AD-A096 350 NAVAL POSTGRADUATESCHOOL MONTEREY CA F/6 13/1

1HEAT EXCHANGER OPTIMIZATION. CU)

UNCLASSIFIED SEMGCP EDRC

2~h~~hhE
mohhhEEEmhEEIm
EEEEohhhhhEmhE
EohEEEEohEEEEE
o NAVAL POSTGRADUATE SCHOOL
Monterey, California

Yh1
R~ 7 1981

THESIS A

/ HEAT EXCHANGER OPTIMIZATION

by

Conrad P. Hedderich 7

. Septembew 1980

Thesis Co-Advisors: M. Kelleher

C- Vanderolaats
Approved for public release;distribution unlimited.

S60

-,- t! ':i1.6 076 "

 UNC LASS IF TIED
SECURITY CLASSIFICATION OF THIS PAGE (W9In Data Entered)

REPORT DOCUMENTATION PAGE .,,?ADC ,is'rNnORI

1. REPORT NUMBER OVT ACCCSSION NO 3. RECIPIENT'S CATALOG NUM6ER

4. TITLE (and Sublitio) S. TYPE Or REPORT a PERIOD COVERED

Heat Exchanger Optimization Master's Thesis;

September 1980
I~. PERFORMING ONG. REPORT PON9914lI

7. AUTNOR(e) S. CON TRACT OR GRANT POIUMIER(e)

A Conrad P. Hedderich
9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT, TASK
AREA 6 WORK UNIT NUMERS
Naval Postgraduate School
Monterey, California 93940
II. CONTROLLING OFFICE NAME AND AORESS 12. REPORT DATE

Naval Postgraduate School '3.1q80

Monterey, California 93940 NUMER
18. A
188
14. MONITORING AGENCY NAME AOORESS(II dilleren tlrai Cntroillind Offie) IS. SECURITY CLASS. (of tile rffo)

Unclassified
IS. OECL ASSI FICATIONt OOWNGRAOING
SCHEDULE

IS. DISTRIBUTION STATEMENT (of lhe Repor)

Approved for public release; distribution unlimited.

17. DISTRIBUTION STATEMENT (of the abstract entered In Stock 20, It differenttm Report)

II. SUPPLEMENTARY NOTES

IS. KEY WOROS (Conthn on evwe*, #$do II 010l000 4 Ideni'd &bb ech mmber)

Heat Exchanger Design Using Numerical Optimization

HEDSUP Design
Heat Exchanger Optimization
20. AISTRACT (Cenelmee OR
en , ree e if nd e6p MINItdeu9r by lock 4n ,Mbe)

A computer code was developed for the analysis of air-cooled

heat exchangers and was coupled with a constrained function mini-
mization code to produce an automated air-colled heat exchanger
design and optimization program with many new capabilities.
A general iteration free approximation method was used for
the analysis which calculates the mean overall heat transfer
coefficient and the overall pressure drop for many flow
DO AN7 1473 EDITION OF I NOV 61 isOISOLETU

2/11 0102"014"6011 ICURITY CLASSIFICATIOR OF TMIS PAGE (When Dots SIerd)

.. I
. . .. ..I11 ..Ir
 #20 - ABSTRACT - (CONTINUED)

arrangements, taking into account the variation of the heat

transfer coefficients and the pressure drop with temperature
and/or length of flow path.
The code is not limited to surfaces found in the litera-
ture, but will accommodate any triangular pitch bank of finned
tubes in multiple-pass configurations.

DD ar%
3 1 UNCLASSIFIED
S14~ o0101-14-6601 2 SI*R9W.,, PIATO 4Pe*..
?"Is P&awe Does 10
 Approved for public release; distribution unlimited.

HEAT EXCHANGER OPTIMIZATION

by

Conrad P. Hedderich
Lieutenant, United States Navy
B.S.M.E., United States Naval Academy, 1973

Submitted in partial fulfillment of the

requirements for the degree of

MASTER OF SCIENCE IN MECHANICAL ENGINEERING

from the

NAVAL POSTGRADUATE SCHOOL

September 1980

Author (,A ,PQ

Approved by: 77-// {. , / °
Thesis Advisor

Co-Advi sor

Chairman, Depa ent of 1'6canical Engineering

Dean of Science and Engineering

3
 ABSTRACT

A computer code was developed for the analysis of air-

cooled heat exchangers and was coupled with a constrained

function minimization code to produce an automated air-cooled

heat exchanger design and optimization program with many

new capabilities.
A general iteration free approximation method was used

for the analysis which calculates the mean overall heat

transfer coefficient and the overall pressure drop for many

flow arrangements, taking into account the variation of the

heat transfer coefficients and the pressure drop with tempera-

ture and/or length of flow path.

The code is not limited to surfaces found in the litera-

ture, but will accommodate any triangular pitch bank of finned

tubes in multiple-pass configurations.

4
 TABLE OF CONTENTS

I. INTRODUCTION------------------------------------------1

A. BACKGROUND---------------------------------------- 12

B. REVIEW-------------------------------------------- 13

C. METHODOLOGY--------------------------------------- 17

D. OBJECTIVES---------------------------------------- 18

I. NUMERICAL OPTIMIZATION-------------------------------- 19

A. BACKGROUND---------------------------------------- 19

B. CONSTRAINED FUNCTION MINI!'IZATION (CONMIN)--------20

C. CONTROL PROGRAM FOR ENGINEERING

SYNTHESIS (COPES)--------------------------------- 27

II.HEAT EXCHANGER ANALYSIS------------------------------- 30

A. INTRODUCTION-------------------------------------- 30

B. PROBLEM FORMULATION------------------------------- 30

C. PERFORMANCE CALCULATION PROCEDURE----------------

1. Log Mean Temperature Difference------------- 3

2. Determination of Reference Temperatures --- 36

3. Correction of Reference Temperatures

for Given Arrangement------------------------- 38

4. Uncorrected Tubeside Heat Transfer

Coefficient----------------------------------- 40
5. Calculation of Wall and Associated
Resistances----------------------------------- 42
6. Calculation of Airside Heat Transfer
Coefficients---------------------------------- 42

7. Calculation of Fin and Surface

Efficiencies--------------------------------- 4

8. Correction of the Tubeside Heat Transfer 4

Coefficient-----------------------------------4

5
 9. Calculation of Air and Tubeside
Pressure Drops----------------------------- 46

10. Objective and Constraint Functions----------48

IV. RESULTS--------------------------------------------- 52

A. BACKGROUND------------------------------------- 52
B. CASE STUDIES----------------------------------- 54

1. Case one----------------------------------- 54

2. Case Two----------------------------------- 60
V. CONCLUSIONS---------------------------------------- 65

VI. RECOMMENDATIONS------------------------------------ 68

VII. FIGURES--------------------------------------------- 70

APPENDIX A: USER'S MANUAL------------------------------- 89

APPENDIX B: SAMPLE USER'S INPUT------------------------ 115

APPENDIX C: SAMPLE OUTPUT FROM COPES-------------------- 126

APPENDIX D: ANALIZ PROGRAM LISTING---------------------- 139

LIST OF REFERENCES--------------------------------------- 184

INITIAL DISTRIBUTION LIST------------------------------- 188

6
 LIST OF FIGURES

Figure

1. Flowchart for ANALIZ ---------------------------- 70

2. Numerical Optimization Techniques --------------- 71

3. Usable-Feasible Direction ----------------------- 72

4. One Dimensional Search -------------------------- 73

5. Two-Variable Design Space with Initial Design --- 74

6. Two-Variable Design Space with First Iteration 75

7. Two-Variable Design Space with Second Iteration 76

8. Two-Variable Design Space with Third Iteration 77

9. Two-Variable Design Space with Fourth Iteration 78

10. General Configuration of Air-Cooled Heat

Exchanger --------------------------------------- 79
11. Thermal Conductivity of Dry Air at
Atmospheric Pressure ---------------------------- 80
12. Thermal Conductivity of Saturated Water --------- 81

13. Viscosity of Dry Air at Atmospheric Pressure 82

14. Viscosity of Saturated Water -------------------- 83

15. 4 Row, 2 Pass Arrangement ----------------------- 84

16. DRATIO Constraint ------------------------------- 85

17. THETA Constraint -------------------------------- 85

18. Design Optimum - Case Study One ----------------- 86

19. Design Optimum - Case Study Two ----------------- 87

20. Two-Variable Function Space --------------------- 88

7
 NOMENCLATU RE

English Letter Symbols

.2
A - total heat transfer area, in

cp - specific heat, BTU/lbm-°F

- heat capacity rate, BTU/hr-F = m p

D - diameter, in.

f - friction factor

F - LMTD correction factor

2
ft/sec
gc - acceleration of gravity, 32.2

H - corrected heat transfer coefficient, BTU/hr-ft -F

h - bank height, in.

J - Colburn factor

k - thermal conductivity, BTU/hr-ft-°F

L - length, in.

- fin height, in.

m - mass flow rate, ibm/hr

m - /2H /k t , ft
o f
n - number of (used with appropriate subscript)

N - number of tubes

p - pressure, psi

P - pitch, in.

Pr - Prandtl number

- heat transfer rate, Btu/hr

r - radius, in.

R - heat transfer resistance, hr-ft2-F/BTU

8
Re - Reynolds number

s - distance between adjacent fins, in.

S - fin spacing center-to-center, in.

t - fin thickness, in.

T - true temperature, deg.

U - overall heat transfer coefficient, BTU/hr-ft 2-F

w - bank width, in.

Greek Letter Symbols

tT - temperature difference, deg.

- surface efficiency

- fin efficiency

- viscosity, ibm/ft-hr

0 - density, ibm/ft 3

- temperature correction

Subscripts

a - air

c - cold

f - fin

ff - free face

h - hot

i - inside

j - reference number, I or II

- limiting

L - longitudinal

m - mean

o - outside

9
p -pass

r -rows

t -transverse

T -total

w -water

x -cross-sectional (flow)
1 -entering

2 -leaving

-O ambient

Superscripts
-uncorrected

* -estimated

i -initial

1) 10

7".
 ACKNOWLEDGMENTS

The author wishes to express his sincere appreciation to

Professor Matthew Kelleher for his advice and guidance

throughout this project. Special thanks go to Professor

Gary Vanderplaats, whose last minute addition to COPES/CONMIN

rescued the design program.

The author wishes to thank all of the W.R. Church Computer

Center and the Dudley Knox Library for their timely assistance.
Final acknowledgments go to my wonderful family: my

wife, Linda, and daughter, Rebecca, whose understanding and

help throughout the long project made the effort bearable.

11
 I. INTRODUCTION

A. BACKGROUND

The cooling of fluids by passing ambient air over extended

tube surfaces is a relatively recent development in heat

exchangers. Its application has come about cautiously, due

to the usual reluctance to change from well established and

well documented methods, i.e. the shell-and-tube heat exchanger.

However, concern for the environment and economic pres-

sures have necessitated the use of air as a coolant. Smith

(l] has listed some typical advantages of direct cooling with

air as compared to cooling with water in a shell-and-tube

exchanger:

a. Eliminates the problem of temperature rise in, and

pollution of, water resources.
b. Enables plant location to be independent of a water
supply.

c. Eliminates the necessity of much coolant piping.

d. Reduces heat exchanger maintenance costs by eliminating

the need of descaling water-side surfaces. The
mechanical drives will operate in a noncorrosive
atmosphere.

e. Eliminates water treatment.

f. Limiting coolant temperatures is unnecessary.

g. Enables installation of exchangers at elevations above

other operating equipment at no penalty, thus reducing
ground area requirements.
With air colling becoming more and more competitive with
water; even when water supplies are plentiful, an automated

air-cooled heat exchanger design package (which could be used

12

A6
for trade-off studies, first cut analysis, and conceptual
design), would be of great use.

The design of an air-cooled cross flow heat exchanger is

a complex task requiring the examination and optimization of

a wide variety of heat transfer surfaces. Studies have shown

that a poor choice of either the heat transfer surfaces or

design parameters can more than double the costs charageable

to a heat exchanger [2].

For the optimized design of heat exchangers with the

computer; reliable, but fast, calculation methods for the

mean overall heat transfer coefficient and the overall

pressure drops are needed for the following reasons:

a. Conventional simple methods using mean values of

temperatures as reference temperatures can lead to
undesirable errors [3].
b. Numerical stepwise integrations are prohibitively
time consuming.

B. REVIEW

A numberof heat exchanger design methods have been proposed

to determine the optimum heat exchanger design. Bergles, et

al. [4], performed an evaluation of different objective func-

tions for compact heat exchangers with different heat trans-

fer surfaces, but the same specifications. The method did not

include any actual optimization techniques, but results did

show that a great improvement in heat exchanger performance

can be made by proper selection of design parameters.

13
 Q 1 076

The riethod of Fax and Mills [5), used Lagrange multi-

pliers to optimize a heat exchanger design under specified

constraints. This technique required that the objective

function and constraints be expressed explicitly and be differ-

entiable throughout the range of interest. The total num-

ber of constraints had to be less than the total number of

variables, and all constraints had to be equality constraints.

The method was obviously restricted to a very limited number

of problems.
For a unit section of an air-cooled exchanger of standard

length and width equipped with specified fans, both Schoonman

[6], and Joyce [7], used factorial searches of fin spacing,

number of rows and air rate to maximize the ratio of heat

transfer to cost. Because the number of sections must be

rounded to an integer, this method only appears suitable for

large exchangers.

Nakayama [3], used a similar approach requiring the plotting

of heat transfer coefficients and pressure drops which pre-

sumably could be programmed into a factorial search.

Kern [8] derived an analytic expression for the annual

cost as a function of air rate and number of rows. (This was

done by using P constant tubeside heat transfer coefficient

and an arithmetic mean temperature difference.) The optimum

was found by setting partial derivatives equal to zero.

Alternatively, if the airside is assumed to control, the

optimum allocation of the components of total cost might be

14

iI
found through geometric programming, as illustrated by Auriel

and Wilde [9]. Oshwald and Kochenberger [101, also presented

a geometric programming method for heat exchanger optimization

and used it to select heat exchanger fluids considering power

requirements, cost, tube diameters, velocities, temperature,

and other physical properties.

After discussing factorial, univariate, and random search

methods for an optimum design of a shell-and-tube exchanger,

Briggs and Evans [11, 121, discuss a "logical search method";

or what Peters and Nicole [13], call "heuristics". With this

method, an engineer makes use of selecting design variables

close to the optimum to obtain an optimum design. The heuris-

tic method is less scientific, and is useful only when computer

time and storage are at a premium. Due to the large number

of discrete variables encountered in air-cooled heat exchanger

design, Peters and Nicole [13], chose to base their cost-

optimizing design programs on heuristic algorithms (starting

close to optimum), specific to the equipment under considera-

tion.

Mott, et al. [21, discuss a computerized procedure for

designing a minimum cost heat exchanger. The method minimizes

a cost index expressed as a function of fluid pumping power.

The algorithm imposed no constraints.

To this point, no mention has been made of applying the

concepts and techniques of nonlinear programming to optimizing

the design of heat exchangers. However, Palen, et al. [14],

in 1974 proposed using the Complex Method [15], for the heat

15
 exchanger optimization problem. They found a minimum cost

shell-and-tube exchanger by varying six geometrical parameters.

The Complex Method requires several feasible starting designs

before optimization can be performed.

Johnson, et al. [16], coupled an existing she.l-and-tube

condenser design code with a constrained function minimization

code to produce an automated marine condenser design program

of vastly different complexity.

The most complete work to date has been accomplished by

Afimiwala [17]. He has applied various nonlinear programming

methods of optimization to the heat exchanger design problem;

including an experimental interactive graphical approach and

exterior penalty function techniques. The gradient based

search methods of Davidson - Fletcher - Powell and conjugate

gradient were used for the resulting unconstrained minimiza-

tions. The exterior penalty method is extremely useful, since
an initial solution satisfying the constraints is not required.

The gradient based search methods are efficient when considering

computer time.

Finally, Fontein and Wassink [18] utilized the complex

method of Nelder and Mead [19], and a steepest descent method

(20], for optimizing a shell-and-tube exchanger.

*It can be seen that although there are many methods that

have been presented for heat exchanger optimization, each of

the methods has its own limitations; none is completely general.

Of all the design procedures cited above (those of which are

applicable to cross flow air-cooled heat exchangers), all are

16
limited to the 120 individual surfaces found in the open

literature [21] for the calculation of the air-side heat

transfer coefficient and friction factor. Therefore, the

designer is faced with choosing an optimum surface from a

number of individual optimal designs calculated from one of

the above methods. In addition, the above methods treat the

overall heat transfer coefficient as a constant, or they become

involved with time-consuming numerical stepwise integrations

in an attempt to account for the varying heat transfer coeffi-

cients.

This paper tries to bridge this gap by presenting an

optimization routine that: selects an optimal surface, takes

into account the varying heat transfer coefficients and fric-

tion factors across the exchanger, performs each analysis in

an iterative-free manner, and may start with an infeasible

design.

C. METHODOLOGY
With the Control Program for Engineering Synthesis and

Constrained Function Minimization (COPES/CONMIN) optimizing

scheme, a nonlinear optimization program is available that is

capable of optimizing a wide class of engineering problems

[22,231. Therefore, for the heat exchanger design problem,

it was necessary to develop a subroutine, which given a start-

ing design, would analyze an air-cooled heat exchanger, and

which would be suitable for coupling with the optimizer,

COPES/CONMIN.

17

.... .
 Figure 1 illustrates the procedure by which the heat

exchanger was analyzed. Initial input consisted of a com-

plete listing of all design parameters, whether known or

estimated. Those that were estimates, i.e. unknowns, were

later passed to the optimizer as design variables.

The analysis scheme and the optimizer will be discussed

in much greater detail in the following chapters.

D. OBJECTIVES

The objectives of this thesis are two-fold.

The first objective was to develop a computer code,

hereafter referred to as ANALIZ, which would analyze an

air-cooled heat exchanger given any initial design. The

analysis scheme was to: be iterative-free, take into account
the variation of the heat transfer coefficients and the pres-

sure drop with temperature and/or length of flow path, and

finally, be written in such a manner that it could be coupled

with an existing optimizer COPES/CONMIN.

The second objective was to actually couple ANALIZ with

a numerical optimization program. This would produce a

detailed design program which would have the capability

to determine an optimum surface, while optimizing the objec-

tive function such as size, weight, cost, etc.

18
 II. NUMERICAL OPTIMIZATION

A. BACKGROUND

Almost all design problems require either the maximiza-

tion or the minimization of some parameter or function.

This parameter shall be called the design's objective function

[241. For example, the problem may call for a heat exchanger

with a minimum volume. The expression for volume would be

the design objective function. For the design to be accepta-

ble, it must satisfy certain design constraints. For exam-

ple, an air heater must be designed so that it will fit into

a given space. Therefore, the engineer must set design con-
straints on the maximum size of the exchanger.

If the objective function could be easily formulated

analytically, the maxima or minima could be found by using

the methods of differential calculus. However, the limita-

tions of this method are obvious.

Another numerical method that would be satisfactory for

small scale problems would be an iterative solution technique.

A computer program could be written containing a series of

nested iteration loops that would vary the design parameters

and solve the problem for a variety of values for each of the

parameters. For other than small, easily formulated problems,

the cost in central processor (CPU), time would be prohibitive.

Over the last twenty years, many numerical optimization

techniques have been developed specifically for computer

19
utilization. These techniques usually do not require a

specific algebraic equation, but rather any computer algorithm

to which design variables can be input and from which the

objective function and design constraint values can be

determined is acceptable. For this reason, nonlinear pro-

gramming methods were chosen for the air-cooled heat exchanger

design. Some of these techniques were summarized by Shah et

al. (25] in figure 2.

1. One-Dimensional Search Methods. Two of the most common
of these search methods are the golden section [261
and quadratic interpolation [27]. The former iso-
lates the minimum in regions of successively decreasing
size, the latter performs a series of iterations
approximating the objective function as a quadratic.

2. Multidimensional Unconstrained Search Methods. These

unconstrained searches can be performed by a sequence
of one-dimensional minimizations in the proper direc-
tions.
3. Multidimensional Constrained Search Methods. A common
method here for enforcing the constraints in an optimi-
zation scheme are based on the sequential penalty func-
tion method. These techniques convert the constrained
optimization problem into a sequence of unconstrained
problems. This is accomplished by applying either an
exterior or interior penalty to the objective function.
The Complex Method [15], locates the optimum based on
an intuitive approach in a n-dimensional space defined
by the independent design variables. The method of
feasible directions is used primarily for inequality
constraints.

An optimization program based upon the Augmented Lagrangian

Multiplier Method and the method of feasible directions was

chosen for this research project.

B. CONSTRAINED FUNCTION MINIMIZATION (CONMIN)

Vanderplaats [22], developed an optimization program,

CONMIN, based on the method of feasible directions which is

20
capable of optimizing a wide variety of engineering problems.

CONMIN is a FORTRAN program, in subprogram form, that opti-

mizes a function subject to a set of inequality constraints.

The following definitions will be useful in the following

discussion:
1. Design Variables - those parameters which the optimiza-
tion program can change in order to improve the design.
2. Design Constraints - those parameters which must not
exceed given bounds for the design to be acceptable.
3. Objective Function - the parameter which is going to
be minimized or maximized.

The general nonlinear inequality constrained optimization

problem can be written mathematically as follows [28]:

Minimize F(Y) (1)

Subject to:

gj(X) < 0 j - I,NCON (2)

1 '<
-- X.1 <-- X1 k = ,NDV (3)

where

x1

= 2 (4)

XNDV

The vector X is the vector of design variables, with NDV

equal to the number of design variables. The objective function,

21
F(X), given by eq. (1), as well as the constraint functions

given by eq. (2), may be linear or nonlinear functions of

the design variables. They shall also be explicit or implicit

functions of R, but must have continuous first derivatives.

NCON is the number of constraints. X1 and X ui

1
are the lower
and upper bounds or side constraints placed on the design

variables. Side constraints could be included in eq. (2),

but are treated separately for efficiency. Equality con-

straints are not dealt with by CONMIN, but will be treated

separately by a multiplier method.

CONMIN requires that an initial vector of design variables,

X, which may or may not yield a feasible design, be specified.

