7/29/2019 Control of SOFC With Fuzzy Logic

1/5

Control of Solid Oxide Fuel Cell for Stand-Alone and G rid Connection using FuzzyLogic TechniqueAbhishekR. Sakhare Asad DavariMS Control Systems. WW TechMontgomery,WV 25 136

abhishek-sakhare@hotmail.com adavari@wvu.eduECE Dept. WW TechMontgomery, WV 25 136

Ali FeliachiLDCSEE Dept. WWMorgantownWV 26506afeliachi@mail.wvu.eduKeywords: Fuel Cells, DCDC Converter, DC/AC Inverter, Fuzzy Logic, Grid.

Abstract - Fuel Cells are very promising sources ofelectricity; environmentally safe and very efficient. Thefuel cells have numerous applications: houses,industries, hospitals, vehicles etc. Another facet of theseFuel cell applications is distributed generation, theimplementation of various power generating resources,near the site of need, for reducing reliance on gridpower. Fuel cells generate electricity from hydrogen bya chemical process. In this paper a solid oxide fuel cellmathematical model is adopted. The solid oxide fuelcells are very efficient and the technology is most suitedto applications in the dist ributed generation. The mainaim of the paper is to achieve the control of the fuel cellfor stand-alone and grid connection. To achieve a gridinterface and stand-alone working by designingsuitable power conditioning units. The power-conditioning unit is needed for the processing of theraw power output of the fuel cell in order to make itusable. The power conditioning unit might consist ofonly a DCDC converter or the tw o stages of a DC/DCconverter and a DC/AC inverter. For the stand-alonepart the concentration is on the controlled DC powerthus, only a boost conver ter @C/DC) stage is used. Forthe grid interface of the solid oxide fuel cell controlledAC power is needed at the interface point, thus, bothstages; boost converter as well as the inverter @C/AC)are needed. A power conditioning unit is designed forthis solid oxide fuel cell and for fuel cells in general.The fuzzy logic control strategy is used for designingthe controllers.

I. INTRODUCTIONFuel Cell (FC) generate electricity from hydrogen by achemical process and their emissions are water [1,2]. TheFC can serve as an emergency source of energy in theevent of a long-term power outage. FC could be used asportable power systems. The FCs are finding use in everyaspect because of their clean and efficient way ofsupplying electric power. The FCs are used in the stand-alone purposes at homes, hospitals, industries and now arefinding their use in numerous vehicles. The FCs arereplacing the batteries and in the current trend arebecoming the most widely used resources. This stand-alone use of the FCs is one aspect of their applications.

Another aspect of the use of fuel cells is DistributedGeneration (DG). DG is to provide electricity to acustomer at a reduced cost, more efficiently with reducedlosses than the traditional power generating systems. Thebenefits that DG could potentially provide, depending onthe technology are reduced emissions, utilization of wasteheat, improved power quality and reliability and deferral oftransmission or distribution upgrades [1,2]. The ever-increasing need for electrical power at different sites on acontinuous basis (stand-alone) and for DG (grid-connected) with rapid progress in power deregulation haveattracted much attention towards FCs.The Solid Oxide Fuel Cells (SOFC) are particularlyattractive because they are the most efficient (in terms offuel input to electricity output). The technology is mostsuited to applications in the DG (stationary power). Thehigh operating temperature produces heat suited well tocogeneration applications. SOFC do not contain noblemetals and do not utilize liquid electrolytes which can beproblematic and expensive [1,2].The output voltage of FCs at the series of the stacks isuncontrolled DC voltage, which fluctuates with loadvariations as well as with the changes in the fuel input. Ithas to be controlled by a D O C onverter. The controlledvoltage thus obtained is then fed to the DC/AC inverter.The power obtained fiom the inverter is to be incorporatedinto the grid to interface the FC with the grid for DGapplications. The inverter acts as the grid interface. Thevoltage and the current at the inverter output needs to beconditioned for the grid connection of the FC. This paperuses a developed model of the SOFC and designs thecontrol strategies and controllers for the DC voltagecontrol via the DCDC converter and the current control ofthe DC/AC inverter to for the grid interface. Two separatecontrollers are designed for these purposes. The fuzzylogic control scheme is employed for the design of the twocontrollers.

