Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.

1007/s12206-024-0437-x

Journal of Mechanical Science and Technology 38 (5) 2024

Review Article
DOI 10.1007/s12206-024-0437-x
Modeling and simulation of heat pipes:
review
Keywords:
· Heat pipe
Ji-Su Lee1, Seok-Ho Rhi1 and Sun-Kook Kim2
· Two-phase flow
· Numerical analysis 1 2
School of Mechanical Engineering, Chungbuk National University, Cheongju 28644, Korea, LS Electric
· Modeling and simulation
· CFD (computational fluid dynamics)
Co., 1 Songjung-ro, Gyeonggi-do 28439, Korea

Correspondence to: Abstract Heat pipes have been extensively studied using various methods, such as
Seok-Ho Rhi MATLAB, AMESIM, and commercial CFD software. Early numerical models employed the
[email protected]
thermal conductance approach, which oversimplified the characteristics and performance of
heat pipes. Newer models comprise the thermal resistance model, which emphasizes two-
Citation: phase heat transfer, AI-based approaches for predicting flow patterns and thermal characteris-
Lee, J.-S., Rhi, S.-H., Kim, S. K. (2024).
Modeling and simulation of heat pipes: tics, and the CFD model, which accounts for phase changes and two-phase flow utilizing the
review. Journal of Mechanical Science VoF and phase change models. Although the thermal resistance model demands fewer com-
and Technology 38 (5) (2024) 2591~2612. puting resources, it has limited visualization of the flow pattern and wick structure. In contrast,
http://doi.org/10.1007/s12206-024-0437-x
CFD models offer advantages in visualizing the flow pattern and thermal characteristics but
have limitations in terms of consuming computing resources and considering heat transfer from
Received April 11th, 2023 wick structures and mass transfer rates caused by phase changes. Consequently, most simu-
Revised December 30th, 2023 lations are validated with experimental results. Innovative approaches for phase changes in
Accepted January 29th, 2024 heat pipes and wick structures are necessary to address these challenges.
† Recommended by Editor
Tong Seop Kim

1. Introduction
Heat pipes, known for their exceptional heat transfer efficiency, are extensively employed to
dissipate high heat flux generated from various sources, including CPUs, LED modules, and
power devices. These devices leverage a two-phase flow of the working fluid, involving proc-
esses like boiling, evaporation, and condensation, as depicted in Fig. 1 [1].
The intricate working mechanism of heat pipes involves loading heat at the evaporator's sur-
face, boiling and evaporating the working fluid, transferring vapor bubbles to the condenser
section, and efficiently transporting heat through a two-phase process. This continuous cycle,
driven by the phase change of the working fluid, makes heat pipes highly effective for manag-
ing and dissipating high heat fluxes [1-3].
Design considerations such as the shape of the flow channel, exterior design, and method of
returning condensed fluid distinguish various types of heat pipes as shown in Table 1. Tradi-
tionally, fluid return is achieved through capillary force, where the wick structure aids in fluid
movement, or gravity-driven processes, with specialized designs like rotating heat pipes and
innovative systems such as magnetohydrodynamic and electrohydrodynamic heat pipes further
expanding their applicability [2, 4-7].
Beyond thermosyphons and standard heat pipes, there are specifically modified heat pipe
designs to meet specific engineering requirements. An exemplary instance is the rotating heat
pipe, commonly deployed in rotor applications, where centrifugal force is harnessed to ensure
the efficient return of the condensed working fluid to the evaporator. This astute design choice
guarantees reliable heat transfer performance even in dynamically challenging environments,
making it invaluable for diverse industrial applications [1, 2, 5].
Furthermore, the progression of heat pipe technology has given rise to innovative systems
that leverage distinct physical forces for efficient fluid return. Magneto-hydrodynamic heat pipes
© The Korean Society of Mechanical
Engineers and Springer-Verlag GmbH seamlessly integrate the phase change behavior of the working fluid with the dynamic Lorentz
Germany, part of Springer Nature 2024 force. This force is generated through electric currents induced within the fluid and an external

2591
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Table 1. Types of heat pipes according to the returning method of the

condensed working fluid [2, 4-7].

Fig. 1. Operating mechanism of a heat pipe [1, 2].

Table 2. Working fluids of heat pipes [2].

Fig. 2. Operating limits of heat pipes [7-9]. Melting point Boiling point Latent heat
Working Useful range
at atmosphere at atmosphere of vaporization
fluid (°C)
(°C) (°C) (kJ/kg)
magnetic field, synergistically enhancing fluid movement and Helium -271 -269 21 -271 to -269
promoting the migration of vapor while facilitating the return of
Nitrogen -210 -196 198 -210 to -160
condensed liquid. The Lorentz force, a constituent of the Lor-
Ammonia -78 -33 1360 -60 to 100
entz equation involving particle charge, velocity, and magnetic
Acetone -95 57 518 0 to 120
field strength, drives this augmented circulation. Electrohydro-
dynamic heat pipes exploit the interplay between an electric Methanol -98 64 1093 10 to 130
field and conductive fluid. The electric field prompts orches- Ethanol -112 78 850 0 to 130
trated motion of charged particles, enhancing convective heat Water 0 100 2260 30 to 200
transfer through intricate fluid flow patterns. This phenomenon Mercury -39 361 298 250 to 650
leverages charge-field interactions to optimize fluid dynamics, Cesium 29 670 490 450 to 900
significantly boosting heat transfer rates. Osmotic heat pipes Potassium 62 774 1938 500 to 1000
speculate a mechanism where fluid motion is powered by os- Sodium 98 895 3913 600 to 1200
motic pressure gradients across a semi-permeable membrane.
Utilizing solvent migration from areas of low to high solute con-
centration [1, 4-7]. [1, 2, 4, 7-9].
The choice of working fluid significantly influences the oper- To ensure structural suitability, critical aspects of heat pipe
ating temperature and maximum heat transfer rate, and various design and optimization include considerations such as ther-
heat pipe designs cater to specific engineering needs. Table 2 modynamic state, geometrical parameters, and adherence to
shows the operating temperature and thermal characteristics of limits defined by factors like the viscous, sonic, entrainment,
working fluids. As shown in Fig. 2, those material properties and boiling limits [1, 2, 4, 7-10]. These considerations play a
contribute consisting various operating limits of heat pipes such pivotal role in determining the performance and efficiency of
as the capillary, viscous, sonic, entrainment and boiling limits heat pipes, guiding researchers in developing effective design
which make harder to define operating limits of working fluids methodologies. In particular, understanding and addressing

2592
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

challenges related to the viscous limit, sonic limit, and entrain- making it one of the most common types. Its simple shape and
ment limit are paramount in achieving optimal structural con- flexibility make it suitable for diverse devices and operating
figurations for heat pipes. conditions. Loop heat pipes (Fig. 4(b)) transfer heat from the
Ensuring compatibility between the material of the heat pipe evaporator to the condenser through circulation of working fluid.
and the working fluid is imperative for optimizing performance. Pulsating heat pipes (PHPs) which is shown in Fig. 4(c) contain
Incompatibility between the working fluid and the materials of working fluid within a turned pipe, and the vapor bubble forma-
the case or wick may induce chemical reactions, leading to tion in slug/plug flow pattern rapidly pulsates in the internal
internal corrosion and the release of non-condensable gases. region to facilitate efficient heat transfer. Variable conductance
The internal corrosion results in a reduction in the thermal con- heat pipes (VCHPs) which is shown in Fig. 4(d) are filled with
ductivity of the case or wick material, thereby impeding heat working fluid and non-condensable gas. The region of the con-
transfer. Furthermore, the generation of non-condensable gas- denser varies according to the heat input capacity that controls
es complicates the circulation of the working fluid, posing chal- the volume of the non- condensable gas inside the heat pipe.
lenges in interrupting the flow of vapor from the evaporator to Vapor chambers and heat pipes for solar energy collectors
the condenser. Thus, the compatibility of the working fluid (Figs. 4(e) and (f)) represent other specialized heat pipe de-
stands as a pivotal design consideration. Basilulis and Busse signs, with vapor chambers commonly applied to smart de-
conducted comprehensive compatibility tests, and the out- vices for heat transfer and heat pipes designed specifically for
comes are delineated in Table 3 [11]. solar energy collection [7-9].
These phase changing behaviors of the working fluid inside The paper extensively explores the theoretical underpinnings
the heat pipe are significant clues in predicting thermal charac- of heat pipes, emphasizing sophisticated modeling and simula-
teristics as the heat transfer quantity and efficiency can be tion techniques. As illustrated in Fig. 5, the modeling of heat
improved via two-phase flow patterns. For this reason, many pipes encompasses the thermal conductance model using
experiments on the working fluid’s behavior, with various vari- thermal resistance, an AI-based model, and a two-phase mod-
ables, have been conducted. The flow patterns in vertical wick- el based on phase change [12-14].
less heat pipes are shown in Fig. 3. Which can be applied to Traditionally, the modeling of heat pipes heavily relied on
other types of heat pipes [8, 9]. thermal resistance models, widely used but exhibiting limita-
In Fig. 3, the bubbly flow occurs at the beginning of evapora- tions in accurately predicting subtle variations and capturing
tion, showing a large number of small bubbles. The bubbles the nuanced behavior of working fluids. Recognizing these
rise due to buoyancy force, grow, and coalesce. After the bub- limitations, there is a growing interest in utilizing numerical
bly flow, large and bullet-shaped bubbles are observed in the simulations as a powerful alternative [15, 16].
flow region, typically referred to as a slug/plug flow. The churn In recent years, the landscape of heat pipe research has
flow represents the transition process from slug/plug flow to been significantly improved by numerical methods, especially
annular flow. In annular flow, a liquid film occurs at the wall, with the aid of advancing computing resources. This shift to-
and the vapor core forms in the middle [7]. wards numerical simulations represents a departure from the
Furthermore, heat pipes are also designed based on their resource-intensive nature of traditional experimental ap-
shapes and purposes of use, catering to diverse engineering proaches. The integration of numerical simulations into the
needs and applications. Fig. 4 [1-4] illustrates a range of heat analytical framework provides a more comprehensive under-
pipe designs according to their shapes and purposes. The standing of the complex interplay between heat transfer, fluid
conventional heat pipe (Fig. 4(a)) is widely recognized for its
straightforward manufacturing process and affordable price,

Table 3. Compatibility data between the working fluids and the materials
[11].

Working fluid
Materials
Water Acetone Ammonia Methanol
Copper RU RU NU RU
Aluminium GNC RL RU NR
Stainless steel GNT PC RU GNT
Nickel PC PC RU RL
Refrasil fibere RU RU RU RU
RU, recommended by past successful usage; RL, recommended by litera-
ture; PC, probably compatible; NR, not recommended; GNC, generation of
gas at all temperatures; GNT, generation of gas at elevated temperatures, Fig. 3. The schematic representation of flow patterns of vertical wickless
when oxide present. heat pipes [7, 9].

2593
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

(a) Conventional heat pipe (b) Loop heat pipe (c) Pulsating heat pipe

(d) Variable conductance heat pipe (e) Vapor chamber (f) Heat pipe for solar energy collector

Fig. 4. Type of heat pipe according to their shapes and purpose of use [1-4].

