Anda di halaman 1dari 57

Marzo 1.

982

Predict storage-tank heat transfer precisely


Use this proced re to deter!ine the rate of heat transfer fro! a vertical storage tank "hen shortc t !ethods are inade# ate.
Jimmy D. Kumana and Samir P. Kothary , Henningson , Durham and Richardson , Inc. Heating or cooling storage tanks can be a major energy expense at plants and tankfarms. Though many procedures for calculating such heat-transfer re uirements ha!e been published [1 , 3 , 5 , 7 , 8 , 10] , the simplifying assumptions that they use can lead to significant errors in computed heat-transfer rates. This is of concern because efficient si"ing of tanks , insulation , heaters and coolers depends on accurate estimates of heat transfer to and from the !arious tank surfaces . #nd the ultimate !alue of being accurate increases as energy costs continue to rise. The procedure presented here determines the heat transfer to or from a !ertical-cylindrical storage tank seated on the ground - like the one in Fig. 1 . It includes the effects of tank configuration, li uid le!el , ambient temperature and wind speed, as well as temperature !ariations within the tank and betweeen air and ground. # partially worked example shows how to use the techni ue , and how to do the calculation on a computer. The theory. $torage tanks come in many different shapes and si"es. Hori"ontal-cylindrical and spherical tanks are used for storage of li uids under pressure % atmospheric tanks tend to be !ertical-cylindrical , with flat bottoms and conical roofs as shown in Fig. 1 . The example presented here is for the latter configuration , but the procedure applies to any tank for which reliable heat-transfer correlations are a!ailable.

Fig.1 - Typical vertical-cylindrical tank &or the sake of simplicity , we assume that the tank contents are warmer than the ambient air , and that we are concerned with heat loss from the tank rather than heat gain. 'ut the method may , of course , be applied to either case. (onsider , then , the categories of surfaces from which heat may be transferred across the tank boundaries ) wet or dry side walls, tank bottom , and roof. In the context used here , "wet" refers to the portion of the wall submerged under the li uid surface , whereas "dry" refers to the portion of the wall in the !apor space , abo!e the li uid surface. In general, the heating coils would be located near the bottom of the tank , in the form of flat "pancakes". Therefore , the temperature of the air *or !apor+ space abo!e the li uid le!el may be expected to be lower than the li uid itself . ,xpereience has shown that the a!erage bulk temperatures of the li uid and !apor space may be significantly *i.e. , more than 5 F ) different , and they are treated accordingly in our procedure. Use of differnt liquid and vapor temperatures is an important departure from the traditional approach , which assumes the same value for both.

Therefore , the temperature of the air *or !apor+ space abo!e the li uid le!el may be expected to be lower than the li uid itself . ,xpereience has shown that the a!erage bulk temperatures of the li uid and !apor space may be significantly *i.e. , more than 5 F ) different , and they are treated accordingly in our procedure. Use of differnt liquid and vapor temperatures is an important departure from the traditional approach , which assumes the same value for both. 6ur basic approach is to de!elop e uations for calculating the heat loss from each of the four categories of surfaces , and then add the indi!idual heat losses to get the total heat loss. Thus ) &or dry sidewall ) &or wet sidewall ) &or tank bottom ) &or tank roof ) Total ) qd = Ud Ad (TV - TA) qw = Uw Aw (TL - TA) qb = Ub Ab (TL - TG) qr = Ur Ar (TV - TA) Q q d + q w + qb + q r , ., .. , ./ , .0 , .1

7hen using these e uations in design or rating problems , we either assume the !arious temperatures for typical conditions or determine them by measurement. The area !alues are also easy to obtain ) !d = " D (L - Lw) ! w = " D Lw !b = " D2 / !r = (" D /2) (D2/ + d2)0,5 , .2 , .3 , .4 , .5

The complications arise when we try to estimate the o!erall heat-transfer coefficients Ud , Uw , Ub and Ur , for the four surfaces of the tank. &or the tank geometry chosen , these can fortunately be calculated from the indi!idual film heat-transfer coefficients in the con!entional manner , using published correlations.

The overall coefficients. Table I shows the component coefficients for each surface . The o!erall heat-transfer coefficient for the dry sidewall of the tank *Ud+ is calculated as the sum of the resistances of !apor film , fouling , metal wall , insulation *if any+ , and outside air *con!ection plus radiation+.

Ta$le % %ndivid al fil! heat-transfer coefficients Type * rface


Inside #all construction Outside Fouling

&ry "all hvw tM + t I kM kI


-1

'et "all hLw tM + t I kM kI


-1

(oof hVr tM kM
'f.hAr + hRr
-1

)otto! hLb tM kM hG hFb


-1

'f.hAw + hRd 'f.hAw + hRw

hFd

hFw

hFr

Note : Tank roof and bottom are uninsulated.

The outside-air heat transfer coefficient * !Aw+ is a function of wind !elocity as well as temperature gradient. 8ata on the effect of wind !elocity and $ T have been presented by Stuhlbarg [10] and Boyen [2] . ith a little bit of manipulation , their data were replotted , yielding the !wind enhancement factor !" "#) in #ig. $ .

'y definition ) ## = !Aw / !$Aw = !Ar / !$Ar , .*-9+ Therefore , once the outside-air coefficient for still air * !$Aw+ is known , the o!erall dry-sidewall coefficient at !arious wind !elocities can be computed as ) %&'d = 1/!Vw + %&/'& + %(/'( + 1/("#.!$Aw + !)d) + 1/!Fd $imilarly , the o!erall coefficients for the wet sidewall, bottom and roof surfaces are ) %&'w = 1/!Lw + %&/'& + %(/'(+ 1/("#.!$Aw + !)w) + 1/!Fw %&'b = 1/!Lb + %&/'&+ 1/!G + 1/!Fb %&'r = 1/!Vr + %&/'& + 1/("#.!$Ar + !)r) + 1/!Fd , .*--+

, .*-.+ , .*-/+ , .*-0+

*q.13 and 1 assume that the roof and bottom are not insulated , which is generally the case in temperature climates. 7e shall now re!iew correlations for the indi!idual heat-transfer coefficients needed to obtain the o!erall coefficients. %ndivid al fil! heat-transfer coefficients. The film heat-transfer coefficients may be di!ided into four categories ) con!ection from !ertical walls , con!ection from hori"ontal surfaces , pure conduction , and radiati!e heat transfer. 7ithin each category , correlations are presented for se!eral flow regimes. +ertical-"all fil! coefficients. These apply to the inside wall *wet or dry+ and the outside wall *still air+ . &or !ertical plates and cylinders , :ato et al. [+] recommend the following for li uids and !apors ) (,- = 0,138 ,Gr0,3+ . (,.r0,175 - 0,55) where , ),% * (.r / 0 and ,Gr 0 101 &or isothermal !ertical plates , ,de [ ] reported the following for li uids ) (,- = 0, 15 (,Gr . ,.r)0,25 where , (.r 0 100 and 10 / (,Gr . ,.r) / 101 and for gases ) (,- = 0,0215 . ,Gr0, 0 . ,.r0, 7 (1 + 0,5 . ,.r0,+7)-0, where , (.r 5 and (,Gr . ,.r) 0 101 &or !ertical plates taller than 3 #% , $tuhlbarg [10] recommends ) h where , %) / (,Gr . ,.r) / 101 ,orizontal-s rface heat-transfer coefficients. These coefficients apply to the roof and inside-bottom surfaces of the tank . The bottom is assumed to be flat. &or surfaces facing up [8] ) (,- = &or surfaces facing down ) (,- = ),%+ /(Gr . ,.r)0,33 , .*-5+ ),+, . - . .-0,75 (,Gr . ,.r)0,25 , .*-4+
0

