Turbulent separated inert and reactive flows over a triangular bluff body

Application Challenge AC2-12 © copyright ERCOFTAC 2019

Test Data

Overview of Tests

The description of the validation test rig strictly follows the original papers [1,2,3]. The test data contain LDA, CARS and gas analysis flow measurements for several operational conditions. For the sake of completeness, there are several alternative / additional experimental data (provided below), which replicate the inert and reactive bluff-body flows and can be used independently or in addition to the Volvo test – rig data.

A flexible modular combustor with optical access has been developed to generate experimental data for model validation. It was designed to enable the use of non-intrusive optical measurement techniques and to allow various combustion systems to be studied in an idealized fashion. The test set-up consists of a straight channel, with a rectangular cross-section, divided into an inlet section and a combustor section as shown in Fig. 3. The inlet section is used for flow straightening, turbulence control as well as fuel and seeding injection. A triangular – shaped bluff body was used for flame stabilization as shown in Fig. 3.

The air entering the inlet section is distributed over the cross-section by a critical orifice plate that, at the same time, isolates the combustor acoustically from the air supply system. Gaseous propane is injected and premixed with air 0.07 m downstream of the critical orifice plate by a multi-orifice, critical flow, fuel injector. The turbulence level in the combustor inlet is controlled by installing grids, honeycomb and/or screens at several axial locations in the inlet section. The examined experiments were run with premixed propane-air mixtures. A premixed combustion system was chosen to simplify the experiment, avoiding gradients in the fuel-air mixture and the influence of droplet evaporation or high speed jets from gaseous fuel injection.

The validation rig was designed to make several hours of test runs possible in order to allow for extensive traversing with a variety of sampling measurement equipment. The combustor section is of a modular design with the walls split into a number of interchangeable sections. The side-wall elements are either air cooled quartz windows for optical access or they may consist of pure water cooled sections. The upper and lower walls are water cooled, but may be replaced by other designs to make other laser diagnostic techniques, such as laser induced fluorescence, LIF, or particle image velocimetry, PIV, possible. The combustor section ends in a circular duct with a larger diameter and the acoustic outlet condition can be considered as a “sudden expansion”. The flame holder used in the present study had a triangular shape (Fig. 3) and was chosen because of the simple geometry, the resemblance to real afterburner geometries and because strong instationary flow and combustion phenomena were bound to occur.

Gas analysis and instrumentation

Local concentrations of combustion products were sampled with a traversing probe. The probe was mounted in the center of the combustor and could be used at three different axial positions where vertical profiles of combustion species could be measured. The gas analysis equipment consisted of chemiluminescence analyzer for Nox, two nondispersive infrared instruments for CO and CO₂, a paramagnetic analyzer for O₂ and a flame ionization detector for total unburned hydrocarbons (UHC). All instruments, except for the UHC detector, were able to cover the whole range of concentrations in the flame. The maximum level of UHC that could be measured with reasonable accuracy was 4000 ppm.

The water content, equivalence ratio, wet species concentrations and flame temperature were calculated from the measured gas composition, assuming equilibrium chemistry and conservation of species. The accuracy of a single species concentration measurement was estimated to be 5% and the accuracy of the equivalence ratio and calculated flame temperature to be about 10%. Fluctuations of the static pressure were measured with a piezo-electric pressure transducer located at the bottom wall, 0.03 m downstream of the reference position. Frequency spectra were recorded and analyzed with a HP 355660A analyzer. The airflow and fuel flow were measured with critical and sub-critical orifices yielding an accuracy of 3% respectively.

Figure 3: Schematic drawing of the validation rig

The combustor was run over the range conditions: combustion pressure 100 kPa, inlet temperature 288 – 700 K, air mass flow 0.6 – 1.8 kg/s and equivalence ratio 0 – 1.1. These conditions yield an inlet Mach number range of 0.04 – 0.2 and a Reynolds number range, based on the hydraulic diameter of the combustor, of 10,000 – 50,000. The selected conditions for the gas analysis tests are shown in Table 2.

All tests were conducted with non-vitiated air. Propane, 98% pure, was used as fuel. To reduce the risk of leakage at the quartz windows, the tests were always conducted slightly below ambient pressure. The turbulence level just upstream of the bluff-body was kept at an approximate level of 3-4%, as verified by the LDA-measurements. The selected combustor geometry is prone to combustion oscillations. However, the equivalence ratio for the gas analysis tests were chosen so that the RMS value of the dynamic pressure fluctuations were below 1% of the absolute pressure level.

Table 2: Gas analysis test conditions.

The gas analysis data have been performed at the following axial positions: x = 0.15 [m], x = 0.35 [m], x = 0.55 [m].

The origin of the co-ordinate system is in the center of the flame holder base surface as indicated in Figs. 1 and 3.

The files referred to below contain the following measured parameters: Y-position [mm], inlet temperature [K], equivalence ratio [-], mole fractions of CO [%], CO₂ [%] and O₂ [%], NOX [ppm], Temperature [K].

