“IN- [359 UNCLASSIFIED Defense Documentation Center Defense Logistics Agency Cameron Station e Alexandria, Virginia UNCLASSIFIED Fa a a a a a rn a ee ee — nr eS Se SESS AON ACE SSE ET AD/A-092 218 HEAT TRANSFER DESIGN AND PROOF TESTS OF A RADIOISOTOPE THERMOELECTRIC GENERATOR Barley. Beck Civil Engineering Laboratory (Navy) Prepared for: Naval Facilities Engineering Command November 1974 DISTRIBUTED BY: Hational Technical information Service U. S. DEPARTMENT OF COMMERCE memati i VNU 301 00403786 40 be gle 2 RL | 8 4 ab <0 eee ate 0) A dns uy Urclassified SECURITY CLASSIFICATION OF THIS PAGE (hen Date Friered) REPORT DOCUMENTATION PAGE 1. REPORT suMBER TN-1359 TITLE (and Sudtetic) GENERATOR ~ AUTHORs Earl J. Beck 9 PERFORMING ORGANIZATION NAME AND &ADORESS CIVIL ENGINEERING LABORATORY Naval Construction Battalion Center Port Hueneme, CA 93043 1) CONTROLLING OFFICE NAME AND ADDRESS Naval Facilities Engineering Command Alexandria, VA 22332 HEAT TRANSFER DESIGN AND PROOF TESTS OF A RADIOISOTOPE THERMOELECTRIC BEFORE COMPLETING FORM { READ SNSTRUCTIONS | Vl) ya CO x TYPE CF REPTST & PERIOD COVERED 6 PERFORMING 28G \2 GOVT ACCESSION BO, 3) RECIPIENT'S C4TALOG KUMBER / 'e Final; Nov 1972—Jul 1974 REPORT NUMBER 8 CONTRACT OF GRANT NUMBERS) 10 PROGSAM ELEWENT PROJECT. T&SK AREA & WOR] UNIT MUMSERS 63724N, Y41IW1, 43-016. 12. RESORT DATE November 1974 13) NUwWSER OF FaSES Th MONITO= NS AGENCY NAME & ADDRESS a8 istferent trom Controlling Ollice Unclassified < 2) 1§ SECCRITY CLASS fol thes report) TSe CECLASSIFIC TION DOSNGRADING SCREDULE 16 DISTRISE TION STATEMENT fof thre Reporte Approved for public release; distribution unlimited. 17 DISTRIBUTION STATEMENT (ol the ahs-ract entered im Block 20. +f different trooy Report) TR SUPPLEwWEATARY NOTES 19.9 WEY WORDS (Continue on reverse tide ef nerves sary and identcty hy Block nuvte heat transfer, natural convectior, under-ocean heat rejection surface. Thermoelectric generator, RTG, radioisotope powered, ciectrical power, enderwater power, DD mers 1473. EDITION OF 1 NOW S555 OBSOLETE Reproduced ty NATIONAL TECHNICAL INFORMATION SERVICE usc moment cf Commerce held, VA. 22151 ABSTRET™ (Continue on reverse side it necessary and identity by block nur ber Unclassified In support of a larger effort of the Nuclear Division, Naval Facilities Engineering Command, CEL undertook to design, build, and test the heat rejection portions of a large 2-kw(e) radioisotope thermoelectric generator (RTG). The design was optimized to produce the lowest practicable temperatures at the cold junction of a Sarge number of thermoelectric heat-to-electricity conversion elements. The geometry was largely defined continued SECURITY CLASSIFICATION OF T= PAGE When Date Entered) ee oii, ———— ipa Mae eat Cie ‘a viestes Th ~ fa ig ny A Cd po at hi cle eae ERT ARS) ai ie nia ma i Le rae ae eae em Be RTS 1 ow ee Unclassified SECURITY CLASSIFICATION OF THIS PAGE(Mhen Dete Entered) 20. Continued by the size, shape, and required number of thermoelectric elements and by their deployment at the upper end of a large pressure-resistant hull. The work showed - the capability of the 12-fianed convectors to maintain a temperature below 90°F at the inner face of the convectors both when the unit was vertical and when ulted &0 degrees from the vertical. The solid copper showed no signs of corrosion; the potential corrosion problem is discussed in some detail in the report, as are related problems of flow, protection, and pussible fouling from masine growth. Unclassified SECUBITY CL ASSICST ATION OF THiS SAGEM rn Date Frrered) Rann) dieeokali paler ani! Aegan i eset Men INTRODUCTION As one of several activities supporting the Nuclear Division of the Naval Facilities Engineering Command in developing a large 2-kw(e) undersea radioisotope thermoelectric generator, the Civil Engineering Laboratory undertook in late 1972 the design optimization, construction, and testing of the necessary heat rejection moduies. All known methods of converting heat to electric power inherently rust reject a certain amount of the availabie heat to the environment. Because of their relatively low efficiency, thermoelectric generating elements reject a relatively large amount of heat, some 28 to 30 kw for a 2-kw(e) unit. The cold waters of the deep ocean are excellent for receiving heat, and the potential exists for attaining a near-mexintum efficiency by keeping the cold junction of the thermoelectric conversion units as near the ambient ocean temperature as is practicable. The colder the cold junction, the better the efficiency and, probably, the life of the elements. The practical foundation for convection heat transfer, entirely adequate for the pianned use here, is well laid in extensive theoretical and empirical work in the open laterature. This work was specialized and extended for the ocean applications in a comprehensive study conducted by and for the Civil Engineering Laboratory by Braun in 1965.* This -study included resclutica of former discrepancies in publi~hed works and extended the resulting correlations to a design method and tests in shallow and deep water in the ocean off Southern California. This work so completely summarizes previous efforts that it is the only reference to be used nere. It is far too comprehensive to be abstracted, so its results will be used directly with a minimum of discussion; they relate to design, eorrosion, and marine fouling of heat transfer surfaces as well as consid- eration of the probable height of a heated plume over a natural convector. Certain assumptions are made by Braun, page Ti-25 ‘‘Optimium Fin Geometry in Free Convection,’’ some of which, while appropriate for industrial heat transfer devices, were not ail appropriate in this appli- cation. Assumption: 1. Steady-state heat flow. Valid, and realistic in the ocean. Assumption: Das Homogeneous fin material and constant fin thermal conductivity. Valid, at least for the first trial designs. The possibility of incorporating ducting (hect pipe type passages) within the fins may be ccnsidered later, which would invalidate this essumption. C. F. Braun Co., Alhambra, CA. ‘ ‘Study of heat transfer and fouling of heat transfer surfaces in the deep ocean-final report,’’ Contract NB 32274, 26 November 1965. ote ‘os wie i a Se TS TE Assumption: 3 Constant heat transfer coefficient over the face of the fin. Not valid, but a useful first approximation in developing a fin cross section using the technique of an analog computer based on resistive paper. Assumption: 4. Unifor2 temperature in the surrwunding fluid. Obviously not strictly valid, but a useful approximation as in Assumption 3. Assumption: 5. No temperature gradients occur along the length and across the thickness of the fin. Invalid, but approx- imately correct for very high conductivity in the fins (copper, silver, etc.) and low surface coefficient. Assumption: 6. Temperature at the