CHAPTER 6 SYSTEM IDENTIFICATION AND CONTROLLER IMPLEMENTATION

Transcription

1 133 CHAPTER 6 SYSTEM IDENTIFICATION AND CONTROLLER IMPLEMENTATION 6.1 INTRODUCTION Process models have become useful for guiding plant operation, safety scrutiny, design of security systems, process design and control systems design. The two main methods which have been generally used to develop process models are i)empirical or data based modeling and ii)modeling based on the underlying physics and chemistry of the process (Liu et al 2000, Prempain et al 2000) Empirical Models Development In empirical modeling, the following procedure is adopted 1. Collection of experimental data from the process. 2. Specifying the relationship structure between variables. 3. Usage of a numerical technique that finds parameters for the structure. 4. Validate the model using real time data set. Hence Empirical modeling depends on the availability of representative data for model building and validation. For the development process model trial and error approach is adopted and information in terms of process knowledge is not required.

2 Mechanistic Models Development The developments of mechanistic models follow a procedure: 1. Fundamental know ledge of mass and energy balance equations of the process are used to define the model structure 2. Perform experiments to determine the parameters of the model 3. Collection of data from the process to validate the model Hence mechanistic modeling does not require much data for model development, and it is not subject to the unconventional behavior in data. This modeling requires a fundamental understanding of the physics and chemistry governing the process and also time consuming Comparison of Empirical and Mechanistic Models Mechanistic models can provide realistic in predictions and analyzing. A mechanistic model offers the opportunities to test the sensitivity of the process to meaningful entities are heat transfer coefficients, activation energies, catalyst poisoning, etc., The parameters of mechanistic model are derived based on various assumptions and ideal process conditions. Hence such sensitivity studies cannot be performed and mechanistic models are not often used to design processes (Chern 1989). Empirical models are frequently used in process controller designs. The model based controllers only require accurate and validated model. Little inaccuracy in the model development can be overcome by the Conventional tuning methods. If a validated mechanistic model is available, it will be meaningless to reject it in favour of an empirical model.

3 135 Due to the complexity of many processes, mechanistic modeling is indeed very expensive in terms of human effort and expertise. As the mechanistic modeling approach forces a detailed examination of fundamental process behaviour, some of the cost is recovered in terms of better knowledge of process behaviour. In practice, empirical modeling can be expensive as well. It requires large amounts of representative data, and in many instances, these can only be acquired by disturbing the process via planned experiments. The advantage with empirical modeling lies in the fact that empirical modeling will deliver some form of working model in a shorter time. Therefore what the model is to be used for. If it is to design control algorithms, then empirical models will do. However, if we need a model to design a new process or one that can be used to trouble-shoot a process that is behaving poorly or a model that is capable of pointing towards fundamental improvements in process operability, then it is best to develop a mechanistic model. The empirical modeling has been chosen in the research work for identification of gasifier model and controller development. Two different downdraft biomass gasifier systems have been considered to develop model and controller namely i) 135Kg/hr downdraft biomass gasifier system and ii) 6Kg/hr downdraft gasifier system. 6.2 EXPERIMENTAL STUDY ON BIOMASS GASIFIER SYSTEM (135Kg/hr) The open top downdraft biomass gasifier of capacity (135kg/hr) set up in R.V College of engineering, Bangalore, as shown in Figure A12.4, installed by NETPRO has been chosen for the present work to develop a static model and control. Experiments have been carried out in this system to

4 136 study the effect of moisture, effect of temperature, effect of air flow rate, effect of grate frequency of rotation, on the gas yield Definition of Gas Yield Gas yield is very important for the discussion as all the parameters like moisture, gas flow rate, size of the wood depends on it. Yield directly depends on the calorific value and the average calorific value (Karkezi et al 2006). The average energy conversion of the wood gasifier is about max % and can be defined as Equation 6.1. % yield calorific valueof thegas / kg of the fuel Avg calorific valueof the1kg of thefuel (6.1) On an average 1 kg of biomass produces about 2.2 m 3 of producer gas at S.T.P. In this process it consumes about 1.2 m 3 of air for combustion. For complete combustion of the wood about 4.2 m 3 of air required. Thus biomass consumes 33% theoretical stachometric ratio for wood burning. Then average energy conversion is written in Equation MJ 2.2(m ) * 4.5 m 3 % yield 62% MJ *1(kg) kg (6.2) The calorific value of the gas was estimated by the average gas composition and their individual calorific value Effect of Moisture on the Gas Yield In combustion systems any water content in the fuel must be driven off before the gasification the detailed evaluation of the moisture is given in the Appendix 4.