LDA (Laser Doppler Anemometry) measurements

A two-component LDA system was used for the measurements. The system consisted of a 5 W Ar+ laser with a fiber optic link to the transmitting optics where two beam pairs were formed. The geometry of the Validation Rig allowed forward scattering to be used, which gave higher performance compared to back scattering. Both the transmitting and receiving optics were mounted on a positioning table to ensure alignment. Two photomultiplier tubes (PMT) were mounted directly at the receiving optics and the signal was fed to the signal processing equipment and data acquisition system which were placed in the control room. The PMT signals were ﬁrst treated by specially designed acoustic-optic preprocessors that suppress noise. The frequency determination was then done by two counters (Dantec 55L90a). It must be pointed out that the preprocessors had to be used to get a reasonable data rate with the counters under these measurement conditions. The Doppler frequency data were collected together with the arrival time on a general purpose data acquisition system. Velocity determination and validation were performed on a PC. Small seeding particles were used to be able to capture turbulence information. The particles were generated by a ﬂuidized bed ﬁlled with 0.05 μm gamma alumina particles.

Temperature gradients during combustion caused index refraction variations in the gas, thereby bending of the laser beams. Therefore the probe volume had to be comparatively large and a beam diameter of 300 μm of the probe volume was chosen to ensure that the beams crossed. Coincidence ﬁltering was used to obtain simultaneous readings of the two components. The coincidence criterion was that the two counters should be active at the same time. A velocity pair was accepted only if the latter counter had started to count before the ﬁrst one was ready and that they both yielded a validated velocity.

Table 3 shows the selected test cases. The flow velocities, velocity fluctuations and the frequency spectra have been measured. All tests were conducted at atmospheric pressure and with nonvitiated air. Propane, 98% pure, was used as fuel for the reacting test cases. The inlet turbulence level was controlled by honeycombs and screens which gave an approximate level of 3 - 4%. The selected combustor geometry is prone to combustion oscillations. However, the equivalence ratio was chosen so that the RMS value of the dynamic pressure fluctuations was below 1% for all selected test cases. The reference temperature was 288 K (600 K for the preheated case) and the reference pressure was 100 kPa at a distance 200 mm upstream of the reference position.

Table 3: LDA test conditions.

The LDA measurements have been performed at the following axial positions and the centerline: x = -0.2 [m], x = -0.1 [m], x = 0.015 [m], x = 0.038 [m], x = 0.061 [m], x = 0.15 [m], x = 0.376 [m], y = 0.0 [m] (Centreline).

The origin of the co-ordinate system is in the center of the flame holder base surface as indicated in Figs 1 and 3.

The files referred to below contain data of the following measured parameters: mass flow [kg/s], equivalence ratio [-], X-position [mm], Y-position [mm], Temperature [K], axial velocity [m/s], normal-velocity [m/s], axial velocity RMS [m/s], normal velocity RMS [m/s].

CARS (Coherent Anti-Stokes Raman Scattering) measurements

Temperature field measurements have been obtained using the CARS technique.

The 2λ-CARS system

Figure 4: . The difference between conventional CARS and 2λ CARS

The CARS principle can briefly be described as follows. By exciting N₂ molecules, under certain geometrical constraints, with a short pulse (~10 ns) of intense laser light of specific wavelengths, the molecules will emit light spectra that contain information on the temperature. Fig. 4 illustrates the difference between conventional CARS and the two-wavelength CARS used for these measurements. In 2λ-CARS the broadband dye laser is replaced with a narrowband, twowavelength dye laser beam. An improvement of about 35 times is obtained in 2λ-CARS as compared to conventional CARS since the laser power can be concentrated on two wavelengths. The higher signal intensity of the 2λ-technique was used to decrease the probe volume by increasing the beam separation.

By dividing the amplitude of the two peaks measured by the receiver and comparing the ratio to a precalculated theoretical temperature dependent ratio, the temperature could be evaluated one order of magnitude faster than with the standard method of comparing full theoretical spectra. To handle the problem with pulse to pulse variation in the laser, a reference oven was used to separate the variation from the true CARS signal. The non-resonant background was not separated from the signal. Instead, a calculated value of the background was used in the evaluation program, based on the expected mixture of the combustion gases for each fuel-air ratio. An error of 15% in the estimate of the non-resonant background gives an error of 1% in the temperature.

The system layout and the main dimensions of the mobile CARS unit are shown in Fig. 5. The unit is based on a Quantel SA 10 Hz YAG/dye laser system. The two optical bread boards, containing the laser system and the receiver, are placed on a remotely controlled traversing unit. This unit, positioning the CARS probe volume, moves the laser system and the receiver simultaneously. The position of the probe volume is determined with magnetic sensors to an accuracy of less than 5 mm.

Figure 5: The CARS system setup.

Table 4 shows the CARS test cases. Measurements were carried out for the temperature field for three inlet conditions that were also used for velocity measurements with LDA.

Table 4: Test cases for field measurements with CARS.

Alternative/Additional experimental data

It is worth noting that several alternative experimental data exist in the literature, which can be used as alternative or/and accompanying data to the considered measured data [1,2,3].

Shanbhogue et al. [5] presented a very useful review and experimental observations of the lean blowoff of bluff-body stabilized flames with a detailed description of instabilities in both inert and reactive wakes. Fujii et al. [6,7] presented LDV measurements (both inert and reactive), which were carried out in an open circuit, forced flow type of wind tunnel at ambient conditions (P = 100 kPa, T = 280 K, Re = 17500 and M = 0.03). An equilateral (D = 0.025 m) triangular rod was placed inside the channel passage of 0.05 m square cross-section. Nakagawa [8] presented an experimental interferogram useful to compare the flow visualization of the vortex shedding behind a triangular rod. Experimental data of the surface pressure coefficient distribution on the sides of a triangular rod can be retrieved from the work published by Tatsuno et al. [9]. Sanquer et al. [10] presented details of energy spectra in the turbulent non-reactive and reactive wakes of the bluff-body flow, including triangular and square cylinders.