ENGINEERING RESEARCH INSTITUTE THE UNIVERSITY OF MICHIGAN ANN ARBOR Final Report SSW END-PLATE STRESS STUDY SK. Clark. J. C. CoOk T. R. Beierle. Project 2658 WESTINGHOUSE ELECTRIC CORPORATION PITTSBURGH, PENNSYLVANIA August 1957

'r r UtKK

The University of Michigan ~ Engineering Research Institute TABLE OF CONTENTS Page ABSTRACT iii OBJECTIVE iii SUMMARY OF RESULTS 1 PROCEDURE 1 A. Internal Pressure Test 1 B. Transverse Loading Test 3 C. Axial Loading Test 4 INSTRUMENTATION 4 QUANTITATIVE RESULTS 6 APPENDIX. COMPARISON OF EXPERIMENT WITH ANALYSIS 12 FIGURES 15 REFERENCES 23 ii

The University of Michigan ~ Engineering Research Institute ABSTRACT Structural tests of the bottom-plate assembly of the Westinghouse S5W reactor were carried out using stress-coat and resistance strain gauges. Quantitative results are presented, using structural modeling laws to apply data from the models tested to the actual reactor barrel. OBJECTIVE This series of tests was designed to supplement and, where possible, to confirm the results of an analytical stress analysis on the bottom (end) plate of the Westinghouse S5W reactor shell, Strains were to be measured on models of the structure at points which appeared to be critical, based on the analysis, as well as at other selected points. Strain measurement was to be by both stress-coat and electric resistance strain gauges. Loadings were to be those considered in the analysis, namely, longitudinal loads on the modules, axial loads on the modules, and pressure on the entire structure. Reference should be made to Westinghouse drawing 513F412 for detailed shell dimensions. Additional information is contained in Ref. 2. iii

The University of Michigan * Engineering Research Institute SUMMARY OF RESULTS 1. The largest values of stresses measured in the model and applied to the prototype occur during transverse loading at or near the junction of the first pass through tubes and the bottom plate. These are presented in Table IV. 2. Stresses arising from the dead weight of all modules acting on the structure are all very small compared with the allowable stresses. 3. Stresses due to 30-psi or 50-psi internal pressure are all below allowable limits. Experiments indicate that at 150-psi internal pressure certain regions will exceed allowable stresses. 4. In general, all gauges read strains which are linearly proportional to load on the model so that good linearity of the structure, such as exists, indicates that the model-prototype laws are applicable with small error here. 5. Some success in correlation of experiment with theory was obtained on those portions of the structure which are relatively clean and simple, such as the toroidal shell. Correlation is poor in regions of very complex structure, as might be anticipated. PROCEDURE Half-size models of the bottom plate of the S5W reactor shell were constructed of both aluminum and plexiglass, for purposes of structural testing. Three types of tests were performed: first, that in which the modules are loaded transversely (simulating a sideward shock); second, that in which the shell is subjected to an internal pressure; and third, that in which the shell is subjected to vertically axial loads, simulating the dead weight of the modules. A detailed description of the testing techniques and procedures follows. A. INTERNAL PRESSURE TEST Three possible internal pressure conditions exist for the prototype shell, namely, internal pressures of 30, 50, and 150 psi. Since the law re

The University of Michigan ~ Engineering Research Institute lating model stresses and pressures to prototype stresses and pressures is CrM =(P (1) where am = model stress, ap = prototype stress, Pm = model pressure, and pp = prototype pressure, then proper stresses will be read directly from the model at pressures corresponding to the prototype pressures, or alternatively, 6p = (P I)m, (la) when one desires to use model pressures differing from those in the prototype. In this case air pressures of 50, 40, 30, and 20 in. of mercury were used as internal pressure in the model, and the prototype stresses were computed by first determining the model stresses from the appropriate strain-gauge data and then treating these by Eq. (la) to obtain stresses existing in the prototype shell. These stresses in the prototype are presented in Table I for the three conditions of prototype internal pressure previously listed, namely, 30, 50, and 150 psi. The method of treating the raw strain-gauge data to obtain stresses in the model is given in the section of this report titled "Instrumentation, since it is common to all types of tests. Pressures used in the model shell were restricted to 50 in. of mercury or less for safety reasons, since considerable energy was stored in the shell in the form of compressed air at these pressures. Pressure loading by water or oil was not thought feasible due to the necessity of thorough protection of the strain gauges. Rubber plugs were used to fill the first-pass stubtube holes during this test. Figures 1, 2, 3, and 4 show views of the experiment after the readings had been taken. During this test, a zero-pressure strain reading would be taken on a particular gauge, after which pressure would be slowly increased and strain read at 20, 30, 40, and 50 in. of mercury. These net strain readings were then all corrected to 30-psi internal pressure, assuming linearity, and an average value taken of the four results. In this manner zero drift was not a serious factor since the time involved was approximately 5 min, and by this averaging process very close reproducibility of results was obtained. In chronological order of completion, the next test was that of loading the plexiglass model after stress-coat had been applied to it, the loading