5 137 The moisture content of the most biomass fuel depends on the type of fuel, it s origin and treatment before it is used for gasification. Moisture content of the fuel is usually referred to inherent moisture plus surface moisture (Redding 1979). The moisture content below 15% by weight is desirable for trouble free and economical operation of the gasifier. Higher moisture contents reduce the thermal efficiency of gasifier and results in low gas heating values. Igniting the fuel with higher moisture content becomes increasingly difficult, and the gas quality and the yield are also poor. Reduction of the temperature below the optimum may result in incomplete combustion of the fuel giving rise to the emission of the tars and creosote which may condense in the flue, especially if it is long or includes the changes of direction and particulates. The water may also recondense in the flue and the gradual accretion of materials leading to the potentials for eventual blockages of fire. Many biomass gasifiers are designed to operate on very low moisture content feedstock perhaps to 10%- 15% (Rosenzweig and Parry 1994). Other technologies, such as anaerobic digestion, fermentation, hydrothermal upgrading and super critical gasification all make use of feed stock in aqueous medium and are particularly suitable for very high moisture content biomass for which drying is unnecessary. Biomass mainly in the form of wood is the oldest form of energy used by humans. Gasification is the process of conversion of solid carbonaceous fuel into combustible gas by partial combustion. In the present study, deals the moisture with the reaction.for the dry biomass feed regular usage of drier can be done to remove excess moisture preset in the wood, first to know the moisture content percentage desiccators is continuously used.

6 138 The best method of removing water adhering to solids is drying under reduced pressure when exhausting desiccators, a filter flask trap should always be inserted between the desiccators and pump. The vacuum should be applied gradually and should not exceed about 50 cm of mercury. If low pressure desiccators are used then steel wire cage must be provided. Experimental results from the Appendix 4 shows the fact of moisture and the yield relationship. Each trial Charcoal present at the bottom of the reactor is replaced before wood is fed in to the boiler. Wood is taken experimentally tested for the moisture as shown in Table A 4.1 and Figure A Table 6.1 Yield relation with moisture No of trials Yield (%) Moisture (%) Trial Trial Trial Trial Trial Trial Trial Trial Trial Trial

7 139 Figure 6.1 % Yield versus %moisture The Figure 6.1 and Table 6.1 clearly indicates that gas yield of the biomass looks consistent at about 10% of moisture and significantly relates that when yield is high, % moisture is low which indicates that gasification efficiency was high at low moisture and high yield concentration Effect of Size of the Wood on Gasification The fuel size affects the pressure drop across the gasifier and power that must be supplied to draw the air and gas through gasifier. Large pressure drops will lead to reduction of the gas load in downdraft gasifier, resulting in low temperature and tar production. Excessively large sizes of particles give rise to reduced reactivity of fuel, causing start-up problem and poor gas quality. Acceptable fuel sizes depend to certain extent on the design of gasifier. In general, wood gasifier work well on woodblocks and wood chips ranging from 80x40x40 mm to 10x5x5 mm. For charcoal gasifier, charcoal with size ranging from 10x10x10 mm to 30x30x30 mm is quite suitable. Size decides the performance activity of the gasifier.

8 140 Table 6.2 Yield characteristics with size No trial Size of the wood(mm) Yield (%) Trial Trial Trial Trial Trial Trial Trial Trial Trial Trial Figure 6.2 %Yield and size of the wood on different trials The Figure 6.2 and Table 6.2, clearly indicates that as the size of the wood piece was large, yield obtained reduced and the moisture content of the wood also high. It was observed from the trial the size 25 mm of the feed the yield remains to be constant and varies slightly. Hence optimum size of the feed can be considered is 25 mm.

9 Effect of Various Feed Materials on the Yield Wood is quite suitable fuel for fixed bedgasifiers. Though the ash content is low (less than 1 %), but because of high oxygen content, the calorific value is low (4.5-5 MJ/m 3 ). Peat is the first stage of coal formation. It cannot be utilized unless air dried to reduce moisture content to 30 % or less. Its calorific value (around 5.98 MJ/m 3 ) is slightly greater than wood. Rice husk among the feed materials has lowest calorific value when dried like 3.75 MJ/m 3. Rice husks during moist weather tend to give low yield compared to, the other biomass feed materials. Compared to all the feed materials used for the gasification purpose coconut shell has highest calorific value and when dry condition gives highest yield compared to others (Kaupp and Goss 1983).It has calorific value of 7.20 MJ/ m 3 (Appendix 8). Table 6.3 Yield and calorific value relation with different feed materials Different feed materials Calorific value MJ/m 3 wood 4.5 Coconut shell 7.20 peat 5.98 Rice husks 3.75 %yield

10 142 Table 6.4 Yield relation with different feed materials Feed materials % yield Wood Coconut husks Peat Rice husks Figure 6.3 %Yield and different feed materials From Tables 6.3, 6.4 and Figure 6.3 it was clear that coconut shell has highest calorific value and high yield compared to other three feed materials Effect of Tar on the Yield Tar effect is very important in the case of biomass gasification. The concentration of tar varies with respect to the yield percentage. Gasification is a proven technology that has been used for over 100 years to convert carbonaceous feed into gaseous products of useable heating value. A unique downdraft gasifier has been designed and fabricated to generate synthesis gas

