Monday, April 27, 2015


Why out of so many Boiler controls, steam Temperature control is so critical???

Functional aspects of superheat temperature control types in utility boiler. by: Swapan Basu*

*Author:  Power plant Instrumentation and control handbook Elsevier ( )

*Systems & Controls (Consulting Engineers I&C) Kolkata India. (

 With so much pain, we had to face the natural calamity major earthquake in Nepal. Lets pledge that we are by their side and arrange to do something /to donate for the noble cause and show our solidarity with people of Nepal, and parts of India who had suffered a lot for  the major earthquake total toll over 5000 lives!!!!!!!!!!! 

Now let us go back to the discussions. Earlier we had discussed PID controllers and DEB controls. This time our discussions will be on Fuzzy controller  and neural network.

Functional aspects of superheat temperature control types in utility boiler.
Swapan Basu*
*Author Power plant Instrumentation and conrol handbook –Elsevie ( )
*Systems & Controls (Consulting Engineers I&C) (

Complexity, influence of various inter active parameters, non linear load dependent process response coupled with time delay make superheat temperature control in a thermal power plant the most challenging job. Philosophy behind controlling superheat temperature is to maintain correct balance between operational efficiency and life expectancy of Boiler and turbo-generator (BTG). There have been many approaches such as conventional PID with gain scheduling & feed forward signal, Direct Energy Balance, State Variable controller, Predictive controller based on system model, Fuzzy logic controller, Adaptive fuzzy  neural controller etc. In this paper functional aspects of various superheat temperature controller types have been discussed briefly to get some idea to compare various approaches and select the one most suitable for particular application.

Key word: Gain scheduling, State variable, Adaptive, Predictive, Fuzzy control, Neural network.

After Last weeks' post:

7.0    Fuzzy control approach:

Accurate mathematical model may be not only little costly, but,  at times it may not be practicable also.  As it is not always possible to understand the system correctly—especially when process operates over wide range of conditions, with disturbances, it is extremely difficult to accurately develop mathematical model. Also, state variable approach has its limitation due to availability of all states and associated measurements, so, Fuzzy approach may offer a better solution. Unlike Boolean set of 1 or 0, , fuzzy logic represents a continuous spectrum with the help of fuzzy sets. In line human conclusion and decision, fuzzy logic (developed by Zadeh) utilizes, inexact information in non mathematical approach[9]. In fuzzy approach a set is defined in terms of its membership. Fuzzy controller keeps fuzzy rules in its knowledge base and apply the same on the process inputs to give output based on fuzzy reasoning process. Let there be a typical fuzzy set “A” with various elements (say a,b, c, d)  having membership functions 0.2,0.3, 0.6 & 1. Naturally in the set element with membership function 1 is a full member and others are partial members.  To limit the discussions, lets take the error and rate of error and in line with fuzzy logic  put them in linguistic  classifications viz. Large negative(LN) to Large Positive (LP) in seven classes with Zero as middle. Referring  to fig no. 7A, an error 40%  will have   0.4 SP & 0.6MP & Rate of change of error 15% will have 0.4 ZE and 0.6 SP  as shown in fig 7a[10].  Clearly control quality depends on overlapping of subsets and linguistic functions. For understanding of fuzzy approach and fuzzy expert fig.  7 may be referred to. Fuzzy control takes heuristic  approach e.g. if temperature error is low (error SN) and rate of change of error is medium (MN), then decrease cooling water slowly. These small medium are imprecise magnitude of temperature  and decrease little is coarse corrective action which is derived from its knowledge base. However on account of its closeness due to human thinking it  has wide acceptance. The control will be better understood with coarse fine control.

As discussed in cl. 6.0 normally lrage plants like in  SC & USC there will be two stages of control, where primary spray (larger quantity)  can be used as coarse and secondary spray  as fine control. Fig 8A shows the processes 1&2. Process 1 i.e. primary control is regulated in conventional control as cascade control where set point is derived from MS flow as shown in fig 8B. Similarly  Secodary spray control has been shown in fig 8C where MS temperature set point can be a fixed set point or can be derived from MS flow (especially during low load condition). These two controls with fuzzy has been shown in fig 8D&E.