The University of Michigan ~ Engineering Research Institute being the same as the loading in the transverse or shock-load type of test. For this purpose a concrete block foundation was constructed and the plexiglass shell assembled into the wooden loading frame furnished by Westinghouse. Next the steel tubes simulating the modules were inserted and were loaded with rods and turnbuckles in such a fashion that equal reactions were felt by all stub tubes at the shell structure. Loads in each rod were monitored by inserting in each loading rod a ring dynamometer made of brass tubing which had two strain gauges attached at its midpoints and which had been previously calibrated, This arrangement is shown in Figs. 5 and 6. The plexiglass shell was sprayed with the proper grade of stress-coat and loads sufficient to cause stub-tube reactions of 60 lb at the shell were applied. At this point cracks clear enough to photograph were noted. The cracks are shown in Fig. 7; the loads are directed downward. Sensitization was accomplished by spraying CO2 onto the model, after which the pictures of Figs. 8 and 9 were taken. These show further cracking around the first-pass stub-tube holes and also around the upper edge of the toroidal shell itself. A complete sketch of the stress-coat crack pattern was made by hand (Fig. 10), which indicates the pattern after sensitization. Based on the results of the stress-coat test of the plexiglass model,.10 more strain gauges were added to the aluminum model in places where early stress-coat cracking appeared, and these gauges were oriented in a direction perpendicular to the crack direction. thus attempting to obtain the direction of maximum strain. These gauges are numbered 106 through 115 on the coded drawing previously transmitted to Westinghouse. B. TRANSVERSE LOADING TEST The aluminum model was next mounted in the wooden frame and the whole assembly placed in the bed of a Baldwin-Southwark, 60,000-lb-capacity, hydraulic testing machine. Since it was not possible to load all of the steel tubes simulating modules at once, they were loaded two at a time in such a fashion as to produce a stub-tube reaction at the shell of 500 lb each. Tubes were loaded in pairs one after the other, and finally the sum of all strains read was taken to represent the effect of loading all tubes simultaneously. This technique presupposes linearity of the structure, which is almost certainly true. Again, zero drift of the gauges was held to a minimum by reading one gauge at a time while loading was taking place, and again reproducibility was excellent. Figures 11 and 12 show this test in progress. The net prototype stresses are computed on the basis of the equation Cp = am v_ am (2) where

