The configuration of the flow field, as depicted in Fig. 2, is irregular.
Therefore, a body-fitted coordinate system, generated by a grid generation approach, is employed. Accordingly, the physical domain is transformed into a computational domain that consists of the equally spaced, square grids. Weng (1989) detailed the procedure, which is not presented here.
The computational domain is selected to be xin = -7, xout = 13, and ywall = 4. The upstream and downstream positions are determined via several numerical experiments to meet the requirement that applying boundary conditions at these positions sho uld not impact the flame structures. Then, a set of numerical tests is conducted to ensure further that the resultant solutions are grid-independent. Table III presents test results. The cases shown in the first column are the same as those in Fig.
25, which will be discussed later. If the number of cells exceeds 218×115, then the variation of resultant peak temperature, the variable most sensitive to the size of the grid, over the entire computational domain becomes insignificant by the increasing number of grid cells.
Therefore, this work uses 218x115 grid cells. The grid is much finer than that, 112×51, used in the earlier study (Chen and Weng, 1990).
CHAPTER 3 Experimental Apparatus Setup
Basically the experimental setup is incorporated with the present combustion model. The experimental setup consists of three major elements in the apparatus, which are the wind tunnel, the porous sintered cylindrical burner and measurement instrumentations. They are described in detail as follows.
3.1 Wind Tunnel
According to the simulation, the tunnel is used to provide a laminar, uniform oxidizer flow to the porous cylindrical burner, and the fuel is injected from the surface of the burner. It is open-circuit and oriented vertically upwards. A schematic configuration of the wind tunnel is shown in Fig. 3. There are five components in the wind tunnel: (1) a blower, (2) a diffuser, (3) flow straightener, (4) a contraction, and (5) a test section. Concepts of making wind tunnel mainly are from Yang et al. (1999).
3.1.1 Blower
The airflow in the tunnel is provided by a variable-speed (frequency controlled) blower (Type TB-201, C-F Company), whose outlet is connected to the main part of the wind tunnel via a flexible 60 cm long and 10 cm diameter plastic ductwork, which the end is coupled to a diameter 346 mm and 75 cm long cylinder, which is designed according as AMCA 210-85 standard (shown in Figs. 4 and 5). The blower is driven by a frame motor, which is controlled by an inverter drive. A
frequency converter (Type M36V2P07, DYNAGEN, J-C Company) is used to control the rotational speed of blower to get the desired velocity.
The frequency of the blower and corresponding velocity in test section shown in Fig. 6. In order to avoid the influence of vibration, the base of wind tunnel and blower are separated (shown in Fig. 7).
3.1.2 Diffuser
A 30 cm long diffuser has an inlet cross-section area of 12x12 cm2 and the outlet one is 40x40 cm2. The expansion ratio based on area is 1:11.
3.1.3 Flow Straightener
The airflow from diffuser section is really unstable before entering into contraction section. The flow straightener is used to make it more stable. The flow straightener section, which serves to insure that the flow to test section is laminar and uniform over the entire cross-section, consists of honeycomb and screen. The screen is mounted to decrease disturbances and make flow uniform. The honeycomb is added to utilities of the screen, like reducing turbulence effect. With appropriate combinations of screen and honeycomb characteristics and putting them at optimum position, it can achieve the goals mentioned above.
3.1.4 Contraction Section
The test section cross-section area 24x4 cm2, therefore the contraction ratio over the contraction section is 16.6:1. Its purpose is to promote a uniform field in the test section. The design criteria are
demanded to shorten the duct and reduce boundary layer thickness along the wall as possible.
3.1.5 Test Section
The test section has a cross section area of 24x4 cm2 and a length of 30 cm. It is made up four sides. In the front and two connecting sides, they are equipped with quartz-glass plates for observation windows.
The rear side is made of stainless steel plate for supporting the burner to insert. The downstream of the test section is connected to a diffuser (500 mm), which can reduce the amount of exhaust gases from test section. The vent follows after the diffuser to outdoor. The rear part of the vent can resist high temperature by adhered fins to inside the vent as heat exchanger. The outside of vent is connected with water-cooling system, including cooler, water tank and pump (shown in Fig. 8). The front part of inside the vent is also very important since a series of instruments are set up to measure the heat release rate (see Sec. 3.3.6).
Measurement in the test section is to confirm the velocity uniform and stable. There are four sets of pitot tubes and a fixed static pressure hole in the front of the burner of the test section (see Figs. 9 and 10).
Change different connection in four pitot tubes in order (fixed the static pressure hole) under the same inflow velocity after a long time. If the pressure difference is closely equal, it means that the flow is uniformity and stabilization (see Fig. 11).
