ML17159A347

Dynamic Behavior of Metals Under Tensile Impact.

ML17159A347
Person / Time
Site:	Susquehanna
Issue date:	04/30/1969
From:	Schultz A COMMERCE, DEPT. OF
To:
Shared Package
ML17159A348	List: ML17159A347 ML18026A289
References
AFML-TR-69-76, NUDOCS 9806020238
Download: ML17159A347 (113)

Text

sy sir AFML-TR-69-76 PART I DYNAMIC BEHAVIOR OF METALS UNDER TENSILE IMPACT PART I. ELEVATED TEhhPERATURE TESTS ALBERT B. SCHULTZ Department of hfaterfals Engineering lQ g)

University of lllfnofsat Chicago Cfrcle Chicago, Illinois ~~ra)M~.

DOC s ~tLN1P j: 'Ug 191969 I '.,;:;

bsssr's ' 'sI r

1 fml j'L',J Ls

>Ii>

unclassified Secutit Classification DOCUMENT CONTROL DATA - R 8 D (Secuttty classlllcetlon ol title, body ol abstract and tndestnd annotation must be entered when the orerell teport ls classllled I. 0RIOINATINO AC TIVITY (Colporete author) Ze REPORT SECURITY CLASSIFICATION Board of Trustees of the University of Illinois Unclassified tb, CROUP Box 4348, Chicago, Illinois 60680 5~ REPORT TITLE Dynamic Behavior of Metals Under Tensile Impact Part I: Elevated Temperature Tests

a. oE5cnlpTIYE NoTEs (Type ol report and lncfuslre Ates)

Summa Re ort - 1 March 1967 to 28 Febru 1969 S. AU THOR(5I [trttet name, retddle tnltlel. last name)

Albert B. Schultz 5 RKPORT OATC Ta ToTAL No, or pAocs Tb. No. or RErs April 1969 36 CONTRACT OR ORANT NO ee ORICINATOR'5 RCPOIIT NUMOCRISI F33615-67-C1283 b, PROJECT NO 735 c TASK NO. 735106 eb oTHER REpoRT Hoist (Any othet numbers that aNy be east+led Ibis tepott) transmitt II approv

~

the 5UPPLEMKNTARY NOTES fo ign M

'his and d

e ument r

ents

'i is subject to oreign

'si AFML-TR-69-76 Part s

),

15. 5PON5ORINO cial export controls

'rc als Ma made MILITAIIYACTIVITY I

Air Force Materials Laboratory Research als and each L

'th prior and Technology Division, Air Force System Command Nri ht-Patterson AFB Ohio Is. *OSTRACT The mechanical behavior of metals subjected to uniaxial tensile impact at elevated temperatures is reported. Tests were conducted on annealed 1100 aluminum at 200', 350', 550', and 800'F; annealed 2024 aluminum at 200', 450', and 600'F; and annealed C1010 steel at 430', 700', 1050', and 1400'F. The materials exhibit a wide range of dynamic behavior, including some in which the stress required to produce a given level of strain is significantly lowered by dynamic loading. The ratios of the dynamic ultimate stresses to the static are found to range from 0.71 to 6.0.

This abst ct is sub) t to special export controls an ach transmi a to fore governm s or fore atio s ay be ma o y wit ri r app al the ta an er cs Divi on ), ir rce teria La rato y, ight Pa erso Air orce ase hio 45 DD,"..".".,1473 Unclassified Sccuhty Classilication

Unclassified Security Classification KEY WOROS LINK, A LINK n LINK C ROLC WT ROI C ROLE WT Plastic wave propagation Dynamic material behavior Unclassified Security Classlflcation

c ~<

DYNAMIC BEHAVIOR OF METALS UNDER TENSILE IMPACT PART l. ELEVATED TEMPERATURE TESTS'LBERT B. SCHULTZ toforei 'o This docy is subj rnmen ial r for gn n na ntrols ay ma only ch al appr 'lof t M s and ra Divi i AM), 'rce te-ri borato right-Patterson Air Force Base, Ohio 45433.

FOREWORD This report was prepared by the Department of Materials Engineering, University of Illinois, Chicago, Illinois, under USAF Contract No. F33615-67-C1283. The contract was initiated under Project No. 7351, "Metallic Materials", Task No. 735106, "Behavior of Metals". The work was administered by the Metals and Ceramics Division, Air Force Materials Laboratory, Directorate of Laboratories, Air Force Systems Command, with Dr. T. Nicholas, MAMD, project engineer.

This xeport covers work conducted from 1 March 1967 to 15 January 1969. The manuscript of this report was released by the author 30 January 1969 for publication This technical report has been reviewed and is approved.

W. J. TRAPP Chief, Strength and Dynamics Branch Metals and Ceramics Division Air Force Materials Laboratory

ABSTRACT The mechanical behavior of metals subjected to uniaxial tensile impact at elevated temperatures is reported. Tests were conducted on annealed 1100 aluminum at 200', 350', 550', and 800'F; annealed 2024 aluminum at 200', 450', and 600'; and annealed C1010 steel at 430',

700', 1050', and 1400'F. The materials exhibit a wide range of dynamic behavior, including some in which the stress required to produce a given level of strain is significantly lowered by dynamic loading. The ratios.

of the dynamic ultimate stresses to the static are found to range from 0.71 to 6.0.

This abstract is subject to special export controls and each transmittal to foreign governments or foreign nationals may be 'made only with prior approval of the Metals and Ceramics Division (MAM), Air Force Materials Laboratory, Wright-Patterson Air Force Base, Ohio 45433.

TABLE OF CONTENTS SECTION PAGE I INTRODUCTION EXPERIMENTAL PROCEDURES

'II RESULTS: 1100 ALUMINUM IY 2024 ALUMINUM C1010 STEEL VI- DISCUSSION REFERENCES 12

LIST OF ILLUSTRATIONS AND TABLES FAILURE PAGE Resistance versus temperature for 0.02 inch wires. 13 Experimental observations, 1100 aluminum, 200'F Stress-strain data, 1100 aluminum, 200'F Experimental observations, 1100 aluminum, 350'F 16 Stress-strain data, 1100 aluminum, 350'F 17 Experimental observations, 1100 aluminum, 550'F 18 Stress-strain data, 1100 aluminum, 550'F 19 Experimental observations, 1100 aluminum, 800'F 20 Stress-strain data, 1100 aluminum, 800'F 21 10 Experimental observation, 2024 aluminum, 200'F 22 Stress-strain data, 2024 alumintan, 200'F 23 12 Experimental observations, 2024 aluminum, 450'F 24 13 Stress-strain data, 2024 aluminum, 450'F 25 14 Experimental observations, 2024 aluminum, 600'F 26 15 Stress-strain data, 2024 aluminum, 600'F 27 16 Experimental observations, C1010 steel, 430'F 28 17 Stress-strain data, C1010 steel, 430'F 29 18 Experimental observations, C1010 steel, 700'F 30 19 Stress-strain data, CI010 steel, 700'F 31 20 Experimental observations, C1010 steel, 1050'F 32 21 Stress-strain data, C1010 steel, .1050'F 22 Experimental obse'rvations, C1010 steel, 1400'F 34 23 Stress-strain data, C1010 steel, 1400'F 35 TABLE I Summaxy of results 36

S E

SECTION I INTRODUCTION The mechanical behavior of metals subjected to impact loading has been examined frequently over the past thirty years. Nevertheless, only limited amounts of data have been collected, and the conclusions reached concerning such behavior have often been the subject of controversy. Mean-ingful investigations of impact behavior are difficult to design and the results are difficult to interpret. Under impact conditions, the external load on the test specimen is often not a measure of the stress within the specimen because the kinetic energy that must be imparted.to the specimen may be comparable to or much larger than the energy required for deforma-tion. Moreover, stress-wave propagation effects may prevent the achieve-ment of a homogeneous state of deformation within the specimen. When the complexity of the mechanics of such tests were not considered adequately the validity of conclusions reached in investigations of impact behavior were brought into question [1]. To account for the complexity of impact testing requires test conditions for which analysis of the observations can be made without unreasonable assumptions concerning test mechanics.

In two earlier papers, a technique for the determination of material properties under impact loading was described which lends itself to analysis without unwarranted assumptions. Observation of a succession of constant velocity transverse impacts, each on the center of a separate long thin wire specimen of the material to be studied, is used to infer material be-The technique permits study of large strain behavior in uniaxial 'avior.

tension without neglecting wave propagation phenomena. The first paper [2]

described the technique and outlined the analysis accompanying data inter-pretation. The second paper [3] extended the analysis, and described the application of the technique to a study of the room temperature behavior of 1100 aluminum. Twenty-one additional series of tests have been completed using this'echnique and the results obtained concerning the impact behavior of metals will be described in two parts. Part I will describe the results obtained in eleven series of tests conducted at elevated temperature.

Part II [4] will describe the results obtained in ten series of tests con-ducted at room temperature on, materials after different amounts of cold work. In both Part I and Part II, it was found that some materials in some states exhibit a dynamic stress-strain curve which falls below the same curve determined in slow-speed tests. Although this finding is not unique to the present investigation, it is unusual. As a consequence, in both parts, the results are presented in more detail than might otherwise be appropriate.

The present paper presents results on the impact behavior of annealed 1100 aluminum at 200', 3SO', 550 , and 800'F; annealed 2024 aluminum at" 200', 4SO', and 600'.F; and annealed C1010 steel at 430', 700', 1050', and 1400'F.

Reviewing the analysis accompanying data interpretation, which was presented in the earlier papers cited, three main assumptions concerning test mechanics were made. It was assumed th'at the wire had negligible bending stiffness, that the state of stress was'ne-dimensional, and that material behavior could be described by a single stress-strain relation applicable over the range of strain rates encountered in the tests. Let engineering stress and strain be denoted by a, c; mass density by p; impact velocity by V; maximum longitudinal particle velocity by u; angle of deformation behind the transverse wave front by g; and longitudinal and transverse wave speeds by c and c. The analysis showed that the relations among these variables depend on the ordering of the wave speeds.

