【溫馨提示】 dwg后綴的文件為CAD圖，可編輯，無水印，高清圖，，壓縮包內文檔可直接點開預覽，需要原稿請自助充值下載，請見壓縮包內的文件及預覽，所見才能所得，請細心查看有疑問可以咨詢QQ：414951605或1304139763
Available at v h ch, ch, significantly cheaper than liquid fuels since it is left behind as 2 and additionally can be produced from waste [1]. Hydrogen as overpressure. Thus, it is of great interest to developers, the All these situations can be simulated on chassis dynam- ARTICLE IN PRESS INTERNATIONAL JOURNAL OF HYDROGEN ENERGY 33 (2008) 863–869 C3 Corresponding author: Tel.: +41448234679; fax: +41448234044. fuel for fuel cells as well as for I.C. engines is likely to play an important role in future vehicle propulsion technology. The development of hydrogen-powered vehicles is also driven by ometers embedded in climatic chambers. However, it is only possible with enormous effort to keep such climatic cham- bers so airtight that the evaporative emissions can be 0360-3199/$-see front matter in C0 X m C3 G;out , qm G qt ? m C3 G;vent_in tm C3 G;car C0 X m C3 G;out . e1T qm G qt denotes the change in mass of gas G within the cell, P m C3 G;in the sum of all mass flows of gas G into the chamber and P m C3 G;out the sum of all mass flows of gas G out of the chamber, m C3 G;vent_in the mass flow into the chamber from ventilation and m C3 G;car the source flow of interest. All variables are functions of time. A mass flow of gas G into the chamber occurs if this matter is found in the ambient air. Thus, the mass flow of ventilation air and the concentration of gas G in the intake air need to be ARTICLE IN PRESS INTERNATIONALJOURNALOFHYDROGENENERGY33 (2008) 863–869864 Drivers aidAirstream ventilator Max. flow: 100’000 m 3 /h Internal ventilation, air conditionning heat exchanger V intake * * V leakage Fig. 1 – Sketch of climatic chamb Chassis dynamometer V air,vent_in * * * V exhaust V air,vent_out er with ventilation flows. In cha location. Natur be determine m air ? r air V air . (2) ARTICLE IN PRESS OGEN The contained mass flow of gas G then is m C3 G ? c G r G V C3 air , (3) where c G is the concentration of gas G and r G its density. Since C3 the da duct of air density r air and volume flow V air . C3 pro the cell. It is thus sufficient to measure the air inflow. addition, since the concentration inside the climatic mber is homogenous, it needs to be measured at just one ally, a flow of gas G as the inflow m C3 G;vent_in cannot measured directly. By assuming ideal gases, it may be d as follows. Any mass flow of air m C3 air is the C3 into measured. The second mass flow into the chamber is the evaporation from the vehicle, which is of interest. There are different possibilities for flows of gas G out of the chamber: C15 Intended ventilation. C15 Leakage. The doors of the chamber as well as channels for cables and pipes are not airtight, so some air leaks. Most climatic chambers operate with a slight overpressure to ensure that air flows out at all openings, since inflowing humid air, when operating at low temperatures, would cause dangerous ice formation and additionally disturb the humidity control of the chamber (Fig. 1). C15 If the vehicle is running and is propelled by a system that consumes air (engine or fuel cell system), either the corresponding air supply can be from outside the chamber or air from the room is used. Since the exhaust gases are typically led outside the chamber and measured there, the latter case is also an outflow for the mass balance of gas G. It is obviously not possible to measure the mass flow and concentrations of gas G at all the outflow locations, but this problem can be bypassed by the following approach. The chassis dynamometers for exhaust emission measure- ments are equipped with fans for the cooling of the vehicle. Together with the ventilation of the air conditioning of the cell, this can cause such high turbulence that the concentra- tion of gas G in the room can be considered to be homogeneously distributed. In other words the mixing time constant in the room must be significantly lower than the air exchange rate. It must be ensured that no dead zones where ventilation is poor exist inside the climatic chamber. In other words, in most cases where chassis dynamometers are installed in climatic cells, the rolls of the dynamometer as well as the breaking electric motor are in an under-floor compartment that is contained in the cell. It must therefore be possible for this compartment to be ventilated intention- ally by opening covers and adding additional ventilators. If the concentration of gas G inside the chamber is indeed homogenous and measured, this concentration also holds for all the outflows of the cell. As long as the pressure remains stable within the cell, which is controlled by the ventilation, the total mass flow of air out of the cell is equal to the flow INTERNATIONAL JOURNAL OF HYDR tests take place in a climatic chamber and do not last for ys, it may be assumed that both temperature and pressure remain stable, and thus that densities are constant. It is thus sufficient to