The University of Michigan ~ Engineering Research Institute Lm = load on model, lb, Lp = load on prototype, ib, am = scale of model, and ap = scale of prototype, which equation again presupposes linearity of the structure. The results of the strain-gauge readings on this test were converted to model stresses, which were in turn treated by Eq. (2), and the resulting prototype stresses are reported in Table II. Co AXIAL LOADING TEST The axial loading test was performed to determine the effect on stres ses in the prototype of allowing the dead weight of the modules to rest on the stub tubes or on the area of the stub-tube spacer plate immediately adjacent to the stub tubes. For this, 18 legs were manufactured which could be bolted to the flange of the aluminum model. This model was then placed in a horizontal position on the bed of a 120,000-1b-capacity Riehle screw-type testing machine, and loads applied axially by means of loading two symmetrically located stubtube holes using two plugs and a bridging cross beam. This test is illustrated in Figs. 13 and 14. Again, superposition with its implicit assumption of linearity was necessary since it was only possible to load two stub tubes at a time, and the strains resulting from these several loadings had to be summed to obtain the resultant strain at any gauge location for simultaneous axial loading of all stub tubes. Aside from changes in mechanical aspects of this test, the techniques used were identical to those used for the previous tests, except that the magnitude of the axial load reaction at each stub tube on the stub-tube radial spacer plate was set at 1000 lb. INSTRUMENTATION As previously stated, instrumentation during this test was primarily by means of wire-resistance strain gauges, In those places where space permitted, rosettes of type AR-1 were installed, but for the most part gauges of type A-18 were used. In many cases it was found possible to install these gauges in right-angled pairs, and these were helpful in determining maximum shear stresses by assuming that these right-angled pairs were oriented in the directions of principal strain. The remainder of the gauges were single. In all, 105 gauges were installed by Westinghouse and 10 by The University of Michigan, making a total of 115 strain readings for each loading condition, The gauge leads were installed here after receipt of the model from Westinghouse. A single common side was wired to each gauge, while separate leads from the other leg of each gauge ran to a bank of multiposition switch

The University of Michigan * Engineering Research Institute boxes. A battery-operated Baldwin-Southwark Type L strain-gauge bridge was used to read strain. Three or four gauges became inoperative due to shorting or broken grid wires during the test, but these were simply cut out of the circuit when the short could not be located. The strain data from the rosettes were reduced in the conventional manner, such as is given by Hetenyi.l The radius of the largest Mohr's stress circle was determined for stresses in the model from the strain-gauge data, assumming that stresses through the shell thickness are zero, and this maximum shearing stress was then treated by Eqs. (1), (la), or (2) as applicable, to obtain stresses in the model. For the gauges occurring in right-angled pairs, it was assumed in all cases that these pairs were oriented in the directions of maximum and minimum principal strain. Taking this as a starting point, Hooke's law for a biaxial stress field (assuming that azo the stress component perpendicular to the thickness of the shell is zero) may be written ex = E (ax - [ay) ~~1 ~~~~~~~(3) ey = E (y- ax) and from these one may find the associated stresses E x = 1 -, 2 (ex + key) by = 1 - 2 (ey + ex) (4) oz = O Three possible Mohr's stress circles now exist. These have radii representing the maximum shear stress as follows: (a) x-y plane TEmax 2 T (e - ey) (b) x-z plane max:2(i-_ ) (eE+ Iey

The University of Michigan ~ Engineering Research Institute (c) y-z plane Tmax E (ey + iex) Using 10 x 106 as the modulus of the aluminum model, each of the above maximum shear stresses was computed for each pair of perpendicularly oriented gauges for each loading. For a particular loading, only the largest shear stress as found above was reported as the maximum shear stresso Every effort was made to orient single gauges in the direction of maximum principal strain, but since the structure was very complex, no assurance could be given that this was indeed the case. For this reason, results from single gauges were transformed into an equivalent normal stress acting in the model by direct multiplication of the indicated strain by the modulus of elasticity, and this resulting stress was then treated by Eqs. (1), (la), or (2) as desired, It should be emphasized that this type of result does not give an accurate quantitative measure of stress at a given point, particularly shear stress, but that extremely high values of this quantity might be an indication of trouble which should be investigated in some detail. QUANTITATIVE RESULTS Quantitative results of the tests performed are given in the tables which follow as stresses in the prototype under the various conditions listed. Where shear stress is reported, it is the radius of the Mohr's circle of largest diameter which exists, as explained in the section on instrumentation, assuming that the stresses through the shell thickness are zero. For gauge locations, reference should be made to Westinghouse drawing 513F412, which has been coded at The University of Michigan and privately transmitted to the sponsor.