For example, for the most commonly occurring ordering, all c > t'., the relations are a = a(c) c u = - c(c')dc' V = - 2(1+c)cu - u (3) tan$ =- u +

V (1+c)c (4) c(c) =-p1 Gc da

~ ~

(s) 1 a c('c) p (1+c) (6) '

is the strain in the wire before impact occurs. These six equations involve seven variables, so that if one additional relation is supplied from experimental observation, the relation of all other variables may be determined. Similar sets of equations govern for other orderings of wave speeds, as described in [3] . In the experiments the relation between V and g and that between V and c can usually be determined, providing two ways in which to determine a(c).

I ~ ~ ~

The critical transverse impact veloci:ty is that which produces a maximum strain level'orresponding to a horizontal tangent of the engineer-ing stress-strain curve. Strain levels above this cannot be propagated into the wire, and failure occurs at the point of impact for impact velocities above critical. Local necking occuring at the impact point prevents this failure from being instantaneous, so that the achievement of the critical velocity is indicated by a very rapid fall-off of the maximum strain level with increasing impact velocity. The critical transverse velocity is a parameter of material behavior in tension which is directly observable.

The corresponding maximum longitudinal particle velocity would be the critical velocity for longitudinal tensile impact, and. may be inferred from experimental observations in the same" manner as is stress-strain behavior.

The following section of this paper will describe modifications to the experimental procedures from the earlier work. Subsequent sections will present the experimental observations and the behavior which they infer for each, of the-three materials. The final section will summarize the behaviors found.

l g g ~ ~

SECTION II EXPERIMENTAL PROCEDURES The experimental procedures were described in the two papers cited.

Briefly, the transverse impact of a projectile traveling at constant velocity with the central section of a long thin wire is observed by multiple-flash stroboscopy over a period of 1.5 msec after. impact. From the photographic records of a series of such tests, each at a different velocity on a separate specimen of the material under test, the relation-ship between the angle of deformation behind the transverse wavefront and the impact velocity, and that between the maximum strain in the wire and the impact velocity is determined. Strain is measured optically,. using bands. of a marking agent over the gage section. Minor alterations in the present work are that the wire length has been increased to 32 feet, and the wire diameter varies from series to series in the work described in Part II.

The wires were heated by passing an electric current through. them in both

. static and dynamic elevated temperature tests. Temperatures were deter-mined from resistivity measurements and checked with temperature-sensitive paint. The variation in resistivity over the temperature range encountered was determined by placing coils of wire in an argon atmosphere in an oven, raising the oven temperature slowly enough to ensure thermal equilibrium, and observing temperature and coil resistance. Compensation was made for ven temperature gradients and lead resistance. The resistivities obtained e shown in Figure 1.

Static stress-strain behavior was determined as follows. A twelve foot long, annealed wire was suspended horizontally between a load cell and a winch. The wire was placed within a U-shaped channel so that air currents over the wire would be uniformly distributed. The channel opening was to one side. Current was passed through the wire, and after thermal equilibrium was reached, the winch drum was rotated slowly, taking up the wire. Test time was of the order of minutes.. Stress was determined from load'ell readings, and strain by observing the distance between gage marks in a 100 in. gage length. Catenary effects were negligible'. The current in the static tests was initially adjusted to duplicate "that used in the dynamic tests.. Voltage, current, and wattage were observed during the test, and varied very little. When necessary, the voltage was manually readjusted to maintain constant wattage. As the largest strain observed in these tests was 0.20, the method gave acceptable reproducibility.

Temperature variation was checked using bands of temperature sensitive paint along the wire, and insuring that the transition temperature of

. the paint was reached simultaneously everywhere. When steel wires were heated to luminescence, visual observation showed the glow to be uniform along the test section. Temperature measurements made with the paint agreed with those calculated from resistivity measurements. In the Figures in which static stress-strain properties are presented, only that portion of, the curve in which the stress increases is shown. Portions

~ 1 g ~

of the engineering stress-strain relation in which stress decreases with increasing strain before fracture would not be relevant when wave propa-gation occurs, as they imply imaginary values of wave speeds.

Dynamic tests at elevated temperature were made using the same heating scheme, but with the longer test section. In the central section of the wire, a flat surface under the wire substituted for the channel, permitting observation of the impact from above. Uniformity of temperature distribu-tion was checked in the same manner as for,the static tests. Prior to impact, the wires were annealed and then brought to thermal equilibrium at the test temperature. The photographic image of the undeformed wire was made seconds before impact in order for any creep or thermal expan-sion strain in the wire prior to impact to be excluded from strain measurements. The opening in the channel permitted the wire to deform freely upon impact. Heating current was supplied until after the impact observations were made.

It is assumed that the error in impact velocity determination is .

negligibly small, that angular measurements are accurate to within O.S',

and that strain measurements are usually accurate to within 0.01. In certain series of tests, depending on temperature, marking agent pro-perties, and photographic contrast between banded and unbanded sections, strain can be measured only to within 0.02. In the two highest tempera-ture test series on steel, strain could not be measured with acceptable accuracy at all. These situations and those in which pronounced data scatter occurs will be noted in discussing individual series results.

SECTION III RESULTS'100 ALUMINUM The ma'terial used in these tests is from the same lot as that used earlier [3]. Wire diameter was 0.02 in., and prior to any testing, wires were heated to 800'F and held for 3.5 minutes to anneal. Room tempera-ture stress-strain behaviors for the material in the as-received condition and after annealing are given in [4].

200'F (Fi res 2 and 3) - Figure 2 shows velocity-strain and velocity-angle observation for this series. Predictions of these from static stress-strain behavior are also shown. 'elationships The smoothed curve of velocity-strain observations shown in the figure was used to infer dynamic behavior, implying the modified velocity-angle relation also shown. Despite some scatter in observations, the two independent observations are self consis-tent. Figure 3 shows static and dynamic stress-strain behavior.

Static properties showed little variability from specimen to specimen.

450'F (Figures 4 and 5) - Figure 4 shows observations and static predictions, and Figure 5 static and dynamic stress-strain behavior.

Static tests were reproducible. Observations depart markedly from static predictions, but the two sets are only partially self-consistent. The velocity-strain data infers a dynamic stress-strain curve higher (12,000 psi ultimate stress) than that of the velocity-angle data. Velocity-angle data are considered the more reliable for this series.

550'F (Fi res 6 and 7) - Static tests were reproducible. Obser-vations depart'substantially from. static predictions. The observed velocity-angle relation is used to infer dynamic behavior, leading to results consistent with velocity-strain data..

800'F (Figures 8 and 9) - Static properties show some variability, t e range zn xcate zn igure 9. The observations exhibit a large amount of scatter. To illustrate how potential ambiguity of the results can sometimes be resolved the procedure by which dynamic behavior was inferred will be described. Three possible of the velocity-strain relation, labelled A, B, smoothed'epresentations and C, are assumed. Dynamic stress behavior is inferred from all of them, as well as the corresponding velocity-angle relationship.

(In cases A and B, the computations must take account of the existence of longitudinal shock waves in the response. The details of this are described in Part II [4].) The three result-ing velocity-angle relations are shown in Figure 8, and the stress-strain behavior in Figure 9. Curve C produces velocity-angle predicitions most consistent with observations. Independently, a 6

i

~ a

~ I ~ ~

smoothed representation of the velocity-angle relation was used to infer dynamic stress-strain behavior as well as velocity-strain behavior, and yielded consistent results. Finally, the dynamic stress-strain relation is used.to predict again the velocity-angle and velocity-strain relationships, which serves to check computations.

Table I summarizes the behavior of 1100 aluminum in these tests, illus-trating how the observed dynamic behavior differs from static. At the'hree lower temperatures, ductility is reduced by dynamic loading, but it is sub-stantially increased at the highest temperature. The-ultimate stress in every case is raised by dynamic loading, up to a factor of 2.7.

Investigations of the elevated temperature dynamic behavior of 1100 aluminum using several different experimental techniques are described, for example, by Nadai and Manjoine [5], Alder and Phillips [6], Bailey'nd Singer [7], Chiddester and Malvern [8], Lindholm, et al [9], and Suzuki, et al [10). Nadai and Manjoine found the ultimate stress raised by factors of 2.1 and 4.8 at temperatures of 392'F (200'C) and 752'F (400'C) over a strain rate range 10 per sec. Alder and Phillips obtained factors up to 1.45 over a range of approximately 30 per sec. at similar temperatures.

Bailey and Singer obtained 2.4 over a range of 10 per sec. at 752'F for high purity aluminum in agreement with Suzuki, et al. Lindholm, et al, whose results agreed with the more limited data of Chiddester and Malvern found the ultimate stress raised by a factor of 3.2 over a strain rate range of 10 per sec. at 750'F. The present results are in agreement with these other findings. It should be noted that only references [5] and [9]

report behavior in tension.

SECTION IV 2024 ALUMINUM The material used was commercially drawn to 0.02 in. diameter from heavier stock wire. Prior to any testing the wires were heated to 600'F and held for 3.5 minutes to anneal. Room temperature stress-strain behaviors in the as-received and annealed conditions are given in [4].

200'F (Fi ures 10 and was used ll - The velocity-strain relation shown to xn er ynamic behavior, with similar results obtain-able from the velocity-angle relation. The variability in static properties cannot account for the departure of the observations from those predicted from static behavior.

450'F (Fi res 12 and 13) - The velocity-strain relation was use to infer behavior, and is consistent with the angle obser-vations.

600'F (Fi ures 14 and 15) - The velocity-angle relation was used to infer behavior, and produced satisfactory agreement with strain observations. A closer fit to the velocity-strain observations indicates less rate sensitivity (ultimate stress 17,400 psi) .

