Under these conditions the integrated system was able to produce cold and dehumified air that is in the safe range for server racks. In this case the return air from server is considered for first stage cooling air which has lower relative humidity and higher potential for evaporative cooling.However, the system is able to provide only 97CFM of air flow under these conditions, which is less than the maximum air flow demand for each server rack. This suggests that the current integrated SOFC-LDD system cannot continuously provide all the power and cooling required for the server rack if it is operated continuously at 12kW. However, both the server demand dynamics and day-night weather variations, could lead to dynamic conditions that with some storage of the concentrated solution could be well-matched to the server demands for power and for cooling and dehumidification. Next step will consider these dynamics for several real-world weather and server power. For the second weather condition with 35˚C and 45%RH the system can produce 96CFM cold and dry air for each server rack. In this case the outside air is considered for the first stage cooling air. Please note that the SOFC system is running based upon design operating conditions for the system. Further investigation is needed to evaluate the SOFC system available heat under different operating conditions that could produce higher quality exhaust heat likely at the expense of electrical efficiency. The results are presented for two scenarios; in first scenario, liquid desiccant provides dehumidified air for one single server rack for the entire year,indoor weed growing accessories the fuel cell runs at full load, and the dehumidifier is sized based on server rack, and the regenerator is sized based on FC exhaust.
In second scenario, liquid desiccant provides dehumidified air for two rows , Fuel cell runs at 70% load, dehumidifier is sized based on cooling required for two rows, and regenerator is sized based on equivalent of two rows fuel cell exhaust. The reason for row level fuel cells running at 70% and in rack level running at 100% is that when fuel cell provides power for only one rack, it is more likely that the single server rack runs at full load, however, when several fuel cells provide power for 20 racks, the load of serves will be balanced to the data center overall utilization .Figure 64 to Figure 69 show dehumidifier results for Illinois. In dehumidifier, two air streams are send to the system, one is the supply air which is a mixture of outside air and return air and the second stream is the cooling air that is send to the dehumidifier to keep the temperature constant by evaporative cooling. Figure 64 and Figure 65 show the supply air and cooling air inlet and outlet temperature. The temperature of the streams stays almost the same even though the dehumidification is an exothermic process, which is due to the indirect evaporation of water to the second air stream. Figure 66 shows the amount of return air. For Illinois during May to September dehumidification is needed. Other time of the year the humidity of supply air is within the acceptable range. As the outside air humidity is higher the use of return air increases due to its lower humidity compared to outdoor air. However, the return air percentages depend both on temperature and humidity of outdoor air and return air to keep the supply air humidity within the acceptable range for servers. Figure 67 shows the humidity ratio of inlet and outlet supply air. The SOFC systems installed at NFCRC are BlueGEN cogeneration systems. BlueGEN is a commercially available CHP unit, built and sold by SOLIDpower. Figure 87 shows the eight BlueGEN units as installed in the NFCRC laboratory.
Operating on natural gas, each unit can produce power modulated from 500We to 2kWe ; however, it achieves its highest net electrical efficiency of 62% at a 1.5kWe output. The BlueGEN systems are typically operated and controlled remotely by SOLIDpower. However, through an online human machine interface, the power output profile and as a result fuel utilization are controllable by the user. To be able to install the BlueGEN we need to provide six connections: fuel , electrical, flue gas, internet, water, and drain connections. BlueGEN requires an uninterrupted supply of natural gas, and it only operates on the second family of gases . Gas supply pressure to the BlueGEN appliance must be 9–20mbar so an Elster Jeavons gas pressure regulator is used to avoid fluctuations due to variations in the main gas supply pressure.The BlueGEN systems require an electrical connection to successfully operate during startup and to export the electricity produced during normal operation. For electrical connection the grid availability is 208VAC 3 phase, 60 hertz. Each fuel cell needs 120VAC 1 phase, 60 hertz. 3 transformers are used where two of them provide electrical connection needed for six BlueGENs and one of them runs two BlueGENs. The main water supply pressure must be at least 1bar and no more than 10bar . During operation, the appliance can consume up to 30litres of water per day, depending on heat and condensate recovery. The BlueGEN appliance has an internal water storage tank for the process water and this tank will fill from the main water supply intermittently. The water consumption rate may be up to 1 liter per minute when this tank is filling. Operation of water needs to be uninterrupted. Wastewater is ejected from the BlueGEN appliance under pressure . The main water supply connection, drain, and overflow rejection connection configuration can be seen in Figure 88. Water is distributed in a manifold to each fuel cell line and each fuel cell has its own valve for water shut down.
