Influence of Turbulent Flow on the DeflagrationtoDetonation Transition in Hydrogen–Oxygen–Air Mixtures in a Pulse Combustor M. S. Assada, Kh. Al’khusanb, O. G. Penyaz’kova, and K. L. Sevruka a
Lykov Institute of Heat and Mass Transfer, National Academy of Sciences of Belarus, Minsk, Belarus National Center for Aeronautics Technology, King Abdulaziz City for Science and Technology, Saudi Arabia email: [email protected] Received June 14, 2012; in the final form, July 16, 2013
Abstract—The effect of turbulization of a hydrogen–oxygen–air mixture flow on the deflagrationtodeto nation transition in a pulse combustor (PC) is studied. The parameters of operation of the PC with flame front propagation in a quiescent and strongly turbulized mixtures (Re Ⰷ 104) are compared. It is shown that, in case of a quiescent mixture no detonation occurs because of a small length of the PC. The presence of intense pulsations (Re > 2 ⋅ 104) created by elements of special configuration in the mixing chamber promotes the formation of a detonation wave, the velocity of which depends on the fueltooxidizer equivalence ratio. Keywords: deflagrationtodetonation transition, turbulence, predetonation distance, pressure, detonation velocity, thrust DOI: 10.1134/S199079311402002X
Recent years have seen intense research in the field of detonation combustion of fuel in various technical facilities, particularly in pulse detonation engines for obtaining high specific power and thrust [1–9]. The idea of using detonation combustion in power plants has been suggested as early as 1940 by Ya.B. Zel’dovich . Such great interest in detonation combustion is due to a higher thermodynamic cycle efficiency as compared to the known practically implemented cycles (Brighton, Humphrey, etc.) and the possibility of achieving better performance (thrust, power, etc.).
The experiments were performed using a pulse combustor (Fig. 1) comprised of operating (1) and mixing (2) chambers, a system (3) for supplying the oxidizer (oxygen, air) and fuel (hydrogen), a system for ignition of the combustible mixture, and a thrust transducer (not shown). The pulse combustor was rig idly fixed on platform 4, moveable only along base 5. For managing and monitoring the PC facility, it was equipped with a gashandling panel, control cabinet, and indication and data acquisition system. Operating cham
To develop an effective pulse detonation engine, it is necessary, in the first place, to provide a proper orga nization of the operation process, determine its pat tern, and establish the relationships between various fac tors and parameters affecting the ignition, combustion, and onset of detonation, such as the method of supplying the components, flammable mixture formation mode, geometric characteristics and design features of the com bustion chamber, thermodynamic characteristics, hydro and gasdynamics phenomena, etc. The present work investigates the effect of turbu lence of the combustible mixture under conditions of weak ignition on the deflagrationtodetonation tran sition in a pulse combustor (PC), a prototype of the pulse detonation engine. Mixtures of hydrogen with oxygen and air in the proportions specified in the table were tested. 181
Fig. 1. General view of the pulse combustor.
ASSAD et al.
Experimental and calculated values of the key parameters of the PC workflow for a quiescent and turbulized mixtures of hydrogen with oxygen and air Cycle Experimental repetition conditions frequency, Hz
D, m/s φ
base 2 base 3 base 1 mean (L = 157 mm) (L = 157 mm) (L = 157 mm)
2.5 2.86 2.86 2.86 2.86 2.86 2.86
0.943 0.982 1.02 1.524 1.814 0.673 0.737
1256 1236 1172 1342 1266 1098 1198
1670 1342 1180 1331 1353 1040 1524
1 1 1 1 1
1.02 1.02 1.524 1.814 0.737
ber 1 was a sectionalized steel tube, forming, together with the cavity of mixing chamber 2, a 740mmlong smooth detonation tube with a constant cross section of diameter d = 25 mm. The PC design provided a separate supply and mix ing of the combustible mixture components (hydro gen, oxygen, air) in mixing chamber 2, which housed three special configuration sections planted on a semi axis with a swirler and connected with the gas supply lines through fastacting valves. The joint action of the mixing chamber elements ensured an intense turbuli zation and swirling of the flow. The end face of the mixing chamber played the role of a thrustproducing wall, at which an automobile spark plug was mounted. To ensure a simultaneous filling of the detonation tube with the combustible mixture components, hydrogen was supplied somewhat later because of its higher mobility. Depending on the mixture composi tion and the amount of each reagent, the delay was determined and controlled with the help of a PCcon trol software code. The ignition system consisted of an ignition coil, electronic switch, and automobile spark plug. The control cabinet included a commercial com puter, oscilloscope, controller, two power supplies, auxiliary relays, and converters, required for a proper functioning of the PC. The PCcontrol code provided the input of the required initial workflow parameters over a wide range of conditions, generation of com mands for respective actuators, monitoring of the time evolution of various characteristics of the process, and processing and storage of the data obtained.