Table I summarizes the behavior of 2024 aluminum in these tests. This

~ ~

~

alloy, which in most circumstances is rate insensitive at room temperature,

~ ~ ~

exhibits considerable rate sensitivity at elevated temperature. In every

~ ~

case, ductility is reduced by dynamic loading. The most striking feature

~ ~ ~ ~

of its behavior, however, is that at 200'F it exhibits a negative rate sensitivity, the tensile stress at ultimate strain in the dynamic tests being 0.71 times that in the static tests. At 600'F, the ultimate stress is raised by a factor of 3.5 over the static ultimate stress.

Bailey and Singer [7] report a raising of ultimate stress by a factor of 1.8 at 662'F over a strain rate range of 5 x 10'er sec. for a similar alloy. Both Lindholm, et, al [9] and Green and Babcock [ll] report com-parable results for 6061 and 7075 aluminum, both of which are believed rate insensitive at room temperature. Suzuki, et al [10] report negative rate sensitivity in an aluminum -3.5 percent copper alloy at 392'F for large compressive strains. They show flow stress decreasing with strain rates in the range 0.2-3.5 per sec. and then increasing with rates up to 30 per sec. for strains larger than 0.35. This behavior may be associated with precipitation rates.

~ > ~

~

SECTION V C1010 STEEL The wires useci were commercially drawn to 0.02 in diameter from heavier stock. Prior to any testing, the wires were heated to 1400'F and held for 15,sec. to anneal. Holding for longer times did not change.

the properties of the wire. Room temperature stress-strain behaviors in the as-received and annealed conditions are given in [4], Tests were made to see .that scale formation in any of the heating processes had negligible effect on static behavior or the weight per unit length.

430'F (Fi ures 16 and 17) - The behaviors inferred from velocity-strain and from velocity-angle observations are not completely con-sistent, although the differences are not large. Both inferences are shown in Figure 17. Static test data predict linear elastic behavior up to an impact velocity of 2100 in. per sec. The velocity-angle observations in this region indicate yielding occurs below this velocity. A dynamic yield stress of approxi-mately 30,000 psi is indicated, compared to the 45,000 psi static

, yield stress. The unusual shapes of the static predictions in Figure 16 arise because shock waves would occur strain behavior governed.

if static stress-700'F (Fi ures 18 and 19) - The behaviors inferred from the two sets of observations are not consistent, and a dynamic stress-strain curve in reasonable agreement with both could not be found. Therefore, two possible inferences are shown, neither completely satisfactory. Both indicate the dynamic -yield stress to be above the static, and both indicate dynamic stresses at larger strains fall below. the static. The curve inferred from strain data is assumed the more reasonable, despite the scatter in the observations.

1050'F (Fi ures 20 and 21) -. Static tests exhibit the variability shown. No marking agent was found, either for this or the next serzes, which had properties suitable for use in strain measure-ment. Behavior is therefore inferred entirely from V-g observations.

1400'F Fi res 22 and 23)" - Strain measurements could not be made, ut the velocity-angle observations clearly indicate pro-nounced rate sensitivity. There was a small increase in g with time in many of the experiments, the amounts shown in Figure 22.

This cannot be explained if dynamic stress strain curve.

behavior is governed by a single (In contrast, increases in c with time are sometimes explainable on the basis of rate independent behavior. Such increases appear when wave propagation speeds are very low. See p. 345 of (3].) The increase in g with time seems to indicate that the material is sufficiently rate sensitive to

~

~ ~

begin to require a more complex behavioral model to describe test mechanics. The two interpretations of the observations shown in Figure 22 do not infer (under the assumption that the behavior is rate independent) much difference in the dynamic behavior.

A summary of the behavior of C1010 steel is given in Table I. This material's dynamic behavior varies in a complex way with temperature. At 430'F the dynamic yield stress is lower than the static, while at the other three temperatures it is raised. The dynamic ultimate stress, on the other hand, is lower than the static at 700'F, but higher at the other three temperatures. There is often a considerable difference between the shape of the static and the dynamic stress-strain curve. Dynamic ductility is from one quarter to three times that determined statically. The strain rate sensitivity at the highest temperature is pronounced, the dynamic ultimate stress being six times the static.

Nadai and Manjoine [5] reported the ultimate tensile stress of mild steel at 1472'F (800'C) to be raised by a factor of 5.5 over a strain rate range of 10 per sec. They found negative strain rate sensitivity over various strain rate ranges at 392'F, 752'F, and 932'F. The maximum lowering of the dynamic stress was by a factor of 0.6. Alder and Phillips [6] reported the compressive stress in a 0.17 percent carbon steel to be raised by a factor of 1.16 by only a five-fold increase in strain rate. Suzuki, et al [10], in their comprehensive report on com-pressive behavior of metals when deformed in a cam plastometer, found that raising strain rate two orders of magnitude caused a 30-50 percent increase in stress at 1472'F over a range of carbon content from .08-.15 percent. For a .15 percent carbon steel, they found negative rate sensi-tivity for large strains at 392'F, and ranges of both positive and negative rate sensitivity at 752'F.

10

~ y ~

SECTION VI DISCUSSION The tests described here are not constant strain rate tests. Because of. the wave propagation in the specimens, strain rates vary with both distance along the wires and time. Average strain rates probably most often fall into the range 10 -10 per second, but when shock waves propa-gate, for example, strain rates may be considerably above these figures.

Under certain conditions, they may be much smaller. In the analysis od the data it is assumed that a single dynamic stress strain curve can describe material behavior over the range of strain rates encountered. The behavior determined under this assumption is in good agreement with the results of other investigations in which behavior was determined under more nearly constant strain rate conditions. Only in the highest temperature tests on steel, in which the dynamic ultimate stress appeared to be raised by a factor of 6 over the static, were there definite indications that a single dynamic stress-strain curve might not suffice to describe material behavior.

It is possible this was also the case in steel at 700'F.

The test results indicate a wide range of behavior in metals subject to tensile impact loading at elevated temperatures. It has sometimes been stated that at high strain rates, metals either behave the same as at low strain rates or else the stress at a given strain is raised by dynamic loading. However, the present results, and those described in Part II, confirm that the stress at a given strain is sometimes substantially lowered by dynamic loading. By comparing the results of the present ten-sile tests with the results of other investigations of dynamic behavior in compression, it can be seen that the two behaviors appear to be similar.

hfetals which exhibit little or no rate sensitivity at room temperature may exhibit considerably rate sensitivity at elevated temperatures.

In these tests, the critical velocity for transverse tensile impact differs from the value predicted from static behavior by factors ranging from 0.67 to 3.9. For critical longitudinal velocity, the factors range from 0.41 to 5.5. Ultimate strains under dynamic loading differ from those found statically by factors ranging from 0.28 to 3.2. Ultimate stresses differ by factors ranging from, 0.71 to 6.0. The energy that can be absorbed by a material can be determined from the area enclosed by its stress-strain curve. Again, large differences between static and dynamic behavior in this respect were found to exist, with the ability to absorb energy under dynamic loading sometimes considerably larger and sometimes considerably smaller than under static loading.

11

REFERENCES

[1] W. Goldsmith, Impact, Edward Arnold, London 1960> Chapter VII.

[2] A. B. Schultz, P. A. Tuschak, and A. A. Vicario, Jr., "Experimental Evaluation of Material Behavior in a Wire Under Transverse Impact,"

Journal of Applied Mechanics, Vol. 34, No. 2, Trans. ASME, Vol. 89, Series E, June 1967, pp. 392-396.

[3] A. B. Schultz, "Material Behavior in Wires of 1100 Aluminum Sub)ected to Transverse Impact," Journal of Applied Mechanics, Vol. 35, No. 2, Trans. ASME, Vol. 90, Series E, June 1968, pp. 342-348.

[4] A. B. Schultz, "Dynamic Behavior of Metals Under Tensile Impact.

Part II: The Effect of Cold Work," manuscript in preparation.

A. Nadai and M. J. Mangoine, "High-Speed Tension Tests at Elevated Temperatures Part II and III," Journal of Applied Mechanics, Trans.

ASME, Vol. 63, 1941, pp. A-77-A-91.

[6] J. F. Alder and V. A. Phillips, "The Effect of Strain Rate and Temperature on the Resistance of Aluminum, Copper, and Steel to Compression," Journal of the Institute of Metals, Vol. 83, 1954-1955, pp. 80-86.

71 J. A. Bailey and A. R. E. Singer, "Effect of Strain Rate and Temperature on the Resistance to Deformation of Aluminum, Two Aluminum Alloys, and Lead,'ournal of the Institute of Metals, Vol. 92, 1963-64,pp. 404-408.

[8] J. L. Chiddester and L. E. Malvern, "Compression-Impact Testing of Aluminum at Elevated Temperatures," Experimental Mechanics, Vol. 3, 1963, pp. 81-90.

[9] U. S. Lindholm, L. M. Yeakley, and R. L. Bessey, "An Investigation of the Behavior of Materials Under High Rates of Deformation," Southwest Research Institute Report, May 1967, 38 pp.

[10] H. Suzuki, S. Hashizume, Y. Yabuki, Y. Ichihara, S. Nakagima, and K. Kenmochi, "Studies on the Flow Stress of Metals and Alloys," Report of the Institute of Industrial Science, The University of Tokyo, Vol. 18, No. 3, March 1968, pp. 139-240.

S. J. Green and S. G. Babcock, "Response of Materials to Suddenly Applied Stress Loads, Part I: High Strain-Rate Properties of Eleven Reentry-Vehicle Materials at Elevated Temperatures," GM Defense Research Labora-tories TR 66-83, November 1966, 110 pp.