The design process continues iteratively as:

xq+l x-q + c* S (5)

q
where _S is a vector search direction, a* is a scalar quantity

which determines the amount of change in X and q is the itera-

tion number. At iteration q a direction Sq must be found

which will reduce the objective (usable direction for mini-

mization), without violating any constraints (feasible direction,,

see Figure 3 q
[30]. Once _S is determined, eq. (5) becomes a

one-dimensional search problem in whcih s* must be found

such thatgF(X) is at a minimum (see Figure 4 [29]), a new

constraint is encountered, or a currently active constraint

(gj(X) = 0) is encountered again.

The design problem at iteration q+l becomes one of

finding a usable-feasible direction, Sq , and a move parameter

22
a*. This process is illustrated geometrically by Johnson

[30]. Consider a condenser problem with just two design

variables, X 1 and X 2 , where

X= condenser tube outside diameter,

X= tube pitch to diameter ratio.

Let the objective function be condenser volume, VOL (X).

Assume that the tube bundle diameter must be greater than a

given value, BDmin, and that the cooling water pumping power

must be less than a given value HP ma x . Figure 5 illustrates

the design problem geometrically.

It should be reiterated here, that while Johnson's exam-

ple starts with a feasible initial design, ® , this is not

a requirement and CONMIN is capable of optimizing given an

infeasible initial design. This is obviously a great benefit.

The optimization process begins by calculating the gradient

of the objective function by using finite difference. Each

design variable is perturbed by .01 in a single forward step.
The gradient of the objective function, 7F, shown in

Figure 6 is simply the vector of the first partial deriva-

tives with respect to the design variables; that is:

-F AVOL

VF(X) V VOL = 1 (6)

9F ]AVOL
DX 2 AX2

23
Therefore, because no constraints are active or violated at
S.VF defines the direction of steepest ascent. Because

it is desired to minimize F, the greatest improvement can be

made by moving in the negative gradient direction so that

= - F - 7VOL . (7)

With the value of S now determined, a search is per-

formed from until the minimum F is found at on Figure

6. This is accomplished by taking several values of -Xq + l in

eq. (5) and interpolating for the a* which will give the mini-

mum value of F.
The second design iteration is begun at G by again

perturbing X to find VF. Instead of moving in the direction

of steepest descent, a conjugate direction, developed by

Fletcher and Reeves (311, is chosen with this method, S is

calculated as follows:

- = - F(X) q + 17F(Y) q12 q (8)

see Figure 7.

The Fletcher-Reeves method is used in order to speed

convergence. With the new conjugate direction, a search is

performed in this direction until a constraint is encountered.

This occurs at ( of Figure 8 on the pumping power constraint.

At ( , with the HP constraint active, not only is VF

found, but the gradient of the active constraint is also

computed, again using finite difference. The requirements on

24
 the new search direction are now twofold; it must reduce the

objective function and, at the same time, not violate the

active constraint. This is solved by using the method of

feasible directions developed by Zoutendijk [32], and imple-

mented by Vanderplaats and Moses [33].

The problem of finding the new S can be stated as [34]:

Maximize a subject to the constraints

VF(X) S+ 3 < 0 (9)

Vg.(X)-S + j. < 0 j = I,NAC (10)

s _ 1 (ii)

where -gj(X) = - -HP and NAC is the number of active constraints

(in this case NAC = 1).

If equation (9) is satisfied and a is positive, the search

direction will reduce the objective function and is defined

as a usable. If equation (10) is satisfied and is positive,

S is a feasible direction, because in this direction, no con-

straints will be violated if only a small move is taken. See

Figure 3. J is defined as the push-off factor for the active

0.

constraint and causes the design to move away from the con-

straint. e. must be greater than or equal to zero in order

to maintain a feasible design. If the maximum value of 3 from

equations (9) through (11) is zero, then there is no direction

that will both reduce the objective function and also be feasi-

ble. Therefore, the current design is, at least, a local

25

b
minimum. In Johnson's example, a usable-feasible direction
exists and a cne-dimensional search leads to 0 in Figure 8
where the minimum bundle diameter, BDmin' constraint is met.
From 0 it should be noted here that CONMIN had informa-
tion regarding the linearity of the BDmin constraint and,
therefore, in (10), has set 9. = 0 to allow S to follow the
constraint as shown in Figure 9. The one-dimensional search
along this constraint is carried out until no further design
improvement is realized. This occurs at 0 "

This discussion of CONMIN would not be complete without

citing the program's limitations. NDV directly affects the
computational time required to reach the optimum. Since the
calculation of gradients required for each design variable at
the beginning of each design iteration is found by a finite
difference step, which requires a complete pass through the
analysis portion of the program, there is a subsequent in-
crease in CPU time as NDV increases. Also, due to the inter-
action between design variables, as NDV gets larger, convergence
slows during the optimization process. Vanderplaats [243,
recommends that for most problems of general interest a prac-
tical limit of NDV = 20 be imposed. NCON does not present
the same problem because gradient information is calculated
simultaneously with VF and then only if the constraint is
active or violated.
CONMIN offers no guarantees that a global minimum has been
reached. Therefore, to lend some assurance, the design is

26
started with several different initial vectors until the

same optimal design is reached.

Although CONMIN performs very well with inequality con-

straints, equality constraints such as:

hK ( ) = 0

cannot be dealt with directly, but must be treated separately,

using a different method, which will be discussed in the

following sections.

C. CONTROL PROGRAM FOR ENGINEERING SYNTHESIS (COPES)

Recall that CONMIN was written in subroutine form,

Vanderplaats [23], has developed a main program which greatly

enhances the use of CONMIN.

For this main program, COPES, the user must supply an

analysis subroutine titled ANALIZ. The subroutines CONMIN

and ANALIZ are then used by COPES to optimize the objective

function subject only to the inequality constraints.

ANALIZ must be organized into three segments: input,

analysis and output. Based on the value of a counter, ICALC,

ANALIZ performs the proper function in sequence.
The COPES program currently provides four specific capa-

bilities:

1. Single Analysis - one cycle through the program, as

if ANALIZ was executing alone.
2. Optimization - minimization or maximization of the
objective function with constraints and side-
constraints imposed.

27
 3. Sensitivity Analysis - used to explore the effect of
changing one or more design variables on one or more
functions.

4. Two Variable Function Space - provides tables of data

of all specified combination of two design variables.

However, a recent addition to COPES (still in the develop-

mental stages), has put the Augmented Lagrangian Multiplier

Method (ALMM) at the disposal of the programmer. Because of

its good rate of convergence and its theoretical properties,

the ALMM is preferred for equality constrained problems [35].

Therefore, COPES will call the various subroutines in order

to optimize the objective function; subroutine ANALIZ for

necessary analysis information, CONMIN for an optimum based

on inequality constraints, and a subroutine (yet to be named),

utilizing ALMM for an optimum satisfying the equality con-

straints.

For detailed explanation of the ALMM, its background and

mathematical derivation, consult reference [351; but for now

consider the equality constrained problem:

Min f(K) (10a)

Subject to hk( ) Q Ic= I,NECON (10b)

where NECON is the number of equality constraints.

Define the "modified" Lagrangian function as:

m
L(X,+) = f(h) + ik k
i=l

where A i is the Lagrangian multiplier. The problem can now

be stated as:

28
 Min L(X,X) (1 Oc)

Subject to hk(X) = 0 (10d)

Then, according to Lagrange, if a A can be found for which

solves the problem stated above, then X is also the solution
to the original problem, eq. (10a) and (10b).
The new problem is solved by the conventional exterior
penalty function method, because this is believed to be one
of the most efficient algorithms for the solution of such
equality constrained problems (35].

29
 ti

II. HEAT EXCHANGER ANALYSIS

A. INTRODUCTION

To meet the objectives of this thesis, an analysis program

for an air-cooled heat exchanger must be coupled with a

numerical optimization scheme to produce a complete, detailed

design package. COPES/CONMIN has greatly simplified this task.

This analysis program must be written in subroutine form,

titled ANALIZ, and organized into three segments: input,

execution, and output. The analysis subroutine must also:

1. Take into account the variation of the heat-transfer

coefficients and differential pressure drop with
temperature and/or length of flow path.

2. Be iterative free, if possible.

3. Be written in such a manner that the optimizer will

play a role in surface selection.

With the number of design variables approaching the prac-

tical limit, the importance of an iterative free analysis

subroutine cannot be over-emphasized. The reason being, that

at the beginning of each design iteration in CONMIN, the calcu-

lation of all gradients (each design variable and active con-

straint) requires a complete pass through ANALIZ. Therefore,

the computational time required by ANALIZ directly affects the

time required to reach the optimum.

B. PROBLEM FORMULATION

The air-cooled heat exchange: is shown in Figure 10. A

cross-flow arrangement with bot. fluids unmixed was chosen.

30
Cool air enters the heat exchanger at temperature T ,
c1
pressure p., and constant specific heat cpa The cool air

makes one pass through the exchanger as it flows over an isos-

celes pitched bank of finned tubes. The air is heated by

water, in single phase,

' at an entering temperature of b1 ,
and constant specific heat cpw
The analysis of theair-cooled, cross-flow heat exchanger

centers about the first law of thermodynamics and on the heat

transfer equation. These equations as they apply to the

exchanger of Figure 10 are as follows:

3 = m ac p (Tc2 ) (12)
ca c 12

Q = m c (T -Th2) (13)
w h1 2

Q5 = U A AT (14)

where ma, T and mw, T are the fluid mass flow rates and

exit temperatures of air and water respectively. U is the

true mean overall heat transfer coefficient based on the out-

side root tube area, A is the total heat transfer surface

area of the exchanger used to compute Um , and ATm is the

mean temperature difference of the given exchanger.

The object of the analysis, therefore, is to determine

Q31 Q 4 and Q5 given an initial listing of values for the

design parameters. The list includes the following:

31

. - I
pip_

Tubeside mass flow rate, ibm/hr

Entrance temperature of hot stream, OF
Exit temperature of hot stream, OF
Specific heat of hot fluid, BTU/lbm -F

Air mass flow rate, ibm/hr

Entrance air temperature, OF
Exit air temperature, OF
Entrance air pressure, psi
Specific heat of air, BTU/lbm-°F
Tube inside diameter (ID), in.
Tube outside diameter (OD), in.
Fin height, in.
Fin thickness, in.
Fin spacing, in.
Transverse Pitch, in.
Longitudinal Pitch, in.
Bank height, in.
Bank width, in.
Cross-flow arrangement
Fin type
Number of rows
Number of passes
Given heat transfer rate, BTU/hr
Thermal Conductivity of tube material, BTU/hr.-ft.-0 F
Thermal Conductivity of fin material, BTU/hr.-ft.-OF
Among the initial listing of design parameters above,
there are parameters that are known and will remain constant

32
throughout the design problem. Also, there will be those

parameters that are unknown and can vary, i.e. design varia-

bles. On the way to determining the various heat transfer

rates, other information will have been computed. This inform-

ation includes the objective function, constraining functions,

and other design data, such as the number of tubes per verti-

cal row.
The optimizer will then manipulate the design variables

in order to find an optimum, while at the same time, performing

a heat balance, that is:

Q = Q3 = = Q5 (15)

where Q may be some given heat transfer rate.

C. PERFORMANCE CALCULATION PROCEDURE

With the temperatures, mass flow rates, and specific

heats all specified in the listing of design parameters;

whether they be constant or variable, the only unknown quan-

tities on the right hand side of equations(12) through (14),

are Um, A, and ATm . They will be determined as shown in

Figure 1.

1. Mean Temperature Difference (MTD)

For many flow arrangements, various approahces for
determining MTD, mainly using diagrams, are available [36],

which have proven very useful in manual design efforts. For

computerized design, however, an explicit, approximate equation

is desirable in order to achieve a fast, sufficiently accurate

33

-i
calculation of the mean temperature difference of a given

flow arrangement.
Roetzel, et al. [37], presented such an approximate

equation together with empirical coefficients for nine

counter-current cross-flow arrangements as they apply to

air-cooled heat exchangers.

Roetzel used the familiar equation for the MTD of the

given flow arrangement ATm:

AT =
m F AT zm (16)

where AT, is the limiting case of pure countercurrent flow:

m
(Th -T - (Th -T
AT z h1 c2 h2 c1 (7
(17)
m (Th Tc2)
ln (Th2 -T)

and F is a correction factor determined by a different set

of coefficients for each flow arrangement. Roetzel reported

the following function suitable for F:

amkln k
F n ai ( -v sin(2 iarctan R) (18)
i=l k=l m

where is the dimensionless LMTD.

AT z
_ m (19)
m h c

34
 T -T
R -
1 h2 (20)
c2 c1

and the coefficients ai,k of the approximating equation (18),

were calculated using a standard least squares estimation

program [38], and are reported in reference [37]. The assump-

tion that both streams were unmixed was used in their calcu-

lation.
When more than four tubeside passes are used, it is

assumed that the heat exchanger has approached the limiting

case of pure counterflow and F is set equal to one [2].

Having determined the MTD, the remainder of the analy-

sis procedure follows Roetzel's [39] general approximation

method for determining the mean overall heat transfer coeffi-

cient, Um, for any flow arrangement while taking into account

the variation of the heat transfer coefficients and the

pressure drop with temperature and/or length of path.

Before continuing with specific analysis procedures,

a brief summary of Roetzel's general approximation method is

in order.

The local overall heat transfer coefficient based on

the outside root tube area can be written as follows:

1
u Ao Ao ln(r /r.) (21)
+
S 2kL + Ho n

where nf is the efficiency of the extended surface.

In order to determine the individual convection heat

transfer (film) coefficients, H. and Ho according to the

35
conventional methods, the coefficients would be considered

constant, and the necessary fluid properties for their calcu-

lation would be evaluated at some mean bulk temperatures,

TThb andT Cb

However, the film coefficients are not constant, but

vary with temperature and/or length of flow path. Roetzel

has taken these variations into account with the use of

corrected reference temperatures. Two sets of corrected

reference temperatures are determined: T T and

Th Tc Therefore, for each set of corrected reference

temperatures, the film coefficients are determined in the

conventional manner using the reference temperatures in place

of the bulk temperatures.

With the film coefficients, H. I Ho, H. and H
1, 0 1. 0
two local overall heat transfer coefficients can be calcu-

lated from equation (21), U I and U I.

Finally, the true mean overall heat transfer coeffi-

cient is calculated as:

1 = 1[ 1 + (22)
U-
m UI UII

2. Determination of Reference Temperatures

The reference temperatures for a pure counterflow

heat exchanger must first be determined from (39]:

T= I (T " cl )
hT Th 2 (T T (Th T
hI 2 h1 h2 L
(23)

36
 h h
Th
+(Thl-Th2)h(T
S
T AT 1

Li
1

(T
(Th 2
T
2
-Tl)
)
2
T(2
] (24)

T' h (T -T (T -T-C
CI cl C2 l (Th-Tc2 ) - (T -T (25)

= T + (T -Tl)
IhI (Th 2-
2
) 1 (26)

cII 2 (Th TC 2 (Th2 Tcl)

AT = (ThlTc2).78868 .(Th2_Tc).•21132 (26a) .

I1 1h h 2 72

TT + (Th -T 11
A2 - T )(Th
c7 (26b)
11• 2 1

Equations (26a) and (26b) were derived by Roetzel in references

[40] and [41]. For the special case where the fluid heat
capacity rates, C, are equal, that is,

m = mC
h CPC

the term

T. (Th2-Tc(
I Th-Tc )- hTh2-Tcl

capacty
, ar equa, rtes, tht is
of equations (23) throuh (26), with j = I and II, becomes

.78868 or .21132 respectively. c

37
 3. Correction of Reference Temperatures for Given
Arrangement

With inlet and outlet temperatures fixed, pure counter-

flow yields the highest mean temperature difference. There-

fore, for any other arrangement, the temperature difference

would be smaller. Thus, the corrections are applied in the

following manner (39]:

Th TA

TThI T
h I

Tc = T'
c +P c
I

TTCI = T'ci + c
c

where +, and wv are the temperature corrections of the

Ck
hot and .old streams, respectively. The corrections are

calculated as follows:

T 1- AT /LTz 1
h (Ch/3c)

AT
Fl-T /AT
m
1
cj 1 + (Cc/Ch) 2/3

The corrected reference temperatures are now used to

determine the thermal conductivities and absolute viscosities of

38
fluids for later use in the calculation of the film coeffi-

cients. Thermal conductivity and viscosity data are usually

presented in tabular form. However, for use on the computer,

an explicit, simple approximate equation with temperature as

the independent variable, was desirable. Water and air were

chosen as two fluids that were likely to be involved in an

air-cooled heat exchanger design. Figures 11, 12 and 13

indicate that the thermal conductivities of air and water

and the viscosity of air can be approximated by a second order

polynomial. The viscosity of the hot tubeside fluid, water,

must be treated specially, due to the following considerations.

Calculations for the tubeside heat transfer coeffi-

cient in the laminar, transition and turbulent regions all

invove
he crretion Li/~ 14
involve the correction, (p/pwall)" [42]. In the past, an

iterative procedure was required to determine the inside tube

wall temperature in order to evaluate wwall* Roetzel [43],

has developed an iteration-free method for determining this

correction. Roetzel's method assumes that the tubeside
fluid's viscosity follows Andrade's Correlation, that is:

= c e /T

Using the viscosity data of saturated water from reference

[44], the coefficients were determined through regression

analysis with the resulting equation:

= .01339 exp( 2 7 1 5 .7764) (27)

-M=oe-39 ex ( 39
where temperature is in degrees Rankine and viscosity is in

ibm/ft-hr. Equation (27) was plotted next to the experimental

data with results that indicate that water does indeed satisfy

Andrade's equation, see Figure 14.

4. Uncorrected Tubeside Heat Transfer Coefficient

The tubeside heat transfer coefficient is calcu-

lated from one of three Nusselt-type empirical equations

[42], as follows:
For laminar flows, Reynolds Number < 2,100

HD~ = 1/3 1/3 D /.14

k i 1.86 Re Pr ( LL/) ( --wall
H-- 4 (28)

For transition regions, 2,100 < Re < 10,000

SHD
i 2/3 D1 2/3] 1/3 __ )14
[Re -125] [1 + (L) Pr "
'wall
(29)
For turbulent flow, Re > 10,000

HD 023 Re "8
Pr 1 / 3
k " (--L)'14
w 30)
0

Therefore, before any calculations can even begin,

the Reynolds Number, Re, must be computed to determine the

type of flow.

The Reynolds Number will be calculated as:

Di mh
Re.J Ax Uh.

40
D i , mh is supplied in the initial list of parameters and
P is calculated from equation (27). This leaves only the
cross-sectional flow area, Ax , to be determined, where:
D.2
A = Np
x 4

The number of tubes per pass, Np, is determined geometrically,

having been given the tube bank height, number of rows, num-
ber of passes, transverse pitch, diameters, and fin height
initially. Figure 15 details the procedure of finding Np"
Mathematically,

N
=I n rows D t
Np n -- [-Df 2 /Pt
--
pass

and would be a rational number. This rational number is

then truncated to an integer number of tubes.
With Re calculated, the Prandtl Number,
Cp, 1 h.
Pr. 3 k
kh.

is computed from given and previously determined thermo-physical

properties.
The uncorrected tubeside heat transfer coefficients,
.14
H!j, that is; H. , without the factor (p/w) , can be
calculated from the proper choice of equations (28) through

(29).

41
 5. Calculation of Wall and Associated Resistances
Equation (21) can be written in a more general form
as:

U- 1
Ri + R + o (31)

where Ri, R and R are the inside wall and outside heat
z.
w 0
transfer resistances respectively. Additional resistances,
such as contact or fouling, can also be added here.
In comparing equations (21) and (31), the resistances
can be computed as:

R. Do (31a)
1j D i H.

Rwall r0 ln(Do/D
kwall i ) (31b)

R0. H 1 flf
Hoj (31c)

6. Calculation of Airside Heat Transfer Coefficients

In order to have the optimization program play a
significant role in the selection of an optimized surface
for a finned tube heat exchanger (which is one of the objec-
tives of this project), an explicit equation for H involving
tube and bank design parameters as independent variables is
a necessity.
In the past, comparison methods, as described by Shah
[45], were used to choose the best surface from a list of

42

J
surfaces for which experimental heat transfer and friction
data existed. The data is presented in graphical form,
where Friction and Colburn Factors are plotted as a function
of Reynolds Number. With J, ma , c and Pr known, the film
a p
coefficient can be computed.
Therefore, in previous heat exchanger optimization
programs, a given surface described by its pitch, outside
fin diameter, fin thickness, fin spacing, and outside tube
diameter, had to be chosen beforehand. After the surface
configuration had been selected, expressions for f and J
were obtained by fitting polynomials to the experimental
data described earlier.
Briggs and Young [47] provide a means of getting
past this obstacle with an improved convection heat transfer
correlation for air flowing across triangular pitch banks of
high finned tubes.
Briggs and Young expanded the work of Ward and Young
[48], to cover a total of 18 differently configured finned
tube banks in order to determine the effect of fin thickness
and tube pitch on the airside heat transfer coefficient. The
heat transfer data for the high-fin tube banks were correlated
to give:
H D
Nu Re .718 Pr1/3 s 296 (32)
k. .1378 Re(r2(

where s is the distance between adjacent fins and z is the fin

height. Equation (32) is based on tubes having a wide range

43

. ........- 1.
of fin heights, fin thicknesses, fin spacing and outside
tube diameter and can be used to predict H for a bank of
0.J
tubes six rows deep. Figure 11 of reference (48] is used to
correct H' for banks of other than six rows.
0.
7. Calculation of Fin and Surface Efficiencies
In order to calculate the outside heat transfer
resistance, which will be used to calculate the correction
for the tubeside heat transfer coefficient and finally, the
local overall heat transfer coefficients, the extended sur-
face efficiency, nfl, must be computed.

The surface efficiency accounts for the temperature

drop from the root to the tip of the fin, due to the thermal
resistance of the fin material. Thus, even though the heat
transfer has been increased by the additional area of the
extended surface, the area is not as effective as if it were
to be all at the root temperature.
The surface efficiency can be expressed as (49]:

Af
f

where , the fin efficiency for a radial fin, can be found

from (491:

2 r °0 1 (mrf)k 1 (mro ) - k1 (mrf)11 (mro)

7mr f2_ro2) [Io(mro)k 1 (mrf) + I1 (mrf)k 0 (mr)

The finned area, Aft and the total heat transfer area, A, are
computed as follows:

44
p.
.!
 2 2
As = NT wf (Df -D

A = NWrf(D f -D ) + (1- rft)Dor]

With R O calculated from equation (31c), the correction to

the tubeside film coefficient can be made.