2453 2453 2379 2707 1805 1172 2211
maxi DCJ, m/s u, m/s mum
0.219 0.179 1.712 0.197 0.139 0.362 0.456
1.795 1.857 3.612 1.837 1.716 2.161 2.221
0.008 0.113 0 0 0.0003
0.143 0.555 0.441 0.319 0.0554
2289 2327 2332 2523 2593 2131 2211
50.34 51.36 48.14 56.08 60.30 43.24 47.15
Re 22293 22850 21836 22242 22891 22672 23312
EXPERIMENTAL PROCEDURE To determine how turbulence influences the defla gration todetonation transition, two modes of the combustion process in the PC were realized: (1) in a turbulent flow of the combustible mixture created over the entire length of the PC by a complex regular geometry of the mixing chamber elements; (2) in a quiescent combustible mixture, immobility of which was provided by closing the outlet section of the PC with a Mylar diaphragm of thickness ≈0.1 mm, which was torn after ignition by rising pressure in the chamber. The diaphragm was placed at a flange equipped with a ball valve for purging and displacing the atmospheric air and residual gases of the previous cycle with the combustible mixture. The gases com prising the combustible mixture were admitted to the PC through the mixing chamber for 1 s while the ball valve was open. Then, the ball valve was closed and the quiescent mixture in PC was ignited. The evolution of the process was monitored with four PCB Piezotronics piezoelectric pressure sensors and 54 ionization sensors spaced along the length of the PC. The pressure sensors were installed in line along the body of the operating chamber at distances of L = 157 mm apart. The distance from the spark plug to the nearest pressure sensor was 195 mm. The ionization sensors were installed at the working chamber casing in an orthogonal manner, at the points of intersection of nine equidistantly positioned sections with six longitu dinal lines evenly spaced (every 60°) around the perim eter of the tube, numbered counterclockwise (Figs. 3b and 4b). Thus, the ionization sensors formed eight identical measuring bases of length L = 78.5 mm each.
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY B
INFLUENCE OF TURBULENT FLOW ON THE DEFLAGRATIONTODETONATION
RESULTS AND DISCUSSION The conditions of ignition and the values of the detonation velocity D and thrust over a wide range of equivalence ratios φ for the propagation of the wave in mixtures of hydrogen with oxygen and air are pre sented in the table. It lists the calculated values of the Chapman–Jouguet DCJ velocity for the same condi tions. Figure 3 shows the dependence of the wave velocity for an intensely turbulized (Figs. 3a and 3c) and a quiescent (Fig. 3b) hydrogen–oxygen–air com bustible mixture with φ = 1.02. Combustion of a quiescent mixture. Experiments demonstrated that, in the absence of turbulent flow in the combustor, the combustion wave is somewhat accelerated, with the velocity reaching a maximum (~570 m/s) within the first measuring base, but it is insufficient for the formation of a shock wave strong enough to initiate detonation. As shown in Fig. 3b, the expanding combustion products generate compres sion waves and cause the movement of the gas ahead of the flame front; however, strong shock waves needed to accelerate the gas and to initiate detonation do not arise due to the lack of time and insufficient length of the PC. This behavior is explained by the fact that, when a quiescent mixture is ignited by a weak source, a laminar flame front arises, which moves at a low speed relative to the gas ahead of it. Despite the fact that a smallscale turbulence arises, which, according to Damkohler and Shchelkin , extends the com bustion zone and enhances the heat and mass transfer, the absence of intense vortex pulsations (largescale turbulence) in the flow and an insufficient length of the PC make the transition to detonation impossible. Deflagration to detonation transition in a moving mixture. In the presence of intense vortex pulsations in the combustible mixture flow with the same equiva lence ratio φ as in the case of a quiescent or lowturbu lence mixture, the flame front velocity increases RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY B
1000 Flow rate, L/min
The fueltooxidizer equivalence ratio was calcu lated from the respective thermochemical equations. The amounts of oxygen, air, and hydrogen for a given interval of the cycle were determined from the mass flow meter readings multiplied by the duration of opening of the respective fastacting speed inlet valve connected to the mixing chamber of the PC. As an example, Fig. 2 shows the time histories of oxygen, air, and hydrogen consumption over a cycle. The cycle interval of interest to us, indicated by the vertical lines, corresponds to φ = 1.814. The experimental procedure involved control and variation of the initial parameters of the process, such as the mass flow rates of oxygen, air, and hydrogen, the total number and repetition frequency of cycles, the duration of the voltage pulse applied to the spark plug, the time delay between the start of gas introduction (oxygen, air and hydrogen) and the ignition of the mixture.