~ ~ ~ I

.20 2024 AI.UHIKUH

.IS IIOO ALUHIKUH LL la)

.IO I

CO CO LLl lL C IOIO (HULTIPLY ORDINATES XIO)

.05 500 1000 TEHPERATURE, 4 F

FIG. 1 RESISTANCE VERSUS TEMPERATURE FOR 0.02 INCH WIRES 13

~ ~

P

.20 -404 0

IIOO ALUHINUH, 2000 F 0 STATIC PREDICTION OYNAHIC, SHQOTHEO 4 o EXP ERIHENT 0

0

.IO -20' V- e r+

0

~>o o

o 0 0o 0 0

~W o4 0

0 0 V,IN/SEC 1000 2000 5000 4000 I

FIG. 2 EXPERIMENTAL OBSERVATIONS, ll00 ALUMINUM, 200 F

o, PSI rrr j

8000 4000 II00 ALUNINUN, 200'F STATIC DYNANIC

.05 .IO .I5 .20 FIG. 3 STRESS-STRAIN DATA, 1100 ALUMINUM, 200 F

~ J

-40'IOO ALUMINUM, 350' STATIC PREDICTION

-20

DYNAMIC, SMOOTHED EXPERIMENT V-a

~P o

~ ear+

0 o o 0

I000 2000 3000 FIG. 4 EXPERIMENTAL OBSERVATIONS, 1100 ALUMINUM, 350'F

~ ~ ' ~

o,PSl 8000 r

t

/I /

4ppp llpp ALUNINUN, 550' STATIC DYNANIC

~ 04 .08 FIG. 5 STRESS-STRAIN DATA, 1100 ALUMINUM, 350 F

~ ~ ~

Al UHINUH, 550' STATIC PREDICTION V-0 DYNAHIC, SHOOTHED

-20'l00o EXPERIHENT oo I 0 0

0 0

V,INISEC IOOO 2000 5000 FIG. 6 EXPERIMENTAL OBSERVATIONS, 1100 ALUMINUM, 550'F

~ ~

//

6000

//

//

4000 C

/

I

/

2000 IIOO ALUMINUM, 550' STATIC DYNAMIC

.02 .04 FIG. 7 STRESS>>STRAIN DATA, 1100 ALUMINUM, 550 F 19

-400 IIOO ALUMINUM,800'

~~ "C STATIC PREDICTION DYNAMIC, SMOOTHED rr o pro o ~<'o EXPERIMENT r Wo r - ~~8 o ~a~

j'rr 20o 0$

~ A V-o o

~>>

0 0 0 0>>ared

<<(geest>>

<<~~~~>>0 0

Q V,IN/SEC I000 2000 3000 FIG. 8- EXPERIMENTAL OBSERVATIONS, 1100 ALUMINUM, 800 F

8000 B

(HYPOTHETICAL) IIOO ALUHINUH, 8004 F I

I I DYNAHIC

'II I

I A 4000 I I tHYPOTHETICAL)

/g r ~ et+

STATIC

.02 .04 .06 .08 FIG. 9 STRESS-STRAIN DATA, 1100 ALVMINUM, 800. F

~ < g ~

0

400 2024 ALUMINUM,200' STATIC PREDICTION DYNAMIC, SMOOTHED EXPERIMENT 0

00 0 0 0

-200 0

0 0 V-e 0

0 Q 2000 4000 6000 FIG. 10 EXPERIMENTAL OBSERVATION, 2024 ALUMINUM, 200'F

<<r,PSI 30,000 20,000 2024 AI.UHINUH, 200' STATIC I0,000 DYNAHIC

.02 .04 .06 .08 FIG. 11 STRESS-STRAIN DATA, 2024 ALUMINUM,.200 F

2024 ALUHINUH, 450' STATIC PREDICTION DYNAHIC, SHOOTHED EXP ERIHENT 0

v-e

-200 0

0 0

2000 4000 FIG. 12 EXPERIMENTAL OBSERVATIONS, 2024 ALUMINUM, 450 F

~ ~

~ I 20,000 r r IO,OOO 2024 ALUMINUM,4SO' STATIC DYNAMIC

.02 .04 FIG. 13 Sl'RESS-STRAIN DATA, 2024 ALUMINUM, 450 F

Y q 0

-200 2024 ALUMINUM, 600' STATIC PREOICTION r,

DYNAHIC, SMOOTHEP o EXPERIMENT

.05 0

O O~

0 0 ~~~~. V,IN/SEC 2000 4000 FIG. 14 EXPERIMENTAL OBSERVATIONS, 2024 ALUMINUM, 600 F

20,000 rrr

/

/I I 2024 ALUMINUM,600' l

STATIC IO,OOO DYNAIIIIIC

.02 .04 FIG. IS STRESS-STRAIN DATA, 2024 ALUMINUM, 600 F

4p4 CIOIO STEEL, 450'

~ STATIC PREDICTION DYNAMIC, SMOOTHED c EXPERIMENT 0

IO 20o V-0 V-.e

/

(PRON V-.O) 2000 4000 FIG. 16 EXPERIMENTAL OBSERVATIONS C1010 STEELp 430 F

4

~ PSI

{FROH V-~)

~ tFROH V- <)

/]

I CIOIO STEEL, 4300 F STATIC

~ DYNAHIC

.04 .08 FIG. 1'7 STRESS-STRAIN DATA, C1010 STEEL, 430 F

-40'IOIO STEEL, 700 F STATlC PREDlCTION DYNAHIC, SHOOTHED

-20'-

. o EXPERINENT

+

zr+

~~

r (FROM V-.<j o

j tFROH V-+)

0/

r 2000 4000 FIG. 18 EXPERIMENTAL OBSERVATIONS, C1010 STEEL, 700 F

~,PSI 60,000 (FROH V-LP DATA) rr r tFROH V-< DATA)

I l

CIOIO STEEL, 7000 F STATIC DYNAMIC

.02 .04'.06 FIG. 19'TRESS-STRAIN DATA, C1010 STEEL, 700 F

CIOIO STEEL, l050' STATIC PREDICTION DYNAMIC, SMOOTHED o EXPERIMENT

-40'0o V-V or r

r<

r r r r r r

V- ~

r rr r r 2000 4000 6000 FIG. 20 EXPERIMENTAL OBSERVATIONS, C1010 STEEL, 1050'F

~,PSI 60,000 r

r r r r r rrr

~ si+

~ es>

rr r~

r

/

I I

~

CIOIO STEEL, I050o F STATIC DYNAMIC

..04 .08 FIG. 21 STRESS-STRAIN DATA, C1010 STEEL, 1050 F 33

"C ye' t

CIOIO STEEL, I400' STATIC PREDICTION

-40'200 DYNAMIC, SMOOTHED z I EXPERIMENT V

(e(

r> < ~

r

.05 tB)

V-c

~ essa tA)

IB) V,IN/SEC FIG. 22 EXPERIMENTAL OBSERVATIONS, C1010 STEEL, 1400 F

e,,PSI 40,000 rr r //

/ /

50,000 r / /

//(e) /

/I / /

I I

I. I I I I

20,000 I -

I CIOIO STEEL, I400' I I I I STATIC I I I DYNAMIC I

I I

I I

II IO,OOO I

I.

I II

.02 .04 FIG. 23 STRESS-STRAIN'ATA, C1010 STEEL, 1400 F 35

TABLE I

SUMMARY

OF RESULTS RITICAL TRANSVER CRITICAL LONG. ULTIHATE STRAIN ULTIHATE STRESS, VELOCITY, IN/SEC VELOCITY, IN/SEC l03 PSI DYN. 0YN. DYN.

HATERIAL STATIC DYN. STATIC STATIC DYN. TATI TATI DYN. STATI TATI DYN. STATIC IIOO ALUHINUH 2004 F 4300 4300 I.O -I620 -1470 .9I .20 .075 .38 9.4 I0.8 l.l5 3504 F 2570 3200 I.I6 -760 -l020 I.3 09 044 49 4.5 7.8 I.7 5504 F I500 2800 l.9 -350 -790 2.3 .05 .03 .60 2.7 6.7 2.5 800' l050 2700 . 2.6 -200 -860 43 .02 .064 3.2 1.8 4.8 2.7 2024 ALUHINUH

.200' 6800 4800 .71 -2340 -I380 .59 .077 .05 .65 28.0 20 .7I 450' 3400 4500 l.3 -830 -I440 .04 .033 82 I4.5 20.5 1.4 600' I500 3700 2.5 -280 -940 3.4 .03 .025 .83 6.0 20.8 3.5 CIOIO STEEL 430' 5300 3400 .67 -I640 -740 .45 .03 .28 45 48 I.07 (FLOW 7004 F 5200 3600 .69 -l720 -703 .4l .07 .07 I.O 52 42 .8I l0504 F 3000 6300 2.I -800 -2500 3.1 .04 .I 2 3.0 24 57 2.4 I400' I!50 4IOO 3.9 -230 -I260 5.5 . .02 .04.4 2.2 6.5 39 6.0

4

~Qp, 0

e 0

NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS TECHNICAL NOTE 3935 HYDROGEN-OXYGEN EXPLOSIONS IN EXHAUST DUCTING By Paul M. Ordin

SUMMARY

The ignition of hydrogen-oxygen gas mixtures at a pressure of 1 atmosphere in 5 2-foot-diameter duct resulted in detonation combustion.

The detonation static pressure at an oxidant-fuel mole ratio of 0.82 was about 315 lb/sq in. abs (pressure-rise ratio of 21) . -The use of water curtain sprays distributed through a substantial section of the duct did not prevent a detonation but did reduce the peak pressure to 200 lb/sq in. abs. The detonation could 'be prevented by adding suffi.cient carbon dioxide to place the gas mixture out of the flammable range. The use of smaller quantities of carbon dioxide resulted. in a reduction in l

the peak detonation pressures. The total pressures exerted on various designs of 90 steel elbows by the detonation were about 900 lb/sq in. abs (pressure-rise ratio of 60). A design stress of 38,400 psi and suitable supporting members for the exhaust duct elbow contained the detonation without any damage to the structure.

INTRODUCTION The design considerations of a rocket facility may involve the firing of 'rocket engines into large ducts for several reasons. The use of a duct for the rocket exhaust may permit a reduction of the noise output and also allow for the cooling and chemical treatment of the exhaust gases.