The University of Michigan ~ Engineering Research Institute TABLE I Final Stresses: S5W End-Plate Pressure Test Stresses given are those which will exist in the prototype, as determined from test and the modeling lawso Gauge or Gauges Type Ts3, psi * T150 psi 8, 9, 10 Rosette 424 706 2,120 13, 14, 15 Rosette 536 894 2,680 19, 22, 23 Rosette 357 595 1,785 27, 28, 29 Rosette 2522 4205 12,610 43, 44, 45 Rosette 652 1088 3,260 80, 81, 82 Rosette 1030 1720 5,150 91, 92, 93 Rosette 319 532 1,595 1, 2 900 pair 1096 1827 5,480 3, 4 900 pair 2160 3600 10,800 16, 17 900 pair 3340 5567 16,700 18, 30 90~ pair 4150 6917 20,750 25, 26 90~ pair 1215 2025 6,075 35, 36 90~ pair 2550 4250 12,750 37, 38 90~ pair 915 1525 4,575 39, 42 90~ pair 3750 6250 18,750 40, 41 900 pair 2015 3360 10,075 47, 48 900 pair 1000 1667 5,000 49, 50 900 pair 2520 4200 12,600 51, 52 90~ pair 1800 3000 9,000 53, 56 90~ pair 692 1150 3,460 55, 57 900 pair 470 787 2,360 58, 64 900 pair 655 1090 35275 65, 66 90~ pair 885 1475 4,425 68, 69 900 pair 1060 1770 5,300 70, 71 900 pair 555 925 2,775 72, 95 90~ pair 473 1715 2,365 74, 75 900 pair 535 890 2,675 83, 84 900 pair 560 930 2,800 85, 87 900 pair 440 730 2,200 89, 90 900 pair 550 920 2,750 94, 96 90~ pair 935 1560 4,675 97, 98 90~ pair 316 527 1,580 100, 101 900 pair 1675 2790 8,375 104, 105 900 pair 470 780 2,340 Subscripts refer to the internal pressure in si at which the prtabove stresses will exist in the prototype.

The University of Michigan ~ Engineering Research Institute TABLE I (concluded) Gauge or Gauges Type a30o, psi s50, psi p150, pSi 5 Single gauge -440 -730 -2,200 6 Single gauge -895 -1490 -4,475 7 Single gauge 1140 1900 5,700 11 Single gauge 430 720 2,150 31 Single gauge 1170 1950 5,850 32 Single gauge 420 700 2,100 33 Single gauge -1260 -2100 -6,300 34 Single gauge 770 1280 3,850 21 Single gauge 1010 1680 5,050 54 Single gauge 260 430 1,300 59 Single gauge -730 -1220 -3,650 61 Single gauge 570 950 2,850 62 Single gauge -810 -1350 -4,050 63 Single gauge 650 1080 3,250 67 Single gauge 640 1070 3,200 73 Single gauge 660 1100 5,300 76 Single gauge 790 1320 3,950 77 Single gauge 600 1000 3,000 78 Single gauge 400 670 2,000 79 Single gauge -950 -1580 -4,750 86 Single gauge 690 1150 3,45~ 88 Single gauge -720 -1200 -3,600 99 Single gauge 160 270 800 102 Single gauge 60 100 300 103 Single gauge 520 870 2,600 The results of the stress-coat tests have been discussed in the section on "Procedure" and have been presented in Figs. 7, 8, 9, and 10. Numerical results from the transverse loading test are presented in Table II, based on a model load of 500 lb at each stub tube on the spacer plate, and treated with the modeling laws using 11,000 lb as the load on each actual stub tube in the prototype.