3.2 Porous Sintered Cylindrical Burner
3.2.1 Burner Structure
The requirement of experimental burner has to be able to sustain the high temperature. It is designed to have inner and outer parts, respectively. The outer part of burner is a replaceable porous sintered stainless steel (20, 40, 70μm pores, respectively) with a length of 40±
0.5 mm. Its inner diameter is 20±0.5 mm, and an outer diameter 30±
0.5 mm (see Figs. 12 and 13). The advantage of this design is that the burner replacement can be easily performed whenever clogging or damage on the porous burner surface occurs due to burning for a long time. The inner of the burner is a cylindrical brass rod (see Fig. 14) with an internal water-cooling groove and fuel supply groove. The outer part is screwed on the inner part. The internal water flow is used to cool the burner to prevent damage from the porous surface structure. The cooling device of the burner includes a water tank, pump, cooler and connected-piping (shown in Fig. 8).
3.2.2 Burner Equipped to Test Section
There are two parameters (half fuel-ejection surface and full fuel-ejection surface respectively) to be measured. Therefore, for the front half side cylinder surface fuel-ejection condition, coating several thin layers of high temperature resistant paint on the backward portion of the burner surface in order to prevent the fuel ejection into the wake region.
(1) The painted surface needs to dry at least 2 hours.
(2) Screwing the burner into the insert and mounting the burner closely to the wall of test section. Connecting the cooling water and fuel pipelines to the burner.
(3) Noticing to adjust the burner so that the uncoated burner surface is facing the direction of the airflow.
(4) In addition, it’s important to test whether there is any leakage from the painting surface of the burner. Instead of fuel pipe to air or nitrogen piping, and then immerse the burner into a water container and open a valve to let gas flow to observe whether there are bubbles from the painted surface.
Methane (CH4, 99.99%) is used as the fuel, and its flow is controlled and measured by a flow meter. A digital mass flow controller (Type MC-2100E, LINTEC, shown in Fig. 15) is controlled by high capability of microprocessor inside. The sensor flow rate signals and command signals are digitized by 16 bit A/D converter to process and operate inside CPU (including: temperature compensation, linearity compensation and control signals operations), and then transformed flow rate signals and command signals into analog signals by 16 bit D/A converter. The fuel ejection velocity is calculated by dividing the fuel volumetric flow by the available fuel ejection area of the burner surface.
3.3 Measurement Instrumentations
3.3.1 Nozzle of the AMCA 210-85 Standard
The measurement of inlet velocity at the test section adopts the AMCA 210-85 standard of nozzle-method to measure volume flow rate and then to deduce flow velocity. The standard is adopted to establish uniform methods for laboratory testing of fans and other air moving devices by AMCA (Air Movement & Control Association Inc.). There
are three nozzles (Ö: 10, 15, and 30 mm) and four sets of pitot tubes distributed equally inside the cylinder (shown in Fig. 5), which are incorporated to measure local velocity of the cylinder. Then it can get total volume flow rate to deduce velocity of test section by dividing cross section area of test section. The precision is within 3% when velocities are ranged from 0.21 m/sec to 3.3m/sec precision, but it becomes 5% as the velocity is smaller than 0.21 m/sec.
3.3.2 Digital Video
A digital video (Type DCR-TRV17, SONY) is used to record the flame profiles, such as envelope, side, wake flame, and flame lift-off. It is fixed on appropriate position to catch flame variations. The special function applied is night-shoot to record various flames in the dark laboratory. All imagines recorded on the tape of the cassette have to transmit to computer to process. The digital video is connected to computer by IEEE 1394 card, and then imagines processed by Ulead Video Studio software to show a series of flame structures in different velocity regime.
3.3.3 Thermocouples
K-type thermocouple (wire diameter, 1 mm; compensation wire diameter, 0.65 mm) is used to measure the temperature in order to get the corresponding density in the exhausted duct. It is made of Ni-Cr/Al-Ni alloy material, compensation time 1 second and measured temperature up to 1200 ℃. The measured signals are connected to data acquisition device (PC Recorder, M-SYSTEM) to transform analog signals into
digital signals by series port RS-232 to computer with MSRS32-E software.
3.3.4 Pressure Transducer
Pressure drop received from bi-directional pitot tube (see Figs. 16 and 17) in the vent is converted into micro-voltage signal by pressure transducer. Via this procedure, the averaged velocity in the duct can be calculated. And pitot tubes in the test section and in front of the nozzles of the cylinder are also measured by the pressure transducer (Type PF-MPS2, POUNDFUL).