, .*-1+

, .*-2+

, .*-3+

),01 /(Gr . ,.r)0,25

, .*.9+

'oth e uations apply in the range 22107 / ,Gr . ,.r / 321010

-o!enclat re.
! 63 D d g #rea of heat-transfer surface , ft $ % Ab for bottom , Ad for dry wall , Aw for wet wall , Ar for roof. $pecific heat at constant pressure , 'T<=lb- # 8iameter of tank , ft. Height of conical roof at center , ft. #cceleration due to gra!ity , .17 2 108 #%/!2 0-3999999 Indi!idual film coefficient of heat transfer , 'T<=ft $h-# % !Aw for air outside the walls , !Ar for air above the roof , !$Aw and !$Ar for still air , !Lw for liquid between the walls , !Lb for liquid near the bottom , !Vw for vapor near the walls , !Vr for vapor near the roof. &ouling coefficient , 'T<=ft$ h-# % !Fw for liquid at the walls , !Fb for liquid at the bottom , !F4 for vapor at the walls or the roof. Heat-transfer coefficient for ground , 'T<=ft $ h-# Heat-transfer coefficient for insulation , 'T<=ft $ h-# Heat-transfer coefficient for metal , 'T<=ft $ h-# Heat-transfer coefficient for radiation , 'T<=ft $ h-# % !)b for bottom , !)d for dry wall , !)w for wet wall , !)r for roof. Thermal conducti!ity , 'T<=ft-h-# % 'G for ground , '( for insulation , '& for metal wall. Total length for heat-transfer surface , ft. Total length for wetted surface , ft. ?rashof number , L3. 52. g. 3.$T / 42 @usselt number , !.D/' or !.L/5 Arandtl number , 63.4/' ;ate of heat transfer , 'T<=h Indi!idual rate of heat transfer , 'T<=h % qb for bottom , qd for dry wall , qw for wet wall , qr for roof. Temperature , # %TA for ambient air , TL for bul& liquid , TV for vapor , TG for ground , Tw for inside wall , Tw7 for outside wall. Temperature difference , #. $urface thickness, ft % t' for insulation , t( for metal. 6!erall heat-transfer coefficient , 'T<=ft $ h-# % Ub for bottom , Ud for dry wall , Uw for wet wall , Ur for roof. 7ind enhancement factor. >olumetric coefficient for thermal expansion , #)* >iscosity of fluid , lb=ft-h 8ensity of fluid , lb=ft+ . ,missi!ity.

hF hG h( h& h) . .w (Gr (,(.r Q q 2 $2 t ' ## 3 4 5

.# ivalent coefficients for cond ctive heat transfer. The wall and insulation coefficients are deri!ed from the thermal conducti!ities ) h&= -& / % & , .*.-+ h( = -( /%( , .*..+ The coefficient for heat transfer to and from the ground is the coefficient for heat conduction from a semiinfinite solid [1] ) hG = 7 . -G / " .D , .*./+

Fo ling coefficients. The coefficients !Fd , !Fw and !Fb apply to the !apor and li uid at the wall, and the li uid at the bottom of the tank , respecti!ely. These are empirical, and depend on the type of fluid and other factors such as tank cleaning . ?enerally , !Fd is the greatest of the three , and !Fb the least , indicating that the greatest fouling resistance is at the bottom of the tank.

Fo ling coefficients. The coefficients !Fd , !Fw and !Fb apply to the !apor and li uid at the wall, and the li uid at the bottom of the tank , respecti!ely. These are empirical, and depend on the type of fluid and other factors such as tank cleaning . ?enerally , !Fd is the greatest of the three , and !Fb the least , indicating that the greatest fouling resistance is at the bottom of the tank. Equ vale!t "#eff " e!t f#r rad at ve heat tra!$fer . The coefficient for sidewalls and roof depends on the emissi!ity of these surfaces , and is gi!en by [8] ) h)= , .*.0+

7ith these relationships , we now ha!e the tools to calculate heat transfer to or from the tank. 89!:P.8. #'( (hemical (orp., has a single manufacturing plant in the <.$. , and exports a high-!iscosity specialty oil product to ,urope. The oil is offloaded in Aort (ity , and sored in a flat-bottom , conical-roof tank rented from BCD Terminal (o. Etd. The tank is located outdoors and rests on the ground . It is e uipped with pancaketype steam-heating coils because the oil must be maintained abo!e 50 F in order to preserve its fluidity. 6ther pertinent data are ) tank diameter is 20 #% % tank height is 8 #% *to the edge of the roof+ % roof incline is %&'- ! (er f##t % tank sidewalls are 3/1+ - in carbon steel % insulation is 1 1/2 -in fiberglass , on the sidewall only. BCD Terminal (o. , does not ha!e metering stations on the steam supply to indi!idual tanks, and proposes to charge #'( (hemical for tank heating on the basis of calculated heat losses, using the co!entional tables [*] , and assuming a tank wall temperature of 50 F. ,he pro-ect engineer from .B/ /hemical decided to investigate how 0123s estimate would compare with the more elaborate one described in this article.

&irst, the engineeer collected basic data on storage and climate. 6il shipments from the <.$. , arri!e at Aort (ity approximately once a month , in 100,000-g89 b8%:!;7. 8eli!eries to local customers are made in 8,000g89 %8<'%r-:'7 , three times a week on a!erage. The typical !ariation in tank le!el o!er a %)-da* period is known from experience. The ambient temperature goes through a more complex cycle, of course. 7ithin the primary cycle of 3+5 d8=7 , there are daily temperature !ariations. 'ut in the seasonal cycle, heat supply is re uired only during the winter months, when temperatures fall well below 50 F. 7ind conditions at the storage site are not as well defined, and therefore much harder to predict. Howe!er, we can assume that the wind speed will hold constant for a short period of time, and calculate the heat loss for this unit period under a fixed set of conditions. The wind speed to be used must be based on the known probability distribution of wind speeds at the site. The procedure for determining the annual heat loss consists of adding up the heat losses calculated for each unit period *which could be 8< !>-r , 12 !>-r7 , 2 !>-r7 , or 30 d8=7 , as appropiate+. This example demonstrates the calculation of heat loss for only one unit period , 12 !>-r7 , using an ambient temperature of 35 F , a wind velocity of 10 ?3!. , and a liquid level of 50@ . ,he other data required are given in Table II . 4ote that the liquid temperature is controlled at 55 F to provide a 5 F margin of safety.

&ata for /)0 0he!ical 0o. 1 e2a!ple


Physical Properties
&ensity 1 l$:ft
%

Ta$le %%
/ir
7.78 7.28 7.779 7.7181 7.772

3i# id
5.68 7.6 57.77 7.12 7.777771

+apor 4
7.78 7.28 7.779 7.7181 7.772

*pecific heat 1 )TU:l$ -F +iscosity 1 cP Ther!al cond ctivity 1 )TU:ft-h-F 0oefficient of vol !etric e2pansion per F /ss !ed fo ling coefficients &ry "all 'et "all (oof

11777 877 11777

)TU:ft+h-F

)TU:ft+h-F )otto! Ther!al cond ctivities Metal "alls %ns lation ;ro nd * rface e!issivity 'all and roof Te!perat res +apor in Tank 87 3i# id in Tank 88 F < tside /ir =8.77 ;ro nd 57 4 *ince the li# id has lo" volatility 1 the vapor space is ass !ed to $e !ostly air. $ince the Arandtl and ?rashof numbers occur repeatedly in the fim heat-transfer coefficient e uations , and remain relati!ely unchanged for all the conditions of interest , let us first calculate their !alues . Thus , for the li uid phase ) (Gr = (.r = .3.52.g.3.$T / 42 = 6..4 / ' = ;1,, . .3. $T 585 7.9 17.77 7.728 7.87 )TU:ft-h-F 877

$imilarly , for the !apor phase , (Gr = .3.52.g.3.$T / 42 = (3r = 7.28

%,;) < %)7 . L3. $T

7e can now calculate the indi!idual film heat-transfer coefficients , using the appropiate L and $T values in the 5rashof)number equations. ,his is an iterative process that requires initial estimates for wall and ground temperatures , plus wall temperatures. ,#eff " e!t f#r va(#r at wall *!Vw) . #s an initial approximation , assume that the wall temperature is the a!erage of the !apor and outside-air temperatures ) 2w = *19 F /.+=. G 5218 >F 1.99.?12 Then find the ?rashof number ) (Gr = 1,102107 2 (L - Lw)3 2 (TV - Tw)

,mploying *q. 15 , find the @usselt number and then the coefficient , 7.7181 ft . 58.77 .w = 25.77 ft then , (,- = 0,138 ,Gr0,3+ . (,.r0,175 - 0,55) = hVw = ,,- . ' / (L - Lw) = 921.1 7.881
'T<=ft$)h)6#

,#eff " e!t f#r l qu d at the wall *!Lw) . Here , neither ,.r nor *,Gr ,.r+ falls within the range of the applicable correlations **q. 1+ , 18+ . Eet us try both , again using an a!erage for Tw ) 2w = (TL + TA) / 2 = (Gr = 17, 7 .L3.(TL - Tw) = 58.77 1.=8.?79 H&