The University of Michigan ~ Engineering Research Institute TABLE II Final Stresses: S5W End-Plate Transverse Loading Test Gauge or Gauges Type Tmax, psi Gauge Type amax, psi 8, 9, 10 Rosette 4,840 5 Single -2,120 13, 14, 15 Rosette 4,970 6 Single -520 19, 22, 23 Rosette 8,990 7 Single -2,610 27, 28, 29 Rosette 2,425 11 Single 3,770 44, 43, 45 Rosette 6,790 31 Single 10,400 80, 81, 82 Rosette 6,040 32 Single -2,255 91, 92, 93 Rosette 10,100 33 Single -99130 1, 2 90" pair 2,330 34 Single 17,765 3, 4 90~ pair 11,510 54 Single 2,170 16, 17 90" pair 153887 59. Single 2,915 18, 30 90" pair 2,460 61 Single 20,050 25, 26 90~ pair 2,600 62 Single 1,925 35, 36 90" pair 19,720 63 Single 8,770 37, 38 900 pair 3,675 67 Single 140 39, 42 90~ pair 27,470 73 Single -6,105 40, 41 90~ pair 19,470 76 Single -4,455 47, 48 90" pair 3,280 77 Single 935 49, 50 90~ pair 6,140 78 Single 2,475 51, 52 90" pair 2,990 79 Single -1,980 53, 56 90~ pair 600 86 Single -800 55, 57 90" pair 1,650 88 Single -4,760 58, 64 90" pair 1,060 99 Single -850 65, 66 90" pair 835 102 Single 11,030 68, 69 90" pair 21,560 103 Single 3,440 70, 71 900 pair 4,650 106 Single 14,300 72, 95 90~ pair 5,210 107 Single 770 74, 75 90" pair 8,630 108 Single 26,370 83, 84 90~ pair 3,680 109 Single 63,060 85, 87 900 pair 6,890 110 Single 35,340 89, 90 90" pair 4,010 111 Single 30,770 94, 96 90" pair 8,476 112 Single 3,160 97, 98 90" pair 5,970 113 Single -1,925 100, 101 900 pair 4,620 114 Single -1,350 104, 105 90 pair 6,500 115 Single 4,540 21 Single -715 Results from the axial loading tests are given in Table III in terms of stresses in the prototype. These tests were run at a model load of 1000 lb at each stub tube, and assuming a prototype dead weight of each module equal to 348 lb, 9

The University of Michigan ~ Engineering Research Institute TABLE III Final Stresses: S5W End-Plate Axial Loads Gauges Type Tmax, psi Gauge Type 'Oax psi 8, 9, 10 Rosette 366 5 Single 392 13, 14, 15 Rosette 376 6 Single 413 19, 22, 23 Rosette 353 7 Single 338 27, 28, 29 Rosette 767 11 Single 413 43, 44, 45 Rosette 293 31 Single 243 80, 81, 82 Rosette 479 32 Single 89 91, 92, 93 Rosette 167 33 Single 594 1, 2 900 pair 276 34 Single 502 3, 4 90~ pair 1016 54 Single 96 16 17 900 pair 1000 59 Single 139 18, 30 90~ pair 1163 61 Single 83 25, 26 90~ pair 493 62 Single 47 35, 36 90~ pair 747 63 Single 135 37, 38 900 pair 278 67 Single 163 39, 42 90~ pair 1170 73 Single 223 40, 41 900 pair 1074 76 Single 227 47, 48 900 pair 366 77 Single 145 49, 50 900 pair 1204 78 Single 246 51, 52 900 pair 131 79 Single 432 53, 56 900 pair 186 86 Single 91 55, 57 90~ pair 209 88 Single 174 58, 64 900 pair 268 99 Single 53 65, 66 900 pair 323 102 Single 77 68, 69 900 pair 278 103 Single 571 70, 71 900 pair 239 106 Single 956 72, 95 900 pair 119 107 Single 684 74, 75 900 pair 294 108 Single 920 83, 84 900 pair 289 log Single 2383 85, 87 90~ pair 128 110 Single 1.396 89, go 900 pair 330 111 Single 2207 94, 96 900 pair 303 112 Single 306 97, 98 900 pair 144 113 Single 319 100, 101 90~ pair 99 114 Single 468 104, 105 900 pair 93 115 Single 527 21 Single 420 Due to a lack of space in the bed of the testing machine, it was not possible to load each of the stub tubes separately. However, the central tube, when loaded separately, showed a maximum strain in gauges 85-87, which resulted 10

