Version of Record: https://www.sciencedirect.

com/science/article/pii/S0306261918301818
Manuscript_3846814ccc87c140d7d2db6b1668bce8

Experimental Evaluation and Thermodynamic System Modeling of

Thermoelectric Heat Pump Clothes Dryer

Viral K. Patel*, Kyle R. Gluesenkamp, Dakota Goodman, Anthony Gehl

Oak Ridge National Laboratory†

Oak Ridge, TN, USA

patelvk@ornl.gov, gluesenkampk@ornl.gov, goodmandk@ornl.gov, gehlac@ornl.gov

*Corresponding Author

ABSTRACT
Electric clothes dryers consume about 6% of US residential electricity consumption. Using a solid-state

technology without refrigerant, thermoelectric (TE) heat pump dryers have the potential to be more efficient than units

based on electric resistance and less expensive than units based on vapor compression. This paper presents a steady

state TE dryer model, and validates the model against results from an experimental prototype. The system model is

composed of a TE heat pump element model coupled with a psychrometric dryer sub-model. Experimental results had

energy factors (EFs) of up to 2.95 kg of dry cloth per kWh (6.51 lbc/kWh), with a dry time of 159 min. A faster dry

time of 96 min was also achieved at an EF of 2.54 kgc/kWh (5.60 lbc/kWh). The model was able to replicate the

experimental results within 5% of EF and 5% of dry time values. The results are used to identify important parameters

that affect dryer performance, such as relative humidity of air leaving the drum.

†
Notice: This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy.

The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government

retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to

do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research

in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Patel 1

© 2018 published by Elsevier. This manuscript is made available under the Elsevier user license
https://www.elsevier.com/open-access/userlicense/1.0/
 1 INTRODUCTION

Approximately 80% of households in the US have a clothes dryer, and 30% of these dryers are at least 10

years old [1]. Typical dryers use a tumble-type drum with air pushed through by a blower to dry clothes. The state of

the art includes electric resistance (ER) dryers, vapor-compression heat pump clothes dryers (VC-HPCDs) and

condensing dryers. ER dryers use a resistance-heating element to raise the temperature of ambient air, which is then

passed through the dryer drum to collect moisture, before it is vented to the outside. In VC-HPCDs, the heat pump is

a refrigeration cycle that includes a compressor, evaporator, condenser and expansion valve. Air circulates

continuously in a closed loop; it is passed over the evaporator to condense moisture from the humid air leaving the

dryer drum, and then over the condenser to heat up the dried air before it re-enters the drum. Condensing dryers

typically use an air-to-air heat exchanger to dehumidify the air from the dryer. Since some models are ventless,

installation is easier than conventional vented ER dryers. Of all the above, the VC-HPCDs are the most energy-

efficient. Although they are based on mature technology and are used extensively in Australia and Europe, they have

had poor market penetration in the US, with the major barriers being high cost and longer dry times [2]. There is

therefore a significant potential for advanced clothes dryers to provide energy savings over standard ER models [3].

A review of recent research on advances in clothes drying is given here. It includes: modeling and

experimental work on the fabric drying process itself [4, 5], improving the performance of existing ER dryers using

advanced control and termination [6, 7], performance characterization and analysis of VC-HPCDs [8, 9], including

VC-HPCDs that utilize alternative working fluids [10-12], optimization of components in condensing dryers [13, 14],

conceptual dryers that utilize hot-water heat exchangers [15, 16], and early-stage research on the use of TE elements

in a cabinet-type clothes dryer [17]. Based on this review, the literature is focused on improvements to existing

technology, with limited research on the use of TEs for clothes drying applications.

The fundamental process of moisture removal from all kinds of materials has been studied for decades. For

clothes dryers, moisture removal from fabric is of primary interest, as described by Yadav and Moon [4, 5] for

example, who developed a theoretical model that was validated with experimental data from a compact tumble-type

dryer. The analytical model accounted for all the major components of the dryer. Some of the simplifying assumptions

for the complex heat and mass transfer processes that occur during drying were: (1) uniform fabric material properties,

(2) homogenous dispersion of moisture content within the fabric, (3) uniform instantaneous moisture distribution

Patel 2
within working fluid and (4) wet-bulb temperature of working fluid equal to fabric temperature. These assumptions

allowed the transport coefficients to be approximated and the process to be modeled successfully. Various model input

parameters were used to compute the temperature and moisture levels of the air at the drum exit at each time level,

along with the total drying time. These were then used to determine the total energy consumption. Experiments

performed on the instrumented clothes dryer accounted for variation in load size/type, initial/final moisture content,

and ranges for ambient conditions. The modeling and experimental results for variation of fabric moisture content

with time were consistent with well-known trends from the literature. A common basis for comparison of dryer

performance was the specific moisture extraction rate (SMER), which was defined as the amount of moisture removed

from the fabric per unit of total energy consumed during the drying process. Overall, the experimental and numerical

results for SMER were in fair agreement.

In addition to understanding fabric drying, modeling has also been used to investigate the effect of variation

in control strategy on the overall performance of clothes dryers. Ng and Deng [6] developed a new control method by

using a combination of mathematical modeling and experimental validation to determine the equilibrium moisture

content relative to the ambient environment (rather than the drying environment, as is commonly used in traditional

termination control methods). This was then used as the termination point for drying; by accurately predicting the

termination point, the drying time was reduced (by avoiding over-drying) by 13%, resulting in energy savings during

the clothes drying process. Similarly, Stawreberg and Nilsson [7] have shown that there is potential for energy savings

by using a specific control strategy when tumble drying small loads of fabric. They developed a mathematical model

validated by experimental data for various drying loads and reduced air flow, which was used to determine the drying

time and SMER. The model was then used to test two control strategies with the smallest drying load. The first control

strategy involved reducing the heat supply to the dryer (to lower temperature and reduce heat losses) and allowing for

the same drying time as the reference test with the larger load. The second control strategy was to reduce the heat

supply and lower the air flow by 20% (to increase the air residence time in the drum, leading to an increase in the

moisture content at the drum outlet), with the same time constraint as the first strategy. Both strategies had a goal of

increasing the SMER. The results from the model showed that the SMER for drying a small test load could be

improved by 6% when the using a specific control strategy, but the drying time was equal to that of the larger load.

The performance and energy efficiency of VC-HPCDs have been the subject of many previous works,

including that by Ganjehsarabi et al. [8], who conducted an exergy and exergoeconomic analysis of a VC-HPCD using

Patel 3
actual thermodynamic and cost data. Using this method, they could determine the effect of varying the main operating

parameters and their effect on overall exergy efficiency and total exergy destruction of the cycle. They also identified

the components in the system with the highest exergy efficiency and of highest exergoeconomic importance. Other

recent research in advanced VC-HPCDs was conducted by Cao et al. [9]. They developed a two-stage prototype VC-

HPCD and compared its energy consumption and energy factor (EF) to those of a commercially available ER dryer

(from the US market) and hybrid heat pump clothes dryer (from the European market). The prototype utilized three

advanced technologies to improve performance: a two-stage vapor injection refrigeration cycle, compact heat

exchangers with optimized design and a high-efficiency brushless DC fan motor. The vapor injection cycle had a

larger heating capacity than a conventional single-stage system, since more refrigerant was circulated through the

upper stage. The compact heat exchanger capacity and performance were maximized with numerical modeling tools.