AI-based analyses may face challenges in fully representing

the complexity of the underlying physics and fluid dynamics,
necessitating a cautious interpretation of their results. This
multifaceted approach, combining advanced numerical simula-
tions with AI-based analyses, opens up new possibilities for
more accurate and nuanced insights into the intricate dynamics
of heat pipes with phase change. The integration of these
methodologies not only addresses the limitations of traditional
models but also aligns with the evolving landscape of computa-
tional tools, making significant strides in the analysis and de-
sign of heat pipes [13, 19-21].
In the subsequent sections, detailed exploration will be con-
ducted on the challenges associated with modeling heat pipes,
Fig. 5. The methods for the modeling and simulation of heat pipes [12-14].
addressing potential solutions, and emphasizing the signifi-
cance of numerical simulations in advancing understanding of
dynamics, and phase change phenomena within heat pipes heat pipe behavior and design.
[17, 18].
Furthermore, the exploration of artificial intelligence (AI)-
based analysis methods has emerged as a noteworthy avenue 2. Methodologies of modeling and simula-
in heat pipe research. Recent investigations have showcased tion for heat pipes and their difficulties
the utility of artificial neural networks (ANN) in predicting and 2.1 Methodologies of modeling and simulation
modeling heat pipe behavior. For instance, many studies have for heat pipes
demonstrated the promising application of AI techniques, con-
tributing valuable insights into the prediction of flow patterns, Designing heat pipes typically involves expensive and time-
thermal characteristics, and optimal design of heat pipes [13, consuming experimental evaluations. To mitigate these disad-
19-21]. vantages, researchers have turned to numerical modeling and
However, it is crucial to acknowledge the fundamental limita- simulation techniques for heat pipes. However, despite signifi-
tions of AI models, particularly in capturing the intricate and cant advancements, mathematical modeling and numerical
dynamic nature of heat pipe phenomena. Despite their promise, analysis of heat pipes still present technical challenges [16].

2594
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

The numerical analysis model of the working fluid is based through CFD has emerged as a powerful tool. By simulating
on continuity equations and the Navier–Stokes equation (Eqs. the complex two-phase flow and phase change processes
(1) and (2)). However, for heat pipe analysis, the energy equa- within heat pipes, CFD simulations enable researchers to gain
tion (Eq. (3)), the evaporation-condensation relationship, and deeper insights into the intricate fluid dynamics and heat trans-
the heat transfer relationship according to the phase of the fluid fer phenomena [12, 25].
are also required. Analyzing the two-phase flow in a closed In the pursuit of accuracy, researchers are continually explor-
system poses numerous issues, making numerical studies of ing novel techniques to overcome the challenges associated
heat pipes challenging [12, 16, 22, 23]: with the wick structure and phase change simplification in the
phase change model. Advancements in numerical methods,
∂ρ ∂ optimization algorithms, and parallel computing hold promise in
+ ( ρu i ) = 0 , (1)
∂t ∂x i reducing the computational burden while retaining the neces-
∂p ∂τ sary level of detail for reliable predictions [12, 26].
∂
∂t
∂
( ρu i ) + ( ρu i u j ) = − + ij + gi + Fi +
∂x i ∂x i ∂x j
∂
∂x j
( )
−ρu i 'u j' (2) To address the limitations of both approaches, a multidisci-
plinary approach that combines the strengths of different nu-
∂
( ρE ) + ∇i( v ( ρE + p ) ) merical methods is gaining traction. For instance, researchers
∂t
(3) may utilize the thermal resistance model for initial estimations
⎛ ⎞
⎝ j
( )
= ∇ ⎜ k eff ∇T − ∑h j J j + τ eff • v ⎟ + Sh .
⎠
and then validate and refine the results using the phase
change model with CFD. This integrated approach allows for a
more comprehensive understanding of heat pipe behavior
In early attempts, heat pipes were analyzed based on ther- while optimizing computational resources [12, 27, 28].
mal conductance obtained from experiments. In this approach, Despite the ongoing challenges, the field of heat pipe model-
the heat pipe was modeled as a solid with thermal conduc- ing and simulation continues to evolve rapidly. Researchers
tance, validated through experimental results. Although this are encouraged to push the boundaries of existing methodolo-
method yielded accurate results, it had limitations in investigat- gies and embrace emerging technologies to unlock new possi-
ing the sensitivity and characteristics of the heat pipe. The bilities in understanding and optimizing the performance of
methodologies for the modeling and simulation of heat pipes heat pipes. With collaborative efforts from experts in numerical
are illustrated in Fig. 5 [12-14]. modeling, fluid dynamics, and thermal engineering, the heat
pipe's full potential can be harnessed to address the ever-
2.1.1 Computational methodologies growing demands of efficient heat transfer in various applica-
Thermal resistance with the two-phase heat transfer model tions [12].
played a significant role in the early days of heat pipe analysis,
primarily because of the limitations in computing resources. 2.1.2 AI based analysis
While this approach provided a simpler and computationally The artificial neural network (ANN) is the first AI algorithm to
more efficient solution, it also introduced certain challenges in be applied for the modeling and optimization of heat pipes. As
accurately predicting thermal characteristics and overall heat depicted in Fig. 6, a network resembling the neural structure of
pipe performance [10, 13, 24]. the brain is constructed by the ANN to generate output condi-
On the other hand, the phase change model utilizing compu- tions for multiple input scenarios. These models are catego-
tational fluid dynamics (CFD) represents another numerical rized based on various training algorithms, notably including
method extensively employed for investigating heat pipes. CFD
possesses the capability to solve complex phase change and
multiphase flow phenomena, but it comes at the cost of sub-
stantial time and computing resources. To manage these re-
source demands, researchers often resort to simplifying the
geometry of the simulation, opting for two-dimensional or sym-
metry models. Although advancements in the phase change
model have led to improvements in time consumption and com-
puting efficiency, challenges still persist due to the inherent sim-
plification of the wick structure and phase change dynamics [12].
As computational technology continues to advance, re-
searchers are keen to strike a balance between accuracy and
efficiency in heat pipe analysis. The thermal resistance model,
despite its limitations, still holds relevance in certain scenarios
where computational resources are constrained, and a quick
estimation of thermal behavior is necessary. However, for
comprehensive and detailed analysis, the phase change model Fig. 6. Basic artificial neural network architecture [13].

2595
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Table 4. AI based modeling in heat pipe research. pulsating heat pipes using artificial neural network (ANN) and
regression/correlation analysis (RCA). The ANN achieves R2
Approach description Author/year Methodology overview
of 0.89, RCA with dimensional parameters has lower accuracy
Demonstrated promising
Simulated a vertical Khandeka (R2 = 0.38), and a third model using a dimensionless Kutate-
simulation results with
copper PHP using ANN et al. [13]/2002 ladze number shows high accuracy (R2 = 0.95).
secured data
Applied ANN for Reported a 4 % relative
Wang et al. [30] presents a novel artificial neural network
Li et al. (ANN) model for predicting closed pulsating heat pipe (PHP)
simulating and error between modeling
[19]/2011
experimenting on PHPs and experiment performance, achieving high accuracy. Using input parameters,
ANN achieved R2 of 0.89, it covers various working fluids and operational conditions. The
RCA with dimensional study introduces a lumped parameter method to evaluate
Introduced thermal parameters had lower accu- evaporation and condensation temperatures, contributing valu-
Patel et al.
performance models using racy (R2 = 0.38), third model
[30]/2018 able insights for PHP applications.
ANN and RCA using Kutateladze number
showed high accuracy Wang et al. [31] propose an artificial neural network (ANN)
(R2 = 0.95) model to predict thermal resistance in closed vertical meander-
Adopted various algorithms TLBO demonstrated more ing pulsating heat pipes (PHPs) using water. The model ad-
Rao et al.
for heat pipe design, reliable results in optimizing dresses shortcomings in existing correlations. With optimized
[20]/2015
emphasizing TLBO heat pipe design parameters, the ANN achieves high accuracy, exhibiting good
Achieved high accuracy agreement with experimental data. This model provides a
with various working fluids promising tool for predicting PHP performance under various
Presented a novel ANN
Wang et al. and operational conditions, conditions, thereby enhancing thermal management in high-
model for predicting closed
[30]/2019 introduced lumped parameter
PHP performance
method for temperature
tech applications.
evaluation Wang et al. [21] conducted the optimization of the heat pipes'
Addressed shortcomings in performance based on big-data/machine-learning techniques,
Proposed an ANN model and their report proposed the unidentified data-driven algorithm
Wang et al. existing correlations,
for predicting thermal
resistance in closed PHPs
[31]/2019 achieved high accuracy to predict the performance of heat pipes.
with optimized parameters Rao et al. [20] undertook a comprehensive investigation in-
Optimized heat pipe Introduced an unidentified volving a variety of algorithms to attain the optimal design of
performance using Wang et al. data-driven algorithm for heat pipes. Within the scope of their research, the teaching–
big-data/machine-learning [21]/2021 predicting heat pipe
learning-based optimization (TLBO) algorithm emerged as a
techniques performance
pivotal element, substantially enhancing the reliability of the
obtained results. The outcomes derived from the application of
the multilayer perceptron neural network (MLPNN), radial basis the TLBO algorithm for optimizing heat pipe design underwent
function (RBF) neural network, and convolutional neural net- a comparative analysis with results obtained from alternative
work (CNN), and are operated using supervised learning. ANN algorithms, namely, NPGA (niched pareto genetic algorithm),
is primarily employed in prediction, control, modeling, and pat- GEM (grenade explosion method), and GEO (generalized
tern classification. Due to these factors, artificial neural net- external optimization). The comparative assessment discerned
works (ANNs) hold a favorable position in predicting heat pipe superior results achieved through the TLBO algorithm when
behavior. By training the ANN through supervised learning with contrasted with those obtained from the NPGA, GEM, and
diverse variables like dimensions, working fluid types, wick GEO algorithms.
characteristics, materials, and more, it can effectively predict
performance metrics of heat pipes, including thermal resis- 2.2 Difficulties of modeling and simulation for
tance and temperature distribution, showcasing outstanding heat pipes
performance [13, 29].
Table 4 demonstrates examples of heat pipe modeling The numerical analysis of heat pipes poses several formida-
through the application of artificial neural network (ANN). Fur- ble challenges [12, 13, 22, 32]:
ther elaboration on each case is provided below. Khandeka et 1) Multifaceted nature of phase change: Heat pipes involve
al. [13] presented this approach and simulated vertical copper complex phase change phenomena, including boiling, evapo-
PHP. They reported the simulation result using the ANN ap- ration, and condensation, within a closed system. To compre-
proach, which seems to be very promising if sufficient data are hensively analyze heat pipes, it becomes imperative to handle
secured. the intricate interactions and transitions between these phases.
Li et al. [19] simulated and experimented on the heat transfer This complexity demands significant computing resources and
performance of PHPs using ANN, and they obtained the result intricate mathematical formulations to accurately simulate
that the relative error between the modeling and experiment pressure gradients, temperature distributions, and heat transfer
was 4 %. during the phase change processes [17].
Patel et al. [30] proposes thermal performance models for 2) Modeling wick structure: The wick structure in heat pipes