11 143 from low bulk density biomass materials such as wheat straw, corn cobs, wood chips and sorghum stalks etc. Gasifier performance was evaluated using distillers dried grains with soluble (DDGS) and wood chips as the feedstock. Gas Chromatograph (GC) was employed to analyze syngas quality. Gasification of DDGS resulted in a syngas composition (mol %) of 19% H 2 and 20% CO; as for wood chips gasification, 20% H 2 and 15% CO (Rabou L P L M 2005) Both had high concentration of N 2, which ranged from 50% to 60%. Small amount of O 2, CO 2 and CH 4 were also included in the syngas. A Tar and Particulates sampling and analysis system was also configured to measure the tars and particulates concentration in the syngas. Tar concentrations were 4.20 g/nm3 for wood chips and 1.65 g/nm3 for DDGS. Under a given set of circumstances (especially long idling periods) this can lead to tar formation and clogging of cooler/cleaners and engines (Reto et al 2000). Efficient cleaning of the gas and correct adaptation of the products of biomass gasification to the specific requirements of the gas combustion systems are prerequisites for use in gas-fired engines, gas turbines and fuel cells. Tar compounds can be effectively removed by increasing the gas temperature or by catalytic cracking over nickel. However, even for wood gasifier there is still no economically viable solution of this tar problem. The tar problem is classified into major two categories. 1. Heavy tars-condensation leads to fouling. < 350 C tar dew point is critical for the parameters 2. light tars- problem is the Naphthalene crystals on gas engine control valve blocks the continuous flow of the biogas (Milligan 1994).

12 144 One of the major obstacles for the application of biomass gasification systems is the high tar content of the producer gas that prevents the use in engines, turbines or fuel cells without further gas cleaning Effect of Temperature on the Yield Temperature effect is high which can be studied from the thermometer which is calibrated compulsorily. On an average the temperature of gas leaving the gasifier is about C to C if the temperature is higher than this (~ C) it is an indication that partial combustion of gas is taking place. This generally happens when the air flow rate through the gasifier is higher than the design value. The temperatures at various heights inside the reactor were recorded using calibrated K-type thermocouples (Inconel 600, < 1200 ºC 5 ºC) as shown in Figure A The total length of the reactor is 3.2 m. diameter of the thermocouple is about 6mm and length is 3 m. The Thermocouple has a digital temperature indicator and the temperatures are read out directly. It was positioned vertically inside the reactor. Then the thermocouple was moved in steps of 20cm, and the temperatures were recorded accordingly Rate of Flow of Gas The rate of flow of the gas yield depends upon many factors like temperature, %moisture and load current. In fact it is indirectly depends on the nature and type of the feed work is mainly concentrated on the % moisture with rate of flow. In this case venturimeter connected at the bottom of the reactor decides the rate of gas yield. Venturimeter used must be initially calibrated to meet the requirement. Venturimeter is used to measure the rate of flow through a pipe. Venturimeter consists of a converging portion, throat and a diverging portion.

13 145 The function of the converging portion is to increase the velocity of the fluid and temporarily lower its static pressure. The pressure difference between inlet and throat is developed. This pressure difference is correlated to the rate of flow. The expression for theoretical flow rate is obtained by applying the continuity equation and energy equation at inlet and throat section, and assuming the fluid to be ideal is given by Equation 6.3 Qth A2 2gh 1 (A2 / A1)2 (6.3) Where A1 and A2 are areas at inlet and throat & h static pressure difference between inlet and throat section in terms of mm of water. Calibration of flow meters Equation 6.3 relating flow rate to the differential pressure cannot be applied directly in practical applications. All the flow meters need calibration for that a known quantity of fluid is passed through the flow meter and the differential pressure across the flow meter related to the actual mass flow rate through a discharge coefficient given as the ratio of actual to theoretical mass flow rate. Two methods of knowing the actual mass flow rate aremeasurement of time for collection of a finite volume of fluid and measurement of mass collected in a certain amount of time. Procedure: Adjust the discharge. Note down the pressure difference hm. calculate the theoretical discharge Q th. Note down the time for collection of 600 liters of water in the measuring tank and determine the actual discharge Q act. Also calculate the actual discharge by measuring the mass collected in 300 sec. with the help of load cell and data acquisition software provided. Calculate the coefficient of discharge C d. Repeat the procedure for at least ten mass flow rates for both venturimeter. The static

14 146 model has been proposed based on the above experimental identification of the biomass gasifier plant. 6.3 DEVELOPMENT OF STATIC MODEL FOR BIOMASS GASIFIER SYSTEM (135Kg/hr) Downdraft biomass gasifier system with capacity (135kg/hr) has been considered to develop a static model and controller. Table 6.5 Recorded biomass gasifier plant data No Trial F A (m 3 /h) f g (s/s) H p (%) F hb (kg/h) ER (%) T ( 0 C) CO/CO 2 Ratio (%) Trial Trial Trial Trial Trial Trial Trial Trial Trial Trial Trial Trial Trial Trial The many experiments done with biomass consumption, temperature and CO/CO 2 ratio by varying the air flow rate (F A ), the frequency of motion of the grate (f g ) and moisture(h p ). Some experiments data reported in Table 6.5.

15 147 The experimental data observed from this gasifier characterizes this process as a non-linear and slow process, and hence the development of model has become a challenge. The proposed model must be representing the non-linear characteristics of the process. Certain simple mathematical equations have been developed for the static model of the gasifier operations by adjusting the mathematical relations between the variables with reference to the recorded data in order to equate the original plant data. Four subsystems namely biomass consumption, Equivalence ratio,co/co 2 ratio and temperature for static model of biomass gasifier were developed using the MATLAB. One of the objectives of the control model developed here is to tune the controller Biomass Consumption (F hb ) The flow of biomass was supplied in a randomly through top of the downdraft gasifier to maintain the gasifier permanently full of biomass. The amount of biomass consumed for the gasification process being considered for monitoring. It depends upon the air flow rate (F A ), frequency of rotation of the grate (fg) and moisture content (Hp). The relations of these variables on flow of biomass were expressed in Equation 6.4 and simulation of this equation shown in Figure FA 1.4 F hb = 1/3 13. f Hp g (6.4) Figure 6.4 Subsystem for biomass consumption