8. Correction of the Tubeside Heat Transfer Coefficient

With the tubeside heat transfer coefficient dependent

on the wall temperature, the dependence has either been 1

neglected, or the wall temperature has been calculated with

an iterative technique in the past. Roetzel [43], has pro-

posed an improved iterative-free method for finding the "Seider-

Tate corectin ( Uwal) .14
Tate" correction, (/., when the tubeside fluid

obeys Andrade's viscosity correlation.

From Roetzel's formulation:

Pwall j 2

where

*
waljwallB
u. .07ln-~~--~)-LC.
-.07 ln-- B.
r 1 -TTh

T - /T*c/Thj
/ 1
h.
T
T.

T
C.
3 - 1
Bj h. wl -1 BjTh

V-7J j wall j

45
where

B. = H! A (Rwal R
Swall 0.
JJ

(Note that all temperatures are in degrees absolute.)

All the parameters on the right-hand side of equation

(21) are now available. Therefore, the two reference over-
all heat transfer coefficients, U I and Uii, can be calculated.

The mean overall heat transfer coefficient follows easily

from equation (22).

With UmI Q 5 can be calculated, with the heat balance

to be performed by the optimizer.

9. Calculation of Air and Tubeside Pressure Drops

The final calculations before computing objective and

constraint functions for the optimization problem involve

the pressure drops in the heat exchanger.

The basic equations that will be used for the calcu-

lation of the pressure losses are as follows:

a. Tubeside (421

A *2 w n P
f i mh - z+(nP-i *2
m (3
h
2Pg-c= T pDi
P (xT) z) + 2 g c AZ
xT (33)2

where z = .14 below Re = 2100 and z = .25 for Re greater than

2100.

b. Airside [501
(
fn2
0 r c
= (34)
Apo
Aff gc

where A is the minimum flow area.

ff
46
 The friction factor for the tubeside flow, fi' is

taken from Figure 9.5 of reference [42], assuming fully

developed flow. For use on the computer, an explicit expression

for f. was obtained by fitting a line and an exponential to

the experimental data of Figure 9.5. This follows:

Re < 1000 f. = .5/Re

26 94
Re > 1000 f .003215 Re

Just as in the case of the airside film coefficient,

for surface optimization on the computer, an explicit equation

for the airside friction factor, fo, is a necessity. Robinson

and Briggs [501, presented such an expression for fo for air
flowing across triangular pitch banks of finned tubes. Robinson
and Briggs' work closely parallels that of Briggs and Young

[47]. The Robinson-Briggs Correlation:

f = 18.93 Re.316 t)-.927 (t 515 (34a)

0D0 PL

covers the range of tube sizes and pitches used in air-cooled

heat exchangers (50].

Therefore, with Re I and Re,,, the four peference

pressure drops, Apij and Apoj' may be computed from equations

(33) and (34). Following Roetzel's general approximation

method:

47
iAl

S
Lpi
LUI
+±1/ U [U~
+ .

UII
1

For a gas, an additional correction is needed because the

density in equation (33) is strongly dependent on pressure,

which is changing through the exchanger. Using the inlet

pressure as reference:

! PoII 1I
01"P
+-11 , 1 +1
U1 UII 'LU1 UI I

( 2'p
2Ao[) 1/2 (35)
-) I'
p1

All the necessary information from an analysis view-

point has now been calculated. Functions needed for the

numerical optimization process shall follow.

10. Objective and Constraint Functions

The objective functions available for minimization

are defined as follows:

a. Volume = wh[Df + (n-l)PL cos ]

where

S = arcsin (Pt/2PL) (36)

2 - DO 2
(Df __
) t)Do7]
b. Heat Transfer Area = f + ( --
T[ 2 S S

c. Air Horsepower = a ma
p

48
 d. Airside Pressure Drop

e. Tubeside Pressure Drop

Limitations were imposed on the following inequality

constraints in order to keep the design within practical

physical bounds.-

a. The diameter ratio,

Df
DRATIO Df
0

must be kept reasonable. This can also be accomplished

to some extent by placing side constraints on the
design variables, Z and D See Figure 16.

b. The optimizer must be prevented from driving the

tube thickness,

TUBTH (D -D.)/2
o i.

to zero.
c. The tubes must be kept from touching in both the
longitudinal and transverse directions,

TOUCHN = Df - Pt

TOUCHL = D - PL
f L

TOUCHL and TOUCHN, therefore, must be kept below zero.

d. Reasonable temperature profiles must be maintained
at both ends of the exchanger. See Figure 10-7 of
Ref. [441,

PROFH = Tc2 Thh

PROFC = T - Th

that is, PROFH and PROFC, must be negative.

49

1.
 e. The number of tubes per vertical row, VROWR, shall
not be allowed to go below 2.

f. The free face area, that is, the minimum flow area
for air, must obviously be greater than zero,

DELSFF = [projected tube area] - hw

where the projected tube area, STOTAL, is

Ir
Df t w ]
= N[Dw + Df
STOTAL
0 5

and N is the number of tubes per vertical row. DELSFF

must Ee less than zero.
g. The airside and tubeside pressure drops must be kept
within design constraints.

h. From experience, the argument of the square root of

equation (35),

ARG5 1 - 2-p-
pl

has been driven below zero. It must therefore be

constrained.
i. To maintain an isoscele pitch bank, the angle, 0,
as shown in Figure 17 and as defined in eq. (36), must
be constrained. One such constraining value is:

Om = arcos (Df/ 2 PL) (37)

The arguments of the arcsin and arcos of eq. (36) and

(37) must be constrained from going beyond 1.

j. The heat balance described by eq. (15) is performed

by three equality constraints:

QRATIO = Q5/0 (38)

QRATIO1 = 3/0 5 (39)

50
 QRATio2 = (40)

All constraints are set equal to 1.

Besides the constraints described in paragraphs a. through

j., above, lower and upper bounds are placed on the design

variables to assure a reasonable design.

51
 IV. RESULTS

A. BACKGROUND

Case studies were chosen as the best way to test the

capabilities of the program for Heat Exchanger Design using

Numerical Optimization (HEDSUP). The design problems posed

were made as realistic as possible.

1. Capabilities

HEDSUP currently has the capability to design for

nine different configurations of triangular pitch banks of

finned tubes:

TYPE 1 - 1 ROW, 1 PASS

TYPE 2 - 2 ROW, 1 PASS

TYPE 3 - 3 ROW, 1 PASS

TYPE 4 - 4 ROW, 1 PASS

TYPE 5 - 2 ROW, 2 PASS

TYPE 6 - 3 ROW, 3 PASS

TYPE 7 - 4 ROW, 2 PASS

TYPE 8 - 4 ROW, 2 PASS

TYPE 10 - PURE COUNTERFLOW

TYPE 10 will include exchangerswith a configuration of n rows,

n passes, where n can go from five to 20.

The banks must be constructed of high-finned tubes

(Q > .0625 in. [42]) with the fins having a rectangular pro-

file of constant thickness. Additional profiles can be inserted

into HEDSUP quite simply, provided that its fin efficiency can

52

t
be expressed explicitly as:

f(Z,t,H,k)

See Subroutine FINEFF of the program listing, Appendix D.

At present, HEDSUP can provide the design parameters

for an air-cooled heat exchanger optimized for any one of the

following design objectives:

(a.) Minimum Volume

(b.) Minimum Heat Transfer Surface Area

(c.) Minimum Air Horsepower

(d.) Minimum Airside Pressure Drop

(e.) Minimum Tubeside Pressure Drop

Additional design objectives can be used, provided that they

can be expressed explicitly as a function of the design varia-

bles and they are added to the common block. It should also
be pointed out that any design variable may simultaneously be

a design objective as long as it conforms to the restrictions

of both; For example, an exchanger may be designed for mini-

mum bank height.

The airside fluid is restricted to dry air. The tube-

side fluid is presently limited to water in single phase.

Other tubeside fluids can be used by HEDSUP, provided that

their visocisites obey Andrade's Law and the fluid's thermal

conductivities can be expressed explicitly as a function of

temperature. The fluid's specific gravity would also have to

be placed in the denominator of eq. (33).

53
B. CASE STUDIES

1. Case One

a. Problem Formulation

An air-cooled heat exchanger is to be designed

for minimum volume with a heat transfer rate of 10,000,000

BTU's per hour. Water is to be cogled from 200 0 F to 125°F.

Dry air will enter the exchanger at 95°F and leave at 130 0 F.

Specifications call for a fan that can produce a pressure

difference of two inches of water.

b. Design Variable Framework

From the list of design parameters in Section

III.B, the design variables must be singled out, identified

to COPES, and given side constraints. All parameters must

be given an initial value. Only the values of the design

variables will change.

Assuming constant specific heats,

C = 1.0 BTU/lbm-"F

cpa = .24 BTU/Ibm-OF

the mass flow rates of both fluids can be determined from

eqs. (12) and (13), because the heat transfer rate and tem-
perature differences are given,

mw =- QyT = 133333. ibm/hr

ma - c = 190476.2 lbm/hr
a CcAT
c

54
 A cross-flow arrangement, fin profile, tube

material, and fin material must be chosen.

The design variables for this example are, there-

fore:

.232 < D. < 2.325 in. D. = 2.0 inches

.24 <D <2.5 in. D =. 2.5 inches '

.0625 in. < z < .46 inches

.01 < t < .0235 in. t - .023 inches

.08 < S < .125 in. Si = .111 inches

0.0 < P < 4.0 in. P 2.125 inches

0.0 < Pt < 4.0 in. Pt = 4.0 inches

0.0 < w < 500 in. wi = 490 inches

0.0 < h < 500 in. hi = 350 inches

The side constraints on the design variables are of a practi-

cal nature with the exception of the lower bounds on fin

height. Recall that the use of eq. (32) is restricted to

high-fins. High-fins will also tend to keep the fluid unmixed,

which was an assumption used when defining the coefficients

aik' used in eq. (18).

c. Constraint Framework

From the problem statement, the airside pressure

drop must be less than two inches of water or .0722 psi,

55
 0 < a
pa < .0722ps

From a practical standpoint:

0.0 < 0 < 1.3

1.0 < DRATIO < 2.5

.018 < TUBTH < .18 inches

-00 < P < .14 psi

The equality constraint

Q51
.

where

Q5 = U A ATm

Q 10,000,000 Btu/hr

will satisfy the heat balance.

The bounds on TOUCHN, TOUCHL ARG5, ARG7, ARG8,

DELSFF and VROWR were discussed in Section III.C.10. Con-

straints on PROFH and PROFC were unnecessary because all

temperatures were specified in the problem statement.

As with design variables, constraints must be

identified to COPES by location in the common block. See

the global catalog, Appendix A.

56
 d. Methodology
Ideally, a three dimensional design matrix can
now be constructed of optimum exchanger designs with minimum
volumes. The matrix would be constructed by first holding
the tube and fin materials constant and varying the configura-
tion, i.e. Type 1, Type 2, Type 3, etc. Next the tube material
would be varied with the fin material and exchanger configura-
tion held constant and so forth. However, for case study one,
the tube material will be chosen as copper, k = 200 BTU/ft-hr-°F,
and the fin material will be aluminum, k = 118 BTU/ft-hr-°F.
Also, in order to simulate an actual trade off
study, the constraint framework will be fixed throughout the
individual case studies.
Problems arise in constructing the matrix when
trying to determine the true minimum volume design for each
configuration. Unfortunately, the choice of initial design
parameters (starting point), coupled with the input parameters
for ALMM, will sometimes lead to entirely different optimum
designs with volumes differing by over 100%.
The ALMM parameters include the initial multiplier,
CC, the multiplication factor, CMULT, and the maximum multi-
plier value, CCMAX. Experience has shown that setting

CMULT = 2.0

CCMAX = 1000.

will suffice for almost all problems. However, there is much

"artwork" involved with the choice of CC. From experience,

57

1 ,-
an initial multiplier of 10 works well when starting far
from the equality constraint, i.e.:

.8 < QRATIO < 1.2

However, when approaching the heat balance, i.e., the equality

constraint, a CC = 100 is necessary for the heat balance to
converge. Equation (15) is considered satisfied when the heat
transfer ratios (eqs. (38) through (40)) are less than 0.1
percent.
Therefore, it is obvious why the chosen initial
design is so critical. Together with the choice of CC it
will determine how and to what design the optimizer will con-
verge. As an example, see Table 1. Notice the calculated
heat transfer for the initial design. This value is the product
of Um , A, and ATm calculated by ANALIZ using the initial design
parameters, some of which are mere estimates. Recall it is
the job of the optimizer to vary the design variables in order
to bring Q5 equal to Q, Q3 and Q4 and at the same time minimize
the objective function.
In order to remove some of the "artwork" and
try to ensure a true optimum design, the following procedure
is recommended to generate the design matrix:
(1) Begin with a Type 2 configuration; input the initial
design and constraint values enumerated in sections
(a) through (d) above; let CC = 10.; execute.
(2) If the heat balance of the resulting design has not
converged, but is within 20%, use the design results
as a new starting point and set CC = 100 (if the heat
balance is not with 20%, let CC = 10.).

58
 TYPE 7 EXCHANGER CONFIGURATION OPTIMUM DESIGN

RUN I RUN II

DESIGN INITIAL OPTIMUM INITIAL OPTIMUM

VARIABLES DESIGN DESIGN DESIGN DESIGN

D. inches .6786 .5392 2.0 1.168

Do inches .7201 .5767 2.5 1.278

Z. inches .1618 .128 .46 .294

t ,inches .0217 .0208 .023 .023

S inches .08 .08 .111 .08

Pt4 inches 1.044 1.02 4.0 2.557

PL inches 1.044 .841 2.125 1.899

h ,inches 342.1 238.4 350. 172.9

w ,inches 286.6 186.2 490. 253.

oQ BTU/hr 21,988,624. 9,999,925. 20,650,352. 9,999,811.

Volume, ft3 213.12 72.92 553.2 153.83

Table 1

59
 (3) Repeat step (2) until convergence. NOTE: CC may
be adjusted up to 150 when approaching convergence.
The design should converge following the use of
CC = 100. If too much adjustment of CC is necessary,
reaching the optimum from the starting point is
unlikely.
(4) Ensure a minimum design by beginning step (1) with a
different initial design.
(5) After finding an optimum design for a Type 2 con-
figuration, use that design for the starting point
for a Type 3 configuration. This assures a reason-
able starting point, probably close to the optimum
for Type 3.
(6) Repeat steps (1) through (3) for a Type 3 configuration.
(7) Repeat steps (5) and (6) for the remaining configuration.
e. Design Matrix
The design matrix is presented in Table 2. The
optimum design is a Type Four configuration and is shown in
Figure 18. Typically, for this case study, when starting far
from the final design, COPES/CONMIN would require approxi-
mately 1700 calls to ANALIZ to reach an optimum. However
when beginning from a reasonable starting point with CC = 100,
only 600 calls were needed. Note that each call to ANALIZ
requires approximately .06 seconds of CPU time on an IBM 360/67.
2. Case Study Two
a.- Problem Formulation
An air heater is to be designed to fit into a
space 81 X 24' X 4'. The heat exchanger is to heat 1,000,000
lbm/hr of dry air from 750 F to 130 0 F. 256,000 lbm/hr of water
at 2000F is available. Design the heater so that the required
air horsepower is at a minimum.

60
 0)(n Ln r
O N M(N
Hr- 0 - P-4 co 0 * N
4-4Lf Lo 0 0 0 as
0i aN
0 Ln n qv ONa

LA~~c N( LA I e

rOO LA In 0 0 0 a%~ N 0 0 i -
U) QN ON 0n-4
4 ON

E-4 (N
co 0T 'N C
( -q I-n n .-4 0 0D0 n 0 N (N
2 ~a. . . .4 N O-I N

:3- U*O -T CN L ON4

a)
0N r- q . ------
CD .-4
N 0 *-
r- CN k.0 co 0
(D 0 (N
4 U q a

. IV co co -1~O
CJN 2N 0 Na.-N A I

taO
0- 14W r- kZ4 %10 -W 000

N
IVL~ N
w N 0 0 *n ~ ON M
-4 0 1-4 -4 (N4 0-
C
r--4
>4 V r4 0IIIn -
in a a 4 (N N 1- D 1- n
as C) (DUn r- a% (mN No 0
mrn 0 I- 0% (D 0 0%0
(N~~~-4 *( *
.- 1 1-4 (N4 (N 0 '-
(N1

N 0 M hn e
N N~ flA ON

S (N U) (N NW * * i -4
co C o 04 0o VN Lna

E- ON LA
m %-

2n V) L
C4LA
e(N tI4N
0D
an 0
U) U4 U) U4 U) U) U a)
U)n
ww
N ) w w w N) wn aw . L
m~ 0 (N m U) W) rU w -W
II 444) 4) OJ4
m - 'n. . u u u

61
 b. Design Variable Framework

Assuming constant specific heats,

i= 1.0 BTU/lbm-°F

c = .24 BTU/lbm-OF

eqs. (12) and (13) will yield the required heat transfer

rate and the outlet water temperature.

C*
Q (T -T ) 13.2 10
6 BTU/hr

mh Cpw T -
Ch Q

Th - C nh = 148.44 *F

Therefore, the design variables are as follows:

i
.232 < D.1 < 2.325 in. DI1 = 2.0 in.

.25 < D < 2.5 in. D = 2.5 in.

.0625 in. < z < = .46 in.

.01 < t < .0235 in. ti = .023 in.

.08 < S < .125 in. Si .111 in.

i

0.0 < P < = 2.125 in.

L L
i
0.0 < Pt < W Pt M 4.00 in.

62
 c. Constraint Framework

When designing for minimum horsepower, the opti-

mizer will naturally try to drive the design to a maximum
volume in order to reduce airside pressure losses. Therefore,
it is reasonable to constrain volume as follows:

0 < Volume < 768 ft 3

The other constraints are:

0.0 < 0 < 1.3

1.0 < DRATIO < 2.5

.018 < TUBTH < .18 in.

0 < Apw < .14 psi

and for the heat balance:

Q5
1

d. Design Matrix
With the tube and fin materials fixed, as in case
one, the matrix is presented in Table 3. The optimum design
is a Type 4 configuration, as shown in Figure 19. In case
study two when starting far from the optimum, the optimizer
called ANALIZ approximately 1900 times. When starting close
to the final design, for example using the design for a Type 3
configuration as a starting point for the Type 4 design,
COPES/CONMIN only required 596 calls.

63
 -4 E- C N( r
t.4 00~ '0 i
I E-4 00 . N 0. 0 ko 0o
Lr
£44 . q14 ( N N
C4 r
$.4 04 M II __ _ __-

oo12 E- LA r .m. N
C..
~-49 N
04~~' U) 0
m% 0% (N
C
00*
(N
0
0C
0
0 ( Ln
N ..(,30N ~(
$4 -4 0-

m 0 0 0 (
'4q 0% (NI
MV4 0N r,

-4) -

Iq .04 0~

~
Im ~ IMo4ma e N 0
.14O 0

NH Hn 1* C. CN

02'.

U r-4 r- r (NI 0 ..

N r-4
H r E- E-4 Cl '-4 ( (N H* N N

(a1 0

U- N N- m-

00- 0 U) r- %

'U
0

Ln wH0
-g
___ I -4 m'-.0

N ~ N
o1 if) - l) C) c 4l) Cl4 -

o - E-00 0 ) 0 0 0 .

o a ~.~ . .~ . ~ . .64
 V. CONCLUSIONS

The intent of this investigation was to couple an analysis

program with a numerical optimization scheme, COPES/CONMIN,

to produce a complete, detailed design program for an air-

cooled heat exchanger, (HEDSUP). In addition, the analysis

program was to be written such that:

1. the variation in the film coefficients with temperature/

length of flow path would be taken into account
2. the surface would be optimized
3. it would be iterative free and thus minimize the CPU
time required during an actual trade off study.
The results from test cases using ANALIZ coupled with
COPES/CONMIN in its present form were unsatisfactory. Although

COPES/CONMIN could optimize the objective function satisfying

the inequality constraints, a reliable heat balance could not

be obtained. The solution to this problem was the addition

of the ALMM option to COPES. In this way, the method of

feasible directions, which works best with inequality con-

straints, was used to satisfy the inequality constraints.

The multiplier method, which works best with equality con-

straints, was used to perform the heat balance.

The results of the case studies show that HEDSUP will

yield reliable designs for various design objectives and

problems with only some trial and error application of the

initial Lagrange multiplier. However, precautions must be

taken to overcome the relative minima that plague this design

65

--i -- mlI li i m ,, , , . i.ih if

i=a .
problem. Table 1 shows vividly the problem of relative

minima with one "optimum" design having a volume over 100%

greater than the "true" optimum.

The value of numerical optimization in a design problem

of this size cannot be overemphasized. For example, in case

study 1, the problem is taking place in a nine-dimensional

design space and intuition on how an optimized design "should"

turn out is quickly lost. Figure 20 helps to illustrate this

point. When only varying two design variables, h and w,

Figure 20 shows that, when beginning from the initial starting

point with a Type 7 configuration, a design satisfying both

the equality constraint and the air pressure drop constraint

could never be found. However, Table 2 indicates how the

numerical optimization routine has varied the other seven

design variables in order to shift the constraints and yield

an optimized design.

The results show that AHDOP did vary the surface design

variables: D i , D0 , Z, t, So Pt and PL' in order to produce

an optimum heat exchanger. This capability of surface opti-

mization is dependent upon the use of the Briggs-Young and

Robinson-Briggs Correlations, eqs. (32) and (34a), respec-

tively. The reliability of the correlations as compared to

the "conventional" method is questionable. Actual experimental

data for a particular tube and pitch will always be the most

useful in predicting pressure drop and film coefficients of

across banks of finned tubes. However, the correlations

66
 mentioned above cover the ranges and pitches used in air-
cooled heat exchangers, and should therefore be sufficiently
accurate in predicting H and AP

II
67I

I *1
i . .i "i
nlU
..
.
 VI. RECOMMENDATIONS

In addition to the insight that this investigation has

given into the generation of an automated air-cooled heat

exchanger design, it has also generated an awareness of this

investigation's shortcomings. As mentioned in the review of

previous work in this area, each of the optimization methods

has its own limitations; none is completely general. Pre-

sented herein are recommendations for improving upon and

furthering development of HEDSUP.