The results showed that the prototype dryer could achieve a 59% energy savings and a 143% improvement in EF

compared to the ER dryer. It was also able to achieve a 25% energy savings and 33% improvement in EF compared

to the commercially available heat pump clothes dryer. Although VC-HPCDs are relatively widespread and offer

significant energy-savings, their use of hydrofluorocarbon-based working fluids, which have high global warming

potential (GWP), is cause for concern; recent efforts in VC-HPCD research have therefore considered other low-GWP

working fluids, such as CO2 in transcritical cycles [10-12].

Besides VC-HPCDs, some studies have been performed to understand and improve the efficiency of

condensing dryers. A good example of work in this area is that of Cochrane et al.[13], who proposed that surface

tension elements (STE) be used to replace the air-to-air plate heat exchanger condensing surface in the dryer. The

objective was to use the STE configuration to enhance dehumidification compared to the conventional heat exchanger

condensing surface and reduce overall energy consumption. An analytical model was first developed to predict vapor

removal from the STE. An experimental prototype was then constructed based on the results of the analytical

prediction and implemented into a condensing dryer system. The study showed that in addition to an improvement in

the energy efficiency rating, the dryer also operated at a reduced temperature and required less time to dry a given

load. Cochrane et al. concluded that optimization in the analytical model and incorporation into condensing dryers

would further reduce residential energy consumption.

Do et al. [14] conducted an experimental study to fully characterize the performance of a closed-cycle

condensing dryer with an air-to-air heat exchanger. They evaluated the energy consumption of a closed-loop tumbler

Patel 4
dryer as a function of the electric resistance heater capacity, drying air flow rate, cooling air flow rate, dry time and

water condensation rate. The parametric study showed that greater heater power produced shorter dry times, higher

air temperature and higher water condensation rates. However, the drying air flow and cooling air flow did not

significantly affect the drying performance.

Other research in advanced clothes dryers has been conducted by Bansal et al. in modeling [15] and

experimental development [16] of a novel dryer based on a heat exchanger using hot water (potentially sourced from

process heat of external combined heat-and-power plants) as the medium. For the experimental development, a

conventional once-through dryer was modified by replacing the electric resistance element with a water-to-air finned

tube heat exchanger. The experimental performance of the modified and unmodified dryer was evaluated for the same

operating conditions. Their new dryer concept was found to have shorter dry times (of 15-18 min) with lower moisture

extraction rates for the same total power input, resulting in an 11% improvement in energy efficiency compared to the

conventional unmodified electric resistance dryer.

Unlike the VC-HPCDs mentioned above, TE elements are a purely solid-state heat pump technology

consisting of two distinct semiconductors sandwiched together in a thin layer. When a DC current is applied, a

temperature difference is created between the two sides of the element [18], and the TE can be used as a heat pump.

In the context of clothes dryers, the cold side of the TE elements can be used to condense moisture from the humid air

leaving the dryer drum, and the hot side can be used to re-heat the dried air before it re-enters the dryer drum. This

can lead to the development of a new type of heat pump clothes dryer with fewer moving parts than VC-HPCDs and

without any refrigerant. Liu et al. [17] recently studied the modeling and performance of a closed-loop TE dryer that

was used to dry flax fiber cloth. A one-dimensional mathematical model for the temperature distribution based on

thermal analysis of the TE elements was first presented. An experimental prototype was also developed consisting of

a TE unit (containing 40 TE elements) and a perforated-wall hexagonal drying cabinet. The cloth was suspended on

hangers in the cabinet, and air-flow was directed through the cold side and then the hot side of the TE unit before

passing through the dryer cabinet in a closed air loop. The effects of cabinet air temperature, initial TE input power,

and the initial wet weight of the fabric on the drying rate and SMER were analyzed. The results showed that the drying

rate and SMER were complex functions of the operating parameters and typically increased at the beginning of the

tests before decreasing with time. The initial wet weight of fabric ranged from 2.1 - 4.1 kg, and drying time was on

the order of 4 h.

Patel 5
 Considering the above, the objective of the research in this paper is to evaluate a novel energy-efficient TE

clothes dryer as an alternative to existing electric resistance and vapor-compression heat pump clothes dryers through

a combined modeling and experimental study. Existing literature shows limited work with respect to clothes drying

using TE heat pumps. The results from the current work are valuable for further advancement of energy-efficient

technology, as they include performance characterization, determination of energy efficiency and drying time and

identification of important parameters that affect overall performance. The experimental performance data presented

here demonstrates, for the first time, that TE heat pump clothes dryers can achieve practical results with improved

energy efficiency compared to baseline dryers. The modeling framework presented accurately reflects the measured

performance, and will help to bridge the gap between research, development and implementation.

2 MODEL DEVELOPMENT

The main objectives of the modeling effort were to create a valid model that could be used along with

experimental inputs to (1) accurately predict performance (such as dry time and energy factor) of the TE heat pump

clothes dryer, (2) identify the most important operating parameters which affected performance, and (3) determine the

effect of major system changes (such as air flow configuration). A steady-state, system-level, coupled psychrometric

and TE model was created in Engineering Equation Solver (EES), building upon a previous model [19]. The model

predictions were compared to the experimental measurements (the experimental setup is described in §3) for several

trials. Although both closed-loop and open-loop (or vented) configurations were initially considered, an optimization

study [20] revealed that the open-loop system performed better and was therefore preferred. The dry time of the open-

loop system was found to be significantly shorter than the dry time of the closed-loop system for the same heating

capacity, primarily due to the difference in the moisture removal rate from the system. Instead of the moisture

condensing out (as in the closed-loop system), it was carried out in the vented air in the open-loop system. The

expulsion rate of moisture in the open-loop system was simply greater than the condensing rate in the closed-loop

system. The shorter dry time resulted in lower energy consumption for the open-loop system compared with the

closed-loop system, even with the same average power consumption. Figure 1 shows the process schematic for the

open-loop configuration.

Patel 6
Figure 1. Schematic of TE dryer model showing state points in open-loop configuration (vented). Banks A, B

and C are assemblies of TE elements.

The schematic in Figure 1 shows the most important components in a typical tumbling-type clothes dryer: the rotating

drum, blower, lint filter and ducting. Also shown is the TE heat pump assembly with hot/cold sides, which is unique

to the TE dryer. In a conventional dryer, the electric resistance heating element would be in place of the TE heat pump

assembly (although installation location can vary). In a VC heat pump clothes dryer, the various components of the

heat pump (evaporator, condenser, compressor, expansion valve) would be in place of the TE heat pump assembly.

