Preliminary Investigation of Flame Configurations Formed by Sonic and Subsonic Hydrogen Jets

Page 1

F BY SONIC AND SUBSONIC EZERCGEN ,- 11 4 tstx,„,---, has be conducted to obtain a qualitative under- i L. standing, of :lame c=fiEtz-,stions formed by kfdrogen jets. Results suggest ‘1,1 finct=te Aril mtensionn of the Enr.11-scale test are prose. SAN D144. SYSTEMATIC DECLASSIFICATION REVIEW OP. DECLASSIFICATION STAMP

Page 3

Faperimentation was necessary to establish correctnocs of this model as well as to ascertain, arm:ling the model to be correct, the number of nozzle diameters downstream to the mass :axing phase. In addition, the diameter and length of the flame as a function of nozzle diameter and cond".tions were studied. A crude experiment was set up to =mine the approximate variation of flame configuration frem a hydrogen jet as a function of stagnation pressure, density, and nozzle diameter. This preliminary study was intended to provide an immediate tentative answer to the testing problem as well as to examine feasibility and value of further model studies. Figure 1 depicts erperimental arrangements. Hydrogen passes through control valves out pert the heater to the exit nozzle. Nitrogen was provided for rinsing the system before and after runs. To obtain higher exit velocities the heater was employed in the flow system forward of the upwar3-directed exit nozzle. Stagnation pressure ahead of the nozzle was detervined by a head probe situated above the boater. This pressure was communicated to a Heise gage on the control board. Heater power was monitored by means of a voltmeter and an ammeter also located on the control board. Power ineut was controlled by a variac. An unsuccessful feature of the systm was a salt-water injector installed at the controls. It was hoped that the injector would render the flame lumincua by forming smell salt particles through spray evaporation. Apparently, however, most of the salt particles were deposited in the line to the nozzle, and only 0 er, ell rannunt of flame colcration was achieved. of the hydrogen jet was achieved by using electrieery heated wires. The wines were attached to t terminnl board in such a manner that ignition could be attempted at varicus heiehte above the nozzle. The entire board could be moved up and down, as well. as in and cut, from the nozzle. Igniting the hydrogen jet pro' ed simple. V-ehaped niehrom, wires were situated with the point in line with the nozzle edge. The heater wires were oveTvolteged ao that, after imgmitlos.,

Page 5

they melted and fell out of the flame, thus avoiding the possibility of their acting as flame holders. The hydrogen flame was not visible in daylight, although achievement of ignition was apparent from the roaring of the flame. Ignitions were accomplished from 2/8 to 6 inches above both 1/6— and 1/16—inch nozzles, using excess chamber pressures of 12, 6, and 3 psi. However, when using the 1/16—inch nozzle and an excess chamber pressure of 18 psi, difficulty was encountered in achieving a stable flame. Although the hot wire ignited the jet, the flame went cut as soon as this wire melted and dropned. No further attempts were made to secure a stable flame. It should be noted that while the hydrogen jet proved easy to ignite, no spontaneous ignitions occurred. Apparently, the jet temperature was in all cases below the auto ignition tee:pc:nature (1050°C), even though low flew rates and high heater currents were empleyed. 3se Ovafieuretien and Dimenslens As mentioned in the previous section, the hydrogen flame was invisitle in daylight, although dirtertien effects brought about by the hot burned gases were apparent. Flare characteristics were examined by photography at night. A canvas screen was mounted about the jet stand to reduce wInd interference. Photographs were obtained by Mr. Dale Fantle of the Optical Meacurements Divimicn, 5216, who used two Graflex cameras equipped with remote—control solenoid shutter trigFlame dimensions and run conditions are ennumerated in Table I. A typical flame photograph (tleck and white coey of color film negative) is shown in Fig. 2. Table Il contains estimated initie1 jet velocities and densities, and mass flow computed by two methods. Computaticnal methods, together with sources of the obvious discrepancies, are discussed in Appendix A. Viewed in darkness, burning was observed to start abruptly some distance above the nozzle. Az inner cone of unburned gas wan visible in the lower portion. Fr- m initial burning it exp. paneled slowly to achieve maximum breadth which was maintained for a considerable length Alto the

Page 6

Table I - RUN COUDITIOIZ AND FLAME DIMAISIONS 2 12 1/16 215 1.88 0.78 3.13 1.. 7.5 varied during run

Page 7

Table IT - COMPUTE) NOZZLE fart CONDITIONS gamputed fro Laval teuatiol. Nara Flow from Tank Pressure Change

Page 8

Fig. 2 - A Typical Flame flame. It then pulled into complete the visible limit in an indistinct fashion. At low stagnation pressures the shape was markedly necked down to the base of the flame. The lower portions of the flame were light blue while the upper showed tinges of orange. This flame configuration proved very stable. Viewed fran a few feet away, no flicker or movement was apparent. No attempt was made to measure the flame temperature; it was apparent, however, that its maximum was in excees of 1350°C, since extraneously placed nichrome heater wires melted in the flame. Distances from nozzle to lower flame edge, as well as initial and maw-tam flame breadths, are plotted in Fig. 3. Distances to the maximum flame breadth are plotted in Fig. 4. Several interesting trends are revealed here. Flame geaetry is chiefly dependent on excess stagnation pressure, since it is apparent that it is only weakly, if at all, dependent upon exit velocity or to,:peratlYr i par a:. In addition, the distance to the flame from the orifice dots not seal with nozzle diameter, but stems to be independent of it. Tbe same appeere true of the initial flame breadth, eepsoially if the difference in novae diameters is taken into account. rins.117,

