard biodiesel. An oil droplet formed in a micro channel is a key factor affecting the reactor performance; the smaller the oil droplets are, the larger the contact area between oil and methanol phases obtains. This work presents a development of the technique for monitoring and controlling an average of the oil droplet size in the micro channel by a direct inverse neural network (DINN) controller cooperating with a high-speed image processing in real time. A proposed control system consists of a DINN controller for tracking the desired size of the oil droplet by manipulating an oil feed flow to the reactor and a neural network estimator for compensating process-model mismatch errors. It was tested in the real-time conditions with a prototype microchannel reactor that is a 7.5×7.5 cm, transparent acrylic sandwich frames with 0.5 mm radius of the channel width. The control algorithm coded by LabVIEW is embedded into a field programmable gate array (FPGA) real-time controller is used to interface with the high-speed camera and an HPLC pump for feeding the palm oil. The results of servo performance showed that the proposed controller successfully enforced the oil droplet size to the desired target with a better performance compared to those of the PI controller.