1. HEDSUP should be expanded to include the capability

for two-phase tubeside fluids. Mott, et al. [21 discusses a

method involving two-phase tubeside fluids that would be

compatible with HEDSUP. The modular design of ANALIZ will

aid this effort.

2. Research in the area of numerical optimization using

discrete variables would benefit HEDSUP immensely. With dis-
crete variables, the design of the exchanger could be accom-

plished with "off-the-shelf" materials. At present, the use

of the optimizer is restricted to continuous variables with

continuous first derivatives. The ability to work with dis-

crete variables would also eliminate the need for the design
matrix. The optimizer could optimize for type configuration,

fin profile, fin material and tube material.

3. Additional research with ALMM is needed to remove the

"artwork" involved with choosing an initial multiplier and

68

44
 thus increase reliability and hence reduce CPU time. The

research should be concentrated in two areas: 1) debugging

the new optimizer with AFMM and 2) scaling of design variables.

4. The addition of cost as an objective function would

increase the attractiveness of HEDSUP. Mott, et al. [2],

and Fontein and Wassink, [18], have presented much useful

information in this regard.
! 5. Mechanical constraints such as tube bursting stress
and tube vibrations should be included in HEDSUP.

69
 VII. FIGURES

.Inpu't

Calculate Log Mean

F Temperature Difference

Calculate Gerrected
FReference Tempueratures

Calculate Number of
-Tubes & Tubeside Flow Area

Compute Uncorrected Tubeside

Heat Transfer Coefficient

Calculate W~all & Associated

Resistances

Compute Uncorrected Airside

Heat Transfer Coefficient

Correct Airside Heat

Transfer Coefficient

Cculate Fin & Surface

F _Efficiencies

Calculate Fin & Total

Heat Transfer Areas

Calulate Aonsrints

Coect Turbelsfl

Comtutt Ma vr J
Presure 1rp

70
 ca C c: -4*4d

E-4 >3
+3 +X4W"-4
LC0 :5
-Z

0 d) v
(D r4
E-4 4 0 "-1
w go 0
~~.,W-
~ to-
C) 0 0>
0l004 4J
5.-4 "
W4O -

'--4E4> 3

E -4 C-C
o ,4

00
-4

-4
0 z

E4
44 0

oo
'-4r Co
4- $4 -16e
00)t 0o

woo
-4 -C 0540
t
oo 43cq

r-44-2 Q 10f
0 0d

0 Figure 2

71
 I- A

S VI

o~ X~
V
0* (D~- oq
w :,w -L

U)W U jf

U) CX. cw<
ct -j w () W _j i
U) a) D 5

V) co
- cf
-

I. LAL z C7
(n -_c wi

L5~. cc LLJ

<L 0 0(n .

AN-t=
xf r0ZC

Figure 3
72" %c
ONE-DIMENSIONAL SEARCH IN DIRECTION Sq -

!I
F(X)

Gz(X)=O

eH
b
dG 3 (W)0O.
1

F(X)zmin
G1(X)= 0 -
G2 (X)xO
1•

Figure 4

73
 SIDE CONSTRAINT
2.2
VOL
1 CONTOUR OF CONSTANT VOLUME

2.0

S1.8

~1.6

VO

Uj 1.4
cc

1.2

1.0 o in o 11 1 11
0.0 0.2 0.4 0.6 0.8
TUBE OUTSIDE DIAMETER (INCH)

Figure 5

74
 SIDE CONSTRAINT
2.2

VOL 1 CONTOUR OF CONSTAN4 VOLUME

2.0VO

LVO

S1.6

LVO

.4

1.2

0mi-80-0

1.0 - -L*~---- IF A- I v 0kL -0.6

0.0 0.2 ~0.40.08
TUBE OUTSIDE DIAMETER (INCH)

Figure 6

75
 SIDE CONSTRAINT
2.2

V1 CONTOUR OF CONSTANT VOLUME

2.0

IA
LaVOL

S1.6

LVOL

1.2

0.0 0.2 0.4 0.6 0.8

TUBE OUTSIDE DIAMETER (INCH)

Figure 7.

76
 SIDE CONSTRAINT

2.2L
VOL1 CNOUR
~ OF CONSTANT VOLUME

2.0

1.8

z1.6
dCVO

~1.4
Uj

1.2

' Opn 81 1
1.0
0.0 0.2 0.4 0.6 0.8
TUBE OUTSIDE DIAMETER (INCH)

Figure 8

77
 SIDE CONSTRAINT
2.2

VOL
1 CONTOUR OF CONSTANT VOLUME

2.0

1.8

' ~. O ran0
D =

1.6.0. . . . .

VOL

TUBE OUTSIDE DIAMETER (INCH).

• Figure 9i

78
II

n Passes

Figure 10. Configuration of Air-Cooled Heat

Exc hanger

79
 +

+.++- +

U-. ++
i
U1" +

..
.
a:
0
111 ++
+

+
++
++
+-I+
+++4!
+ ,
+
ai:+ +
(U) + L+

I-I
++ + +r 1
mi I-+ +

+4
0 ++
+

0 0 0 0 0 0 m 0 0 0 a
0 0 w N CO 17 V, m' fu

(..) dW31
Figure 11

80
 ""+ + +"i I
+ +

++
S++
IZE'

S+ I
F 1
+ 4

'7- - I
Ld +

H ++:

ci: ~1

Figure 12

81
 4.
++
++
+

w! En~
+
In + +
IL +
Q. ++ I
+
- +
+
_+
+

I " +
++ + 0-
+ 1 F-
cc ++,
++
+
++ -+
++ +

+ +
+

+
•

0) co N w Ifl v P1 cuJ

(j SOP) dW3I
Figure 13

82
06

al *
CE Lo

IL 89"Ti-

08 >

( m~ **CV

(J) dW3

839
 4 ROW, 2 PASS ARRANGEMENT

Df

-
Air

in
out

Figure 15

84

L a t. lso-m-a.o---oom-i --
 D___ Df

Figure 16

Figure 17

85
CASE STUDY ONE

DESIGN OPTIMUM

.128"

1" 201"

Figure 18

86
 CASE STUDY TWO

OPTIMUM DESIGN

0.9"

I - ' 81.85"

l 1. 81 ,

+ .0235"

4 K93"'

3.65"

Figure 19

87
 f/
U

o / 4

L.

>
CU

!.I!

Figure 20

88
 APPENDIX A

HEAT EXCHANGER DESIGN USING NUMERICAL OPTIMIZATION (HEDSUP)

USERS MANUAL

In order to execute HEDSUP, it is necessary to provide

formatted data for COPES [25], followed by formatted data

for the ANALIZ portion of the program. This section defines

the data which must be supplied. The data is submitted in

segmented blocks. All formats are alphanumeric for TITLE

and END cards, F10 for real, and I10 for integer data when

supplying COPES. For ANALIZ, formats are F14 for real and I10

for integer data.

While the COPES user's sheets define the data in formatted

fields of ten, the COPES manual does provide means of simpli-

fying this task through unformatted data input.

The included GLOBAL CATALOG defines objective functions,

design variables and constraints along with their location

for ease in compiling the necessary COPES data.

89
 GLOBAL CATALOG

GLOBAL FORTRAN
LOCATION N'AMkE DEFINITION
1 FMDOT Fluid Mass Flow Rate, lbm/hr
2 FLTMP1 Inlet Fluid Temperature, *F
3 FLTMP2 Outlet Fluid Temperature, *F
4 AMDOT Air Mass Flow Rate, lbm/hr
5 ARTMP1 Inlet Air Temperature, *F
6 ARTMP2 Outlet Air Temperature, *F
7 TUBEID Tube Inside Diamter, in
8 TUBEOD ITube Outside Diameter, in
9 FINHT lFin Height, in
10 FINTH Fin Thickness, in
11. IFINSP Fin Spacing, in
12 PITCHL ILongitudinal Pitch, in
13 IPITCHN Transverse Pitch, in
14 BAN1CW Bank Width, in
15 BANKH Bank Height, in
16 VOLUME Volume, ft .
17 JAREA Heat Transfer Area, ft 2
18 jFHP Air Horsepower, HP
19 PPA Airside Pressure Drop, psi
20 PPW Tubeside Pressure Drop, psi
21 DELQ1 Q 5Q,BTU/hr
22 DELQ2 Q5- Q4, BTU/hr
23 DELQ3 4- Q 3 , BTU/hr
24 DRATIO Df/D 0
25 TUBTH Tube Wall Thickness, in
26 TOUCHN ITip-to-Tip Clearance, Transverse
27 TOUCHL ITip-to-Tip' Clearance, Longitudinal

90
 GLOB&L CATALOG (Cont.)

28 PROM Tc - Thlt_ 0 F
29 PROFC Tc, -.Tho O
30 ARG5 1 - 12AP,/P~]
31 QQT BTU/hr_
3Z2 DELPHI max radiansJ

33_____ QRATIO Qs/&

35ARG8i DfI 2 PL
36 i VROWR Number of Tubes per Vertical Row
37 DELSFF Negative of Free Face Area
38 j THETA arcsin (ARG7)

91
co

E-4

'.4

92a
 C3

< 4,

go 0

4A4

C3"
-=C

C -

0 ..4
41 4)
tw Nn .-

.0 u 0

Cs 's . 0 s

00 .4 ..
C6 ~ ~
r4s I0 ~ 1
go 400

Ii x0 u*t .
0. La4 .Q
z

44. -

93 ~0 0.
 014

aV
UA

$.0
Q 14
a > 4
41 W
00 0 (
0 w 4

aa CU c

-4 0 -4 1

-4 0 U3 U3 "a
4J* -4 a0 u
= m 0 (U
5 '4 C )
.6 41 14 0 1 ='
00 V 2 C6 -4 0 41
o o > A 04

X7 re 14 .4 m. C: w

o~t 0 UZ U
O .6J 4 0 w 0
ej w. 0 (U4 -

Q -4NA.) JA ".a
-P4 r- 1. M C3 06 r6

z gnU -- 4 0

0.A ( -4 4 0
al Ad A44 A4 0
Ad
00 00U . U U 1

Z 4 U U *-

-94
 w 03
0 0

.0 Ci Cd C.
to 0 0.

ci .6J 4
.0 U ~ 1 L
2.
+ w C)
-4 0 C)l. C
w 0

4:
~14
-) 4
02
>,
~.4 0 00 IV0
) .
-In tU
~~1dc
+ L - -4 -)4 03 2.2 C

0 12 coN ., -

Q') Ai C7 4-1 (U.0

L
co 02 44 -4 4
c 14 - ) 1-0 m2 -
0 0 2 -4U C
>0 - ) C

~
- ~ 2Li ~ c e . 0 i -
41~~14 "- 4-4

0N i 0 k& C) O
0 Ca'u"4 4 ) L 0 v2 M. a

C) * 024.4 c A C2 .0. C
ca4 E-Z -40 4 a0 o -M 4
0 44 C)-0M
44 " i .
.6) 00d C3 C) V 4 -4 U

4 0 &l, W) &4 0L %.A.

140 C; .1 ) 0 C) C) 0 C)
0 J- Aj W k-4 &'W0
U 1.' U 2 0 i 4 * 02 . 4

V4 '.4 w..- 2- 4 0 02 0 2 - 2-
0 0Q = 0 1 0 -4 0
w) 4
C) -q 4 Li
u i Zn
l U C)
w 2 02 0 ~C
0-4 4 . U
) 0 . z.

P-4 ). - i
000 L L .j. ) i L )

C-4 en 2
NO C) ~ P,0 ODC
-4 0

.~. ' L .0 ~ d 950

04: -
AD-A096 350 NAVAL POSTGRADUATE SCHOOL MONTEREY CA F/6 13/1

EEEEEEEEIIIII
HEAT EXCHANGER OPTIMIZATION. (U)
SEP 80 C P HEDERICH
UNCLASSIFIED

IIIIIIIIIIIIIIlfflfflf
EIEEEEEEEEEII
EEIIIIIIEEEEEE
EIIIIhIhEEEIhE
IIEEIhhEEEEIhE
EEEEEEEEE*IIIIJ
 4.4
A0
-4

E-44 a

E4.4

C/3

-4 E-4
a
CI

(v A
0E-4.

tn C)dP

0
* 96
 4.0
0
4..
6J I4

0 00
-H ta4

-4 0 0 1 4
o %1

o ~ ~ 0-
0
1
0
4
J~
f
~ i
.
J-6

r. A0
0 1 0 0-4.

o rJ 4 4 .4 4.
-4 41 9:
41
1-I~ 0 ~ j*4 S.
-4 41 -a
0 C
to

N~r .- M4 U - j

-4. . toif

f
" E-4. .. 40 41 4 0 0
E-4
W16 0
w o. 0o 141 0 4

CL 71 w. to1 W .61
03 4 0 4. cis I 0 0
13 0
al = 0
--

0
41 .4
U~a
0a. ~ > '
4 0c
-
0
0
~ 00
04
-4
U
P.4 04
M " = A.
F-' 0 0 0-0 6J L;

Aj i64 0 4A . 4 -M

L~ .4 0 04 0
0 -4 5-4 1 E -4 J-J 4 F 1 4
0 I6 0 > .6 ..) -
m 0 0 Vj U

0- -4 w 4 0 -1 4 01

~ if f U 0 0 041
 A4 t

4N 41N l
1-4 V4 a
pq0 > -4 "4 0

to 0o a -d

-4- 04 t
&. -0 4 4, 4 04

to 0)t > -4 4t 44 t
-P4 Q .4 N -
o 4 CO C: -4
0 00
-4 $a > 0
AU -r4 -. 4 4) "4 * '
d) 0) Qo ca to

U) q). '4 Q 0 0
0 -4 ~4 -4

Wo a~ .4
c 0 ,4
-13 4 t o U 44 ) .44
to 5. 0 U to S
"4 to1 0 4 t

CO 0 -2

too 0 S A A
0 'to CO (U to to

v~ Q V- -4 UA r'44
ow o .~
at .~ " o 't

0 -4 0 U c0
^3 f-4
CU t
zU -4 c
'0 0 4 > 4 '4I Io
% o .6 ,a 0 4.4
to
Li -4 4 '4,4 0 5.a 4.)
aS - 4 -M t
-4 a4 MA
Z ~ ~0 t cU 0
+J 0

C
5.4~~
to
0
to
5. t
5.~>, U
~Q>, '

- )
:. 5..
0 t ~5 to
U
'44
4 '4
t
.
ado V)Q 4 -
dc~5. A - t o t
* 4 I05 o

to 44 t -' i to98
 A2 0 4ci

.00.

~~6d
4 3 4

0 L

> .0 0
u- IA 0 4i

L~~~c . w
4

ca 4 10 -
-* 02

(U i
tA -02 -. .4 L

E-44

.3 430 1
r3 . >0(3

m2 0 w~

Z - ..-4 en

-4 99 0 0 02

02 0WO
 -4 4

DI -4

44 cn4 -4 > c

z m 0 4 1

U3

C-41

40

w w U
> >. (U1

zoz
w 0
Z4 L4)1
w~
QA

I4-, (

Cf0C C
0 0
C's ajn

(U z ~ (100
 JA

.0

o - o

c.0

t,6
4.1 C

0 0
ec CU

z CU C,-

U ..
.i
....
 0 0C

E-4 - 4

41 E44

E1 A-i
02~

C4

E-4 -

03 02
CA U0 0 a

to U

Wa -- q C
La 1. 0 C

V2,. E- 3 a a

C Cu CC
Li ~ C ).w

w 4 14 1
en
~-a
4- i

Cu Z L Li102
 0 .

0 0u

og
.0 0 5.
1 0 0

4) L1
44 44
- 0.4 L

1.4. U4 +a
o0 c0al
Cu~ -0so
.0 -11 0
V9 v 41
+Cf
cn -j 4

0(
* 0 w 0 L)>
91. * C4.

I 0 0.

sw
Q7 41

U~ 4)0

0 4.1 u~

*~t C - U
0i*

Ci2 C3 zZ
W L Ci1 0 .
 41

co

'a 0

Vu -4
0

29

L- 0-

lug 0
tj 4A

1.44 0M

j 0 ca t

- .3C

4.4 C
i~.

a 104
 (- .J -

= k - c

" Aji

U-4-

(n :)43 -4 cc0
(a
C I-q 04 u z

4- E-1 J. 0 -

z 4 0 W w E
a4 '-T . 0
cc4 CO C'4
E-4 -4.

E-4 '0) C *v4 c) o

a4 = = m. C 00E40-
Ez o- uc C
o
cn-

1050
 -.4

un

~~

Co ;Zo
106 .0 0Qz. z

W 0 w 0 C

CN -44 =. ,-
cn AJ - 0.

-,4 c c 1 -4 .,.
0. 4.
C~ J U4
CUE- C.1 J 0 U- -

-E-4.
B4

CA 2 -4 H) Q M

1062
 O.

Lfn

'.0
C1 w u

W4 IT

00) (0 u

0U =

0- 4-4 4- 4

.0- .14 0 - U) )

107 . --
 -- 7

0))
Co co
41 u.

4.)

0))'r. C Co

04 cc. r .r4~
F4 -4 -

0)
4j~ 0 -
914 4 a.

108 0.-
 14 14 -4 N M C

02A 0 0 w

01 0 0 0 0 0 0 0
W) 04 C4 mo -T C4 CO CO
CO 00

z,. r-4 M -4 '. CTLn

CO e'J--

010
 CU 40
co :-4 w

W 0 00
u -4 0

0.4*4 CLS~
1~ 44 44
0
S0d 0)
U2U 4

39

0--f4

001
 '-44

~x.1

PQ E-4

-4 I

cc44.

4.J 4Q

w~ 41

04 4 H 0

z 0 u Q-
0 m
0~~ '. 0
1-4LI,~ 0 u a.
C%4 -4 j4

w
2 A4
I 4.
I 'b
E--

u E-4 -4
CA L4O4. .4
440U
 '-C4

Co.

ca
41 -

'-4
CCu

z %

-4-
1--4

-4 E-'--

112
 -44

C44

V4~~- enN
o0
N
41I -.4 -r

"4 -4 -H - .

4 -4

*-4
UI0 W L"

-4- -4 -4 *0-
E-40
Nr U
1 it it
od. 0'
0 )44 ~
wD P4J 0 . .

1.4 113 -0 .~
 o4 4 42
-4 c
-4

0)0
004
CL-
o0 Ai
JJ

-4 C

114u
 APPENDIX B

SAMPLE USER'S INPUT

0 o

an~ u

z0

C115

in 0
APJA 1.11.0.

04-W4 C

b C N

UN !> C

1160 LI
 en- 4 C .

~C4 4 -4 a 0C 0

C%4 LI,
C4 04 0 0

~O-4 00 0 0 0 0

117
 C- 0

00

00 0

118
:.'~~ -I- -

C4 C.)
c4

0U

4.V)
o 0 -N

119,
 I
CN C1 0

fi CD- S* S

* ci ' C*
ou-

00 C'Nd 'Z
-4 C1

00

o~~ ~t~ I~

120
 1C 4 4,
5 C.

04 en

C
o
1: 0C ~0 ~ C4

cQ +

121
 C14

-*D 03 i 0

-. ]
C., m ~

CC4
Z-3

I ,.-12

122
 Cl~ C, CA,

c i.
-4

C.)~

1-44

00

12
 I-. - ~ -4 e- 4 w -

1-4 u

- -- - -

4 - I

- en

U12
or 0n 44 LL.~~I

---------

-4

-tL
 APPENDIX C

SAMPLE OUTPUT FROM COPES

£AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA*AAAAAAAAA

0s/16/;O 13.5%..3L
FILE: SAy DATA TI NAVAL bOST:PRADUA7E SCHOOL
I

FCCCCCC, 30OC000 PPPPPPP EEEEEEE JSSSSSS

a u P p E

C
E u
8a p
; FPPPP'p IEE:
E
issssss
S
C 0 u P E S
Ccccccc Occcooo p EEEEEEE SSSSSSS

C UN TR J L P ROG RA M
FO0R
E NG IN E E RING SYN T HE S IS

T IT LE
AIR HEATER DESIGN - OPTI'IZE VOLUME.
I CARC 1.AG4ES C~F CUIljT(UL 9AFA

CARD IMA GE
0
L) AIq HEATER )ESIGN - PTIM1ILE VOLUM4E.
2) 299
2) 2
3) S BLOCK( C
4) Ot 8 0 v 10
41 0 80 0 LO
5) S 3LOCK 0
6) Co
6) 0.
7) 0.
7) 0
81 S 6LOCi% E
9) 9 L6 -1.
II S WI~CK F
11 23292.j 1 5,.6786.68
Ili.2 2.5 .731

126
FILE: Si V DATA TI NAVAL. P]StGKtAUUATE SHCOL

131 .uO?:2 I .t20 .16

14) .i1P. 0211,. J235
141 .uk .0217 .0235
15 • .Jd .125 .08
16)'
16l O. t.o)t
Lo J. 4'. 1.34*
17) 0...., 1.0.4
Is) 0.0 SJG. t,!:)*
18 ) %. 503• 285.
19) 0. 530o. 342.
201
211 S BLOL K G
1,1
21) 1 .
22) 2,a
22) 2 8
23) 3.9 3 9
23)
24) 4,10
24) 4 10
251 5,11
25). 5 L
261 6,1Z
26) 6 12
27) 7,13
271 7 13
28) 8,14
28) 8 14
293 S,15
291 9 15
30) S rLOCK H
31) IL
31) 11
if24 24
33) 1.,0.,2.5
331 L. 0. 2.5
341 25
343 25
351 .Olv.,.18
351 toLd 0. .18
36) 26,27
361 26 27
3f1 - L.04.23J, U., (.
31) 0. 0..O)23
0.
38)
3d1) 19 Ii.
341 - 1- 0+210 O. , • 022
39) -1.J+20 0. .0722
40) 20
401 20 *
41) -1.. ,23,J., • 14

127
FILE: SAV DAA H NAVAL POST4ACUATE SCHOL
41.) -L.0*20 0. 914
42) 30
423 30
43) .Tu.,1.U#Zu
43) .7 0. 1.0.20
44) 3dt38e
44)' 31 38
45) 0.J@J.0 ,L.3 k).
451 J.0 0.0 1.3 0.
#6 1 34933j
461 34 35
473 -L.JW. ,1.
4d) 36
48) 36
49) 2. 90. ,1. 0 7
49) 2. U. 1.0+7
SO) 37
50) 37
513 -L JC2J it..0.
52) 33,33,2
52) 33 33 2
53) 1. t-1 , 1.0+20
53) 1. .1 1.0+20
54) E4 0
1 T ITLE :
AIR HIEATE:% OESIGN OPTIM4IZE VOLUME.