200 0 0
200 300 Time, ms
Fig. 2. Example of measuring the volumetric flow rates of hydrogen, oxygen, and air for a 59.9% H2–16.6% O2– 23.6% N2 mixture (φ = 1.814).
manyfold, and the onset of detonation occurs in almost all cases at a distance of 20 to 22 diameters of the combustion chamber, i.e., at 500–550 mm from the ignition source (Figs. 3a and 3c). Note that, while at φ = 1.814, the wave velocity at the third measuring base reaches 1805 m/s (less than the Chapman–Jou guet velocity), at φ = 0.982, the detonation wave at the same distance from the ignition source becomes over driven, propagating at a velocity of 2453 m/s. This means that, for a stoichiometric mixture, the positive influence of turbulence on the process of detonation onset is stronger for lean and rich combustible mix tures. Figure 4a shows the dependence of the detonation wave velocity on the distance along the tube and on the angle of positioning of the sensor over the perimeter of the measuring cross section for a rich (φ = 1.52) mix ture of hydrogen with oxygen and air. It is seen that, at the same moments of time, detonation wave passage was recorded at different points along the length and cross section perimeter of the tube. This indicates that the detonation wave has a complex structure, with its velocity varying not only along the length of the cham ber, but also in time and space. For a flame path length from the spark plug to the outlet to the atmosphere of 740 mm and a cycle repe tition frequency of 25 Hz, for example, the flame front should propagate at an average velocity of not less than 18.5 m/s. This is about an order of magnitude higher than the laminar flame speed in PC conditions. Such a high velocity can only be achieved in a highly turbu lized combustible mixture, in which the temperature, pressure, and gas velocity increase sharply, while the flame front becomes curved, with a larger surface area, thereby causing the normal burning velocity to increase to a supersonic level. As a result, compression waves catch up with each other, forming a shock wave, which initiate detonation. Vol. 8
ASSAD et al. 50
40 2309 m/s 30
1189 m/s 1189 m/s
4 3 2 1 0
30 Time, ms
6000 Line 6 5000
Line 1 Line 2
Line 5 Line 4
Line 3 Velocity, m/s
3000 DCJ 2000 2309 m/s 1000
1189 m/s 0
600 Distance, mm
Fig. 3. Diagram of combustion wave propagation in a .47% H2–23.1% O2–29.9% N2 mixture (φ = 1.02): (a) turbulent flow, (b) quasiquiescent mixture (signals from pressure sensor), and (c) detonation velocity for turbulent flow (signals from ionization sensors).