16 Equivalence Ratio (ER) Equivalence ratio was calculated based on as F hb, F A, Hp and the type of material expressed as a function of a material factor (m b ) which represents the amount of air needed to obtain combustion of 1 kg of dry biomass. Which was estimated based on stochiometric calculations. It can be computed automatically from the elemental composition of the biomass. The expression for equivalence ratio is as follows Equation 6.5 and simulation of this equation shown in Figure 6.5. FA ER = 100 Fhp(1 Hp) mb (6.5) Figure 6.5 Subsystem for equivalence ratio CO/CO 2 ratio The CO/CO 2 ratio depends on ER and Hp. When Hp is low, the ratio increases with ER to reach a maximum. The expression formed for the CO/CO 2 ratio is written in Equation 6.6 and simulation of this equation shown in Figure 6.6. CO/CO 2 ratio = (0.3Hp 0.5) ER (6.6)

17 149 Figure 6.6 Subsytem for CO/CO2 ratio Temperature (T) Temperature is related to the quality of produced gas very closely. It depends on ER, Hp and F A. The expression for T is Equation 6.7 and simulation of this equation shown in Figure 6.7. The type of biomass was expressed as a function of two factors, m a and m c. T = 33 m. ER f m Hp c g a (6.7) The Equations some constant values were assumed in order to fit with experimental data. Figure 6.7 Subsystem for temperature Using the above simulink models of the various sub-systems, the complete steady state model of the biomass gasifier was developed. For true

18 150 simulation, the model needs to adapt to the particular un-modeled characteristics of a particular run. In order to adapt, specific model parameters were fine tuned to make the model output match the measured experimental output. The gasifier model for control is shown in Figure 6.8. This model has been used to validate the rules and membership functions of the fuzzy controller. Figure 6.8 MATLAB simulink model of the gasifier 6.4 FUZZY MODELLING A static model for the gasifier system has already been developed. But, the accuracy of this model was less with certain input conditions. It was less efficient and does not resemble the plant very closely for some input conditions. This is the reason to choose the fuzzy modelling approach. For a system like gasifier, the model must be representing the non-linear dynamic characteristics of the process. Here, a fuzzy system is modeled with three

19 151 inputs and one output based on the experimental data reported in table 6.5. The three input variables are hp (moisture content), flow and grate and the output variables are CO/CO 2 ratio and temperature. Inputs are given to the fuzzy model in the operating region of the gasifier, it give the corresponding outputs, based on the membership functions and the rules written. Fuzzy logic uses the whole interval between True and False to describe human reasoning. A logic based on only the two truth values True and False is sometimes inadequate when describing human reasoning. This model is based on the fuzzy logic. The steps of establishing a fuzzy model are as follows. a) Divide the tested data into two parts. The first part is modeling data used to establish a fuzzy model. The other part is validation data which is used to validate the precision of model. b) According to the modeling data, determine the input and output variables, the corresponding domain interval, the number of fuzzy subset, the quantification factors, and the membership functions. c) Quantify the modeling data, calculate the membership degree of every data in every subset, and the maximum membership degree determines the data belonging to a subset. d) Based on the knowledge of the process, write the fuzzy rules. The modeling data are used to validate the model, and then fuzzy rules are adjusted. e) Validate the model using validation data, adjust the fuzzy rules. And the final fuzzy rules can be obtained.

20 152 Figure 6.9 Membership function for H p Figure 6.10 Membership function for f g Figure 6.11 Membership function for flow

21 153 Figure 6.12 Membership function for temperature Figure 6.13 Membership function for CO/CO 2 The two linguistic variables for H p and these were high and low. The five linguistic variables chosen for f g like f1, f2, f3, f4 and f5. Similarly, for flow, Five linguistic variables were chosen like EL, VL, B, VH and EH. One of the output variables temperature has seven linguistic variables were chosen like EL, VL, L, B, H, VH and EH. The other output variable CO/CO 2 ratio was chosen with five linguistic variables such as TL, L, N, H and TH. The membership functions of process variables were shown in Figures 6.9, 6.10, 6.11, 6.12 and 6.13.

22 154 in Table 6.6. The rules were constructed for gasifier modeling and it is reported Table 6.6 Rules for plant modeling Sl.No. IF h p and grate and airflow Then temperature Then co/co 2 1. low f1 EL EL TL 2. low f1 VL VL TL 3. low f1 B B TL 4. low f1 VH VH TL 5. low f1 EH EH TL 6. low f2 EL EL L 7. low f2 VL VL L 8. low f2 B L L 9. low f2 VH H L 10. low f2 EH EH L 11. low f3 EL EL N 12. low f3 VL L N 13. low f3 B B N 14. low f3 VH H N 15. low f3 EH EH N 16. low f4 EL EL H 17. low f4 VL VL H 18. low f4 B B H 19. low f4 VH VH H 20. low f4 EH EH H 21. low f5 EL VL VH 22. low f5 VL L VH 23. low f5 B B VH 24. low f5 VH VH VH 25. low f5 EH EH VH