11.THE OVERALL SYSTEMThe FC converters will play an intricate role in FCtechnology. The block diagram of the overall system isshown in Fig. 1. The unregulated output voltage of the FC

0-7803-828 1-1/04/$20.00 02004 IEEE 551

mailto:abhishek-sakhare@hotmail.commailto:adavari@wvu.edumailto:afeliachi@mail.wvu.edumailto:afeliachi@mail.wvu.edumailto:adavari@wvu.edumailto:abhishek-sakhare@hotmail.com

7/29/2019 Control of SOFC With Fuzzy Logic

2/5

is fed to the DCDC boost converter. The voltage isboosted depending upon the duty ratio. The unregulatedvoltage is converted to a regulated DC voltage by adjustingthe duty ratio of the boost converter which in turn isadjusted by the controller. The boost converter respondsfast to the changes in the duty ratio and regulates the DCvoltage. The fuel flow needs to be adjusted, which takeseffect gradually and controls the output voltage. Thecontrol of fuel flow is out of the scope of this paper.

FuelCell

by a control signal provided by the controller; a FuzzyLogic Controller (FLC) to keep the average output voltageat the desired level. The output of the DCDC converter atthe load is compared with the reference (desired) voltageand the error signal generated is one of the inputs for theFLC. The change in the error (the derivative or the rate ofchange of error with respect to time) derived from the erroris the other input which is fed to FLC. The FLC basedupon the inputs and the rule base generates a command

DCIDC--+ Converter ,

signal to adjust the duty ratio of the PWM generator of theconverter. The voltage output will be set based upon thisnew duty ratio value and adjust the unregulated DCvoltage coming from the FC to a regulated DC voltagewith a fixed average value[ll]. The response time of theDCDC converter is very short compared to that of thereformer. Thus for the fast system response, initially the

DCIACConverter Convertex

Controller Controller

Fig.1 Overall System Block DiagramThe output voltage of the DCDC converter is filtered andfed to the inverter to produce the AC output for gridconnection or a load. The inverter is the connecting linkbetween the FC and the grid. A controller adjusts theinverter current for incorporating the FC into the powergrid by matching it with a reference current to source asine current into the grid. The current is controlled by theswitches of the DC/AC inverter. In this case the duty ratioof the pulses fed to the four switches of the bridge inverteris controlled. The controller adjusts the duty ratio on acontinuously.

111. THE SOFC MODELFor the mathematical model [2] that is used in this paperthe stack voltage is given by the equation shown:VFC =W OR I

V o :Open circuit reversible cell potentialE :Std . eversible cell potentialN :Number of cellsF Fara4 s consi .I :Stack currentx 1 Hjldrogen partial pressureThe Nernst and Ohms Law were applied to obtain theseequations (considering Ohmic losses). This is a dynamicmodel the explanation of which can be found in [2].

JY.HE FUZZY LOGIC CONTROLLERSA. DCDC Converter Control for Stand-alone Working

The DCDC converter control block diagram is shown inFig. 2. The DCDC power converter is switched on and off

converter is controlled for load variations.

-r IPWM Generator

Fig. 2 Boost Converter Control LoopThe inputs to the fuzzy controller are the error, e@, andchange in error de@).The controller output or the changein the control signal is the change in the duty ratio. Theduty ratio change is fed to the PWM generator, whichchanges the duty ratio accordingly and adjusts the outputof the converter. The equation (2) represents the error andthe change in error in the mathematical form.

e(W=v,, -v,, (2a)de(k)c (e(k)-e&-l)) (2b)T

e(k) is theerrorde(k) is thechangein errorV, is thereferenc8CvoltageV is thevoltagedt theterminalsf theDCDCconveter

whereT=Samplingime

The three variables of the FLC; the error, change in errorand the change in the control signal have sevenmembership functions (MF) each. The fuzzy partition ofMFs for the three variables are as shown in the Fig. 3. Thefuzzy variables are expressed by linguistic variablespositive large (PL), positive medium (PM), positivesmall (PS), zero (Z), negative small (NS), negativemedium (NM), negative large (NL). The linguisticdenominations for the MFs are same for all the threevariables [3,6,11].

