Process Control

Physics-Based Modeling

My process control experience is centered on physics-based modeling with a strength for chemical processes. However, in learning the basics of process control, I had many opportunities to develop physics based models for perturbation and control. Pictured left is a simple model for a cruise-control system. This comprehenive model accounts for disturbance variables (DV) such as cargo/passengers, hills, and wind. The sensor (speedometer) measures the process variable (PV, speed) changes as the acutator (gas pedal) rejects deviations caused by DV's.

Chemical Process Control

I worked through several control-oriented problems in my efforts to learn and master process control. Pictured left is a PID controller affecting the cooling of an exothermic reactor. To maximize the production of A, the temperature of the reactor needed to remain constant while not exceeding 400K as to prevent a runaway reaction. To achieve this, I had to improve the tuning by adjusting Kc, τI, and τD by trial and error until the controller displayed a 10% to 15% overshoot in response to a reactor temperature set point step from 300K to 320K.

Arduino Project

Another project I worked on involved a thermocouple, a microcontroller, and a motor. The objective of the project was to design and build a controller for a miniature heat activated 'escape vehicle'. My team of 3 built a shoebox-sized car with an Arduino Leonard controller and a small DC motor that was powered by two 9-Volt batteries. A K-type thermocouple read and relayed the temperature at a point on the car. When the temperature rose above a specified setpoint, the Arduino signaled the motor to drive the car away from the disturbance. The car stopped when the temperature reading dropped below the setpoint.


We ran controlled tests by submerging the thermocouple in hot water to achieve a set initial temperature of 160 degrees Fahrenheit. The Arduino was programmed to power the motor when the thermocouple registered temperatures above 110°F so the car would move away from the heat source while cooling the thermocouple with convective heat transfer. We also created a physics-based model of the system using a simple energy balance around the tip of the thermocouple. When comparing the physics-based model to the collected data, variances showed more than convective heat transfer occurring - even with minimal externalities affecting the system. We tried to use the physics-based model to come up with guess values for our first order plus deadtime model. By taking the energy balance differential equation and linearizing it, we tried to get tuning parameters for a FOPDT model, namely Kp (process gain), τp (process time constant), and θp (process time delay). This failed and stands as an area for improvement for any continuation of this project.


We heated the car past the setpoint by pouring water on the thermocouple, which signaled the motor to actuate and drive our system away from the disturbance. This method did not yield consistent results. We found that the thermocouple had some residual materials stuck at the base. This extra material acted as a buffer to heat transfer and is a primary source of error in our first 7 trials. To get more consistent results, we heated the car up to the same initial temperature before switching on the motor to let the car run till the temperature cooled below the setpoint. We also trimmed away the extra PVC around the thermocouple, allowing for more airflow. The resulting data had less error compared to our model, with a difference of ~700. The reduction in error confirmed that convection was the primary method of cooling the thermocouple. Knowing this allows for easier control, as a different motor could be programmed to vary a response to disturbances instead of a binary response. There are pieces of literature that we could also look into that use a Fuzzy -PI Controller to optimize the control of the system along with having an energy efficient system. These types of controllers have been implemented in systems such as wind generation farms (Ray et al) and speed control in electric vehicles (Khooban et al) both of which have similar design/control concepts as our system.