<sing *q. 1+ 8<d 18 , we get two estimates for the heat-transfer coefficient ) k= 7.12 NPr = 585 hLw = (0, 15.' / Lw).(,Gr . ,.r)0,25 = hLw = (0, 5.' / Lw0,75).(,Gr.,.r)0,25 = 7.975 1.518
'T<=ft$)h)6# 'T<=ft$)h)6#

To be conser!ati!e , we use the higher !alue ) hLw = 1.518 'T<=ft$)h)6#

,#eff " e!t f#r va(#r at r##f *!Vr+ . 7e consider this a flat plate with ) ft Diameter 27.77 and use *q. 20 , again with ) 2w = 52.8 F 7.7181 (Gr = 1,102107 . D3 (TV - Tw) = hVr = (0,27.'/D).(,Gr.,.r)0,25 = 1.15.?79 7.185
'T<=ft$)h)6#

,#eff " e!t f#r l qu d at ta!k b#tt#- *!Lb+ . #ssume that the ground temperature * TG+ is 5 F above ambient , and use an average of liquid and ground temperatures as a first appro7imation for the tan&)bottom temperature 8 2w = (TL + TG)/2 = /2L + T8?b + 5 ) = 5918>F

Then , figure the ?rashof number , and use *q. 11 to get the coefficient ) (Gr = 17, 7.D3.(TL - Tw) = 8.88.?76 (Gr . ,.r = hLb = 2.8=.?79 1.178
'T<=ft$)h)6#

,#eff " e!t f#r #ut$ de a r at r##f *!$Ar+ . #ssume ) 2w7 = Tw since the roof is uninsulated , and get the coefficient for still air from *q. 11 ) (Gr = 1,12107.D3.(T"7 - TA) = 1.15.?12 h=Ar = (0,1 .'/D).(,Gr.,.r)0,33 = 7.66= 'T<=ft$)h)6# ,#eff " e!t f#r #ut$ de a r at wall *!$Aw+ . #ssume that the temperature drop across the film is one-fourth of the drop from the inside fluid to the outside air *a!eraged for the wet and dry walls+ , and use *q. 15 to find the coefficient ) $2 h=Aw = -3,1=0 G 51=98 >F (Gr = 1,12107.L3.AT = 7.81
'T<=ft )h)6#
$

9.19.?12

,#!du"t #! "#eff " e!t$ f#r .r#u!d / -etal wall/ a!d !$ulat #! *!G , !& , and !(+ . These are straightforward from *q. 21-23 ) h& = '& / %& = h( = '( / %( = hG = 8 . 'G / B.D 657 7.225 7.172
'T<=ft$)h)6# 'T<=ft$)h)6# 'T<=ft$)h)6#

(adiation coefficients for dry and "et side"all 1 and roof @ !)d / !)w / !)r0 . #s for the outside-air film coefficients , assume that ) 2w7 = TA + 0,25 (Tb-9' - TA) where , Tb-9' is the temperature of the li uid or !apor inside the tank , if the surface is insulated. &or the uninsulated roof , assume that ) 2w7 = TA + 0,5(TV - TA) Then , 2w7 = 2w7 = 2w7 = =8.98 57.77 52.8 F F for the *insulated+ dry sidewall , for the wet sidewall ,

for the roof. F <sing , . .0 , find the coefficient for each of the three cases ) h)d = 'T<=ft$)h)6# 7.989 h)w = h)r = 7.989 7.968
'T<=ft$)h)6# 'T<=ft$)h)6#

0losing in on res lts.

Table III summari"es the heat-transfer coefficients just calculated , including the corrections for wind - !$Aw and !$Ar are multiplied by 3,3 and 3,1 , respectively , based on data for 10-?3! wind in Fig. 2 . $ubstituting these indi!idual coefficients in *q. 11-1 , we obtain the U !alues listed in T8b9; (((. T/)3. %%%

,eat-transfer coefficients after first iteration


0oefficient hVw hLw hVr hLb hG hAAr hArC hAAw hAwC h& h( hF h) '> I &or 10-?3! wind. 7.81 1.68= 657 7.225 11777 7.9868 ).%,%? 7.81 1.68= 657 7.225 877 7.9895 ).%707 11777 7.9681 ).%+,1 877 ).);@@ 657 657 7.66=8 2.789 &ry "all 7.8818 1.518 7.18=9 1.178 7.172 'et "all (oof )otto!

7hat remains to be done J 7hen we began the calculations , we assumed that the outside-wall temperatues were related to the bulk-fluid temperatures by ) 2w = TA + 0,5 . (Tb-9' - TA) 2w7 = TA + 0,25 .(Tb-9' - TA) for uninsulated surfaces for insulated surfaces

In order to calculate accurate coefficients for heat transfer , we must now obtain better estimates of these wall temperatures. This re uires an iterati!e procedure that can be programmed and run on a computer. &or dry wall , the rate of heat loss is gi!en by all three of the following ) qd = Ud.Ad.(TV - TA) , .*.1+ qd = !Vw . Ad . (TV - Tw) qd = (!)d + !Aw).Ad.(Tw7 - TA) $ol!ing *q. 25 8<d 27 for Tw7 yields ) 2w7 = (Ud/(!)d + !Aw)).(TV - TA) + TA $imilarly , sol!ing *q. 25 8<d 2+ for Tw yields ) 2w = TV - (Ud/!Vw).(TV - TA) , .*.5+ <sing the same approach , now calculate Tw and Tw7 for the wet wall , and Tw for the roof and bottom of the tank. To find the correct wall temperatures , use the initial estimates of U and ! !alues in *q. 28 8<d 21 *and in the parallel e uations for the other surfaces+ to get new Tw and Tw7 !alues . Table IV shows these temperatures after a second iteration . <sing these new temperatures , recompute ?rashof numbers , indi!idual heattransfer coefficients and o!erall coefficients , and then iterate again to get a new set of Tw and Tw7 !alues . hen the current and previous iteration3s temperature estimates are the same "within a specified tolerance) , the iteration is completed. , .*.4+ , .*.2+ , .*.3+

T/)3. %+ *econd iteration yields closer te!perat re esti!ates


Te!perat re Tw 1 !$ de0 / F %teration 2 1 &ry "all 56.7 52.8 'et "all 82.9 58 (oof =8.98 52.8 )otto! 8=.9 59.8

Tw7 1#ut$ de0 / F

2 1

=8.9 =8.98

=6.8 57

=8.98 52.8

Table V lists the indi!idual and o!erall coefficients after the second iteration . .lthough it is clear that additional iterations are needed , let us accept these values as sufficiently accurate for the present purpose . Then we can obtain the total heat-transfer rate * D+ by using the U !alues in *q. 1-5 and summing. T/)3. +

,eat-transfer coefficients after second iteration


0oefficient hVw hLw hVr hLb hG hAAr hArC hAAw hAwC h& h( hF h) &ry "all 7.56= 7.=19 1.759 657 7.225 1777 'et "all 7.98 7.=19 1.759 657 7.225 877 (oof 7.181 7.=1 7.96 657 1777 )otto! 7.619 7.172 657 877

7.98 7.9815 7.98 '> 7.1=9= 7.1688 7.16=6 7.7898 I &or 10-?3! wind. Table VI shows the calculated heat-transfer rates through each boundary , and the total rate . @ote that the roof and bottom of the tank account for only slight heat loss , despite being uninsulated.

T/)3. +% (ate of heat transfer d ring nit period


* rface U/ Ar;8 / 2T3&ft+h-F ft+ $2 1 F q/ 2T3&h

&ry "all 7.1=92 1878 18 =158.975 'et "all 7.1688 1878 27 5991.58 (oof 7.16=6 =18 18 99=.71 )otto! 7.7898 =15 18 512.128 T<T/3 =658 91=28.= @ote ) Total for 12-! 3;ri>d is 111.10 ETU. This , of course , is for the unit period of time , when wind speed is 10 ?3! , the tank is half full , and the air is 35 F . Table VII shows how the results of unit-period heat losses can be tabulated and added to get the cumulati!e heat loss for a month or year. 6f course , this re uires climatic data and tank-le!el estimates for the o!erall time-period.