fin base is uniform, and there is no contact resistance at the fin base. Valid, first part approximately correct for high-conductivity materials such as copper; second part valid for most construction methods, including brazing, casting, and soldering. Assumption: is There is no heat source within the fin itself. Valid for solid fins. With internal ducting, as with heat pipe channels, not valid. Assumtion: 8. Heat transfer from the fin end and sides is negligible. Valid for all but thickest and shortest (in direction of water flow) fins. The usual industrial criterion for heat transfer surface optimization is cost, and the methods outlined in the reference are besed on cost considerations. For the present design, the relatively low cost of the heat rejecticn surfaces no matter how constructed appeared trivial compared to the probable costs of the remainder of the RTC design. There- fore, a design method based on the following criteria was adopted: (a) Within the physical constraints of size based ca the number and size of the pre-existing thermoelectric units, provide a ninimum base temperature at the cold junction of the thermoelectric units. (b) Provide as unifora a base tempereture as practicable, which means a fairly heavy sectica of high-conductivity material. (c) Provide the necessary strength in longitudinal compression for the intended overpresst-es. The thermoelectric units are fabricated by an intense swagging process and can provide the necessary strength in the radial direction. (d) Provide a maximum of protection against corrosion. (e) Provide a maximum of protection against marine growth. (f£) Fabricate by a method that will insure optimum cuality control. For the immediate demonstration experiment, fabricate to as nearly the final design as practicable by a low-cost method. aie Litas ea sad aie iy tee : kGotiess an: ule: ‘prs have wid ‘be et Liat! aia ten ual oat: sais dendaads t04 ite, rise cheb eee sae wanes : saa: 409 > eabvaltien eae fi : 20d We Doand ey! aoberelen add nt baal sve: wad abe. Sid at suey att Ae Sees Wel Qawetaeliy ed yp Metesh- soeestg) a@sgo%, ~ > Seles baba sag boos pura wades wed swisex (ort apa rae) ooreaeter pad ~sindt pa_rced ° Wee As eiips. “shiladerg. wen eg Od dey ‘bey phe ive "etveains ater to ae, ne oneaen i er asad 5 at art iad ao) aed ‘Berg emilee 9g Ce cinseasee: a re alyssa lepereds atl sae seceta ie | kebaas ni we 702" “iy Be eee sheped Wad bre 2n9987q) eats eonsiel ne Ye i Cr er felis ons ma) 4g | maui at ianehoe ; wipaaerors bis Bat § Feeauee maple pith pe 7 etn aan itomett ii 238m | ad ae tne Anak Se eR OS The remainder of this report describes the design methods, the final design selected, the results of controlled tests in a large tank of seawater at CEL, and the results of a 30-day immersion in Port Hueneme Harbor. DESIGN AND OPTIMIZATION OF THE HEAT TRANSFER MODULES Previous criteria provided by the sponser, required that some 28 kw of heat be rejected in 12 vertical modules of the overall dimensions shown in Figure 1. All of the useful heat to be converted in part to electricity must be transferred radjally from a high temperature heat pipe in am area as shown. To provide for compression strength in the longitudinal direction, at least 3/16 inch of most copper alloys would be required, not crediting the additional strength and rigidity provided by any fins used in the final design. A thimble thickness of 1/4 inch was arbitrarily selected to utilize readily available materials. Based on a design overpressure of 10,000 psi (approximately equivalent to a 20,000-foot immersion depth in seawater), the resulting longitudinal compressive stress in the 1/4-inch-thick thimble is: n(D2/4)P a p2 P n(D2 - d2)/4 p2 - a2 16,500 psi* (1) for the dimensions shown in Figures 2 and 3, or about 1/6th the probable compressive strength for structural copper; stronger alloys would provide an even greater margin, as would any heavy fit.s--an almost certain eventuality in this type of convector. The heavy section is desirable to produce even temperatures at the minor diazeter, d, and a long life in the presence of corrosion. Seawater is an excellent cooling medium; however, for the relatively high unit area heat loading of 2,600 watts on a 0.308-square-foot base area, (2600) (3.4) R709 2 ane 22,700 Btu/ (hr) /(£t*) Q/A an extended surface consisting of a number of vertical, heavy fins will almost certainly be needed to produce 2 low base temperature. As a first trial, the following calcuiatio; was made of the temper- ature drop through 70-30 copper-nickel: * See List of Symbols after Appendix B. Litunig ada Pret }rinee wai Pi we 4 pas at és iditheat iq bivew ayetlo Tignwata $ vedo ett at | mAh ieee Wp trae Feonls Aa 28 Wie Hl bh ese ae pabg ene Pa) ay, 2 ‘ow ateterksah, ef aoldodn ¢veed ea Sotheyncs’ a) GED Phil ad wae ay @8hs anil ii bel, Loli sts onan, any et eer Clone wae a: ee 7 Oy : - va guage ets $3, oie i chivitates ate 10% qavecss auitbom gottens tan been te igh: i ; ) Sand jontormctis-PUE Ds wt an Au, 008, 50, ue aan ’ Abas) areas WANS eS fsobarai ha ' Je tr se OPT fendi eave: sekkey, Date ebay a pene prerie a a DE, sini P (tals Samm. m3 tam seme " x gabeol (ed we 7 nn 3809 ® Mt iret rE SSSR NO otedy we ah mee: *. xQ _ (0.25)(8,850) x -K A (200) (0.378) 2 or 29.6°F (2) for a 1/4-inch-thick thimble. While not unreasonably high, it would produce base temperatures well over 100°F with natural convection heat transfer to 40°F seawater. Much better results would be obtained with copper. For the Cu-Ni case, the temperature drop from the base would be "approximately as shown schematically in Figure 2. This figure also shows a first estimate of the full-scale cross section of idealized heavy fins. The final shape and height, H, selected were eventually modified to reduce manufacturing costs and to avoid interlocking of the fins of adjacent modules on the prescribed base circle (21.25 inches diameter) for the 12 modules. The resuits are slightly nonconservative compared with the ideal shape (Figure 2) with ample fillets at the base and parabolic cross section. That is, the more expensive, longer fins would provide slightly better cooling; the difference is marginal. The bare thimbles would have a base temperature well over 200°F, which would be undesirably high. An extended surface (fins) to produce additional heat transfer area is clearly indicated. Because of the high heat loading, the optimum fins will require: (a) heavy cross section, to avoid excessive temperature drop along their length, (b) a material of high thermal conductivity for the same reason, and (c) as close spacing as practical without reducing flow to the base. The actual value of the film coefficient, h, is determined by the vell-known dimensioniess equation as refined by Braun (1965) for seawater: Nu = £(Pr) (Ra)" (3) in which f is an experimental constant and the exponent n is typically 1/4 up