3.3.5 Oxygen Analyzer and Pretreatment System
Oxygen analyzer (Mode 755A O2 Analyzer, shown in Fig. 18) is used to measure oxygen consumption in the vent. It should be calibrated and zeroed before testing, using 99.99% pure nitrogen as the zero gas and the air as the span gas, which is composed of 21% oxygen and 79% nitrogen (all gases produced by J-R Gas Company). Exhausted gas should be filtered and cooled completely in the pretreatment section to avoid high temperature and other suspended grains to damage the analyzer. It may reduce oxygen concentration and result data unstable.
The pretreatment system is constructed of two sets of series glass wool filter, a membrane filter, and a cooler and micro pump to introduce gas inlet (shown in Fig. 19). The whole schematic piping arrangement of pretreatment system is shown in Fig. 20. The measured signals are connected to data acquisition device (PC Recorder, M-SYSTEM) to
transform analog signals into digital signals by series port RS-232 to computer with MSRS32-E software.
3.3.6 Heat Release Rate Measurements
The flame strength is quantified by the heat release rate. Such measurement is carried out in the exhausted duct, which includes: (1) K-type thermocouple, (2) a robust bi-directional probe connected to pressure transducer and (3) a gas sampling probe connected to the oxygen analyzer with the pretreatment section (shown in Fig. 21). These instruments are used to measure the temperature, velocity and oxygen concentration of oxygen of product gas, respectively. All the sampling gases are first introduced to the pretreatment section for cooling and filtering before they go into the oxygen analyzer. The whole schematic configuration can be seen in Fig. 22. The signals will be collect via a data acquisition system, and then these data are handled by 586-PC to calculate the heat release rate.
3.4 Procedure of the Experimental Operation
(1) Calibrate the instruments to make sure the stabilization and accuracy of their performance before performing the experiment.
(2) It usually takes times for completion of warm-up for the apparatuses that include blower, mass flow controller and O2
analyzer.
(3) The blower has to be operated for 30 minutes until the flow uniformity and stabilization are achieved. Stabilization depends on pressure difference (four pitot tubes vs. 1 fixed static pressure
hole) in front of the burner in the test section is stable.
(4) The digital mass flow controller needs to be operated for 15-30 minutes because of thermal-sensor type to let flow more accurate.
(5) Check if any fuel gases leak from pipelines by suds. It’s very important procedure for this achromatic, flavorless, toxic and flammable fuel.
(6) Turn on the flow of cooling system to the burner and vent.
(7) Start the computer program (software: MSRS32-E) used to collect desired data via a data acquisition system (PC Recorder), and then these data are handled by 586-PC using software to calculate the heat release rate.
(8) Turn on the ignition device, which is produced spark by the way of 3000 volt high voltage. Turn it off until the flame is established. Note that remember to ignite first before let the fuel input, or it may be exploded.
(9) Open the valve of methane fuel vase and keep inlet pressure up to 8 psi, then turn on the stop valve of mass flow controller.
The alarm light of mass flow controller displays green that means fuel passes pipelines to the burner. Set the fuel flow rate to the desired amount in liters per minutes. First choose a certain fixed value of fuel supply to the burner, and then increase slowly the airflow velocity in order to get various flame types.
It means that change the inflow velocity as a parameter under a fixed fuel ejection rate.
(10) In the low velocity regime, the envelope flame is expected to
appear. Once the flame is established, permit several minutes for the flame to stabilize or adjust the blower speed if necessary.
Next, gradually increase the inflow velocity to get the other types of flames. After that, change another value of fixed fuel ejection rate to continue this step. Then it can gain profiles of different fuel flow rate as increasing inflow velocity.
(11) Digital video is fixed to position to get the same observation view of test section to catch all images from testing.
(12) Using the front half side fuel-ejection of cylinder surface instead of full side fuel-ejection burner to repeat the procedure from (8)-(10) steps.
(13) Also change the fuel ejection rate as a parameter under a fixed flow velocity. Carry out the similar procedure as the above.
3.5 Uncertainty Level Analysis in the Experiment
Uncertainty analysis is carried out to estimate the uncertainty levels in the experiment. Formulae for evaluating the uncertainty levels in the experiment can be found in several papers (Kline and Mcclintock [1953]
and Moffat [1982]) and textbooks (Holman [1989], Fox and McDonald [1994], and Figliola and Beasley [1995]). Accordingly, Table IV summarizes the results of all uncertainty analyses.
