Measurement of Hydrogen Mixing Process by High Response Hydrogen Sensor Toshio Mogi, Ritsu Dobashi Graduate School of Engineering, The University of Tokyo Takashi Nohmi HySafeNohmi I’d like to talk about measurement of hydrogen mixing process by high response hydrogen sensor. 7th International Conference on Hydrogen Safety September 13, 2017, Hamburg, Germany

Background In Japan, we are on the way of 2020 Tokyo Olympic Games and under the words of clean energy. Hydrogen energy applications are prospectively developed: in Fuel Cell vehicle (FCV), cogeneration type of Fuel Cell, combustion of biogas in electric power generation plant etc. Hydrogen Low ignition energy (0.019mJ) Extensive flammable region (4-75vol%) Easy leakage and high diffusivity Properties on safety We are on the way of 2020 Tokyo Olympic Games and under the words of clean energy. Hydrogen energy applications are prospectively developed: in Fuel Cell vehicle (FCV), cogeneration type of Fuel Cell, combustion of biogas in electric power generation plant etc. However, hydrogen has some serious properties in comparison with other combustible gas, For example, ignition energy is 0.019mJ at stoichiometric mixture. It’s flammability range in air is 4 to 75 % by volume of hydrogen. And hydrogen is easy leakage and high diffusivity. Leaked hydrogen or unintended concentration of hydrogen as a cloud may escape and lead to an accident, so the leak should be detected as soon as possible for the safety sake. Leaked hydrogen or unintended concentration of hydrogen as a cloud may escape and lead to an accident, so the leak should be detected as soon as possible for the safety sake. 2

Background GLOBAL REGISTRY Global technical regulation HYDROGEN POWERED VEHICLE 100mm 5.2.2.3 Fuel cell discharge system: [At vehicle exhaust system’s point of discharge, the hydrogen concentration level shall not exceed 4% average by volume during any moving three-second time interval during normal operation including start-up and shutdown.] This is one of the very important examples. According to the Global technical regulation on hydrogen fuel cell vehicles (FCV), fuel cell discharge system at the vehicle exhaust system`s point of discharge, the hydrogen concentration level shall not exceed 4 % average by volume during any moving three-second time interval during normal operation including start-up and shut down. Because, FC stack need to washout by the concentrated hydrogen as the purge gas. So, how to exhaust gas without exceeding 4% is the most concerns. FC stack need to washout by the concentrated hydrogen as the purge gas and how to exhaust gas without exceeding 4% is the most concerns. 3

Objective In order to detect hydrogen behavior in real time, mass spectrometer system with differential pumping stage was selected to develop real time monitoring. To evaluate this monitoring system, we measured hydrogen behavior in a tube modeling FCV hydrogen discharge system composed of plastic tube with pressure gage, Mass Flow controllers and Solenoid valves. Variety of simple experiments, injection, mixing, change flow rate and change tube inside diameter were carried out to control the hydrogen concentration in nitrogen instead of the air. So, In order to detect hydrogen in real time, mass spectrometer system with differential pumping stage was selected to develop real time monitoring system. To evaluate this monitoring system, we measured hydrogen behavior in a tube modeling FCV hydrogen discharge system composed of plastic tube with pressure gage, Mass Flow controllers and Solenoid valves. Variety of simple experiments, injection, mixing, change flow rate and change tube inside diameter were carried out to control the hydrogen concentration in nitrogen instead of the air. 4

Hydrogen Sensor 5 The sensor based on mass spectrometer system Sampling by 150-250μm SUS capillary tube ↓ Skimmer Ionization chamber Ionized by electron bombardment method Mass spectrometer Schematic diagram of Sx system This figure shows the hydrogen sensor. It is the same as the previous presentation. So, I want to cut the detail explanation of the sensor. 5 Sx Hydrogen Sensor

Comparative experiments Experimental Setup Next, I’d like to explain the comparative experiments using general use sensors and our sensor. This figure show the experimental setup. We used a catalytic combustion type sensor and thermal conductive type sensor as the general use sensors. HX is old model and response time is slower than Sx, but it can record the external signal. These sensors are located on the jet axis in front of the nozzle. The solenoid valve is open by switching signal from the pulse delay generator and hydrogen is released in the open space. Recorder and HX are also started by the signal. Sensor ・Catalytic combustion sensor ・Thermal conductive sensor 6

Comparative experiments Time history of hydrogen concentration This graph shows the time history of hydrogen concentration detected by each sensor. HX can detect hydrogen quickly. Catalytic combustion type sensor is not bad, but can not follow the changing hydrogen concentration. On the other hand, thermal conductive type, waveform becomes blunt. 7

Experimental Setup 8 Vertical Mixing Apparatus 4mm L=10～12400mm To investigate the hydrogen diffusion process in tube, we used two mixing type apparatus. First, this figure shows experimental setup of vertical mixing. Tube length is changed from 10mm to 12.4m. The sampling position is near the exit of the tube. H2 and N2 flow rate are controlled by the mass flow meter. Firstly, we used high-speed solenoid valve, its response time is from 0.4 to 3 ms. After, we used various type valve, such as ball valve, needle valve. L=10～12400mm 8