In conventional cascade controls both are controlled with respect to Temeprature difference and both  would be active almost all the time on account of the process delay. So both are coupled and may cause interfence and stable control is very difficult to achieve especaily during load changes. Fuzzy expert systems developed based on human experience can be used to improve the peroformance with decoupling effect and intelligently incorporating gain scheduling[11]. Primary controlled system can be decoupled from the secondary control (say) as per following Fuzzy rule for example
·      If valve position Vs is withiin say 30-70% Valve position of the primary valve Vp will not be changed. The normal set point of Vp may be 30%. The spray during this period will be effected by Secondary pray.
·      For valve position Vs < 30%, Vp may be closed and for Vs >70%, Vp may be opened wider as per tempearature difference control as shown. So decoupling rule[11] (Ref: figure 10D) may be:
o  When Vs is withiin say 30-70%, Vp will be in original position
o  When Vs < 30 or > 70%,following rules apply:
§ When T2 >=  high temperature set value  (Say 480°C) Cascade control shall be effective for superheater safety.
§ When < High temp set value(Say 480°C), Valve position control shall be initiated for decoupling.
In secondary spray control gain scheduling is incorporated utilizing fuzzy infernce logic as shown in figure 8E  Standard triangular membership function with linear rule would be applied to two inputs viz. Error and rate of change of error as shown in fig 9 shall be used.  As shown in fig 8E fuzzy control system output is added to the conventional cascade controller output. The beauty of the gain scheduling is that whenever the difference will be more, gain  from fuzzy controller will be larger so that final output is predominated by the fuzzy output. On the other hand when gain will be less then gain from fuzzy controller will be less and cascade loop will be more effective. As indicated earlier there will be seven member ship function (LN to LP) for each of error e and Δe.  Finally gain scheduling is supervised based on final MS temperature. Fuzzy systems also have a few limitations.  Fuzzy systems does not have any common frame work to deal with different problem.  Also Human expert play an important role in developing a fuzzy controller/expert system[9].

8.0    Artificial Neural Network & Fuzzy approach:
In intelligent control system application use of Neural network and Fuzzy control cannot be overestimated. In the following control application limitation of Artificial Neural Network(ANN) falling to local limit and drawback of Fuzzy controls of requirement of experienced operator can be somewhat be overcome. Neuro-fuzzy systems which use ANNs to determine their properties (fuzzy set and fuzzy rules by processing the data samples[12]). In this application Adaptive neuro fuzzy inference system(ANFIS) membership functions are extracted from the data set which describe the system behaviour. ANFIS learns from the data set and adjusts as per error created. As ANFIS uses neural  network to implement Fuzzy rule, it improvise the system with only fuzzy rule.

Setting up initial small step of training with zero goal error after a few nos. of iterations it is possible to generate fuzzy inference system automatically. A small comparisons amongst Neural network control (NNC) Fuzzy control (FC) and Adaptive Fuzzy Neural Network AFNNC) have been presented below[12]:
Table 2 Performance Comparison of various systems[12]
Control method
 75% load
100% load
Delay time(S)
Adj time(S)
SS Error(%)
Delay time(S)
Adj time(S)
SS Error(%)

9.0    Conclusion:  
Grossly three factors viz. Load (MS flow) Heat transfer from Flue gas and spray quantity are mainly responsible for variations in SH temperature. However on account of various interactive loops, non linear load dependent process response coupled with time delay make superheat temperature control in a thermal power plant very difficult and challenging  job. So depending on system design, cost effective systems could be adapted to improve system efficiency and reduce fatigue and failures. Small improve in efficiency may have large impact on cost saving and lesser pollution.
[1] D E B(R) coordinated control of steam temperature for once through supercritical boilers-METSo ( 2009
 [2] State controller with observer  CCI Sulzer valve –white paper 2003
[3] Advanced control of steam superheat temperature on a utility boiler  by B. Gough Andritz Automation (
[4] Boiler steam temperature control—optimizing efficiency and lifetime  by T.K. Seal ,T.K Nandy & Anoop K NTPC
 [5] Steam Temperature control  blog: September 8 2010
[6]Power station and process control system: modern Control Algorithms in Process control systems : observer based state control Mauell:
[7]Adaptive predictive expert control of superheated steam temperature in a coal fired power plant by R.R.Perez,A. Geddes & A. Clegg International journal of adaptive control and signal processing wiley online library ( 2012.
[8]Design of superheated steam temperature control system based on ADRC-PID for ultra supercritical unit by: J.Guo & X.jiang : international journal of advancement in computing technology (IJACT) vol 4 num: 21 November 2012.
 [9] Induction of Fuzzy rules and membership functions from training examples  by: T.P. Hong & C.Y. Lee : Elsevier: fuzzy sets and systems 84(1996) 33-47.
[10] Superheater Temperature control using Fuzzy algorithm  y : G.A. Pereira,V.K. Prabhullachandran & U. Krishnan ERDC Thiruvananthapuram India
[11] Intelligence –based hybrid control for power plant boiler by: W.Wang, & J Zhang : IEEE transactions on control systems Technology Vol 10 no.2 march 2002
[12]Design Research of an adaptive –fuzzy neural controller by: P.niu,G Li & M Zhang :Journal of advances in information Technology, vol.2 No.2 May 2011.

       we meet again next week.
        “stay tuned for a new post next week…” 

      Best wishes for you ! An appeal to help the mankind in distress!!!!
         CAN I HAVE FEEDBACK FROM YOU AT my mail id : 

No comments:

Post a Comment