Auditoría de control interno.

2.6.1 Burner system setup

Laminar flame speed measurements were carried out by utilising the jet-wall stagnation flame configuration coupled with particle imaging velocimetry (PIV) technique. The burner consisted of a converging nozzle with 22 mm exit diameter and was shrouded by a 3 mm wide annular nitrogen flow to prevent air entrainment from disturbing the flame. The burner flow impinged on a flat, water-cooled brass surface with a diameter of 100 mm, and the distance between burner outlet and plate over the nozzle diameter ratio (L/D) was varied over the range of 0.6 to 1.0 to stabilize the flame. The jet-wall stagnation setup is shown in Fig. 2.1.

An in-house fluidized bed seeder was employed for PIV velocity measurements.

The geometry and mechanism of the solid particle seeder is described in Appendix A1.

The seeder contained an internal swirler and was used in conjunction with a magnetic stirrer. The seeding particle used in the experiment was hydrophobic AEROSIL^® Amorphous Silica R812 S with a size distribution in submicron range (~ 0.3-0.4 μm) and a density of 0.05 g/cm³. The compressed air flow was dried and filtered to remove any moisture before fluidizing the particles to prevent agglomeration. A bypass valve was installed upstream of the burner to vary the flow rate of the mixture at the nozzle outlet, so the stretch rates can be varied while maintaining the same mixture

Experiments

Figure 2.1: Jet-wall stagnation configuration for laminar flame speed measurement

2.6.2 Acetone/methane/air mixture preparation

Figure 2.2: Flow delivery system for (a) acetone/methane/air and (b) acetone/air mixture.

Bypass

Seeder MFC

Burner Plate

MFC Air Air

Needle valve

Heated

MFC CH4

Thermoregulator Dreschel bottle

Vaporiser

Mixer ^MFC

Air

(b) Acetone/Air (a) Acetone/CH4/Air

Cooling coil

Pump Acetone Heated

N2

Acetone Stagnation

plate

N2 co-flow Thermocouple Water inlet

Burner

Water outlet

22 100

mm

Experiments In this experiment, the acetone seeding concentration in methane/air is defined as the percentage of acetone in total fuel by mole α = X_a/X_f. The equivalence ratio of the mixture is calculated based on the total mixture of methane/acetone/air. Liquid acetone was vaporised into gaseous form before seeding into the methane/air mixture by utilising a bubbling system typical of those used in acetone PLIF setup in the flow delivery system, as shown in Fig. 2.2a. A 250 ml Dreschel bottle containing HPLC grade liquid acetone (99.8+% pure; Fisher Scientific) was immersed in the fixed temperature thermoregulator. The desired acetone vapour was then obtained by flowing air at a constant rate through the Dreschel bottle regulated by an Alicat^® mass flow controller (MFC). The acetone vaporisation rate was calibrated using the long average weighting method with different flow rates at fixed bath temperatures. The temperature of the hot bath was monitored by an unsheathed fine gauge K-type thermocouple and the variation shown was ± 0.5 K. High temperature, chemical-resistant PTFE tube was used at the outlet of the Dreschel bottle for acetone vapour flow delivery and was heated with OMEGA^® rope heater. The partial pressure of acetone was maintained below the saturation pressure at any point along the line to prevent condensation. Measurements with a fast FID showed that the acetone vaporisation rate becomes inconsistent at high gas bubbling rates due to the disequilibrium between the liquid and air chamber. Hence, only low flow rates of air (<1 l/min) were used as bubbling air for acetone seeding. The vapour-pressure line and the calibration of the acetone vapour mass flow rate is shown in Appendix A2. Methane (99.5% pure; BOC) and primary air flow rates were regulated by Alicat^® and Bronkhorst^® MFCs respectively, which deliver ± 1 % full scale accuracy.