2596
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

plays a crucial role in generating capillary forces that facilitate the performance of the AI model is heavily dependent on the
fluid flow and phase change. However, accurately modeling quality and diversity of the data it has been exposed to during
the intricate geometry and behavior of the wick structure pre- the training phase [20, 21].
sents significant challenges. It often requires a separate im- Another factor that warrants attention is the relatively limited
plementation of heat transfer analysis and the capillary function, availability of meta-models and comprehensive research con-
further adding to the complexity of the numerical analysis [33]. cerning heat pipes. While AI has showcased its transformative
3) Incorporating boiling mechanisms: Boiling is a critical potential across various domains, its application in the context
process in heat pipes, leading to the generation of vapor bub- of heat pipes requires meticulous validation and calibration. As
bles, which significantly impact the heat transfer characteristics. the field continues to evolve, more efforts are needed to de-
Incorporating these boiling mechanisms into analytical meth- velop sophisticated meta-models that can further enhance the
ods can be quite challenging. Due to the complexity and com- predictive accuracy of AI-based systems [13, 19, 20].
putational intensity involved in simulating bubble generation, it Given the complex and ever-evolving nature of heat pipes
is common for certain numerical models to simplify the analysis and related phenomena, continuous validation becomes a
by focusing solely on evaporation and condensation phenom- paramount consideration. Validation serves as a crucial step in
ena. Consequently, such simplifications might limit the accu- assessing the reliability and performance of AI-based predic-
racy of predictions, especially in cases where boiling signifi- tive models. Regularly verifying the model's outputs against
cantly influences heat pipe performance [12]. real-world observations allows researchers and practitioners to
4) Complexity and experimentation: Despite considerable identify potential discrepancies and fine-tune the AI algorithms
advancements in numerical modeling, the inherent complexity accordingly. This iterative process of validation and refinement
of heat pipes often necessitates experimental validation. The is vital in ensuring the long-term viability and usefulness of AI-
numerical analysis must cope with a wide range of parameters based predictions in practical applications [13].
and conditions, making it difficult to create all-encompassing
models. As a result, experimental verification remains a key
component of the overall analysis process, bridging the gap
3. Thermal resistance models with two-
between theory and real-world application [12].
phase heat transfer in heat pipes
Over the years, various numerical methodologies have been The utilization of numerical and mathematical analyses em-
employed for heat pipe analysis, ranging from local analysis to ploying thermal resistance models confers significant advan-
one-dimensional and two-dimensional simulations of thermo- tages, providing a direct and pragmatic approach to predicting
syphons without wick structures and pulsating heat pipes outcomes for commercial heat pipes. The thermal resistance
(PHPs). With the advent of multi-core computing, researchers model, owing to its simplicity and reliability, has garnered wide-
have expanded their efforts to include three-dimensional PHPs, spread application as an effective tool for heat pipe design.
thermosyphons, and loop heat pipes. Utilizing the volume of Nonetheless, it is not without limitations, particularly in its inca-
fluid (VoF) model, current numerical models can analyze entire pacity to account for the phase change phenomenon occurring
heat pipes, yet challenges persist in precisely modeling the within the heat pipe. For these reasons, in this paper, we aim to
wick structure, which significantly affects the overall heat pipe briefly address the thermal resistance with a two-phase model.
performance. Despite the computational complexities and in- Utilizing the thermal resistance models, certain studies have
herent assumptions in numerical analysis, it remains a valuable attempted to calculate the thermal properties of heat pipes by
tool, often supplemented and validated through experimental incorporating the two-phase heat transfer model. In this ap-
investigations [12, 13, 22, 32]. proach, the pipe and wick structures are modeled using ther-
AI-based predictive models offer a promising edge by ena- mal resistances, while the inner region of the heat pipe is ana-
bling rapid predictions for a wide array of heat pipes and vari- lyzed with respect to two-phase flow heat transfer using corre-
ables, given that they have undergone proper training. This lations listed in Table 5. The implementation of this modeling
advantage stems from the inherent ability of AI systems to and simulation can be easily achieved through numerical engi-
process vast amounts of data and discern patterns that might neering tools [23, 24, 34].
not be immediately apparent through conventional methods. Several researchers have explored the thermal resistance
The potential efficiency gains are particularly enticing in indus- model with two-phase heat transfer in the context of heat pipe
tries and fields where timely and accurate predictions are of analysis:
utmost importance [13]. Ali [34] and Lee et al. [23] utilized AMESim 1D modeling
Nevertheless, the quest for precise and reliable predictions software to study and model heat pipes AMESim provides a
demands considerable efforts. Developing a robust AI model convenient method for modeling systems through its integrated
necessitates an extensive collection of data and exhaustive components. In their study, they utilized thermal conduction
training on diverse cases representative of real-world scenarios. and a two-phase pipe model to simulate a heat pipe. They
Each unique case demands its own tailored model to ensure simplified the wick structure using effective thermal conductivity,
optimal results. As such, the process of training these models while implementing the internal phase change and heat trans-
can be resource-intensive and time-consuming. Additionally, fer model within the heat pipe using the program's two-phase

2597
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Table 5. Two-phase heat transfer correlations [35]. Ghajar et al. [24] proposed a hybrid CFD-mathematical
model. They employed the CFD model to calculate the heat
Condition Name Application Tested fluids No.
leak in the compensation chamber arising from two-phase heat
Shah
Vertical tubes Water, CfCs, HCs 1 transfer. Additionally, the thermal resistance model was math-
(1979)
ematically utilized for the other region of the heat pipe.
Cavalini and
Condensation zecchin
Horizontal CFCs, HCFCs,
2
Arabnejad et al. [36] conducted a numerical analysis of U-
tubse HFCs shaped PHPs, deriving the governing equations (continuity,
(1974)
Traviss Horizontal R12 (CFD), R22 momentum, and energy) analytically. They solved the charac-
3 teristics of two-phase flow within PHPs using an implicit solver,
(1973) tubes (HCFC)
Chen Water, ethylene demonstrating precise predictions of heat transfer rates within
Vertical tubes 4
(1966) glycol, etc. an accurate range.
More than 13000 Noh et al. [9] studied the thermal optimization of PHPs with a
Vertical tubes
Boiling “VDI” steiner data sets developed one-dimensional model that considered two-phase
and taborek tubes 5 flows. Analytically calculating the two-phase flow using govern-
(1992) Horizontal More than 13000 ing equations for liquid slugs (incompressible) and vapor plugs
data sets (compressible), their model was validated against experimental
data.
No Equations Hansel et al. [28] investigated conventional heat pipes using
⎡ ⎛ x0.76 (1 − x )
0.04
⎞⎤ the Sockeye code. The code solved governing equations dis-
hTP = hLO ⎢(1 − x ) + ⎜ 3.8 ⋅
0 .8
⎟⎥
⎜ ( Pred )
0.38
⎟⎥ cretized by the finite volume scheme, considering capillary
1 ⎣⎢ ⎝ ⎠⎦
pressure and heat transfer. The study successfully obtained
λl
hLO = 0.023 ⋅ ReLO 0.8 ⋅ Prl 0.4 ⋅ void fraction, pressure profiles, and temperature distribution,
Dh
0.8
validating the results by comparing them with analytical solu-
hTP = hLO ⋅ ⎡⎣1 − x + x ρl / ρ g ⎤⎦ tions and experimental data.
2 The thermal resistance model with two-phase heat transfer
λl
hLO = 0.05 ⋅ ReLO 0.8 ⋅ Prl1/3 ⋅
Dh offers the advantage of considering two-phase heat transfer
F1 with relatively fewer computing resources. However, it still fac-
hTP = hLO ( Prl ) [1 − x ]
2/ 3 0.9

0.23 F2 es challenges in visualizing two-phase flow behavior and sim-

3
λl plifying the wick as a solid conductance. Despite these limita-
hLO = 0.023 ⋅ ReLO 0.9 ⋅ Prl 1/3 ⋅
Dh tions, it remains a valuable tool for analyzing heat pipe behav-
hTP = S ⋅ hNcB + F ⋅ hcv ior and is often complemented with experimental validation [9,
4 λl 23, 24, 28, 34, 36].
hcv = 0.023 ⋅ ReLO 0.8 ⋅ Prl 0.4 ⋅
Dh The reason why the thermal resistance with two-phase mod-
h = 3 hcv 3 + hNcB 3 el is essential is that the CFD (computational fluid dynamics)
hcv = hLO ⋅ FTP
modeling of heat pipes has evolved based on the assumptions
1
of this model. Such CFD analysis is an overly simplified ap-
FTP = proach, which involves simplifying the evaporation/condensa-
[A + A ]
0.5
1 2

−2.2
tion model and heat transfer model. For instance, Kaya et al.
5 ⎡ 0.37
⎤
0.01 ⎛ ρ ⎞ [8] conducted an analysis of a loop heat pipe by modeling the
A1 = ⎢(1 − x ) + 1.9 x0.6 ⋅ (1 − x ) ⋅ ⎜ l ⎥
1 .5

⎜ρ ⎟⎟
⎢ ⎝ g ⎠ ⎥ wick's effective thermal properties and simplifying the heat
⎣ ⎦
⎡⎛ h ⎞
−2.0 transfer and flow of the working fluid. Solomon et al. [33] mod-
⎡ 0.7 ⎛ ρ ⎞
0.67
⎤⎤
A2 = ⎢⎜ VO ⎟ ( x ) ⋅ ⎢1 + 8 (1 − x ) ⎜ l
0.01
⎟⎟ ⎥⎥ eled a nano-fluid heat pipe using a screen mesh as the wick,
⎢⎝ hLO ⎠ ⎜ ⎥⎥
⎣
⎢
⎣ ⎝ ρg ⎠ ⎦⎦ where the wick was modeled with effective thermal properties,
and the liquid and gas phases were separated to analyze the
flow and heat transfer mechanism of the working fluid.
pipe component. The thermal conduction model is essentially In this manner, the modeling of heat pipes has advanced
applied to both the wick and the heat pipe case. The internal based on the relatively straightforward thermal resistance with
phase change is computed based on internal pressure and two phase model, and to date, many assumptions and model-
temperature. The heat transfer model between the working ing are still being carried out using this technique. Therefore,
fluid and the wall is determined using the equation provided in the thermal resistance with two phase model holds a crucial
Table 5 [35]. Additionally, the internal mass flow rate of the position in heat pipe modeling methods.
working fluid is modeled using the program's internal function, The utilization of thermal resistance and mathematical models
taking into account latent heat at that pressure. This approach in analyzing heat pipes presents several challenges and limita-
allows for a comprehensive consideration of internal circulation tions that warrant careful consideration [8, 10-14, 27, 31-33]:
and heat transfer. Inadequate modeling of two-phase flow: One of the primary