measure the volume flow of air and the concentration of gas G to determine its mass flow. For the chamber it correspondingly holds that m G ? c G r G V ch . (4) The index ch stands for chamber. Assuming that the volume flow of air out of the chamber is equal to the inflow and that the distribution of gas G in the chamber is homogenous, Eqs. (1)–(4) give qc G;ch qt r G V ch ? c G;vent_in r G V C3 air;vent_in t m C3 G;car C0c G;ch r G V C3 air;vent_in . e5T And this can be solved for the mass flow of the source, thus the car m C3 G;car ? qc G;ch qt r G V ch C0 c G;vent_in r G V C3 air;vent_in t c G;ch r G V C3 air;vent_in . e6T So, the system emissions as mass per time unit can be calculated by knowing the chamber volume, the density of gas G (thus temperature and pressure) and measuring the volume flow of air into the chamber as well as the gas concentration of G inside the chamber and in the air intake. As pressure and temperature in inflow and outflow are alike, densities for both flows can be considered to be equal. 2.2. Measurement equipment A commercial gas chromatograph (Reduction Gas Analyzer (RGA3), Trace Analytical Inc., California, USA) was used to measure H 2 inside the climatic chamber. The RGA3 is an ultra-trace level gas detection system capable of monitoring low ppb concentrations of reducing gases such as H 2 . The instrument consists of a microprocessor-controlled gas chromatograph which utilises method of reduction gas detection. Synthetic air preconditioned by molecular sieve 5A ? and SOFNOCAT to remove H 2 O and reaction impurities (CO and H 2 ) is used as carrier gas. Aliquots of air samples are flushed with a rate of 20ml/min over a 1ml sample loop. After equilibration, the sample volume is injected onto the columns. Sample components of interest are separated chromatographically in an isothermal mandrel-heating col- umn oven. The chromatographic precolumn (Unibeads 1S, 60/80 mesh; 1=8 00 C230 00 ) is mainly used to remove CO 2 ,H 2 O and hydrocarbons. Subsequently H 2 and CO are separated by the analytical column (molecular sieve 5A ? , 60/80 mesh; 1=8 00 C230 00 ) and pass into the detector which contains a heated bed of mercuric oxide. Within the bed a reaction between mercuric oxide (solid) and H 2 occurs and the resultant mercury vapour in the reaction is quantitatively determined by means of an ultraviolet photometer located immediately downstream of the reaction bed. The columns are kept at 751C; the detector is heated to 2701C. The amount of H 2 in the air sample is proportional to the amount of mercury that is determined. ENERGY 33 (2008) 863–869 865 During the quasi-continuous observations of the H 2 con- centration in the test chamber, measurements were taken every 2min. At the beginning and end of each test cycle the ambient air concentration (concentration of the ventila- tion inflow) was measured for 30min. Typically the concen- trations were very constant over the short time of one test cycle and in the range of the mean of 576C694ppb at Duebendorf [9]. Two high concentration reference gases (50 and 100.2ppm H 2 ; Messer Schweiz, Switzerland) were dynamically diluted with zero air to the range of interest by means of a dilution unit (MKAL diluter, Breitfuss Messtechnik GmbH, Harpstedt, Germany). The dilution unit was indirectly referenced against the primary gas flow standard of the Swiss Federal Office of change substantially; thus average concentration during one sampling step is approximated by the mean of the values So the mass emitted during the sampling interval k is m G;car;k ? r G ec G;ch;k C0 c G;ch;kC01 TV ch t V C3 air;vent_in;k T C18 C2 c G;ch;k t c G;ch;kC01 2 C0 c G;vent_in;k C18C19C19 . e9T Mathematicallymore complexbut also more accurate is the discretisation by solving the differential equation (5) analyti- cally for one time step, what needs certain assumptions. Here there is freedom to assume all input signals (i.e. V C3 air;vent_in etT, c G;vent_in etT, m C3 G;car etT) as arbitrary functions of time. ARTICLE IN PRESS INTERNATIONALJOURNALOFHYDROGENENERGY33 (2008) 863–869866 measured at either end of it. The mass balance results in m C3 G;car;k ? c G;ch;k C0 c G;ch;kC01 T r G V ch C0c G;vent_in;k r G V C3 air;vent_in;k t c G;ch;k t c G;ch;kC01 2 r G V C3 air;vent_in;k . e8T * m G,car ( t ) t Case: early peak Case: constant * m G,car ( t ) Metrology. The different mixtures of the two high concentra- tion standards showed excellent agreement with each other and the NOAA/GDM scale [10]. Detection limit for H 2 was C610ppb and the standard uncertainty of the measure- ment 5%. 