Operation of rocket engines with various propellant combinations has produced hard starts and explosions. The nature of the chemical propel-lants and starting systems and. the design of operating va1ves and related.

hardware and. of injection systems all affect the tendency to promote explosions. If a rocket engine is either enclosed in or sealed to the exhaust duct~ the duct will contain the products exhausting from the rocket engine, and somewhat the same conditions will exist in the duct as in the rocket chamber. This possibility may result in explosions in the exhaust duct.

Explosions" involve two combustion processes dependent upon the con-ditions that exist in the conta'ner. The explosion'ay result in a flame

I l

I 4

NACA TN 3935 or combustion wave that travels at a few hundred feet per second, or in a":

detonation wave that travels at many thousand. feet per second. The pre~-

sures associated. with a detonation wave are considerably higher than structures designed. to withstand. normal combustion pressures. 't those obtained with normal combustion and. could. result in the failure of",

is therefore desirable to know the conditions under which a detonation may,'.

develop in a large duct and. the characteristics .of a detonation of rocket propellants. An effort was made to carry out the studies in a configuia" tion simulating a rocket facility.

'p This report presents results of an investigation at the NACA Lewis.>

laboratory to determine whether the ignition of a rocket propellant mix-"

ture at atmospheric pressure and in a large duct would, give rise to explo sions with velocities and pressures characteristic of a detonation.

propellant combi'nation was selected.'because of its wide,':

The'ydrogen-oxygen range of explosive mixtures and the possibility of its consideration as, a useful rocket propellant. The experiments were carried. out in a pipe.

2 feet in diameter and. approximately 30 feet long. The large length was.

used to ensure sufficient distance for the 'buildup of a detonation and the laxge diameter.-to reduce the wall effect. The velocity and. pressureY were measured. to determine the nature of the explosion. Additional ex-were conducted. to determine the end load. pressures exerted. on,;."- 'eriments 2-foot-diameter elbows and the stresses developed in a thin-walled duct because of a detonation wave. Methods of preventing the formation of a detonation and. of reducing the possible maximum pressures were investigated, by the use of water and. carbon dioxide introduced. into the duct with the hydrogen and oxygen.

THEORETICAL PROPERTIES The theoretical values of detonation pressures and. velocities for the hydrogen-oxygen combination were obtained. from reference 1 and. are presented. in figures 1 and. 2. For an initial pressure of 1 atmosphere>

a peak detonation pressure of 265 lb/sq in. abs is obtained. at an oxidant fuel mole ratio of 0.5. The detonation velocity at this composition is 9200 feet per second. The detonation pressures of the stoichiometric hydrogen-oxygen m~ure with various quantities of nitrogen were obtained from reference 1 and, are presented. in figure 3. The addition of nitrogen to the hydrogen-oxygen mixture decreases the peak detonation pressures.

However, large quantities of nitrogen are required, to produce a substantial in the pressure. Approximately 60 percent by volume of nitrogen 'ecrease in the mixture will reduce the pressure from 265 to 196 lb/sq in. abs.

The limits of inflammability of mixtures of hydrogen, air~ and. carbon dioxide or nitrogen were obtained. from reference 2 and. are presented in figure 4. The flammable range for hydrogen and. air is between 4 and 72 percent hydrogen. For the hydrogen-oxygen mixture, the f1anumble rang~

is 4.6 to 93.9 percent hydrogen, and. the detonation limits vary from 15

5g 1

i93S NACA TN 393S r.

.n a to 90 percent hydrogen (ref. 2). The reduction of the oxygen concentra-res- tion 'belov 8 percent in sn air-hydrogen-carbon-dioxide mixture will make s .

the mixture nonflammable, vhereas for a hydrogen-air-nitrogen system, a of reduction of oxygen concentration to 'belov 6 percent is required. to reach "r

4g the nonflammable range.

ay cket ura- APPAEVZUS The experimental variables~ snd to a lesser degree, the apparatus, is "-..~r.. have a pronounced effect on the development of a detonation from a flame or explosion {ref. 3). A detailed description is therefore considered

<plo- valuable in understanding the results of the experiments. The apparatus he was set up in an open field and designed to permit the controlled flow of lS oxygen and. hydrogen gas into a steel duct. The propellant flov entered the- duct through an injection plate which was sealed. to the duct. The le Ias duct was fitted with a torch igniter and instrumented. to record. the duct pressure and gas velocity. For the studies involving the effect of vater on the detonation, various spray bank configurations were installed. in the duct. The a@4tionsl studies on the effect of'nd. loads due to det-onating pressures were carried out by adding various designed. elbovs to the existing apparatus. The elbows were instrumented. to measure the end.

ssures The problem of structural loans caused. by detonating pressures vas investigated. briefly by measuring the stress in s thin-walled uc

.gated.

which vas attached. to the existing apparatus. A series of runs vas also e t he mach with carbon dioxide gas introduced into the system. A sufficien number of bottles were manifolded. to permit the desired flow rate of gaseous carbon dioxide into the duct ~ The schematic diagrams in figures 6 and 7 indicate the essential features of the appsratus.

Duct Piping'and Elbovs ant The duct consisted. of several sections of 2-foot-diameter 3/8-inch seamless pipe. The duct was sealed to the 6-inch injection plate through a steel conical section 2 feet long. The propellant injection plate con-led sisted of two 14-inch pipe openings for the gss inlet. The duct for the yen initial experiments was 27 feet long from the injection platesection to the exit.

With the addition of the elbows to the end. of the straight the ltial total lengths were increased to about 34 feet. For the experiments to

>en determine the stress developed, in a thin-vslled duct a 2-foot-diameter duct of 14-gage (0.0747-inch) stock was made snd attached to the existing straight section of the pipe. A sketch of the thin-valled duct is given in figure 6 'A number of 90 elbows were investigated to determine the

~

end load pressures. Schematic diagrams of the elbows studied are pre-sented. in figure 7. Figure 7(a) shows a 90 straight miter made oi'

NACA TN 3935, 3/8-inch material, 7(b) a 90 straight miter with an elliptica1 ring at the intersection of the two cylinders of 14-gag material, 7(c) a sec-tioned mitered. elbow of 16-gage material, 7(d) a sectioned. mitered. elbow of 12-gage steel reinforced with meta1 fins, 7(e) the same elbow with thrust supports welded. to the side, 7(f) the same elbow with several of the reinforced. metal fins removed,, and. 7(g) a 90 turn with a dished head on the extended. horizontal section.

Propellant System

~

Two gas cylinder manifolds supplied the oxygen and. hydrogen to the duct. Lines 2 inches in*diameter were used. for the propellant system.

The propellant flow rates were controlled. by means of the upstream pres- .-.

sure through critica1-flow orifices. The pressure was controlled. through'.

a diaphragm regulator and was turned. on and. off by a remote operating valve. Check valves were installed just upstream of the injection plate ".

to prevent any backflow during a detonation. ln addition, a heU.um flush'.

system was installed. in the hydrogen line to permit flushing of the line ""..

and duct between runs.

'ater Systems To study the effect of water on extinguishing the explosion or reduc-ing the magnitude of the pressures, water injection at the following three positions in the duct (fig. 6) was investigated:

(1) Position 1: jet-wheel station. The design of many full-scale rocket facilities includes the introduction of water through spokes into the hot core of the rocket exhaust. The function of this water spray is to cool the rocket exhaust gases to saturation. A similar ~ater spray system was installed in the detonation apparatus to determine the effect~

if any, the jet-wheel flow had on the quenching of the explosions. The jet spoke station was located. 3 feet from the injection plate, (2) Position 2: two spray sections positioned. 5 feet apart. The introduction of water sprays from two sections 5 feet apart was investi-gated to determine the effect of the increased. cooling on the explosion.

The first spray section was located 8 feet from the injection plate. Low-pressure swirl-type spray nozzles were used. at each station.

(3) Position 3: five spray sections positioned. 1 foot apart.

number of spray sections was increased, and. the distance between sections reduced. to 1 foot. The first sect'on was 8 feet from the injection plate.

For the initial runs~ the water was supplied. to the various sections from a single header located. above the duct. With the use of the more com-plicated spray systems the header was placed. inside the duct.

1" NACA KN 3935 Carbon Dioxide System

,To permit the study of the effect of carbon dioxide gas on the hydrogen-oxygen explosions a number of bottles of carbon dioxide gas were manifo3.ded, and the gas was led. into the duct at the water jet-wheel station. The jet-wheel water flow was not used. for these tests, cylinders were commercial cylinders designed. to empty in from 1.8 to 2.0 seconds. The carbon d.ioxide flow rate was maintained. by connecting the desired number of 'bottles to the manifold and. adjusting the hydrogen and oxygen flow rates to fill ratio the duct within 1.8 seconds.

schematic In this manner diagram of the sys the desired dilution was.obtained. A tern is shown in figure 8.

ignition System The ignition system consisted. of a propane-oxygen torch mounted, on the duct 3 inches from the injection plate. The flow of propane and.

oxygen to the igniter was preset and. controlled. by pressure regulators and critical-flow orifices. A sparls was used to ignite the mixture. The operation of the igniter involved. two steps, establishing a sparg and introducing the propane and oxygen flow. For most of the runs the com-bustion of the hydrogen-oxygen mixture was initiated. 'by the spar}r, alone.

INSTRUMENTAL ON D tonation-velocity measurements were made with ioniz~tio~

inserted in the gas strea at 5-foot intervals. The impulse formed, by the combustion gases was recorded on an oscil the shorting of the gap by impulses and. the position of gaps lograph. From the time between velocities between the gaps could be determined duct, average explosion quantities of water sprayed. into the duct Runs'made with considerable of the resulted in the short-circuitingthese runs were obtained ionization gaps befoxe the run, from the static however the velocity data for P

pressure traces.

Static pressure during the passage ofpickups the detonation wave was meas-pressure of the stxain gage type.

ured by catenary diaphragm-typetransducer was recorded. on an oscillograp] .

