Decentralized control

Controllers for decentralized control and regulation are usually positioned locally to the process and connected to the control room. 

This chapter is organized in the following way. First, the general model of the CSTR process, based on first principles, is derived. A linearized approximate model of the reactor around the equilibrium points is then obtained. The analysis of this model will provide some hints about the appropriate control structures. Decentralized control as well as multivariable (MIMO) control systems can be designed according to the requirements. 

The diagram above shows an interactive MIMO system, where the controlled variables, outlet flow temperature and concentration, both depend on the manipulated variables. In order to design a decentralized control, a pairing of variables should be decided. A look at the state Eq.(23) suggests the assignment of the control of the temperature to the cooling flow and the concentration control to the reactor inlet flow. In this case, the internal variable Tj may be used to implement a cascade control of the reactor temperature. Nevertheless, a detailed study of the elements of the transfer matrix may recommend another option (see, for instance, [1]). 

Based on the linearized models around the equilibrium point, different local controllers can be implemented. In the discussion above a simple proportional controller was assumed (unity feedback and variable gain). To deal with multivariable systems two basic control strategies are considered centralized and decentralized control. In the second case, each manipulated variable is computed based on one controlled variable or a subset of them. The rest of manipulated variables are considered as disturbances and can be used in a feedforward strategy to compensate, at least in steady-state, their effects. For that purpose, it is t3q)ical to use PID controllers. The multi-loop decoupling is not always the best strategy as an extra control effort is required to decouple the loops. 

Decoupling. K is determined to decouple some blocks of process/ manipulated variables. This is usually complemented by a decentralized control or a pole placement strategy. 

The local control of the CSTR by using centralized and decentralized control has been also analyzed. The decentralized control is studied with a cascade control with two PI primary and secondary controllers. From the block diagram, the step response to change in the concentration and temperature references are deduced. A short reference to the decoupling problem is also discussed. 

A. Marchetti, M. Amrhein, B. Chachuat, and D. Bonvin, Scale-up of batch processes via decentralized control, ADCHEMIFAC, 2006, p. 221. 

Many of the aforementioned heuristic decentralized control synthesis approaches rely on engineering judgement rather than rigorous analysis. On the other hand, the implementation of advanced, model-based, control strategies for process systems is hindered by the often overwhelming size and complexity of their dynamic models. The results cited above indicate that the design of fully centralized controllers on the basis of entire process models is impractical, such... 

Ricker, N. L. (1996). Decentralized control of the Tennessee Eastman challenge process. J. Proc. Contr., 6, 205. 

Much progress has been made on the design of decentralized controllers [8]. A methodology has been developed to translate overall performance and robustness specifications into specifications on individual loops. Designing the individual loops according to these specifications guarantees the satisfactory performance of the overall system. 

FIGURE 15.70 Block diagram of a 2 x 2 process with single-loop controllers applied (decentralized control). Note that C represents a controller and P represents an inpnt/ontpnt fnnction. 

The recommended tuning procedure for a single PID loop can be extended to tuning the singleloop PID controllers applied for decentralized control of a MIMO process. The first step in tuning... 

RGA provides a measure of interactions caused by decentralized control. Let assume that the input My controls the output y,. Let the corresponding gain be g,j when the other loops are open, and when the other loops are closed, so that the other outputs are constant, except for the considered loop. By definition the element Ay of the RGA is given by ... 

A multivariable controller computes the value of each manipulated input based on the value of all controlled outputs. The design of multivariable controllers is not an easy task. Moreover, multivariable controllers requite accurate plant models. Therefore, most of process systems make use of a simpler structure, where the value of one manipulated is computed based on the value of one controlled output. This is called decentralized control, or multi-SISO. The decentralized controller has the following diagonal structure ... 

The design of a decentralized control systems involves two steps ... 

Let G = diag gi-) be the matrix containing the diagonal elements of G. The decentralized controller is diagonal K s) = diagik,) and each loop has the transfer function L, = g,A . 

In other words, gives the apparent disturbance gain as seen from loop / when the system is controlled using decentralized control. Thus, the condition to avoid input constraint follows directly ... 

The choice of the controlled and manipulated variables can be based on engineering judgement, or on methods that are more systematic. In any case we recommend those that regard in the first place the material balance. This gives valuable insight into the number of control degrees of freedom, functional controllability, I/O pairing for decentralized control. 

In the last chapter we developed some mathematical tools and some methods of analyzing multivariable closedloop systems. This chapter studies the development of control structures for these processes. Because of their widespread use in real industrial applications, conventional diagonal control structures are discussed. These systems, which are also called decentralized control, consist of multiloop SISO controllers with one controlled variable paired with one manipulated variable. The major idea in this chapter is that these SISO controllers should be tuned simultaneously, with the interactions in the process taken into account. 

Song et al (2006) proposed a multivariable purity control scheme using the m-parameters as manipulated variables and a model predictive control scheme based on linear models that are identified from nonlinear simulations. The approach proposed by Schramm, Griiner, and Kienle (2003) for purity control has been modified by several authors (Kleinert and Lunze, 2008 Fiitterer, 2008). It gives rise to relatively simple, decentralized controllers for the front positions, but an additional purity control layer is needed to cope with plant-model mismatch and sensor errors. Vilas and Van de Wouwer (2011) augmented it by an MPG controller based on a POD (proper orthogonal collocation) model of the plant for parameter tuning of the local PI controllers to cope with the process nonlinearity. 

Perea-Lopez E., Grossmaim I.E. and Ydstie B.E. 2000. Dynamic modeling and decentralized control of supply chains, Ind. Eng. Chem. Res., 40(15), 3369-3383. 

KANBAN AS A DECENTRALIZED CONTROL SYSTEM FOR JUST-IN-TIME... 

In considering kanban as a decentralized control system, the following control parameters are necessary number of kanbans in circulation number of units in the kanban standard container and kanban delivery cycle a-b-c (where b is number of deliveries per a days and c indicates the delivery delay factor as an indication of replenishment lead time). For example, 1-4-2 means that every 1 day the containers are delivered 4 times and that a new production order would be deUveted by the 2nd subsequent delivery (in this case, about a half-day later, given four deliveries per day). 

Open HMI with access to decentralized control components... 

---