to Grashoff’s number of about 107, and 1/3 above 109. The lower value is associated with laminar flow, the higher value with turbulence. For the fin length, L, dictated by the design and the physical properties of seawater at pressures equivalent to 20,000-foot depth: ee = 0.95 Btu/f1b) (°F) 3 ) = 64 15/ft- L = 1.66 1b/{£t)(hr) K = 0.385 Btu/(hr) (£2) (F/£t) Bae ser tS ee fOr Le Ooo Et g = 4.17 x 100 ft/(hr) (hr) Pro = cp u/K = 5.68 (dimensionless); at 1 atmosphere, this is about 4.5 ads 701: (Ye osotatb “sa! WOkW beweqaes. va: wheats abtedo rag, Ca ety cs Nata abe uf ‘oui oT, hte rey a ae et ee ged” aly, Ve Neanrecaeth vheosindtind lneestay eh oligos eas yveat Cu) 3 tact thw taltie red: te PMs aapiol vheds nie ts: 2 nel got ne press i favs ie y Peta i it ots. i 7 tee sh od went ; ane ely: iy) lait ¥ o ee ala et contieeast o) f dewhalY toon why ety de, whe seen a es ee * pig as Md baie Sis ae | (ait VLieaiees wh a tewetions Galt. Mire 7mvND ery i Pals aNd a eye" puta oi Me: 7 Le te i vith i awe White. fay qodwil aad Vat? seat a a barelone dn lasibyile: wit foal method oft 3¢ beaeradh yo “yhigns) WER, ens fi ets 199 22200 0% en slLivippy oa tue ne'rty *- or ini da i ee y ~ i A ae De: saat x ic) aa Wor x bic ; While the design is for the deep ocean, say 20,000 feet, the immediate results must be related to 1 atmosphere; comparable values of the important physical properties and resulting dimensionless numbers are tabulated in Tabie 1. Because of the small change in physical prop- erties in the direction of improved heat transfer with depth, results from the 1-atmosphere tests are slightly conservative--netal surface temperatures will be slightly lower in the actual deep-ocean applications. The £(Pr) is shown as a function of Pr, Figure 3, and the values for the shallow and deep water indicated in the curve. For the remainder of the report, shallow water values only will he used. OPTIMIZATION OF EXTENDED SURFACES--ITERATIVE APPKOACH Referring to Equation 1 and Figure 1, it is obvious that the only physical aspect of the heat rejection surface relezted to h is the height, L, which is fixed by the dimensions of the thermoelectric elements to be used. The other important paraneters are the physical properties of the fluid, the temperature difference (At), and the exponent--which is assumed to be 1/4, as the Grashoff’s number is about 109, the approximate upper limit for laminar flcw (Braun, 1965). Table 2 gives some estimates of the effect of varying the surface temperature on the convection film coefficient, h, the resulting heat rejection of a bare cylinéer, and the increase in area per each of the 12 modules necessary to reject the design value of 8,850 Btu per hour per module. Because all of the heat transferred from the base area thrcugh the extended fin surface must be moved through the fim retal with a loss in temperature proportional to the fin conductivity, there is an obvious tradeoff between long, slender fins (dimension H, Figure 2) to obtain a maximum usable area and short, stubby fins to produce a high surface temperature (minimize temperature drop) and a resulting high convection coefficient, h. Because the heat transferred is proportional to the area and to h, and because h varies as the 1/4th power of the temperature difference, there is no simple linear relationship allowing a direct approach to establishing proportions. Using criteria listed earlier, a first design of a possible shape is snown, Figure 4. This stubby fin has approximately 2.5 times the area of the base cylinder and fron Figures 5, 6, and 7, this fin should have a surface temperature of about 130°F when transferring the necessary heat to 40°F seawater at low pressures, if made of copper. A similar, much longer Sin made of aluminum (K = 1,500, Table 3) would have temperatures approximately as shown in Figure 8. This is the first result of the use of an analog computer which utilizes a high resistance paper to simulate heat conduction and 2 millivolt meter to read voltages, which simulate temperature. In Figure 8, the surface was divided into ten segments and the surface temperature estimated for each. The method is very rapid and powerful, in thet simple shapes can be rapidly produced, and substitution of various fin materials can s idan, baveLt ghcoa tes gical. aed FL Keiee Bint 2 Suara ewan ab * watt oy) bie webrh tyes - 2 ry ’ soethiaw? ‘tide espa ater ‘3 waver tlevitfie « baw aolisobnes Yaad. a WORT Tee Hil 6 erat’ i. ‘thd weary be made by varying the current to achieve the desired voltage drop across the base 1/4 inch. At the design heat load, a copper cvlinder 1/4 inch thick has a At of about 2°F. All other potentially useful materials are of lower conductivity, so they will have a greater temperature drop; for aluminum, this is 3.6°F. The actual resistance paper used for these and most later iterations is shown in Figure 9, and includes not only the field with pricked points at the grid intersections, but a varying resistance area allowing simula- tion of the resistance provided by the film coefficient, h. By biasing the outer border to 40°F with an external electrical resistance it is possible to obtain direct voltage readings at any point on the surface which are equivalent to the temperature. Figure 10 shows the same cross section, with temperatures plotted for a first single nodule designed to provide a prototype for the full-scale 12-module system, shown in Figure 11. This single module is shown schematically in Figure 12 in partial section full size, and in elevation 1/4th size, with some details of its construction. Its method of construction allowed two desirable features: a complex, ‘ideal’ shape at low cost, and a copper sheath te which could be directly attached a single constantan wire forming a thermocouple junction by which precise surface temperatures could be obtained. The succession of steps followed in checking the design was: (a) Prepare, bias, and take readings of a preliminary resistance paper half section, four times full size. (b) Build the prototype, lead-filled single module and equip with electric heaters and a heat transfer i1:quid consisting of approximately two parts water am’ one part ethyl alcohol, enough to cover the electric heaters while in the vertical position. (c) Test the lead-filled prototype in a small tank at two tempera- tures of water--one at about 70°F, ambient fcr the Port Hueneme area, and one at about 40°F, using ice water. (d) Repeat the ambient water tests with the module tilted 30 and 60 degrees from the vertical (a small trunnion was used to main ‘n stability). (e) Fill in the field of temperatures, using an electrical bias to reproduce the measured values. Figur-s 10(a) and 10(b) show the section temperatures at mid-elevation; the temperatures underlined are measured. The values obtained are recorded in Tables 4 and 5. There was no indication of local overheating