The current from the pressure The hooP stress in the thin-walled duct was obtained. by es on thee duct and recording the output on th oscillog aph The gages locations of the pressure probes and. ioniza t ion gaps are indi~~t~d in figure 6.

Typical Pressure traces are onshown in figure 9 along with the method.

the figures. The records indicate a of obtaining the values plotted. result of a detonation w t pressure steep r which rise was the

NACA TN 3935

'he pulse was extremely sharp, the sudden rise probably induced. vibration" in the pressure pickup with the d.iaphragm oscillating about the actual pressure. In addition, since the resolution was quite indefinite, a line extrapolated back to the initial trace, as shown in figure 9, was used to obtain the detonation pressure. A calibrating voltage~ corresponding" to an established. pressure~ was impressed. across the leads, and. a normal d,isplacement vas obtained. on the film. The d.istance of 'the at the initial trace was then compared. with the calibrated cU.splace extrapolated-'ine ment~ and the actual detonation pressure vas obtained. Values incU.cated.:.

by the peak of the trace are approximately 20 to 30 percent higher than the extrapolated values.

Prior to the test all the valves and. instruments vere checked.. A 35-millimeter camera vas used. to take pictures of the oscillograph traces'~

during the run. Pressure calibration constants vere placed. on the oscil-';,

lograph trace, and. the film speed was adjusted. to 60 inches per second.. ",

The desired. pressures vere established. in the propellant flov lines, and. '<

the remote operating valves were opened. for a specified time which would, the duct vith the hydrogen-oxygen mixture to an initial pressure of osphere. The time of flov of the gases varied from 1.3 to 1.8 sec-on . The propellant valves vere then closed., the camera and. instruments put on, and. the spark ignited.. The instruments were shut off immediately *,.'"'fter the run, and. the duct vas flushed. with helium. For the experiments in which vater was introduced into the duct, the vater flov vas established.

before the propellants vere introduced. into the duct. For the experiments with carbon dioxide the carbon dioxide vas introduced. into the duct at the same time and. for the same duration as the propellants.

Detonation of the hydrogen-oxygen mixture occurred in all runs in which sufficient carbon dioxide vas not used.. The introduction of water into the duct did not quench the detonation but did lower the peak det" onation pressures. A summary of the data is given in Cable I. The ini-tial runs were made without water sprayed into the duct and at an oxidant-fuel mole ratio of 1.2. The detonation pressure at station 2, which was 8 feet 9 inches from the igniter, vas 329 lb/sq in. gage and. increased.

to 357 lb/sq in. gage at station 3~ which vas 5 feet from station 2. A second, run under the same conditions gave pressures of 316 and. 322 lb/sq in. gage at the two stations. For the third. run, the oxidant-fuel ratio was reduced to 0.84~ and. the pressures obtained. were 290 and. 286 lb/sq in. gage at the tvo stations. The detonation velocity vas about

~

0 feet per second. for the iniCial runs made at an oxidant-fuel rat>>o 1.2 and. increased. to about 7700 feet per second, at the oxidant-fue>

~

'ACA TN 3935 ions ratio of 0.84. The remaining runs were made at a constant oxygen-fuel

'ratio of 0.84 with the variables including the amount and position of ine water injection, the amount of carbon dioxide, and. the structure and.

design of the steel elbovs. Runs 4 snd. 5 were made vith the addition of Dg vater, introduced at the jet-wheel position. The water flow was 17 pounds per second. The detonation pressures measured, at instrument station 2 sd vere 350 and. 375 lb/sq in. gage and decreased. to 222 and 243 lb/sq in.

"e- gage at station 3. The detonation velocity of the first of the jet-wheel.

runs (run 4) increased from 7700 to 8160 feet per second betveen two areas, and the initial velocity for the second. run was 6610 feet per second.

F

~e The next tvo runs, runs 6 and. 7, were made with'the water introduced.

st position 2 (two spray banks 5 feet apart). The total water flov was 13.3 pounds per second.. A detonation took place in each of the two runs I

vith the pressure increasing from about 263 lb/sq in. gage at instrument

es station 3 to.308 lb/sq in. gage at station 5. The detonation velocity 1- decreased in traveling dovnstream from station 2 to station 5 from 9520 to 7620 feet per second for run 6 and. from 12,700 to 10,900 feet per second. for run 7.

.d.

Runs 8 to 11 were made with the vater spray system 3, vhich consisted.

of five spray banks vithin 5 feet. The detonation pressures measured. for P

run 8 were 161 lb/sq in. gage at station 2 and. 123 lb/sq in. gage at sta-tion 3. The water flov rate vss 26.4 pounds per second.. For r'un 9, the

.ts vater flov was increased to 34 pounds per second~ and. the pressures ob-shed. tained. vere 121 lb/sq in. gage at station 3 and 207 lb/sq in. gage at nts station 5. The detonation velocity decreased. from 8560 to 6660 feet per "he second from station 2 to station 5. Runs 10 and ll were made.,with the water flov reduced. to 17.2 pounds per second, and low detonation pressures vere obtained.. An additional pressure probe located. at station 1 in the conical approach section, 5 inches from the igniter, indicated pressures of from 118 to 145 lb/sq in. gage. The pressure at station 5 for run 10 vas 201 lb/sq in. gage and. for run 11 was 172 lb/sq in. gage. The det-onation velocities averaged. about 7000 feet per second from station 3 to station 5 for the two runs.

Because of apparent failure of large 'quantities of vater sprayed.

into the duct to quench the detonation~ the studies vere continued. with the use of carbon dioxide as the inert diluent. The carbon dioxide was introduced into the duct through the jet-vheel station (water spray posi-tion 1) at the same time the propellants were introduced. into the duct.

The first run with carbon dioxide~ run 12~ was made at an oxidant-fuel ratio of 0.84 and. with water sprayed. into the duct through spray system 3 (five bank sprays). The water flow was 17.2 pounds per second..

The flov of carbon dioxid.e into the duct was set for 27.2 pound.s per second, a rate which vould. result in, an oxygen concentration in the duct

l s

NACA TN 3935 of 6.8 mole percent at the design fuel flov of 1.58 pounds per second. of .q oxygen and. 0.117 pound per second, of hydrogen. The mixture did not ignite or produce a detonation. The folloving run~ run 13, vas made without carbon dioxide and. at the same oxidant-fuel ratio as the previous runs but at a lover propellant flov rate. (The lover flov rate was used. to permit more flexi'bility in the time of operation.) The ignition once again resulted. in s detonation. The pressure measured at station 1 was 121 lb/sq in. gage and increased. to 223 lb/sq in. gage at station 5. Tvo ~h additional runs, runs 14 and. 15, vere made with carbon dioxide introduced. j into the duct at the same time as the-propellants, and. in each case com-bustion did not take place. For run 14~ the oxygen concentration was reduced. to 6.9 percentp and. for run 15~ to 5.9 percent. To study the effect of reduced. quantities of carbon dioxide in the mixture~ two runs vere made with the resultant oxygen-hydrogen mixture vithin the flammable range. Run 16 vas made with a carbon dioxide flov rate of 8.3 pounds per seconds and. run 17 with a carbon dioxide flow rate of 2.2 pounds per sec-onds In each case combustion resulted.. The only pressure reading that vas available for run 16 indicated a pressure of 90 lb/sq in. gage at .. '!

station 1, and the two readings obtained. for run 17 were 60 lb/sq in. gage'.

at station 1 and. 116 lb/sq in. gage at station 5.

The second. phase of the investigation was conducted. with the aim of o'btaining information helpful in the design of the structure to contain the detonation. The test model vas modified. by the addition of various 90 0 turns at the end. of the straight section af the existing duct. The 90 elbows were instrumented with pressure pickups and. the traces recorded on the oscillograph. The first elbov investigated. vas the standard. 90o miter shown in figure 7(a). The pressure probes were located on the hor-izontal section and. in the end, axially with the duct. Tvo runs were madel'uns 18 snd 19'ith water introduced into the duct. The water 4'e flov rate was 17.2 pounds per second., and. spray position 3 was used. p The average side-on (static) pressures for the tvo runs (recorded by pressure probes mounted on the outer wall of the duct) vere about 112 lb/sq in. gage at pressure probe station 1 and about 191 lb/sq in. gage at station 5. The face-on (total) pressures on the elbov were 543 lb/sq in. gage for run 18 and, 620 lb/sq in. gage for run 17. The elbov -'y vas made of 3/8-inch steel and vas not distorted. in any way.

For further study of'he eff'ect of the detonetfng pressures on the y) duct material~ a thin-walled. stra'ight duct wss attached, to the encL of the .4y existing straight section (fig. 6) and. the stress developed. was measured.

F The hoop stress on the duct vas measured. by means of strain gages cemented on the surface of the thin-valled. duct. Tvo runs, runs 20 and 21, were made with the thin-valled duct. The oxidant-fuel ratio wss 0.84, and vater spray position 3 was used.. The water flov vas 17.2 pounds per sec-ond. The side-on detonation pressures developed (198 lb/sq in. gage at station 3 and. 172 lb/sq in. gage at station 5) vere comparable to the  !'

values obtained. in the previous runs, and the stress developed in the

0 A

NACA TN 3935 of wall was about 38,000 psi. The fact that the thin-walled. duct was not nite distorted in any way indicated that the duct could. probably take a higher stress.