2.3. Analysis methodology As described in the previous section the low concentrations of the gases of interest cannot be measured with high time resolution, i.e. within seconds. The equipment described above allows a sampling rate of 2min. Thus Eq. (6) needs to be solved discretely. The most direct and simple approach of discretisation is replacing the derivative of the chamber concentration by the difference of the last two measured values. For time step k this results in qc G;ch qt et ? kTTC25 c G;ch;k C0 c G;ch;kC01 T , (7) where T is the sampling interval [11]. Since both ambient concentration of gas G and ventilation air flow typically change very little over one time interval, it does not matter if the values at the beginning or the end of the sampling interval are used. The chamber concentration, however, may (k-1)T kT (k-1)T Fig. 2 – Most extreme possibilities of time functi Hence, if necessary it might be possible to measure the ventilation air flow at high time resolution and use that time function for the calculus, but usually this flow is reasonably constant. The ambient concentration of the gas G typically is constant too if not working downwind of a huge non-uniform gas source. Of course the time function m C3 G;car etT of how the vehicle emits the gas G is unknown. If the total mass emitted during one time step m G;car;k is given, the most extreme case for the calculus is if all of it is released immediately after the time interval starts or immediately before the time interval ends (peak functions, Fig. 2). The ‘‘a(chǎn)verage’’ case happens if the vehicle is constantly emitting gas G. For benchmarking the quality of this methodology in Section 3.2, Eq. (5) is solved subsequently for all three assumptions. In the case of the early peak, the solution of Eq. (5) for the time t ? kT is c G;ch;k ? c G;vent_in t c G;ch;kC01 t m G;car rV C0 c G;vent_in C18C19 e C0eV C3 =VTT , (10) and thus, the mass of gas G emitted in one time period is m G;car;k ? rV c G;ch;k C0 c G;vent_in e C0eV C3 =VTT t c G;vent_in C0c G;ch;kC01 ! . (11) For the late peak case we obtain c G;ch;k ? c G;vent_in tec G;ch;kC01 C0 c G;vent_in Te C0eV C3 =VTT t m G;car rV , (12) m G;car;k ? rVec G;ch;k C0 c G;vent_in tec G;vent_in C0c G;ch;kC01 Te C0eV C3 =VTT T. (13) t t Case: late peak * m G,car ( t ) kT (k-1)T kT ons of gas release for benchmarking. And for the average case of a constant emitting source m C3 G;car etT?m G;car;k =T: c G;ch;k ? c G;vent_in t m G;car rTV C3 t c G;ch;kC01 C0 c G;vent_in C0 m G;car rTV C3 0 @ 1 A e C0eV C3 =VTT , (14) m G;car;k ? rTV C3 V c G;ch;k C0 c G;ch;kC01 e C0eV C3 =VTT 1C0e C0eV C3 =VTT C0 c G;vent_in 0 @ 1 A . (15) Even though Eqs. (11), (13) and (15) look rather different, their outputs remain similar as long as the sampling interval Tis small compared to the ventilations time constant V=V C3 .So the quality of this method rises if both the sampling interval values belong to this test equipment. heat exchange units, etc., are difficult to describe. Thus a test where a well-defined volume of helium was released im- 3.2. Identification of volume flow and validation If the volume flow of the ventilation is not possible to be measured directly, but is constant over time, it is possible to determine it by the following test. As above, a certain volume of a measurable gas such as helium is injected into the cell (with ventilation running). After the mixing phase in the cell the helium concentration will follow Eq. (5) or one of its solutions (10), (12) or (14) with a zero source activity of the car m C3 G;car etT?0. Measurements are shown in Fig. 3. Subtracting the background concentration and building the logarithm of the He concentration result in a straight line for the first 2000s where concentrations are reasonably above the detection limit. The gradient of this straight line is directly the air exchange rate, thus V C3 =V. Its inverse is the above discussed air exchange overall model. A known amount of helium (or hydrogen) can be released either in one moment as above or repeatedly if ARTICLE IN PRESS INTERNATIONAL JOURNAL OF HYDROGEN ENERGY 33 (2008) 863–869 867 mediately and its concentration was measured subsequently, while external ventilation was closed but internal circulation was on, allowed the chamber volume to be estimated from the dilution ratio. The value was found to be 256m 3 with a standard deviation of 8m 3 . 0 500 1000 1500 2000 2500 3000 0 10 20 30 40 50 Helium concentration in air exchange test He [ppm] 3.1. Determination of chamber volume Estimating the air volume in the chamber by geometrical means is quite difficult, since the volumes of car, ventilators, and ventilation are small. Realistic examples are given in the next section, where both the different methods (Eqs.(11), (13) and (15)) and different sampling intervals are applied to the same test data to highlight how accuracy depends on the different parameters of the system. 3. Example and sensitivity analysis The test examples described here were conducted in the climatic cell chassis dynamometer of Empa. All numeric time [s] Fig. 3 – Determination of air exchange ra equipment allows and the calculus (Eqs. (11), (13) or (15)) using measured values must give the amount released. This was repeatedly checked. 