2598
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

challenges lies in the accurate representation of the complex neering applications [13].
two-phase flow behavior within the heat pipe. The fundamental
limitations of simplified models often hinder a comprehensive
understanding of the intricate fluid dynamics involved.
4. CFD modeling analysis in heat pipes
Limited visualization of two-phase flow: While thermal char- 4.1 CFD modeling and two-phase flow analy-
acteristics, thermal resistance, and maximum heat transfer can sis for heat pipes
be effectively simulated, visualizing the two-phase flow poses
additional difficulties. Expressing the profile of void fraction The investigation of numerical studies concerning two-phase
obtained through simulations requires additional efforts and heat transfer in heat pipes has yielded insightful findings. As
techniques to enable a comprehensive visualization of the flow early as 1973, Rohani and Tien [37] developed a two-phase
patterns. numerical analysis model for gas-loaded heat pipes based on
Time-intensive implementation of analysis models: Develop- the energy balance equation. The study revealed that the flow
ing the analysis model demands significant time and effort, as results were remarkably similar to those observed in simple
each governing equation must be meticulously incorporated heat pipes. Subsequently, Harley and Faghri [38] presented a
into the computational program. The complexity of the model transient numerical solution for rotating heat pipes, uncovering
can lead to computational overhead, impacting the overall complex vapor flow patterns that exhibited significant differ-
simulation time. ences from conventional heat pipes.
Divergence risks in simulation: The simulation process is The advent of enhanced computing resources and computa-
susceptible to divergence, particularly when the solver strug- tional fluid dynamics (CFD) has facilitated numerous attempts to
gles to accurately capture the significant density difference explore the two-phase heat transfer phenomena. These simula-
between the liquid and vapor phases. To mitigate such issues, tions have been conducted under specific assumptions [12, 39]:
relaxation factors are often applied, but this may introduce 1) All phases of the working fluid are incompressible.
uncertainties in the results. 2) The phase change inside the heat pipe is an interaction
Necessity of experimental validation: To ensure the reliability between evaporation and condensation.
and accuracy of the thermal resistance model and its predic- 3) Evaporation and condensation occur at the saturation
tions, experimental validation is essential. While simulations temperature.
can provide valuable insights into thermal characteristics and 4) The boiling process is neglected.
heat transfer, experimental investigation is obligatory to cor- In line with these assumptions, numerical analysis can be
roborate the model's performance with real-world data. performed using the multi-phase model, with many studies
Addressing these critical points in the thermal resistance employing the volume of fluid (VoF) model—a popular ap-
model requires a multidisciplinary approach and the integration proach to analyze multiphase fluid flows [30]. The VoF model
of advanced techniques. Incorporating more sophisticated two- assumes that each phase is a permeable continuum, and it
phase flow models and visualization methods can enhance the has been widely applied to the analysis of evaporation and
accuracy of predictions and offer a deeper understanding of condensation in heat pipes.
heat pipe behavior. Furthermore, efforts to optimize the imple- The CFD simulation for heat pipes is divided into three main
mentation of the analysis model can lead to more efficient and stages: pre-processing, solver, and post-processing (Fig. 7).
streamlined simulations, reducing computational overhead. During pre-processing, the user generates a grid from the
To overcome the challenges related to divergence risks, re-
searchers can explore alternative solver approaches and cou-
pling strategies. By refining solver settings and implementing
robust numerical techniques, the accuracy and stability of the
simulations can be improved.
Collaboration between numerical modelers and experimental
researchers is paramount in advancing the field of heat pipe
analysis. Experimental validation not only establishes the cred-
ibility of the thermal resistance model but also enables re-
searchers to gain insights into real-world phenomena that
might be challenging to simulate accurately [23, 24].
As technology continues to advance, the integration of cut-
ting-edge computational methods, such as machine learning
and artificial intelligence, may offer new avenues to enhance
the accuracy and efficiency of thermal resistance models for
heat pipes. Embracing these emerging technologies will un-
doubtedly drive innovation and facilitate the development of
highly efficient and reliable heat pipe systems for various engi- Fig. 7. Flow chart of the CFD simulation for heat pipes.

2599
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Table 6. Definition of the volume fraction in the VoF model [38]. Table 7. Mass transfer rate of evaporation and condensation [21].

α q (volume fraction) State of fluid Process Condition Phase Mass transfer rate
0 Empty
ml→v = −0.1⋅ α lρl
( Tl − Tsat )
Liquid
0 < αq < 1 One or more other fluids Tsat
Evaporation Tl > Tsat
1 Full
ml→v = 0.1⋅ α lρl
( Tl − Tsat )
Vapor
Tsat

computer-aided design (CAD) file and assigns boundary m v→l = 0.1⋅ α v ρ v

( Tsat − Tv )
Liquid
names. The pre-processing file is then imported into the solver, Tsat
Condensation Tv < Tsat
where boundary conditions, models, and solving methods are ( Tsat − Tv )
Vapor m v→l = −0.1⋅ α v ρ v
designated. In the post-processing stage, data are analyzed Tsat
through a grid independence test, and users can verify the
simulation results [12, 40].
varying pressure and temperature conditions [12].
After the evaporation/condensation model was imbedded in
4.1.1 VoF (volume of fluid) model and the phase
the CFD simulation, which is the Lee model. Some studied
change model
used to apply Lee model as the evaporation/condensation
The VoF model is the Euler–Euler approach to analyzing
model and it is expressed as Eq. (7) [25]:
multiphase fluid flows and is solved under the assumption that
each phase is a permeable continuum. The VoF model has
∂
been applied to the evaporation/condensation analysis of heat
∂t
( )
(α v ρv ) + ∇ ⋅ α v ρv Vv = mlv − mvl (7)
pipes. The continuity equation applied to the volume fraction of
the two or more phases of fluid Eq. (4) is calculated and by
using it, the behavior of the multiphase can be predicted. The where αv is the volume fraction of the vapor, ρ v is the density
summation of all of the fluid’s volume fraction is 1, as shown in of the vapor and Vv is the velocity of the vapor. The phase
Eq. (5), and the algorithm of the VoF is described in Table 6 change rate of the two-phase flow can be calculated by the
[41]. equations below: mlv means the evaporation rate of the fluid
and mvl , means the condensation rate of the fluid, as shown in
∂α q Table 7 [12].
+ ∇ • α q vq = 0 (4)
∂t
4.1.2 Grid dependency of CFD
∑α q =1. (5) In the CFD simulation, grid dependency is one of the most
important factors affecting the results. One of the problems
caused by the grid is expressed in Fig. 8. As shown in Fig. 8,
To predict the phase change rate inside the heat pipe, many the liquid–vapor interface is not smoothly connected; this is
studies have applied the simplified energy equation and Clau- why the interface is calculated based on the grid and nodes.
sius-Claperyon relation to the simulation. The Clausius– For this reason, if the size of the grid is too big to chase the
Clapeyron relation defines the line separating the two-phase liquid–vapor interface, the numerical analysis is diverged or
between the liquid and vapor on the pressure-temperature fails to get correct results [12].
diagram, which is shown in Eq. (6) [34]. Fig. 9 shows another problem caused by the grid. This situa-
tion commonly occurs when the grid is bigger than the size of
dp L the bubble. and it is caused that the solver fails to catch the
= . (6)
dT T ( vv − vl ) temperature of phase change at nodes which are the evapora-
tion of Tl > Tsat _sat and condensation of Tl < Tsat [12].
In the early stages of simulating heat pipes, numerous stud- As shown in Figs. 9 and 10, the solver can’t catch the behav-
ies have opted to utilize user defined functions (UDF) to incor- ior of the interface between phases, the local pressure drop,
porate evaporation/condensation models into their simulations. etc., if the generated grid is too coarse. To prevent these prob-
UDFs are specific functions within Fluent, a commercial analy- lems, the user conducts the grid independence test using a
sis program, that facilitate the application of evapora- number of models which have grids with different sizes and
tion/condensation models based on energy balance and the then finds the optimal number of grids by checking data includ-
Clausius-Clapeyron equation to each grid of the computational ing the temperature gradient, pressure drop or other results
model. This approach allows users to model complex physical [12].
phenomena in their simulations, even if the commercial soft- In addition, the mesh qualities and its types can affect the
ware lacks support for such physics in its basic settings. The simulation results. The types of mesh are generally distin-
versatility of UDFs empowers users to define the thermody- guished by the shapes which are represented by tetrahedron
namic states of the working fluid accurately in response to and hexagonal mesh. In the early day of CFD simulation, the

2600
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

(a) Eulerian VOF SLIC (simple line interface calculation) method (b) Eulerian VOF PLIC (piecewise linear interface calculation) method

Fig. 8. Interface between the liquid and vapor of the VOF model [12].

(a) Failure to chase bubble generation (b) Succeeds in chasing bubble generation

Fig. 9. Problem of CFD for vapor generation caused by the grid size [12].

correctly. Moreover, the methods mentioned above consume

more computing resources, requiring more time. Therefore,
considering the advantages of the simulation for saving design
time, the increase in the time of the simulation leads to another
problem [12].

4.2 CFD modeling and simulation for heat pipes

The thermosyphon and the pulsating heat pipe (PHP) offer
distinct advantages when it comes to analysis and experimen-
tal verification through computational fluid dynamics (CFD)
modeling. These advantages stem from the absence of a wick
Fig. 10. The simulation results of the interface between the liquid and vapor
according to a number of elements [12].
structure in both heat pipe types, which simplifies the modeling
process and allows for a more focused examination of other
critical factors [40-48].
bad qualities of mesh which were caused by the significant In the case of the thermosyphon, its wickless design elimi-
different distance between nodes affected that the solver di- nates the need to account for complex capillary forces and
verged or the unusable difference are expressed in the results. wick characteristics in the CFD model. Instead, the analysis
However, As developing a new method for the solver and ap- can concentrate on the fundamental mechanisms of heat
plying the relaxation factor, the dependence of mesh qualities transfer involving the working fluid, gravity-driven flow, and
has been gradually decreasing and the robustness of solver phase change phenomena. This streamlined approach not
has been insured [12, 42]. only makes the modeling process more straightforward but
At present, to fix these problems, moving mesh is adopted in also enhances the accuracy and reliability of the results. As a
the interface of the two-phase and level-set (LS) function or the result, thermosyphons have become a popular choice for re-
front-tracking (FT) method applied in the simulation. Neverthe- searchers and engineers seeking to study heat pipe behavior
less, there are problems with chasing the liquid–vapor interface and optimize their performance through CFD simulations [49-

2601
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Fig. 11. The number of CFD analytical papers according to the types of Fig. 12. The comparison of temperature distributions and thermal resis-
HPs. tance between the experiment and simulation [43].

56]. They used the 2D model and simulated a two-phase flow and
Similarly, the pulsating heat pipe (PHP) shares the advan- phase change using the VoF model and the evapora-
tage of being wickless, presenting an opportunity for efficient tion/condensation model (UDF). Then, they validated the simu-
and focused CFD modeling. With no wick structure to consider, lation results by comparing the experimental results which
the simulation can emphasize the dynamics of fluid flow and were the thermal resistances of the thermosyphon according to
phase change, enabling a deeper understanding of the heat the heating power. As shown in Fig. 12, Tevap is the tempera-
transfer processes within the PHP. This advantage becomes ture in the evaporator, Tadiabatic is the temperature in the adia-
particularly valuable in PHPs, known for their intricate flow batic section, Tcond is the temperature in the condenser, and
patterns and pulsating behavior. By concentrating on the es- subscripts of CFD and EXP mean the CFD simulation and
sential physics involved in these wickless heat pipes, CFD experiment respectively. Their results showed simulation mod-
simulations of PHPs can yield valuable insights into their per- el had good agreement with the experimental data within 1.93
formance and potential areas for improvement [40-48]. ~ 13.07 % error and the simulation results could show the two-
The prevalence of analysis and simulation efforts centered phase flow pattern inside the thermosyphon.
around thermosyphons and PHPs can be attributed to the con- Asmaie et al. [44] developed a simulation model for heat
venience and accuracy offered by wickless heat pipes in the pipes using water and CuO/water nanofluid. They employed
CFD context. Engineers and researchers can efficiently study the phase change model through user defined functions (UDF).
the thermal characteristics and behavior of these heat pipes The simulation and experimentation were carried out on the
without the additional complexities introduced by wick struc- heat pipes using water and nanofluid. Subsequently, the au-
tures, which are present in other types of heat pipes [24, 33, 36, thors compared the simulation results to the experimental data
40-62]. and found that they exhibited a high degree of agreement. The
Fig. 11 further highlights this trend, showing a notable con- maximum temperature difference observed was approximately
centration of research activities around thermosyphons and 11 °C at the end of the evaporator. Following the investigation,
PHPs in the context of CFD modeling. This visualization un- the researchers utilized the simulation to optimize thermosy-
derscores the practicality and effectiveness of analyzing wick- phons. The outcomes revealed that the maximum heat flux
less heat pipes through CFD, as evidenced by the significant achieved with CuO/water nanofluid was around 46 % higher
body of work focused on these two types of heat pipes [24, 33, than that attained with water.
36, 40-62]. Seo et al. [45] modeled and experimented on the conven-
tional thermosyphon which is 1000 mm in length, 20 mm in
4.2.1 CFD modeling of thermosyphons outer diameter and 16 mm in inner diameter. To simulate the
The CFD analysis for thermosyphons has been conducted thermosyphon, they used the VoF model and phase change
with various boundary conditions. Cases in which analyses of model by UDF. The experiment was also conducted to visual-
thermosyphons were conducted are shown in Table 8. As ize the flow patterns in the thermosyphon. After the simulation
shown in Table 8, in the early days of the simulations of ther- and experiment, the results of both flow patterns were com-
mosyphons, the evaporation/condensation model was applied pared, and it showed that the patterns of the bubbly flow and
for the analysis by the user-defined function (UDF), and then the slug/plug flow were observed at the evaporator section and
the Lee model was adopted as the phase change model after the churn flow was dominated at the adiabatic and condenser
being embedded in the program. section.
Fadhl et al. [43] carried out a numerical simulation of a Zhongchao et al. [46] experimented on and simulated the
closed thermosyphon which used water as the working fluid. conventional thermosyphon which was 9.52 mm in external