T/)3. +%% * !!ing losses for nit periods yields heat loss for =7 days
Period 1 2 = 3i# id level 1 A 87 87 5= TA =8 29 52 F 'ind speed 1 !ph 17 8 7 ,eat loss 1 )TU 1111975 =921579 521891 -

52 9= 89 86 67 59 Total for =7-day period B

88 58 67

=7 27 18

7 121=68 7 81=891787

0o!parison "ith other !ethods. #erstin and $treet F1G offer a !ery simple method for calculating heat loss from tanks. &or a tank with 1,5 i< of sidewall insulation, and a wind speed of 10 ?3! , the recommended o!erall U *based on ' = 0,011 for the insulation+ is 0,1 for AT = +0 HF and 0,1 for AT = 100 HF. .d-usting these values for ' = 0,028 and AT = 17 HF , as in our e7ample , yields an overall U of 0,20+ ETU/#%2!-HF . ,he total e7posed surface is 3.331 #%2 "tan& bottom not included) , and thus the overall rate of heat transfer by their method is 8

Q 9,.92 x ///- x -3 111666 )TU:h This compares with a heat loss of 8.113 ETU/! *for the exposed surface+ calculated by the procedure of this article-see Table >I . Thus their method yields a result /-O too high in this case. $tuhlbarg [10] takes an approach similar to that proposed here , but his method differs in how the outside tankwall film coefficient is computed. $tuhlbarg recommends the use of a manufacturerNs data table , and does not explicitly distinguish between the bulk li uid temperature in calculating the proper heat-transfer coefficient. The algebraic method of Hughes and 8eumaga [5] resembles the one presented in this article in many ways. 'ut it does not recogni"e differences between li uid and !apor temperatures inside the tank , nor does it account for the interaction between $T and wind speed in calculating a wind)enhancement factor . #inally , even though their procedure requires iteration , the focus of the iterative efforts is to get better estimates of fluid properties , not tan&wall temperatures. 0oncl sions. 6ur engineer at #'( (hemical was able to negociate a significant reduction in the heating charges proposed by the BCD Terminal (o. , which had used a shortcut method for its estimate , because the procedure presented here is rational and defensible. # rigorous solution of the interations can easily be reached on a digital computer or e!en a programmable calculator , and the effort pays off in better design or operation criteria. :ar- .iBoCicD , 8ditor. (eferences. -. #erstin &. , and $treet ?. , "#pplied (hemical Arocess 8esign" . .. 'oyen K. E. , "Thermal ,nergy ;eco!ery" . /. (ordero ;. , "The cost of missing pipe insulation" . 0. ,de #. K. , "#d!ances in Heat Transfer " . >ol. 0. 1. Hughes ;. , and 8eumaga >. , "Insulation sa!es energy". 2. :ato, @ishiwaki and Hirata , Intl. K. , of Heat and Lass Transfer. 3. :ern 8. M. , "Arocess Heat Transfer". 4. Aerry ;. H. , and (hilton (. H. , "(hemical ,ngineersN Handbook". 5. ;ohsenow 7. L. , and Hartnett K. A. , "Handbook of Heat Transfer". -9. $tuhlbarg 8. , "How to 8esign Tank Heating (oils". The / thors. Ii??= D. 5-?8<8 is (hief Arocess ,ngineer at Henningson , 8urham and ;ichardson , Inc. , A.6. 'ox -.300 , Aensacola , &E. /.131 , where he is in!ol!ed in all aspects of engineering design . He holds a '. Tech. , degree from the Indian Institute of Technology , and an L.$. , from the <ni!ersity of (incinnati , both in chemical engineering. Lr. :umana is registered as a professional engineer in four states , and is acti!e in #I(h, as chairman of his local section. He holds a patent on distillation of fuel-grade ethanol. J8?ir .. 5>%!8ri was a process engineer with Henningson , 8urham and ;ichardson , Inc., when this article was written . He recently joined Hoffmann Ea ;oche Inc., @utley , @. K. 93--9 , where he is engaged in designing chemical process plants. Lr. :othari holds a '.$. , degree from L. $ayajirao *India+ and a n L.$. fro the <ni!ersity of (incinnati , both in chemical engineering . He belongs to #I(h, and has written se!eral technical articles.

J8?ir .. 5>%!8ri was a process engineer with Henningson , 8urham and ;ichardson , Inc., when this article was written . He recently joined Hoffmann Ea ;oche Inc., @utley , @. K. 93--9 , where he is engaged in designing chemical process plants. Lr. :othari holds a '.$. , degree from L. $ayajirao *India+ and a n L.$. fro the <ni!ersity of (incinnati , both in chemical engineering . He belongs to #I(h, and has written se!eral technical articles.

Marzo 1.98

Predict storage-tank heat transfer precisely


Use this proced re to deter!ine the rate of heat transfer fro! a vertical tank "hen shortc t !ethods are inade# ate.

Jimmy D. Kumana and Samir P. Kothary , Henningson , Durham and Richardson , In

Heating or cooling storage tanks can be a major energy expense at plants and tankfarms. Though ma procedures for calculating such heat-transfer re uirements ha!e been published [1 , 3 , 5 , 7 , 8 , 10] , the sim assumptions that they use can lead to significant errors in computed heat-transfer rates. This is of con because efficient si"ing of tanks , insulation , heaters and coolers depends on accurate estimates of h to and from the !arious tank surfaces . #nd the ultimate !alue of being accurate increases as energy c continue to rise.

The procedure presented here determines the heat transfer to or from a !ertical-cylindrical storage tan the ground - like the one in Fig. 1 . It includes the effects of tank configuration, li uid le!el , ambient te and wind speed, as well as temperature !ariations within the tank and betweeen air and ground. # par example shows how to use the techni ue , and how to do the calculation on a computer.

The theory. $torage tanks come in many different shapes and si"es. Hori"ontal-cylindrical and spherical tanks are storage of li uids under pressure % atmospheric tanks tend to be !ertical-cylindrical , with flat bottoms a roofs as shown in Fig. 1 . The example presented here is for the latter configuration , but the procedur any tank for which reliable heat-transfer correlations are a!ailable.

&or the sake of simplicity , we assume that the tank contents are warmer than the ambient air , and th concerned with heat loss from the tank rather than heat gain. 'ut the method may , of course , be app case.

(onsider , then , the categories of surfaces from which heat may be transferred across the tank bound or dry side walls, tank bottom , and roof. In the context used here , "wet" refers to the portion of the wa under the li uid surface , whereas "dry" refers to the portion of the wall in the !apor space , abo!e the surface.

In general, the heating coils would be located near the bottom of the tank , in the form of flat "pan

Therefore , the temperature of the air *or !apor+ space abo!e the li uid le!el may be expected to be lo li uid itself . ,xpereience has shown that the a!erage bulk temperatures of the li uid and !apor space significantly *i.e. , more than 5 F ) different , and they are treated accordingly in our procedure. Use o liquid and vapor temperatures is an important departure from the traditional approach , which assume value for both.

6ur basic approach is to de!elop e uations for calculating the heat loss from each of the four categori surfaces , and then add the indi!idual heat losses to get the total heat loss. Thus ) &or dry sidewall ) &or wet sidewall ) &or tank bottom ) &or tank roof ) Total ) qd = Ud Ad (TV - TA) qw = Uw Aw (TL - TA) qb = Ub Ab (TL - TG) qr = Ur Ar (TV - TA) Q qd + qw + qb + qr

7hen using these e uations in design or rating problems , we either assume the !arious temperatures conditions or determine them by measurement. The area !alues are also easy to obtain ) !d = " D (L - Lw) !w = " D Lw !b = " D2 / !r = (" D /2) (D2/ + d2)0,5

The complications arise when we try to estimate the o!erall heat-transfer coefficients Ud , Uw , Ub and U four surfaces of the tank. &or the tank geometry chosen , these can fortunately be calculated from the film heat-transfer coefficients in the con!entional manner , using published correlations.

The complications arise when we try to estimate the o!erall heat-transfer coefficients Ud , Uw , Ub and U four surfaces of the tank. &or the tank geometry chosen , these can fortunately be calculated from the film heat-transfer coefficients in the con!entional manner , using published correlations.

The overall coefficients. Table I shows the component coefficients for each surface . The o!erall heat-transfer coefficient for th sidewall of the tank *Ud+ is calculated as the sum of the resistances of !apor film , fouling , metal wall , any+ , and outside air *con!ection plus radiation+.