VC heat pump clothes dryers also generally utilize a closed-loop configuration. The TE heat pump assembly is divided

into banks A, B and C, which are independently powered and controlled. The locations of the state points illustrated

in Figure 1 are given in Table 1.

Patel 7
 Table 1. Locations of state points illustrated in Figure 1

State point Location

[1] Ambient air / blower inlet
[2] Blower outlet / TE hot side inlet
[3] TE hot side, between banks A and B
[4] TE hot side, between banks B and C
[5] TE hot side outlet
[6] Drum inlet
[7] Drum / lint filter outlet
[8] TE cold side, between banks B and C
[9] TE cold side, between banks A and B
[10] TE cold side outlet / exhaust

In addition to the above state points, the hot side heat sinks for TE banks A, B, and C are labelled [3, 4 and

5], respectively, in Figure 1. In the same way, the cold side heat sinks for TE banks A, B, and C are labelled [10, 9

and 8], respectively. Between state points [5] and [6], an empirical heat loss parameter accounts for system energy

losses to satisfy an energy balance in the steady-state model. This heat loss was modeled as being proportional to the

difference in temperature between [6] and ambient, in the amount of 15 W/K.

When the dryer is operating in the open-loop configuration, air from the ambient first enters the blower inlet

at [1]. It flows through the blower and enters the heating side of the TE assembly at state point [2]. Sensible heating

of the air occurs as it passes state points [3] and [4], and it exits the TE assembly at state point [5]. The hot air enters

the drum at state point [6] where it collects moisture from the wet fabric. The humid air leaves the drum and passes

through the lint filter at state point [7] to the inlet of the cooling side of the TE assembly. It flows through the cooling

side of the TE assembly and is dehumidified as it passes state points [8] and [9], and condensation occurs. It then exits

the TE assembly and is exhausted to the ambient at state point [10] (for the experimental trials, the ducting for the

blower inlet and TE assembly exhaust were kept far apart to prevent exhaust air from re-entering the system).

2.1 Model inputs and parameters

The system-level thermodynamic and heat transfer model had inputs and parameters that fell into two

categories: inputs that were varied for each trial and global parameters that applied to all trials, as shown in Table 2.

Parameters were experimentally measured, found from manufacturer data, or explicitly specified as fixed values in

Patel 8
the model during the calibration. The “Fitted” column refers to global parameters that were assumed to be constants,

based on previous experimental observation. The assumed values resulted in consistent model predictions across all

trials. The model inputs that were specific to each test were the starting and final mass ratios of water to cloth

(represented by and , respectively) and the fabric dry mass of cloth ( ). The inputs and parameters specific to

the TE assembly were the current ( ), electric resistance ( ), approach temperature ( , the difference between the

temperature of a given TE heat sink and the temperature of the air flowing over that heat sink), Seebeck coefficient

( ), thermal conductivity ( ) and TE geometry; these values are described further in §2.2. The inputs and parameters

for the various psychrometric state points were the air flow rate, ( ), relative humidity ( ), ambient temperature

and RH (represented by and , respectively) and empirical thermal losses (represented by ,

which was applied proportionally to the temperature difference between the given state point and ambient). These

values are described further in §2.3. The model was divided into three subsections as illustrated in Figure 2. The first

subsection consisted of the model inputs and preliminary calculations of TE properties. The second subsection

consisted of various TE and psychrometric equations, which were solved simultaneously. The third subsection

consisted of the model outputs which included key performance parameters.

Table 2. Various model inputs and parameters and their sources

Fitted Measured
, ,

Inputs varied by Current to TE banks: , ,

(none)
trial ,
Drum rotator motor power, !

Blower motor power, " = $% &

TE electric resistance, = $8Δ :; (Eq. 4)
-7/ = $% -6/, -6/& (Eq. 8)
= 15 */, (Eq. 9)
Approach temperature, = $8= >? =@ ; (Eqs. 5 and 6)
-9/ = 80%
Global parameters Seebeck coefficient, = 0.000419 /, (manufacturer data)
-10/ = 86%
Thermal conductivity, = 1.75 */ ∙ , (manufacturer
4 5 = 1.25
data)
TE geometry (manufacturer data)

Patel 9
 Inputs
Starting moisture content,
Final moisture content,
Mass of dry cloth,
Air flow rate,
Current to TE banks, , ,
Ambient conditions, ,
Drum motor power, !

Blower motor power, " =$

Seebeck coefficient,
Thermal conductivity, TE thermal conductance:
, = $ , C, , Eq. (2)
TE branch length, C
TE branch base area,

Simultaneous equation solver

TE cooling capacity: = = $ , , ,, , Eq. (1)
TE power consumption: DE =$ ,
TE heating capacity: =@ = $ = , DE , Eq. (3)
TE electric resistance: =$ DE,@ , DE, , Eq. (4)
Hot side approach: @ = $ =@ , @ , Eq. (6)
Cold side approach: =$ = , , Eq. (6)
Hot side enthalpy at state point j: F-G/ = $ F G H 1 , =@ , , Eq. (7)
Cold side enthalpy at state point j: F G = $ F G H 1 , = , , Eq. (7)

Outputs
Drum inlet temperature: 6 = $ F 6 , I-6/
Drum outlet RH: 7 =$ -6/, -6/ , Eq. (8)
TE energy: JDE = $ DE , K

Dry time: τ = $ , , , I 6 , I-7/ , Eq. (11)

Energy factor: JM = $ , K, DE , ! , " , Eq. (12)
Spec. moisture extr. rate: NOJ = $ " , K, DE , ! , " , Eq. (13)

Figure 2: Model flow diagram showing each subsection.

Patel 10
2.2 Thermoelectric model
In the TE heat pump assembly, electrical energy was supplied to power the banks of TEs, resulting in heat

transfer between the air and TE heat sinks, and mass transfer due to condensation of water vapor on the cold-side heat

sinks. The equations governing the TE effect and their incorporation into the model are described below. An ideal TE

is shown in Figure 3.

Figure 3. Ideal TE couple with p- and n-branches.

For the above ideal TE with positive (p) and negative (n) branches, the cooling power, = , at steady state is [21]

S
= =% H & H ,8 H Q; H
P Q R
(1)
2

where is the Seebeck coefficient (subscripts are: p, for the positive branch and n, for the negative branch), is the

current flowing through the branches, is the cold-side temperature and @ is the hot-side temperature. The electrical

resistance of the two branches in series, , and thermal conductance of the two branches in parallel, ,, are defined as

CP TP C T P P
= + ,= + (2)
P CP C

where C is the length of a given branch, is its base area, T is the electrical resistivity and is the thermal conductivity.

The heating power, =@ , can be determined from = and the electrical power consumed by the TE, DE , as follows

=@ = = + DE (3)

where DE is the product of voltage and current, or the product of the current squared and resistance, DE = = S

(from Ohm’s law). From the above equations, the heating and cooling power of the TE are a function of applied

current, hot and cold side temperatures, TE material properties (Seebeck coefficients, electrical/thermal

conductivities), and TE geometry (length and area). To incorporate them into the system-level model of the TE clothes

Patel 11
dryer, these quantities were specified as inputs (in the case of current), defined as parameters from manufacturer data

(Seebeck coefficients, thermal conductivity, and geometry) and experimentally measured (electrical resistance, hot

and cold side temperatures, approach temperature), as shown in Table 2.