~

The allowable stress used for the design of most of the elbows in-vestigated. (thin elbows) was about 38,400 psig (48~000 Xjoint efficieny s of 0.8). Xn addition~ to obtain the maximum detonation pressures for the Zwo tests all the runs vith the additional elbovs were made without the use

-ed of vater in the duct and at an oxidant-fuel ratio of 0.84. Run 22 was made with the single 90 miter elbov constructed. of 14-gage material (fig. 7(b)) . The side-on pressure at station 6, lip1 feet upstream of the end of the elbov, vas 294 lb/sq in. gage, and. the face-on pressure on the elbov (station 8) was 910 lb/sq in. gage ~ The detonation pressures caused,

)le some distortion in the elbow in that there,was a bulging in the surface of

.'er

>c the elbow axial with the horizontal duct ~ Xn addition, the elbov apparently failed. in bending because of buckling at the flange. A diagrammatic sketch of the distortion is shown in figure 10. The effect of the detonation pres-

'ge sure on a multisection elbow was studied by installing the sectioned. elbow shown in figure 7(c). Run 23, made with this elbov at an oxidant-fuel ratio of 0.84 and. without water in the duct, gave at station 1 a pressure of 157 lb/sq in. gage and. at station 5 a value of 198 lb/sq in. gage, while for the face-'on pressure 'on the elbov (station 8) a value of 1250 lb/sq in. gage was read.. A value of 505 lb/sq in. gage vas obtained. at station 7 on the bottom of the elbow. The pressures and loads vere too great, for the elbow failed completely in bending near the flange. Photo-ied graphs of the elbov after the run are shown in figure ll. The elbow used.

for run 23 vas constructed. of 16-gage material. For run 24, an elbov sim-it ilar to that used. for the previous run was designed., but was reinforced, by metal ribs around various sections and vas constructed. of 12-gage material (figs. 7(d) and 12)i. The detonation pressures were 203 lb/sq in.

gage (side-on pressure) at station 6 and. 910 lb/sq in. gage (face-on pressure) at station 8. A value of'781 lb/sq in. gage was obtained. at station 7 on the bottom of the elbov. The loads exerted by the detonation did not bend. or twist the elbo~ but did. induce several cracks in the hor-izonta1 approach section to the elbov. The elbov apparently withstood.

the impact load. but was questionable with respect to the bending moment near the flanged connection. The bending moment vas probably caused by force exerted. on the bottom of the elbov. Supporting thrust legs, incU.-

cated in figure 7(e), vere velded to the elbow to take the bending load..

.e d; Three runs vere macle, runs 2S, 26, and. 27, with this installation, and. it ed.

proved satisfactory. The average side-on pressure for station 6 was" 248 lb/sq in. gage with a face-on pressure of 870 lb/sq in. gage for run 2S and. 912 lb/sq in. gage for run 26. The pressure at the lover section of the elbow, station 7, taken in run 27 was 728 lb/sq in. gage. No effect of the pressures was noticed on the structure.

I NACA TH 3935 To study the role of the supporting ribs welded ar ou nd. the ebov1 t

'ith the hrust support in position several ribs vere removedve, as indi n catecL b y fi re 1 3, snd. run 28 vas maLe. The pressures obtained. v r gure 272 lb/s/ q n. gage st station 4, 246 lb/sq in. gage at station 6, and a face-on pressure of 882 lb/sq in. gage at station 8 The e lbov vas not s or t ed., but several cracks vere noticed in the structuree v disto vh ere some of thee ribs had been removed. Studies of the elbov indicated, that the o

cracks were probably due to the damaging of the velds.where the ribs had.,

been removed..

The next series of runs, runs 29~ 30, and. 31, was made with s mocLi-fied. elbow consisting of a dished. head. as the end piece (fig. 7(g)). The'.

initial run vith this elbov gave s static pressure of 206 lb/sq n. gage st station 6 and. a total pressure (face-on) of 1020 lb/ssq in. n. gage. The detonati on resulted. in complete ripping open of the vertical duct at the veld: and a d.istortion of the dished. head. to form a sphere of smaller radius. A nev vertical section was installed., and. runs 30 snd 31 were made. The configuration withstood the forces of the detonation, for no further stretching of the h ad. occurred snd. the vertical P c sec s on remained, sa s ac ory. The average pressures obtained for the two runs were 253 lb/sq in. gage at station 6 (side-on) and 786 lb/sq in. gage at the dished.

head. (station 8}.

DISCUSSION The experiments indicated. that with the duct loaded. with hydrogen snd. oxygen the discharge of an ignition source resulted in a detonation.

The experiments vere carried. out with oxygen flows of about 1.2 pounds per second. and. hydrogen flows of 0.12 pound. per second, flows comparable to that from a 400-pound-thrust rocket engine.

The experimental detonation pressures obtained in the initial runs without the use of water in the duct vere higher than the theoretically calculated. values. The theoretical calculations indicated pressure rises of about 19 to 1 (275 lb/sq in. abs), vhile the experimental values vere about 24 to 1 (335 lb/sq in. abs). The errors involved in the interpre-tstion of'he pressure record. and, in the assumption that the du c t vas fill illed. vith the hydrogen-oxygen mixture to a pressure of 1 atmosphere may s"count for the difference in values. The experimental detonation velocity for the initial runs gave a spread of values vhich may be accounted for by the error in interpretation and in the film speed,.

experimental detonation velocity at an oxidant-fuel mole ratio of 1.2r without vater in the duct, vas about 5000 feet per second compared. to the theoretically calculated. value of about 7000 feet per second.. At an oxidant-fuel mole ratio of 0.82 without vater in the duct the experimental detonation velocity vss similar to the calculated value of about 8000 feet per second..

NACA TN 3935 The introduction of water into the duct through'the jet-wheel sta-tion did, not reduce the detonation pressure at station 2, 'but apparently the continued mixing of water, steam, and gas was sufficient to reduce the pressure at station 3. The question of whether the detonation pres-sure would. have been reduced. farther downstream with the use of the jet-wheel water could not be answered., since additional pressure probes were not installed for these runs. It is believed~ however, that the detona-tion pressure would have increased.'to its peak value farther downstream in the duct. The increase in detonation pressure some distance downstream of the water injection position was obtained. with the runs made with water injection at position 2 (two spray curtains 5 feet apart). The use of the two spray sections was based. on the belief that the gases would be cooled.

and. the mixture d.iluted with sufficient steam to result in a lower det-onation pressure. Hesults of the experiments (runs 6 and. 7) indicated. an initial lowering of the detonation pressure to about 260 lb/sq in. gage at station 3 and. increases to 270 lb/sq in. gage at station 4 and to 308 lb/sq in. gage, approximately the theoretical maximum, at station 5.

Apparently the effect of the water was restricted. to a very short volume ned of the duct. The flow of water at water position 2 was 13.3 pounds per second compared. to the 17 pounds per second used in the jet-wheel studies.

.-led The use of a more finely atomized. and distributed water curtain did~

however, indicate a reduction in pressures throughout the duct. The use of water spray position 3 (five banks within 5 feet) served. to both slow the detonating velocity and. decrease the pressure. The distribution of water was more effective -in reducing the detonation pressure than the quantity of water used.. The water flow was varied from 17.2 to 34 pounds per second. and. essentially no difference in pressures was obtained.. The pressures varied. from 167 at station 2 to about 200 lb/sq in. gage at station 5 with this spray system. In all likelihoodp if it were possible to locate sufficient water sprays in the transition region between the combustion and. detonation front, the flame would. be extinguished. and. a detonation prevented..

A further effect of the water in the duct was to decrease the dura-tion of pressure as determined. from, the pressure traces. A plot of the e-ressure-time press history with and. without the addition of water is given in figure 14. In general, pressure existed. for approximately 10 milliseconds when water was used and 20 milliseconds without water. It is probable that sufficient water was present to quench the reaction behind. the det-onation wave.

Th e raatee of buildup to a detonation for the hydrogen-oxygen combina-tion is extremely rapid, as indicated. from the results obtained. from thee pressure probe at station 1. The probe was located. 5 inches from the and the results indicated. a pressure 'ncrease to about 120 'gniter~

lb/sq in. gage or to within 40 percent of the, maximum value.

yl 0'

h 12 NACA TN 3933

'C t's Results of the tests with carbon dioxide as the inert diluent indi cated that the hydrogen-oxygen mixture can 'be taken out of the combustible'.

range by reducing the oxygen concentration to belov 8 percent. The ex-periments ind.icated. that the method. of introducing the carbon d,ioxide is ta'ot critical, since success vas obtained. by merely 1 e ad. ing th e p pes con-aining the carbon dioxide just into the outer edge of the duct (fi 8). ~

The carbon dioxide vas introduced as a gas, and. over 95 percent remained.

as a gas during the expansion. This vas possible becaus f th d. i o e bottles and. manifolds. Increasing the oxygen concentration to 17 and 34 percent in the m~ure by the addition of smaller quantities of carbon dioxide placed. the mixture in the flammable range, and. ignition resulted, in a detonation. The pressures, however, were lover than those obtained. without the use of carbon dioxide.

Results of the second. phase of the study, which involved the design of'quipment to contain the detonation, indicated. that material subjected.

to sudden detonation loads can be subjected to extremely high stresses vithout f'ailing. Tests carried. out with the thin-walled duct (runs 20 and. 21} indicated. that hoop-stress values of about 38 000 psi are con-servva ive. The pressures measured. with the duct configuration wer e st a c.-

or s ide-on pressures~ since the pressure probes were all mounted. on the ti outer wall. Proposed. exhaust ducting configurations for rocket test facilities that require a 90 turn would. subject the elbov to the total or face-on pressure rather than just the static pressure. The initial elbow investigated vas constructed. of 3/8-inch-thick material (fig. 7(a))

and. vas satisfactory under all condi fons. The scaling of'his thickness to a practical field. size exhaust duct would. result in prohibitive thick-ness; it vas therefore considered. advisable to continue the tests using smaller stock. The thickness of the material involved in the design of, most of the remaining elbovs investigated. vas based on the assumption of thin-walled.-duct behavior, and the pressure load taken by hoop tension in the duct. The duration of the dynamic load. was considered to be a fev milliseconds, and. the allowable stress, 38,400 psi, was based. on 160 per-cent of the yield point and. an 80-percent efficiency factor.

The initial elbows fabricated. (figs. 7(b) to (d.)) and. tested. under detonation conditions suff'ered some sort of failure. In general, the principal damage vas caused. by the buckling or cracking of the 90 elbow at the flanged connection to the main duct. The single-miter elbov (fig. 7(b)} in addition to bend.ing near the flanged. connection was boved.

out at the elbow section axial+ with the duct (fig. 10). The total'pres-sure vas 910 lb/sq in. gage, which is equivalent to a pressure-rise ratio of about 63 compared. to a static-pressure-rise ratio of about 21.