Ta$le % %ndivid al fil! heat-transfer coefficients Type * rface


Inside #all construction Outside Fouling

&ry "all hvw tM + t I kM kI


-1

'et "all hLw tM + t I kM kI


-1

(oof hVr tM kM
'f.hAr + hRr
-1

)otto! hLb tM kM hG hFb


-1

'f.hAw + hRd

'f.hAw + hRw

hFd

hFw

hFr

Note : Tank roof and bottom are uninsulated.

The outside-air heat transfer coefficient *!Aw+ is a function of wind !elocity as well as temperature grad the effect of wind !elocity and $ T have been presented by Stuhlbarg [10] and Boyen [2] . ith a little bit manipulation , their data were replotted , yielding the !wind enhancement factor !""#) in #ig. $ .

+w !d 5 17 17 'y definition ) mph mph

4T 5 12.87 9.87 H& H&

'f 5 hAw / !$Aw = hAr / !$Ar =

## = !Aw / !$Aw = !Ar / !$Ar

Therefore , once the outside-air coefficient for still air *!$Aw+ is known , the o!erall dry-sidewall coefficie wind !elocities can be computed as )

%&'d = 1/!Vw + %&/'& + %(/'( + 1/("#.!$Aw + !)d) + 1/!Fd

$imilarly , the o!erall coefficients for the wet sidewall, bottom and roof surfaces are )

%&'w = 1/!Lw + %&/'& + %(/'(+ 1/("#.!$Aw + !)w) + 1/!Fw

%&'b = 1/!Lb + %&/'&+ 1/!G + 1/!Fb

%&'r = 1/!Vr + %&/'& + 1/("#.!$Ar + !)r) + 1/!Fd

, .-/ and -0 assume that the roof and bottom are not insulated , which is generally the case in tempe climates. 7e shall now re!iew correlations for the indi!idual heat-transfer coefficients needed to obtain coefficients.

%ndivid al fil! heat-transfer coefficients. The film heat-transfer coefficients may be di!ided into four categories ) con!ection from !ertical walls , from hori"ontal surfaces , pure conduction , and radiati!e heat transfer. 7ithin each category , correla presented for se!eral flow regimes.

+ertical-"all fil! coefficients. These apply to the inside wall *wet or dry+ and the outside wall *still a !ertical plates and cylinders , :ato et al. [+] recommend the following for li uids and !apors )

(,- = 0,138 ,Gr0,3+ . (,.r0,175 - 0,55)

, .*-1+

where ,

),% * (.r / 0 and ,Gr 0 101

&or isothermal !ertical plates , ,de [ ] reported the following for li uids )

(,- = 0, 15 (,Gr . ,.r)0,25

, .*-2+

where ,

(.r 0 100 and 10 / (,Gr . ,.r) / 101

and for gases ) (,- = 0,0215 . ,Gr0, 0 . ,.r0, 7 (1 + 0,5 . ,.r0,+7)-0,


0

where ,

(.r 5 and (,Gr . ,.r) 0 101

&or !ertical plates taller than 3 #% , $tuhlbarg [10] recommends )

),+, . - . .-0,75 (,Gr . ,.r)0,25

, .*-4+

where ,

%) / (,Gr . ,.r) / 101

,orizontal-s rface heat-transfer coefficients. These coefficients apply to the roof and inside-bottom the tank . The bottom is assumed to be flat. &or surfaces facing up [8] )

(,- =

),%+ /(Gr . ,.r)0,33

, .*-5+

&or surfaces facing down ) (,- = ),01 /(Gr . ,.r)0,25 , .*.9+

'oth e uations apply in the range 22107 / ,Gr . ,.r / 321010

-o!enclat re.
! 63 #rea of heat-transfer surface , ft$ % Ab for bottom , Ad for dry wall , Aw for wet wall , Ar for roof. $pecific heat at constant pressure , 'T<=lb-#

D d g

8iameter of tank , ft. Height of conical roof at center , ft. #cceleration due to gra!ity , .17 2 108 #%/!2 Indi!idual film coefficient of heat transfer , 'T<=ft$h-# % !Aw for air outside the walls , !Ar for air above the roof , !$Aw and !$Ar for still air , !Lw for liquid between the walls , !Lb for liquid near the bottom , !Vw for vapor near the walls , !Vr for vapor near the roof. &ouling coefficient , 'T<=ft$ h-# % !Fw for liquid at the walls , !Fb for liquid at the bottom , !F4 for vapor at the walls or the roof. Heat-transfer coefficient for ground , 'T<=ft$ h-# Heat-transfer coefficient for insulation , 'T<=ft$ h-# Heat-transfer coefficient for metal , 'T<=ft$ h-# Heat-transfer coefficient for radiation , 'T<=ft$ h-# % !)b for bottom , !)d for dry wall , !)w for wet wall , !)r for roof. Thermal conducti!ity , 'T<=ft-h-# % 'G for ground , '( for insulation , '& for metal wall. Total length for heat-transfer surface , ft. Total length for wetted surface , ft. ?rashof number , L3. 52. g. 3.$T / 42 @usselt number , !.D/' or !.L/5 Arandtl number , 63.4/' ;ate of heat transfer , 'T<=h Indi!idual rate of heat transfer , 'T<=h % qb for bottom , qd for dry wall , qw for wet wall , qr for roof. Temperature , # %TA for ambient air , TL for bul& liquid , TV for vapor , TG for ground , Tw for inside wall , Tw7 for outside wall. Temperature difference , #. $urface thickness, ft % t' for insulation , t( for metal. 6!erall heat-transfer coefficient , 'T<=ft$ h-# % Ub for bottom , Ud for dry wall , Uw for wet wall , Ur for roof. 7ind enhancement factor. >olumetric coefficient for thermal expansion , #)* >iscosity of fluid , lb=ft-h 8ensity of fluid , lb=ft+ . ,missi!ity.

hF hG h( h& h) . .w (Gr (,(.r Q q 2 $2 t ' ## 3 4 5

.# ivalent coefficients for cond ctive heat transfer. The wall and insulation coefficients are deri!e thermal conducti!ities )

h&= h( =

-& / %& -( /%(

, .*.-+ , .*..+

The coefficient for heat transfer to and from the ground is the coefficient for heat conduction from a se solid [1] ) hG = 7 . -G / " .D , .*./+

Fo ling coefficients. The coefficients !Fd , !Fw and !Fb apply to the !apor and li uid at the wall, and t the bottom of the tank , respecti!ely. These are empirical, and depend on the type of fluid and other fa as tank cleaning . ?enerally , !Fd is the greatest of the three , and !Fb the least , indicating that the gre resistance is at the bottom of the tank.

Equ vale!t "#eff " e!t f#r rad at ve heat tra!$fer. The coefficient for sidewalls and roof depends on emissi!ity of these surfaces , and is gi!en by [8] )

h)=

, .*.0+

7ith these relationships , we now ha!e the tools to calculate heat transfer to or from the tank.

89!:P.8. #'( (hemical (orp., has a single manufacturing plant in the <.$. , and exports a high-!iscosity speci product to ,urope. The oil is offloaded in Aort (ity , and sored in a flat-bottom , conical-roof tank rente Terminal (o. Etd. The tank is located outdoors and rests on the ground . It is e uipped with pancake-t heating coils because the oil must be maintained abo!e 50 F in order to preserve its fluidity.

6ther pertinent data are ) tank diameter is 20 #% % tank height is 8 #% *to the edge of the roof+ % roof inc ! (er f##t % tank sidewalls are 3/1+ - in carbon steel % insulation is 1 1/2 -in fiberglass , on the sidewa

BCD Terminal (o. , does not ha!e metering stations on the steam supply to indi!idual tanks, and prop charge #'( (hemical for tank heating on the basis of calculated heat losses, using the co!entional ta assuming a tank wall temperature of 50 F. ,he pro-ect engineer from .B/ /hemical decided to inves 0123s estimate would compare with the more elaborate one described in this article.

&irst, the engineeer collected basic data on storage and climate. 6il shipments from the <.$. , arri!e a approximately once a month , in 100,000-g89 b8%:!;7. 8eli!eries to local customers are made in 8,00 %8<'%r-:'7 , three times a week on a!erage. The typical !ariation in tank le!el o!er a %)-da* period is experience.

&irst, the engineeer collected basic data on storage and climate. 6il shipments from the <.$. , arri!e a approximately once a month , in 100,000-g89 b8%:!;7. 8eli!eries to local customers are made in 8,00 %8<'%r-:'7 , three times a week on a!erage. The typical !ariation in tank le!el o!er a %)-da* period is experience.