Experimental Setup 9 Coaxial Mixing 4mm 6mm 10mm 1m Φ1.6mm This figure shows the experimental setup of coaxial mixing. The tube diameter of main flow is 4mm, 6mm and 10mm. Hydrogen is spouted to the main flow from 1.6mm inner diameter nozzle. The hydrogen is sampled at 1m from the hydrogen spouting nozzle. 1m Φ1.6mm 9

Time History of Hydrogen Concentration This graph shows the time history of hydrogen concentration. Inner diameter is 4mm, 6mm and 10mm and the tube length is 1m was used. Plastic tube is connected to vertical mixer and the concentration of H2 is monitored by H2 sensor Sx by time at the end of tube by sampling. The experiment is started by open up H2 solenoid valve to introduce 100% H2 at the flow rate of about 2l/min at the pressure of 0.2 MPa. At the beginning of experiment there are some H2 leak from solenoid valve and observed gradual increase of H2 concentration, but the effect is negligible. H2 flow mixed with N2 flow and should reach about 40 % of hydrogen content in steady state flow. The concentration of H2 has peak top about 60 %. Also the width of peak top becomes wider when the diameter of pipe becomes larger. These peak top might changes peak height and peak width. In this paper we call this as H2 spike. H2：3NL/min，N2：5NL/min，Vertical Mixing 10

Relation between Reynolds Number and Spike Head The vertical mixer attached with plastic tube of 4 mm inner diameter, the length of 12.4 m is used to check Reynolds Number effect. Reynolds Number were changed from 200 to 13000 at the H2 pressure of 0.2 MPa, and Reynolds Number were changed from 15 thousands to 20 thousands at the H2 pressure of 0.5 MPa. The increase of Reynolds Number decreases the area of spike head.　 This means that in turbulent flow the spike head made of the high concentration H2 decrease but still survived in the long tube. Vertical Mixing 11

Effect of tube length Vertical Mixing 12 This graph shows the result that are conducted by changing the length of the tube from 10 mm to 12.4 m. The concentration measurement is conducted at the end of each tube respectively. The spike head remains at the length of up to 12.4 m. The high concentration of H2 penetrates 12.4m length tube without diffusion. Vertical Mixing 12

Comparison of various valves This graph shows the comparison of spike area about various valve type. The spike heads are observed by changing the type of valves. The result shows that the spike heads are observed in all experiment and valves with narrow controller like needle type enhance the spike height and area. Coaxial Mixing 13

Spike area of various gases No Gas Cmax(%) Ceq(%) Time(s) Area(a.u) Cmax/Ceq 1 H2 76.9 14 0.47 0.18 5.5 2 He 39 7.5 0.83 0.17 5.2 3 O2 65.3 43.8 0.21 0.07 1.5 4 CO2 45.5 26.5 0.05 1.7 5 CH4 41 27.2 0.15 0.03 6 Ar 49.7 19 0.32 0.08 2.6 Cmax: Spike top concentration of gas Ceq: concentration of gas after stabilization of flow This table shows the spike area of various gases. The experiment using coaxial mixing equipment with plastic tube at the length of 12400 mm carried out and same spike is observed. The spike heads are observed in all gas experiments and the lighter molecular weight enhances the spike height and area. The spike heads are observed in all gas experiments and the lighter molecular weight enhances the spike height and area. 14

H2 Mixing Model Before this study After this study 15 These figures show the over expecting image of hydrogen concentrations, not CFD results. And the over expecting hydrogen concentrations is analyzed by changing experimental conditions. These behavior show in every case, the spike head is appeared and demonstrated that the high concentration phase start to run as front flow in tube. By mixing of hydrogen with nitrogen in steady state flow, laminar flow with diffusion mechanism by concentration gradient might work to reach equilibrium concentration. But the establishment of steady state flow and the concentration gradient by Fick`s law do not work within second considering diffusion constant of H2. This mixing mechanism is almost like injection of H2 into N2. It seems that Reyleigh-Taylor instability might be comprehensive to understand the H2 spike. Even considering the complex system of FC stack in FCV with bent and narrow path for H2, acceleration and deceleration path of H2, the exhaust gas from FCV is simple pulse and width of less than second. 15

Concluding Remarks Simple H2 purge model experiment to demonstrate the existence of cloud in exhaust gas from FC stack was conducted and at the point of discharge the real time H2 monitoring was took place. Our experimental result shows hydrogen injection within milliseconds demonstrated the existence of hydrogen cloud, the spike. The spike head always appeared, when the pressure H2 on front of the solenoid valve is changed. Changing the kinds of valves, the diameter of tube, the length of the tube, and even the Reynolds number up to 20000, this spike head flow with high hydrogen concentration always appeared. Even other gas showed the similar spike head with high concentration. Finally, I would like to shows the concluding remarks. Simple H2 purge model experiment to demonstrate the existence of cloud in exhaust gas from FC stack was conducted and at the point of discharge the real time H2 monitoring was took place. Our experimental result shows hydrogen injection within milliseconds demonstrated the existence of hydrogen cloud, the spike. The spike head always appeared, when the pressure H2 on front of the solenoid valve is changed. Changing the kinds of valves, the diameter of tube, the length of the tube, and even the Reynolds number up to 20000, this spike head flow with high hydrogen concentration always appeared. Even other gas showed the similar spike head with high concentration. 16

Thank you for your attention Acknowledgment A part of this study was supported by MEXT KAKENHI Grant Number 15K16298. Thank you for your attention mogi.toshio@mail.u-tokyo.ac.jp fin