552

7/29/2019 Control of SOFC With Fuzzy Logic

3/5

NL Nh4 NS Z PS PM PL

I 1Fig. 3 Membership FunctionsRate of Change of Error

Table I Rule Base for FLC-1Table I shows the rule base for the FLC. A rule in the rulebase can be expressed in the form: If e is NL and de is NL,then change in duty ratio is NI,. The inference methodused is basic and simple; the commonly used MIN-MAXmethod is implemented. The output MF of each rule isgiven by the minimum (MIN) operator, whereas thecombined fuzzy output is given by the (maximum) MAXoperator. The rule base is the Mac-Vicar Whelan rule base,the general rule base when the characteristics of the systemare not known [ 6 ] .The numbers of rules set for the DC/DCconverter control are forty-nine: seven linguistic variableseach for the MFs of error and change in error (inputs ofthe FLC). The centroid deki f ica t ion method determinesthe crisp output value from the center of gravity of theoutput MF [3,6].

B. DC/AC Inverter Control for Grid-ConnectionThe inverter must source a sinusoidal current into the grid.The working of the inverter control loop is shown in Fig.4. The control loop consists of the FLC which is basedupon the predicted current control method [9]. he inputsto the FLC are the change in the line current of the inverterover a time period of the grid voltage and the referencegrid source voltage (which is assumed to be a puresinusoidal voltage). The FLC based upon the inputs andthe rule base processes the desired control signal to adjustthe current output of the DC/AC inverter. The controlsignal is fed to the gate drives of the bridge inverter in theform of controlled PWM pulses with a varying duty ratioand based upon the control signal the output current of theinverter is altered. Thus, the scheme calculates the dutycycle required for the switches of the inverter that drivesthe line current to the reference value in one switchingperiod.

The source voltage or the utility voltage V, (the gridvoltage) is used as one of the inputs for the FLC, the otherinput is the current change AIover the sampling time T,(where T, s assumed to be smaller than the period of thegrid voltage) obtained from the current generated at theterminals of the DC/AC inverter. The equation (3 )represents the source voltage and the change in current forthe inverter in the mathematical form.

DC/AC Filteronverter Converter(Mtb aCmt-01PW-MGeneratorA I I

v) ;;s=Fig. 4 DC/AC Inverter Control Loop

where T, is the sampling timeL is the inductor of the LC filtert, s the time at the beginningof T,Vi, is the voltage at the terminals of he inverter

The duty ratio for the single-phase inverters can be defmedas a function of source voltage (V J and the change in linecurrent(AI ) as follows [7,8,9]:

wheredk is the duty ratio for switches SIand S2 over oneswitchingperiodThe single phase inverter can be controlled with theswitches S&. The switches SI and S2 sed to shape thewaveform to follow the reference current. While theswitches S3 and S4 are used to correct the polarity of thewaveform [5]. Hence the Vi,, an be described as follows:ynv = k dc (5 )The equation (4) is used for the FLC and the FLCgenerates the duty ratio as the output used as the controlsignal for the PWM of the inverter.The three variables of the FLC; the source voltage, thechange in inverter output current and the control signalhave seven MFs each. The basic fuzzy partition of MFsfor the variables are as shown in the Fig. 5, 6 and 7. Thefuzzy variables are expressed by linguistic variables

553

7/29/2019 Control of SOFC With Fuzzy Logic

4/5

positive large (PL), positive medium (PM), positivesmall (PS), zero (Z), negative small ( N S ) , negativemedium O,negative large (NL), for the sourcevoltage and the current change. The duty ratio is expressedby the fuzzy variables DO to D6.NL NM NS Z PS PM PL

-1 -1

V. SIMULATIONAND RESULTSOutput unregulated voltage is controlled to a regulated DCvoltage as shown in Fig.8 by a FLC when the loadexperiences a step change (this action might be caused bythe voltage variations by the fuel cell itself rather than theload change). The FLC controls the duty ratio of the boostconverter. Since the response time of the boost converter isvery fast, the output voltage recovers steady state quickly.Similarly for a linear variation in the load (or by the actionof the fuel cell) the FLC is fast to act and brings thevoltage back to the desired average value as seen in Fig.8.