The ambient temperature goes through a more complex cycle, of course. 7ithin the primary cycle of 3 there are daily temperature !ariations. 'ut in the seasonal cycle, heat supply is re uired only during th months, when temperatures fall well below 50 F.

7ind conditions at the storage site are not as well defined, and therefore much harder to predict. How assume that the wind speed will hold constant for a short period of time, and calculate the heat loss fo period under a fixed set of conditions. The wind speed to be used must be based on the known probability distribution of wind speeds at the

The procedure for determining the annual heat loss consists of adding up the heat losses calculated fo period *which could be 8< !>-r , 12 !>-r7 , 2 !>-r7 , or 30 d8=7 , as appropiate+. This example dem the calculation of heat loss for only one unit period , 12 !>-r7 , using an ambient temperature of 35 F velocity of 10 ?3!. , and a liquid level of 50@ . ,he other data required are given in Table II . 4ote tha temperature is controlled at 55 F to provide a 5 F margin of safety. &C 3C 3w 5 Ta-b 5 +w !d 5 tM 5 tI 5 27.77 58.77 25.77 =8.77 17.77 =:16 1 1:2 ft ft ft H& mph in in tank sidewalls thickness insulation thickness *fiberglass - on the sidewall only+ tank diameter , ft. tank height , ft. li uid le!el , ft ambient temperature, H&

&ata for /)0 0he!ical 0o. 1 e2a!ple


Physical Properties
&ensity 1 l$:ft% *pecific heat 1 )TU:l$ -F +iscosity 1 cP Ther!al cond ctivity 1 )TU:ft-h-F 0oefficient of vol !etric e2pansion per F /ss !ed fo ling coefficients &ry "all 'et "all (oof )otto!

Ta$le %%
/ir
7.78 7.28 7.779 7.7181 7.772

3i# id
5.68 7.6 57.77 7.12 7.777771

11777 877 11777 877

)TU:ft+h-F

Ther!al cond ctivities Metal "alls %ns lation ;ro nd * rface e!issivity 'all and roof Te!perat res +apor in Tank 87 3i# id in Tank 88 F < tside /ir =8.77 ;ro nd 57 4 *ince the li# id has lo" volatility 1 the vapor space is ass !ed to $e !ostly air. 7.9 17.77 7.728 7.87 )TU:ft-h-F

$ince the Arandtl and ?rashof numbers occur repeatedly in the fim heat-transfer coefficient e uations relati!ely unchanged for all the conditions of interest , let us first calculate their !alues . Thus , for the l

(Gr = *ea @DE C (Gr =

.3.52.g.3.$T / 42 52.g.3./ 42 /9E . .3. $T 99.8

and , (.r =

6..4 / '

585

$imilarly , for the !apor phase , (Gr = .3.52.g.3.$T / 42 $ea *B+ G (Gr = and , (.r = 52.g.3./ 42 /9E . . . $T
3

1.9.?79

6..4 / '

7.28

6 (%;r8%i>< 1. 7e can now calculate the indi!idual film heat-transfer coefficients , using the appropiate L and $T valu 5rashof)number equations. ,his is an iterative process that requires initial estimates for wall and grou temperatures , plus wall temperatures.

,#eff " e!t f#r va(#r at wall *!Vw) . #s an initial approximation , assume that the wall temperature is of the !apor and outside-air temperatures ) 52.87 # Then find the ?rashof number ) (Gr = (K) 2 (L - Lw)3 2 (TV - Tw) ,mploying *q. 15 , find the @usselt number and then the coefficient , 7.7181 ft . 58.77 .w = 25.77 ft then , (,- = 0,138 ,Gr0,3+ . (,.r0,175 - 0,55) 915.9 hVw = ,,- . ' / (L - Lw) 7.896
'T<=ft$)h)6#

2w =

1.9=.?12

,#eff " e!t f#r l qu d at the wall *!Lw) . Here , neither ,.r nor *,Gr ,.r+ falls within the range of the app correlations **q. 1+ , 18+ . Eet us try both , again using an a!erage for Tw ) 2w = (TL + TA) / 2 (Gr = (K).L3.(TL - Tw) 58.77 1.=8.?79 H&

<sing *q. 1+ 8<d 18 , we get two estimates for the heat-transfer coefficient k= 7.12 NPr = 585 hLw = (0, 15.' / Lw).(,Gr . ,.r)0,25 hLw = (0, 5.' / Lw0,75).(,Gr.,.r)0,25 To be conser!ati!e , we use the higher !alue ) hLw = 1.518
'T<=ft$)h)6#

7.97= 1.518

'T<=ft$)h)6# 'T<=ft$)h)6#

,#eff " e!t f#r va(#r at r##f *!Vr+ . 7e consider this a flat plate with ) ft Diameter 27.77 and use *q. 20 , again with ) 2w = 52.8 F 7.7181 (Gr = (K) . D3 (TV - Tw) 1.12.?12

hVr = (0,27.'/D).(,Gr.,.r)0,25

7.182

'T<=ft$)h)6#

,#eff " e!t f#r l qu d at ta!k b#tt#- *!Lb+ . #ssume that the ground temperature *TG+ is 5 F above and use an average of liquid and ground temperatures as a first appro7imation for the tan&)bottom tem TG = Tamb + 5 = 2w = (TL + TG)/2 = 09.99 59.87 H& >F

Then , figure the ?rashof number , and use *q. 11 to get the coefficient ) (Gr = (K).D3.(TL - Tw) = 1.41,F92 (Gr . ,.r = hLb = (0,1 .'/D).(,Gr.,.r)0,33 = ..4/,F95 1.178
'T<=ft$)h)6#

,#eff " e!t f#r #ut$ de a r at r##f *!$Ar+ . #ssume ) 2w7 = Tw since the roof is uninsulated , and get the coefficient for still air from *q. 11 ) (Gr = (K).D3.(T"7 - TA) = h=Ar = (0,1 .'/D).(,Gr.,.r)0,33 -.-.,F-. 7.689
'T<=ft$)h)6#

,#eff " e!t f#r #ut$ de a r at wall *!$Aw+ . #ssume that the temperature drop across the film is one-fo drop from the inside fluid to the outside air *a!eraged for the wet and dry walls+ , and use *q. 15 to fin coefficient ) AT = 0,25.((TG -T8?b) + (Tw7 -T8?b)) = (Gr = (K).L .AT =
3

0./31 5.99,F-. -,15..5/ 7.871

H&

(,- = 0,138.,Gr

0,3+

.(,.r

0,175

- 0,55) =

h=Aw = ,,-.' / L =

'T<=ft$)h)6#

,#!du"t #! "#eff " e!t$ f#r .r#u!d / -etal wall/ a!d !$ulat #! *!G , !& , and !(+ . These are straig from *q. 21-23 ) h& = '& / %& = 'T<=ft$)h)6# 657.77 h( = '( / %( = hG = 8 . 'G / B.D 7.225 7.172
'T<=ft$)h)6# 'T<=ft$)h)6#

(adiation coefficients for dry and "et side"all 1 and roof @!)d / !)w / !)r0 . #s for the outside-air film coefficients , assume that )

2w7 = TA + 0,25 (Tb-9' - TA) where , Tb-9' is the temperature of the li uid or !apor inside the tank , if the surface is insulated. &or the uninsulated roof , assume that ) 2w7 = TA + 0,5(TV - TA) Then , 2w7 = 2w7 = 2w7 =

=8.98 57.77 52.8

# # #

for the *insulated+ dry sidewall , for the wet sidewall , for the roof.

<sing *q. 2 , find the coefficient for each of the three cases ) h) = h)d = h)w = h)r = 9.909.9/9.9.2-4.33/ 2.1.999 2/3.150 299./3/ 299./3/ 299./3/ 7.988 7.988 7.965
'T<=ft$)h)6# 'T<=ft$)h)6# 'T<=ft$)h)6#

0losing in on res lts. Table III summari"es the heat-transfer coefficients just calculated , including the corrections for wind !$Ar are multiplied by 3,3 and 3,1 , respectively , based on data for 10-?3! wind in Fig. 2 . $ubstitutin indi!idual coefficients in *q. 11-1 , we obtain the U !alues listed in T8b9; (((. T/)3. %%%

,eat-transfer coefficients after first iteration


0oefficient hVw hLw hVr hLb hG hAAr hArC hAAw hAwC h& h( hF h) 7.871 1.685 657.77 7.225 11777 7.988 7.871 1.685 657.77 7.225 877 7.988 11777 7.965 657.77 7.689 2.7=9 &ry "all 7.8986 1.518 7.182 'et "all (oof

'> / 1ft+0 5

).%,%% Ad -,194 Qd *, .-+ 3,417.2 ;,;@1.0

).%17; !w -,194 Qw *, ..+ 5,396.9 T!"#r

).%++, !) /-1 Q) *, ./+ 683.8

Qi ETU/!r Q Q d + D w+ D ) + D E

I &or 10-?3! wind.