To establish a relationship between the typical electrical resistance of a TE module and the temperature

difference between its hot and cold sides, ∆ :, experiments were conducted separately from the fabric drying

experiments. The hot and cold side temperatures of select representative TEs in each bank were measured directly

using temperature sensors embedded in the heat sinks (described further in §3). The electrical resistance of the bank

was determined indirectly from the applied voltage and current, using Ohm’s law. Measurements were then made for

multiple trials where the TE temperature difference was varied. The experimentally determined values of electrical

resistance for a given bank were then divided by the total number of TE modules in the bank to calculate the average

resistance of each module (since all modules in a bank were connected in series). These values are plotted as a function

of the average temperature difference across banks A, B, and C in Figure 4. For purposes of the model, a linear fit of

the data was used, resulting in a simple expression which gave electrical resistance for a TE module in a given bank,

, as a function of the temperature difference across that bank, ∆ :. The same expression was used in the model

for all three banks.

Patel 12
 = 0.00368∆ :; + 1.004 (4)

Figure 4. Linear regression of electrical resistance of a TE module.

As with the electrical resistance, a relationship between approach temperatures and =@ or = was also

determined by conducting separate experiments. The approach temperatures are defined as

@ -G/ = DE -G/ H -G/ (j = 3, 4, 5)

(5)
-G/ = -G/ H DE -G/ (j = 10, 9, 8)

where @ -G/ is the hot-side approach temperature, -G/ is the cold-side approach temperature, DE -G/ is the heat

sink base temperature, and -G/ is the air temperature, all at state point G. For a bank of TEs, the experiments involved

a steady increase in current and measurement of hot- and cold-side heat sink temperatures and corresponding air

temperatures. The current, temperature, and TE properties (from manufacturer data) were used to determine =@ and

= with Eq. (3) and Eq. (1). These were plotted as a function of for each TE bank; linear expressions were found

to relate =@ and = with , as illustrated in Figure 5. It should be noted that approach temperatures for banks A and

B on the cold side were based on limited experimental data due to lack of reliable measurements. As a result, the fits

for those banks were linearly extrapolated for use in the model.

Patel 13
 Figure 5. Hot-side (top) and cold-side (bottom) approach temperatures as a function of qH and qC,

respectively.

The experimental results showed that the range of was higher on the hot side than on the cold side. As shown in

Figure 5, the expressions for =@ and = as a function of that were used in the model are as follows.

Patel 14
 =@, : = 138.578 @ -3/;

=@, : = 100.768 @ -4/;

=@, : = 70.058 @ -5/;

(6)
= , : = 173.368 -10/;

= , : = 128.308 -9/;

= , : = 91.108 -8/;

2.3 Psychrometric cycle model

The model utilized psychrometric functions to determine thermodynamic state points, based on

thermophysical properties of an air-water vapor mixture in EES. The air-flow rate at the drum inlet, -6/, was a

model input that was experimentally measured for every trial (the experimental measurement is described further in

§3). The density of air was used along with to determine the mass flow rate of air, . In the TE heat pump, the

enthalpy at a given state point, F-G/, was a function of the enthalpy at the previous state point, F-G H 1/, , and the

heating or cooling power for the bank, as follows

XY-Z/
F-G/ = F-G H 1/ + (j = 3, 4, 5)
[\ -]/
(7)
X^ -Z/
F-G/ = F-G H 1/ H (j = 10, 9, 8)
[\ -]/

The enthalpy was used to determine other relevant state point quantities such as temperature, dewpoint, RH and

humidity ratio, I.

Drum model. The RH of the air exiting the drum ( -7/) was an important quantity that merited prediction with

reasonable accuracy, so that it could be used in the model with different operating parameters. From previous

experiments, it was determined that -7/ was a strong function of -6/ and the air temperature at the drum inlet,

-6/. Using multiple linear regression, a relationship was established between the experimentally measured values of

-7/ (response variable), -6/ (predictor variable 1) and -6/ (predictor variable 2) during five trials, as shown

in Eq. (8) and Figure 6.

Patel 15
 -7/ = H0.00179% -6/& H 0.001738 -6/; + 1.124 (8)

Figure 6. Comparison between linear regression prediction and experimental measurements for RH[7].

Moisture condensation model. According to previous experimental observation, the at the exit of the cold side of

the TE module ( -10/) was fixed in the model at 86%, and the immediately preceding it ( -9/) was fixed at

80%.

Heat losses. Note also that the was set at 15 W/K and was defined as an overall heat conductance term as

follows.

_ = 8 -6/ H ; (9)

2.4 Performance parameters

The most important quantities to characterize clothes dryer performance are the drying time and the energy

efficiency. The energy efficiency can be expressed as the energy factor (EF) or the specific moisture extraction rate

(SMER). These quantities are defined below. To determine the drying time, the steady state model first computed the

steady-state moisture removal rate, ", from the cloth (kg/s) in the drum, as shown in Eq. (10).

" = -6/8I-7/ H I-6/; (10)

where [6] was the air mass flow rate measured at the drum inlet (in kg/s) and I-6/ and I-7/ were the humidity

ratios (in kgw/kgda) at the drum inlet and outlet, respectively. The drying time, K (in s), was calculated as shown in Eq.

(11).

Patel 16
 % H &
K=4 5 ∙ (11)
"

where 4 5 was a correction factor, was the dry mass of cloth (in kg), was the starting mass ratio of water

to cloth (in kgw/kgc), was final mass ratio of water to cloth (in kgw/kgc), and " was calculated using Eq. (10). The

correction factor 4 5 was used because the steady-state model developed in this work did not account for (1)

initial transient heat-up effects or (2) a reduced drying rate typical of the final stages of the fabric drying process. As

such, the model under-predicted the drying time compared to the experimentally measured values. It was found that

using a constant correction factor of 4 5 = 1.25 in the model allowed for consistent and accurate prediction of

the drying time, compared to the experimentally measured values, as shown later in the results section.

The EF was computed as shown in Eq. (12), which is similar to the US DOE definition of energy factor

outlined in 10 CFR 430 [22].

ij
JM = h k
0.535 i*F (12)
` H a ∙ 1.04 ∙ K ∙ b cDE + ! + " f ÷ 3600
de

where , and are as defined above, DE is the total TE power consumption, cde is the conversion efficiency of

the TE power supplies (set to 90%), ! is the drum rotator motor power (measured value), and " is the

blower motor power (measured value). The factor 1.04 is a “field use correction factor,” and the factor 0.535 represents

a typical reduction in moisture content of the test load during a laboratory test cycle. Both factors are defined in 10

CFR 430 [22] and are included here to allow direct comparison with the EF rating of commercial products in the US.

The dryer standby- or off-mode power consumption, which is used to determine the combined energy factor (CEF)

[22], is also not included in the above analysis.

