M14.7 Suggested Case Study Schedule
It is suggested that these case studies be assigned during the last 4–5 weeks of a typical one-semester course. The suggested time frame is realistic enough to allow for the study of other modules and for short homework assignments during this period.
1. Literature review and motivation for problem | 1 week | 2. Model development (step tests) | 2 weeks | 3. SISO controller design | 3 weeks | 4. MVSISO controller design | 4 weeks | 5. Oral presentation | 4–5 weeks | 6. Written report | 4–5 weeks |
Each problem (unit operation) will have an "advisor" who will answer questions that you have regarding the problem and possible solutions techniques. After you turn in part 1, you will be assigned an advisor. You should probably plan to meet with the advisor once per week, during office hours, during the project period.
An example partial solution to a biomedical engineering control problem (drug infusion control) is presented on the web page.
Literature review and motivation for problem (due after 1 week) Write a short memo (two pages or fewer), giving the following information: The example that you have decided to study. The importance of the unit operation to the particular industry. What are typical sizes or production rates? A literature search with references related to control of the unit operation. A web-based search (using Google or some other search engine) is unacceptable. We are interested in sources published in books, journals, magazines, or conference proceedings. Appropriate databases include Current Contents and the EI Compendex.
The SIMULINK files for each problem are available on the textbook web page. You will apply step inputs to identify the process transfer functions for the next phase of the project. Model development (due after 2 weeks) Write a short memo (one page or under) summarizing your modeling results. Attach the following information to the memo: SISO controller design (due after 3 weeks) Write a short report summarizing the control system design for each independent SISO controller. Please include the following: A description of the design procedure (IMC-based PID, Ziegler-Nichols, direct synthesis, etc.) and the tuning parameters used. Closed-loop responses for each independent loop. It is appropriate to demonstrate the effect of different values for the tuning parameters. How does a setpoint change in a controlled loop affect the uncontrolled loop?
MVSISO controller design (due after 4 weeks)
The RGA analysis:
Based on the RGA, suggest how the variables should be paired in a MVSISO control structure (i.e., y1 paired with u1 and y2 paired with u2). The suggested PI or PID tuning parameters that you would use if the loops did not interact: For example, what tuning parameters would you recommend for loop 1 if loop 2 was open; you may use IMC-based PI or PID, Cohen-Coon, Ziegler-Nichols, Direct Synthesis, or frequency-response-based tuning (satisfying certain gain and phase margins).
A couple of plots:
One for the setpoint response of loop 1, with loop 2 open; another for the setpoint response of loop 2, with loop 1 open. Do you suspect that the control loops must be detuned if all loops are closed? Use the RGA to assist you with this. Are there dynamic reasons that you might want not pair outputs and inputs based on the RGA? What are the practical limits to the magnitude of setpoint changes?
Oral presentation (performed after 4–5 weeks) Prepare a short (fewer than 15 minutes) oral presentation with overhead transparencies. Each group member should make an equal contribution. The problem advisor and at least one other consultant will observe the presentation and ask questions. Final Report (due after 4–5 weeks) Write a short final report including the following: A title page, giving a list of the group members. Include a one paragraph abstract that summarizes the results presented in the report. A short background section describing the problem. Include literature references. Summarize the model developed in the preliminary reports. A short section with the preliminary pairing selection based on RGA analysis. A single-loop results section giving the setpoint responses of each control loop separately (i.e., setpoint response for loop 1 assuming loop 2 is open). A MVSISO results section for the setpoint responses of each control loop. In this, you will show the results of a setpoint change in loop 1 with loop 2 closed. Did you need to detune either loop? Disturbance rejection results. If there are no disturbance transfer functions, assume disturbances to the inputs. A recommendations section. Do you feel that MVSISO control is sufficient? Do you feel that a multivariable technique such as decoupling should be used? Do you feel that the system needs to be studied further using more advanced techniques? The previous reports (with the advisor comments) placed in the appendix of the report.
|
