Fuel Cell Emulator: Design and Implementation, Study Guides, Projects, Research of Electrical and Electronics Engineering

Download Fuel Cell Emulator: Design and Implementation and more Study Guides, Projects, Research Electrical and Electronics Engineering in PDF only on Docsity! TEST FUEL CELL CIRCUIT by Leann Cerven and Nate Schweighart ECE 345 Julio Urbina 5/1/00 Project #51 ABSTRACT As more attention is given to fuel cells and their possible widespread implementation in household applications, more attention is also being given to products that interact with fuel cells. Our fuel cell emulator has the same output characteristics of a real fuel cell, but is a much more feasible option for testing fuel cell products in their design stages. This emulator is portable, doesn’t require a fuel source, can be adjusted for different output voltages, and responds similarly to a fuel cell in terms of its load voltage response to load size and fuel flow changes. Tested values for load values include 0-60 V and 600 W, and maximum load values are estimated at 48 V and 1500 W. ii 2. DESIGN PROCEDURE 2.1 Determining Fuel Cell Curves and Equations The first step in designing a fuel cell emulator is to determine the actual output values of a fuel cell. This means large amounts of research need to be done in order to properly know the behavior of a fuel cell. For example, Fig. 2. is one of the IV curve behavior characteristics we were able to find on a fuel cell. PEM Output Voltage vs. Current for Different Fuel Flow Rates 0 10 20 30 40 50 60 70 0 10 20 30 40 50 Output Current (A) O u tp u t V o lt ag e (V ) 100% Flow 50% Flow 25% Flow Fig. 2. Fuel cell output voltage versus current curves [1]. It needs to be stated that Fig. 2 is the result of a PSPICE circuit model and does not take into effect the 15% preload due to internal operations of a fuel cell [2]. By analyzing this information and these IV curves we were able to decide on appropriate fuel cell behavior. It is obvious that the IV curves of a fuel cell are not linear. It is possible though to break them up into many separate linear regions. Figure 2 also shows only three fuel flow curves when in actuality there are values for every fuel flow between them. In order to approximate these nonlinear curves, they were divided into three separate linear curves. To determine the equations for these three linear portions of the curve, we used the line equation: y=m*x+b This equation can also be represented as: Vout=-R*I +Vin (1) As you can see a linearized IV curve has a slope of (–R). We use this fact to set different resistive values to achieve the slopes we want. As said before, the non-linear IV curve is divided into three different linear portions. By finding values on each of these linear segments, the points can be plugged 2 into the linear equation to determine the equations of the lines. Thus when we have the line equation for a certain segment, like Vout=-1*I +57.25 for example, we know which resistor values and power supply values are needed for the circuit to follow the IV curve. This approach is the most desirable since it closely approximates the actual curves but yet keeps the circuit linear in nature. This method also allows for freedom in shifting the curves. The curve equations can be recalculated depending on the desired output wattage and voltage and fuel flow. 2.2 Design Options for the Circuit The basic circuit for the emulator is shown below in Fig. 3. Fig. 3. Fuel cell emulator circuit The dc battery in Fig. 3 is set to the value of y-intercept found in the equations described earlier. Therefore, we use the parallel resistor combination to manipulate the output values. The electronic load is modeled as the variable resistor at the top of the parallel combination, and the other branch resistors are meant to dissipate additional power when the power rating of the electronic load is exceeded. This means that there needs to be the ability to independently switch on and off each branch of the parallel combination. There are several ways to achieve this switching requirement—including a relay, n-type MOSFET, and p-type MOSFET. All of these options would need to be controlled by a ttl signal. This ttl signal would turn on a diode that would allow the current to flow through the circuit and produce the turn-on voltages needed for the relay or MOSFETs. Since the switch only needs to be switched on or off, and not at a high frequency, a relay switch could work. However, a high power relay would need a 12 V source. An additional power source would not be very efficient and add cost and size. Another point against the relay is the bulkiness and awkwardness in integrating it into our circuit. MOSFETs seem to offer the best