The University of Michigan ~ Engineering Research Institute in an actual shear stress in the model of 1175 psi, and treating this with Eq. (2), the largest concentrated load which can be carried by the central stub tube, based on a prototype allowable stress of 10,250 psi, is P m 4 -? (10,250) = 34,900 lb max 1175 The deflection of the spacer plate with respect to the shell was measured separately by a dial gauge mounted between the shell and spacer plate, and was taken for transverse loads only. Using model loads of 1000 lb per stub tube, the prototype spacer-plate deflection is given by 5p = 5m *Lp am Em (5) p m_ LM ap Ep where 5 = deflection, a = scale, E = modulus of elasticity, L = load, and where the subscripts m and p refer to the model and prototype, respectively. Using a modulus of 10 x 106 psi for aluminum, and 25.5 x 106 psi for steel, a scale factor of am/ap = 1/2, and using the measured model deflection of.515 in., the deflection of the prototype is 5p =.103 in. Finally, since gauges 39, 42, 109, and 110 showed rather high strain readings, rosettes were constructed of Type A-18 gauges and attached at these points. The reduced maximum shear stresses in the prototype obtained from these rosettes are given in Table IV for the case of transverse loading of the modules only. TABLE IV Prototype Shear Stresses due to Transverse Loads, from Rosette Data Area of Gauge Attachment Tmax, psi 39, 42 (117) 19,740 109 (118, 119) 31,110 110 (120, 121) 18,250 111 (122, 123) 18,250 The gauge numbers listed in parentheses in Table IV are gauges added to make rosettes in conjunction with the gauges already present. 11

The University of Michigan ~ Engineering Research Institute APPENDIX COMPARISON OF EXPERIMENT WITH ANALYSIS A review of the analysis was made to pick out those calculated stresses which could be compared directly to measured stresses. Table AI given below indicates those comparisons where warranted. Position of the existing stress is indicated by the gauge position noted in the table, in terms of the gauge number o TABLE AI Measured Calculated Stress Stress, Vol. 12 Gauges ps p Part Page Transverse Loading 61 a = 20,050 c = 2,670 2 6 70,71 T = 4,650 T = 1,850 2 6 75 a r= -9,250,= 9,200 3 7 68,69 v = 21,650 T = 32,800 3 11 T = 14,500 3 11 91,92,93 T = 10,100 T = 19,600 3 13 72,95 T = 5,210 53,56 T = 600 58,64 T = 1,060 85,87 v = 6,890 5 v = C/2 = -2120/2 = -1060 7 T= a/2 = -2610/2 = -1305 T = 3,240 4 13 59 T = a/2 = 2915/2 = 1457 73 C = -6,105 a = 6,o6o 4 13 12

The University of Michigan ~ Engineering Research Institute TABLE AI (continued) Gaugje or Measured Calculated 2 Stress, Stress, Par. P ~Gauges psi psi~ Part Page - 2psi psi 104,105 T = 6,500 T = 9,560 5 5 39,42 T = 27,470 37,38 T = 3,675 94,96 T = 8,476 16,17 T = 13,887 65,66 T = 835 18,30 T = 2,460 25,26 T = 2,600 31 a = 10,400 a = 2,020 6 6 78 a = 2,475 16,17 T = 13,887 T = 2,260 6 6 65,66 T = 835 18,30 a = -1,400 a = -4,260 6 6 43,44,445 = 6,790 T = 4,590 6 10 16,17 T = 13,887 T = 27,000 6 12 65,66 T = 835 T = 14,200 6 14 18,30 T = 2,460 25,26 T = 2,600 35,36 T = 19,720 89,90 T = 4,010 1,2 T = 2,330 T = 3,270 7 5 51,52 T = 2,990 = 3,060 7 5 100,101 T = 4,620 47,48 T = 3,280 3,4 T = 11,510 T = 1,960 7 6 49,50 T = 6,140 Axial Loads Vol. II2 3,4 T = 1,016 = 563 2 3 49,50 T = 1,204 40,41 T = 1,074 13

The University of Michigan ~ Engineering Research Institute TABLE AI (concluded) Measured Calculated 2 Gauge or Stress, Stress, Vol.II psi psi Part Page Pressure Loads 3,4 T = 2,160 T = 8,04o 4 4 49,50 T = 2,520 40,41 T = 2,015 1,2 T = 1,096 T = 2,600 4 5 51, 52 T = 1,800 47,48 T = 1,000 100,101 r = 1,675 Deflections: Transverse Loading Spacer Plate with Respect to Shell 6 =.103 in. 5 =.002 in. 4 12 14

The University of Michigan S Engineering Research Institute '.....,~i ii Fig. 1o Shell, switch boxes, and bridge in place for pressure testo 15 ~ 4 o~~.::~~::-~ o k,'~B~ XA, j~s

