A STUDY OF THERMOELECTRIC DESIGN
CRITERIA FOR MAKINIZING COOLDOWN SPEED
Td. Hendricks
a. J. Buist
Marlow industries, inc.
1021 So. dup
eer Road
Garland, Texas 75082
(2a) ag4-2s21
Summary
‘The transient cool-down performance was analyzed
for a modified two-stage Marlow Industries Model
NU 2020 thermoelectric (TE) heat punp. A forward time
Gifferencing conputer technique was. imposed on a finite
element thermal model of the thermoelectric heat pump
to calculate the transient behavior of the device.
Analysis of the cosputer simlations was made by con
parison with steady-state optimization criteria. Some
oteworthy conclusions were drawn providing guidelines
For optiatzation of TE heat puros for cool-down speed.
A two-stage thermoelectric design wos fabricated
and its transient performance yas tested to valicate
Computer simulations.
Introduction
‘The major emphasis in the design of themp~
electric cool ing devices has been to accomplish a
(given cooling task while minimizing electrical pover
Snput, This minimization of input power inherently
resules in thermoelectric designs wnich have slow
{thermal response. Consequently, low power coolers for
fast-cool applications require further considerations
for inproving cool-down speed through design rather
than simply increasing copter input. pover.
Therma} Vodet_ Description
The basic twowstage thermoelectric cooter in-
vestigated nas a standard Marlow Industries Model
IE 2080 heat pum “K ohocograph of the MI 2020.45
shown fn Figure 1.
Figure 1
Two-Stage Thermoelectric Heat Pump, Model MI 2020
‘The function of the cooler was to cool a .118" x .118"
ys08" alumina substrate mounted onto the top ceramic.
Aakitional thermal masses on the top ceramic, included
‘etellization, solder and a glass bead type thermistor
ith platinunciriaium Teads.
‘The base of the heat punp was assuned to be
ounces on an infinite heat sink. A vacuum
us
environment was assumed to surround the unit during
operations Passive thermal radiation leading from
the environment, in addition to thermal conduction
Yoading from the substrate and thermistor leads, vas
accounted for. No active thermal loading was present.
A finite elenent thermal model of the two-stage
thermpelectric cooler, the substrate, and the
Surrounding thermal environrent vas generated by con-
puter analysis. The various thermal resistances and
Capacteances ware analyzed and similated in the com
puter thereat model. The transient behavior of the
Bctust thermal model was then approximates by a for-
ward tine differencing technique simul taneous]y
iinposed on each node of the finite elenent, thermal
model. The well known thermoelectric equatfons and
Standard neat transfer techniques were utilized to
predict the transient heat fluxes and temperatures
‘within the actual thermal model.
on Requtvenents
‘The goal of the investigation was to aesign a
fast-cooly two-stage thermoelectric device capable of
Feaching @ cold side temperature, Tg, of =20°C from a
base temperature, Ty, of 65°C in mininun tine after
electrical power application, To achieve this goal
the investigation was focused on the three factors
which have 2 major impact on the transient thermal
Performance, and consequently the design, of any fast~
fool themoslectric. They snclude:
(1) Minimization of the thermal mass of both the
hheat pump components and the cooled thermal load
So as to minimize the amount of stored energy
which mist be extracted for 2 given cooTing
effects
(2) Hinfafzation of a11 thermal resistances connect~
fing the various components throughout the heat
purp, and particularly that connecting the
fooled Toad to the heat pump itself. this will
provide the most efficient neat flow paths From
fhe cooled Toad to the heat sinks
(3) Maximjzation of the tine averaged heat pumping
apactty, Gq» of the thermoelectric during the
cool-down period, represented by
Lt
. & ateet o
where t = time
tg = tine to achieve Final cold
Side temperature
Q(t) * instantaneous heat pumped
at cold side of TE heat
pune.Minimization of thensal mass and thermal resis- free
tances were addressed initially since these are fun
damental to designing fest thermal response in any erumaeenee en Teuen)
thermal systen, Wodifications were made subject to Capacity vo. TIME DURING
practical linftattons, The transient performance of Rane leur ‘Cool-Doww PERIOD
‘the modified MI 2020 is displayed in Figure 2, Its
fEransient performance exhibited significantly faster a
co01-down Speed compared to the conventional designs.
‘Therefore, these modifications were incorporated into
the Final design.
TRANSIENT COOL Dow OF MIZaz0 WITH
ANG WITHOUT TERNAL. HABE. MODIFICATIONS
st 2020 iS
7 Neo sGBtercario%s gy
Wins woorrrcarrons —& oa. octet
*
2
E aes
bo. 2.00
“9 Ficune 4
-0 OPrIMM STEADY STATE |
18 COUPLE RATIONVS. aT |
0 FOR A 1W0 STAGE THERWOELECTATE AEAT PUMP |
Tine «aeons Coan ae
COUPLE RATIO
Heat Pusoing Capacity Hoximizat ion pales coer tee teaaere .