Page 9

EXCESS STAGNATION PRESSURE (psi) FIG 3 — PLOT OF VARIOUS FLAME DIMENSIONS VS EXCESS STAGNATION PRESSURE

Page 10

EXCESS STAGNATION PRESSURE (psi) FIG. 4 - PLOT OF LENGTH TO MAXIMUM FLAME BREADTH VS. EXCESS STAGNATION PRESSURE

Page 11

-the maximum diameter flEi its distance to the nozzle scales with orii7ice diameter within the limits of this emperiment; they also show a weak dependence on excess chamber pressure. Distance from the nozzle to the fuzzy flame termination has not been plotted. From Table II it is apparent that this quantity does not scale directly with nozzle diameter. Observation of these trends, as shown in Figs. 3 and 41 together with flame photographs, suggests that the seer flame boundary is much the same regardless of stagnation pressure, temperature, and, to a certain extent, orifice diameter. This is significant in that it ineicates that the full-scale test may be expected to display a cone-shaped flame rather than one suggested by visible behavior of dhamieel rocket exhaust. Other studies of subsonic jets dealing with velocity concentrations, etc., have yielded results similar to those of this experiment . 2/ While supersonic exhaust is undoubtedly a complication, mixing along the outer surface is still to be expected; it is this mixing and slowing action that determines the lower flame boundary and geometry. The ranee of the present e-rncriment is too Baited to permit wide claims. However, by assuming that the mixing of the supersonic jet is similar, although not identical, to that of the subsonic jet, approximate full-scale behavior may be predicted as follows: 1. The distance of the flr•,c boundary above the nozzle will net scale with diameter, but will prove a weak function of exit jet dynamic pressure. 2. The breadth of .the lower edge of the flame will be that of the nozzle plus a matter of 3. While rc firm basis exists for estimating the flame configuraticn during the superscnic phase, the flame can be expected to expand into a cone after the jet becomes subsonic. The total cone angle will probably be somewhat less than that observed in this study (13-15 degrees}.2/ 4. EXpansion into a cone will continue until a me/Amum breadth is achieved; the maximum breadth will scale approxlmtely with orifice diameter and show a weak dependence on mess flux and/or dynamic pressure at the nozzle exit. 5. Detailed behavior of the flame beyond zmoatmml breadth has not3 be determined, Qualitatively, it would be expected to form a cylinder for some length and finally clone on itself at a height which cannot be assigned from this atudly,

Page 12

6. Ignition of the full-scale hydrogen flame should prove simple. A pilot flame or hot wire situated a diameter or more above the nozzle lip Should suffice. Once ignition is achieved, the flame should maintain itself. The work described herein is preliminary; a clean-Lv phase is needed to refine results and remove the objectionable features discussed in the appendix. At the same time the experiment should be expended to include a wider range of nozzle sizes, fuels (ammonia as veil as hydrogen), and stagnation conditions. Prior to performing an expanded averiment, however, assurance of the results of model studies scaling are needed. Theory and other everiments suggest that approximate sealing is

Page 13

In Table II there are discrepancies between mass flow rates as determined by changed of tank pressure and those computed from theory. These discrepancies, as well as the theoretical basis for the cemputations, are considered in the following paragraphs. The method of computing the mass flow rate from tank pressure changes is straightforward. The perfect ger low is used to compute the mass of gas used during the timed run. Fetally straightforward is the method of computing mass flux from stagnation conditioes forward of the nozzle. The Laval nozzle equation is used: Here N is the mass flow rate, A is the nozzle area, Y is the ratio of specific heat capacitors, o is stagnation pressure P o is stagnation density, and P is ambient pressure. This equation For lecrer van? ue:7 of P/Po the mass flow Is cenztant for given stagnation conditiene because of In this ezperiment eteenatien erecslu•e as measured; etagnetion temperature andi density were not. If no heater power is rupplied, stagnation temperature is ambient (291 °1). If it is aeserand that all of the power supplied to the heater appears in heating of the gas stream, the staenatien temperature may be computed by a cemparison with an equivelent no power condition, uath the equation

Page 14

Here, M is the mass flow at the elevated temperature, H is heater power applied, So in the specific heat at constant prescure, To is the ambient temperature, and N is the sass flow rate at ambient temperature. Equation A.2 may be derived using E4 A.1 along with the perfect gas law and the energy conservation eqtation where AT is temperature rise above ambient. Exit velocities, tepperature, etc., may be oom. puted by assuming the flow to be isentropic. The procedure is described in all standard texts. The discrepancies of Table II arise from three sources: experimental errors, heat bases fron the gas stream which invalidate, for lower flow rates, the use of Bo A.2 and A.3, and the fact that the nozzles were designed for constructionel expedience rather than for perfect performance. This last factor is particularly evident in the 1/8—itch nozzle results. This indicates that the proper value of A in Eq A.1 is not known, since a venturi constriction is formed which lietts the effective nozzle area to less than that of the physical opening. Those error sources can be reduced by using totter nozzles and heat insulation. Since some heat loss is inevitable, it is advisable to use tank pressure changes as an index of mass flow rates and eff ective exit tepperatures.

Page 15

Sandia Corporation TM-277-56-51, August 30, 1956. 2. Pat, S., F1uid,A3LallAs of Jets, Van Nostrand Compan7, New York, July 1954. Sons, Inc., New York, January 1947.