2602
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Table 8. CFD analysis cases of thermosyphons.

Type of HP Author/year Analytical method Summary

VoF Working fluid: water
Conventional thermosyphon Fadhl et al./2013
evaporation/condensation model (UDF) analysis: validated in the experiment [43]
VoF Working fluid: water, CuO
Conventional thermosyphon Asmaie et al./2013
evaporation/condensation model (UDF) analysis: validated in the experiment [44]
Working fluid: water
VoF
Conventional thermosyphon Seo et al./2016 comparing the flow pattern and results between the
evaporation/condensation model (UDF)
analysis and the experiment [45]
Working fluid: water
Zhongchao et al./ VoF comparing the thermal performance of the
Conventional thermosyphon
2018 evaporation/condensation model (UDF) thermosyphon between the analysis and the
experiment [46]
Working fluid: water
comparing the thermal performance and flow
Loop thermosyphon Sara/2019 VoF, Lee model
pattern of the thermosyphon between the
analysis and the experiment [47]
Working fluid: water
comparing the thermal performance and flow
Conventional thermosyphon Alaa et al./2019 VoF, Lee model
pattern of the thermosyphon between the analysis
and the experiment [48]
Working fluid: water
VoF comparing the thermal performance and flow pattern
Conventional thermosyphon Ayad et al./2020
evaporation/condensation model (UDF) of the thermosyphon between the analysis and the
experiment under various working conditions [49]
Working fluid: R134A, R404A, R407C
comparing the thermal performance and flow pattern
Conventional thermosyphon Adrian et al./2021 VoF, Lee model
of the thermosyphon between the analysis and the
experiment under various working conditions [50]
Working fluid: water
VoF (firstly, 60 s) and eulerian
comparing the thermal performance and flow pattern
Conventional thermosyphon Arat et al./2021 (secondly, 60 s)
of the thermosyphon between the analysis and the
evaporation/condensation model (UDF)
experiment [51]

diameter, 8.32 mm in internal diameter and 100 mm in length.

They adopt the VoF and phase change model by UDF, and the
results of the simulation were validated with the experimental
results. The maximum difference in temperature between the
experiment and simulation was about 2.5 °C, which meant the
numerical results were in good agreement with the experiment.
After the validation of the simulation, they evaluated the ther-
mal performance of the thermosyphon, and the minimum ther-
mal resistance was 0.552 K/W when the heating power was
80 W.
Sara [47] experimentally and numerically studied the loop
thermosyphon under various conditions. In this study, the VoF
model was applied as the two-phase model and the Lee model
was applied as the phase change model. The results of the Fig. 13. The comparison of temperature distributions between the experi-
ment and simulation [47].
simulation were validated by the experiment as shown in Fig.
13. The differences in the temperature between the experiment
and simulation were small to acceptable, but the maximum The results of the flow visualization showed that the bubbly
percentage of the difference in the pressure drop and the flow at the beginning of the experiment and simulation and
maximum percentage of the difference in the temperature drop then the slug/plug flow pattern appeared in the simulation as
between the experiment and simulation were 18.76 % and the heat pipe operated, but the large bubbles could not be
18.5 %, respectively. The flow visualization of the loop thermo- observed in the simulation. Except for the large bubbles, the
syphon was also conducted and the results were compared. simulations of the loop thermosyphon had good agreement

2603
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

with the experimental data. Table 9. CFD analysis cases of PHPs.

Alaa et al. [48] made the model for the copper vertical ther-
Type of
mosyphon which is 16 mm in outer diameter, 14.2 mm in inner HP
Author/year Analytical method Summary
diameter and 400 mm in length to visualize the two-phase flow
VoF
patterns. In the simulation, the VoF model and Lee model were Xu et al. evaporation/ Working fluid: water
adopted, and the simulation was validated by the experiment. PHP
/2013 condensation experiment: absence [52]
Using this modeling, they analyzed the flow patterns and the model (UDF)
behaviors of the two-phase flow in the thermosyphon. Working fluid: water
Ayad et al. [49] conducted the computational evaluation of comparing the results be-
the thermal characteristics of the conventional thermosyphon. tween the vof model
VoF, mixture,
and the mixture model
The author used the ANSYS Fluent, and the VoF and phase PHP
Lin et al. evaporation/
comparing the flow pattern
change model were applied. To validate the model, the results /2013 condensation
and the thermal
model (UDF)
of the simulation were compared with the experimental results characteristics between
of the other research. It was found that the maximum error rate the analysis and the
of its wall temperature was 9.86 % and the maximum error rate experiment [53]
of its thermal resistance was 8.63 %. Using this model, the Working fluid: water
VoF,
singularity: no
thermosyphon which was 22 mm in external diameter, 20.2 Pouryoussefi heat transfer model
PHP evaporation/condensation
mm in internal diameter and 400 mm in length was simulated /2015 between the vapor
model, heat transfer model
with various variables, including the filling ratio of the working and the liquid (UDF)
was applied by UDF [54]
fluid, a supplied heat of 50, 110 and 280 W and the inclination Working fluid: water
angle. Their results showed that the thermal resistance of wick- Single comparing the flow pattern
Wang et al./ VoF,
less heat pipes had a decreasing tendency by increasing the Loop
2016 Lee model
and thermal characteristics
filling ratio and supplied heat, and the bubbly, slug/plug, churn PHP between the analysis and
the experiment [55]
and annular flow patterns with the inclination angle were well
expressed. Working fluid: water
singularity: no evaporation/
Adrian et al. [50] conducted an experiment and simulation of Flat
VoF,
condensation model, heat
thermosyphons for heat exchangers that used R134A, R404A Sedighi et heat transfer model
Plate transfer model was applied
al./2018 between the vapor
and R407C as the working fluids. The CFD simulation was PHP by UDF comparing the flow
and the liquid (UDF)
carried out with the VoF and Lee models. They performed the pattern between the analysis
experiments and simulations with variables including the work- and the experiment [56]
ing fluid, the filling ratio and the supplied heat, and their results Working fluid: R123
VoF,
comparing the flow pattern
showed the visualization of the flow pattern in the thermosy- Vo et al. evaporation/
PHP and thermal characteristics
phon according to the working fluid, the filling ratio and the heat /2020 condensation
between the analysis and
input, and they reported that the computational results were model (UDF)
the experiment [57]
identical to the experimental results for all the tested conditions. Working fluid: SiO2-CuO,
Arat et al. [51] conducted a thermal performance analysis of Al2O3-CuO, water
thermosyphons which were investigated by the experimental VoF,
their results showed the
and numerical methods. They chose the VoF model for the first thermal resistances under
Zufar et al. evaporation/
PHP various conditions are in
60 seconds, the Eulerian model was employed after 60 sec- /2020 condensation
good agreement between
onds, and the phase change model was adopted by UDF. The model (UDF)
the analysis and the
simulation was conducted with various turbulent models which experiment optimal
were LES, k-kl ω, k-ω, SAS, k-ε and re-stress, and then the VF = 60 % [58]
results concerning the temperature were compared with the Working fluid: water
experiment. Their results are shown in Fig. 14, and every dif- comparing the results
ference in temperature between the experiment and the simu- between the analysis and
Mucci et al. VoF,
PHP the experiment analysis:
lation is expressed. Their results showed that the maximum /2021 Lee model
thermal characteristics
absolute temperature difference values were 12.09 K and according to the no. of
26.07 K for difference turbulence models, and these values turns [59]
were similar to those in other studies.
In every case of thermosyphons, the results of the numerical
analysis were validated with the experiment, and the VoF thermosyphons. The cases of the numerical analysis for PHPs
model was applied as the two-phase flow model. are shown in Table 9.
The predominant choice for modeling pulsating heat pipes
4.2.2 CFD modeling of pulsating heat pipes (PHPs) is the volume of fluid (VoF) model. This model is fa-
The CFD analyses for PHPs can be commonly found next to vored for its effective conservation of volume, preventing solver

