Adding Feedback

The only serious failing of the feedforward technique is its dependency on accuracy. To provide perfect control, a system must model the plant exactly otherwise whatever error may exist in positioning the manipulated variable causes offset. Errors may arise from several sources  

Failure of the computing system to adequately represent the characteristics of the plant 

Exclusion of significant load components from the feedforward system 

The first and second items alone limit the accuracy of practical systems to the vicinity of 1 to 2 percent.  

Some processes, such as the heat exchanger described earlier, are easy to model. But this is not always the case, particularly when mass and heat transfer coefficients must be used over a wide range of operating conditions. Therefore item 3 may be of considerable importance in the more complicated processes. 

---

Orban et al. (1982-2) discovered that in a CSTR within an extremely narrow range of flow rates and input concentrations a system containing Br03, Br" and Mn(II) or Ce(III) exhibits oscillations in the potential of either a Pt redox or Br- selective electrode. Existence of oscillations was predicted by the model calculations of Bar-Eli [Bar-Eli in Vidal and Pacault (1981) 228-239]. The bromate oscillators such as the B-Z reaction were derived from this fundamental system by adding feedback species which enlarges the region of critical space in which oscillations occur. 

Now redo the linearization adding feedback control. For a value of K = 1, is the system locally stable What is the minimum value of K required for local stability Determine if the nonlinear system is actually stable for this value of K by developing a phase-plane plot using various starting conditions. 

Another important consideration when adding feedback is the location of the dynamic compensator. Although lead-lag can be beneficial to the response of a feedback loop, it interferes with manual operation. When an operator changes the output of the controller manually, he likes to see that action reproduced exactly by the manipulated variable. With a lag or lead-lag between the controller and the manipulated variable, several minutes-possibly even an hour-could elapse before the effect of the adjustment is complete. Therefore it is mandatory to arrange the system so that dynamic compensation is out of the feedback loop. 

Nearly four biUion L/yr of ethanol are added to gasoline and sold as gasohol (18). The starting or driveabiUty difficulties have been solved, in part, by the advances in vehicle technology employing fuel feedback controls. 

Another method of identifying the controls needed is to describe the result-producing processes in flow diagram format. This will enable you to identify where the verification stages need to be added and the feedback loops inserted ... 

The bold lines demonstrate the short-term system. It is obvious that no feedback loops exist between antiviral intervention and other elements of the system. Therefore, the total COl can be calculated by just adding the independent cost components. 

You may notice that nothing that we have covered so far does integral control as in a PID controller. To implement integral action, we need to add one state variable as in Fig. 9.2. Here, we integrate the error [r(t) -, (t) to generate the new variable xn+1. This quantity is multiplied by the additional feedback gain Kn+1 before being added to the rest of the feedback data. 

The neutralizer in the previous example might be controlled differently if the main fluctuation in the load occurs in one or two of the streams. Instead of combining all the streams together before they enter the neutralizer, those streams that vary widely might enter an additional holding tank, where they would be neutralized using traditional feedback control. They would then be added to the main neutralizer, which also has a feedback controller. Which system is best can be determined by running an economic analysis (see Chapters 10 and 11). 

As described by Brogan ( ) the addition of state variable feedback to the system of Figure 1 results in the control scheme shown in Figure 5. The matrix K has been added. This redefines the input vector as... 

The dissemination of well-designed educational material constitutes a complement to other strategies, especially as a basis for training measures (face-to-face interviews or feedback systems), its relatively low added cost being a factor to take into consideration. However, disseminating this type of material alone hardly has any effect on prescriber behaviour. 

The feedback signal is a function of the output signal. It is sent to the summing point and algebraically added to the reference input signal to obtain the actuating signal. 

Fig. 3. 2D evolution for the open system described in Fig. 2. The temporal development of total galactic mass (a) and 2D thin disk abundance (by mass) (b) for a set of open stochastic accretion models with primordial gas accreted. From the top to the bottom the curves refer to 44%, 10%, 5%, 1% and no mass added. As shown, the early time development of deuterium is dominated by astration, while the later history (post-star forming peak) is controlled by the infall rate and feedback to the maintenance of the star formation.

Production of the API begins with the selection of a synthetic route, as determined in the development program. Raw materials are added into a reaction vessel. These raw materials as reactants are heated or cooled in the reaction vessel (normal range is from -15 to 140 °C purpose-built vessels are needed for extreme reactions that require lower or higher temperature controls or pressurization of reaction processes). The chemical synthesis reactions are monitored and controlled via sensor probes (pH, temperature, and pressure) with in-process feedback controls for adjustments and alarms when necessary. Samples are withdrawn at dehned intervals for analysis to determine the reaction progress. Catalysts, including enzymes, may be added to speed up and direct the reaction along a certain pathway. 

---