3.6 Experimental Repeatability
The procedures for changing the airflow velocities at different fuel ejection velocities were performed three times to ensure experimental repeatability, coincidence, and accuracy. The transition velocity, which
transforms an envelope flame into a wake flame, is a critical value to investigate the flame behaviors. Table V presents the transition velocities for flame transformation from an envelope flame into a wake flame in the front half cylinder fuel-ejection system as a function of fuel ejection velocity. The table records three measured data, their average, and the error at each fuel ejection velocity. The error is defined as the ratio of the absolute difference between the maximum and minimum values of the three data to their average. Generally, the errors are within an acceptable range (maximum of 6.82%) and the repeatability is quite good except at three points, vw = 1.12 cm/s, 1.23 cm/s, and 1.68 cm/s.
These points are near the two demarcation lines, between region I (vw = 0.9 ~ 1.12 cm/s) and II (vw = 1.23 ~ 1.57 cm/s), and region II and III (vw
= 1.68 ~ 2.8 cm/s). The errors are inevitably large at these critical points. The flame transition processes and their corresponding characteristics are detailed below.
CHAPTER 4 RESULTS AND DISCUSSION
I. Simulation Part
The gaseous fuel used is methane (CH4) and the ambient oxidizer is air. The basic thermodynamic and transport property data, summarized in Table VI, are taken from Chen and Weng (1990) to enable a fair comparison later.
4.1 Comparison with Related Experiments and Simulations
The present combustion model is first validated by comparing the predicted results with the pertinent measurements of Tsuji (1982) and the simulation results of Chen and Weng (1990). Then, the predictions are compared with the measurements and calculations of Dreier et al. (1986).
Figure 23 presents the comparison, by plotting the blow-off limit as functions of -fw (nondimensional fuel ejection rate) and 2Uin/R (flame stretch rate, ks). Notably, this line in Tsuji’s experiment (1982) represents a demarcation line at which the flame is transformed from an envelope flame to a wake flame instead of being extinguished. The predictions of this report are quite close to Tsuji’s experiments (1982) in the regions of high fuel-ejection rate and low stretch rate. Generally, the prediction is much better than that of Chen and Weng (1990), implying that the prediction that is based on a four-step chemical mechanism is indeed better than the one that uses a one-step overall chemical mechanism. However, in the domain of 0.2 < -fw < 0.77 and 91 <
2Uin/R < 376, a significant discrepancy exists between the present predictions and Tsuji’s measurements (1982). Notably, this domain is
located at a transition from the very small fuel ejection rate to the large flame stretch rate. The discrepancy may be attributable to two factors stated in Chen and Weng (1990): the first is the three dimensional effect in the experimental configuration and the other one is a chemical effect.
The blow-off that results from a low fuel ejection rate is very close to that obtained experimentally because it is governed mainly by the thermal quenching of the cylinder surface. The geometric effect should be minor in this branch. However, aerodynamic and chemical limitations greatly affect the blow-off mechanism due to flame stretch. For a given 2Uin/R (~ 376), in the higher -fw regime (> 0.77), the four-step chemical effect seems appropriate even if the fluid flow dominates, whereas it does not suffice to describe the reactions in the regime of lower -fw, such as 0.2
< -fw < 0.77. Better agreement with measurements initiates from -fw = 0.5 and 2Uin/R = 128, which occurs much earlier than that of Chen and Weng (1990), and continuously improves thereafter.
Figure 24 compares the predictions in this study to the measurements and calculations of Dreier et al. (1986). This figure depicts the temperature distribution along the forward stagnation line, where the forward stagnation point is at x = 0. The presented combustion model reproduces the data measured in the experiment, and the agreement is much better than that of their own numerical computation. The temperature profile on the oxidizer side predicted in this study shifts to the left of the experimental data by around 0.2mm; the shift is approximately 0.5mm on the fuel side. Considering the experimental uncertainties, the agreement can be regarded as excellent.
Now, Table VII and Fig. 25 directly compare with the results of Chen and Weng’s simulations (1990). Table VII depicts the inflow velocity range for flames with different appearances. Notably, no side flame exists in this study, whereas neither lift-off nor subsequent late wake flames appeared in Chen and Weng (1990). Apparently, the application of a four-step mechanism shows its influence on flame
Now, Table VII and Fig. 25 directly compare with the results of Chen and Weng’s simulations (1990). Table VII depicts the inflow velocity range for flames with different appearances. Notably, no side flame exists in this study, whereas neither lift-off nor subsequent late wake flames appeared in Chen and Weng (1990). Apparently, the application of a four-step mechanism shows its influence on flame