2.6.3 Acetone/air mixture preparation

At high flow rates, the vaporization rate in the Dreschel bottle was excessive and led to unstable performance. For these cases, the acetone bubbler method was replaced with a vaporiser fitted with an atomizer for high levels of acetone. The liquid acetone was delivered to the vaporiser using a high precision microannular gear pump via the chemical resistant Tygon^® tubing. No bubbles were observed along the fuel line, indicating a smooth and consistent delivery of fuel. The full scale accuracy of the pump

Experiments is ± 1 %. Part of the total air (~ 40 %) was diverted to vaporise the fuel while the rest (~ 60 %) was used to seed the flow. The air flow rates were separately regulated by two MFCs. The vaporiser was heated by an Omega^® rope heater and insulated using high temperature resistant material. A K-type thermocouple was used to monitor the temperature and the heating was controlled using a phase-angled power temperature controller. The temperature within the vaporiser was maintained at ~ 50 ^oC to ensure the partial pressure of acetone stays within the gaseous phase. The vaporised fuel was then mixed with the remaining seeded air in the mixing chamber before passing through the copper cooling coil upstream of the burner. The partial pressure of acetone was kept under saturation pressure along the line to avoid condensation. A K-type thermocouple was fitted at the burner body to monitor the mixture temperature. The schematic diagram of the system is shown in Fig. 2.2b.

2.6.4 PIV setup

The jet-wall stagnation velocity flow field was obtained using a planar PIV system. The mixture flow was uniformly seeded with submicron-size silica particles. A dual laser head, double pulsed Nd:YAG laser (Litron Lasers:NANO-L-200-15 PIV) was used to generate a light sheet with a thickness of ~ 0.5-1.0 mm to illuminate the uniformly dispersed particles in the flow. The vertical plane of the light sheet was generated from the laser beam by using a cylindrical diverging lens. The laser energy used in this experiment was around 40 mJ/pulse at 532 nm with ~ 4 ns pulse width.

The light scattered from the seeding particles in the flow was recorded by a 12 bit, 2048 x 2048 pixels Imager Pro X 4M CCD camera with a pixel pitch of 7.4 x 7.4 μm at double frame mode of 4.5 Hz. The camera was fitted with a 60 mm/F2.8 Nikkor lens coupled with an optical band pass filter centered at 532 nm to minimize the effect of flame luminosity. The timing of the laser system and camera was synchronized by a LaVision^® Programmable Timing Unit Version 9 (PTU 9). The commercial software Flowmaster from LaVision was used for image acquisition and the analysis of image pairs. The field of view was fixed at 35 x 35 mm with a magnification factor M equal to 0.43.

Experiments

2.6.5 PIV accuracy and uncertainty

The timing between PIV pulses ∆t (~ 150-200 μs) was chosen such that the particle movement in the reference flame speed region is within 1/4 of the interrogation window. The cross-correlation is performed using the adaptive multi pass with decreasing window feature, where the initial subregion of 64 x 64 is used before spatially window shifted to the final 32 x 32, with 50% subregion overlapped to optimize the spatial resolution of the velocity field. The seeding density is kept to around 6-10 pixels per subregion to obtain high quality PIV correlations. The peak-height validation method is applied where the ratio of highest peak to second highest peak in the correlation plane is kept at 1.2 for vector validation and to remove spurious vectors. The timing error for the laser is typically small (~ 4 ns) and does not contribute significantly to the error in velocity measurement. Another possible source of

error for PIV is the particle lagging effect. Based on the Stokes number Stk = ρ_pd²U_o/18μL_s, where U_o is the characteristic fluid flow and L_s is the characteristic

length scale, the calculated Stokes number for the particles in the flow is of the order of 10^-5 << 1, indicating the particle lagging effect is not significant. Peak locking effect due to the small signal of less than a single pixel is another possible source of error.