The degree of turbulence in the mixture in the PC was assessed using the Reynolds number Re, which is expressed through the average flow velocity u , kine matic viscosity νΣ of the medium, and the diameter d as
Re = ud . νΣ The average flow velocity was calculated from the consumption rate of reactants (Qair, QH 2 , QO 2 ) and the crosssectional area S: QΣ Qair + QH 2 + QO 2 = , S S whereas the kinematic viscosity of the gas mixture νΣ (m2/s) was calculated as the sum of the partial kine matic viscosities of the components: u =
= rair ν air + rH 2 ν H 2 + rO 2 ν O 2 ,
where rair , rO 2 , and rH 2 are, respectively, the fractions of air, oxygen, and hydrogen in the gas mixture. The values of Re and the mean flow velocity for the investigated gas flows in the PC are given in the table. The large values of the Reynolds number (Re Ⰷ 104) are indicative of a high intensity of turbulent pulsa tions in all cases in the combustion gas mixture flow. This is, obviously, a key factor in the acceleration of the flame front and in the formation of detonation waves, both in time and in space. This conclusion is supported by the results of , according to which, in the case of intense largescale turbulence (with a scale much larger than the combustion zone thickness), the flame propagation velocity is determined only by the
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY B
INFLUENCE OF TURBULENT FLOW ON THE DEFLAGRATIONTODETONATION (a)
360° 300° 240° 180° 120° 60°
2000 0 Base 1
Base 3 Base 2 Base 5 Base 4 Base 7 Base 6 Base 8
/ Velocity, m
(b) Line 6 Line 1
Line 5 360° 60°
turbulence on combustion and transition to detona tion in a smalllength PC (L = 740 mm). It was shown that: (1) for a quiescent mixture and a weak ignition source in the PC, the flame front accelerates, but no transition to detonation occurs; (2) for a highly turbulized combustible mixture flow (Re > 2 ⋅ 104), detonation onset occurs due to intense pulsations and heterogeneity of the flow, fac tors that reduce predetonation distance and accelerate deflagration to detonation transition. Such intense turbulent flow is created by specialshape elements in the mixing chamber. ACKNOWLEDGMENTS The authors are grateful to the King Abdulaziz Center for Science and Technology (Saudi Arabia) and Lykov Institute of Heat and Mass Transfer of the National Academy of Sciences of Belarus for financial and scientific support.
120° 240° 180°
Line 4 Line 3
Fig. 4. Spatial propagation of a detonation wave in the tube at φ (signals from ionization sensors): (a) time evolution of the detonation wave velocity and (b) positioning of the lines of sensors over the tube cross section perimeter.
fluctuating turbulent velocity, being independent of the laminar flame speed, i.e., of the chemical factors, such as the mixture composition, type of fuel, etc. As regards the thrust, for a quiescent mixture, it is practically absent (table). For the combustion of a tur bulized mixture, it exhibits a mixed behavior, depend ing on the mixture composition, stoichiometry coeffi cient, detonation velocity, and the pressure developed in the course of the process. CONCLUSIONS The studies performed revealed facts important for assessing technical manifestations of the influence of
RUSSIAN JOURNAL OF PHYSICAL CHEMISTRY B
1. V. V. Mitrofanov and S. A. Zhdan, in Advances in Con fined Detonations, Ed. by G. D. Roy, S. M. Frolov, R. J. Santoro, and S. A. Tsyganov (Tourus Press, Mos cow, 2002), p. 199. 2. S. Eidelman and X. L. Yang, AIAA Paper No. 1998– 3877 (1998). 3. W. Heiser and D. Pratt, J. Propuls. Power 18, 68 (2002). 4. R. Mohanraj and C. L. Merkle, AIAA Paper No. 2000– 16225 (2000). 5. S. Yungster, AIAA Paper No. 2003–1316 (2003). 6. S. M. Frolov, V. S. Aksenov, and V. S. Ivanov, Russ. J. Phys. Chem. B 5, 664 (2011). 7. Pulse Detonation Motors, Ed. by S. M. Frolov (Torus Press, Moscow, 2006) [in Russian]. 8. G. D. Roy, S. M. Frolov, A. A. Borisov, and D. W. Netzer, Progress Energy Combust. Sci. 30, 545 (2004). 9. V. A. Levin, I. S. Manuilovich, and V. V. Markov, Com bust. Explos., Shock Waves 46, 418 (2010). 10. Ya. B. Zeldovich, Zh. Tekh. Fiz. 10, 1453 (1940). 11. K. I. Shchelkin and Ya. K. Troshin, Gasdynamics of Combustion (Akad. Nauk SSSR, Moscow, 1963) [in Russian].