solution. They are small, efficient, able to handle high power levels, and easily integrated into circuits. Their high frequency switching capabilities go beyond our need, but they are still a cheaper and more feasible option than relays. The next decision was whether to use n- type or p-type MOSFETs. To get the required turn-on voltage for the n-type MOSFET there needs to be either independent power sources (Fig. 4) or an optocoupler combining the ttl input and the power Computer Input Pin 22 Computer Input Pin 21 1 3 2 MUR3040PT MUR3040PT HP 6060B R_var Load Resistor R_var 1 3 2 HP 6674A 60Vdc 3 supply voltage. A p-type MOSFET would simply require a resistor combination in series with the ttl controlled diode described earlier (Fig. 5). Figure 4. N-type MOSFET with independent power sources. Figure 5. P-type MOSFET with Vgs voltage divider. Next, the resistor values that connect to the switches had to be determined. Each fuel flow level has its own voltage versus current curve, and the slope of that curve is equal to the resistance of the parallel resistor combination. The value of the equivalent resistance can be found by: Req = Fuel Flow Slope = (100% fuel flow slope)*1/(fuel flow) (2) The maximum output values determine the 100% fuel flow slope values. Since there is a proportional relationship between fuel flow and output, you can determine the slope of each curve by using Eq. (2). The inverse of the fuel flow is used so that the slopes of the lesser fuel flows will be greater and therefore decay faster as the model indicates (Fig. 2). The electronic load in this project can handle up to 300 W at resistance levels from 0.1 to 1000 . The values of the resistors in the parallel branches had to be higher than the smallest slope value (since the equivalent resistance of a parallel combination is smaller than its components). There were several options for the value of these resistors, and the only crucial consideration was that the resistors had the power rating to support the current that might be put through them at maximum operating conditions. We chose to make the parallel branches parallel combinations themselves in order to obtain small resistor values that could sustain high power levels. Since there were only two analog ports on the S G MTP30P06V 1.4 0.7 2N3055A 510 S HP 6674A 60Vdc MTP30P06V G HP 6060B R_var Computer Input Pin 21 2N3055A D 1100510 10k 1100 Computer Input Pin 22 10k D Load Resistor R_var 12Vdc 10k R_var V2 60Vdc D 470 10k G TTL S 12Vdc TTL 470 4 3. DESIGN DETAILS 3.1 Tested Fuel Cell Curves Since there was not enough time to design and build an adequate load to test the fuel cell emulator at 1500 W, the project was scaled down to 600 W so that the full fuel flow range could be tested. The voltage versus current curves were calculated to go from 0-60V and 0-13 A. Figure 7 shows the calculated voltage versus current curves that we simulated and whose slope values we used for the equivalent resistance. The calculation of these curves is described below. Fig. 7. Calculated and tested fuel cell I-V curves To calculate the 100% fuel flow curve we used the data points in Table 1. Table 1. ESTIMATED DATA POINTS FOR 100% FUEL FLOW I-V CURVE Current (A) Voltage (V) 0 60 0.75 57 6 54 13 47 These points are the intersection points between the three linear portions of the I-V curves. With these points the equations in Table 2 were determined using Eq. (1). Table 2. 100% FUEL FLOW I-V CURVE EQUATIONS Current Range (A) Equations (5)-(7) 0-0.75 -6*I + 60 0.75-6 -0.5454545*I + 57.24 6-13 -I + 60 Greater than 13 0 Tested Fuel Cell I-V Curves 0 10 20 30 40 50 60 70 0 5 10 15 Current (A) V o lt ag e (V ) 100% Fuel Flow 75% Fuel Flow 50% Fuel Flow 25% Fuel Flow 7 The values for the other fuel flows can be determined by simply multiplying the current range and the equation slopes by the inverse of the fuel flow. Appendix 1 shows all the data points calculated for these curves. 3.2 Final Circuit Diagram As described in Section 2.2, the p-type MOSFET circuit option was the best choice. From Eq. (2) and Table 2, the range of equivalent resistance values is 0.545454 to 24 (25% is the lowest that the fuel cell will operate at). With the power resistors that were available, we made a branch of resistors equal to 0.7 and the second one equal to 1.4. Figure 8 below shows the final circuit complete with part numbers and values, and Figure 9 shows the parallel resistor bank combinations that made up two of the three branches in the equivalent resistance parallel combination. Figure 9. Figure 8. Final circuit schematic Parallel branch resistors 3.3 Program Flowchart In order to correctly produce the IV curves that would adequately match our calculated IV curves a program was needed