This paper reviews some of the designs of an airflow controller which in turn maximizes the net power output in a PolymerElectrolyte Membrane (PEM) fuel cell system. Experimental results show that theperformance and the power output of fuel cell depends upon gas flow rate, operatingand fuel stack temperature, humidity and pressure. However net power productionof the fuel cell depends heavily on the oxygen excess ratio in the cathode.
Atime-varying and a complex non-linear dynamic system present many challenges todesign a oxygen and hydrogen excess ratio to an optimum value that maximizesnet power output over a wide range of operating conditions. Performanceimprovement can be done using two modelling approaches: black box modelling anddetailed dynamic modelling. A fuzzy-PID (Proportional IntegratorDifferentiator) hybrid controller and a DC-DC controller have been designed andtested earlier.
The results show that these control strategies can be used toimprove the system’s performance significantly. The control system performancewas determined by comparing the simulated results and flow rate obtained usingfuzzy logic controller. External voltage was kept constant and fuzzy-PID showsbetter response than a manual controller. 1. Introduction Fuel celltechnologies have been identified as a potential solution in solving theproblems of meeting increasing renewable energy demands. Fuel cells transformchemical energy into electricity. Hence, it is an environment friendly energyresource.
Mitigation of environmental pollution and providing energy shortagehas led to fuel cells being considered as elements of alternative energysystem; capitalizing on high efficiencies and low emissions. Fuel cells canprovide large amounts of current hence power, with the thermodynamicrequirement being the appropriate flow of reactants. There are several types offuel cells, each using a different chemistry.
PEMFC is commonly used to powervehicles. Its performance and efficiency still needed to be improved, and theissues of cost, reliability, it is necessary to learn more about the mechanism that causes the performance loss,such as non-uniform concentration, current density distributions, high ionicresistance due to dry membrane, or high diffusive resistance due to flooding onthe cathode. Flow field will require optimum design to achieve high reliabilityand performance. Theperformance of fuel cell and vehicle applications they are embedded into dependon a delicate balance of the incorrect temperature, humidity, reactantpressure, purity and flow rate. Thus, its clear that the performance of thesystem depends upon the controller in place.
For practical applications anaccurate controller model is required. Study of control oriented design andpower management is required for the understanding of system behavior, and forsubsequent design and analysis of model based control system 3.Fuzzy control is one of the useful controltechniques for uncertain and ill-defined non-linear systems. Fuzzy logic basedcontroller has high performance and other numerous advantage. The mostimportant aspect being, the cost of the controller design and implementation. Theproposed hydrogen flow controller modify control active current to the loadchange and setpoint current for controlling to the load 1. A properair-supply system is important for a PEMFC 4. Excess oxygen content incathode may lead to power loss and decreases net power output/efficiency.
Also,if there is oxygen starvation in the cathode it would lead to deterioration.The electrode half reactions are as follows: Anode: H2 (a) 2H+ + 2e- Cathode: 2H+ + 2e- H2(c) Overall: H2 (a) H2 (c) Pukrushpan5 previously designed a model-based controller that lacked experimentalsupport whereas Feroldi 6 proposed a coordinative control strategy thatregulates air flow replenishment and air exiting the system simultaneously.This shows both oxygen excess ratio and cell voltage are under control.
Pukrushpan7, 8 presented a system with transient behavior of air supply system whichwas a single input single output feedback controller. This feedback controllerprovided a stoichiometric air ratio. A semi-empirical model was used. A complex, non-linear and time-varying airflow system can be approximated using principles like mass conservation,theories of thermodynamics, fluid dynamics and heat transfer. Golbert and Lewin9 have already developed a time varying model with its state space matrix,did its controllability analysis and then presented an adaptive nonlinearcontroller.
Stefanopoulou and Suh 10 modelled and analyzed a DC-DC converterwith a compressor. This design was done considering the dynamics of fuel cellsystem. This converter is employed in a model-based control techniques to tunetwo separate controller for the compressor and the converter. They alsodemonstrate the lack of communication and coordination between the twocontrollers. This leads to a tradeoff in achieving the power output objectives. A PEMFC includes a number of components thatare related to controlling the flow, temperature, pressure and humidity. ADC-DC converter transforms unregulated DC power to a regulated one as it isfocused on soft voltage sources that a fuel cell involves. This can also beused to filter current from the fuel cell which could lead to degradation orfuel cell failure.