or insufficient cooling in any of these first tests. ites shine Asti ab bes cal are aeepeiaatrit ils Abba we feared pes gn Pi) anid fe tates: ‘beg. t xa ener 3 3ued, be. 2 re _ easante idl i ail mt anaurer _ toda he, Tye: ay : 1 oSmak saan, : be pokperiis) rain’, oe BY, bee 3. tie a ® a) SP7Ite out sche one” fae Ihde 4 “heeds = = ( Tada wor oa ts Sueer bho Ah ae argh pie TaN wit: nite 089 yosew jae igi mn! - elme 0% S28w $a’ eokaniya I bao’ 6). bat Yule iii sition | 19 a iG at aire ak i Yat atogi3 saeciow, Bayaeaee’ ol? gabe. i pietaqzel ats + aaa ae PR Resets aa ldas out. he seaaen Pt * maak 0), tel as Send ye yer eel ‘Tanel 34 i ee aaipeta a FULL-SCALE PROTOTYPE EXPERIMENT Following successful verification of the design to this point using the single lead-filled module (which for purposes of this report is not considered to be a valid experiment, but a design device), 12 full-scale modules made of solid copper were fabricated (Figure 13). Details of fabrication are shown in Figure 14. In effect, each of the 12 copper modules is a small boiler or upper terminus of a heat pipe heated by three electrical immersion heaters of 1,000 watts nominal rating each. To ailow the use of fins of copper, the best material, and at the same time allow fabrication at reasonable cost, stubby rectangular fins were used. Further, this was necessary to avoid interlocking of the fins on the tight circle prescribed by the design; the loss of the extra area had negligible effect on the base temperatures as will be shown later. When the complete RTG is built, there may be a small advantage in using the complex, higher fins--that is, longer in the radial direc- tion, Figure 2. This would require either interlocking the fins or increasing the diameter of the major circle on which the i2 modules are mounted. A close-up view of the copper heat rejection modules is shown in Figure 15, and an underwater view of the completed experiment with instrumentation as tested in CEL’s 60,000-gallon tank of seawater in Figure 16. All temperatures were read from a manually balanced, direct reading portable Leeds and Northrop potentiometer, with a rapid-response electronic nul? aeter. The calibration allowed reading to within 1/2°F, and esti- mating somewhat more closely. Thermocouple locations are shown in Figure 17. Tests of the full-scale device, Figures 11 and 16, were conducted in ambient temperature water in the vertical position as shown and at 30 and 60 degrees from the vertical, without and with a protective shroud which mounted in the four 1/2-inch holes which can be seen adjacent to the modules, Figure 13. The very large volume of water in the tank made it impracticable to reduce the ambient water temperature appreciably with ice without incurring excessive cost and reducing the salinity. One loading of ice was used, and the data are identified in the tabulation of results. The freshwater tended to stratify on the surface as the ice melted because of its lack of salinity. An effort was made to stir the water in the tank, with uncertain results. The normal temperature gradient in the tank from top to bottom was measured at about 2°F. The only irregularity in experizental procedure occurred when the complete experiment was left in the seawater test tank over a weekend. The single-module lead-filled prototype had as a precaution been vented to avoid pressure build-up in the event of insufficient heat transfer. Provision was made for venting on the full-scale experiment, but the holes were plugged as a simplification, inasmuch as no pressure build- up had been observed in the first prototype. Apparently a faulty O-ring sealing the water-alcohol mixture allowed leakage past the heater elements ess toe et yt in five of the 12 modules. This was observed as an excess electrical load and, while none of the heating elements burned out, they were replaced with new units, and precautionary venting was provided. None of the data included in this report are of tests with defective wet insulation in the heating elements. DISCUSSION OF RESULTS The entire point of the experiment was to validate the capability of the module design to dissipate the required heat, measured electrically, and to provide low temperatures at the locations of thermocouples Ml, M2, and M3 which are on the surface where they would be in intimate contact with the cold junctions of the thermoelectric urits in the actual PTG. The results are tabulated in Tables 1 through 8. A run consisted of adjusting the electrical input to the desired value and observing temperature rise until the inner temperatures were stabilized. Because of the rapid cooling of the seawater, the high thermal diffisivity of the copper fins, and the small mass of the water-alcoho!l mixture contained in the heat pipes, temperature stability was typically achieved in about an hour from cold start-up and even more rapidly with minor adjustments in power input. To insure stability, a typical run consisted of a 1-1/2 to 2-hour stabilization period, followed by taking a single set of read- ings. With the exception of failure of the G-rings containing the alcohcl- water mixture as discussed above, there were no observed irregularities nor unexpected temperature anomalies. While the nominal required heat input to the 12-module experiment was 28 kw total, some tests are for a higher power dissipation, 32 kw. This is slightly over the design heat load, so that the maximum temper- atures recorded are somewhat higher than the values which might be expected in a newly fueled RIG deployed in deep-ocean water at, say, 40°F ambient seawater temperature. The temperatures as recorded in the tank tests are shown in Tables 4 and 5. An appreciation for the linearity of the metal temperature changes with change in ambient seawater can be obtained from Table 6, which gives both the temperatures of the lead-filled single module at two water temperatures, and the differences. With a difference ‘n water temperatures of 35°F, no measured temperature differences greater than 48°F were observed. The most important single temperature, because it is physically closest to the base, is T6. For this location, dropping the water temper- ature by 35°F caused a metal temperature drop of 43°F, a conservative development in assessing the eventual consequence of imuersing a full- scale RTG in very cold, deep-ocean water. Summarizing the results shown in Tables 7 and 8, there was no evidence under any of the conditions of test of poor flow to and around the modules which cause excessive temperatures or pocr cooling. The effect on the inside metal temperatures, those seen by the thermo-electric xacodules, of eaten varie Sp RAY . DPE EN 2 La wn nite yy Re eA oes 4 eee By \ tact ile, sian i ay ma shi ae ( Rk ee Ge Fs el ee i ‘ ya we foie er nape val : f | fis 5 ‘ ; Panes ri \ be * yp ike ia ¥ va r ; Ett SS RE eS A an ER te EN EE tilting the experiment