23 155 Table 6.6 (Continued) Sl.No IF h p and grate and airflow Then temperature Then co/co high f1 EL EL TL 27. high f1 VL VL TL 28. high f1 B H TL 29. high f1 VH VH TL 30. high f1 EH EH TL 31. high f2 EL EL TL 32. high f2 VL EL TL 33. high f2 B L TL 34. high f2 VH H TL 35. high f2 EH VH TL 36. high f3 EL EL L 37. high f3 VL EL L 38. high f3 B L L 39. high f3 VH H L 40. high f3 EH VH L 41. high f4 EL EL N 42. high f4 VL EL N 43. high f4 B VL N 44. high f4 VH B N 45. high f4 EH H N 46. high f5 EL EL H 47. high f5 VL EL H 48. high f5 B L H 49. high f5 VH B H 50. high f5 EH H H

24 DESIGN OF FLC FOR BIOMASS GASIFIER SYSTEM (135Kg/hr) A Fuzzy Logic Controller has been proposed for downdraft biomass gasifier system (135kg/hr) using the developed static model. A fuzzy system is a static nonlinear mapping between its inputs and outputs. It provides a formal methodology for representing, manipulating and implementing a human heuristic knowledge about how to control a system (Claudio 2005, Driankov et al 1996, Yamakava 1993, Ronald et al 2002). The design procedure of a Fuzzy system involves the following steps. 1. Choosing the fuzzy system inputs and outputs 2. Putting control knowledge into rule base. 3. Fuzzy quantification of knowledge. 4. Matching/Determining which rules to use. 5. Inference Step of determining conclusions. 6. Converting decisions into actions. In controller design it is very much needs to identify controlled variable, manipulated variable and disturbance of the process. The block diagram of gasifier control system and variables involved is shown in Figure This gasification plant biomass consumption was mainly depended on the air flow rate and the frequency of motion of the grate. For a given constant air flow rate, as the frequency of motion of the grate increased, the ash removal rate also increased, and therefore the biomass consumption was larger just to maintain the biomass level inside the gasifier. Hence, the air flow rate (F A ) and the frequency of motion of the grate (fg) were selected as the action variables of the fuzzy control. Both were actuated by electric motors. To select the process variables, two aspects needed to be taken into

25 157 account. On the one hand, the heating value of the produced gas was calculated from the average gas composition during each run and only the concentrations of CO and CO 2 could be measured on-line. So, one of the selected process variables for control was the CO/CO 2 ratio because it was important to maintain it under control. On the other hand, the dependency of the quantity and quality of the produced gas on the temperature in the throat zone could be identified from the mathematical model developed for this gasifier. As the temperature profile inside the gasifier could be measured online, the throat temperature (T) was selected as the second process variable and moisture considered as a disturbance. Hence following variables were used to design the fuzzy controller. i) Manipulated variables : Air flow rate (F A ) and frequency of rotation of the grate (f g ) ii) Disturbance: Moisture content (Hp) iii) Controlled variables: Temperature (T), and CO/CO 2 ratio Figure 6.14 Block diagram of the gasifier control system

26 158 The knowledge of experts about biomass gasification could be represented in fuzzy if/then rules in such a way that the fuzzy systems can deal with the indistinguishable and inaccurate biomass condition (Zhao et al 1992). This proposed work, the fuzzy logic method has been adopted with three inputs and two outputs for the capacity (135kg/hr) of biomass gasifier. The three inputs were error T (error in temperature), error CO/ CO 2 and Hp (moisture content) and the outputs were airflow and frequency of grate. Fuzzy rules have been formulated based on error temperature, error CO/ CO 2 ratio and Hp (moisture content) which has converted to non-fuzzy values by defuzzification. These values have been fed to the final control element for control action. The simulation representation of gasifier control is shown in Figure The five linguistic values have been defined for Error CO/ CO 2 variable like very low, low, zero, high, very high, and five linguistic values have been defined for the Error temperature T variable such as very low, low, zero, high, very high. The five fuzzy values were defined for output variable Airflow rate such us extremely low, very low, base, very high, extremely high and also for the Grate frequency variable, the values f1, f2, f3, f4, f5, were defined. The membership functions were shown in Figures Figure 6.15 Simulation of control system

27 159 Figure 6.16 Membership function for H p (moisture content) Figure 6.17 Membership function for error T Figure 6.18 Membership function for error CO/CO 2

28 160 Figure 6.19 Membership function for air flow rate Figure 6.20 Membership function for grate frequency of rotation The rules that have been framed for this developed controller were represented in Table 6.7 and 6.8.

29 161 Table 6.7 Rules for adjusting frequency of grate Sl.No If hp and errort and errorco/co 2 Then grate 1. low Verylow verylow f4 2. low Verylow low f3 3. low Verylow zero f1 4. low Verylow high f3 5. low Verylow veryhigh f1 6. low Low verylow f3 7. low Low low f2 8. low Low zero f2 9. low Low high f2 10. low Low veryhigh f1 11. low Zero verylow f3 Table 6.8 Rules for adjusting air flow rate Sl.No If hp and errort and error CO/CO 2 Then flow 1. low Verylow verylow EH 2. low Verylow low VH 3. low Verylow zero VL 4. low Verylow high VH 5. low Verylow veryhigh VL 6. low Low verylow VH 7. low Low low VH 8. low Low zero B 9. low Low high B 10. low Low veryhigh VL 11. low Zero verylow VH 12. low Zero low B 13. low Zero zero VL