For a more fundamental comparison to other literature, the factors in Eq. (12) can be dropped and the SMER

can be computed as a more general efficiency metric. As previously defined, SMER is the amount of moisture removed

from the fabric per unit of total energy consumed during the drying process, and is computed as shown in Eq. (13).

" ij"
NOJ = h k
K∙b DE
+ + f ÷ 3600 i*F (13)
cde ! "

It should be noted that an important distinction between EF and SMER is that the units of EF are based on the dry

mass of the cloth, whereas the units of SMER are based on mass of water removed from the cloth during drying.

Patel 17
 3 PROTOTYPE TE CLOTHES DRYER EXPERIMENTAL SETUP

The prototype TE clothes dryer is shown in Figure 7; it was a modified donor ER dryer. The heating elements

were removed, and the TE heat pump assembly was installed. All controls and switches on the donor dryer were

deactivated. The location of the TE assembly and ductwork for the open-loop (vented) configuration was used, as

illustrated in Figure 1. Pressure transducers were installed at state points [2, 6, 7, 10] (same as those in Figure 1) to

measure the static pressure in the ducts relative to ambient. A traversing pitot station was installed at state point [6] to

measure the air-flow rate at the drum inlet. T-type thermocouple probes were used to measure air temperature in the

ducts and in between banks in the TE module. Dewpoint sensors were installed at various state points which (along

with corresponding dry bulb air temperatures) allowed the RH to be determined. Compared to the initial prototype in

previous work [19], the final prototype was fabricated with an increased number of TE elements, heat sinks optimized

for high heat transfer/low pressure drop, and high-efficiency components.

Figure 7. CAD model and experimental prototype TE clothes dryer with open-loop configuration (vented).

The maximum systematic uncertainties of measured and derived quantities are given in Table 3.

Patel 18
 Table 3. Uncertainties of measured and derived quantities

Maximum
Quantity Instrument
uncertainty

Temperature Omega TMQSS-062G-6 ± 0.5°C

Type T thermocouple read by NI 9214 module
Static pressure Setra Model 264 Very Low Differential Pressure ± 7.5 Pa
Transducer
Air flow rate Air Monitor Corp. Veltron DPT2500 Transmitter ± 3 % rdg.
with 4” LO-flo Pitot Traverse Station
Dewpoint Vaisala HMT 330 Transmitter with HMT 337
warmed probe for high humidity ± 0.67°C
Measurement temperature
Applied voltage to Sorensen XG150-5.6 DC Programmable Power
Supply ± 2.18 V
TE banks
Current through TE Sorensen XG150-5.6 DC Programmable Power
Supply ± 0.056 A
banks
Mass of test load Sartorius Midrics 1 ±2g
Relative humidity N/A ± 2 % RH
Derived
Humidity ratio N/A ± 0.000606 kg/kg
quantity
TE power N/A ± 17.3 W

3.1 Thermoelectric subassembly

The TE heat pump subassembly for the prototype was made up of numerous “unit engines,” as shown in

Figure 8.

Figure 8. TE unit engine with heat sinks, TE modules, and spacer block.

Patel 19
Each unit engine consisted of four commercially available TE modules, with each module consisting of 51 TE

junctions (or couples) electrically arranged in series and sandwiched between ceramic substrates. The four modules

for each unit engine were clamped between two aluminum extruded fin heat sinks, and an aluminum spacer block

which acted as a standoff. The performance of these modules was measured using a custom evaluation stand. The

design model was used to devise a design consisting of a 6 × 6 array of unit engines (total of 144 individual TE

modules). The TE subassembly was sub-divided into three banks of 48 TE modules each. Each bank had its own

independent DC power supply, and the modules within a bank were connected in series. Each unit engine in the 6 × 6

grid was assembled individually, and all mated surfaces were coated with a heat sink compound to minimize contact

resistances. The unit engines were then installed on a polycarbonate mounting plate and wired according to their bank

and dedicated power supply. With respect to gravity, the hot side of the TEs faced up and the cold side faced down,

allowing gravity drainage of liquid water condensed on the TE cold side. When assembly of the 6 × 6 array was

complete, the mounting plate slid into a clear housing. Thermocouple probes were used to measure the inter-bank air

temperatures, and fine-gauge wire thermocouples were embedded in select heat sinks to measure hot- and cold-side

TE temperatures.

3.2 Experimental procedure

Experimental trials were conducted according to the standard test procedure for residential clothes dryers

outlined in 10 CFR 430 [22]. The standard size load bone dry weight of 3.83 kg (8.45 lb) was first verified using

a scale. A conventional washing machine was then used to uniformly wet the fabric (using the rinse and spin cycles).

The (starting moisture content, calculated using bone dry weight and wet weight) was adjusted until it was at 57.5

± 0.33%. The wet fabric was then loaded in the prototype TE dryer to begin the experiment. The drum rotator motor

and blower were started, and the TE bank power supplies were activated. All measurements were recorded at a rate

of 1 sample/sec. The load was removed from the dryer and weighed periodically, before resuming the trial. The

experiment was continued until the relative moisture content was less than 4%.

Patel 20
 4 RESULTS AND DISCUSSION

Several experimental trials were conducted on the prototype TE dryer; the main variations from one trial to

another were the electrical current applied to the TE banks and air flow rate. The measured values that were used as

model inputs for five trials (air-flow rate and current were averaged over a given trial) are shown in Table 4. The

model inputs from Table 4 were used along with the expressions in §2 (Eqs. (1-13)) to determine the relevant quantities

in the model. The expression for the blower motor power, " , was based on a curve fit between experimentally

measured air-flow rate and power consumption for the open-loop system and was found from previous experiments.

Table 4. Experimentally measured values used as model inputs for each trial

Experimentally measured value

Parameter
Trial 1 Trial 2 Trial 3 Trial 4 Trial 5
– starting moisture content [kgw/kgc] 57.38% 57.32% 57.37% 57.53% 57.50%
– ending moisture content [kgw/kgc] 2.15% 3.15% 3.67% 3.99% 3.30%
[kg] ([lb]) 3.83 (8.45)
-6/ [L/s] ([CFM]) 64 (135) 58 (123) 65 (138) 66 (140) 67 (141)
[A] 1.23 1.18 1.40 1.40 2.00
Electrical currents for TE banks [A] 1.48 1.24 1.90 1.90 2.00
[A] 1.81 1.31 2.40 2.40 2.00
[°C] 22.8 22.9 22.7 22.7 23.5
[%] 43.2 58.0 56.3 49.5 44.0
Drum rotator motor power [W] 240 97.5 95.9 95.6 97.6
Blower motor power [W]
S
" = 0.0107% -6/& H 0.7845% -6/& 114 90.6 121 125 127
+ 25.194

The power consumption for the drum rotator motor in Trial 1 was high compared to other trials because the

built-in motor was being used at an inefficient operating point (this motor normally runs both the blower and drum

rotation; however, a separate blower was used in the prototype system, leading to a very low motor load, and low

efficiency). For Trials 2-5, a right-sized drum rotator motor was used instead, resulting in a much lower power

consumption for the same performance. Sample experimental data for Trial 5 are given in Figure 9, Figure 10, and

Figure 11, including temperatures, dewpoints, and RH at various state points, as well as power consumption of the

Patel 21
major dryer components with time. The TE power consumption was constant because a fixed current of 2 A was

applied to all TE banks.