Fig f Ge n eral test aragmn I ~~~~~~~~~~~~~~~~Fig~o 4~0 Manometer, pressure gauge, and relief vale used in pressure tests f Psssa& k~~~~~~~~~~~~~~~~~~~~ ~~:~iF:-:iiiiii-~~~~~~~ ~ ~~~~~~i~~~~~~~~~~~~~e `~~~~~~~~~~~~~~~~~rsQ + w~~ erI ~~~~-i li~~~~~~~~~~~~~~~~ ~~~~~~~II~~~~~0 6`~~~~~~~~~~~~?~~~~~~~: "~~~~~~~~~~~~~~~:~~t P~~~~~ ~ ~ ~ ~ or pressure lodrg 0 ~~~~~~~~~~~~~~~~~~-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~t P~~~~~r\::iiii~~~~~~~~~~~~~~~~~~~~~ Fig~ )-L Manometer, pressure gauge, and relief~~~~~~~~~~~~~~~~ vav use Ini- prssr tests.::Ir

The University of Michigan ~ Engineering Research Institute 'I.:":,-, ' --- i -i. Ii|l.n,.... Figo 5. General test arrangement for transverse loading of plexiglass sheL10 g i............:.i:::'!.... Fig~ 6 Tes t arrangement for loading of plexiglass shell showing ring dynamometers~ 17

The University of Michigan ~ Engineering Research Institute r] PI i~~~~~~~~~ Cd cO3.4-D C) c6 Cd c4 -I '.r-I u)2

The University of Michigan * Engineering Research Institute WE ~ ~ ~ ~ ~ ~ ~ MO _:i'Tf~i~i:*:_::X00~:*g:0000:.N _ I.. ),~ '' ' ' i''l''' i L 0 tS0i-S.l l i,'': '::: i 4:: tt00i000;i i:tg~:; t;t;k i: li;;;I b. - _ r '::::: t:::E;.t~~~~~~t t t00 T~~i: tt;:: iLX;;Xi4E; "'0! ti~~~~~~t tX X ttlEX - 00 41::;;t t00 0: tf 1 l;::: X ff 00: _l::::::: X e j.: t.:.i:: X s N.'j::::::- ii::.: iE:::;:::tE: i: i:00:EL05::~::l iE::EtE iE 4 i.E:;t:;::: l~ i i-:X:: E ~ iii l:_ ~~i -i. ' 'g#'""E ' 0 ' "':L '' '::::;':: ' L: E:::::: ij::.:: j:j:id:i Fig. Stress-coat cracks in plexiglass shell after sensitization. Fig, 9. Stress-coa cracks in plexiglass shl in:::::::g i j 0:0::........: of th::.a: ':.l seton:.i.::: sni i F: i f...............::j:si: S.S "'.:',., L,, '', 4 j, ' ' ', ',:,.: e.;:.',: ',,, t:,,'..',. '

The University of Michigan * Engineering Research Institute Fig. 10. Overall stress-coat crack pattern on plexiglass model under transverse loading. ~~z~~~ ~~ ~~~~~: i "':~~~~~~~~~~ ~ K I 1(( /Q?7 tsi1 + IK 1 - A I(/ IrjiI 20

- The University of Michigan Engineering Research Institute Fig~ lrl [ ili~.i~~~~~iiiiii~~~~iii l i::..... i? ~~~.~i............. iiiii~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~iiiiiii~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~~~~~i'iiiii~i ---i,#~~~~~~~~~~~~6 Fig. 12G Transverse loading of aluminum shell showing method of load application~ 21 Ir~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~i 17~~~

The University of Michigan ~ Engineering Research Institute Fig. 13o General test arrangement for axial loading of aluminum shell. Fig. 14o Axial loading of aluminum shell showing supporting legs bearing on flat bedplate of testing machine. 22

The University of Michigan ~ Engineering Research Institute REFERENCES 1. Hetenyi, M., Handbook of Experimental Stress Analysis, John Wiley and Sons, New York, 1950. 2. Rowley, J. C., Unpublished analysis of bottom-plate structure of the S5W reactor barrel, August, 1956. Transmitted privately to Westinghouse Electric Corporation.

UNIVERSITY OF MICHIGAN 3 901 5 02827 33431111111 3 9015 02827 3343