2604
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

ration/condensation model. In these cases, the volume of the

vapor and liquid, the temperature at the heat source and the
sink were set as the initial conditions, and then they observed
the bubble–liquid behavior. Even if they ignored evapora-
tion/condensation in the CFD analysis, their results showed
that the flow pattern and behavior of the vapor and liquid were
observed well. Except for the previous two cases, the analytical
method and validation are similar to the case of thermosy-
phons.
Wang et al. [55] conducted a numerical and experimental in-
vestigation of single PHPs with a corrugated configuration.
Fig. 14. Difference of temperature between the experiment and simulation They chose the VoF model and the Lee model in the simula-
[51]. tion, and then the simulations were validated by comparing the
thermal resistances between the experiment and simulation.
They evaluated the performance of PHPs, and their results
divergence. However, its disadvantage lies in the discontinu- showed a 28.96 % decrease in the start-up time and 37.57 %
ous nature of its functions, making precise calculations of cur- decrease in the thermal resistance due to the application of a
vature or normal vectors challenging. Consequently, accurately corrugated configuration located at the evaporator.
tracking the phase interface poses occasional challenges. In Sedighi et al. [56] studied flat-plate PHPs with extra branches
the context of computational fluid dynamics (CFD) analysis of in the evaporator where the bubble pump was connected to
PHPs, a common issue arises in accurately tracing the liquid boost the circulation of the fluid, and they conducted a numeri-
film and phase boundary between the internal wall and vapor. cal and experimental investigation. In the simulation, they used
To overcome this challenge, additional tracking interface mod- the VoF model for the two-phase interface and the application
els at the boundary or a denser grid structure might be neces- of the bubble pump was incarnated by the source term in the
sary. Furthermore, confirming the behavior of bubbles gener- momentum equation. Their results showed the extra branches
ated and grown inside the PHP also necessitates an excep- with the bubble pump make the thermal resistance of PHPs
tionally dense grid structure. It is crucial to note that all these decrease by 11 to 20 % and the numerical results could display
methods come at the cost of increased CFD computational the flow patterns in PHPs compared to the experimental results.
time, introducing significant drawbacks from a numerical analy- Vo et al. [57] carried out the experimental and analytical
sis perspective. Striking a balance between accuracy and study of PHPs with R123 as the working fluid. To incarnate the
computational efficiency remains a critical consideration in interface between the vapor and liquid and phase change, they
PHP CFD simulations [12, 41]. used the VoF model and UDF. They validated the simulation
Xu et al. [52] conducted the CFD simulation for flat plate model by comparing the data between the experiment and
PHPs under various conditions. They chose the VoF model for simulation, and the heat transfer rates predicted by the simula-
the two-phase analysis and the evaporation/condensation tion agreed well with the experimental data within 5 %.
model was applied by UDF. Their results show a very clear Zufar et al. [58] investigated the thermal performance of
two-phase flow pattern inside the PHP, including a bubbly flow, PHPs with Al2O3-CuO and SiO2-CuO as the working fluid. The
slug/plug flow and semi-annular/annular flow. Using this model, PHP which was a closed loop type and four-turned had an
they evaluated the tendency of two different configurations of external diameter of 3 mm, an internal diameter of 1.5 mm, and
PHPs which were the conventional flat PHP and flat PHP with a length of 200 mm. To investigate the PHP, they conducted
micro grooves. The results showed that the PHP with micro the experiment and simulation, and their results showed that
grooves was effective for the heat transfer. the PHP with Al2O3-CuO had 57 % lower thermal resistance
Lin et al. [53] simulated and experimented with the PHP, and and the PHP with SiO2-CuO had 34 % lower thermal resis-
they chose the VoF and mixture model as the two-phase tance when the water was used as the working fluid. The nu-
model, and the phase change model was adopted by UDF. merical results were validated by comparing the temperature of
The simulations were conducted with PHPs of different sizes, the evaporator between the experiment and the simulation.
and the heating powers and thermal resistance of PHPs were The results of the simulation could express the flow patterns
compared during the experiment. Their results showed that the and pulsating of working fluid inside the PHP well.
dominant variable that boosts the performance of PHPs was Mucci et al. [59] numerically studied the effect of the number
the inner diameter. of turns on PHPs. Using the VoF and Lee models as the two-
There are noteworthy points that were made through simula- phase and phase change models, respectively, they investi-
tions conducted without the evaporation/condensation model, gated the flow instability and heat transfer characteristics of
which were written by Pouryussefi et al. [54]. Their research PHPs with different numbers of turns. Fig. 15 shows their vali-
adopted the VoF model as the two-phase flow and the heat dation of the simulation. They reported that the numerical
transfer model between the vapor and liquid without the evapo- analysis for PHPs with 23 turns using mesh A used a calcula-

2605
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Fig. 15. The results between the experiment and simulation according to
turns [59].
Fig. 17. Temperature distribution and difference between the experiment
and simulation [39].

thermal conductivity model is applied for the heat transfer, and

the momentum equation of the porous medium is employed for
the Internal flow model which is calculated by Eq. (8). These
modeling techniques, by disregarding phase changes within
the porous medium and heat transfer due to fluid flow, can lead
to significant errors. Therefore, a common modeling approach
involves adopting correction coefficients to adjust errors when
compared to experimental results. Thus, such modeling meth-
ods fundamentally limit the universal application of the analyti-
cal model [22, 23, 60].

Fig. 16. Modeling of wick structure.

⎛μ⎞
∇p = − ⎜ ⎟ vi + C2 ρ v vi . (8)
⎝α ⎠
tion time of 4 to 6 weeks using an Intel Xeon 56 core computer
and its results had good agreement within 15 %. Thus, they 4.2.3.2 Cases of CFD modeling
thought mesh A was acceptable to use. They obtained and The CFD analysis for other types of heat pipes is shown in
analyzed the flow structure, velocity and temperature profiles of Table 10.
PHPs and could conclude that the PHP had the higher stability Annamalai et al. [39] experimentally and numerically studied
with more turns. a conventional heat pipe which was 1000 mm in length, 31 mm
in outer diameter, and 27 mm in inner diameter. For the nu-
4.2.3 CFD modeling of other heat pipes merical analysis, they assumed the wick-screen mesh was fully
4.2.3.1 Modeling of wick structure immersed with the liquid phase and the core of the heat pipe
In these cases, the wick structure is the most difficult point to was placed with vapor. To analyze the wick structure, they
analyze. The wick structure of a heat pipe acts as a porous used the effective thermal conductivity for heat transfer. Their
medium, inducing capillary action to facilitate the return of the results between the experiment and simulation were compared
working fluid. Therefore, the wick plays a crucial role in heat as shown in Fig. 17. In Fig. 17, the left y-axis represent the
pipe performance, and research on the internal flow and heat temperature, and the right y-axis represent the temperature
transfer within the wick has been actively conducted. However, difference percent between the simulation and experiment, and
the pores within the wick are not uniformly arranged in reality, Tevap is the temperature in the evaporator, Tadiabatic is the tem-
and mathematically modeling these irregularities inherently perature in the adiabatic section, Tcond is the temperature in
introduces errors. Moreover, applying such mathematical mod- the condenser, and subscripts of CFD and EXP mean the CFD
els to numerical analysis demands computing resources of a simulation and experiment respectively, and the maximum
magnitude beyond the current capacity [22, 23]. difference of the evaporator temperature is 13.24 % when the
For this reason, effective thermal properties were applied to supplied heat is 50 W, and the simulation temperature of the
most of the simulations for the wick of the heat pipe, which condenser corresponds exactly to the experiment when the
meant the thermal and flow models were applied separately as heat input is 100 W.
shown in Fig. 16. While neglecting phase changes within the Solomon et al. [33] conducted the experiment and simulation
wick, the heat transfer model assumes that the interior of the of heat pipes that were 19.5 mm in outer diameter, 16.5 mm in
porous medium is completely filled with liquid. The effective inner diameter, and 350 mm in length and which had a screen-

2606
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Table 10. CFD analysis cases of heat pipes.

Type of HP Author/year Analytical method Summary

Working fluid: water singularity: no
evaporation/condensation model.
VoF, wick - effective properties,
Conventional heat pipe Annamalai et al./2011 only the vof model was applied
phase interface (UDF)
comparing the thermal characteristics between
the analysis and the experiment [39]
Working rluid: CuO
singularity: vapor flow and the liquid flow were
VoF, wick – effective properties, modeled separately and it can obtain the
Nanofluid heat pipe Solomon et al./2014
phase interface (UDF) pressure drop at the wick and the vapor core.
comparing the thermal characteristics between the
analysis and the experiment [33]
VoF, wick - effective Working fluid: water the CFD results were
Conventional heat pipe Pooyoo et al./2014
properties, phase interface (UDF) validated by other papers [60]
Working fluid: water comparing the flow pattern
Micro channel separate VoF, evaporation/condensation
Yue et al./2018 and the thermal characteristics between the
heat pipe model (UDF)
analysis and the experiment [61]
VoF evaporation/condensation Working fluid: water comparing the results between
Flat grooved heat pipe Kurt/2019
model the analysis and the experiment [62]
Working fluid: water, nanofluid comparing
VoF, capillary force - UDF
Flat grooved heat pipe Wei et al./2020 thermal characteristics between the analysis and
Lee model
the experiment [63]
Working fluid: water comparing the thermal
Annular heat pipe Song et al./2020 VoF, Lee model characteristics between the analysis and the
experiment [25]
Working fluid: acetone, AnBr2
Brahim et al. VoF, evaporation/condensation model
Flat heat pipe optimization for the flat heat pipe through the
/2021 multi-physics analysis
numerical method without the experiment [64]
Working fluid: water comparing the thermal
Hemishperical shell vapor
Ahn et at./2021 VoF, Lee model characteristics between the analysis and the
chamber heat Sink
experiment [26]
Working fluid: water comparing the thermal
VoF, evaporation/condensation
Flattened heat pipe Sanhan et al./2021 characteristics between the analysis and the
model multi-physics analysis
experiment [65]

mesh wick with a thickness of 1 mm. In the model for the simu- which were applied by the UDF, and the results of the simula-
lation, they used the same methods as Arul Sevan where they tion were validated with the experimental results. Their results
modeled the wick with the liquid and the vapor core separately, showed that the distribution of the wall temperature and liquid
and the simulation validated the previous works. They con- fraction were the key parameters affecting the cooling capac-
ducted a simulation with variables with the supplied heat and ity.
the working fluid. Their results showed that the pressure drop Kurt [62] modeled the flat grooved heat pipe with a phase
and the vapor velocity when the working fluid was Cu-water change heat transfer by COMSOL and MATLAB. The author
was higher than when the DI water had a heat input of 350 W could apply the phase change function in the simulation which
because the nanoparticle was coated in the wick, and it en- was calculated by MATLAB, and then the COMSOL solved the
hanced the nucleation. momentum and energy equations.
Pooyoo et al. [60] conducted a simulation of heat pipes with Wei et al. [63] studied the flow and heat transfer mechanism
screen-mesh wicks, and they used the same method as the of Ω-shaped microchannel heat pipes under zero gravity condi-
above two papers. Their results for the wall temperature were tions that had an outer diameter of 15 mm, a vapor core diame-
in good agreement with the experiment, and they illustrated the ter of 10 mm and a length of 648 mm. To incarnate the analy-
vapor velocity, centerline pressure and wall temperature of sis model, they also used the VoF and Lee model as the two-
heat pipes according to the operating conditions. phase flow and phase change models, respectively. They
Yue et al. [61] studied the microchannel separate heat pipe compared the thermal resistance between the experiment and
with R22 as a working fluid, and they conducted the experi- simulation to validate the results of the simulation, and their
ment and simulation with changes in the filling ratio. In the results could predict the thermal characteristics of the micro-
numerical simulation, they adopted the VoF and Lee model, channel heat pipe well.