Figure 9. Experimental data for Trial 5: temperatures and dewpoints at various state points including

ambient, drum inlet (T[6], D[6]), and drum outlet (T[7], D[7]).

Figure 10. Experimental data for Trial 5: RH at ambient, drum inlet (RH[6]), and drum outlet (RH[7]).

Patel 22
 Figure 11. Experimental data for Trial 5: total power consumed by TE power supplies, blower motor, and

drum rotator motor.

Figure 9 shows that the drum inlet temperature, -6/, reached an average value of 37.1°C for trial 5, which

was similar to all other trials for the TE dryer. These values are considerably lower than typical drum inlet air

temperatures for conventional ER dryers (~52–60°C). This lower temperature affects the ability of the incoming air

to hold moisture, since it is proportional to temperature. As such, having a lower temperature resulted an increase in

the drying time for the TE dryer compared to a conventional ER dryer. The initial and final transient periods of the

drying process are also shown in Figure 9, indicated by gradual increases or decreases in temperature and dewpoint.

During the initial heat-up period (which was a small fraction of the overall drying time), insignificant drying occurred

because the input energy was used to heat up the components of the system. As drying proceeded, a transition to

constant-rate drying (which is well-known in the literature [4]) occurred. Most of the process parameters remained

relatively constant, as did the rate of evaporation. Towards the end of drying, a transition to falling-rate drying

occurred, characterized by a decreasing rate of evaporation. Most of the input energy was used to heat the fabric and

drum, leading to an increase in drum outlet temperature, -7/, and decrease in drum outlet dewpoint, l-7/.

Figure 10 shows that the drum outlet RH, -7/, had an average value of 79.1%, meaning that the air was

effectively able to collect moisture from the fabric as it passed through the drum. This value decreased significantly

towards the end of the drying process, as it became less effective. Figure 11 shows the power input to the TE heat

Patel 23
pump, which was considerably lower (~0.7 kW) than the typical power input to the heating element in a conventional

ER dryer (~ 3-4 kW). As a result, the overall energy consumption was reduced significantly.

The experimentally measured results used to validate the model for each trial are given in Table 5. They

include the air temperature at the drum inlet ( -6/), RH at the drum outlet ( -7/), the total energy consumption of

the TE heat pump, the total energy consumption of the dryer, the total drying time, the EF and SMER. The air

temperature and RH values were averaged for the entire trial. As previously mentioned, this was an approximation,

since it did not account for the initial and final transient (i.e. non-steady) periods of drying.

Table 5. Measured results used to validate model for each trial

Experimentally measured value

Parameter
Trial 1 Trial 2 Trial 3 Trial 4 Trial 5
-6/ [°C] 31.1 30.9 35.9 35.7 37.1
-7/ [%] 75.7 84.6 83.4 81.4 79.1
TE energy [kWh] 0.769 0.721 1.131 1.092 1.092
Total energy [kWh] 1.509 1.257 1.544 1.476 1.466
Dry time [min] 117.3 158.7 107.4 100.1 96.1
EF [kgc/kWh] ([lbc/kWh]) 2.50 (5.52) 2.95 (6.51) 2.38 (5.26) 2.49 (5.48) 2.54 (5.60)
SMER [kgw/kWh] 1.39 1.64 1.33 1.38 1.41

By examining the experimental data in Figure 10, Figure 11, and Figure 12, we see that accurate prediction of

important quantities such as the drum inlet temperature and drum outlet RH is essential. These values directly affect

the drying rate and, ultimately, the performance and energy efficiency of the clothes dryer. To illustrate how the

important quantities changed while progressing through the various points in the cycle, all the model-predicted state

points for Trial 5 (as an example case) are given in Table 6, As above, the locations of all state points correspond to

those in Figure 1 and Table 1.

Patel 24
 Table 6. All model-predicted state points for Trial 5

State point qH [kW] qC [kW] ATH [K] ATC [K] h [kJ/kg] Tair [°C] RH [%] ω [kgw/kgda]
[1] - - - - 43.77 23.50 0.4403 0.007929
[2] - - - - 43.77 23.50 0.4403 0.007929
[3] 0.4898 - 3.534 - 49.98 29.59 0.3075 0.007929
[4] 0.4473 - 4.439 - 55.66 35.16 0.2246 0.007929
[5] 0.3744 - 5.344 - 60.41 39.81 0.1745 0.007929
[6] - - - - 57.79 37.25 0.2003 0.007929
[7] - - - - 57.79 22.53 0.8072 0.01383
[8] - 0.1640 - 1.800 55.71 20.51 0.9134 0.01383
[9] - 0.2411 - 1.879 52.65 20.97 0.8 0.01244
[10] - 0.2910 - 1.678 48.96 18.97 0.86 0.0118

Note that the heating/cooling capacities and approach temperatures are only given at the state points that

correspond to their respective banks in the TE heat pump, as they are not defined elsewhere in the cycle. As mentioned

above, for this steady-state model, the predicted quantities in Table 6 do not represent any particular point in time.

The data in Table 6 show that the model was able to provide detailed information for each state point in the cycle and

could be used to gain insights into the effects of various parameters on the performance and energy-efficiency of the

TE clothes dryer, without having to perform many experiments. The comparison between experimental and model-

predicted drum inlet temperature and drum outlet RH for all five trials is shown in Figure 12.

Patel 25
 Figure 12. Comparison between experimental and model-predicted drum inlet temperature (left) and drum

outlet RH (right) for five trials.

The modeling and experimental results revealed that the TE/heat sink and air temperatures played an

important role in their efficiency, since the TE coefficient of performance (COP) worsens at higher temperature lifts

[21]. For the drum outlet RH, the best results were expected from high RH leaving the drum because it meant that (1)

the vaporization of the moisture on the cloth was maximized for a given temperature and (2) less heat needed to be

removed from the air before the onset of condensation. As illustrated in Figure 12, the model showed a maximum

deviation in drum inlet temperature of within ± 2°C and could accurately predict the RH of the air exiting the drum

by ± 10%.

The comparison between experimental and modeled values of total TE energy consumption (in kWh) and

dry time (in min) for all five trials are shown graphically in Figure 13.

Patel 26
Figure 13. Comparison between experimental and model-predicted TE energy use (left) and dry time (right)

for five trials.