30 162 Table 6.8 (Continued) Sl.No If hp and errort and error CO/CO 2 Then flow 14. low Zero high VL 15. low Zero veryhigh EL 16. low High verylow B 17. low High low B 18. low High zero VL 19. low High high VL 20. low High veryhigh EL 21. low Veryhigh verylow VH 22. low Veryhigh low B 23. low Veryhigh zero VL 24. low Veryhigh high VL 25. low Veryhigh veryhigh VL 26. low Verylow Verylow EH 27. low Verylow low EH 28. low Verylow zero EH 29. low Verylow high EH 30. low Verylow veryhigh EH 31. low Low verylow EH 32. low Low low EH 33. low Low zero EH 34. low Low high VH 35. low Low veryhigh VL 36. low Zero verylow VH 37. low Zero low B 38. low Zero zero VL 39. low Zero high VL 40. low Zero veryhigh VL 41. low veryhigh verylow VL

31 163 This developed fuzzy logic controller of biomass gasifier system (135kg/hr) was based on the static model of the biomass gasifier that has been proposed. The prototype was design for fuzzy controller to prove the theoretical performance Prototype Design of FLC In order to validate the developed fuzzy logic controller performance on gasifier operation a small prototype has been designed with help of microcontroller and C coding. The Data Acquisition unit function was the input parameter converted into the standard range. The signals of standard value were given as time- multiplexed inputs to the ADC that converts each of its analog input into the corresponding digital output. The digital data were sent to Microcontroller for processing. The fuzzy concepts were programmed using C language. The Programming steps involved were acquiring the input parameters such as temperature and, CO/CO 2 in terms of voltage. 1. The Fuzzifying the given inputs to determine the strength of each membership function. 2. Determining the Fuzzy rules that are to be applied. This can be inferred from the Rule Base Inference which was framed based on the Input & Output data. 3. Defuzzifying the fuzzy sets using Center Average method to convert the fuzzified value to crisp value. This gives the output of the controller. The developed program was dumped into the AT89C52 microcontroller using Lab tool- 48 Intelligent Universal Programmer. In LabVIEW the developed model was implemented for receiving the multiple

32 164 data sent by the microcontroller and displaying the measured variable and controller output. The steps involved in controlling the temperature and CO/ CO 2 ratio using the Fuzzy algorithm. The analog input corresponding to the parameters temperature and CO/CO2 were given to one of the eight channels of the 8-bit ADC (ADC 0809). By means of a Timer (555) circuit, the ADC is supplied with a clock pulse of frequency 290 KHz approximately. This is required for the ADC to perform the operation of sampling the analog inputs.at89c51 microcontroller with 12 MHz internal clock frequency selection bits through a port. It also gives the necessary signals for A/D conversion such as ALE, SOC etc to the ADC. The Octal buffer (74LS245) is a digital circuit that increases the driving capability of the data and the address buses of the microcontroller by suitably amplifying the current. Hence it is placed in between the ADC and the microcontroller as shown in Figure The microcontroller acquires the digital data sent by the ADC through one of its data port and feeds are to the Fuzzy code that calculates the controller output. the controller output is sent to DAC (DAC 0800) through the data port P2.The signal and controller output were sent to the PC through its status port of serial port in labview for monitoring the measured variable and controller output. In the bidirectional buffer, data can be allowed to flow in both the directions by controlling the DIR pin. Hardware connections are made in such a way to get lower nibbles as output by disabling it. Thus a byte of data is received by the PC through the status port. The DAC converts the digital input (Controller output) to analog signal in the range of required level. The analog output is sent to the final control element for controlling the process.the flow chart for the fuzzy implementation in microcontroller is shown in the Figure 6.22.

33 165 Figure 6.21 Circuit diagram of the FLC Figure 6.22 Flow chart of the FLC

34 166 The CO/CO 2 ratio was effectively controlled with fuzzy logic controller by adjusting the frequency of motion to control the biomass residence time inside the reactor. The temperature was also effectively controlled with fuzzy logic controller by adjusting the airflow rate. 6.6 EXPERIMENTAL STUDY ON BIOMASS GASIFIER SYSTEM (6Kg/hr) The small scale stratified downdraft biomass gasifier of capacity (6kg/hr) set up in Karunya university, Coimbatore has been chosen for this work to develop a dynamic model and controller. This is a very small size, only suitable for laboratory research. In this proposed work, a single-inputsingle-output (SISO) process control loop has been recommended with an intelligent fuzzy logic controller for the optimization of the process temperature Development of Dynamic Model of Biomass Gasifier System (6Kg/hr) The process reaction curve is the most widely used method for identifying dynamics models (Peter Harriot 1972). It provides adequate models for many applications. The graphical calculations determine the parameters for a first order with dead time model. The process reaction curve is an open-loop method, in which the control action is removed from the controller by placing it in manual mode and an open-loop transient is induced by a step change in the input signal. The transfer functions obtained through empirical modeling for each step input.