Fig.6 Membership Function for Source Voltage andCurrent changeDo D1 D2 D3 D4 D5 D6

0 1

Fig. 6 Membership Function for Duty Ratio[TableIIRule Base for FLC-2

Table I1 shows the rule base for the FLC. The rules are setbased upon the knowledge of the system and the workingof the system. The rule base adjusts the duty ratio for thePWM of the inverter based upon the changes in the inputto the FLC. The number of rules here are 49 based uponthe seven MFs for the source voltage and the change ininverter current respectively. The inference method used isthe MIN-MAX method. Defuzzification is done using thecentroid defuzzification method [3,6].The fuzzy control scheme does not need an accuratemathematical plant model. Therefore, it is applicable to aprocess where the plant model is unknown or ill defined.The fuzzycontrol is also nonlinear and adaptive in natureand offers robust performance under parameter variationsand load disturbances. The fuzzy controller offers goodperformance with relatively fast response time and smallovershoot [3,6].

Fig. 8 Voltage Response (step and linear change in load)

. . , . .. . , . .. . . . .........................................

j i : : : : : : : :. . . . . . . .

. . . . . . . . . . . .

Fig. 9DutyRatio (step and linear change in load)The change in the duty ratio from the original value of 0.5to the final value of 0.59 in case of the step change isshown in Fig. 9. In the case of the linear change the dutyratio changes from 0.5 to 0.67 as in Fig. 9. The FLCresponds to the change in the voltage and adjusts the dutyratio of the switch of the DC-DC converter to a value sothat the boost converter provides the desired averagevoltage.Fig10 shows the controlled sinusoidal current output of theinverter which goes into the utility.

554

7/29/2019 Control of SOFC With Fuzzy Logic

5/5

Fig. 10 Current Output of the InverterVI. CONCLUSION

The FLC is used to control the output voltage of the boostconverter and the output current of the inverter. The outputof the FC generation system has a high voltage fluctuationrate in response to the load variations. The output voltageof the FC changes with the change in load. The FLChandles the changes well and controls the output voltage ofthe boost converter. The output doesnt have anyovershoot. The inverter current is controlled by the FLC tointerface the FC with the grid. The FLC also offers highsystem stability and performance.

VII. ACKNOWLEDGEMENTThis work was sponsored in part by the US DOEEPSCoRWV State Implementation Plan Award.

VIII. REFERENCES[l ] D. J. Hall and R. G. Colclaser, Transient modeliiigand simulation of tubular solid oxide FCs, EnergyConversion, IEEE Transactions on, Volume: 14 Issue: 4,Sep 1999.[ 2 ] Sedghisigarchi, K.; Feliachi, A., Control of grid-connected FC power plant for transient stabilityenhancement. Power Engineering Society WinterMeeting, 2002. IEEE ,Volume: 1 ,27-31Jan. 2002.[3] L. x. Wang, Stable adaptivefizzy Control of non-linear systems, F u z z y Systems, IEEE Transactions on,Volume: 1 Issue: 2 , May 1994.[4] Y. H. Kim and S . S . Kim, Ai?electrical modeling andFuzzy logic control of a FC Generation SJwtem, IEEETrans. on Energy Conversion.

[6] Ronald R. Yager and Dimitar P. Filev, EssentialsofFuzzy Modeling and Control. Wiley-Interscience; 1edition (June 27, 1994).[7] A. M. Hava, T. A. Lipo, and W. L. Erdman, UtilityInterface Issues for Line Connected P W M Voltage SourceConverters: A Comparative Study IEEE Applied PowerElectronics COI$and Exposition,pp .123-132, 1995.[SI R. Wu, S . B.Dewan and G.R. Selmon, A PWM ACto DC Converter with Fixed Switching Frequency, IEEETransactionson Industry Applications, Vol.26,No. 5, p p .880-885,September/October1990.[9] R Wu, S . B. Dewan and G . R Selmon, Analysis of aPWM AC to DC Voltage Source Converter under thePredicted Cment Control with a Fixed SwitchingFrequency, IEEE Transactionson Industry Applications,V01.27,NO .4 , pp . 756-764, July/August 1991.[IO] Yasuhiko Dote and Richard G. Hoft, IntelligentControl Power Electronic Systems, 1998, OxfordUniversity Press.[Ill Abhishek R. Sakhare, A. Davari, A. Feliachi,Control of Stand Alone and Grid Connected Solid OxideFuel Cell using Fuzzy Logic, System Theory, 2003.Proceedings of the 35th Southeastern Symposium on, 16-18March 2003 Pages: 473 - 476.

[5]N. Mohan, T. M. Undeland and W. P. Robbins. PowerElectronics Converters, Application and Design. 198 9,John Wiley and sons Inc.

555