Though limiting the current drawn from the fuel cellenhances the performance of the fuel cell but it degrades the voltageregulation performance of the DC-DC converter. A physics based model wasdeveloped in which hydrogen dynamics, humidity and temperature dynamics areneglected as they are slower than the air flow dynamics. The importance of airsupply arise due to its parasitic loss. 2. Fuel Cell System Models This section describes the fuel cellsystem model developed by various people in their experiment and analysis. 2.1 Extremum seeking controller This model 5, 8 is used for the analyzing the impact of oxygen supply,temperature, and water content on power output. The model described manifoldfilling dynamics, compressor inertia and partial pressures.
The components thatthe system has are fuel cell stack, a compressor, manifolds (cathode andanode), on air cooler and a humidifier. There are nine state variables whosegoverning equations can be classed into both manifolds and the compressor.Conservation of mass and energy laws make the governing equations in the manifoldswhereas inertial dynamics of motor and compressor make up for the governingequations in compressor. The state equations formed in the cathode are asfollows: Using conservation of mass of water at cathodeside. The state equations formed inanode manifold are as follows: The compressor supplies the airto the cathode that consumes the energy from thr motor which is generated fromthe fuel cell.
Mass flow rate of the air is determined by the inertialdynamics. 2.2 Fuzzy control of Fuel Cells The components of a typical fuel cellinclude gas diffusion layer, fuel stack, catalyst layer, membrane etc. Theredox reaction involved in the process is as follows: Oxidation H2 2H+ + 2e- Reduction O2 + 4e- + 4H+ 2H2O Electrical energy is obtained from the systemwhen a significant amount of current is drawn, but the actual cell potential isdecreased. There are several irreversible losses present in the actual fuelcell system. These losses are called overpotential or polarization. Theseoriginate from three sources namely activation, ohmic and concentration.Activation dominant at low current density, ohmic varies directly with currentand increases with it over the range of current density.
Concentrationoverpotential lies at the higher range of current density where the cellvoltage is at a minimum. Aclosed loop feedback control system is used for actuating error signal. Thedifference between input and output signal is termed as error. The feedback isfed to the controller basically containing a proportional or an integral or aderivative or a combination of pair of controllers.
Thomya and Khunatornproposed a fuzzy logic controller that controls the active current from the fuelcell by controlling the flow rate of hydrogen. A typical fuzzy controllerconsists of five different steps involving:· Defining the input and output variables· Designing the fuzzy control loop· Fuzzification · Inference· Defuzzification Generally,the inputs are error e(k) and change in error ce(k). The controller outputis called as ‘Read Current’. The relation between these variables is asfollows:e(k) = set point current – read currentwhere, read current is the hydrogen flow ratefrom the feedback loop which is proportional to terminal load and ?Ukis referred to as change of duty ratio by fuzzy controller.Ce(k) = e(k) – e(k-1) Fuzzyvariables are expressed as PB (positive big), ZO (zero) and NB (negative big).Rules for writing the fuzzy variables are in table 1. For example:· If e(k) = PM and ce(k) = ZO, then u(k) will bePM· If e(k) = NB and ce(k) = PM, then u(k) will beNB Thedefuzzification method gives the output from the center of gravity of theoutput.
A membership function is set up to test controller design. 2.3 Adaptive control strategy Thefuel cell used in the experimentation by Zhang, Liu, Yu and Ouyang is amembrane-humidifying PEMFC with an active area of 150 cm2, 24 Nafion112 cells in series and 1kW power output operating at a temperature of 65ºC anda pressure of 30kPa. There are majorly four control subsystem that are usednow-a-days, which are air/hydrogen supply, the humidity of reactants, stacktemperature and power output. The humidifying membrane allows the gas to havesufficient contact with the saturated vapor.
This makes the humidifier thatmuch more important. A cooler is added to remove excessive heat during thereaction. A proportional controller tracks with the pressure in the cathode andanode. Some assumptions are made before modelling the air-supply subsystem.These are (1) gases obey the ideal gas laws that is due to operation of thesystem at the low pressure (2) coolant carry away the generated heat due torelatively small volume of the stack. (3) properties of the gas remain same atthe inlet and outlet of the cathode (4) gas in cathode is fully humidified (5)extra water carried out by exhaust gas (6) flow channel and diffusion layerlumped as a single volume (7) gas pressure in the air tank kept constant. Agas flow regulator was designed by them that replenishes air into the stackcathode.
