See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/305608515

CFD simulation of a Pulsating Heat Pipe using ANSYS FLUENT

Technical Report · July 2016

CITATIONS READS
6 9,402

1 author:

Umberto Costa Bitencourt

University of Minnesota Twin Cities
3 PUBLICATIONS 6 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Pulsating heat pipe View project

All content following this page was uploaded by Umberto Costa Bitencourt on 24 July 2016.

The user has requested enhancement of the downloaded file.

 CFD SIMULATION OF A PULSATING HEAT PIPE USING
ANSYS FLUENT

A PAPER SUBMITTED AS A PART OF A RESEARCH

PROJECT ENTITULED “DEVELOPMENT OF A PULSATING
HEAT PIPE FOR ENHANCED HEAT TRANSFER”
By
UMBERTO COSTA BITENCOURT
under orientation of
PROF. ALI KHOUNSARY

DEPARTMENT OF PHYSICS
ILLINOIS INSTITUTE OF TECHNOLOGY
JULY 2016
 1. INTRODUCTION
A pulsating heat pipe (PHP) [1] is a passive and extremely efficient way of exchanging heat
between a hot source and a cold one. The simulation of a 2D single loop pulsating heat pipe using
ANSYS FLUENT 17.1 is proposed on this article. After doing a lot of research on the theme it can
be noticed that not many articles actually explain how to simulate a pulsating heat pipe. For this
reason, the main point of this article is to make anyone to be able to reproduce this simulation.
This approach starts with the simplifying assumptions to carry out the simulation, and then its
procedure is shown step by step which are also explained physically as they are introduced.

2. PHYSICAL CHARACTERISTICS AND ASSUMPTIONS

The pulsating heat pipe analyzed here is a pretty simple one. First, it has only one loop, second the
analysis is carried out using a 2D model. The filling ratio is 50 %, which is the ratio between the
volume of liquid used and the total volume, but this number can be changed very easily. Three
phases are utilized: water liquid, water vapor and air.

2.1. GEOMETRY
The geometry dimensions are shown in the image 1, it is already shown with half of its volume
filled with liquid water (this is defined at the end).

Image 1: Geometry dimensions in mm

2.2. BOUNDARY CONDITIONS

The boundary conditions are defined as in the image 2. The heat flux is very high because the
walls are too thin, they will be set as 0.5 mm later. The basic process of a PHP is that, when
 the water heats up at the bottom part of the pulsating heat pipe, it increases its temperature,
vaporizing at around 35 °C (because there is a vacuum inside the pipe), and consequently rises
up. When it rises all the way up it touches the condenser and its temperature decreases, making
it condense and now it goes down again.

Image 2: Boundary conditions

3. MOVING TO ANSYS FLUENT

The next sessions show a step by step procedure of how the simulation is done. Most of the steps
come with some physical insight explaining the reason of using specific features.

3.1.GENERAL PROPERTIES
The transient model was chosen since there is a dependence on time and there is no steady state
for a pulsating heat pipe. The gravity was set as 9.81 m/s² in the negative y direction. Nothing
else was changed from the default.
 Image 4: General properties of the model

3.2.SETTING UP THE MODELS

The flow inside a pulsating heat pipe is called slug flow which is characterized for a liquid–gas
flow in which the gas phase exists as large bubbles separated by liquid “slugs” [2]. For this kind
of flow the VOF (volume of fluid) approach is the most adequate because it tracks the interface of
the phases, which is very important to describe this flow [3].
 Image 5: Models utilized for the analysis

Clicking on multiphase twice, the properties are set as shown in the image 6. The three Eulerian
phases are air, water vapor and water liquid, even though there is a vacuum inside the pipe, which
is defined later, air is still defined as one of the phases. The reason for including a phase with air
is that if only water liquid and vapor are defined, the calculation starts as if there is already water
vapor inside the pipe. The implicit Body Force box was checked as well.
 Image 6: Defining the multiphase model

Plus, the energy equation was turned on to allow heat transfer between the phases and the flow is
laminar with the viscous heating box checked. The viscous heating is turned because it includes
the effect of transformation of friction on the walls into heat. To show that the flow is laminar a
conservative assumption is made. It assumes a velocity of 1 m/s, which is high, and then calculate
the Reynolds number using the formula:
𝜌𝑉𝐷
Re =
𝜇

Where:

𝜌 is the liquid water density = 992 kg/m³ (at 39 °C)

V is the liquid water velocity = 1 m/s
D is the diameter of the pipe = 2 mm

𝜇 is the liquid water viscosity = 8.90 × 10−4 Pa.s (at 25 °C)

Using those values, the Reynolds number is calculated as Re = 2230. As Re < 2300 then the flow
is laminar [4].
 Image 7: Defining the viscous model

3.3.IMPORTING THE MATERIALS

The next step is to import the materials needed from the fluent library. To do that, click on materials
and then on Create/Edit.

Image 8: Opening ANSYS materials Database

After clicking on Fluent Database the materials can be copied, and then they will be displayed on
the first window (Image 8). The materials selected here are only the fluids (Water liquid and
water vapor), since aluminum is already default for the solid parts as well as air is for fluids.

Image 9: Selecting materials

3.4. DEFINING THE PHASES AND THEIR INTERACTIONS

Now that the materials were imported, come back to the Models, and then following the steps
shown in the image 10, define which material corresponds to each of the three phases.