from the vertical is shown in Figure 18(a). AJso shown are the effects on these critical metal temperatures of shrouding the modules with a vertical steel chimney in what might be expected to be the worst case; when the experiment is tilted 60 degrees froa the vertical. Tiere was some increase in temperature at all three points of measurement, but no great change which would predict difficulty--the presence of the shroud merely impeded the vertic2l fiow of water in this partially prone position. The effects are, in fact, similar to placing the shroud on the unit when in the vertical position, Figure 18(b). Here, there was some indication of reduced flow to the lower parts of the modules, but essentially no effect at the tcp, Tl. The small decrease in temperature of Tl with the shroud added can be almost exactly accounted for by slightly lower water temperatures. The efficacy of a shroud is almost universaily assumed, because of the apparent ‘chimney’ effect that might be gained. However, a convec- tor has one major difference from a chimney. In 2 typical chimney used to produce draft in a furnace, all of the heat is introduced below the chimney, and the draft produced arises from the difference in density between the heated stack gases and the surrounding cool air. There is, of course, a stack effect with the shrouded convector. However, there is what appears to be a more important second effect common to plumes. Without the shroud, cold fluid (water in this case) can be continually entrained--the higher metal surfaces do not see the warmed water, Fut rather cold water drawn in by the vertical convection currei.c. Our consideration of the use of a shroud for this application arose not from a desire to enhance the flow and produce marginally lower metal temperatures, but to provide protection for the convection modules from cables, etc, as will be necessary in the ocean. its usefulness in this function may well justify the small loss in cooling etfect caused by the inhibition of inflowing cold water. A schematic of a shroud as it might be utilized on a large RTG is shown in Figure 19. In assessing the foregoing, it should be rezembered that with the cold water typical of the deep ocean (36 to 40°F), all of the znetal temperatures would be reduced in proportion. Critical metal terperatures of the order to 70°F or lower would be expected. A discussion of metal selection based on heat transfer, fouling, and corrosion considerations is contained in Appendix A. The results of a 30-day immersion in Port Exeneme Harbor at nominal full power are discussed in Appendix B. It should be noted here that in neither the Laboratory tank tests nor those in the harbor was bare copper exposed, as would be expected with high velocities. The use of copper is satisfactory on that basis. i i i i i i drat | jet nh a | & Cheats 7 7 [ayen a i ie? te | woe i i chia se ‘ 4 « he oF WAY -¥e § Mig Wek! oh B i q 5 | Le bee a : rs NSS f \ H : i ee fam in ali i ROK te oe aadaon aes esiies > mete. ‘ast heenloinau ess, hr a ak, anetT. ale foos oo, vet Rall go beaten boatain sei aia Kamu 193 tunes towtts Gregae tons son Be se ¢ =f vifeimitines od ned (eens td nt gaits) ig ia well a) Se a a ~“ stitaw, howvew od son jon ab: arre® in 49 4R3IuS grok saved Pua tyra ails wa. 9287/3) bs cadens v7 iT Bran w ani hinibadet wit tem pate bets binge & Fe ase » Ses 76 ‘oa ni Joalee ont tpn at 4} fh Leo ti chats sKy wid ite yi ‘suo! ’ So IRoord anon bowie vi tos tse econ Ae, hb ciivesaad, #2. hem tadee at anahieratst pend WY erred.“ et sf sede hb Oe; & ‘eaten 3 lf a ry (ae ssh wh boupuns ie Ory veo ht i : a7 j yao bee AP weds arsed. be Peay P J hindde- at ri wv a i blow #2 . eae Aqges a ye8- tee rodye- rh OF ‘wugds cae rOsNIPISE tet see a4. KoYTAS 0 een sit, ood 2d iouiley fossa Wade i im CONCLUS IONS 1. The design utilizing solid, heavy, short fins on the 12 modules should produce cvld junction temperatures im the actual RTG of 70 F or lower in the deep-ccean (cold water), or about 35°F above the ambient temperature. 2. Tiiting of the RTG up to 60 degrees from the vertical does not seriously affect the capability of the convectors to reject the heat produced; therefore, strict verticality will not be a critical requirement in the ocean. 3. The use of a shroud around the convectors to provide a ‘chimney’ effect is not justified in terms of efficiency; the elimination of the cold water more than offsets any advantage from increased vertical convection. 4. The use of a shroud around the convectors to provide mechanical protection from the projecting convectors and other appurtenances such as electrical connectors which might be placed within the convector circle does not materially inhibit heat transfer, and may be justified on the basis of safety in spite of a small decrease in electrical] output. 5. For the flow conditions produced at the power inputs used, the velocities are not high enough to cause washing of the copper surfaces; the use of copper, an optimum material from a heat transfer standroint, is both justified and highly desirable in this application. 10 oh, a a i rd i i i i nt Det és : eg i oe 4 , vasa: ere. nawort ipa inal Lan, RT ART ae . Vente anon acacia all ats | a oe > ; ; ea 1 | i LY We i Amati) y , i ii} al y v ( i ; b i het Vania Ae A | t Na j i} Ab : 4 1 1 xi Asie vy i Hed J ip y _" | “| a4n6i4 ‘sapuljAo aseq ay sawn GB’Eg 40 O!e4 bale ue aney P[NOM uMOYs WaISAS Ulf-7L BYL “IN-ND O€-OL : "Bia ‘OOS = YH UVIM [efdoyeus yO apUU GIP SUT) OUI JI ‘doup asnyesadiua) Pue UO!}IAS SSO19 Uy PasOdosd 40 D)}BWAYIS 'Z ainbi4 t H t ; x ‘MO]4 18aH 40 UONIaIIg a ~~. Ov ay os S - t oN ao) } n . a ' UIE ere sire ene eed LU i dgt'62 BS eel t sey" } ~ { i i | i ‘ulé/bel = H ES i y 5 if f Fa ! ] wy Ebe= N =—S “y “ut 68 *(sayoul EES SNssan Sayou! Q) J ‘YIPua) YUN D14yda}»OWJIY ay) UY sehuoy Apybys uayxer si] 'YBuay uly yeyY DON “UeIIO OUY 0} UO}JIBAUOD JpuMTHU Aq payood aq OF (SYUN Ip IA;BOLWIAY ay] JO UO OUN! PjOd) ease aseq ayy yO SUO/SUBWIP feUdIayX A 'f ainbi4 \ ul ee's Vug Sulp [EN IUAAG SN Oe 11 Poa, th © ; + i if . ; } E 4 1 | \ , ‘ - ve 5 : inet ATP ein ere EA Rp ac - - ae i a ie nae! bbe) eae, i » at 40°F surface temperature \ Note: Repiotted next sheet on log4oa paper. = Design Point A - 40 &0 120 160 200 240 280 Assumed Surface Temperature, ty (°F) Figure 6. Area ratio necessary to attain Gesired base surface temperature. 13 mei iat ene a : ei | ; “UWP gph semilieecebeny cnn Wigtajmeem Bi AR RET be FMT ER AAT aS . cee ; Sania a on nu re) Baal pane | cts nba gag FPN rd 1 irdeto adh gry gl ny a ranean sm mtephn i Le “shasta ar: | i eee 4 ne ‘ 1 ae ¢ aa : : ia ; or? TALC B ah achalasia lt fen beta i nit ay Da Ree i ae a MELT C LM 4 Surface Ratio, Ry _ i) oOo wo 10 20 30 40 50 60 70 80 90 150 205 Assumed Surface Temperature, ty (°F} Figure 7. Data for Figure 6 plotted on log-log coordinates showing extrapolation to very low surface temperatures. 