2607
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

Song et al. [25] and Ahn et al. [26] carried out the simulation phase flow and phase change in heat pipes demands
for special heat pipes that have a short evaporator space with substantial computing resources and time, making the
a large condenser area. To embody their devices in the simula- process computationally expensive [12].
tion, they chose the VoF model and Lee model, and the simu- 3) Boiling process simulation challenges: Simulating the boil-
lation was validated with the experiment. Their results showed ing process within the heat pipe is a formidable task, and
that the circulation of the working fluid could be observed, and in many instances, it is nearly impossible to model effec-
the vapor was well condensed at a large condenser region. tively using CFD [12].
In the case written by Brahim et al. [64], the optimization of 4) Adopting evaporation/condensation model: To address
flat heat pipes which were used for electronic chip cooling was certain challenges, the widely used Lee model, coupled
conducted by COMSOL without an experimental investigation. with the volume of fluid (VoF) model, is adopted. However,
However, in the simulation, they adopted the method wherein accurately defining the magnitude of the evapora-
the thermodynamic properties, phase and flow of the working tion/condensation rate remains a significant challenge [12,
fluid were in the simulation at every time step. 43, 47].
Sanhan et al. [65] conducted a numerical study of flattened 5) Grid dependency and grid independent testing: The analy-
heat pipes for the CPU and GPU cooling that were 2, 3 and 4 sis heavily relies on grid resolution, leading to potential
mm in flattened thickness, 6 mm in diameter and 200 mm in simulation failure or divergence if the solver fails to capture
length. They adopted the VoF and phase change model in the interface between the liquid and vapor phases. As a re-
COMSOL to solve the flow inside the heat pipe. Then, the tem- sult, rigorous grid-independent testing is essential, which
perature profile calculated by the simulation had good agree- consumes additional time and resources [12].
ment compared with the experimental data, with a standard 6) Validation through experimental results: Given the com-
deviation of 1.85 %. Using this model, they found the best flat- plexity and limitations of numerical simulations, research-
tened thickness to be 2.5 mm. ers frequently resort to validating their findings primarily
based on experimental data, which ensures the reliability
and credibility of the results [25, 33, 39, 58].
4.3 Current status and challenge CFD simula-
Addressing these critical points in heat pipe simulations re-
tion for heat pipes
quires continuous advancements in numerical modeling tech-
In the areas of heat pipe simulations, the majority of studies niques, as well as a deeper understanding of the underlying
have focused on wickless heat pipes, such as thermosyphons physics and phenomena at play. Collaborative efforts between
and PHPs. In cases involving wicked heat pipes, researchers experimental investigations and numerical simulations play a
have incorporated effective thermal conductivity and the capil- vital role in refining and validating the simulation models, ulti-
lary function of the wick structure to represent heat transfer and mately leading to more accurate and reliable predictions.
capillary force phenomena [39]. However, a significant chal- Overcoming these challenges will pave the way for the effec-
lenge arises due to the absence of a well-defined magnitude tive application of heat pipes in diverse engineering and ther-
for the evaporation/condensation rate without experimental mal management scenarios, promising innovative solutions to
data. Consequently, simulations demand rigorous validation address complex heat transfer challenges [25, 33, 39, 47, 51,
through comparison with experimental results to ensure accu- 59].
racy and reliability. As a result, the very essence of simulation, The advent of quantum computing offers a potential solution
which revolves around time and cost savings, becomes tar- to these challenges. Quantum computing's unparalleled com-
nished, hindering progress in the modeling and simulation of putational power could accelerate the simulation process and
heat pipes [21]. enable the handling of more intricate models with greater accu-
Nevertheless, with the potential advent of quantum comput- racy. Complex heat transfer phenomena, including heat trans-
ing, a transformative shift in computational power and capabili- fer in wick structures and boiling processes, could be simulated
ties is on the horizon. Quantum computing, with its ability to with unprecedented precision, paving the way for more reliable
perform complex calculations at an exponentially faster pace predictions and design optimizations [66, 67].
than classical computing, holds great promise for addressing Furthermore, quantum computing's capabilities could allevi-
the challenges faced in heat pipe simulations [63, 64]. ate the computational resource and time constraints faced in
Several noteworthy problems impeding the two-phase analy- traditional CFD simulations. Tasks that currently demand ex-
sis for heat pipes using computational fluid dynamics (CFD) tensive computing resources and time could be executed
have been identified: swiftly, leading to significant time and cost savings in the simu-
1) Heat transfer in wick structure: Modeling heat transfer lation process [66, 67].
within the wick structure necessitates additional complexi- As quantum computing technology continues to advance
ties in the simulation. Furthermore, the representation of and become more accessible, its integration into the field of
capillary forces is often highly simplified, potentially limiting heat pipe simulations holds the potential to revolutionize our
the accuracy of the results [33, 39, 63]. understanding of heat transfer phenomena and enhance the
2) Computational resources and time: The simulation of two- performance of heat pipe systems. Collaborative efforts be-

2608
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

tween quantum computing experts and researchers in thermal computational power and capabilities that can revolutionize the
management will be instrumental in harnessing this transfor- field of heat pipe simulations and unlock new frontiers in engi-
mative technology to overcome the critical points faced in CFD neering and thermal management applications.
simulations and unlock new frontiers in heat pipe engineering
and design [66, 67].
Acknowledgments
This research was supported by Basic Science Research
5. Conclusions Program through the National Research Foundation of Korea
In this study, we examined various methodologies for the (NRF) funded by the Ministry of Education (grant number:
modeling and simulation of heat pipes, including thermal resis- NRF-2022R1I1A3054588).
tance and mathematical models, and CFD simulations. While
these approaches offer valuable insights into heat pipe behav-
Nomenclature------------------------------------------------------------------
ior, they often rely on simplified methods and are heavily vali-
dated with experimental data. C2 : Inertial resistance factor
The thermal resistance model with the two-phase heat trans- Dh : Hydraulic diameter, m
fer model demonstrates advantages such as lower computa- E : Energy
tional resource requirements compared to CFD simulations, F : Correction factor
while still considering the two-phase heat transfer inside the gi : Gravitational acceleration, m/s2
heat pipe. However, this model faces challenges when at- hv : Specific enthalpy of vapor, J/kg
tempting to visualize the flow pattern and behavior of the work- hl : Specific enthalpy of liquid, J/kg
ing fluid due to its reliance on two-phase heat transfer correla- h : Convective heat transfer coefficient, W/m2-K
tions, which only consider the void fraction. lv : Latent heat of fluid, J/kg
To address the need for verifying flow patterns, the CFD m l →v : Mass transfer rate from the liquid to vapor phase, kg/s
model is employed to analyze the flow pattern and thermal m v →l : Mass transfer rate from the vapor to liquid phase, kg/s
characteristics of heat pipes. Initially, applying the evapora- Pr : Prandtl number (dimensionless)
tion/condensation model in the simulation required defining the p : Static pressure, N/m2
relationship and equations in the program. Rth : Thermal resistance, °C/W
Despite these attempts, significant challenges persist in the Re : Reynolds number (dimensionless)
modeling and simulation of heat pipes. Simplified effective S : Correction factor
thermal properties are used to represent heat transfer in po- Sh : Volumetric heat source
rous media, requiring separate solution of momentum and heat T : Temperature, °C
transfer in the wick structure. Additionally, simulations often t : Time, sec
require validation through experimental data, as defining the ui : Velocity tensor
evaporation/condensation rate in the program can be complex, v : Velocity of fluid, m/s
and boiling is not adequately embodied in the wick structure. vq : Velocity of fluid in phase q, m/s
Grid-independent testing is crucial, as simulations may fail to v : Specific volume, m3/kg
accurately capture evaporation, condensation, and fluid behav- x : Quality
ior within heat pipes if the grid quality and size are inadequate. xi : Direction tensor
To address these issues, researchers have explored new
methods, such as moving meshes and mixture models. How-
Greek
ever, these methods demand significant computational re-
sources, time, and expertise. Consequently, most modeling α : Permeability, H/m
and simulation attempts have focused on wickless heat pipes, αq : Volume fraction of phase
such as thermosyphons and PHPs. λl : Thermal conductivity of liquid phase, W/m·°C
In conclusion, full modeling and simulation of heat pipes re- μ : Dynamic viscosity, kg/m·s
main challenging, and simulations have often relied on simpli- ν : Kinematic viscosity, m2/s
fied methods. Despite offering valuable insights, they consume ρ : Density, kg/m3
considerable time and computational resources, limiting one of τ ij : Stress tensor
the primary advantages of simulations - saving time in design
and exploration without physical experiments. Thus, compre- Subscripts
hensive analysis of the behavior of working fluid in heat pipes
will likely continue to face challenges unless groundbreaking cv : Convective boiling
methods for phase change in heat pipes and wick structures eff : Effective
are proposed. Advancements in quantum computing hold g : Gas
promise to address these challenges, offering transformative i : Liquid

2609
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

LO : Liquid only performances and characteristics of thermosyphon heat pipe

NcB : Nucleate boiling using alumina nanofluids, Heat Mass Transfer, 57 (8) (2021)
th : Thermal 1275-1287.
TP : Two-phase [17] C. R. Kharangate and I. Mudawar, Review of computational
sat : Saturation studies on boiling and condensation, International Journal of
v : Vapor Heat and Mass Transfer, 108 (2017) 1164-1196.
VO : Vapor only [18] H. M. Maghrabie et al., Numerical simulation of heat pipes in
different applications, International Journal of Thermofluids, 16
(2022) 100199.
References
[19] Jia-qiang E., Y. Li and J. Gong, Function chain neural network
[1] B. Zohuri, Basic principles of heat pipes and history, Heat Pipe prediction on heat transfer performance of oscillating heat pipe
Design and Technology, Springer International Publishing, based on grey relational analysis, J. Cent. South Univ. Tech-
Cham (2016) 1-41. nol., 18 (5) (2011) 1733-1737.
[2] P. Nemec, Gravity in Heat Pipe Technology, IntechOpen [20] R. V. Rao and K. C. More, Optimal design of the heat pipe
(2017). using TLBO (teaching–learning-based optimization) algorithm,
[3] F. Li et al., Two-phase flow visualization and heat transfer Energy, 80 (2015) 535-544.
characteristics analysis in ultra-long gravity heat pipe, Energies [21] Z. Wang et al., Advanced big-data/machine-learning tech-
16 (12) (2023) 4709. niques for optimization and performance enhancement of the
[4] B. Zohuri, Other types of heat pipes, Heat Pipe Design and heat pipe technology – A review and prospective study, Ap-
Technology, Springer International Publishing, Cham (2016) plied Energy, 294 (2021) 116969.
431-449. [22] P. Gang, F. Huide, Z. Tao and J. Jie, A numerical and ex-
[5] M. R. Doshi and G. Y. Eastman, Osmotic heat pipe: problems perimental study on a heat pipe PV/T system, Solar Energy, 85
and promises, Letters in Heat and Mass Transfer, 4 (6) (1977) (5) (2011) 911-921.
429-436. [23] J. S. Lee et al., 1-D modeling of two phase flow process in
[6] T. B. Jones and M. P. Perry, Electrohydrodynamic heat pipe concentric annular heat pipe and experimental investigation,
experiments, Journal of Applied Physics, 45 (5) (1974) 2129- Processes, 10 (3) (2022) 493.
2132. [24] M. Ghajar, J. Darabi and N. Crews, A hybrid CFD-
[7] W. G. Anderson, Different types of heat pipes, Thermal & Flu- mathematical model for simulation of a MEMS loop heat pipe
ids Analysis Workshop (TFAWS 2014), Cleveland, USA (2014). for electronics cooling applications, J. Micromech. Microeng.,
[8] T. Kaya and J. Goldak, Numerical analysis of heat and mass 15 (2) (2005) 313-321.
transfer in the capillary structure of a loop heat pipe, Interna- [25] E. H. Song, K.-B. Lee and S.-H. Rhi, Thermal and flow simu-
tional Journal of Heat and Mass Transfer, 49 (17-18) (2006) lation of concentric annular heat pipe with symmetric or asym-
3211-3220. metric condenser, Energies, 14 (11) (2021) 3333.
[9] H. Y. Noh and S. J. Kim, Numerical simulation of pulsating heat [26] J. H. Ahn, S. H. Rhi, J. S. Lee and K. B. Kim, Thermal investi-
pipes: parametric investigation and thermal optimization, En- gations of hemispherical shell vapor chamber heat sink, Ener-
ergy Conversion and Management, 203 (2020) 112237. gies, 15 (3) (2022) 1161.
[10] P. J. Brennan and E. J. Kroliczek, Heat Pipe Design Hand- [27] J. P. Longtin, B. Badran and F. M. Gerner, A one-dimensional
book, B&K Engineering, Inc, USA, N81-70113 (1979). model of a micro heat pipe during steady-state operation,
[11] P. Wallin, Heat pipe, selection of working fluid, MVK160 Heat Journal of Heat Transfer, 116 (3) (1994) 709-715.
and Mass Trasfer Project Report, Lund, Sweden (2012) 1-7. [28] J. E. Hansel, R. A. Berry, D. Andrs, M. S. Kunick and R. C.
[12] D. Sun, J. Xu and Q. Chen, Modeling of the evaporation and Martineau, Sockeye: A one-dimensional, two-phase, com-
condensation phase-change problems with FLUENT, Numeri- pressible flow heat pipe application, Nuclear Technology, 207
cal Heat Transfer, Part B: Fundamentals, 66 (4) (2014) 326- (7) (2021) 1096-1117.
342. [29] A. G. Olabi, S. Haridy, E. Sayed, M. Radi, A. Alami, F. Zway-
[13] S. Khandekar, X. Cui and M. Groll, Thermal performance yed, T. Salameh and M. Abdelkareem, Implementation of arti-
modeling of pulsating heat pipe by atificial neural network, Pro- ficial intelligence in modeling and control of heat pipes: A re-
ceedings of 12th International Heat Pipe Conference, Moscow, view, Energies, 16 (2) (2023) 760.
Russia (2002) 215-219. [30] V. M. Patel and H. B. Mehta, Thermal performance prediction
[14] H. Mroue, J. B. Ramos, L. C. Wrobel and H. Jouhara, Ex- models for a pulsating heat pipe using artificial neural network
perimental and numerical investigation of an air-to-water heat (ANN) and regression/correlation analysis (RCA), Sādhanā, 43
pipe-based heat exchanger, Applied Thermal Engineering, 78 (11) (2018) 184.
(2015) 339-350. [31] X. Wang, B. Li, Y. Yan, N. Gao and G. Chen, Predicting of
[15] Z. Mahboodi and H. Nemati, Experimental analysis of inclined thermal resistances of closed vertical meandering pulsating
heat pipe thermal resistancen, IJHT, 33 (4) (2015) 155-160. heat pipe using artificial neural network model, Applied Ther-
[16] L. A. Shuoman, M. Abdelaziz and S. Abdel-Samad, Thermal mal Engineering, 149 (2019) 1134-1141.