Although design modifications for improving
factors (1) and (2) are intuitively clears the pro- .
cedure for maxinizing G, 18 considerably fore couplex. estates
This is due to the fact that Q,(t) vartes consider-
ably during cootadown. The typica} transient. be-
havior of Gt) is Tiustrated tn Figure "Note that
the shaded region represents che total eneray, Fy
extracted by the cooler top ite top cota surface
diring the tine trerenent te and ts equel to the
integral portion of equation (1}.
The criteria for maximizing heat pumping capacity
for any given steady-state condition are well. known.
The process can be reduced to optimally selecting 13
‘and Rs where Tis the electrical current, A isthe TE
pellet length/eree and-R is. the ratio of thermocouples
prim 1canpeneseH
from each stage to its adjacent, next higher stage, ‘ ae
Both of these peraneters depend only on the boundary
Penperatures of the TE heat pump. They can be selec~
ted to produce maximum heat pumping capacity for any a
given base temperature and oT, temperature differential 2 s = 6
from hace to tone
: DELTA T «DEG 0)
120The problen in maxintzing the dynamic heat pump-
fing capacity in transient design is that aT starts: ae
zero and tnereases to a final at, aTy. The optinun
values for TA and R throughout this excursion are
shown in Figure 4 for "steady-state conditions. The
values at sty, Th, = 23.4 amps/en and Ry * 2.19,
represent the quantities one woutd select to optinize
Steady-state performance for the final iT, Howevere
a significant variance exists throughout the cool=dovn
period and it Ys'not obvious witch 1h and R will
maximize Qe.
1 optintzation
Cateulations of G, were made for various values
of 1. and holding coupte ratfo to Ry. The resuits
are illustrated in Figure 5, It ts interesting to
note that, although the optimum steady-state 1k varied
from zero’to Th, the value that produced the maximun
Us Dips mas approximately 55% higher than 19,. The
fe ys s
symetry of this curve suggests that the penalty for
being above or below Taz 1s equal, thus favoring the
eft side of the peak from power input considerations.
Counle Ratio Ootintzation
Cateutations of Q were made hotding 1. to Diy
and varying R around 85. The results are {Tlustrated
‘in Figure 6, The value that aaxinized Us Rr vas
very close but slightly (eSt) betow R,. ‘Unlike Thy,
the value Ry fell within the steady-state range of
volues iTlustrated fn Figure 4.
FroURE s
TIME AVERAGED WEAT euMpING
CAPACITY VS. NORWALIZED TN
For = e18
Couple ratio is directly assocfated with the
magnitude of'a cooler's ultinate steady-state ats
Thus as Ris decrensed StontTicantly below Rye &
point wiTl, be reached where it fs no longer possible
f0 achieve aty. This essentially 4s why the curve
drops rapidly to the Teft. The fall-off to the right
4s less sevare and is due’prinariiy to the proportion-
aity'smaiter top stage and thus lesser Qo (th.
Experiventel
A special cooler, Nodel $1154, nas des toned
maintaining [1 and Ras close es possible to Lin and
zy respectively. Coapored to the original HI £020,
one thermoelectric couple was added to the top stage
giving the cooler an of 7/3 » 2.33 thereby aporoach=
ing closely the value for Ry.
Test results are_shown with the computer sfnu~
lation in Figure 7. This data {lustrates not only
the progress achieved In overall cool-down speed’ Fron
Figure 2 but also serves to validate ee computer
simutations.
Ficure 6
TIME AVERAGED MEAT PUMPING cAPREETY
WS. NOaMALZe0"crur“= Sarto
FoR ths 96. 2 aneS/2
_ ass
* ase
24 on
7
'
ase an 1
3 as as '
= = 1
Bam ose :
IS ace / _ '
f '
an ' aa !
' :
az : ae :
! '
_ ‘ '
as oe
Dine ae
arConclusions FrouRE 7
Analysis of a small, low-power two-stage thermo- ACTUAL EXPERIVENTAL TRANSIENT PERFORMANCE
electric heat pump vas mide inorder to modify it to Gr SP 1150 VS. COMPUTER SIMLATED TRANSIENT
{improve cool-down speed. The following procedure was PERFORMANCE
derived for re-defining the thermoelectric design
Paraneters derived from Steady-state optimization: a
1. Determine the optimum efficiency, steedy- oo 1S 2 oe Gnes/cm
state design paraneters for the final 21 RoBi
iay'end Rye se
2. Adjust Ry downward by 5%. ey computer
3. Adjust Ta, upward by 558. 30. + lta
‘The above procedure applied to the specific case ze DATA
studied inproved Q- and thus cool-down speed by 25%.
Further study is undervay to test the above procedure
for general applicability as a guideline for caoT=
down speed design optinization.