14 300 FP coe Renn aay Se iummnnnciaie = = Ds fe) Ato. = 2150.5 - 48.7) = 3.6°F (Approx. K = (2,700){2/3.6) = 1,500 912 92.4 e934 {ors e953 97.4 a ———— ® @ Figure 8. epresentatme results from analog computer. Values shawn would be for metal with conductivity of aluminum. any eG | bo iil vee Apu Aad ee i} 1 ae ‘ Figure 9. Actual high-resistance surface anziog computer with conzector wires. 16 pil clad Sahn pie A tay bette ih x 41 (seawater! a 90 ate pee 109 e109 2106 e102 WT) / Lacation of Thermocouples TI T2 Test data are underlined. All numbers are temperature, OF Temperatures not under- lined were filled in using & electrical analog computer. Note that in spite of the areater area than tm the stubby copper fin, the base temperature ts almost 30°F higher, due to tne relatively poor conductivity of tead_ The results with lead are typical of what couid be expected with copper-nicke! alloys. Figure 10(a}. Computer output for fins made of lead. Underlined values are experimental, and computer was used to fill in the freld. 7 7 a he te yc. ; ee oe il | a ‘ . My ieetiey KUM June Lae anager CAA At lee APY ay ORO: Bie qebvryne cee lett hf Samii bays 5 ny i : 4 Roe roa i wacyeie eeeteres gate Ty EAL Hh” ROR A : ; bps . \ oe my eh mits A , Hig i a ee Ww ’ i 7 . ‘ {aa ‘ ‘ H 1 Bia * seat ¢ x 1 i 1, CHERCES eit "¢ re ps i y A J My wv i - j j Four times full sze Nater temperature, 40°F Tamperature profile obtained by electrical anclog comput, with temperature drop based on the trermal conductivity of copper and estemated film transfer coefficients, venfred in previous experiments. All numbers are temperature, °F, Note: The computation method requires 91 6 90.7 certain assumptans which may change the aciual values up as down by 1 to 2 dearees; the fractional =mperatures weuld not be justified on that dasis. However, the method does provide accuracy of the order indicated in establishing temperature gradients. Figure 10(b). Comparable computer output for stubdy copper fins used in the 12-module experiment. All values shown are calculated. 18 ti mya SH ant pe “et tht nee dp sine ct i t f i : F Ip , : ‘ i i i i i ya ie i y i 1 es i ay i i witha Yer ei vale velba wf ae vejrise Laon na car i mip f eR TE cyte ba mine tid ty POS aT aloe | y Fara ‘ \ 1 ( . } } ‘ \ 5 ooh ; f i - Ih : hy Ai ea i pay i A / Pon i a a RR ORE yh POMPE ATCT ete & < “i aD VL Pe al ATS RNY OY RNAS HALE RASS BP hI aU SOPRA LI RUSBY Fle ae AL LOL HAVUReR NK 3 3 s a 3 § : é x Figure 13_ Full-scale 12-module heat transfer experiment; showing upper ends of heat pipes, with one finned module in place. 19 Sail 12 fins, formed copper sheet, tinned inner surface virgin lead filled — fh i! ‘ ' nda ee # i he hts wher | i ’ vay ie ous ryan al t : po LU nc oe 9 Gell! po bictllel hOB G Ta epee “S8|NPOW 30 JapowU a/ed5-)jNy yO s}leap UONeOGey ‘b| aunBi4 Tine eves Daye Tada = 3] : Cswbasenerseors [Boerectotemnness aonwese | Noon 2% IH, ea tl lane {t 4,2 I, Kr : ° on g } ° 4H A (ovis oho | eealy . ee TMUSIa Fert ase ts ol O iP, | arian {ieee tecedbiat | | leet | | af bre Tee O=b3d pa TSO TT TE (420 OF ener ahem T Twi i iveare e's Cy Uirares | Fes mu 0) p fe Ps ben omy: S47 setae ves lane mr ansand / ‘ . tov ower! [ eis fe Se amet ie F Bens | | | | IP TOA 8) mo at ee Se emniy gre SAAN DIN 3 ghd Sia maton heb! oor aN = Woirecoeemevnr fhe t « @ Wvsts IH Rey sree OS BIivee > NG ASD wre Ae SS om Se ORE OS ee ene . —— | — A i / 0 + = 22 i i Bees ez as ares ta -as At is 14 > AVAL CIVIL ENGINEERING LAB i ESR ALIAS a RAR Fv (CF As (sere AEE Are tL Hanh ha rw raat amor Eh Per at ch lim h ing w up of finned modules showing close packi . Close- length of fins. igure 15 F 23 CRP rege sitesi eee i pense ing show! module experiment in NCEL saltwater test tank instrument leads and electrical Dower cable. - igure 16. Full-scale 12 F 24 LL. shroud cwii | Notes: (1) ; = ; i aes | t pa pe co ea i} , Tei! : \y : en i ts Lig : x3 La ‘i e He bal iu i i, a> 5 ii f Sut } aS ? y i i i Te? i i i 1 ‘ : . f | J ‘i : t “! + 1 ’ is RUE Iah ye Renmalnrcee sey ment oat ek INE WORE sno mf i eh) p : } hem ik i es LC meet P| i} ay iG ey ae ™ } = { J ) ca Niyiiy Figure 22. Closeup of RTG heat transfer modules immediately after removal 30 day immersion. from Port Hueneme Hzrbor, Port Hueneme Harbor, -adherent scum on partially rusty simuleted hull. * 1on RTG upon removal from 30 day immers Figure 23. showing non 30 TOMAS 2 Sila Voit (5 th Hint aay jtevety een pmtt reta OM. diaasasi havvanedho yaahcgbi en oo ina iy hit on bias eoreaelbne sa babici st + t aol Figure 24. Closeup of RTG heat transfer module fins, following drying; showing early indication of marine growth. 31 *pasn sft € UofT zenby (ueaw) - faajum daap 103 tw Cada oo N ‘$2098 MoTTeys a0y S*0 ice S*0 (SE°0) (01 * €£°L) (8S€°0)(,01 X ES‘L) lel = z0l * €L*1 = (01 * 68°Z)(9"0) (Ot X ESL = (01 X 99°z) (8S"0) —eee—e—e————— 9°0 8S °0 x . x ° 20! 98°? 20! 99°C g?! x O4 3?! base! at I) 89°S Si) 60! x GEC *L OL X 80°1 6 129909 UT yadaq 1004-00040 (wIy [) 419304 moTTeYs —— Jajenees daeq pue motTeyg uy sadjoweireg razsueay yeoy aAt eaeduoy y dojouvirg "| eTqeL 32 Table 2. Calculating the Effects of At in Shallow Seawater Tests (Independent variable; surface temperature, t Assumed b Surface Q/w.0. Fin Area Temperature, Fins Ratio, Res tw (Base area to Reject, of fin) 8,850 Btu 2 : if Based on Approximation I ES j=) ZS ind ial || 5 So iT =) Cas | io wr w_ =h a mean 160\ fe tw e.g., for to = 80, h 2 Based on Q = hace, where A = 0.378. c : s c Re 1s defined here as surface area ratio of a particular fin system to a plain cylinder of same root diameter, with average surface temperature as shown. Valid approximation at depth in ocean. In shallow water, boiling would keep th below about 216°F. 33 Table 3. Thermal Conductivities of Some Potential Building Materials Thermal Conductivity, k [(Btu/ (hr) (ft2) (°F/in. )] Silver Copper Aluminum alloys Magnesium alloy 90-10 Copper-nickel 70-30 Copper-nickel Lead, pure Monel 174 Stainless steels 90 to 180 Hastelloy (d) 145 Titanium, pure 118 Typical alloy 60 2 Shown for comparison; has highest value of any known metal. Maximun. Typical. 34 a ae H ay yr “au thw Seangit aed go ad | he wei ‘pried etnies yo} Hpenda 4 ae Ni «hoe anneal ee 4a: ie hye LAE cS WA YEN AP NORM MRAM WE ASH DEORE RC AAS panuyzquoo o0f 99 paddtyTy ¢ssuppeaa on (VW) OL dandy uy uMmoys sduppeay ‘uzW Q{ AO0jJ uo tamog JJO Aamog JolS = dinyelraduaj 193BM F9dF Jo sdeq aaiy. fTeoz aja 4aKedqS syauuay OL Cv (jnduy samod MyA-Q°E FopNpow op tuys soanqeaodway, asay, E261 Ttadv ¢ jo uoyavynyey, ~ suoF yd} A0J (40) oAN VAIdWOL 7, OTqey OO OVO Ocol OcOl OLOL 0004 0S60 060 Oup 35 ae Lea reteeteiti om oman, aera ater abit nt 4 *satdnoosow19y JO uo, Roo], 1074 (RP) QOL aanByty oog % JjJO TamMog il j 00v1 OSEL Orel OLCL reo -}349A 03 yIeq poddrTy, OleL OOEL "urWw QZ AOJ popuszzeun IJOLT BaAdAsep —CQ poddTy, syiBwsy 2 : i Outpt. po BMOT VO FOP C1) oan Utoduoy, > : c (ponut ju0)) “Hh OLY], 36 eu ae + 1 thy i" We Us iveatda F i a ni by Nh y ; lin ibaa erp persmamirerer ne nr emt T a ' a) ; i itae ut ’ i ; 1 i i 1 ’ 2 i i : ii aya i i F é i A Ay y { i ; ; : Tt TU fs ‘ 3 s ; i ‘ at HacHtt i} LECH P RMS ETA PAGRIPETR ST NIRA Ha a. pen seen aegRy Bhey Um R-RE PT TTD hatind me tem RE { } | } ponuy qu0y Oct *sooizop O€ PIITTI tssupypear ON OLLt OOtL OsOL 0701 ofol O10l exeren! 