The full scale PIV measurement error can be determined by the ratio of the nominal correlation peak value (0.1 pixel) [43] to the maximum displacement permitted, namely 1/4 of the final interrogation window [44]. Since the final interrogation area used in all the image processing is 32 x 32, the accuracy of the axial and radial velocity measurement is determined as ± 1.25 % full scale. The number of image pair taken for each condition was at least 250 - 500.

Experiments

2.6.6 Laminar flame speed determination

The planar 2-D velocity vector field obtained using PIV is analyzed to determine the reference stretched flame speed and the imposed strain rate. An example of the 2-D velocity vector map and a flat flame image are shown in Fig. 2.3a and 2.3b respectively. From the burner outlet, the centreline axial velocity decreases towards the stagnation point, reaching a minimum velocity as it approaches the upstream boundary of the preheat zone, before accelerating through the flame due to thermal expansion. To determine the strain rate K, the centreline axial velocity profile from the velocity map is extracted, as shown in Fig. 2.3c. The minimum axial velocity upstream of the thermal mixing layer is identified as the reference burning velocity Sref. The velocity gradient immediately preceding the reference point is determined as the axial strain rate K. The flow pattern is reflected in the radial velocity profile at the reference position (Fig. 2.3d), where the linear radial velocity gradient can be more accurately be employed to determine the axial strain rate, as K = 2a [42].

Based on the variation of the reference flame speed with K, the unstretched laminar flame speed can be determined by using the methodology of either linear or nonlinear extrapolation to zero stretch rate. The non-linear extrapolation method was introduced by Tien and Matalon [45] due to the non-linear relation between S_ref and K as K → 0 based on the asymptotic analysis using potential flows. However, Vagelopoulos et al. [28] showed that if the Karlovitz number Ka = α_mK/(S_u^o)², is lower than 0.1, where α_m is the thermal diffusivity of the unburnt mixture, laminar flame speeds obtained from the linear extrapolation under these conditions yield sufficiently accurate results. The typical range of strain rates of ~ 100-250 s^-1 derived from these experiments yield Ka<0.1, so the linear extrapolation is sufficient.

Experiments

Figure 2.3: (a) Velocity vector map for an image pair (b) Image of a jet-wall stabilized flame (c) Axial velocity profile indicating the reference flame speed (d) Radial velocity profile used to derive the strain rate.

Burnt product

Radial profile at location Sref

(a)

Unburnt

reactants Centerline Axial profile L

D

L

D

-10 -5 0 5 10

-0.4 -0.2 0 0.2 0.4

Radial velocity (m/s)

Radial coordinate (mm) Center of burner

a Radial vel.

gradient (d)

0 5 10 15 20

0.1 0.4 0.7 1 1.3 1.6

Axial velocity (m/s)

Distance from nozzle (mm)

S_ref

Axial vel.

gradient ^K (c)

(b) Wall

Numerical analysis

0 50 100 150 200

10 20 30 40 50

Strain rate K (s^-1)

Reference flame speed (cm/s)

S_u^o linear

φ=1.07 φ=0.9 φ=0.8

S_u^o nonlinear

Figure 2.4: Example of (─) linear and (---) non-linear extrapolation to obtain the unstretched laminar flame speed for 9% acetone/methane/air mixture.

The extrapolations of unstretched flame speed have been performed using both the linear and non-linear method. The linear extrapolation yields slightly higher unstretched flame speeds by ~ 1-2 cm/s as demonstrated in Fig. 2.4, which is within the experimental uncertainty. Chao et al. [32] reported that the accuracy of linear extrapolation method can further be improved by increasing the nozzle-plate separation distance L or decreasing Ka. In this experiment, the use of a 22 mm diameter nozzle allows a relatively larger nozzle-plate distance as compared to the 14 mm diameter used in [42, 46]. The current jet-wall setup also provides a lower strain rate range than the opposed-jet configuration [30] and subsequently lower values of Ka, which favours the use of the linear method in deriving the unstretched laminar flame speed. Therefore, the subsequent flame speed results are reported based on the linear extrapolation method.