Introduction

According to some random chemical engineering professor in London, bioethanol is the single most important product for the future of the chemical industry. Apart from its large-scale use as a biofuel, bioethanol has the potential to unlock a fully renewable chemical industry, through production of biobased ethylene, one of the two main feedstocks for current organic chemicals and materials. It also kills COVID viruses, so, you know, good stuff. Bioethanol is typically produced via fermentation with the yeast Saccharomyces Cerevisiae, which has been used for this purpose by humans since the dawn of civilization.

You have been put in charge of the Design of Renewables Using Novel Kinetics (DRUNK) project which will deliver a control strategy for a continuous bioreactor to produce ethanol from lignocellulosic glucose. The reactor has very complex kinetics and temperature is controlled via a cooling jacket, as the cells will die or underperform if the temperature is not maintained within design range. Since production has already been delayed two years because of some “global pandemic”, the bioreactor has already been installed and you have only two weeks to design and test a control strategy before production commences. Simulink (MATLAB version R2021b) must be used to carry out the calculations and simulations, and your predecessor team already developed a Simulink module to simulate the bioreactor, but they were sacked because their control strategy included a tap, a long transfer line and a huge jug in their office.

Project Information

The yeast fermenter has been developed as a bioreactor module (S-function) in Simulink. This mathematical model is a first-principles model of a continuous fermentation bioreactor for the production of ethanol from renewable feedstocks (e.g., lignocellulosic biomass). It accounts for various nonlinear characteristics of the process, including complex reaction kinetics with product inhibition, a detailed energy balance for the bioreactor and cooling jacket, the oxygen mass transfer within the fermentation broth, the effect of oxygen availability on by-product (i.e., glycerol) formation and the effect of temperature dependence on the kinetic parameters,

oxygen mass transfer, and by-product formation.

The components tracked in this bioprocess include:

•    The total cellular biomass ( )

•    The viable cellular biomass ( )

•    Glucose () as the carbon substrate

•    Ethanol () as the product of interest

•    Glycerol () as the by-product

•    Oxygen (O)

•    Feed Temperature

•    Bioreactor Temperature

•    Inlet coolant and jacket temperatures

Ethanol is produced during the exponential phase of the microbial culture (growth-associated production), but also a significant amount of ethanol is produced after the cells have reached their stationary phase (non-growth-associated production) (Aiba et al., 1968). Once ethanol production is decoupled from growth, lower temperatures favour product synthesis since the protein unfolding and enzyme deactivation effects are less profound (Aiba et al., 1968).

Like ethanol, glycerol  is  produced during the exponential  phase of the  microbial culture (growth-associated production), as well as after the stationary phase (non-growth-associated production). Glycerol is the main by-product of the process, which can account for up to 5% of the carbon source in industrial processes and is produced by the cell as a protective measure against the osmotic stress induced by the reduced water activity in the fermentation broth as a  result of increased ethanol concentration  in the  broth  (Alfenore et al., 2004; Hallsworth, 1998; Jones & Greenfield, 1986). Glycerol and polyols contribute to the thermal protection of proteins against denaturation and cell death (Amillastre et al., 2012; Back et al., 1979). Oxygen availability  has been shown to affect glycerol formation, where improved aeration strategies lead to reduced glycerol production (Alfenore et al., 2004; Costenoble et al.,  2000).  Two  dependencies  are  considered  for  the  calculation  of  the  equilibrium concentration of oxygen, namely its dependence on the temperature inside the bioreactor and on the oxygen that is supplied by an air inflow stream. Both dependencies are captured by Henry's law

Glucose is assumed to be consumed for three distinct activities, namely (i) cellular biomass growth, (ii) ethanol production, and (iii) glycerol formation. Oxygen consumption takes place in the fermentation broth for cellular biomass growth and is assumed to follow saturation kinetics.  Elevated temperatures have been shown to induce cell death  in yeast cultures (Amillastre et al., 2012).

The reactor is equipped with thermocouples, flow meters and a super-fancy in-line HPLC to monitor the inlet and outlet conditions (temperature, flowrate and composition). It has also a mass spectrometer at the outlet to measure the composition of the outlet. This leads to a lot of variables, including:

•    Dilution rate

•    Inlet glucose concentration

•    Coolant flowrate

•   Air flowrate

•    Feed temperature

•    Coolant inlet temperature

Control Objectives and Problem Statement

You are tasked with controlling the bioethanol production process’s two key variables:          ethanol concentration (to keep distillation costs low) and bioreactor temperature (to keep the little yeast cells happy). You will do this by manipulating the coolant temperature and            glucose concentration in the feed (your two manipulated variables). Unfortunately, the          control panel was originally developed at UCL. To avoid putting the plant at risk, you might   need to modify it in order to make an efficient and robust control system that will keep the     plant running smoothly.