CCNTRCL PARA4FTERS;
CALCJLATICi (J9L NCALC = 2
Nu;46E UF aij3AL JE3 I;N VAR I ABLES9 N~D -
INPUT I%'F1Wl~r[U'l P-i.1r Cwu., IPNPUT = J)
DEBUG PRINT C6UE, IPUBG a 0

CALCULATIC4 CJNTK(,lL, NJCALC

VALUE '4EA,4i 4'
I SIJN.!;LL 4, ALYSIS
2 OPT 1 4 1ZA ItL4.
3 SEN-*S! TI Yi TY
4 Twt-V;,k LLC FXJ4CTIGN- SPACE~
5 OPTIA~ji .N'I~T
6 APPKo0XI4ATE OPTIAIZATION

**OPT1'41ZATION 1N1 0R'4ATIJ)N

GLOeAL VA1AILE vWU:4'ER OF UE'JECTIVF a 16
MULTPLI1l N(.'J~ATIV! INDICATES 91NVIZATION) a -0.OOCE 01

128
 FILE: SAy DATA ri NAVAL P3STURADU4TE- SCHOOL
CCJN#IN PARAMETERS (IF LEAO# CONM[N )EFAJLT WILL OVER-RIJE)
IPRIf41 I TM4AX Aj I A
Ik4 Ni CAL IT R4 L INJ 6J '4AC ' 01 Ni)G
a du j LU a 0 11 1
F 1F DCH.4 CT CTMIN
0.0 0.o 0.0 0.0
Cr1. CTLMIN THETA PHI
0.0 0. 1 0.0 0.0
DELFLN DA3F UN AL PH AX ABOsil
C.0 O.J 0.0 U. 0
DES If'IN VARIABILE [;iF JR 'IAT lo'4
Nol-ZE$K) !1r1IAL V4LJC: wILL 'tofAI~DE 9r1JULL INPUT
0. V. L Ua Er% UIPEt IN I TI AL
NO. r3uUX.) 5BJUNr) VALJE SCALE
I 0.232'))S u'J 3.2325JE 01 Q.67bbJE 00 0.3
.2 O.2503;Oe 00 U.220JOE 01 0.720LJE JO j.) 1
3 0. 62 50u E- iL 1.JI
6jl J.L6OJJE JJ 0. )
4 O1k 3.)J0E-jI J.21 70 J 'JL .2J5.e~ 0.3
5 Q.-30J00E-oQ u.1153UC )3 O:)J- 0.3
6 0.0 0.'.]0u.L OL U. L-..JJZ J1 0.)
7 0.3 Je 430JOF '31 0.104U0E J1 0..)
8 0.0 0 53003Y- 03 J.2'3533E ji O.)

I 10
9 0.0

DESIGN VARIAdLES
0. V.
NJ.
0.5JUUJE 03

GL C6AL
VAK. N U.
4UL TI PLfI 4G
F A:ru.p
0.342%)JE 03 0.3

I I 3.L0JJJE 01
2 2 0 . 100 111E I1
3 3 9 3L 3J CJE 01
4 4 10 O.10003F 01
5 5 11 0,LJJOE QI
6 6 12 %)-10000 11
7 ? 13 a.10jj)UE 01
8 814 0.100002 31

CCNSTRAt"4T 14FORATION
THEAE ARE LL CJ4 TRAI:%T SETS
.LCBA L LLAL L I'mE AR L94ER -4RALI ZAT 1J'1 P
10 VA-k. I VA.-4 e 1D31,11 FAC TJAitf4
1 2'. Z)
3 .1')000F 11 0. 1JuijE 1I o.2*i.K
.5 25 0 3 U.1 3000:-01 U. L i)dJt -J L 2 Jii .
5 26 27 0 -O.1104 Ili J.L1JuYl; to 3.
78 4' 0 0 -5.11 3539 it, 0. I ~ c.T.~2
0 U 0 -0.LIUQ,)E th ,1i.J)J: 1 ~ .r
9 30 0 0 0.7J00(J- 00 J. 7303:3) 011...:
10 3d 38 0 003 J* 1 a J0C 03 0. loioi Il

129
FILE: SV JA TA TI NAVAL POSTC;ADUATE SCrIOfIL
1Z 34 35 0 -0. 10 100~C-I 0.IO001Y7 01 0.1.V11)3' 01
16 36 1 1.)
) Jo Jt: )L J. 2 jU 0- ti U. L ) J~ It
id 37 a 0 -0. 1r~' 16 LI ~
J.1JJ 0.0I
19 33 j. 2 U.LOJOJL 'I L J. louic J 0. 1UJ3I Lk
TCTAL N~U40ZR IF* Cl4.iTAAI?4EU PARA4U ERS L3

**ESTINIATLO JAIA STLRAC;E REQUIRC9ENTS

REAL I NT EE
INPUT EXECtiTLdf AVAILABLE 14PUT EXEILUTI J;J AVAILABLE
106 55JJ *)ci 1 14b 1330
I ~A I -CCOL ED HiE4T EAC HAt4GER 0-) 11 41 AT IC.4

INPUT DATA

TUSESIDE PARAMETERS
MAS FQL kTi 133333.3125 L-34/4IR
OUTLE:T TE.40t.AT'J$.Ez 125.JJJ JEG F
SPECIFIC )-EAT= 1.0J00 o3Tu/Lo4-F

AIRSI!)E PARA:* ETEiRS

MASS FLCW RATE= 1190476.0000 LtIM/I4'l
INLET TejPCP-%TU~t= 95-J,) ) ; F
O'JTLtT TEA*'PArUm L 1 3J.JLJ jEG F
SPECIFIC I#-ts-.f U.2403J hTU/L~M-F
INLET PVAESSUREm 14.J000 PSI

TUdiE :;E0METicf
TUBE lrlJIJE 0IA.2 0.8700 INCHES
T BECUTS)E
jA.=1.363J) 1!4CHES
F NAAH f 0E
J.46.13 IlCtlZS
FIN SPAkI.;, CE'lrZR-Tj-CL?.JrE~m 3. L1110 1N.4CA ES

TUdE ARRAN;ME 4T

130
FILES SAy OATA TL NAVAL PUjTGRAOUATE SCiICUL

Th4ANSWAESF PITC~is 2.1250 IINCIFS

BANK( I E 61- v6.ooo0 LfN.iE-S
BAN4K '41J iH 2da.Ojk0J INCrItES

a I~NTEGER V'ARA ME TEQS

TYPE CF CR.JSS FLOY, %RAAGLMENTx 7
TYPE CF FIN ',%fLLE2
NUMHER OF RJ.S=
NU~V3ER JF PASSES= 2

MISCELLANEOJS VARIAU3ES
GIVEN I-EAT TkANSI- F.1 RATE= W0jo)20)j .3003 ITIJ/tiR
* THEhRIAL CU'J[JLJT I wi r JF T0I3E 'lATERI AL= 220.JO203 TU/-R-FT-F
THERIVAL CC.JULTly OF' FIN :4ATFYIAL= i8.JJUU 6TU/HR-FT-F
GAS CONSTANJT= 53.34JU FT-LBF/LiiN-R

* PENALTY PARAMETERS
2. CCCOC0000 0.0 0.0

OPTIM4IZATION' PARAMrETERS
.00.0 0.0 0.0

*CC a0.100J3F 32 CAULT - 0.200002 01 CCMAX -3.IOOOJE 34

KOINIT
7 I

* * CONI'4N DEtIE C IS I,,:I TIAL XI I ) .3T. V 08 11

X III u U. 2 3:PjE-Ct;IV,-ICi) J.2173E-JI.
XII I IS SET E-jLJAL TJ VJ 1) F(]k I x 4
KCUNTz 1 JELs 0.77434F )I 08Jz 0.75310t J2 3'iJI z J.723C9E; 02
-3.43510E J-J.4 5Th1. L-U ~-5 .0J) UO-0.2355RE JI
-O.471L1F JJ-.1i~ 7L JL:)-37kl x-.1oF A-J..5j47JE JJ
-0.1004,+E U3-J.qi-Ji jO-..16713E 06 J.77PV4E JO
0.0 o.u U.J 0.0 3.0
0I.0 U. % 0.0 000 0.0
0.0 3.3 0.0 0.0 0.0
0.0 0.0 0.U 3.77434E U1

131
FILE: SAy OATA TI NAVAL P3ST AAOUArE S^.HOOL
KOUNTz Z
KOUNT z 2 .)EL-s 0.22126F OL UR1J 0.71452C 32 tM3JI a 0.72922E 02
O.LM17,JL-J'i-J.3 7 72 LJ-.'? 1-L'..1AP .33-i .o:Oi: )lI
-J.4tt47 IJ 3-ij. Lb )1)t:3-~J~b~C3 ~i JL-3 .:, JU. JO
0.0 U.0 0.0 1.0 0.0
0.0 .3.31 0. '0.0 0.3
0.0 0.0 0.0 3.0 0.0
0.0 3.0 0.3 0.55358E 01
KOJNT= 3
KOUNTz 3 r'ELm C.7C33ilE 00 003= 0.711,74E 02 Oi a 0.72243E 02
-0.454 77C JJ-J. 4 id );t -. 36LC-J ~3 JJ-3.J3.-
1bJI
~..4
~-J.['i-il
j j~ i- .3~U~ tJJ1-J.14~-15t 3t-4. 531 5E J3
03.11315E -Jid JO-0.16493E J6-J.733LE-JL
0.0 3.3 0.3 ).3 3.0
0.0 J.)k 0L 0.0 0..0
0.0 0.0 0.0 3.0 0.0
0.0 0.0 0.0 0.48320E 31
KOUNTz 4
KOU-NTz 4 )EL= 0.1 C33UE 31 0,3J= 0.71241E 32 3031 a 0.71464E J2

-0.4543oE C-~uLl )L-J.33j',"E CJ-J.L5U3OE )1-).5JJJ1E J3

-0.IIiC9E J3-J. i*;',93d Jo-0.1b316E 0,5-J.5165LfE-J
0.0 3.3 U.3U 3.0 0.0
0.0 3.0 0.0 0.0 U.3
0.0 0.0 0.0 3. 0 3.0
0.0 0.0 0.0 0.37990E )L
KOUNiTz 5
KOUNTs 5- OELz 0.1C3483E U1 03J= 0.71294E 032 )S331 0.11459E 02
-0.45-951E JJ-0.4132)C OU-U.1&-S4E-C-.9835E J0-'J.JA53E .1
-0.4i436E U3-d).i'~iI.3 JI-0.36-3)cit: 0-.bU3 Jl-'J.t)J303E 33
-0.11 30CE U3-J.9' 18 O-0. 15 39 E J6-0.*502! CE-Q1
0.0 3.0 0.0 :3.3i 0.0
0.0 3.3 0.0 0.0 0.3
0.0 3.0 0.0 :).0 0.0
0.0 3).31 0.3 0.27942E 31
KOUNT= 6
KOUN T b DEL= 0.941263E 00 C3J-- O.'1338E )2 0331 z 0.71448F 02
-0.17171-2-3. 13 J3-U.u-733E '31
-0.49436E 03-3. to 1 it 31-). 3c-4DJ u-). I4,-)9E 31 -J..JOLJE )J
-3.1 L30 L 33U ~3 o-J.4 70164E-Ul
JC3-31
0.0 0.0 3. .0 3.0
1
0.0 0.3 0.0 0.0 3.0
0.0 J.0 0.0 0.0 3.0
0.0 0.0 %).0 0,L8529E UL

132
FILE: SAY DA TA T AVAL. P3STGRAUUATE SCHOOL
XGUNTa 7
KUUtVT:x 7 L)EL= 0. 18313E 01 UBJal 0.71402E! 02 rTiI O. 71445E 0?
-0.25423 it )-.
-- 3 .i-J-)*3 2-.'-u.iI JJ).,sOIJ3E 31

0.0 U.J. O.o J.3 0.0

0.0 0.0 0.0 0.0 0.0
0.0 0.0 0.3 3.0 0.0
0.0 0.J 0.0 0*21066E-01
KOUNT= d
KUUNT= 3 OELz 2).751.7E 01 11-J=' 0.71433E 32 JtSJI 0.71445E 02
-0.45452L- UJ-0.4L411L JJ-J.li60~jE )L
"001JE-1Jd~AE
~0.243c4E-C2-J.41 4.?- E-02-0 .J--2Z iF- 3'7-J.4I 3 4E JJ--) .o51J.3E .)L
-0.49.3uiE ilU* ~i-'i.3,)i
l J) J5-). 14))9f; il-i .5JO0IE JO
0.0 0.0 0.0 2).0 0.0

-0.0 5'~-2O.J 0
J-.0 3~iK~-.0i .,3
U.000
0.0jj~ 0;L
.7 -. 130E0 6 02.

KOUN1T= 90 )EL= C.25450E 01 063j= Oo71530E 32 OBjI 0.71446E 02

-0.454472E JJ-J.41.iJ ii-)lj6I
E.1
-0. ~~ ~ ~ ~
6EO- ~ .e 4~14 7!J-.1)0
i3 J-Jb.f3
U44kk J-o.LItl.2*; oI-Q.3J~k24EF Ji'20l(EJ
-0.1L1JZl0-J.i'o JJ-o.16C)32E 06 0.4125E-3li
0. .0 0003.000
0.0 '3.0 J. 0.0 .
0.0 3.0 0.30 0.0
0.0 0.0 0.0 -0.832E 00
KOUNT= LO
KOtJNT= LO . EL= 0.25459.E 01 0LJ: O.?I53 0 E 22 0001 =0.71599E 02
-0.45271E O J-Jo..~l 9$? 2- ~j~3)l1~J-.110E~-.
-0.49:.d Fk0J-.2-)
* -, L 01-U.3-, 24E 3.-). 149-;)2E 01-U). >JU4E 00.
-3.1L.344E )3U9~LJO-i.
,5220 J .31824E-31
0.0 0.0 0.0) 2.0 0.0
0.0 loo 0.0 0.0 0.0
0.0 J.0 0.3 0.0 0.0
0.0 00 0.0 -0.88302E 01

KOUNT= 1

KON1 )L .. 274E() iJ 0753E 2OB1133~6E0

FILE: SAV JA TA TI NAVAL P4JSTROUATE SCHOIOL
KOUNTs 12
KOUNT= 12 )FL= 0.2593E 01 tJrSJ= 0.71691E 02 )PiJl - 0.71594E 02

-0.4946it J.J-J..4'213-3 1ioLJ1).I,' 0-.£ijjt ii

-3.1L34.,E .3 242JE-01
J653.Ji~-3-.L~1~
0.0 3.0u. j .0 3.0
1.0 3.3 U.0 3.0 U.0
%).0 3.3a 0.0 0.0 J. 0
0.0 0.0 0.0 3.42860E 31
KOUNTx 13
KCIJNrz 13 DEL '31714 1E 01 OJ=J 0.71739E 12 )13JI a 0.7L776E C2
-0.45173k 0.12t 3OVn E33d74 0J-0.L3t-.i5F )1

-0.49.*A-tE )J0 .I'd JI) . -. 3' ' O)*IV 31-3.531 roE )0

0. 0.U 3.0 0.0
00 .3 0.3 0.0 0.0
0.00.0.3.57EU

KCUNT= 14 DEL= O.lc953E 01 Ufl3= 0.71757E 32 0 8J1 .717 75 E CZ

-0.0984 3J-.3oOj 0.0 3.0i 3.0

0..0 0.3 0.0 0.0

0.0 0.3 0.0 3.87260E 00
KGUNT= 15
KCJNT= L5 - EL= 9. 16739E 01 18J= 0.71774E 02 381=0.71774E C2
-0.4517bE 0d)J1'23 0-J.2C5J.7E-3l-).o97J7r- )-U.1660E )L
-0.4-.$.
0.)). t1J~31 -. 3~4 0-3
U * I4)i9 01-u.'.3L37E 33
-0.11372E J3-J.9P)4JL )O-,).1649/E -. 12-0
0.0 3.3 0.3 0.0 0.0
0.0 0.3 0.0 3.0 0.0
0.0 3.u 0.0 3.0 13.0
0.0 0.3 0.0 -3.80129E 03
KUOUN Tz 16
KCUNT= 16 JEL= 9.3 32q5E 01 OGJ= 0.71799E )2 3831 0 .71773E C2
-30.45L7JE U-J.41 12*) 10-0.2u?''bE-1-.39'7HE J3,-u.iocdijE .31
-0.1171L
.jJ-J.94'Iiej JL.-J.ib-t' 3 1 -J.410437U Jl-.wi~
0.0 3.3 0.3 U.0 0.0
0.0 0.3 0.0 0.3 0.0
0.0 0.3 13.U 0.0 3.0
0.0 0.3 0.0 -0.413JSE 01

134
FILE: SAy DATA 11 NAVAL PUST3RADUATE SCHOOL

KQUNT z 17 VEL= 3.2124JE 01 nu9J 0.71,303E 32 iJ1 z .71924F 02

-0.451 34C OU Oi '21S-LJ i~ t13-3 * Ju12E M

0.0 J. 3.3 .0 .
0.0 0.3 0 0.0 .
k.0 3.0 0.0 3.0 3.0
0.0 U.0 0.0 -0.20047E 01
KUUNT- 1d
KOUN7% 18 flEL= .).21353E 01 OBJZ 0.71317E '32 OLIJI a0.71823F 02
-0.451 34E 33-J.41V.I/ JO) 2 ~7o
~13 j-J . lm'i2E 'I
3. 1o.'8u
J -JIJ.I4ii -)i-) ' : J -. 1 L j
-0.4949'2E J).~~LJl--.31'fl7t: J-l. 14101F J-.:)LcE JO
0.0 3.') J'i.1o23 EJ 3.t3 3.0:-J
0.0 0.3 0 3.0 J .0
0.0 0.3 0.0 3.0 3.0
0.0 0.0 3.0 0.12912E JO
KOUNT= 19
Kf)UNTz 19 DEL= ).42-)76F 01 083J= 0.71837E 02 05J1 = 0.71822E 02
-0.451 34E 03-).41'4 6C i.2~ TE1-.897~ 33-i * 13Lj'1,E )1
-0.494.92E JJ0li'tJ-~A~EO3-.i.8%E Jl-3.51llSE 33
0.0 3.3 . 0. U0 a.u
0.0 0.0 0.0 0.0 ).u
0.0 3.3 0. 3 3.3 0.0
0.0 0.3 0.0 0.442668 F 1
KOULN T= 20
K()UNJT 23 DEL= 0.43olIk )I 01!= 0.719266 02 CJBJI 0.71821E 02
:0.45134.E 3J-J.411'6E 30-3. 223 3cE-'JI-U. d9?lju: J3>).l-2 )I
-0.49492E jJ-j.16iJt )13j9TEQJ3b+%JU-).:)ji1E DO
-0.11361iE 3i -J.L55 O-.1 9E Jo ).(,72u4E-J?2
0.0 3.) 0.0 J.0 J3.0
0.0 0.3 0.0 0.0 J.0
0.0 0.0 3.0 0. 1 3.0
0.0 0.0 0.0 0.8727-3E 01
KOUNT= 20 DEL= 1).43011E 01 CUJ= 0.71894E '32 3931
I OPTJ.IATILY) AL-SJLTL

OBJECT IV FJNCT1[;,*
GLOeAL LOCAT UiN 1,i FJ NCTI CN VALUE 0.71521E 02

DESIGN VA4.IAiLES

135
FILE: SAV DATA T1 ~ NAVAL. PJSTG5tADUATE SCHOOL
Do V. GLU13AL L'lvE R UPPER~
to %4c VAR. NJ i,)Jlll) V AWE 114JUNC
I I I 0.2320jE 0:) U.338JJE J0 1.2 325Cf. 31I
2 2 5 0. 2%.)0E 00 J. 5 155qE 30 .). 25J00J' 0
3 3 9 0. 025#X) E-U L 0. 1496 4L a0 0.1IOJ no hI
4 4 10 0. i49 9 9E - 2 u. 2 7 )L -0 1 )..e I r.C-%L
5 5 11 0*794)9E-OL %J.74 y )E-0L 5).U1510 11
* 6 8 12 0.0 0.eS3f3ts -00 J.3,)g9Y ()Lj
7 713 0.0 090. i6 %ouit 0l . 4.1joo% )I
8 8 14 0.0 a*1dbul 30 j3*; 9 Yj,
9 9 15 000 0*236O4E U3 J .49999E J3

DESIGN COASTRAI;NTS
GLOiIAL L OriR U PPE:.-
1D VAR. NO4. B~U)JW0 VALUJE a 1)J1
4u
1 24 0.103U0E 01 1.45L3E 31 3.251CUiF 01
3 25 0.18303_E-01 0.1hil-47-OL J.lIIJJOE 30
5 26 -0. ILJjE Lb -3. LS2 )J J.0
6 27 -0.LLOOJE 16 -O.%J0J !.3
7 19 -0. 1100;)h lb 3.1 1J3 -jL U. UZ2 C;CE-,) I
a 20 - 0. I I )E IS 3.13.15co' 0. L*14Q0C :)U
9 30 0.300OjE 00 3 .9 614 A E i 3 U. L 1)0r it)
10 38 0.u 0. u5b66E )3 Q. I.0IJ,: 01L
12 34 -0. 1330J E 01 a . bi043:E 00 u. I J )) 31-
16 36 0. Z3J) E 0 1 J. 2 2 6. C J3 U. I J) *EJ)
i8 37 -0. 1LOQ E L4o -). 16529E 05 0.0
19 33 0. 1OOE 01 0. 99933~E JJ .
JoLOJOE L16