6 (%;r8%i>< 2. 7hat remains to be done J 7hen we began the calculations , we assumed that the outside-wall temp related to the bulk-fluid temperatures by ) 2w = TA + 0,5 . (Tb-9' - TA) 2w7 = TA + 0,25 .(Tb-9' - TA) for uninsulated surfaces for insulated surfaces

In order to calculate accurate coefficients for heat transfer , we must now obtain better estimates of th temperatures. This re uires an iterati!e procedure that can be programmed and run on a computer. &or dry wall , the rate of heat loss is gi!en by all three of the following ) qd = Ud.Ad.(TV - TA) , .*.1+ qd = !Vw . Ad . (TV - Tw) qd = (!)d + !Aw).Ad.(Tw7 - TA) $ol!ing *q. 25 8<d 27 for Tw7 yields ) 2w7 = (Ud/(!)d + !Aw)).(TV - TA) + TA $imilarly , sol!ing *q. 25 8<d 2+ for Tw yields ) 2w = TV - (Ud/!Vw).(TV - TA) , .*.5+ , .*.4+ , .*.2+ , .*.3+

<sing the same approach , now calculate Tw and Tw7 for the wet wall , and Tw for the roof and bottom o

To find the correct wall temperatures , use the initial estimates of U and ! !alues in *q. 28 8<d 21 *an parallel e uations for the other surfaces+ to get new Tw and Tw7 !alues . Table IV shows these temper a second iteration . <sing these new temperatures , recompute ?rashof numbers , indi!idual heat-tran coefficients and o!erall coefficients , and then iterate again to get a new set of Tw and Tw7 !alues . h current and previous iteration3s temperature estimates are the same "within a specified tolerance) , th completed.

T/)3. %+ *econd iteration yields closer te!perat re esti!ates


Te!perat re Tw 1 !$ de0 / F Tw7 1#ut$ de0 / F %teration 2 1 2 1 &ry "all +?.% 52.87 @,.; =8.98 'et "all ,0., 58.77 @?., 57.77

*(%;r8%i>< 2) 8 #or wet wall , roof and bottom tan& , the rate of heat loss is given by all three of the following 8

&or w;% w899 , the rate of heat loss is given by all three of the following 8

qw = Uw.Aw.(TL - TA) qw = !Lw . Aw. (Tw - TA) qw = (!)" + !Aw).Aw.(Tw7 - TA)


$ol!ing Eq. +7 a!d +8 for Tw7 yields 8

, .*.1P+ , .*.2P+ , .*.3P+ , .*.4P+ , .*.5P+

2w7 = (Uw/(!)w + !Aw)).(TL - TA) + TA


$imilarly , sol!ing *q. 25L 8<d 2+L for Tw yields 8

2w = TL - (Uw/!Lw).(TL - TA)

&or r>># %8<' , the rate of heat loss is given by all two of the following 8

qr = Ur Ar (TV - TA) qr = !Vr Ar (Tw - TA)


$ol!ing *q. 25LL 8<d 2+LL for Tw yields 8

, .*.1PP+ , .*.2PP+ , .*.5PP+

2w = TV - (Ur/!Vr).(TV - TA)

&or b>%%>? %8<' , the rate of heat loss is given by all two of the following 8

qb = Ub Ab (TL - TG) qb = !Lb Ab (Tw - TG)


$ol!ing , . .1PPP and .2PPP for Tw yields 8

, .*.1PPP+ , .*.2PPP+ , .*.5PPP+

2w = TL- (Ub/!Lb).(TL - TG)

,#eff " e!t f#r va(#r at wall *!Vw) . #s an (%;r8%i>< 2 , the wall temperature is) Tw 5 56.1 H& 1.71.?12 Then find the ?rashof number ) (Gr = (K) 2 (L - Lw)3 2 (TV - Tw) ,mploying *q. 15 , find the @usselt number and then the coefficient , 7.7181 ft . 58.77 .w = 25.77 ft then , (,- = 0,138 ,Gr0,3+ . (,.r0,175 - 0,55) 928.5 hVw = ,,- . ' / (L - Lw) 7.586
'T<=ft$)h)6#

,#eff " e!t f#r l qu d at the wall *!Lw) . Here , neither ,.r nor *,Gr ,.r+ falls within the range of the app correlations (*q. 1+ , 18) . 2w = 82.8
3

H& =.51.?76

(Gr = (K).L .(TL - Tw)

<sing *q. 1+ 8<d 18 , we get two estimates for the heat-transfer coefficient k= 7.12

NPr =

585 7.599 1.775


'T<=ft$)h)6# 'T<=ft$)h)6#

hLw = (0, 15.' / Lw).(,Gr . ,.r)0,25 hLw = (0, 5.' / Lw0,75).(,Gr.,.r)0,25 To be conser!ati!e , we use the higher !alue ) hLw = 1.775 'T<=ft$)h)6#

,#eff " e!t f#r va(#r at r##f *!Vr+ . 7e consider this a flat plate with ) Diameter 27.77 ft and use *q. 20 , again with ) 2w = =8.98 H& 7.7181 (Gr = (K) . D3 (TV - Tw) hVr = (0,27.'/D).(,Gr.,.r)
0,25

2.12.?12 7.199
'T<=ft$)h)6#

0oefficient for li# id at tank $otto! @!Lb0 8=.9 H& Then , figure the ?rashof number , and use *q. 11 to get the coefficient ) (Gr = (K).D3.(TL - Tw) = 9.89.?78 (Gr . ,.r = hLb = (0,1 .'/D).(,Gr.,.r)
0,33

2w =

5.98.?78 = 7.799
'T<=ft$)h)6#

,#eff " e!t f#r #ut$ de a r at r##f *!$Ar+ . #ssume ) =8.98 H& since the roof is uninsulated , and get the coefficient for still air from *q. 11 ) (Gr = (K).D3.(T"7 - TA) = 1.16.?11 h=Ar = (0,1 .'/D).(,Gr.,.r)0,33 7.=11
'T<=ft$)h)6#

2w7 = Tw

,#eff " e!t f#r #ut$ de a r at wall *!$Aw+ . #ssume that the temperature drop across the film is one-fo drop from the inside fluid to the outside air *a!eraged for the wet and dry walls+ , and use *q. 15 to fin coefficient ) AT = 0,25.((TG -T8?b) + (Tw7 -T8?b)) = (Gr = (K).L .AT =
3

8.9= 1.22.?1= 11999.62 7.889

H&

(,- = 0,138.,Gr

0,3+

.(,.r

0,175

- 0,55) =

h=Aw = ,,-.' / L =

'T<=ft$)h)6#

,#!du"t #! "#eff " e!t$ f#r .r#u!d / -etal wall/ a!d !$ulat #! *!G , !& , and !(+ . These are straig from *q. 21-23 ) h& = '& / %& = 'T<=ft$)h)6# 657.77 h( = '( / %( = 7.225
'T<=ft$)h)6#

hG = 8 . 'G / B.D

7.172

'T<=ft$)h)6#

(adiation coefficients for dry and "et side"all 1 and roof @!)d , !)w , !)r0 2w7 = 2w7 = 2w7 = h) = h)d = h)w = h)r = 9.-20 9.-90 9.-53 290.519 293.290 290.-20 299./3/ 299./3/ 299./3/ 7.959 7.987 7.958
'T<=ft$)h)6# 'T<=ft$)h)6# 'T<=ft$)h)6#

=8.9 =6.8

# #

for the *insulated+ dry sidewall , for the wet sidewall ,

=8.98 # for the roof. <sing *q. 2 , find the coefficient for each of the three cases )

Table V lists the indi!idual and o!erall coefficients after the second iteration . .lthough it is clear that a iterations are needed , let us accept these values as sufficiently accurate for the present purpose . Th obtain the total heat-transfer rate *D+ by using the U !alues in *q. 1-5 and summing. T/)3. +

,eat-transfer coefficients after second iteration


0oefficient hVw hLw hVr hLb hG hAAr hArC hAAw hAwC h& h( hF h) '> &ry "all 7.586 7.889 1.858 657.77 7.225 11777 7.959 *, .--+ ).%+0) 'et "all 1.775 7.889 1.858 657.77 7.225 877 7.987 *, .-.+ ).%1%) (oof 7.199 7.=11 7.968 657.77 11777 7.958 *, .-0+ ).%?%;

I &or 10-?3! wind. .ffect of "ind velocity and 4T #! heat-tra!$fer rate 1F .. +0 +w !d 5 'f 5 4T 5 17 mph 18.9 H& hAw / !$Aw =

17

mph

11.1

H&

hAr / !$Ar =

Table VI shows the calculated heat-transfer rates through each boundary , and the total rate . @ote th and bottom of the tank account for only slight heat loss , despite being uninsulated.