0S60 “JjJO tanog 04760 "dots = vinjzeradusay aaqem Sao, jo seq 0”60 aaiy. fyedpqyzean 31eIS ouTL ' syxreUsy oanqvaoduoay, (anduy aomod My-1*E taTNpow oy, suys) i soanjesoduay sol €261 THady » Jo voz aetnqey “¢ 9TqQPLY, WyalO//2MAALDRL MENA I ACHAT Pacem a ky conan ram reine 37 iH 7 i 0 + ‘ i it i! i 1 : " wae sate < o vet seagate acer lm Nyda aaa 4 mena eek ean Pye RA v i si Uae ie 4 ‘ i 1 Te) nay j i y 3 i i ; j F i AY i ~s : i - ? a it ‘\ } a 1 H i Ree f i i a i Be aa tbat \ ' ! y ; a t wy "SUOFIBOOT aydnodowrayy 4oy (LV) OQ] PANDEY vag, *JjO Jamog *‘wieM But jje3 gaqem ‘9u03 92T TLV | "81 GLI OL LOL cel OOo! elt LOL L6 "6 écl OSEL ZLt OOM 2G I BG ZZt | OveL ELI 86 86 96 lt | ocel *TBOFIAVA 0} yoeq paddty Orel col G'C8 td tZ ZOl OOtL col 18 12 LZ ZOl OSszl Z9L 62 OL G°69 901 Orel t9L cL 49 99 tOl OcZl *pasoto aATeA pue 909 poddfy Olcl O9l 08 G°L2 G*2ZL Lol OOZL O91 18 SL G2 00l OSLI LOL G*6L Gl Gl 2°86 OvLt LL v1 1 ct LL syiewsy out a SUOTILVOO'T IO4J (4) dAnjrvaodwoy, (ponuz juog) “Gg aT qvy, 38 Att nee ‘a my Wis ae 7 4 sje ys phe binkrmdrmlele @y ee ix i = 4 3% Se be sera Seer f t f Hes h Wi claire eats ARS Place Table 6. Comparison of Metal Temperature Changes With Water Temperature Change for a Single Vertical Lead-Finned Module (3-KW power input) Temperature (°F) at 1100 Hr 90 Thermocouple Location@ Temperature Difference (GE) 1400 Hr Tl T2 T3 T4 TS 48 T6 43 T7 46 T8 38 T9 (water) * See Figure 10 (a) for location of thermocouples. 39 s/c am rep brah comer tid oma ok ae ae Le aR stqunsonasts Jo merase ae £ Sa = Se u ; a | Table 7. Tabulation of 13 August 1973 Test Temperatures For Full- Scale, 12-Module Heat Transfer Unit (60,000-gallon seawater tank; 28-KW power input; 1130-hr readings) Thermocouple Number Temperature (°F) 2 See Figure 17 for thermocouple locations. Thermocouples M13-M17 are on adjacent module and are similar in location to M4-M9 shown in Figure 17. 40 a a WSR Sty Msvei, ofovormeedgeil! “Got atenly saz ube faesetha ao aim TINGE vet quesenedt. a at crite ois byte Ti age ae sMurpiad 2 ayqeg, tai asodiuo' 4 ‘sofdnorouUdyy JO UONB 0] JOy Z[ DUNT] 29g 5 pnoiys yum 02'04 | OL'02 ug saoutap 09 . mn q ; por ZEOL | Z'SO1 7701 wug soaadop o£ payn ug 1601 £01 LEI REEYN ; ; nor £001 +501 Wo I9A, A a un tL e tol Veot e7 SUONEBIO'T 10) (44) 2anaBsodtus J, Qndutdamod Ay-Z— yuE. wEMEas UOT|E34).00'09) JIUMLIS HU PON vores “OOO'OD UE MUL JaysUTTL WEP APNPOW-Z LE IPPLIS- TN soy saamesodway ¢261 daquiaidas ¢] purzy yo uonrinquy °g Sy LF OOFT OSEI Out] 41 i nue} Tels Tee Fy ee Wait ie + th we sinew Hele ul mr yy .; ’ \ , . ; , f ie ts ; i Vr iC ere eeu ] “0 ri ciliolata Ur AA oy f ‘ i ) ii) r peal ‘yostrdtioy yoy.) 4 ‘dae uy *suOKMIO, UNO OUI aOy Lf ony ag 1 P44 / 1} OOS Cel | Ue] PCOl | U'bo] b'Y6] U'ZOL | 792 “ed | o'sol 7 rOL | O'LOT O'COT O'COL |] O'TULE | HLIL |] 7 FOI 0 £2 3ny 01 uado uadgy 009 £'O# C'OR O'7R £670 | PL ATW SI uosdg usd | 19 B'9S 3 $69 BOL OLE P6700 | $2 UWS ado wade | oud OL j SOL Vk ; SRO'O bl "WY trdg O'R POS OS RYE Se i O90 | Po Wot wd BAL OS 06S : Ltd i OSL L710 | #2 Ges ZI wd ORL | § p'9S | 06% Hrd 'sd SHO'O | #4494 St O'RL | UadE RL 9S VOs Lye OSL SkO'O | +2994 FI OL | urdo Ved | 092 O'6S O19 wa Ore 8'6L O'OR PL 4d ed O29] ode yeh Powe | ues GiHS b'PL bod | OOK S'O# rh on ed hon i 4JOQgaryy OUMOUONEY - 4 (998/15) get | et | of | st] ot er | zi ol 6 x é 2 $ U EE) | IPL anny i po SUOLKIOT JOY (4),) Soamimgad tia (andut aomod AyyegZ) AOQIR HY OUDIUON EY Ut soanyrsoduiay WoL jo olryngry ‘6 dL 42 Appendix A METAL SELECTION FOR NATURAL CONVECTORS IN THE OCEAN Pure copper is not often used in direct contact with seawater in heat transfer devices, so its selection here should be explained. For hundreds of years, it has been recognized that of all the known metals and finishes, copper is the optimum for inhibiting marine growth. A frequent]y quoted rule of thumb has been that copper-bearing alloys typically are useful in inhibiting marine growth in proportion to the percent of copper present. The sheathing of wood sailing ships with copper to prevent attack is further justification for this observation. However, the usual application fer nonferrous materials in industrial heat transfer has been condenser tubes--for which copper without alloy is inadequate. This is because of the high velocities required in condenser tubes to give the good heat transfer necessary to make condensers of a reasonable size and cost. As a rule of thumb, there is no apparent corro- sion of copper at velocities below about 4 to 6 ft/sec, depending upon tube size, no matter how long it is immersed or exposed. At significantly higher velocities, the surface ‘washes’; a continually new layer of copper is exposed and removal rates in excess of 0.002 inch per year from this velocity effect are typical. For the heavy fins that are optimal in high capacity convectors, (for reasons given in the cesign section), such washing rates would not be prohibitive. However, estimates based on Grashoff’s number (107) might be expected to produce maximum velr ities of a few feet per second. From these low velocities and the unconfined nature of the flow, it was quickly determined that copper was the best available material because of its excellent properties in all other respects--corrosion, fouling, and high thermal conductivit:;. Since no washing of the (slightly oxidized) con- vectors has been observed to date, the choice appeers to be justified insofar as these experiments are concerned. Also, the tests in Port Huenete Harbor provided further evidence on this as well as the all- important consideration not within the scope of the laboratory-type experiments--fouling in the actual ocean environment. 43 “ai | ern aa 3s) Spe eat ieee - Ree TA iO ec -twoteny® 9 Bei ane anh nl, spay Meat 800.0 Io Resse oh eres akouges. agi ah pean Th ove | den be el Retee unimaon.itoen faokdove ngiags — (Gt) acne 2'Fioddnt) vo baeed-enteml saw | Wage: tq 208} we a to aerad “Lee mumh eon alt eu XE want, wit Yea wipe hank} no oan ade ee Chan fe bores yet ‘ghdntiavn sasd init ae fae tke gi tiot mkguawabemtadqens xorg, 12 re ey 7 i (bestbina hdo sesh hed te “archiva on AvoT A Nanas wth ronth JBporveodos gy Naiedded “fis! si te Taw ge ak) ino aanabive, rel yaw, eer caaaezodal ody te, aiteroe lait hea ase ~ Appendix B 30-DAY IMMERSION TESTS IN PORT HUENEME HARBOR Following completion of the controlled laboratory tests reported in the main body of the report, it was decided to immerse the unit under power in the Port Hueneme Harbor for a month. This additional testing would have several potential benefits: determination of RTG temperature at lower water temperatures than could be readily obtained in the test tank, the incipient formation of biological] scum, and the temperature effects of small tidal currents acting across the copper finned heat rejection surcéaces. The instrumentation for these tests was the same as used in the earlier laboratory tank tests, except that the test unit was fitted with a carefully calibrated Savonius Rotor Current Meter at the same level as the heat transfer surfaces, Figure 20. One thermocouple failed (opened) before any readings were taken, but since it was very expensive to remove the unit from the water and replace, the thermecouple was not repaired.