For water rejection line, overflow and wastewater will be collected in a tank through a sloped pipe and then will be pumped to the drain. Figure 89 and Figure 90 show the pump used for fuel cell water rejection connection, manifold, and valves for fuel cell supply water, respectively. Looking at SOFC dynamic data, it is observed that BlueGEN electrical efficiency is the highest, roughly 64%, at full load. BlueGEN efficiency is the lowest at 30% at 300W of output . The BlueGEN system is recommended to run between 1500W and 500W. At the lowest recommended power, the efficiency is roughly 43%. As expected, at loads lower than the nominal power current density drops, and because of lower polarization, voltage increases. Experimental results show that BlueGEN control system is designed to keep air flow constant and that it changes the fuel utilization and fuel flow proportional to the output power. BlueGEN stack temperature difference is designed to stay at 50℃ at full load. The experimental results show that by reducing the load, the temperature difference decreases. The exit temperature drops below the inlet temperature at 700W. At low current density the endothermicity of reformation reaction is higher that exothermicity of electrochemical reaction which causes temperature drop along the cell. Note that while electrochemical efficiency increases at part-load, fuel utilization is lowered to maintain the overall thermal balance so that the overall system efficiency drops at part-load conditions. At these part-load conditions the fuel cell stack operates at overall endothermic conditions, which must be matched by heat from the thermal oxidizer that converts the unspent fuel at the lower fuel utilization conditions.In another dynamic test,rolling benches the BlueGEN system was tested through the following profile shown in Figure 95 in 24h for 42 days. During this cycle, the power decreases between 1500W to 500W with 100W increments. In each power step the system is given 1 hour to reach to steady state and then it has 15 minutes to ramp down to the next power level. At this writing the BlueGEN system has been running for more than 6000hr from the time of installation at the NFCRC. Figure 104 shows the power profile of BlueGEN system over this 6000h operating period. After 4000 hours of steady-state operation the system runs the first dynamic load Figure 93, then after a couple of days of steady operation it goes through the 24h dynamic load cycles for 1000hr followed by the one-week dynamic cycle. Figure 105 shows the voltage change over the 600hr operation. The system operation is divided to three region and Table 22 shows the degradation results for each of the regions of operation. During the first 1000hr of operation as the system is new the degradation rate is twice that of the regular steady-state operation. Also, the system degrades twice as fast as steady-state operation while operating under highly dynamic operating conditions.
As the system degrades, the fuel consumption increases and as a result the overall system efficiency drops. For cell level tests, experiments were carried out using a commercial anode supported solid oxide cell. The cell anode side consists of two layers, a thick Ni-YSZ anode support acting as the mechanical support, current collector and gas diffusion layer, and the other is the functional layer with a more dense structure near the electrode-electrolyte interface . The electrolyte consists of 8% mol Y2O3-ZrO2 on which Gadolinium Doped Ceria oxide barrier layer is deposited to prevent the formation of insulating SrZrO3. The cathode active layer was made up oflanthanum strontium cobalt ferric oxide in contact with the cathode current collector of similar thermal expansion. The cell had an active area of 12.6cm2 . The experimental setup of the single cell used for each of the three 1000hr tests is shown in Figure 107. The furnace keeps the temperature at constant. The JV scan and EIS measurements were performed with a Zahner Zennium impedence analyzer. Zanher positive terminal connects to the cathode and the negative terminal to anode. The voltage response is also measured via the same system. During EIS measurements constant loads of 0.1, 0.5, and 0.9A/cm2 were applied to the cell. For applying constant loas of 0.5A/cm2 during the 1000hr polarization a load box is connected to the setup. The fuel cell air and fuels are supplied by Voegtlin Red-y smart series mass flow controllers. The purity level of hydrogen, nitrogen and ammonia used in the experiments are 99.999%, 99.999% and 99.99% respectively. Nitrogen and hydrogen are stored as compressed gases while ammonia is stored in liquid form in commercial cylinders.The single-cells were tested with three different fuel composition as shown in Each test runs on a pristine single cell SOFC. Cell 1 on pure H2, cell 2 on H2-N2 case, and cell 3 on NH3 directly, which includes the internal decomposition reaction . A nominal operating temperature of 750°C is chosen for the tests. After mounting each cell, each SOFC cell is reduced in H2 and N2 atmosphere. Each cell runs for 1000hr at 0.5A/cm2 to compare long term effect of different fuel on SOFC degradation. Dynamic current density – Voltage scans were performed every 100hr from Open Circuit Voltage to 700mV at 5mV/s with 1mV resolution. The JV scans are followed by three EIS measurements at 0.1, 0.5, and 0.9A/cm2 and 200mA perturbation signal for a frequency range of 1KHz to 200KHz. The quality of EIS spectra was verified using the Kramer Kronig’s test. The overall test duration of test for each cell was 47 days. Table 23 to compare the degradation effects between in-situ and ex-situ ammonia reforming with that of pure hydrogen for a duration of 1000h. Each test runs on a pristine single cell SOFC. Cell 1 on pure H2, cell 2 on H2-N2 case, and cell 3 on NH3 directly, which includes the internal decomposition reaction . A nominal operating temperature of 750°C is chosen for the tests. After mounting each cell, each SOFC cell is reduced in H2 and N2 atmosphere. Each cell runs for 1000hr at 0.5A/cm2 to compare long term effect of different fuel on SOFC degradation. Dynamic current density – Voltage scans were performed every 100hr from Open Circuit Voltage to 700mV at 5mV/s with 1mV resolution. The JV scans are followed by three EIS measurements at 0.1, 0.5, and 0.9A/cm2 and 200mA perturbation signal for a frequency range of 1KHz to 200KHz. The quality of EIS spectra was verified using the Kramer Kronig’s test.