The comparison between experimental and model-predicted TE energy use is an important result because

unlike other energy usage in the system which does not vary greatly (i.e. blower and drum motor energy usage), the

TE energy consumption depends on many factors such as the current and temperature difference (as shown in Eqs. (1-

6)). During a given trial, the temperatures on the hot and cold sides of the TE change with time as the fabric is dried,

resulting in a change in the heating/cooling COP and capacities. The coupled nature of the model makes it versatile

enough to account for these various inter-related quantities. The dry time is also an important metric; it is closely tied

to the energy consumption, and accurate prediction via modeling is essential. The results show that model prediction

of drying time was within ± 5% of the experimentally measured one.

The final comparison between experimental and model-predicted values was for the energy efficiency of the

TE dryer, which was characterized by the EF and SMER using Eqs. (12) and (13). As indicated in the denominators

of Eqs. (12) and (13), the total energy consumption included the AC electrical energy consumption of the TE power

supply, the blower, and the drum motor. The trials conducted so far reached an EF of 2.95 kgc/kWh (6.51 lbc/kWh)

and SMER of 1.64 kgw/kWh and the maximum deviation between model and experiment was < 5% for all trials. The

differences between model prediction and measured experimental results for the key state points in the system are

summarized in Table 7. The + and - signs indicate quantities that are over-predicted and under-predicted by the model,

respectively.

Patel 27
 Table 7. Differences between model prediction and experimental measurement for key parameters in each

trial

Trial 1 Trial 2 Trial 3 Trial 4 Trial 5

-6/ [°C] +1.46°C +0.42°C +0.60°C +0.39°C +0.18°C
-7/ [%] +6.9% +0.4% -2.0% -0.3% +1.6%
TE energy [kWh] +2.0% -2.4% +1.9% -0.6% -2.9%
Total energy
-1.4% -3.3% -0.4% -1.5% -3.9%
[kWh]
Dry time [min] +1.5% +2.9% -0.7% +0.1% -3.2%
EF [kgc/kWh]
+2.2% +3.9% +0.8% +2.0% +4.4%
([lbc/kWh])
SMER [kgw/kWh] +2.2% +3.9% +0.8% +2.0% +4.4%

To put the above energy-efficiency results into context, the current minimum CEF requirement for a standard

size, vented, electric clothes dryer in the U.S. is 3.73 lbc/kWh for an 8.45 lb load of fabric [23]. The results demonstrate

that the EF of the TE heat pump clothes dryer (6.51 lbc/kWh) can be up to 74.5% greater than a baseline ER dryer

which meets the minimum energy efficiency standard. Although the corresponding dry time is relatively high (159

min), this can be improved in future work through variation in the control strategy, optimization, and selection of

high-efficiency components. A detailed techno-economic study of the economic viability of TE dryers is beyond the

scope of this article. However, the following trends point to a promising future for the technology: clothes dryers

consume a large amount of energy, minimum energy-efficiency standards are increasingly stringent, and alternatives

to vapor compression heat pumps for clothes drying may be needed due to the increasing restrictions on vapor

compression refrigerants and the increasing flammability of new refrigerant alternatives. In addition, costs of

nonflammable, solid state TE technology have been consistently falling for decades. By avoiding the need for vapor

compression refrigerants, the TE-based dryer may become a preferred heat pump dryer configuration.

5 CONCLUSIONS

An experimental and modeling study of a TE clothes dryer was conducted. Energy factors of up to 2.95

kgc/kWh (6.51 lbc/kWh) and SMER of 1.64 kgw/kWh were achieved experimentally. The model results were accurate

within 2 K for key temperatures, within 10% RH for key RHs and within 5% of TE power consumption, dry time, and

Patel 28
EF/SMER. The results also show that the RH and air temperature at drum exit are key factors in the overall

performance of the system. The modeling methodology used in this paper can be applied to TE dryers. It can be used

to compute the efficiency and dry time of a real-world TE-based clothes dryer. In addition, it can be used to design

the components by accurately predicting the expected state points for the cycle.

Although the efficiency of TE heat pumps are traditionally considered unfavorable compared with vapor

compression, the experimental findings of this work show that TE heat pumps can be a good match for the clothes

dryer application. The modeling results provide insight into why this surprising result is possible: having multiple

banks of TE modules allows most of the heat pumping to be performed at a low temperature lift. Consider that for a

given drying cycle, the overall system lift is 27 K (exhaust air is cooled down to 10°C and drum inlet air is heated up

to 37°C), the three individual TE banks only provide 20 K, 15 K, and 10 K of lift. The inherent modularity of TEs

also allows for flexibility in design of the heat pump.

NOMENCLATURE
2
A area [m ] or TE bank A (name of TE bank that cools air just before the exhaust)

AT approach temperature [K]

B TE bank B (name of middle TE bank)

C TE bank C (name of TE bank that heats the air just before the drum), correction factor

CFM cubic feet per minute

DC direct current

EF energy factor [lbc/kWh] or [kgc/kWh]

ER electric resistance

FMC final moisture content of cloth, expressed as mass ratio, yf = mw/mc [kgw/kgc]

h specific enthalpy [kJ/kg]

I DC current supplied to TE modules [A]

K thermal conductance [W/K]

L branch length [m]

mc dry mass of cloth [kg]

mass flow rate [kg/s]

Patel 29
 " mass flow rate of water vapor leaving dryer system (net of outflow vs inflow) [kg/s]

P power [W]

q heat transfer [W]

R electrical resistance of TE modules [Ω]

RH relative humidity [-]

SMC starting moisture content of cloth, expressed as mass ratio, y0 = mw/mc [kgw/kgc]

SMER specific moisture extraction rate [kgw/kWh]

T temperature [°C]

TE thermoelectric

V voltage [V]

volume flow rate

VC vapor compression

y mass ratio of water to cloth, mw/mc

Greek

α Seebeck coefficient [V/K]

ρ electrical resistivity

λ thermal conductivity [W1m-1K-1]

τ dry time [min]

ηPS power supply electrical efficiency (power conversion from AC to DC)

ω humidity ratio [kgw/kgda]

Subscripts

0 initial

1-10 state points as defined in Figure 1

c cloth

C cold side of TE module

da dry air

e electrical

f final

Patel 30
H hot side of TE module

n n-type semiconductor

p p-type semiconductor

w water

ACKNOWLEDGMENTS
This work was sponsored by the U. S. Department of Energy’s Building Technologies Office under Contract

No. DE-AC05-00OR22725 with UT-Battelle, LLC. The authors would like to acknowledge Mr. Antonio Bouza,

Technology Manager – HVAC&R, Water Heating, and Appliance, U.S. Department of Energy Building Technologies

Office.

Assistance in engineering design and specification for the TE modules and measurement of TE module

performance were provided by Uttam Ghoshal, Key Kolle and Ayan Guha of Sheetak, Inc.

The authors would also like to acknowledge Guolian Wu of Samsung Electronics America for his valuable

input on the research.

REFERENCES
[1] IEA, 2009, Residential Energy Consumption Survey (RECS), U.S. Energy Information Administration, Dept. of

Energy.

[2] Denkenberger, D., Calwell, C., Beck, N., Trimboli, B., Driscoll, D., and Wold, C., 2013, "Analysis of Potential

Energy Savings from Heat Pump Clothes Dryers in North America," Ecova and Collaborative Labeling and

Appliance Standards Program (CLASP).