* It was originally planned to set the entire RIG on the bottom on a hardwood pallet; a survey of the proposed area by CEL divers indicated suitable near-level areas. However, early attempts to stabilize tne RTG on the bottom showed that the bottom was neither level nor stable within a reasonable distance of the wharf. It was eventually suspended by a heavy synthetic line just off the bottom in about 25 feet of water (Figure 20). The bottom of the RTG was sufficiently close to the harbor bottom to stir up mud upon its retrieval. The heat rejection surfaces were about 19 feet below the water surface at mean tide. Typical terperatures taken during the 30-day immersion period are shown in Table 9 with varying currents. In the earlier laboratory tests, the power level was varied from the nominal 32 kw by changing the voltage from a portable generator. In these tests, power output was determined by the CEL dock-side voltage, yielding a heat dissipation of approximately 8 kw or about 11% less than the nominal. In Table 9 the temperatures taken in the harbor, where the experiment was exposed to water some 18°F colder than in the earlier laboratory tank tests, are compared with data taken 10 August 1973 in the tank (last column). All things being equal, it would be expected that typical metal temperatures in the harbor would be about 18°F colder than in the tank cue to the colder water. Also, the current effect, while probably slight and variabie, should be discernible. To allow direct comparison, the temperature differences are increased by the 18°F water temperature difference, Figure 21. * It was a surface temperature reading in a partially instrumented module and was not important in the analysis. 44 i] a Siwy: 7 ‘at ae Baas eeriene to me Th ywbis. OF iadodiiaite ® | watt Wht Oh (RaOTD PRdrsb od Tig, aie OTT wet acrn read ie ae Fae | eee fuvstraey palo sie Saba riser: mek p28 “tei YUE Wiehe “out gdaw ny ionepony vik ‘wl bes ‘Lidertedee ee vi pein au! rdersihiciaiee Sad hea us jonaotel elas pacity, as apn . Sistem. i like pia nk: iene 8 cose iri Mets yaoteredal selires whos Oh qinite tahoe Tara. Reel Ave Sy WK LOE Sr MUN OF asda Aah eee faamn healed ined Lope od. Kivgw tt ytmots aaboa. e dong et? ah Had dokhhs RPA pueda ad bheM Nod ted we ehh he white ae ait pea ie: Sree, oh eT ee SH? ond Wesaaa doe th Wed he it rabtbnrsag tp ae | i susie ig tag AORN, oily | ye club hae oie betnuenadent eink tse i a shaken ane Pi eye a DO ee lenin Lae aye: ite ing Re ee In Figure 21 the average of thermocouple readings for the important temperatures measured by 1, 2, and 3 junctions of the thermoelectric units in a complete RTG are plotted against measured velocity, irrespective of the direction of water flow. These are important data points because they provide the design temperature for the cold junctions in tne RTG. Point A is the average of the same readings for the August 1973 tank experiments, without flow. The velocities indicated are those at the beginning of a set of readings, but frequently they varied by 100Z% during the time required to take 18 thermocouple readings of the tenperature. The actual variability in average velocities for several minutes before a series of readings would be responsible for the small variations shown. Fluctuations in line voltage would also produce random variations; for the harbor data shown, there was no control of power input, nor was it ponitored continuously. In surmary, low velocities of the order of 0.1 to 0.3 ft/sec enhance the heat transfer from a convector in the ocean, but not significantly in terms of overall electrical generation efficiency. The important point is that any change, while transient and unpredictable, will always be conservative in that cold junction temperatures will be depressed and generator efficiency enhanced in the presence of ocean currents. NOTES ON CORROSION The experimental work discussed in this report is directed only to the heat transfer, particularly to provide a minimum temperature at the cold junctions of the RTG’s thermoelectric units. The planned tests required only short-term, periodic immersion during which no significant corrosion would be expected. Surface treatment of the large steel shell, Figure 13, was entirely cosmetic and did not include extensive precleaning or priming. The ends of the heat pipes, which would have been capped with pure copper for long-term immersion, were capped with a more readily available, less expensive, and more easily machined aluminum alloy. The 30-day immersion tests discussed in this Appendix provided a2 brief check on the corrosion handbooks. Since aluminum alloys are electrically much more active in seawater than either copper or steel (with E;* of -0.80, -0.70, and -0.40 volt, respectively, for aluminum alloy, steel, and copper), the aluminum caps corroded severely. There was some corrosion of the steel adjacent to the copper fins and none of the copper (Figure 22). Several alternatives, none of which will probably afford complete protection from corrosion for all parts of the RTG, are available in the final RTG design. One solution might be to cover all the RTG surfaces with heavy copper plating, either electrolytic or rolled. Welding would introduce holidays in a rolled: plating, so electrolytic plating might be the more acceptable of these approaches. Electroplating the entire * Ey is the no-current voltage versus a saturated calomel electrode. H g 45 aon piers bay sigsonis} tts od avevis City oat shor iweteotkh #? ma Sts! Te res tage 7 an eSen fy : ert ne soe bey ni, oak oa Ait eben kee e om | fae3st bergen) bre joelencdy ofits 493 a J LS Bae tetamreah ME bith: ao Ie agmst pe raxee ys nbaio dy abeonery “ets of beoa i tralia en addy gates rede vowel 9hF i] Se ek ts Te Are tates, le Pe faweg ee tao Ve but fon Bit bein binders: peooed!. 4 Wkuies Hah tbsp ‘pesdhy ( hep rn By bsqqed Sia pioheysn wiss* snot 4 PT (\Votks sal thska Vilsne sive bad Pies pS ages A AS Jolie h bablwedy Hiareged eta owt Naedb ates { faak S24 29elin OVniAe seett .7anodas aa ; ie (iiaker fawra i on YeNhtTe Bead ve O71 Bitte ah cays om ‘ote a nate 5 uts Yo) yisvisssqes