The input and output variables for the air regulator varies withtemperature, electromagnetism, humidity and gas pressure. Non-linearity isobvious with the system and has an influence on the dynamics of the airflow andcan be assumed as time varying. This has a very complex design method thatinvolves a lot of solving of partial differential equations. To optimize theperformance of the air supply sub system an adaptive controller is used. Thiscontroller helps to alleviate the dynamic variation that arise from varyingtime. This involves placement of closed loop poles keeping the desired resultsin mind. The operation process involves some goals that needs to be achieved:(1) quantification of air flow (2) definition of ideal kinetics of cathode air(3) designing a controller in which the airflow responds to the commandsaccurately and dynamically.
Toreduce the complexity of the design of the digital controller, the model of thegas flow regulator was designed using a discrete time equation. The air flow isdirectly related to power output. Dynamic air flow can be attained using twomethods as done by two groups, Gelfi and Suh-Stefanopoulou, one used thecapabilities of a compressor of a fuel cell system using a blower. The otherused a dynamic coupling between a compressor and fuel cell in which the motorwas driven by fuel stack. This group involved Zhang, Liu, Yu and Ouyang, andthey did three tests of the design of the air flow system designed by them.They checked the step response in all the tests they did and found out thestability of the system after drawing the results from the graph. 2.
4 DC-DC Convertors Suh andStefanopoulou used a fuel cell with an active area of 280 cm2 with381 cells and a power output of 75kW. This is applicable for automotive andhousehold purposes. They proposed a fuel cell reactant supply system thatconsists of a supply manifold, hydrogen pressure control, an air flow controland a humidifier cum temperature controller. A PI controller was developed forthe system. They investigated how a DC-DC controller can be used to filter fastloads (current). It can be used to filter out fast load changes.
Model wasdesigned, and the dynamic states were formed according to the flow rate incathode. They considered nonlinear static functions as well to achieve a real-timesituation. The DC-DC converter transforms the fuel stack power to the outputrequirements that connects the device. They used a boost converter as it can beused in PEMFC applications. 3. Results and Conclusions 3.1 Extremum seeking approach The results of a simple staticfeedforward (sFF) controller and a static feedforward with a PI (sFF + PI)controller were compared to evaluate the performance of the extremum seekingcontroller proposed by the researcher. The sFF is used to measure thedisturbance input and compute the compressor voltage output.
This takes a formof a typical feedforward compensator which is equal to the inverse of the DCgain. A PI + sFF is used to measure the stack current. It also adjusts themotor voltage about the optimal value in the feedforward loop.
This approachhas a drawback i.e. the optimal value identified in the lookup tablecorresponds to steady state values only.
Also, the temperature of the fuelstack must be 353K with a water content of 14. Only if these conditions are metthe optimality will be guaranteed. Simulation results indicate that the optimaloperating point can account for the time varying and non-linear parameters suchas stack temperature and water content using the compressor voltage online. Thisconstraint approach allowed them to explicitly set constraints on reactantoxygen and membrane hydration.
That enabled them to select more aggressiveextremum seeking parameters that eventually increased convergence speed with asatisfactory overshoot. The conditions were defined, and the outcome of theexperiment was that the optimal values were found by the steady state analysisand the final values were 22.5 kW net power with an oxygen excess ratio of 2.
64and the motor voltage corresponding to these values is 151 V. The net poweroutput from the extremum seeking approach was more than either the sFF or sFF +PI at steady state. This indicates that the ES algorithm produces more netpower despite the variations in flow rates, temperature in fuel cell stack andthe non-linearity, the ES produces superior output when compared to both sFFand sFF + PI.
3.2 Fuzzy Logic Controller 3.3 Adaptive control strategy PEMFCcould become a vital alternative to internal combustion engine if an enoughpower output and the need of load is met. To achieve this all the subsystems ofthe fuel cell must function effectively and be dynamically co-operative witheach other. Non-linearity and time varying characteristics are a hinderanceduring thus operation. An adaptive control strategy has been proposed by theresearcher to run the air-supply subsystem for air replenishment in cathode.
Some of its findings show that the designed adaptive control strategy couldlead to good system performance and stability. The performance of the systembecame better and net power output of the system was increased. 3.
4 DC-DC converter Alow order fuel cell system model was developed with a DC-DC converter using allthe physical principles mentioned by the researcher along with stackpolarization. They captured the inertial dynamics of compressor, partialpressure, and manifold filling dynamics. Relation between air supply and infuel cell and voltage regulation was developed. A model based controller wasdesigned to regulate the oxygen excess ratio and the bus voltage usingdifferent methods and architecture. They compared the design on the performanceof the system at different operating conditions (medium to high load range).
Itwas verified that the linear decentralized controller device can achieve betterresults over a wide range of net power.