OUTPUT OATA

TUBESIDE PARAMETERS
MASS FLrW4 RAT--! 133333.3125 Ll,4/HR
INLET T[-.p~jAEz 2,o0.033Y.) EG F
OUTLET T4E,)"ATuKLz 12-'oU3)j JEG F
SPECIFIC I-EATZ 1.0303 BrU/LE54-F

AIRSIOE PARAMETERS

136
FILE: SA'V OATA Ul NAVAL PusrGI(AOUATE SCHOOL

MASS W~Li RATr- 1190476,003J3 L,3%1/HA4

INLET TEPFT,)i 9.i30CJ OE4' F
OVfLIE7T~ ue 1.30.01
t 'lo.hG F
INLET VAESSUREa 1'.e000o PSI

TUBE IGEJMETRY
TLtJE lNS5t9F 0IA.a 0.538d I'CHES
TUBE UL(SIL~t OLA.s 0.5,756 INCHIES
FIN I-EIGT= O.L299 INtii
F IN 7I*~=J.UL201 Ir~iits
FIN SPAt.IN,, CETR-T-CLt4T i= 0.08OdO INCIfES

TWiE AqtRANGEME.4T
TRaANSVERSE PITCA=I 1.022~3 1CCES
LONGITUJIN*L PlITZ~l' 0.6314 INCHES
BANK HL1iHTz 236.U435 141CNFS
B A14K wIJTHt= 13b .0704 1NaCHES
* TllETAs 0.656
OELPHIs 033,

INTEGER~ PAAi4ErERS
TYPE CF CFCSS FLOm ARA.'GE4ENTz 7
TYPE CF F14J PJ*-ILE I
NUMB~ER jF itu,:Sz 4
NU'lTkA OF P-.3sES2 2
VERTICAL RQiWS CO'ITAlN 229 TtWiES

10TSCELLA.'EOUS VARIA~iLES
GIVEN 14E.'& TrtANSFEIA 9A- 10003000.0000) STU/ril
Tt4ERC44t CClJLr~vtTv' CF- TUiE. '4ATL'71%Lz ZZ0.000)0 3TU/4R-FT-F
T*#EP'AL C0t41)iCTLIvTY CrF PIN MAF1A.= I.JU03 3TU/-4R-FT-F
GAS CCNjTAN1TZ 53.3W)t FT-LW/L&-R&

PENALTY PAXA'4ETF' S
IZ.COOOOOOOU 0.000

137
FILE: SAV DATA Ti NdAVAL PJStGRAOUAtE SCHOOL

OPrIMILAFiuN% PA~RAMETERS
0.0 0.0 0.0 000

CTtIER VALUES
HEAT TRX*45FCs4 RArt%
QUOTLz 9,s, iQJ U/HR
31)
QOUT2z ))vi99Si.JJO)j

HEAT TR~ANSFER IF4N~

OELCIx 672J.JJ(.) :ITU/M+
OEL~js 4.Ujuv'
PENALlYz 11.9919
CCNSTPAINTS
0IA'1ETE~ R~ 1.4513
TUBF THICKIEJSSm 0.0184 INCA-.S
TOUCHI'4 -0.146S')
TCUCHL= -0.0)021
PROFH- -7-).U003 I-
PROFC- -3,).Uh)U F
VALUES TO BE OPTIMIZE)
VOL UM E 71.321;. ZJ.FT.
AREA= 10105.421'4 ,:J.FT.
AIR HPz '4.1296 HIP
03P =
AIR PRZ!SSU~z Or% 0.0710 PSI
TUBESII)E PMIS .JE )i~di~ U.13,34 PSI
QSJECTI1L iJN-.TION= 71.8214
IPR~UGRAM CALLS TO ANAtLIZ
ZCALC CALLS
1 1
2 115U
3 1

138
 APPENDIX D

HEDSEJP PROGRAM LISTING

Wf 0..- I
t 0.w &a I-0W

mu.100

U. N -. W

~-U

NU. i- 2

W- -a0 L <w2

NL M40- . z W

-. cZ4 c I.- Ne -
M. 2L)FP. 0 CLL. U P) 4 .u

.L.
4= J 0 Z 0L
.- wZ O 4

--C
-im
4W
T-4W s
winLj
-j" w(LIcc C W0- 4 I.)-U W- z W.
0-2~D -z.U
dN~0WI ~

W a La I. CQ W &4a. tuLP..
w- 4 .
 OOOOOZ4O0O4O0OOOOOOOOOOOOCOOOOOOw OwmOma, OOO 0.

009
i it CC
0U *) fn igw
_j ->L*

C* WIt
ow-
L .1 m I.
co d% c-Wcc

C3 *.-C -4 Ztl
z9.IU ZO-0 V)
L 30 30
10 0 1- * I0 c:WL
I-- zu x-x *f
2 MLL u I..

Ll-1 I-. .-. l0 - ZLI-u

(n 0AC
u (A ) - T 3Z0 f.I--s0
29- z .Jc 40 <. -. 1W Wc U.w

< 0j s- 4 a- CA 4c -0 0.I.
60-j 111 4'-' WI- Ms-Z
0 IO& WM LL >. ~n
U.- %n m x Z WI- v) co - ONV I-- w
xII)1 -IA II ; C3J9W 0.0.J1
w 0
LU = -j LX")c
u i.u- -U m - Ui-o ~ U. <
LflO Z I/S oW. t M l It- w P-I >.I--or
33 ft Z Oe 4C (A~
1--lc - 49 t U.21
0- 4A M CL Z u.'-I- 0 W^ . 4
I-'.j In V) 0A <W 0 %. w - uy L. w a a w
lo20tu .- 0 e ZU. ft. UR5.
OCW x X
=A W -m mb n Mo9.- -4 L .. w 4
. I if-
ILul 1.
b0 3.J1/
- L) C7 LJOIA0U U)M- =&(A
ZIJ- Cow) Wc I.-IA = - -- U .J
- -W/ I- -.
11. UJ "1110 .J L.n < -9-OU. 0 ~ -1 4m3 wm U. 4
4nJ = Z J .U. 01/U U. III<Wl- 9- >. N-.~ M 't
32V) ct 0Z 0~ =- w) U - 0 w MW40.0..M .- I- -
W.J" *d 0. III II OCM 09-2rc -4 N M- 00 aI~9 1
W4 I- Z--0 Z9W4p 0a- 2J< - Zo ~W
LL 0 I.%LZ-D
0
w -~ E > I.- rx q. wa .4- - ~ U -
0-4 w- I a. e CA:. o- -dI9 w III All
I 3.1~~0U 0 0 Lu90 = 0 W-..- -4N f1'

014 0- II ).OOOIA W 1 :14 SU01- ~z w w -W -r-t w

uO M~4-.w 10- w-ZI
9-it 4w
z 101i >0-40 0
C) co4 kn 11--

"d 41 49l- - 9-0./

at CA 0 4 * 9 I 99--9
CL-co1z N = w r1-4 0 .1 0
CL U)
I IU)UU)

140

W.-. - -I
 0 CQ '%70%0000O0000 O"M f (In n VIrlt
f o

IU,. C
a.
ci
I-- I

r4 . ..
U.W
L *G. P--
U.CL 4

.-. z j. . ce M

xw c. 0
S1 .- 0 W-
LL. *
-1 Cc ON

I z: * ZL) )I-
ufl * O< Z<
4~Z p)- ow u 02
2 0 Ii CLW~ Z - 4

1A 2A 01o- w * -J C
CD t) ..J co) *~ WL
=0M
0z Ne >. me I *,0
0
-i g . LU LL c-cu *& Iw CL 0
9L N# x0 LL tui
Z- l-LU c
)ILA U W L*- - flO.-
m~ -C,
f-, CO (A I - x ML. $- _I O-J -

0L ZLo-j I- J -... J..cL

>. -J).J J - x4 0 1^ U.U
-U .j II I M. I. a.L

CY W )-I - e LP$-(5&. I-- - W.j-.

I LU 2 )0 lVI-WILO
a-I*0. a WW10- WU WU--o
00 0Z I-- x Incem Q I-- L) x
.O. 0 1.-~ 32: LLJ~aU4
-4
lop * Z4 -- f-J w wQU.Q
t* 49 49
*i . r- LL 0JLA ft4 t
w w w w w 4x LU w IL z O w.

:6ac I C) *ju 0=%*

IXJL U.

141
 -0-I-4 -

U. * 0 - Z -
In .01 *i I"~Z
w * UOCDZ Z U.- 0
. OWui * W1-.. I-o d4 X IA.
**-IZW
x0 x 1-(d) W. 0.
4a l-OPJ- a Z... XI- U)
0 0(w 1-Z 0 . .J .- W I-2 4
wi 0. a0 ZWJWUJW * 0 LLO wU
0. N * Z - I-WOCO 0. C; I
x -0 0 C4 L I.- mcc 0 5- Z0(A
LU 0A c Z al.- WUI-- 0 4f)(
tn
0. 40i - W4ZCL(, 9 0. -4t OU. C:
0.LU
4 1% In-W Q am0>
. LULL
A
u 49 0 4A m" 4W 0L 0) oL ft <0 0
z LA LU tA U.L MX20 (A - . S- 3CLL cL3 1
wJf 32z4c . UWLULUWL x * S d
W N 0 M. wJ ..--. U - * 2 Z-4 1-x 2
LU.- .... JWiZ <4.J=Z3 x 0 4Z Z.1
UL. V U. Cc UNCM-< x .0 >
LU ft (A.JO. 4%.. 2 a C
* CC WL .j X Xwm4~o 4 0 . LU )V- 4A
* *0 a i 0 J oc CILUb" 0 SI- ~-Leo 0
*LLU t6 .-I W 2-P&C(A - . .4< 20 U.00 x
X (A 3c z LL I1-Wuj2rW 0. a . -1 49 y L -cc l'-
-C M Lu CAULWWW L3 0 co Wil
L(CIA Z
Nm =5 nLU XvL w L20l I
Z L itlA C - L.
1-.Oju cc
LU 0. zw.4w u Z .u
<c 2 I
* ON 1- 4^l-WUJU. 4JL
- z b Um OIL x CL.
*j - .I z. ZLL6>.gc LU < u -.. Z( ....JWW 'A
*U. Cw xz W.LW3 I- "aW4Vfl4 - -. ZCD Z

.01 0OUc~ 4'C^ ... LLCW. IwZ 1 iL-

z G.
= LU -M'ZmI- U-. Ln Ne w I.-0.-OC. LU .Z w
l^ L 0. W-4Z b-u- L16Lw LU 0~ tZ-tiz Mtn U.
Lu
ceW-
W ..d A
V)iOcWU.
LU u) U.I-Lm 1- coW U ) ~L(A- -ILUWW L.
Wu WU..L4 0 m X-04ru. OiI~ 0A
m JO X4* 0
ceJOUM ~ -i Zx- JC.3'C (le:1-
I- 0 m 0 cc i.Je x .J %...-LU U. *.".-
x 1-. Z-30 O0 WZ w. 4- - ZU>-J. 4 4~
cc
W~w X- OWU- 0 lo - - 3c IL wz I (A WC
LU I- ILA
LL. WC Z wu)I-s-.. M On .- wrwj OWL. x
CL Z - 0 Mu.=m M ILL U I-4 - WU o-@ w L)W .j WOL <
2: D . I- 0.- CAXZ44 LU co LU a -WZ~c&. 0-mo
Lu 0 X W J -W. LU Z1--ft.Mac 0- L. U. b-4 -$- uLuWu 0
1-)-C LUA WO > 1-40 ZI- D D- 4A ULW.W)cd WIl-1- C
_j0 c rm -C Wco
4 u - 1.- ce Wie-EC-.- Cc L iL
in Lu LI-4 T' tuAI49 _.j 1- PI .JW LUJ %, 0 7- y(A eWW~ co
LU w *a0.- n1 I-WU4W 11 W-4". b.. 1-- 0 .0 L I.- - OLNV :D
1- O-.. Z -Ccow 3 -Of-utj u 4 0 UJW.7 Q3ZI
0-LUw Uj M=-4 J ILLI. - -w* J L i-LU-40 ZWJO
LU 1-1- W - -1 4 w W U. c 9 .. Jt1- WC MWL
WZuu - I- CC &AU. 0 XaCCULL 0-1 -)) * .-. W4O > LU 0I
o ZX - XC3c * LUZr~0
0
- 4 - * CC:) wc ..J
L3 I-jWOu n *bnr 2"LIOLU 91-0 U)
Q 0. weluJTo 1-n -
flhJoy tU 01,~.- W 0 4 OU. 0 * CU.j>M cc w
0~ el 4 1. 2Wli> 0 4AT-YT7z 0 ... * -4 4W WOWL
* LU jf .1 wb -J"* 0 LU 1-.J.. 1- 0 < * tnxL3..JC 1-tmI- .
* ZWJ.JC. 0- Ox.4-1 *a. C-. * - 0 1-&- .J:) <1-4x ..
o I- CW4c * I..
ld -.. 0 :..JfAm-ULU,4 a H 0 Do .41- C(U..
a 1-zj -a 0U 4w Ut* 9-0 3 0 0)COl-t mul. Lu
* j- * uq.ttuO 0 q uA Cie WJWj

*xu. ~ *U.U) 1-. >i-vflc S 0 .U.U. Uum

142
 cc L
As x 0 LU
j l7- LU 0
0- 1- Z 3 t.
I LUj 0 Y. Lu Lu

IZ LUcc Z W c-
L.. * O) ft W - D

LL* L 4 X
U. ZU at W I- u
Lu .Wu
* * LU 0 <
0 Lu z u
LL I-- Lu LU
:I
1--J 49 (Al
zi -~ Z CC
*ULu 4z 4 < 0M
WU.. -C t- <
U. 0i ul LUJ W a
4j~ s-W~ I. - LUJ

Ox 400.
qc Lu 47

'-. 7 u >- -j 4

0~ t7 L)
WJO 4 Xn 0- 4u
7- u -W x U. M

I.-CC IA 0 $.4n W OU

1.30 GAL *

circl. ccLUZ L UW
O-x X u 0.>-fU0Z ~ U. Z 0 I.- 1- Uu(i''
LU aJ -4 4...
Z. LL WU. 4 i ccin

UW (A 0j .W 4
(^- .0 WE U. 4 < X * * C..
a- 0
-w % U Licc- 0 - i I I N -
az cc - i Lw .-. I.- 4MC0 * 4c""
Zee X: Z I-4 lu J ~ aI U..in W
'L UL 4.-
U ft 1 *% *z ft
WZ =4 W. U. U. W
Zw 43 - fl- Z _j4iZ
-0 w>4V LUC
- N u
OT I- 74 0.0 *9U I
Q:3a UU0 ccQQU UUUUUU Mc 4U UU.J -4 UUU U^ NUU

UW LUI f 4U n - IUd U. &w 4f33-

z" ar 0J.CA LLf. uio j . A t143-U

I- 0 d eUC Z a UX - wub
mui uj~~zc Z -"C NN O
 z z
NO 0 M- W

LU . 0 LU. 0 Z I-
a-~ *K x 0 z wl
0aW 0. 0 Li. (i M 00-
Wwb- at 4 4 f 0 x ZZ
* fLL= 0f V) M 4l u. 4'
Af

Q- a, 1 1^) 410
-do- * ae Z 14 w..
* U424- Z -= -L . x
*u W4N LuWZ% -_ zZ0 C
I-U . "-0~ 4~ Z Z *
@ 4x 0 0 1-- 1 CLZ

CLW. J-AZ
U. W I.- W *>
1.-I- I-QC< M Z LiU M- 0 0 -
ZU <W 1- OKx 1
LUwu W CC~. (A
/ I ew le Wt y
=000 4 0/ ce L) 0) M I-W

- z2 .JWI- L- C 4 _J00
U.0 Lu of- ow t U. Lo' C.L
Lo Q l - - (A) - K LUJ 0 LU * 4.
wwu ( LuwO- Z-J- x L. ce c4 ..
o - o)cca 4 WZN I-LL ZviW 4idi
Q= U. 0I-uJeL)u i Pr%i ILA uCD UZ
L-U. Zu.. %% o- --- 0 4dZ .4 _

K * LLJL 0 0 wI-n3 0 t- w4 _J z Z
.ice 0Oa '-4 -
0~- OZ+ U. 0 Cem LL. U-.
L3 -' - d of W 0- 4LU Lu..
-ft U
W mL3- XL)'- 0 490 LUWO 0 0.
LUW .1 .- 0 In -. r &CC 0 0 i. x
WO """ - ... u Q 04 .W x .-J= .-
-. U Ow UW ..j :IL.- 0 ft (A=U.L .4- - .
moC U=XW co o 4 wu% Lu U.. JI- ft 6- -t L
LU.-. LUJ i-= M m WOU4- cc co cc 0
a x3 .-4
COW 0 b.- - I- LU Ut.)- M M> ~MA 0 Vd)"% -.4
mn A . %- %k > Z N V)3 $-(A I-(A LU zo wr CL
v)M -4 N P"WZ 0 0 0 4cf I/) . . 0 CIO z Z--
b- W&C.~ 0 n InWi Wk m0.4 W0a n W(n qc.j
I- co - LU LU v g'u2M cr * -0. ?-) :.li _j ccLu j
Line -4 01-0 w 4V cn*
At 0-Wui 0-1 1-0 A44 on U.
LU a W4Wn
uI-U M WUZ-
U (-W I- z W_* -
-tw 0-aC( F- I- r Q .4~ W.UZ 0 WL/) Uj 0 - *1
xI LU -4 N UWJ4 - <CC- 1"I CA.u -LUJ O/W U.4 a.
0-
= "0I ) k-5
*..)D 4x 4 - 490 ex 4/)4 of
- 4 cc Ic :: 2 Z ..JU 149 _57 CL .4c ZK U-
4-)
U~ Q cgWitA a -0 M -~~
.JCL M4 -l ..jo < e ~j
.-4 0- 14 II - 1..- 4 4311. LU.. LZ
" = U u. cc L
VU4. - U Z A NbLuN -ji Owi .jo UCC isa it
* L .. N .4L -i= W- < 0 "t .-. U) N
CCU "J' N Ix "I U) .1 - U>-.1 _ -0 1-- 5-
O.-i LP"- =ZZU - *O.JK .Jtl
z I- atJ 0 4o0
.4 O x I.-w-0 !1 e U... l OL W4~ M oe 4LU" WOO0
LUU =
CIE) M W4 Iu.-a uU -Wi.) CQO I-=(O

* §.-LLAW LWJ 3W
(t
30U:N ~ o
w
* 0-WO => C=

U ~ ~U UUIUUUMJU uU.)
uuv QUQQU U.QuuU UUUOtJUU U

144
W000oOO0'00o0o0"00000000 0Oo0Ooo00o-o
-ooPoooe
NN"NNN mNNNNN
mM mcnMMMMM M C ii fr

Z ZL. W

z w -

-z 1-6- z w
49 - C4 I

z :9 Wz 3E
le L0 -
z D o

4A- w 4 i 0
(fl.-ia( ..o x Z
vi wD 0"i <
m wJ4 Z 3
I- 2 0 0r
i w cct 49 0

tA 0. KJ 2
Uj W 40 b- 0
z wL "-Z xK -j
Uj le 040 z2 Z-J LL
-j t- Ko KZ:mQ
- U. _ 9-c I.-
-
tA u
19 0.. - 0. C m.&. e
0. 0 4 J% W - I

a LO 32 LU .%0 E m- T.
44 Wi. 0L i -i c . z
-j-C 3k
xf wL - 0- V)
NI Q I-= 7 A LL WJ

x C4 I-- I.- -- IN M~ :D V -+ = w 0.cf

CA * *
o- 1-K . 0 0 0 I-3- U. 0 4.- 3i = LUZ Nm N LLWj
ce w n- m 0 - m 3.- ! + 4 i -
,W b-CL 0 E3eP(7 CYr-1 4 Iez 4A I " 0 0 -ec
da 14 14. % rCY -d
4w of IA in0 Wt- 0 0 0L4
UN* n -.4 NflM I I i
.J I.- ZXLu. C tau L1J 2i
C. X4 1 1.- N < cy
0 in
%A -* Z U Wj no o wo
4 0 0 I- ZN 2 u .. a.* le CD P c&
U
Q 0 cm0 0 CC W 03 N 0 m %A-3 D4
44 04 0 Y ( N N CK if Kd 7W CC
a V)i% U
a U .S W
U KO 1.- 1 2j
z .-i c
M~ V)0 N K"'ZI- Z 0-. K " xnin x Wh

O Zo O O4949- O - 20. t 0 - c min-1 3 X-

a~ zo
3C L C0- 0-0 1- 099-I

IMLL C3 Z.J 6--1 - x-99

U~ ~ ~ QQ QLUUUUUUUU4UUUQOU
VUU UuqUUU Q QuuU

145
 ce 14
LU I

o Mo
Z x
x T

- ~ 0

OZ Q

w~~u ccO

LU LUJ LU

.4 4cr
-i XW1- 0
- M o. CA

0. ..j

- Z -- U tvU

00 --

. a4
Ma
%-.
z>.z
-- On L
Nc

N -a 0-ZL .*

I.- Z - 0 -1.. JZUJ

X
N Zt Z 6- Z 4i xW co 3c
-oi. L
9LL ow ZO0- x W0. Z LUOx=gL
I2
X- _j 0 0 U = Z-C Q* CL

I1- n I.-01.- LU I- LUJ 14 )1*C

. I-U.4.J SN - - - - X Z ~70 - fA (
ZO3 N4-b M0 w1 LUDI-_ 0"* +
bwt % 8 z: ; I ad ZU - Z *
N 4- Z 4,)4 LUI-M Ul x t
Ic I- ZU i -DU W a %*- u.0 :)4 _j 4 '
ccc Z w4 w-P 4 -0 0 CO)0UZ
M OWN > M I- M
4 .9 02' 0. c i 4
-a Ii 1st 1 11 0n .. 1r- it -4 .4 a-
NN Q-a W nIa X WZ .- 4W I At .
ZI 4 4 4Z 0 4 xv-cDZ% ac Ito u
LL U. Wl- M.
0w-I I -.. Z WD- 9 LU <4 %m
0 0 0 cic Wu4 -i Lw..j ag- O.J-_j Ww W
cec 4- W WU. $A wO 0 XWWO ZO-O0Z W &
0L 0. .J4 4 4 0 0 i- 1-g i.0 Z'-Z 4>- > 4 4 <4 x

Op-a Lu U.S

U UQSU Q U Q Q U w ;. Q t.NQj WQOQWQM Q UQ Q QQ

146
mm MODWooooQ7o (oo%(
9oaoo nooooo0000000

I -3

L4

-y-4

L-Jo
c-a

Lw I "

- W. . .