Image 10: Defining the three phases

The phases are defined as in the table 1:

Phase Fluid
1- Primary phase Water vapor
2- Secondary phase Water liquid
3- Secondary phase Air
Table 1: Description of the phases

As an example, clicking twice on phase 1, it can be defined as shown in the figure 11, it is a good
habit to name the phases with their actual names. This process is repeated for the other two phases.

Image 11: Defining the primary phase

Now, the interactions between the phases need to be defined. Clicking twice on phase interactions,
there are only two possible interactions: Mass transfer and surface tension. Both play an important
role in a PHP. First, the mass transfer accounts the effects of condensation and evaporation
happening inside the pipe, second the surface tension is important because the tube diameter is
very small and the liquid ends up sticking to the walls (as expected).
To set up the mass transfer, follow the image 12. Then, clicking on edit, the saturation temperature,
which is the temperature that the liquid starts to turn to vapor and the vapor to liquid, is defined as
35 °C (308 K) at a pressure of Pa.

Image 12: Setting up the mass transfer mechanism

To set up the surface tension, follow the image 13. Even though the surface tension changes with
temperature, here it is considered constant (0.07 N/m), for simplification purposes, which is
approximately the surface tension of water at 39 °C. Wall adhesion must be included since the
main feature of a PHP is the fact that the water sticks to the walls.

Image 13: Setting up the surface tension

3.5. CELL ZONE CONDITIONS

On this section the vacuum inside the pipe is introduced. Clicking on “Cell Zone Conditions” and
then on “Operating Conditions” set the “Operating Pressure” as 4000 (this is the saturation
pressure of the water at a saturation temperature of 29 °C, as defined in the image 12) as shown in
the image 14. The other parameters used are the default ones.

Image 14: Including the vacuum effect

 3.6. BOUNDARY CONDITIONS
Following the scheme from the image 2 the boundary conditions are introduced to ANSYS. Named
selections should be created during the meshing to facilitate this step. After clicking on “Boundary
Conditions” the window that popped up should look like the image 15.

Image 15: Boundary conditions

Again, following the image 2, the boundary conditions are defined. As an example, let’s set the
evaporator conditions. Clicking on “evaporator”, its type is “wall”, then clicking on edit, configure
it as in the image 16. The Wall Thickness used here is 0.5 mm.
 Image 16: Boundary condition for the evaporator

3.7. INITIALIZING SIMULATION AND PATCHING LIQUID AND AIR

POSITIONS
After configuring the other two boundary conditions the last step before starting the simulation is
initializing it and attribute initial positions for the liquid and air for the beginning of the simulation
as shown in the image 1. Configure the initialization window as shown in the image 16 and hit
“Initialize”. Set the liquid and air volume fractions as zero because in the next step they are going
to be patched.
 Image 17: Initializing solution

After initializing, under the group “Adapt” hit “Mark/Adapt Cells” and then “Region…” as in
the image 18.

Image 18: Mark/Adapt Cells localization

The main point of marking cells is to create regions that later will be filled with one of the phases.
For this simulation, two regions will be marked, the bottom half for water liquid and the top half
for air. As an example let’s mark the liquid water region. The first thing to do is define the
coordinates that the region will sweep and then hit “Mark”. To make sure everything is right hit
“Manage…” and the regions created will be under the Registers tab, click on one of them and then
“Display”. The geometry should turn to red as in the image 19. Do the same for the air region.
 Image 19: Marking liquid water region

The last thing to do is patch these two regions. This only can be done if you initialized the
simulation before. Hit “Patch…” and configure it as in the image 20. Selecting the phase liquid
and the variable as “Volume Fraction” with its value of one, basically it’s been told to the software
that initially that region filled with only liquid water. The same process is repeated to the top part,
which is filled with air.
 Image 20: Patching regions

When both patching are done, the work can be checked on the tab “Results”, clicking on
“Graphics” and then on “Contours” twice. After the new window pops up, follow the image 21 to
see the phases on the geometry.

Image 21: Checking the phases distribution

 3.8.RUNNING THE SIMULATION
The biggest issue with simulation of multiphase flows is that the time step needs to be sufficiently
small to capture the movement of the particles, and at the same time it needs to be big to reduce
the computational time. For the simulation purposed here, different time steps were used and their
main issues are summarized in the table 2. At the end, a time step of 5x10-4 s was the best choice
since all the others were diverging. Furthermore, the number of steps was chosen as the number
needed to have 6 seconds of simulation.

Time step (s) Issue

0.1 Diverges very fast
0.01 Diverges
0.001 Diverges after some time
0.0005 Takes a long time but doesn’t diverge
Table 2: Issues for different time steps

Image 22: Running the simulation

 4. RESULTS
The computational time was around 20 hours, using a laptop with 4 cores. Maybe this approach is
not the best one, however the results seem adequate. As expected:
a) The water sticks to the walls
b) Liquid columns are formed
c) The fluid oscillates
The following images are a synthesis of the results. In the image 23 there is a legend to better
understand the image 24.

Image 23: Legend for the next images

 Image 24: Liquid distribution at t = 1 s, 2 s, 3 s and 5.7 s respectively

5. REFERENCES

[1] Akachi H., U. S. patent # 4921041, 1990.

[2] Wikipedia, Slug flow.
[3] LearnCAx, Introduction to Multiphase flow modeling in ANSYS FLUENT.
[4] Wikipedia, Reynolds number.
[5] Ashutosh Kr Singh, Numerical Analysis of Performance of Closed Loop Pulsating Heat Pipe

View publication stats