The next elbov investigated. vas a multimiter el'bow 'shown in figure 7(c) ~ The use of a multimiter long elbow in place of'he sharp single-miter 90 elbow vould permit the gradual transition of the stress to the

1 duct and thus prevent the discontinuity stresses at the guction from exceeding the membrane stress in either part. A single run with this elbow resulted in'evere buckling near the flanged connection (fig. 11) ~

The stress concentration was considerably higher on this duct than on any of the others investigated because the material used. was 16-gage instead of 3.4-gage. The pressure on the section of the elbow axial with the duct was 1250 lb/sq'in. gage, while on the bottom of the elbow section a value of 505 lb/sq in. gage was obtained. The pressure 1250 lb/sq in. gage was the highest detonation pressure recorded in the program and. may not be a correct value because of the destruction of the elbow'. The fourth elbow investigated was a multimiter elbow similar to the previous one but con-structed of 12-gage material and reinforced by meta1 ribs welded. to the structure (fig. 7(d)).. The detonation {run 24} resulted in cracks in the horizontal section near the flanged connection to the horizontal duct anL was believed. to be caused. by the excessive bending moment.

To determine the effect of thrust-supporting members on the elbows, the next runs were made with the rib-reinforced multimiter elbow modified to include two supporting members (fig. 7(e)). This configuration resulted.

in completely satisfactory operation. The total pressures on the elbow axial with the duct were about 900 lb/sq in; gage {pressure ratio of 61}.

A subsequent run {run 28} was made with some of the supporting ribs removed (fig. 13), but the results were not conclusive, since a consider-able number of cracks developed by the detonation were along the areas where the ribs had. been removed. It was believed that when the ribs were cut away the structure was weakened.

The final elbow configuration investigated consisted. of an L-shaped pipe with the dished head. axial with the duct (fig. 7(f)) . The detonation (total} pressure measured at the Lished. head for the first run was 1020 lb/sq in. gage'and was sufficiently large to distort the dished. heaL and rip open the vertical duct. The distortion was an extensionof the center of the dished head. toward a spherical shape of smaller radius. It was believed. that the weld on the vertical section was faulty, because after it was repaired. the next two runs resulted in an average tota1 pres-.

sure of 786 lb/sq in. gage, and no further damage was done to the struc-ture. The higher value of pressure obtained. in the first of the three runs may be in error part+ because'of yielding of the metal.

'S

SUMMARY

OF RESULTS An investigation to determine the detonation combustion pressures of hydrogen-oxygen mixtures at atmospheric pressure in a 2-foot-diameter duct gave the following results:

NACA TN 3935

.. The spark ignition of hydrogen-oxygen gas mixtures in a 2-foot er duct at s pressure of 1 atmosphere resulted in de tona on tlono A

. The use of water Jets and. water sprays distributed through the id. not prevent a detonation but did reduce the peak pressures.

. The transition zone from normal combustion to s detonation for hogen-oxygen mixture is extremely short; therefore,

."oduce sufficient d.iluents to prevent a detonation.

it is difficult Detonation was.

ed, by the addition of sufficient carbon dioxide to make the mix-

>nflammable.

Equipment designed to contain the detonation should, consider pressure-rise ratios of about 25 and total-pressur e--rise ra os t 60; ti i

The design stress of materials to contain detonations can be,.

".ably higher than used. for normal applications because of the Ly short exposure time. Values of design stress about 160 percent iormal curve were completely satisfactory.

The use of thrust support members for 90 turns was necessary to e ive bending moments in the horizontal piping.

ight Propulsion Laboratory ional Advisory Committee for Aeronautics Cleveland~ Ohio, January 9, 1957 REFERENCES

. Bernard., and von Elbe~ Guenther: Combustion, Flames snd. Explo-

>s of Gases. Academic Press, Inc., 1951.

, H. F., and Jones, G. V.: Limits of Inflammability of Gases Vapors. Bull. No. 279, Bur. Mines, 1939.

<ilhelm: Explosions and. Combustion Processes in Gases. McGraw-Book Co.~ Inc., 1946.

r

~ ~ ~

~ ~ ~ ~ . ~ ~

~ ~

~ . ~

~ ~ ~ ' ~

~ ~ ~ I ~

~ ~ , ~ ~

~ ~ ~

~ ~ ~ ~

~ I ' ~

I~

I

~ ~ ~

~

~ ~ ~

I

~ I I ' ~ ~ .

~ I ~ ~

' ~

~ ~ ~ ~

' ~

280 260 Q

240 0

g 220 200 0 .8 3..2 1.6 2.0 2.8 5.2 Oxidant-fuel mole ratio Figure 1. - Detonation pressure against oxidant-fuel mole ratio for hydrogen-oxygen mixture.

initial pressure, 1 atmosphere.

t 4

t tli

I~~

I~ ~

~ I~ ~

II ~

I ~ I I II ~

I ~ IS ~ ~

- ~ ~ ~ ~ II ~ ' ~ ~ ~ I ~ ~ ~ II

NACA TN 393S 20 16 O

E d

o 12 8

.0 2 3 4 Moles of nitrogen, Y Figure 5. - Effect of excess nitrogen on hydrogen-oxygen explosion. Initial pressure, 1 atmos-phere. Mixture composition, 2 moles of hydro-gen, 1 mole of oxygen, Y moles of nitrogen.

80 Nonf lammab le 60 dl O

I P

40 Q

0 F lammab le Nitrogen 20 Carbon dioxide 0 20 40 60 80 100 Carbon dioxide or additional nitrogen in original atmosphere, percent 20 18 16 14 12 10 8 6 4 2 0 Oxygen in original atmosphere, percent Figure 4. - Limits of inflammability of mixtures of hydrogen, air, and carbon dioxide or nitrogen (ref. 2).

Propane Oxygen cylinder cylinder Pressure regulator Orifice Remote Remote operating operating valve valve Or ifices Remote Infector operating Igniter valve Duct 11ydrogen cylinders Helium cylinder Check valves CD-4591 Orifice Pressure Remote regulator operating valve Oxygen cylinders Figure 5. - Propellant and igniter system.

V A-A B-B and Position 1 B'-B'osition 2

(get-wheel station) (two sections positioned 5'part)

Spray nozzle with 9/32" lead and orifice Pressure probe Thin-walled test section Strain gage flow ~

A g B~

Igniter 4'0" 5'1 51 -" 51 2'I5" AJ I

BJ Bl ~

Ionization gap Pressure probe 3" Header Mater flow ~

2'1 a - ~:

0 11~1' 1'41 9"

~CD-4 602 J

21 3II C-C Position 3 C-C (five sections positioned l'part)

Figure 6. - Duct, instrumentationp and water spray systems.

4, CA p4 4'

l r

lm "I

NACA TN 3935 21 Station 8

Pressure Station 8 probe Pressure probe Station .'40 7 I I ill Pressure ~

33- probe 28ll (a) 90o Straight miter; 3/8" wall.

(b) 90 Straight miter; 14-gage steel.

Station 6 tation 8 15" I 3~

is" Station

~00-4303 (c) Sectioned miter; 16-gage steel.

7 Figure 7. - Steel elbows investigated.

0

~ ~

~ ~ ~

~ ~

~ ~

~ ~ ~

~ ~ ~ ~

~ ~ ~

gg(~%

II ~ ~

t ~ ~ ~

a, r~

A i.

0 g

0 I

E i

I ~

C>c I"~

~ ~

I ~ ~ ~

~ I I I ~ ~ ~ I

~ '

0

.8 Carbon dioxide bottles CD-4606 Carbon dioxide bottles Figure 8. ' - Carbon.diaxide;a@tern'."

gj'.>> ' ~ ...w

< J~ ~In

~

NACA TN 3935 25 e~

Ionization trace Pressure trace Pressure trace C-"43B18',

Detonation-Velocity Calibration Detonation velocity= Distance between robes D Film speed 5 ft D 60 Ec I 12 sec s ~

Pressure Measurement Pressure J

I (a) Run 3. Film speed, 60 inches per second.

Figure 9. - Typical pressure and ionization traces.

s

'l II NACA Tg ':jg Pressure traces C-458lS ').,

Pressure Measurement Experimental detonation Calibrated pressure pressure D (b) sun lO.

FIgure 9. - Concluded. Typical pressure and ionization traces.

a~

Bulge Bending Figure 10. - Failure of single 90o miter elbow.

>x A

. ~

I 28 NACA TN 39 P+

'I ~

+*

e/

4 4 ~@V

~ 1 gW

/ C-39042;

~ ~i ~

~x'

'P'

~ ~ 7 Tl

, ~

I C

L~ l l~g~

J

~'\ 'p I

t /

4

~ "w p ~V) g1 '~QC-39046 Figure ll. - Ihmage to multisectionel elbow constructed of 16-gage steel (fig. 7(c)) ~

AC NACA TN 3935 29

+++<

+A'~+

+~+~i I~

++i

+p

' +)

~ .:-.-':

~~ '

~4 I

a C-59043 Figure 12. - Multfsectioned elbow with supporting ribs (fig. 7(d)).

~L I'

NACA TN p9

~ lh P "hI .C r

/g h

I

'fj

"..,'i L

~

P44h 'W I

l Ay

'l 4

~ f 'y Pl I%

1P h

P I h g P9290.'ibs removed.

- Multisectioned elbov plus supports with several Figure 13.

r A )4 4

+kW I aaaaaaaaraaraaaa Eaaa

~ EEEEEEEEEEEE EEEEEEE rara EEEEaa

~ Eaaaagaraar

~ EEEEEIEEEEEE rara EEEER a

~ aaragggggEE ++ERR I ~ ~ ~ \ ~ -' e 7

4'i a p