'A Ci a~ 3il UN
0. m. 0. U -.

*~L 0. 4 ~ci- 10w

S. - .
WI I :2Z2 U,0
ffl~ I w
'. 4 3 -. )--
0 w 0c 0-
C) j 7 z X -4 0> = M- 0
in 0 w. U LL a.
-q I- CL I.. cy c ). I #
~~M a e- . a.>
... + >1 N tn C% 01 N~ei
. 0
9O q9 :m * 4 -4 N0- .4 m0 n
>0 U t- j '.0a -
w hi 0 a. - 00 o -a O - >--.
00 u a.u N a ac wI u ii. 4'-
j a. LL- L- w >j
uiLw )w >
11 a.
.-
.0 I.a.a t- t - a:: 0-
a0- a
*. 4 ac in - i t.-"- 0 0" %- 0 b4" 0 0

LL LL 0 . . a. a . CC 1 39 .3 :1 w 3- 3C 3 : 3 :X 3C M C
w toA ~ a O '~ 0 0O'. ' 0O W
0 i- ., .

Q OUOO ZIuWut jQQ L . 0 Q

NN U UN 0& W W w WW WW14W7
444444444 44 44444444(o(nol a11 1 04441 lr444444*-4
to,*44
- -t 4 -tt -t4, U.t
-V- -'e't- t iZ *t -t -r t

*- %t4 U..Z* *Pef LL-M

0% .0 N 4x x
% WX- 11 @. U.NQ tNU'.

cc.u OtANiZ . A a
z~ ~3u.l- 4j CL 0.-a~
o 0 4 3ULLIw N% -
- .JOLW 0 4 0 W.4fLWL)..t
XUWI -2cDUj Ad*wv~z.-,

t 4mo OXAJ-%f - I

.4W0 V 4-Www"W4CZ

M ~ '4 -u ... WZ
kf 00

W- I-LIV
% 0 * -t-
4~'~
4 0 O tA ~ 0)(HZ=OWx
.J W~ 4IIXto Ou. eb3 *-.4

( ~~. ~ ~. 8~ ~ ~I-C.- ~ X ~9-~~X ~ I ..9

-~~o
- - - In 04LL4
JL 4.qf-( II7

* *UU I.,-0 W%0ZWW%%Z'-

*~4X C C4O4 - .- CI-40 Z-

2 9-Z)-40.-Z Wi-Z9-zm

0 ~
U.. 0U-, 0 -j ul.- %a Z

000- 0U 04LU 0tur

0i 0 . N %v "0 00

0J 0 x ~ O .. t.Mm- - =PI.-)0
.4 - C-

m CO qM~uwX1 -MIM148-
 0 0(%GO(30%7
wwoo O0O000O00OO-4.4-4

O1: cu 4 _t )(%"e
LA it. Nc<fl

-o i-c~ 0 %. 0-0 0w-~.

0 0
x.J 4 4 U. -t u.Uft -U.9-)'x _

Do -. if a' w 4 L % 4

Q_.. W9lJz 4 mCe t0.4 *. -m o i

ft >_7D.4 -4a 7t1.cI-L
Ll.) pw>01- u. r* IZ-W Omf .L '0
O ~'ftw
O ft Wt0. -%t e(cz
=lZI-Oz )( 1+L .O e..Snn
OQQ0L~O tfl -4 L#1
Ox 't.. 0.4u
0. N: Q .'f Zfnlg '%4
. %...~
).LR "aO.j 0 b OU <W w-4X(. *U.=Z .. O
I-.a Z(4tn X cc'%UPt ",.t 64CL
tnZ
W ) c W1
U ~ - *..J .- . 4 Ob
cmj -0co 0. LL tui.. %. iIU.Lg.t t
m fIWZU.% x 4,L 0 00-)~ O.w
U4 .. - L4-=~* c
Uma
X _ .< 1,.- ( q. If o-w-4p-t- iO--a
-~ Ut
M W - -.. tzU. *0 c L
.. tLu,x cc- 4o ZOX * X-tcr=L. saW
%%O".JW0'~t . 9- M.-40 M-4 L3= 40
%-.Cf O.OWtI - 1,- -4 _.Jm-. 11 -UC -51 l4-
8o. <o'. 0,% * J in W -U.. ""- .U.WC -LUWLLO)
4A UtdZx a V r 07 o *- .9-( -W
WLteoUOU. ce 4 i,- ino(-4 ~ m V
m~uIWZ
uowqcW4mr .1I Lu c Coo If4 *Ouif"IEC -. ill :
)9-IJ.(A> -%9- 4
<- .- Z%.cru< 0.%00.9-
wU 4A X(.U -4w L M. I-.Z_- -=
rU.4cnAnU.I x MCW~~h. 4.*
.40C - I cf4- Z C 9-ccm=0 ocr
% O1%d-Z & 0 = . UWX(WT0'.
Wuu.uW -U.1% < - - 4<
Cqx<)Ur-4
CLLOZW I~m m. I:-
-U 4 m
Wf 4rcJ--.= >-4 X >.Oait
.0%tuJ- -
L-. J=f9 .- - Lu CC - 1J4 -X)

mwMUZTP-4 0. OU., . 4V-W)COZtZt- K

= -(a = 4%4 9 = = ,.% I-~ .4-4-q

-4 f N * .. 4- NUNN% =>0-4p-§- O-W 90CL '4

X6_(nwd 0- I x xj.----- - -Z.O- 11 W.i/)4 Lp),*

o.d a emO. 0 *t*t .0.-(JO*<Cyzu.Zwu -4
*%-4* 4 11 cv-4%*0S 9 ftv44ru.jCF vru.
fit * U.J..J .- t-t4t0.4-t.t(x( -.- 4Z..JL ft f6

%2S#%WU ft"U.% % .=U.LLLU.O.-LLLLU. _ _ %. 1-W

1~-*I
%%Wi U
Ifa C0% .. U.A LAV%-t-mM
.%.%.%% NZ.-4U) 0. inCL
.... ... <.JW .j- - .#.
wrI *W%~O k
WIof ~I -W -O( z 0411 -40c

4/9u.4ZZu..-4.0La&. .4LU. U.LLLU.LU.LU..LL-4cc4 ft"NW:: W.

o0 o 1 Ooooooo

149
 #.-

4c I

.- 4

00 .

o =(

.J(3
U- n2

Eo I

-~ 150
 II
cr.

U.

I-

qJIJU a~

Xui151

ce~, ofc
000 00000

00 00

z
LU

I-.

Lu

Lut

I *1

UO

15
wwwwwu

U-

z I-N
0 9 1
1.0

>Jt 0
0- 3c .

x0 fn
z w

'.0U
Lip-

-~ 153
U LU.LLLL.UL LLU. WWULAL

U.

4A4

U-U

zz az
-o) -W

tg- z. m

0W z 4 o

;n- WM -. I-
40% VI -
tmo z
04 0W 0

a cc
L)U
-*C

W W O a L
00
w O no0 ne
no -Owl

U. XLOU. W-0Z
MW 0 -- 3)cI
z Ow~ .z = z
CD W >.I. 0 ft-Z

U.w) ..JC "m. UJ-

U)
W WW 49Zi
I- 4z

40WW

154
 LU

.4.

4
LUZ

z %
I-L
r. a
LU 0 LU

I--
w6z *.CCL

01155
O00000000O0000000000000000
ui~

z I-

L QwXW

-J 00 of

0.. z

I
a.- Njd

L110. W.
* U.C Z. 0. 3

z- Li LLW
CC Wzc ce. n

0.-Z, zL 0 Z N 0
I-. 3-..i Mlqw" C 13. I.*
Oo.JI OI-> r $6 )0W
ZOL VMIOI OLL~wju . C6aw0. V
4 w- -i u..ox 2 +~ i-u

044- 400 ( < IW -4 0 N -I 0

WW. Ul.W MN gU LU.AW 0 %%-0M 0. UJ,
0. lU-. -U) O2Wfta, 0 : so
xm mm- R..W..JI4W LU N -d 1.- 0j
zx W-j~ U. OW Zw0. N O 00. -... J~
Ows.d
0 iIUIVI .0
J~O.JL 0 . 4 x- a. I. ZU.
Z
ft a4
4A0.U Jqtw n I'0 I- N wS0 x
Oc z 4o00 .-. If If zO-4NI0 GOfl-.
Iz Lu V LA.U. 11 1 rlol (IYUL.IL .W - N 0# X . $.0.
- 0. W.-
= v. - ft*
L)D.J O O 0.0 w. m m n L. -i - -
W. 1- z W3 33 3
LvJIUJLUxW U L