3.3. Evaporation test and accuracy analysis Real hydrogen system emission tests were conducted with a hydrogen vehicle. The test shown here included a parking phase from 1 to 2523s, then a test ride up to 3842s, where another parking phase is monitored up to 7100s (Fig. 4). The left plot in Fig. 4 shows the hydrogen concentration measured with a 2min interval. On the right side the emissions of the car for each time interval are displayed. 0 500 1000 1500 2000 2500 3000 -4 -2 0 2 4 log(He - He ambient ) He [ppm] time constant V C3 =V, and if one of the chamber volume or ventilation volume flow is known the other can be deter- mined. Here, with the given chamber volume the volume flow is found to be 0:5605m 3 =s with a standard deviation of 0:005m 3 =s. The volume flow was found to depend on ambient pressure, and this identification should thus be repeated on the day of the evaporation tests. In addition, if the volume flow of the ventilation is known by measurement, similar tests can be used to validate the time [s] te by helium injection experiment. difference of two measured values. These errors are, however, ARTICLE IN PRESS because 0 2000 4000 6000 8000 -0.05 0 0.05 0.1 0.15 Mass H 2 [g] time [s] Fig. 5 – Cumulative hydrogen emissions. OG They are calculated applying the different methods and assumptions, i.e. Eqs. (9), (11), (13) and (15). For the given situation with a chamber volume of 256m 3 ,a ventilationvolume flow of 0:5605m 3 =s (giving an air exchange time constant of 463s or 7.72min) and a sampling rate of 2min, the accuracy results are as follows: The approximate formula (9) and the accurate formula (15), both assuming that the vehicle emissions are constant over one sampling interval differ less than 0.5% from each other. The values calculated by the worst case equations (11) and (13), assuming short emission peaks at the beginning or end of the sampling intervals, produce errors of 14% and C012%. As can be seen from the overall characteristic of the mass emission curve (Fig. 4, right), however, it is very implausible that the emissions of the vehicles are peak-like and those peaks exactly synchronised with the sampling. Thus, the real accuracy locally, when emissions start or stop, may be as uncertain as C012% to 14%. The overall or aggregated emis- sions, however (as displayed in Fig. 5), will show a much higher accuracy in all practical cases. 0 2000 4000 6000 8000 600 800 1000 1200 1400 1600 1800 2000 Hydrogen concentration Conc (H 2 ) [ppb] time [s] Fig. 4 – Evaporation test: left: chamber concentration, right: calcula left and right curve appears similar, the right one is sharper INTERNATIONALJOURNALOFHYDR868 From Figs. 4 and 5 it can be readily seen that this vehicle shows rather small system emissions while running, i.e. 0.0046g after a 21min ride (3842s). Conversely they rise remarkably after system stop. The maximal gas flow reaches 4.32mg/min some 20min (1200s) after engine stop and decreases slightly afterwards. Obviously some parts of the hydrogen system leak after system stop until they are exhausted. Note that all variables such as ventilation flow and ambient concentrations are considered to be constant within each time step. If they vary slowly and their values are measured, this methodology is also applicable with the same accuracy. 3.4. Sensitivity analysis The sensitivity to measurement errors of this method can be analysed by standard error propagation methods [12].It shows that random errors in the measurement of the chamber concentration have a considerable impact on the quality of step by step results, caused by building the 0 2000 4000 6000 8000 -2 0 2 4 6 8 x 10 -5 Massflow of hydrogen form car Massflow H 2 [g/s] time [s] equ. (9) equ. (15) equ. (11) equ. (13) ted emission mass flow. Note: Although the shape of the it includes the derivative of the left curve. 0.2 Cumulated hydrogen emissions ENENERGY33 (2008) 863–869 compensated when the integral emission is built. A systematic error of the concentration values, i.e. a bias between the chamber values and the ambient (or inflow) values would result in an incorrect linear trend underlying the integral signal. Such a trend can be detected easily, if the tested vehicle shows phases with assured zero emissions, such as after being stationary overnight. Alternatively, such bias can be reduced by using the same sensor for ambient (intake) and chamber concentration measurements, what is recommended. In addition this method is sensitive to the ratio of sampling rate and air exchange rate. This sensitivity is highlighted by just neglecting intermedi- ate data points in the above example. In this way, the sampling rate can easily be simulated to be a multiple of the original sampling of 2min. It can be seen in Table 1 that with increased sampling time the range of theoretical uncertainty increases. When the sampling time reaches similar values to the air exchange time constant of 7.72min, i.e. 6 or 8min, then the maximal uncertainty rises above 50%, and the values of single steps thus become somewhat ARTICLE IN PRESS Table 1 – Error between methods as a function of OGEN Sampling time interval 2min 4