35 167 The process reaction curve method involves the following actions to develop the dynamic model of gasifier system, 1. The temperature process was allowed to reach the steady state for certain air flow rate. 2. A single step change was introduced in the input (air flow rate). 3. The input and output i.e. temperature response data are collected until the process again reached the new steady state. 4. The graphical process reaction curve calculations are performed. The development of the dynamic model for this gasification plant has been done using the process reaction curve method. The developed dynamic model was first order system with dead time. As an experimental result the gasification temperature has the highest influence on the efficiency and hence temperature has been considered as a controlled variable with air flow rate as a manipulated variable. The general transfer function of has been represented in Equation 6.8 first order system with dead time G(s) k e s 1 s (6.8) Where the terms k,, and meant that static gain, time constant and delay respectively. Then the values of k, and have been obtained from the real time biomass gasifier process response shown Figure 6.23.

36 168 Figure 6.23 Response of temperature versus time The Graphical calculations are as follows from the real time response: i) Gain (K) = (Final steady state value- Initial steady state value)/ step change = (620-50) / (115-15) = 570/ 100 = 5.7 ii) Time constant ( ) = time for the response to reach temperature T1 T1 = 63.2 % of (change in process variable) + offset = 63.2 % of (620-50) + 50 = C Time constant = 80 minutes = 4800 seconds

37 169 Based on the real time parameter s adaptation on the first order transfer function for the biomass gasifier system has been drawn in Equation 6.9. G(s) = s 1e (6.9) The step change from (15-115) lpm has found that optimum region for controlling the particular gasifier. The various controller like PI,PID, Fuzzy logic controller and Self tuning fuzzy logic controller were proposed using this developed model. 6.7 CONVENTIONAL CONTROLLER DESIGN Proportional-Integral-Derivative (PID) algorithm is the most common control algorithm used in industry presently. Often, people use PID to control processes that include heating and cooling systems, fluid level monitoring, flow control and pressure control. PID controller is not an adaptive controller, hence the controller has to be tuned frequently and whenever load changes. Auto- tuning of these controllers becomes difficult for complex systems (Liu G.P et al 2000). In order to prove the drawbacks of conventional controller in downdraft biomass gasifier a little attempt is made to design a PID control which is designed to ensure the specifying desired nominal operating point for temperature control of gasifier and regulating it, so that it stays closer to the nominal operating point in the case of sudden disturbances, set point variations, and noise. The proportional gain (Kp), integral time constant (Ti), and derivative time constant (Td) of the PID control settings were designed using Zeigler- Nichols tuning method. The simulink model of PID control is shown in Figure The conventional controller has not suitable for this type of highly non-linear and slow process

38 170 (Prempain et al 2000). In order to improve the gasifier control process the intelligent control techniques were proposed in this research. Figure 6.24 PID controller for downdraft gasifier 6.8 DESIGN OF FLC FOR DOWNDRAFT BIOMASS GASIFIER This biomass gasifier, a single-input-single-output (SISO) process control loop has been recommended with intelligent fuzzy logic controller technique for optimization of the temperature process. Fuzzy logic controller has been implemented for a small-scale stratified downdraft biomass gasifier to verify the efficiency of fuzzy controller in comparison with the conventional controllers. The temperature error T and change in temperature error T have the inputs for this fuzzy controller and output of the controller was air flow rate. Figure 6.25 shows the triangular membership function for error T, Figure 6.26 shows the membership function for change in error of T. Figure 6.27 shows the membership function of air flow rate. The rules used for the fuzzy controller has been represented in Table 6.9. Where, e(t), e(t) and u(t) meant that error, change in error and control signal respectively. The linguistic values has been defined for the input temperature error, change in temperature error and output airflow such as Negative Big (NB), Negative Medium (NM), Negative Small(NS), Zero(ZO), Positive Small(PS), Positive Medium (PM) and Positive Big(PB). The simulation of

39 171 the fuzzy logic control system is shown in Figure This developed controller has fixed gain, which cannot be tuned its gain automatically with uncertainty occur in the process. Hence self tuning fuzzy controller has been proposed for this process. Figure 6.25 Membership function of error T Figure 6.26 Membership function of change in error T

40 172 Figure 6.27 Membership function of air flow rate Table 6.9 Rules for fuzzy logic controller U(t) e(t) e(t) NB NM NS Z PS PM PB NB NB NB NB NB NM NS Z NM NB NB NB NM NS Z PS NS NB NB NM NS NS PS PS Z NB NM NS Z Z PM PM PS NM NS Z PS PS PB PB PM NS Z PS PM PM PB PB PB Z PS PM PB PB PB PB Figure 6.28 Simulation of fuzzy logic controller

41 DESIGN OF SELF TUNING FLC FOR BIOMASS GASIFIER SYSTEM The self tuning fuzzy controllers (SFCs) to continuously adjust, online, the input/output scaling or gain factors of the ordinary fuzzy controller in order to improve its performance against different dynamic operating conditions. The designed self tuning controller is used to temperature control of a biomass gasifier system with un known parameters according to the feedback inputs, error (e) and change of error ( e), based on the proposed fuzzy rules. The system implementation and tests were carried out using LabVIEW software. The adaptive controller can be thought of as being composed of two loops. The inner loop consists of process and an ordinary feedback controller (Kevin M. Passino and Stephen Yurkovich 1998). The parameters of the controller were adjusted by the outer loop, which has been composed of a recursive parameter estimator and a design calculation. It has sometimes not possible to estimate the process parameters without introducing probing control signal. Notice that the system may be viewed as an automation of process modeling and design in which process model and the control design are updated at each sampling period. A controller of this construction can be called a self tuning regulator to emphasize that the controller automatically tunes its parameter to obtain the desired properties of a closed loop system. Figure 6.29 shows the block diagram of a self tuning controller.