The main control objective is to maintain a target steady ethanol production of 40 g/l. A       secondary objective is to control the temperature of the bio reactor within the range of 30 to 31oC.

In order to achieve this, you can implement an automated control scheme on the glucose     concentration in the stream and the cooling agent temperature. The operation ranges of       these variables are 20-300 g/L and 10-25oC. However, no control can be exerted on the       other input variables as these are determined by upstream processes and broken valves.     Due to this, you can expect disturbances in the input dilution rate and feed temperature (see below). The nominal inlet conditions are

Development of a Distributed Control System in Simulink

The user interface of the bioethanol production process in Simulink is shown below Error!    Reference source not found.(refer to the “Simulink FAQ” provided to you on Blackboard to familiarize yourself with Simulink). The Simulink block comprises 6 inputs (corresponding to  dilution rate, inlet feed glucose concentration, coolant flowrate, air flowrate, feed temperature and coolant inlet temperature) and 9 outputs (total biomass, active biomass, glucose             concentration, ethanol concentration, glycerol concentration, oxygen concentration (liq),        bioreactor temperature, cooling jacket temperature).

Step-by-Step Controller Implementation

The following parts A-F describe a step-by-step procedure for the development of control for the bioethanol production process.

A.  Output signal

Make sure that from the output signals given you can monitor the 2 required variables: bioethanol concentration and bioreactor temperature. Modify as necessary.

B.  Process Identification (Transfer Function Model Identification)

It is useful to identify simple transfer function models in order to devise an efficient control scheme. Towards this goal, evaluate the dynamic response of the DRUNK process, changes in  bioethanol  concentration  and  reactor  temperature  to  changes  in  the  inlet  glucose concentration and changes in the cooling agent temperature. In doing so, pay special attention to the step changes imposed and keep in mind that the process is nonlinear (dependent of the value about which step changes are imposed) and valves can saturate.

C.  Control Structure Selection (Loop Pairing)

Using a Relative Gain Array analysis and other appropriate measures, compare the possible control structures. Select the best structure and explain your choice. It may be useful to consider combinations of the manipulated variables also.

D.   Outlet Temperature Control Loop

Determine an appropriate Master feedback controller for the bioreactor temperature. In particular:

•    Choose an appropriate feedback controller type (try P, PI and PID)

•    Find the critical frequency and maximum gain for a feedback proportional controller using the Ziegler-Nichols tuning method

•    Tune  the  controller  using  the  on-line  Ziegler-Nichols  tuning  method  as  a  first approximation, and then try to improve upon it directly in the plant

•    Test the performance of the controller for a variety of disturbances and set-point changes

•    Observe how your controller affects the bioethanol concentration.

E.   Outlet Composition Control Loop

•    Determine an appropriate Master feedback controller for the bioethanol concentration. In particular, with the temperature control loop open (that is, disconnected):

•    Choose an appropriate feedback controller type (P, PI or PID) and select appropriate controller constants employing the Cohen-Coon tuning rules and then try to improve upon it

•    Test the performance of the controller for a variety of disturbances and set-point changes

•    Observe how your controller affects the outlet temperature

F.   Multi-Loop Control (Loop Interaction; Detuning & Decoupling)

•    The two control loops designed in parts D and E above will interact with each other to some extent. Study the behaviour of the system when both loops are closed for a variety of disturbances and set-point changes

•    Investigate the effectiveness of controller detuning, using simple empirical detuning laws.  If  deemed  necessary,  design  appropriate  decoupling  controllers  to  reduce interaction  between  the  loops.  You  may/may  not  need  to  introduce  one-way decoupling/decoupling in both ways.

Evaluation of Controller Performance

You are to email (ONLY TOic.rdcp@gmail.com– please do not cc the module leader) your best control system (Simulink .slx file) no later than 16:30 on Thursday 17th  February (ONE email per group). On the morning of Friday 18th  February, a sequence of disturbances will become available on Blackboard. You are to evaluate the performance of your submitted control system for this sequence and discuss the results in an addendum to your project report due on Monday 21st  February at 16:30.