T/)3. +% (ate of heat transfer d ring nit period


* rface U / 2T3&ft+h- Ar;8 / F ft+ $2 1 F

&ry "all 7.1527 11878 18.77 'et "all 7.1917 11878 27.77 (oof 7.1619 =18 18.77 )otto! 7.75=9 =15 18 2O2!. @,?+, @ote ) Total for 12-! 3;ri>d is 112.013 ETU. This , of course , is for the unit period of time , when wind speed is 10 ?3! , the tank is half full , and t F .

Marzo 1.982

er precisely

sfer fro! a vertical storage ade# ate.

ham and Richardson , Inc.

and tankfarms. Though many shed [1 , 3 , 5 , 7 , 8 , 10] , the simplifying ansfer rates. This is of concern on accurate estimates of heat transfer urate increases as energy costs

rtical-cylindrical storage tank seated on on, li uid le!el , ambient temperature eeen air and ground. # partially worked n a computer.

cal and spherical tanks are used for indrical , with flat bottoms and conical guration , but the procedure applies to

an the ambient air , and that we are od may , of course , be applied to either

rred across the tank boundaries ) wet ers to the portion of the wall submerged e !apor space , abo!e the li uid

nk , in the form of flat "pancakes".

el may be expected to be lower than the the li uid and !apor space may be gly in our procedure. Use of differnt approach , which assumes the same

m each of the four categories of Thus ) , ., .. , ./ , .0 , .1

e the !arious temperatures for typical

, .2 , .3 , .4 , .5

oefficients Ud , Uw , Ub and Ur , for the tely be calculated from the indi!idual correlations.

at-transfer coefficient for the dry r film , fouling , metal wall , insulation *if

s well as temperature gradient. 8ata on nd Boyen [2] . ith a little bit of ctor !""#) in #ig. $ .

=.= =.1 , .*-9+

!erall dry-sidewall coefficient at !arious

, .*--+

ces are )

, .*-.+

, .*-/+

, .*-0+

generally the case in temperature oefficients needed to obtain the o!erall

!ection from !ertical walls , con!ection hin each category , correlations are

and the outside wall *still air+ . &or ids and !apors )

, .*-3+

o the roof and inside-bottom surfaces of

m , Ad for dry wall , Aw for

0-3999999
$

=ft h-# % !Aw for air and !$Ar for still air , Aw near the bottom , !Vw for of.

at the walls , !Fb for r the roof.

# h-#

h-# % !)b for bottom , !)d

und , '( for insulation , '&

bottom , qd for dry wall ,

& liquid , TV for vapor , TG wall.

etal. Ub for bottom , Ud for dry

#)*

ation coefficients are deri!ed from the

heat conduction from a semi-infinite

and li uid at the wall, and the li uid at he type of fluid and other factors such ast , indicating that the greatest fouling

walls and roof depends on the

o or from the tank.

ports a high-!iscosity specialty oil om , conical-roof tank rented from BCD s e uipped with pancake-type steamreserve its fluidity.

e edge of the roof+ % roof incline is %&'fiberglass , on the sidewall only.

indi!idual tanks, and proposes to es, using the co!entional tables [*] , and /hemical decided to investigate how is article.

nts from the <.$. , arri!e at Aort (ity ustomers are made in 8,000-g89 !el o!er a %)-da* period is known from

7ithin the primary cycle of 3+5 d8=7 , y is re uired only during the winter

uch harder to predict. Howe!er, we can d calculate the heat loss for this unit

tion of wind speeds at the site.

he heat losses calculated for each unit propiate+. This example demonstrates mbient temperature of 35 F , a wind given in Table II . 4ote that the liquid

on the sidewall only+

Ta$le %%
+apor 4
7.78 7.28 7.779 7.7181 7.772

!ostly air.

nsfer coefficient e uations , and remain heir !alues . Thus , for the li uid phase )

e appropiate L and $T values in the estimates for wall and ground

hat the wall temperature is the a!erage

within the range of the applicable

erature *TG+ is 5 F above ambient , ion for the tan&)bottom temperature 8

op across the film is one-fourth of the alls+ , and use *q. 15 to find the

, and !(+ . These are straightforward

0 . #s for the outside-air film

s insulated.

g the corrections for wind - !$Aw and wind in Fig. 2 . $ubstituting these ((.

iteration
)otto!

1.178 7.172

657.77 877

).);@0 !E /-0 QE *, .0+ 439.3

d that the outside-wall temperatues were

btain better estimates of these wall d and run on a computer.

Tw for the roof and bottom of the tank.

alues in *q. 28 8<d 21 *and in the ble IV shows these temperatures after mbers , indi!idual heat-transfer of Tw and Tw7 !alues . hen the n a specified tolerance) , the iteration is

re esti!ates
(oof @,.17 52.8 @,.17 52.8
all three of the following 8

)otto! ,@.1 59.87 -

ure is)

within the range of the applicable

op across the film is one-fourth of the alls+ , and use *q. 15 to find the

, and !(+ . These are straightforward

n . .lthough it is clear that additional r the present purpose . Then we can summing.

nd iteration
)otto! 7.799 7.172 F 657.77 877 *, .-/+ ).)+@;

r rate 1F .. +0 =.=

=.1

and the total rate . @ote that the roof sulated.

period
q/ 2T3&h =1211.59 81186.82 966.75 279.79 ;,@+% , the tank is half full , and the air is 35

1E - 3e PFtrole - (/FF%-/;. .T ;G-%. 0,%M%HU. Pierre ' ithier Q8ITI6@$ T,(H@IA - -53..

Fig. 2.9.1 Typical overall heat transfer coefficients fro! flat steel s rfaces Figure 0.;.% pro!ides some typical o!erall heat transfer coefficients for heat transfer from bare steel flat surfaces to ambient air. If the bottom of the tank is not exposed to ambient air, but is positioned flat on the ground, it is usual to consider this component of the heat loss to be negligible, and it may safely be ignored. 9 &or .1 mm of insulation, the < !alue should be multiplied by a factor of 9.. 9 &or 19 mm of insulation, the < !alue should be multiplied by a factor of 9.-.

Ta$le 2.9.2 .ffect on heat transfer "ith air !ove!ent

2aGle 0.;.. shows multiplication factors which need to be applied to these !alues if an air !elocity is being taken into account. Howe!er, surface is well insulated, the air !elocity is not likely to increase the heat loss by more than -9O e!en in exposed conditions.

>elocities of less than - m=s can be considered as sheltered conditions, whilst 1 m=s may be thought of as a gentle bree"e *about / on th 'eaufort scale+, -9 m=s a fresh bree"e *'eaufort 1+, and -2 m=s a moderate gale *'eaufort 3+.

Ta$le 2.9.= <verall heat transfer coefficients for oil tanks &or bulk oil storage tanks, the o!erall heat transfer coefficients uoted in 2aGle 0.;.@ may be used.

Fig. 2.9.1 Ta$le 2.9.2 Ta$le 2.9.=

%nternational site for *pira2 *a

'ind +elocity .ffects on e2posed1 )are and %ns lated * rfaces.

rfaces

teel flat surfaces to ambient air. sual to consider this component

!ent

y is being taken into account. Howe!er, if the e!en in exposed conditions.

ught of as a gentle bree"e *about / on the

il tanks

ed.

nal site for *pira2 *arco

- (entipoise G .,0-5/1 lb=ft.h