- 0~u-W~K 0W ..
J.. Lf0IL" 0 IL j iI
. Li ~I b-~ 0.WWW.JI.- ILUQ 0 *2 - QIi
N -W

Ifl.JL~ S'S0. I~-Ih~ 0 N156N

 uj

w 4

ww

I.--
LII)
M qr

~~~-1
I- U -

-- l:xI- U 3 '0
22 0 U.0 z 0
0 0.-1.- - 0LW
I *0.1~ U.
-L .- 0- 04 cc a " r% m N m t
16. LL - . w - LL
0 0c (3
0% c I-- 1.- 0 LII *4Otfl *ON *.- 4mE *00404
0 1
IL w WO0 W
LU PeIiN 99eN ern
@ *0 OW .
V) @.L 0 _.j M 0L U CLItNII 1 *
*Ill.4.. r" CL
'. N
m. w -t 4 0 IN ).1 i lgt"igg1 11111gw11~.. . . If 1j lil - to
K K ) 0 Z x 0 z
a0 K KK N (.D -NN 4 .. t-
19 N 00 M .
LL Zn W ~ w w ~ U.

U.)

QNU U Q QUU U wu~ui3U4 U)UU u ulnU

157
 OO0OOOOOCO0OOOCOOO0OO00O00OO0VO000~CD

CK -i Mi " 1- 4-'

P- ew -o o 0 tuM4 O0 040-z - m t
MIODI--41NI *10 0 *1 %1A m 4L 00 .NODN("0of *I o 4,*1o 0 M CL*
* ~*vN~ ~ ~ P ON v~ a *N it0
.0 it it " 4. a" *611N. .0 It ItO U

WUUNUWU
WUIW~U .-d~. ZN6Ie SSIHItU . IjNIU
OOOO OOOO OOO 000000%000Cr-0C000000 LOD0000000000000

I- P i l 0o
Go N

CM o~;- 0
*& f a 0%"

NP-MM (0 . m*
~4
%6 If i 4 P N.N U .E t
I- O

00.4-44t-- 0~ 159% 0 *.n.~oc'a

 III~ul
0 0v- %ur 0 C.00 0 c omcnm
.- %0 tI
w--- *0

-
Z-.ma
O- m Ln- waaIamr lot W

-a-aNNNNN N(3JNNMN PJNNNNW-%JN7%0

o16 0
 ii N
LA.I

z SU

a00I II 4Aw

-11m m -
00l- LU LU
U. 4w w 0) 0 o w0 %
'Aj LUI

ooqjW Q u -4 -4 U. 044

161-
 .4.4-4-4-NNNNrN, r~Mt c' M m 44 -4 4 rt

MID

~- Nr 0.

* N.;.I

o ---d W Ui

NN z* W
M- WLfl LW
- _~ U.
4 ~ .L
X0 LWL L... C0
4C NN. Z WX~0
a-- r (LUj. U. 0 :
ZO. 'DenLU~. - 4-W
-i- 01U. U. CL Oct. -0 ). )W
.4L - 0 LWQ U.2:7
M
N .- W* U.0.W.U.0
** C). LLJO 0 N
.. UJJ- L-JAW - Ob0 cc -
-ll-. NN4 I.. I-w ny1.r LU0 r
CL) ** W W 0. Cl m- 0L
>. - m0
z4* 1-W- u *r1- 4 x~ x
X41lx- N(% al-. ll OU I.- I. I. Q - I *

0.0 J 0
ZZ N -4 M 0N
W WCQ0 * * Ill m m 1 mm 0D
14uo Z;J-j <U W -~- -1 I Il-- - *+ *d
i-& L -u'LUW
&11-m m 0 0LL 44 4- U. . CI)m.
1.-0L -- I. e0cc e L. 4d CC 1 ) I It N if 1 R I 7 I1- i
ZIL~4 0-
W:0CO i a-
1--O J*1.-C 4 . it U
Azcz .91 Q tC 0
.)IU. cc L m WLzSc 1. 4 mUU .0 S *U -
S-. -- I- j~ CLcf 0. 4 j I I f I I N N I
~aiu e~O~ I I 0- 0N 1N4It S-4i

~ ~i-J.ZIZ
hI. - 4 I - I - I162 U
 Cie z

a.
.a-

163
CL

V) 1

C:,

0. 0

Qd -
Z f-.

U 0

(x cca

NM 0 -1 .C x
00 0 nmU."b

uw w4 LO

-4. Z P4 - I--
0 -
"14 U -- lo4JUP- .
11- P" U f - J - l U I. It i
=- 4 LU +4 a 0 Luu
u i

OL Z Ul - ,$ n i-1 i LUp 4 N
CC- U5--X X Is I L C r 4 N - - :
:).- 0d
_ D- - e0 . . t

w .j UP

o ~ ~ 4. U-
UQ. UU UU Q:::

0LI-0 2:2:~ u. 164

0

LU
cc
D

2.

z >I
4 (AS

z~ .1- "" IA (I c "c

04 j * L M W 0 _
zk 0 a.N0 c
<W 3w 0 1.=
Mw - 0 -I
0.> 0 V -j :
Z co. .0 NzL
x -
Q.-. P =D- J w X - =0 0 1- c
mv I.- - N* < -
Z 47- 2 e
0 - U) yc W
y 4
u d WU. M W0. *L W
*L WI j 0 4 cc a

W1 .1 lo 2 > x( 040 0 .
m me tip U. P" c T -t
N. C 1 M i
ccZ > < ) jwzz _mz
_0 X.. SW "0 04. W 1
ao ... *n Nz W
I * w < i ac..J
=4 % LL <> o Z ~-4.U U. 3W

LU W. - 0
>4 WL .JO )U
woa~ w L) U ujUQU Qu u 0 Q4v 0N

~
4J)U. 5.. .. W 3 ~ 1065
 -ap
-- d - 4wNldNNNN NN flM-.-440-

ErZZ 2I

o 4o

4 U .4

zf 4U 0 r- L.

U- - o0 %
*~u u * z

X 0 z
u Z* I.
a W Nt LL
0. P4 *I. a I-o
cc Z0. 0e WL 2
Z 0 J.
1- - CO .d
.-e 4. Q02:2 U.
2- .. XN n CA N. -J LJWOM4

WW W. 0 - = I.- - Qe< u
cc cc L. 0- 1-- 0 N40
4w WM m * 2 a aI 0u LLo
* J- MI. Q Z ig0 I.- I.3 .n
.4 =LJ4- Z~ " d4 x0 L.. Z
2i UX W0 0U Z W
M- * .
-if. co )I W x W- z 4 u N
xW
I'- I- - L U. 40 = 0
2-1 N - - * M
z "~ Zi
&~ * M 0 0 U. U. - -U
4 J- CM Z I 4 a (A OU.0.M
Wu Uz WLZ- 0 W 6 ZZN9*d 4U 0- M40 Z.01A Z
z ..-s . U 2L- 4 0 c o 4W Z O - ZJ " wo
4" bU L W . I- x* Z %. ."w W
cc co 11 V) 4A> M
0 WI- w )p. u *
30 4 a 11 CC 0 (AV Z..
- I-- *11
Zi 2w< 0 X Z X - a U. M w w I- *'U 10 z L
cc0.nt X.~ M wX .NX0 0 I- 1-W X 1
2 48..- OM 0 0 ~ I- 24 LZO.. Zs40LUO"L

0 Z z W
tI . ~ Z 4 0I.-a O 0 1
74~~~z cc-4 gUU0
0W il~ z~U .
XZ n 4A4.0
mRc ~ 4 J 04
I-O0 W n Z O O~

166
 M ~ 4,

zM

co i

-j z
t*N -

11-40 1*0 0 l- t-

cr z 11.0
c o
0o
1--
u
0..
o 1 K3
n-W~w
= *0. 4 -* 0.
'INN 5* *JU - *L.
0~e 0- . L x Q 0
t-J.-Lu
0 c% L0 CCU -D
O-o I-- 01S 0-- Z U
go * 0 1-5* *. f * *.
. -
H"40 i-. -' a a u ao~ w. w
.. N -1 *j U.1 * W8 0 * M
W14A z- S6 Sc j .. 0 z Q-'. a z
w 0 0 W"4 S- 0
"q P"

000 - .. t* - A x. " I-
X)0 a '0 CA 7* MZ -j Z
L.W 1-0 ww Wu.M U. .. U.- - * w
I-S MC0 Z - wL i.0. t
x

)-tu .. - -
L4 U W L

--- wo QQ- 4.) - QQ I

QUO - I

N 'l 41
- -. 167
00000czoooooooooooooooooooozooooooeoooooooooooooo

z-J

vix

IL>
lb . 0 -

0 0 U.

.. 0

.)g U. 0
zo *

OLL ..j - 4

4%0. z ;

94A. 'L2 0 L
-4UL. i 4
LU. 0 cc x 05

ce ~~ z .9z Pw z 0
6*. .). z 0 CL 'l z0 _Ii

re WU .OA 0. U. 4 4n
ccco a0 0 - 4
I. -J3 44 .- * 0 z k *0 44
wo $- K
LU '.- (3 u.4 1.-d + X WW 0 L6 > #A
w~ z~fh-L
A~fl 9 0
Z- 66 Z LL w
2 .- CL1f w4Z -
zZ~ ~ ~ 4 W ow .J *L
zo )-LU 4 4 5LU
-" Z-u J ilO < C.) 49 > cc -a
>0. 04 CL u N UW N I- n~ T 4 ...J cc U. * W .
02 WI-. X LULaO 09 0 0-> 0 >
P:) LU -4 IC 0 WZ LU 0 co I.- 4f U .1c -i -'r UJ
LU9-4U C e-ef1- Z mo Z 00 A otwUZ 4 y .
0 a" L M LL. 12 W%6 uu dot w4l L.
09- 1- X LU 1.-0u2 0 cc .- %n Lu 41
40M
cb- 1. 9 Z b . 0 CA, 0 W L

LAU 0. Liz &. N. ZI - 2

Quu wo UUUUQ oitiJ U00 Ouu u. UUULJ u u t U UUL

168
zzzzzzz

LALI

1691
 OoOOOOOxOOOOOOOOOOOOO

U
4 jbuO c

ujOA
MXCI04%
* -el- *

0 LA.

%m .4 Z-' 0 f/)

I--Z-4

cc0~ > &An U.

o. L.Zu'O z *
LUNO * 0. 1^ 0
Q-90 z cc t

I l x0 3 9 -4 n

co U.- - *O~ uj
uow0 0 - 0I- 4 Z*ZN
0OZ4 ft z a *> 46. 1.I- Z
~0 M4
of I 1 I-- 4d III- - U WU z u
L)~
L * U1 In
0~ w I-

w Luocgt
40 >.. 0 00 u f- LLU
ZU. - 0n M . wu w. cc
UM 1 0-0 U' 0.2 C

(aowVhL z 0 > <w 4 L4a&U Z inQ Z

1- *- LU
W Ux
u a -j _j W 0 0 WZ
30L2.J *0 0 w~ &) 0- M Z D 4 U. WW
Zn^ In Q ui

U.AJ3
- 0 X 0 U- I

UUUUIU
U UUU U UUUU U UU UUU 00UU Q-ou

170
 NM~ tklo-G(O rL%0 -0%-N MU Lop-wcy%0 O mtnfl'Or-W% 0-4"mrtm, P-

10 0n tA , -0

LAI )( ~ ~ i U1%0 n

LU o4 -4 *

cc Z U1 In >

0 II) i
CK 0N 0 o (14.
I GoJIn

cz a ~ . I I * .0 T I
LA4*0. %4 0 00 Oem 1I x
-a 0- 0 N 0%% 0 4
.2 CC0 0-4 - 0 0 4 00
(n LU < I- WUO - -'t W-4 0
3 d X _J_ _.. 41 .J-4 _JO 0-
O 0S * + I 0 0 0 0 -
.t~I AO GA- e tAm 0A 0 4* 0~
2U. cc X0 X 0 * 0 * X* X_ 0 GI
LA. 0 -we" 40 0 in~ FP.- .9- 0 *N (V G
(A wU zi-O X1I0 = GO- 0 X. MO 0 C3
x 0 >'A >U Z 00 N >0 >02
44 OX m0-Ila .0 .meC)- I-
aci aWO
004A o f00 0 o0f" 0 1 I 0
> X o.- 22 7x X1r U- ** 21I $~ 0 0 I
j l 4 4 44 4
L-#1. x c -0
' 40-0 o- .2 4.% ex 0-4 . - 0 0
at L. Xt0 '
*> 0 0 O C 0 *a% qx0 90 or- 0
a ro c0~-o IL0inON 0O>
r M000%.
Ow 0U- .0 a 4 04N 0 O> - gX a0Ik- 0-
S0 ce e~ ("U inA - ar VOO 1 % 4
0 ow I Of'- 0 1I 0 0 0 0 O .. ~ * O
CL -9 1- * uJO
U) W4 x i-* I I -m - .
1-WO 0 O W .; J (E (Z0 (3..WL. jey 0
u :3 ac 3z o- s *%A .0 *0 *U% 00 00 @4. 0
Z 0 4t4 0 (A 0 4AX 4AO "n t Ap- 00, (A 0 GA_ 4n
4x ft" X- Xw *X4 X0 0 X I- X0, x X* x a 0 :9
Iv m
tI lb b-t- 0% 4Ws 0-4 In M -0 4 a -a- 4.4- 40 GA 0
aOw CWZ j. >0 >- 0 21 too 02 Z 0 > >X >0 0 Z
cc 0 0. N 40- ac
:)0 0-U. IX IJ-% 'A.0O U.40 J6 0 U U. &U0 U0 Z) U LL U IAt.
GAO 4W - 49 4 b4sm 0 "0 P-O~ -9 .x 0-
O .4 I.- 40 - 4? -

09 0

171
 Ull Ul% C

"0%tN

-l .%C 0 1I- 0%

M (n n- 0 -- M
x 0( A IL 0 g
4 0 N. 7- 0 F_ JN 00
54') I I' -0 -4Q -Jae

o 0 0 0 - usn
P- -r

T) 40 0 At o ..
S~~ft1 * at% U. % 0

4 0 *n 000 3

Io -0 0L > 04. -O .

I~^ 04 - -4- ~ 5' 00

- J Ig, I t, *Oo0-
-.'04 0* 0..
- NO 04
A* V%8 0% 0 00- 0~
CA . Z : +~-
*e
- 0nl 4^0..
'A cc 90 - 0 0,0 0 .- 00 -. 0
N " -V. I
e_% 0 II- Mo~ U. qe 0 Z*M 70*
0: -0 Z 4 _-0 >M 0 >W% t..
z i 01
~0 M 8A 0 (A
It-0 1 01 - If- O-$ O.t%
ON 0 I 40 0% 04n 0 CU 4 0..4 00'.
0%. 03 0 0 -9 4- 0 > 0OM I.).
U.air 0 000 - . O> _j.- _j.0o 01. 4.34 94 . 0
0. .t0= & 0 owl 0= 0 oar,- a0-0 O41 9L
a . W0 -C 54'- n-0 a CO &U 0* S 0-.<
* - L w mqk -d q cn * an * W" M0- tm3
eIft* IA;0- -J 0 'A A-1 I% U' l * * I .- ~
40 ~p*~U -.19 i 40
of% 0 0 .0j 0* U5040 4b- t
10 11. 0 ~0 10 00%o U .0 * *_ I te* ec*
* t'.gA.0 -( 9 *5 4A Z-#I C4 - 01- A *-.'3 0

x* lZa 4 A X- W- X0 X-Z - U- zoh0X X Z*oy I.-

20 41
>0 v 1 44 44 .- 0 401 a' 4)

Oj a' 1w40- x N
x w 0 49l
. -*0 >- c. 49
0- wo *4( W>
0 0 LL 05 L 0 3 PWU I .0
w(1 W0 et' *>_
Z U.S I" )(IU) - - WA W- - 1-- -WP %- ft*O--

0u')v0 W..JNO
V Q I 00U
Q L.QUQhQ*u 000%00-

og.. 0~. . . 0% .0 LU. * IA .1'1720..

 fen 04

x r- e
40'

040.
33~--

o co

N8. to-44 i
t cc Vra
wtf r-44-

G~ 0

{ S 173
cc z

N
t- LL 0

n.- IA 0

* co
LLo
- -4

I 0L
cc .f .4K
U.
z ij(u . *r
o -" .U coa

'~ I--z 4 LU
~ . WWLLf..W
-CID
-' U. L W

U.o .o WU.U.--C4-'
.a 0 U.
0 Z- Q -~

LA 40- W.LL * IDJ 0 a 0

43 j4 ClWL.0-z
u's- N 0 0 u
Zce =Z WC00f.4 &N -. Np M * N
U.N -A OWce.Z*0 4% 0 Z %0-
1.-.d4-. ALuw %&J-w w W-
0Z "I.
m~b QOZn~Z ZXI p- I cc + 4L w4 9-4 -
act wU -" x Z 0 1,.- Z C 0 II 0 41
it - %wo ft a
W4 - WOO..S IWw i -A U $- W 4 IA - N M Mc 49. $- I- =
CC b .UL 4'-Z XU W A e~ P- sa D
# (>.J A II Z
=4 W.- IWU~b--- CU. L.W w w K 't a-OOUZ
A-j 0 0 ow 44cc'1%CD
ow0 .- I I 9L-
L- WWO~Z3b-I
000

174
 .0

+* --

It-

CID
'0- *

O*

Qt *N O
mpl

00Lr -4 CC
AC
L* e .0* tflt

X *%oI 0*0

- z %6- N0 M

-) W U. -N* W
L6 39 - M~ Qc Q*a
.le

Ifl175
110

OCOOOOOOOO O*OO~

- 0*
C* No

0 M0
a A N O
;4 *1-.

.* 0 000

ONN*

u'i
x ~ ~ LAJUN4
W Nl aN
* (
'0 0 0%J 01'Ntt
oL
fn CD* . N m
Wj -04*+Go
f Com 0%ta

%-Z - 00 M :
)( n -4 0-l 11-0
- ft L - +0* ooW w

UW 0 Q'- 0 N u

-I~l0*~ 176
 o1 CID
* 0

WLU
M*

'00
,or- t-
,no

fl. 6 (o

r 0CO ")4
uj *0
; N-h

:l L 4 .N (n eqf

U*
NNX%2

w ~ u se
0-w w
L6
lie Mot O M

QWQ0

%fl-177
 LIM
co 0

* -*-
NQ -I

I-- N.-

N-*

.014

*0l

2p 'm'

FO. -4 -

LL;4-- 7 1

-4 11

-4-'O 0'I

0 f n If
zc x -0 0 -
.41L 041"* c -'
Ac -44. 'N ex' ( L
1-.4

--- 4. 178
oooooooooooooeooo>>>oo>>oo>>oo>oo>>>>>>oooo

W 0

M z +1 -+

I. - O

8j U lfl-t0-
I- z II Lu 0L %%.U -

D LU I- - 0r

j V) 0 4 jL U .
-a
LU U *0 C=-
cu L.j Ir u - UJ -0 O
I.-LU~ I- Wa M-
0W
UNU U.' tno 0- M
.- N.a kmi U. '-0
0O4 4- LU z ~ * L)
kU UL "d LUCI. U O
0- CieQ- a--.
UiN% ...J LL. L 0-
.0. . NLL i" U .J . .I
ou IZ UJ s ZU 0 1 i
in- :UI-- cc Z - +
LLL.
ou- +. LU
=. -

00~ LU LW. 0Ucj 4 ..

LU- OU u + Z4
W - 4 -a
04A LU 0e c - a- 09 4. n
"- -it m" " D " V -4 .x 0

34 b Wd
Q. . LU m L - -i -0 X-*
.. unL 0 W >-
UM- 0- M-'t 0 <

". M.. U4 4x ". a/. LL m.i

- ' -T Z.
1.-A 2c~ 3 I- ZOW U U n a.
~' W~LU U 01.- 0 0-, Z 0- JO Z 0
zz WW * u f0(.
0'0.tC 0 4%
1- 1-I U. W
_j =z U -LU 0 It
*I n W-.
04 O4 _II-I- I- tv~oM 0-D P- + 1
no WI-~ .!t W-U 0 *'to-U Q I1
cz 0 0 ct M ft #3 %A 0a
L LL,- e~ LU -
MWU Z- -. Wl X I.-L 0u QI r - W.~-oI
raw49 -> < I-4 < 4A CAZ 4 t-ow I- -Z -
000 0 ego - ..m.J m = > =
.0 - >- Ul~ 0 WZ
ZI-t u LUn 01- u -4 -
O-OU W LU . LU -
MLU. I-ta3 0 U9 U I.-
UUUU OUW wUw UUUU U U QQW u UUU UU4

179
Z (
o-I zi0
2 0- wLU i LU
;V Z" Li CLz

wJ4 0e M U.1 0. V
x. 0L~
LZ1 0 0
U LL -
9L <0 0 -t 1^
2x U CO czA- r- z ce 0. :3
- 40 0 w 0 N LUJ Ll. 0 1%1
&..- wU M 0. LA- cc 0p
Ci CL- . M u. w
LLU I' 0"*'
cco. . U _j U.I.
Cc *U cc
20 CI-
.9 4 W0-LU G
bwL j . o...J M -4.LL o

oI.
7Fx .1 U) N0C

U. )V-WA LD M LU LJ co Mao - 0 %. 0..

x) W - b. Li U LL. M W.JjC
rOO .c
T 20.1 L -~ 0 k- 1~ 0
o OW OX~ ui 2
z 't It I
Mce.- tLi LU 4A 6 m4 +J 0.* 0 L)
4(0 N IA-L .9o Go I- > LA- -ic Ie o.j .~4
04 -Z a'-... I- U. 0 (L (U0. a1 0 _j
n. -40 4
_j 4 U.. : U)I- c U. u LU

L > N 24 CC0. 4 t %L u* - -T z

'* - _j WOCL(. .1 0 4 .0 .a
I0. 0. W 0Q.C 0.4 1,_ 4 f (4 4
mw < ow l0 I.Ci0 I.I *N .
ULU 4u . *~
0.0M &~Q LJ0W
<w *00)* 9jt 00
4. -j LI'_ I_ Qd) 0. LL 0 &Z1 .0 ) 3
W W 4wJ (4(4W ) - C31. C.. . n
auJ-a 0j
X-~ i, Z0~.400'O cU U' Z0t
0L L
Mtn 00.1.1c o.$ 0 LU '8. Ow 0fU
. *
-C. UO -Z-I (A a- 2 X' x) -- L = ON LUw
_J % -OO
44 *~ d u~L) Wce :lop Oe o * 0a
e-.j 0L 1-Zcd UOLUN o .. I.- "d M2w 4U- L
C. cc <
1'- .. co cO - "0 W- 00 - 1-r'
zu flJ I- 0 0 4WL Lu CL
0 2
u U)i
0000N 40 4U. 11. ft u. 0 C- 4 ..0-U)
a .. 4
uj0, U I'~-. - Wa- (4 Tp co .- U.1 4 cc~ af 0. * I A LU
I.1)00 W O
mull t-A
U LU4 . W
Mj W W LL _I0 ~L
am4 U)U)
m( p~a4 0 < i. nW*X'004 m
CC
J a-. -L U -- J f- UZ -Z -* U.(
0-1 I L LU 0 L
mcc) A-01.1 4 2 > Z
qRL lo 472 z 0 _

1-O-U. ado-$-- 13 1- (40 co 3wo U.

180
 W
0 ~11

0 U
I-.. w
00. W

-Cl
z (-Q
tnM atc

=U * 0i L

z * w

4z- 0 N

4zl- t%- d t

,
,; . 0 ~
0~ aiA0 1. > -

MCC C.. - A I.0

0 - W 0 L1

A IL -4 UJ 'N 0
Zwu*% -8 inz 0 .1-OS
30 0 La Q4U.0 I-
..)- cc In I.. WJ6> W~z
0- 4U W Z=4 Cc
44a t
wwu
AU xzz

181
xxxxxxxxxx(xxxxxxxxxxxxxxx)(X Xxx()(xX

z 0 %

zo W0
=4
L)
t W
* -

cc. -j - (
4 .... - in

.j I W. I. C 4 WL
U. 24 t- z e4
..J U Wn LU 0w C
2
Oix . - U* ZW lc
QW 10 W W.0
4Z

qc: Lu m Q.) Z U
l' CLI I - L
zQ C3 49 - (A x

4 00.
n u p'-W 0.s *0

I- N -.1 -j *q * 4 N -Z

0 Ne ULU -.1 LU ** 4 W j ) 0

L. *42z z 00 x--Z 4 Lu .. 0
C3 W"WO w V) a. tA as.
ON (j ULL 3-. -. 1 LA2in 0
Qe 0*
04 00 C3C
W8 . b"( 4 2 cc .4U m
2u ..j l LW
w = LWU~ Q U. 0 i~ 9.L )9 x 0
C160 U cc - 0w Z -.1LL.z
0 LU
W--CL N j ~W-Z4 49 * L
S-w4 0 W 5 U. "a 4* 00 LU
4 j U. CC 1-
o- W5 W lo- " Wn
tw wnZ L) .;~ 1-2 inL
*ce WIz
AJP- .0 C) w ccW 0 ~ MU. LU

W tf%- CNL-4-w Z
W1 WW *L

am WJU~ 1O-W M8.OW W 6 4 4 LU

Lu 8.) O - in z ~w u - Oj 0-cr n )- + c
'-.Ou.N oil nw >.0 zZ W =U..j 8. I- LUW5
00MC Us. 0 -0 LU )1-)'- .0 1, * U. a Z 0 n
0i s"i V) 4A W WW bpf- .J U.4 Qw W 0 z..j
0Z 0nn 4110 0 0 0 NL .L Wn-Z-4 4 weZ 4
~W wWOw.ad JUJQ ZW U
LLi a l'- Nw lomII

Uj = 4dI- w aQ
W4W Lu LU LU LU Lu j LU
. (% X 4
. 0 0
ccin I- I.- 1.- .4 N- 0 10- -

182
00000

IL-N

.N

wiS J
0 ie4

18
 LIST OF REFERENCES

1. Smith, E.C., "Air-Cooled Heat Exchangers," Chemical

Engineering, pp. 145-150, November 17, 1958.
2. Mott, J.E., Pearson, J.T., and Brock, W.R., "Computerized
Design of a Minimum Cost Heat Exchanger," ASME Paper
72-HT-26, 1972.
3. Nakayama, E.U., Petrol. Refin., p. 109, April, 1959.

4. Bergles, A.E., Blumenkrantz, A.R., and Taborek, J.,

Proceedings of the Fifth International Heat Transfers
Conference, v. II, pp. 239-243, 1974.

5. Fax, D.H., and Mills, R.R., Jr., Transactions of ASME,

v. 79, pp. 653-661, 1957.
6. Shoonman, W., "A.S.M.E. Symposium on Air-Cooled Heat
Exchangers," Seventh National Heat Transfer Conference,
p. 86, 1964.
7. Joyce, T.F., J. Heat. Vent. Engrs., p. 8, April, 1967.

8. Kern, D.Q., Paper presented at the Second National

Heat Transfer Conference, August, 1958.
9. Avriel, M., and Wilde, D.J., Ind. Engng. Chem. Process
Des. & Dev., v. 6, n. 2, pp. 256, 1967.

10. Oshwald, P.F., and Kochenberger, G.A., ASME Paper 72-WA/

HT-15, 1972.
11. Briggs, D.E., and Evans, L.B., 49th AICHE National Meeting,
New Orleans, La., 1963.
12. Briggs, D.E. and Evans, L.B., Chem. Engng. Symp. Ser.
v. 60, n. 23, 1964.
13. Peters, D.L., and Nicole, F.J.L., "Efficient Programming
for Cost-Optimized Heat Exchanger Design," The Chemical
Engineer, n. 258, pp. 98-111, 1972.
14. Palen, J.W., Chain, T.P., and Taborek, J., "Optimization
of Shell-and-Tube Heat Exchangers by Case Study Method,"
Chemical Engineering Progress Symposium Series, v. 70,
n. 138, 1974.
15. Box, M.J., The Computer Journal, v. 8, pp. 303-307, 1964.

184
 IWO-

16. Johnson, C.M., Vanderplaats, G.N., and Marto, P.J.,

"Marine Condenser Design Using Numerical Optimization,"
TRANS. ASME, J. of Mechanical Design, pp. 469-475, July
1980.

17. Afimiwala, K.A., Interactive Computer Methods for Design

Optimization, Ph.D. Thesis, Mech. Engng. Dept., State
University of New York at Buffalo, 1973.

18. Fontein, H.J., and Wassink, J.G., "The Economically

Optimal Design of Heat Exchangers," Engineering and
Process Economics, v. 3, pp. 141-i49, 1978.

19. Nelder, J.A., and Mead, R., Computer Journal, v. 7,

p. 308, 1965.
20. Fontein, H.J., and Wassink, J.G., Verfahrenstechnik,
v. 8, p. 200, 1974.
21. Kays, W.M., and London, A.L., Compact Heat Exchangers,
2nd ed., McGraw-Hill, 1964.

22. NASA Ames Research Center Technical Memorandum NASA

TM X-62,282, CONMIN - A FORTRAN Program for Constrained
Function Minimization User's Manual, by G.N. Vanderplaats,
August 1973.

23. Vanderplaats, G.N., COPES - A User's Manual, prepared

for a graduate course on "Automated Design optimization"
presented at Naval Postgraduate School, Monterey, CA,
1977.
24. Vanderplaats, G.N., Numerical Optimization Techniques For
Engineering Design, presented at graduate course on
"Automated Design Optimizations," Naval Postgraduate
School, Monterey, CA, 1980.

25. Shah, R.K., Afimiwala, K.A., and Mayne, R.W., "Heat Exchanger
Optimization," Proceedings of the Sixth International
Heat Transfer Conference, v. 4, pp. 185-191, 1978.

26. Himmelblau, D.M., Applied Nonlinear Programming, McGraw-

Hill, 1972.
27. Fox, R.L., Optimization Methods for Engineering Design,
Addison-Wesley, 1971.
28. Vanderplaats, G.N., Method of Feasible Directions,
presented at graduate course on "Automated Design
Optimization", Naval Postgraduate School, Monterey,
CA, 1980.

185
29. Vanderplaats, G.N., Automated Design Optimization, class
notes for a graduate course of the same title presented
at Naval Postgraduate School, Monterey, CA, 1980.
30. Johnson, C.M., Marine Steam Condenser Design Using
Numerical Optimiation, M.S. Thesis, Mech. Engng. Dept.,
Naval Postgraduate School, December 1977.
31. Fletcher, R., and Reeves, C.M., "Function Minimization
by Conjugate Directions," British Computer Journal,
v. 7, n. 2, pp. 149-154, 1964.
32. Zoutendijk, G.G., Methods of Feasible Directions,
Elsevier, Amsterdam, 1960.

33. Vanderplaats, G.N., and Moses, F., "Structural Optimization

by Methods of Feasible Directions," Journal of Computers
and Structures, v. 3, pp. 739-755, 1973.
34. Aerodynamic Analysis Requiring Advanced Computers NASA
SP-347 Part II, Application of Numerical Optimization
Techniques to Airfoil Design, by G.N. Vanderplaats, R.N.
Hicks and E.M. Murmaa, pp. 749-768, March 1975.
35. Imai, K., Configuration Optimization of Trusses by the
Multiplier Method, Ph.D. Thesis, University of California
at Los Angeles, June 1978.
36. Bowman, R.A., Mueller, A.C., and Nagle, W.M., "Mean Tem-
perature Difference in Design," TRANS. ASME, v. 62,
p. 283-294, May 1940.

37. Roetzel, W., and Nicole, F.J.L., "Mean Temperature

Difference for Heat Exchanger Design - A General Approxi-
mate Explicit Equation," TRANS. ASME, J. of Heat Transfer,
pp. 5-8, February 1975.
38. Middleton, J.A., "Least-squares Estimation of Non-Linear
Parameters - NLIN," Share Program Library Agency, Program
Order No. 360D-13.2.003.
39. Roetzel, W., "Heat Exchanger Design with Variable Transfer
Coefficients for Crossflow and Mixed Flow Arrangements,"
Int. J. Heat Mass Transfer, v. 17, n. 9, p. 1037-1049,
1974.
40. Roetzel, W., "Berucksichtigung veranderlicher Warme-
ubergangskoeffizienten und Warmekapazitaten bei der
Bemessung von Warmeastaushern," Warme-und Stoffubertragung,
v. 2, n. 3, pp. 163-170, 1969.

186
41. Roetzel, W., "Calculation of Single Phase Pressure Drop
in Heat Exchangers Considering the Change of Fluid
Properties along the Flow Path," Warme-und Stoffibertragung,
v. 6, n. 1, pp. 3-13, 1973.

42. Kern, D.Q., and Kraus, A.D., Extended Surface Heat Transfer,
McGraw-Hill, 1972.
43. Roetzel, W., "Iteration-Free Calculation of Heat Transfer
Coefficients in Heat Exchangers," Chemical Engineering
Journal, v. 13, pp. 233-237, 1977.
44. Holman, J.P., Heat Transfer, 3d ed., McGraw-Hill, 1972.

45. Shah, R.K., "Compact Heat Exchanger Surface Selection

Method," Sixth International Heat Transfer Conference,
pp. 193-199, 1978.
46. Naval Postgraduate School, Monterey, CA, Report No.
NPS-59KK75071, A Method to Predict the Thermal Performance
of Printed Circuit Board Mounted Solid State Devices,
by M.D. Kelleher, pp. 46-49, 31 July 1975.
47. Briggs, D.E., and Young, E.H., "Convection Heat Transfer
and Pressure Drop of Air Flowing Across Triangular Pitch
Banks of Finned Tubes," Chemical Engineering Progress
Symposium Series, v. 59, n. 41, pp. 1-10, 1963.

48. Ward, D.J., and Young, E.H., "Heat Transfer and Pressure
Drop of Air in Forced Convection Across Triangular Pitch
Banks of Finned Tubes," Chemical Engineering Progress
Symposium Series, v. 55, n. 29, pp. 37-44, 1959.

49. ASHRAE Handbook of Fundamentals, pp. 54-59, American

Society of Heating, Refrigerating, and Air-Conditioning
Engineers, 1972.
50. Robinson, K.E., and Briggs, D.E., "Pressure Drop of Air
Flowing Across Triangular Pitch Banks of Finned Tubes,"
Chemical Engineering Progress Symposium Series, v. 62,
n. 64, pp. 177-184, 1966.

187

- . -0,- - - . .----.----. I
 INITIAL DISTRIBUTION LIST

No. Copies

1. Defense Technical Information Center 2

Cameron Station
Alexandria, Virginia 22314

2. Library, Code 0142 2

Naval Postgraduate School
Monterey, California 93940

3. Department Chairman, Code 69 2

Department of Mechanical Engineering
Naval Postgraduate School
Monterey, California 93940

4. Professor M.D. Kelleher, Code 69Kk 1

Department of Mechanical Engineering
Naval Postgraduate School
Monterey, CA 93940
5. Professor G.N. Vanderplaats, Code 69Vd 1
Department of Mechanical Engineering
Naval Postgraduate School
Monterey, CA 93940
6. Professor R.H. Nunn, Code 69Nn
Department of Mechanical Engineering
Naval Postgraduate School
Monterey, CA 93940

7. Lt. C.P. Hedderich

4508 Revere Dr.
Virginia Beach, VA 23451
8. Professor Wilfried Roetzel 1
Institut fur Mechanik und Thermodynamik
Hochschule der Bundeswehr
Holstenhofweg 85
D-2000 Hamburg 70
Federal Republic of Germany

188

- _ .. -, . ..
,r. . . . ..,. ... . . .,-.. ... .. ._ ,,., ,, ..