42 174 Figure 6.29 Block diagram of self tuning fuzzy control In order to make the system adaptive the input and output gain of the controller were adjusted on-line according to the current states of the controlled process, i.e. gain scheduling using fuzzy logic thereby making them self-tuning FLC s.this logic has been depict in Figure 6.29 in which G e and G ce were proportional and derivative gains respectively in the input side of the FLC and G m has the output scaling gain. The adaptation rules for G ce, were opposite to that of G e. Table 6.10 and 6.11 represented the rules used G e and G ce for the self tuning controller respectively. The self-tuning fuzzy logic controller has been developed and implemented in LabVIEW. Figure 6.30 shows the block diagram of this controller. Table 6.10 Rule base for G e U(t) E E NB NM NS ZO PS PM PB NB VB VB S S M B B NM VB VB S S S M VB NS VB VB VS M VS VB VB ZO VB B VS M VS VB VB PS VB M VS M VS VB VB PM B B S S S M VB PB B B S S S B B

43 175 Table 6.11 Rule base for G ec U(t) E E NB NM NS ZO PS PM PB NB VB VB VB VS VS VS S NM VB VB VB VS VS VS S NS VB B B M S S VS ZO VB B B M VS S VS PS B S S M B B VB PM S S VS VS B VB VB PB S VS VS VS B VB VB Figure 6.30 Block diagram of self tuning controller in LabVIEW 6.10 DEVELOPMENT OF AUTOMATIC ASH HANDLING SYSTEM As a part of the research work the automatic ash removal process of downdraft biomass gasifier system also has been proposed. In wood gasification plant the ash removal process is a major challenge because the accumulated ash affects the efficiency. A clean feedstock has been played an important role from the viewpoint of continuous operation. Time and again,

44 176 there would be problems associated with poor quality bio-feedstock. This matter could be resolved by using a screw arrangement to draw away the residue that settles at the bottom as shown in Figure A Some agroresidue briquettes with a mix of groundnut shell have high ash content.the system involving grate would have difficulty in permitting high ash content feedstock which required a screw based ash extraction system. The char/ash was finally disposed off at ground level using a combination of char/ash conveyor and manually operated bucket elevator. In existing biomass gasifiers, the removal of ash extraction has been done manually (Svrcek et al 2000). The amount of ash collected in the ash chamber varies in proportion to the size and moisture content of the wood and hence automating this process is difficult. The pressure was considered as the control variable because the pressure in the gasifier was directly dependent on the weight of ash accumulated. Fuzzy controller was developed to regulate the pressure Design Of FLC For Ash Handling System The two inputs were selected for the fuzzy controller namely error pressure and change in error pressure and weight of the ash has been selected as one output. Fuzzy rules have been formulated based on error pressure and rate of change of pressure error and it has been converted to non-fuzzy values by defuzzification. These values have been fed to the final control element for control action. The five linguistic values were defined for the Error pressure variable, and five fuzzy values for the output variable were defined. Input Membership Functions were shown in Figure 6.31 and Output Membership Functions is shown in Figure 6.32.

45 177 a. Error of pressure b. Rate of change of error Figure 6.31 Membership function of inputs (a) Error of pressure (b) Rate of change of error Figure 6.32 Membership function of output

46 178 in Table The rules that have been framed for this controller were represented Table 6.12 Rules for fuzzy logic controller If Error If Change in Error Then Controller output I I A I II A I III A I IV A I V B II I A II II A II III A II IV A II V B III I B III II C III III C III IV D III V D IV I D IV II E IV III E IV IV E IV V E V I D V II E V III E V IV E V V E

47 179 To validate the developed fuzzy logic controller performance on ash removal a small prototype has been designed with microcontroller and C coding. The C code for acquiring the input parameters i.e. pressure and, rate of change of pressure is written using Keil Cross- Compiler which gives the corresponding hex code. The hex is dumped into the AT89C51RC microcontroller using Lab tool- 48 Intelligent Universal Programmer. In the PC, C code is developed for receiving the multiple data sent by the microcontroller and displaying the measured variable and controller output. The input parameter is converted into the standard (0 to 5V) range using a signal conditioning unit. The signals of standard value are given as timemultiplexed inputs to the ADC that converts each of its analog input into the corresponding digital output. The digital data are sent to Microcontroller AT89C51RC for processing. The output from the transmitter (4-20mA) is converted into (0 to 5V) using a current to voltage converter shown in Figure Fuzzifying the given inputs is to determine the strength of each membership function. Determining the Fuzzy rules that are to be applied. This can be inferred from the Rule Base Inference which is framed based on the Input & Output data. Defuzzifying the fuzzy sets using Center Average method to convert the fuzzified value to crisp value. This gives the output of the controller. The controller output in the range of (0-5V) is converted into (4-20) ma using a voltage to current converter shown in Figure The interface circuit also shown in Figure A (Appendix 12).

48 180 Figure 6.33 Current to voltage converter Figure 6.34 Voltage to current converter Fuzzy controller was developed to regulate the pressure which offers better settling time and without over shoot.