to take measurements, make calculations and control the lab equipment continuously. Agilent VEE was the program used to accomplish this. Agilent VEE is designed so that the program can obtain data from the lab equipment, such as the Fluke multi-meter, and then use this data to determine the outputs to be sent to the power supply and variable load. S G MTP30P06V 1.4 0.7 2N3055A 510 S HP 6674A 60Vdc MTP30P06V G HP 6060B R_var Computer Input Pin 21 2N3055A D 1100510 10k 1100 Computer Input Pin 22 10k D Load Resistor R_var 1.5 Resistor Banks 1 and 2 5.6 1.5 1.5 6.2 1.5 7.1 8 Figure 5 is a flow chart that shows the different states our program traverses to achieve the needed tasks. Fig. 5. Program flow chart Sense current Adjusting to 48V? Set fuel flow goal to calculated value Set fuel flow goal to knob value Compare fuel goal with present value Increment or decrement fuel flow Determine which curve by using current value Calculate power supply voltage Calculate needed resistance Determine if extra resistors need to be turned on Output 5V to specific analog out Output electronic load resistance and power supply voltage Stop button pressed? Set equipment to zero Display circuit values Yes No Same Different Yes No No Yes 9 4. DESIGN VERIFICATION 4.1 Test Procedure The load that was used to test this emulator was comprised of ten parallel resistor boxes that each had ten 500 resistors in parallel. Each switch could be turned on or off. With all the switches on, the load was approximately 5. The load voltage, current, equivalent resistance, electronic load resistance, and calculated load voltage were all monitored on the Agilent VEE interface to determine if the emulator was acting correctly. There was also a notification of which resistor banks were on. The calculated load voltage allowed us to compare the value of the load voltage to a theoretical value while the program was running. This gave us the opportunity to double check the load voltage quickly at any time during the emulator’s operation. When we believed the emulator was behaving in the manner we had designed, we made our own I-V graph. Since Agilent VEE did not allow us to take a real time voltage versus current plot, the load voltage and current data were taken by hand. We took these values for 100%, 75%, 50%, and 25% fuel flows for loads ranging from infinity (open) to whenever the current was exceeded. The recorded data can be found in Appendix 2. 4.2 Emulator Load I-V Curves Figure 10 shows the result of our experiment. As you can see, there are a few minor discontinuities, but the graph verifies the successful implementation of a 600 W fuel cell emulator. Fig. 10. Emulator load voltage versus current curves. 4.3 Maximum Output Values To check if our circuit would truly handle the 1500 W as it was originally designed to do, we added three 27 resistors and 4 resistor boxes to the load. This dropped the load resistance down to 3.6. With this load we were able to achieve 17.5 A at 43.34 V, or 760 W with little strain on the circuit. Emulator IV Curves 0 10 20 30 40 50 60 70 0 5 10 15 Current (Amps) V o lt ag e (V o lt s) 100% Fuel Flow 75% Fuel Flow 50% Fuel Flow 25% Fuel Flow 12 5. COST ANALYSIS Table 3 is a breakdown of the costs involved in producing our emulator. Table 3. Parts and Labor Costs. Quantity and Part Cost (in dollars) 1 HP 6634B DC Power Supply $4496 1 HP 6060B Electronic Load $2363 1 PC Computer ~$800 1 Agilent HP VEE program license $2600 1 Fluke Multi-Meter $750 7 High Power Resistors ~$25 6 Low Power Resistors ~$2 2 MTP30P06V p-type MOSFET $8 2 2N3055A BJT $1.60 2 Engineering Labor 2*120hours*$25/hour=$6000 Total $17,045.60 13 6. CONCLUSIONS We have successfully designed and built a prototype fuel cell emulator that mimics a fuel cell’s output characteristics. This current design models a 600 W output system, but it can be easily modified to fit the user’s needs. With just a few changes to the equations (as explained earlier), the desired output can be adjusted anywhere from 0 to 1500 W. The major reason that we did not achieve our 1500W goal was because of a lack of sufficient load to handle such large amounts of power at such a low resistance. Another problem that could be fixed is discontinuous IV curves as they move from one part of the curve to the next. This problem stems from the inability of the design to account for circuit losses. This can be fixed by allowing to circuit to correct for these losses by calculating exactly where on the IV curve it should be at and then correct its values until it reaches the correct voltage. But in spite of these small setbacks our design was a large success. We overcame many design obstacles and were able to produce an output that behaved as we intended. 14