[3] York, D., Nadel, S., Rogers, E., Cluett, R., Kwatra, S., Sachs, H., Amann, J., and Kelly, M., 2015, "New

Horizons for Energy Efficiency: Major Opportunities to Reach Higher Electricity Savings by 2030," American

Council for an Energy-Efficient Economy.

[4] Yadav, V., and Moon, C. G., 2008, "Fabric-drying process in domestic dryers," Applied Energy, 85(2), pp. 143-

158.

Patel 31
[5] Yadav, V., and Moon, C. G., 2008, "Modelling and experimentation for the fabric-drying process in domestic

dryers," Applied Energy, 85(5), pp. 404-419.

[6] Ng, A. B., and Deng, S., 2008, "A new termination control method for a clothes drying process in a clothes

dryer," Applied Energy, 85(9), pp. 818-829.

[7] Stawreberg, L., and Nilsson, L., 2013, "Potential energy savings made by using a specific control strategy when

tumble drying small loads," Applied Energy, 102(Supplement C), pp. 484-491.

[8] Ganjehsarabi, H., Dincer, I., and Gungor, A., 2014, "Exergoeconomic Analysis of a Heat Pump Tumbler Dryer,"

Drying Technology, 32(3), pp. 352-360.

[9] Cao, T., Ling, J., Hwang, Y., and Radermacher, R., 2014, "Development of a Novel Two-Stage Heat Pump

Clothes Dryer," Proc. ASME IMECE, Montreal, Quebec Volume 6A: Energy, p. V06AT07A040.

[10] Sarkar, J., Bhattacharyya, S., and Gopal, M. R., 2006, "Transcritical CO2 Heat Pump Dryer: Part 1.

Mathematical Model and Simulation," Drying Technology, 24(12), pp. 1583-1591.

[11] Mancini, F., Minetto, S., and Fornasieri, E., 2011, "Thermodynamic analysis and experimental investigation of

a CO2 household heat pump dryer," International Journal of Refrigeration, 34(4), pp. 851-858.

[12] Erdem, S., and Heperkan, H., 2014, "Numerical Investigation of the Effect of Using CO2 as the Refrigerant in a

Heat Pump Tumble Dryer System," Drying Technology, 32(16), pp. 1923-1930.

[13] Cochran, M., Goodnight, J., Babin, B., and Eckels, S., 2009, "Condensing dryers with enhanced

dehumidification using surface tension elements," Applied Thermal Engineering, 29(4), pp. 723-731.

[14] Do, Y., Kim, M., Kim, T., Jeong, S., Park, S., Woo, S., Kwon, Y., Jung, Y., Lee, J., and Ahn, Y., 2013, "An

experimental study on the performance of a condensing tumbler dryer with an air-to-air heat exchanger," Korean

Journal of Chemical Engineering, 30(6), pp. 1195-1200.

[15] Bansal, P., Sharma, K., and Islam, S., 2010, "Thermal analysis of a new concept in a household clothes tumbler

dryer," Applied Energy, 87(5), pp. 1562-1571.

[16] Bansal, P., Islam, S., and Sharma, K., 2010, "A novel design of a household clothes tumbler dryer," Applied

Thermal Engineering, 30(4), pp. 277-285.

[17] Liu, D., Zhao, F.-Y., and Tang, G.-F., 2008, "Modeling and Performance Investigation of a Closed-Type

Thermoelectric Clothes Dryer," Drying Technology, 26(10), pp. 1208-1216.

Patel 32
[18] Rowe, D. M., 2005, Thermoelectrics Handbook: Macro to Nano, CRC press - Taylor & Francis Group, Boca

Raton, FL.

[19] Patel, V. K., Goodman, D., Gluesenkamp, K., and Gehl, A., 2016, "Experimental Evaluation and

Thermodynamic System Modeling of Thermoelectric Heat Pump Clothes Dryer," Proc. International Refrigeration

and Air Conditioning Conference at Purdue University, West Lafayette, IN.

[20] Goodman, D., Patel, V. K., and Gluesenkamp, K., 2017, "Thermoelectric heat pump clothes dryer design

optimization," Proc. 12th IEA Heat Pump Conference, Rotterdam, The Netherlands.

[21] Goldsmid, H., 1995, "Conversion efficiency and figure-of-merit," CRC Handbook of Thermoelectrics, pp. 19-

25.

[22] 10 CFR 430, 2013, "Energy Conservation Program for Consumer Products," Subpart B, "Test Procedures";

Appendix D/D1/D2, "Uniform Test Method for Measuring the Energy Consumption of Clothes Dryers."

[23] 10 CFR 430, 2017, "Energy Conservation Program for Consumer Products," Subpart C, Energy and Water

Conservation Standards"

LIST OF FIGURES
Figure 1. Schematic of TE dryer model showing state points in open-loop configuration (vented). Banks A, B and C

are assemblies of TE elements. ......................................................................................................................................7

Figure 2: Model flow diagram showing each subsection. ........................................................................................... 10

Figure 3. Ideal TE couple with p- and n-branches. ...................................................................................................... 11

Figure 4. Linear regression of electrical resistance of a TE module. ........................................................................... 13

Figure 5. Hot-side (top) and cold-side (bottom) approach temperatures as a function of qH and qC, respectively. ..... 14

Figure 6. Comparison between linear regression prediction and experimental measurements for RH[7]. .................. 16

Figure 7. CAD model and experimental prototype TE clothes dryer with open-loop configuration (vented)............. 18

Figure 8. TE unit engine with heat sinks, TE modules, and spacer block. .................................................................. 19

Figure 9. Experimental data for Trial 5: temperatures and dewpoints at various state points including ambient, drum

inlet (T[6], D[6]), and drum outlet (T[7], D[7]). ......................................................................................................... 22

Figure 10. Experimental data for Trial 5: RH at ambient, drum inlet (RH[6]), and drum outlet (RH[7]). .................. 22

Patel 33
Figure 11. Experimental data for Trial 5: total power consumed by TE power supplies, blower motor, and drum rotator

motor. .......................................................................................................................................................................... 23

Figure 12. Comparison between experimental and model-predicted drum inlet temperature (left) and drum outlet RH

(right) for five trials. .................................................................................................................................................... 26

Figure 13. Comparison between experimental and model-predicted TE energy use (left) and dry time (right) for five

trials. ............................................................................................................................................................................ 27

LIST OF TABLES
Table 1. Locations of state points illustrated in Figure 1 ...............................................................................................8

Table 2. Various model inputs and parameters and their sources ..................................................................................9

Table 3. Uncertainties of measured and derived quantities ......................................................................................... 19

Table 4. Experimentally measured values used as model inputs for each trial ............................................................ 21

Table 5. Measured results used to validate model for each trial .................................................................................. 24

Table 6. All model-predicted state points for Trial 5................................................................................................... 25

Table 7. Differences between model prediction and experimental measurement for key parameters in each trial ..... 28

Patel 34