IV. Logistics

1.   Duration and assistance. The project is available now. The project will run for two weeks (until 17 February). There will be GTA assistance available both in person and via  MS Teams on  Mondays, Tuesdays, Thursdays  and  Fridays from  0900- 1200. Please wait in the main MS Teams channel for a GTA to take you to a breakout room (for group privacy/commotion reduction). Your course coordinator will be available via Blackboard, or you can make an appointment via email to discuss any group issues.

2.   Groups. Your boss has assigned you to a group of (normally) 4 engineers for the duration of the project. Because of your boss’s abhorrence of paperwork, there will be no switching permitted.

3.  Assessments. The project will be assessed in the following ways:

a.   Controller report (1 per group!) – 80%

b.   Controller test performance (1 per group!) – 10%

c.   Controller addendum/postmortem on the test 10%

4.   Notes

a.  You MUST use MATLAB version R2021b for your design and simulations

5.   Deliverables

a.   Submit your best controller by 17/02/21 at 1630

b.   Submit your report by 17/02/21 at 1630

c.   Submit your Controller Performance Addendum by 21/02/21 at 1630

Marking scheme for the report:

•    Presentation: overall style, clarity and organisation 30%

•    Description of theory 30%

•    Discussion of the control strategy and simulations 40%

Report Guidelines

The following items are to be submitted:

(a) Report

The submission of one report by each group of students will be required. Please use the normal  departmental  template  for  report  writing.  A  penalty  will  be  imposed  on  reports exceeding the page limit. Arial 11-point font (minimum) and 2 cm margins (minimum).

Each report should not exceed 10 pages, including main body text, but excluding the title page, author names, abstract and references.

Do not try to include all of the results you generate but do include meaningful analyses to go with the figures or tables. Adopt compact and efficient ways to present the results. A compare- and-contrast approach may be helpful with overlaid plots from different simulations shown on the same graph with appropriate legends.  For example,  present results from single-loop controller   performance,   performance   with   both   controllers,   and   performance   with decoupled/detuned configurations on the same graph. You may also consider combining results from P, PI or PID controllers and present them on the same plots. This presentation will be both efficient and easy to read.

The following is a suggested structure for the report and the topics for inclusion under the major headings. There is also a suggestion of an appropriate page limit for the different headings.

Table 1. Report structure.

Overall page limit – 10

(i) Abstract and Keywords: The abstract must summarize the content of the report and the main conclusions. This is normally a short paragraph.

(ii) Introduction: The bioethanol process description; general aims and objectives of the reactor and controller design and specific aims for this bioethanol process.

(iii) Theory and Methodology: Only the theory and methods used in the work should be presented (the report is not a tutorial). It can cover the chosen reactor design elements and control strategy and any required details on mass balances, reactor modifications, model identification, feedback control, cascade control and decouplers, etc., with salient equations and  block  diagrams  as  appropriate;  methodology  to  implement  these  strategies  or  any experiments.

(iv) Results and analysis: Results and analysis (not all cases, but the ones that you judge as the  most  important);  sample  calculations;  clear  statements  of  parameters  used  and simulations performed. The use of multiple plots on one set of axes is useful for showing comparisons.

(v) Discussion: The discussions must give insights. It is not enough to just describe the results. One way is to make tables or graphs showing trends and to provide an explanation for the observed trends.                                                                                                                         (vi) Conclusions and suggestions for future work to improve your design.

Controller Test Report addendum. An addendum (max 2 sides A4, Arial 11pt font, 2cm

margins) must be submitted by 16:30 on Monday 21 February. Discussions on the results of the  performance  evaluation  should  be  included  (explanations  of the  observations, what worked as expected, what went wrong and why).

Report Marking

Reports are due to be submitted electronically (Blackboard) by  16:30 on  Thursday 17th February.  Late  submissions  will  be  penalized.  It  is  wise  to  allow  plenty  of time  when attempting submission at peak times in case the electronic submission system becomes overloaded.

Marks will be awarded for quality of the results and presentation of the report. The assessors will look for: Good English and proper grammar including appropriate punctuation marks; organisation of the report in a coherent and logical manner with sections and sub-sections as necessary  under the  major headings;  inclusion of proper citations and references; good formatting  including figures with  legible  lines, appropriate  background,  legend  positioned