I Flow Patterns for Predicting Shell-Side Hgat Transfer RAJESHWAR K. GUPTA' and DONALD 1. KATZ
Universityof Michigan, Ann Arbor, M i .
, l
!. 4,
4.
Coefficients for Baffled Shell-and-Tube &changers The shell side of an exchanger may be divided into three separate zones: longitudinal flow, cross flow, and an eddy zone. Correlation of eddy zone coefficients permits estimation of exchanger performane by considering each zone'seporately
T,
nle&mMm '
of fluid turbulence
and heat transfer on the shell side of bat04 shell-and-tube heat ewhaqen is mefe OOmplatbanisreeogniEedin curmt methods of eondating the data for these ewhaqen. In thib investigation, 9ow patterns on the'shell side were sbld*d in d e r to divide the shell side into rcgiop. of different 0ow characteristics for predicting heat transfer. Data are reported on shell-side flow patterns, heat kander eharactens ' tics, and pressure drops for an all-glass heat exchanger with six different tube bundles. Only
thc baf&
CUB and
the b a e spacings
were varied. Thc fiow pattern studies reveal that the flow on the shell side caa be divided into zonq f thm different flow char. . olongitudinal ac0ow (0, true rnBow (c), and eddy or dead zones (a). Themechanwms ' of flow and heat fu in the eddy wnes are d b t fmm thacinthcoabatwownes. Ce5cienta of heat tr& for the eddy wne are '&?sUlt CalteX oil Refining (India), Ltd.,V i i a p a m a m , India
*
computed fmm the experimental data and equations from the literature for longitudinal and true -flow. Many additional data will be q u i d to wmplete satisfactory wm$ations on this basis. C r d o w heat eanaler measurementa should be made on mdgeshapcd unimaswellas~tangularmcdels. The pmredmpdataarecarelatedaswell.
Experimental D& A single-paw glaa heat exchanger (Table I) was equipped with circulating pump and fhv-meaaufing devices for both the tube and sbfl d e . The b a e s were reconstructed to consist of a braasr u b k sandwich, which prevented leakage at both the tubes and the shell. A water solution of pmpykne\glywl (60 to 80%) was uscd as the hot fluid on the shell side of the exchanger. Wilson plot data were obtained fmm over-all heat tanafer codficienm for a series ofvelocities on the shell side. These shell side film cmllicients are plotted on Figure 1 as a function of the mean Reynolds num-
ber befwcen the window, w, and in ams flow, c e. The drop on the shell side multiplied by the distance
+
betmcnbaWesdividedbythecxhanger lengtb is d t e d aaa fundiw of ma: rate on Fire 2. Flow patterns wax obmved from the movement of foreign particles. Theac patterns were used to divide the heat exchange into thrn zones as portrayed on
F i 3 . Acha~geinflowratechan&ed the fluid motion within the eddy zone but not the p'oportion of the cxehanger mnface in the eddy zone aigni6cantl) Figure 4 gives the pruportiy of the edd wne p l u ~true cmas 0ow zone €or thres bafac CUB. This figure pamila quantitative division of the teat achanger into the thm wnw of heat trader. C o d a t i o n of Heat Transfer Data
The heat transfer &dent on th ahell side, h., is considered to be a cno posite qf that for the thm zones:
4 Figure 1. Experimental shell-side coefficients for six
bundlss
I
lure2. Cwr n of pres
' i .
I.
VOL. 49, NO. 6
JUNE I957
999