2610
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

[32] A. Faghri, M.-M. Chen and M. Morgan, Heat transfer charac- Connecticut, USA (2019).
teristics in two-phase closed conventional and concentric an- [48] A. A. B. Temimy and A. A. Abdulrasool, CFD modelling for
nular thermosyphons, Journal of Heat Transfer, 111 (3) (1989) flow and heat transfer in a closed thermosyphon charged with
611-618. water – A new observation for the two phase interaction, IOP
[33] A. Brusly Solomon, K. Ramachandran, L. Godson Asirvatham Conf. Ser.: Mater. Sci. Eng., 518 (3) (2019) 032053.
and B. C. Pillai, Numerical analysis of a screen mesh wick heat [49] A. M. Al Jubori and Q. A. Jawad, Computational evaluation of
pipe with Cu/water nanofluid, International Journal of Heat and thermal behavior of a wickless heat pipe under various condi-
Mass Transfer, 75 (2014) 523-533. tions, Case Studies in Thermal Engineering, 22 (2020) 100767.
[34] K. Ali, Development of an innovative cooling concept for tur- [50] Ł. Adrian, S. Szufa, P. Piersa and F. Mikołajczyk, Numerical
bofan engines, Master Thesis, KTH School of Industrial Engi- model of heat pipes as an optimization method of heat ex-
neering and Management, Sweden (2018). changers, Energies, 14 (22) (2021) 7647.
[35] SIEMENS, Simcenter Amesim software Library User’s Guides [51] H. Arat, O. Arslan, U. Ercetin and A. Akbulut, A comprehen-
(ver. 2020.2), SIEMENS (2020). sive numerical investigation of unsteady-state two-phase flow
[36] S. Arabnejad, R. Rasoulian, M. B. Shafii and Y. Saboohi, in gravity assisted heat pipe enclosure, Thermal Science and
Numerical investigation of the performance of a u-shaped pul- Engineering Progress, 25 (2021) 100993.
sating heat pipe, Heat Transfer Engineering, 31 (14) (2010) [52] D. Xu, T. Chen and Y. Xuan, Thermo-hydrodynamics analysis
1155-1164. of vapor–liquid two-phase flow in the flat-plate pulsating heat
[37] A. R. Rohani and C. L. Tien, Steady two-dimensional heat pipe, International Communications in Heat and Mass Transfer,
and mass transfer in the vapor-gas region of a gas-loaded heat 39 (2012) 504-508.
pipe, Journal of Heat Transfer, 95 (3) (1973) 377-382. [53] Z. Lin, S. Wang, R. Shirakashi and L. W. Zhang, Simulation of
[38] C. Harley and A. Faghri, Two-dimensional rotating heat pipe a miniature oscillating heat pipe in bottom heating mode using
analysis, Journal of Heat Transfer, 117 (1) (1995) 202-208. CFD with unsteady modeling, International Journal of Heat and
[39] S. Annamalai and V. Ramalingam, Experimental investigation Mass Transfer, 57 (2) (2013) 642-656.
and CFD analysis of a air cooled condenser heat pipe, Therm. [54] S. M. Pouryoussefi and Y. Zhang, Numerical investigation of
Sci., 15 (3) (2011) 759-772. chaotic flow in a 2D closed-loop pulsating heat pipe, Applied
[40] N. Hasan, I. G. S. Tibba, F. T. Mosisa and A. Daniel, Ground Thermal Engineering, 98 (2016) 617-627.
tunnel as renewable energy utilization of ground energy as a [55] J. Wang, H. Ma, Q. Zhu, D. Yiwei and K, Yue, Numerical and
source and sink for building heating and cooling, AIP Confer- experimental investigation of pulsating heat pipes with corru-
ence Proceedings, the 1st International Conference on Me- gated configuration, Applied Thermal Engineering, 102 (2016)
chanical and Materials Science Engineering: Innovation and 158-166.
Research-2018, Maharashtra, India (2018). [56] E. Sedighi, A. Amarloo and B. Shafii, Numerical and experi-
[41] P. Liaw, Y. S. Chen, H. M. Shang, M. Shih, D. Doran and E. mental investigation of flat-plate pulsating heat pipes with extra
Steward, Numerical investigation of slag behavior with com- branches in the evaporator section, International Journal of
bustion/evaporation/breakup/VOF models for solid rocket mo- Heat and Mass Transfer, 126 (2018) 431-441.
tors, 31st Joint Propulsion Conference and Exhibit, American [57] D. T. Vo, H. T. Kim, J. Ko and W. K. Bang, An experiment and
Institute of Aeronautics and Astronautics, San Diego, CA, USA three-dimensional numerical simulation of pulsating heat pipes,
(1995). International Journal of Heat and Mass Transfer, 150 (2020)
[42] P. M. Knupp, Algebraic mesh quality metrics, SIAM J. Sci. 119317.
Comput., 23 (1) (2001) 193-218. [58] M. Zufar, P. Gunnasegaran, K. C. Ng and H. Mehta, Numeri-
[43] B. Fadhl, L. C. Wrobel and H. Jouhara, Numerical modelling cal and experimental investigations of hybrid nanofluids on
of the temperature distribution in a two-phase closed thermo- pulsating heat pipe performance, International Journal of Heat
syphon, Applied Thermal Engineering, 60 (1-2) (2013) 122-131. and Mass Transfer, 146 (2020) 118887.
[44] L. Asmaie, M. Haghshenasfard, A. Mehrabani-Zeinabad and [59] A. Mucci, F. K. Kwame, J. Chetwynd-Chatwin, M. Y. Ha and J.
M. Nasr Esfahany, Thermal performance analysis of nanofluids K. Min, Numerical investigation of flow instability and heat
in a thermosyphon heat pipe using CFD modeling, Heat Mass transfer characteristics inside pulsating heat pipes with differ-
Transfer, 49 (5) (2013) 667-678. ent numbers of turns, International Journal of Heat and Mass
[45] J. Seo and J. Y. Lee, CFD Analysis and visualization of the Transfer, 169 (2021) 120934.
two phase flow in a thermosyphon for a passive heat removal [60] N. Pooyoo, S. Kumar, J. Charoensuk and A. Suksangpanom-
system of a nuclear power plant, Transactions of the Korean rung, Numerical simulation of cylindrical heat pipe considering
Nuclear Society Spring Meeting, Jeju, Korea (2016). non-darcian transport for liquid flow inside wick and mass flow
[46] Z. C. Zhao, Y. Zhang, Y. Zhang, Y. Zhou and H. Hu, Numeri- rate at liquid–vapor interface, International Journal of Heat and
cal study on the transient thermal performance of a two-phase Mass Transfer, 70 (2014) 965-978.
closed thermosyphon, Energies, 11 (6) (2018) 1433. [61] C. Yue, Q. Zhang, Z. Zhai and L. Ling, CFD simulation on the
[47] S. A. Kloczko, Experimental investigation and numerical simu- heat transfer and flow characteristics of a microchannel sepa-
lation of loop thermosyphons, Master's Theses, University of rate heat pipe under different filling ratios, Applied Thermal En-

2611
 Journal of Mechanical Science and Technology 38 (5) 2024 DOI 10.1007/s12206-024-0437-x

gineering, 139 (2018) 25-34. Soek-Ho Rhi is a Professor in the

[62] C. Kurt, Three-dimensional modeling of heat transfer and fluid School of Mechanical Engineering at
flow in a flat-grooved heat pipe, Master Thesis, Bilkent Univer- Chungbuk National University. He
sity, Turkey (2019). received his Ph.D. in University of Ottawa.
[63] A. Wei, X. Ren, S. Lin and X. Zhang, CFD analysis on flow His research interests are heat pipe,
and heat transfer mechanism of a microchannel Ω-shape heat thermoelectric conversion, heat transfer,
pipe under zero gravity condition, International Journal of Heat heat exchanger etc.
and Mass Transfer, 163 (2020) 120448.
[64] T. Brahim and A. Jemni, CFD analysis of hotspots copper
metal foam flat heat pipe for electronic cooling applications, In- Sun-Kook Kim is currently a manager of
ternational Journal of Thermal Sciences, 159 (2021) 106583. global products R&D team, LS Electronic
[65] W. Sanhan, K. Vafai, N. Kammuang-Lue, P. Terdtoon and P. Co., Ltd. He received B.S. and M.S. de-
Sakulchangsatjatai, Numerical simulation of flattened heat pipe grees in Mechanical Engineering from
with double heat sources for CPU and GPU cooling application Chung-Buk National University.
in laptop computers, Journal of Computational Design and En-
gineering, 8 (2) (2021) 524-535.
[66] D. Jaksch, P. Givi, A. Daley and T. Thomas, Variational quan-
tum algorithms for computational fluid dynamics, AIAA Journal,
61 (5) (2022) 1885-1894.
[67] G. Malinverno and J. B. Alberto, Quantum computing CFD
simulations: myth and reality, Proceeding of 1st Spanish Fluid
Mechanics Conference, Cadiz, Spain (2022).

Ji-Su Lee is Ph.D. student in Mechanical

Engineering, Chung-Buk National Uni-
versity. He received his B.S. and M.S.
degrees in Mechanical Engineering,
CBNU. His research interest is heat and
mass transfer in heat pipe.

2612