UppsalaUniversitylogotype

MATVET-AM 23010

Degree project 30 credits

Jan 2024

Increasing the diodicity of

ceramic Tesla valves by
exploiting the design freedom
of additive manufacturing
A study in design optimizations of Tesla valves for
ceramic 3D printing

Udit Sharma Master’sProgramme inAdditiveManufacturing

Master’s Programme in Additive Manufacturing

 UppsalaUniversitylogotype

Increasing the diodicity of ceramic Tesla valves by exploiting the

design freedom of additive manufacturing
Udit Sharma

Abstract
The work presented in this thesis was conducted at Uppsala University and at Fraunhofer IKTS,
Dresden. The thesis aims to study design optimizations for increasing the diodicity and thereby
performance of a Tesla valve, a type of “no moving parts” (NMP) valve, through design freedoms
offered by ceramic additive manufacturing.

Tesla valves are capable of creating a pressure differential across them purely by virtue of mechanical
design, and do not employ any moving parts. By geometry manipulation, Tesla valves enable fluid
to flow in a way that hinders its own flow, thereby creating fluidic resistance and increasing pressure
in one fluid direction, while allowing relatively unimpeded flow in the opposite direction.

The manufacture of Tesla valves in the past has been restricted to simple geometries because
conventional manufacturing processes such as CNC machining are unable to produce intricate
geometries, something that Tesla valves require. With the recent innovations in additive
manufacturing, design of these complex geometries has become feasible but still requires further
research. Prior literature has only explored relatively simpler constructs of Tesla valves, not fully
utilizing the design freedoms offered by additive manufacturing.

In this thesis, ceramic additive manufacturing and stereolithography has been used to manufacture
complex Tesla valves. In addition to just complex design, this thesis also presents design
optimizations that can be utilized for simpler Tesla valves for increasing a metric known as diodicity.
Diodicity refers to the ratio of reverse to forward pressure difference, and a high diodicity of a valve
indicates that the valve is able to hinder fluid flow more effectively in one direction than the opposite.

Additive manufacturing boasts an ability to construct complex geometries, due to the layer-by-layer
process of building the final component. Stereolithography (in particular, ceramic stereolithography)
is capable of producing parts that have high resolution and dimensional accuracy, while also
maintaining desirable material properties, such as resistance to high temperatures and mechanical
durability. Since the envisioned Tesla valve is to be used at elevated temperatures, this makes
stereolithography a viable method of producing Tesla valves for aerospace applications.

Design optimizations were carried out and subsequently verified for effectiveness through fluid flow
simulations and practical evaluations. Certain design optimizations were shown to have drastic
effects on the diodicity of the Tesla valve, and these have been subsequently incorporated into the
designs of the Tesla valve in an effort to increase the diodicity of the designed Tesla valves. For
practical evaluation, the optimized Tesla valves were 3D printed through ceramic stereolithography
and stereolithography and extensively tested on a testing rig, with experimental parameters congruent
to the fluid flow parameters applied during fluid flow simulations.

It was found that the results of the fluid flow simulations and experimental testing were somewhat
consistent with each other, and that it is feasible to produce optimized Tesla valves through ceramic
stereolithography. However, it was found through practical evaluation that certain design
optimizations were found to have little to no effect on the diodicity of the final Tesla valve, with
some optimizations even reducing the diodicity.
FacultyofScienceandTechnology,UppsalaUniversity.Uppsala.Supervisor:UweScheithauer,Subjectreader:FrancescoD’Elia,Examiner:UrbanWiklund

Faculty of Science and Technology

Uppsala University, Uppsala

Supervisor: Uwe Scheithauer Subject reader: Francesco D’Elia

Examiner: Urban Wiklund
List of Abbreviations
1. NMP- No Moving Parts
2. AM- Additive Manufacturing
3. SLA- Stereolithography
4. STL- Standard Triangle Language/Standard Tessellation Language
5. 3D- Three Dimensional
6. CAD- Computer-Aided Design
7. IPA- Iso-Propyl Alcohol
8. CerAM- Ceramic Additive Manufacturing
9. VPP- Vat Photo-Polymerization
10. 2D- Two Dimensional
11. RDE- Rotating Detonation Engine
12. CRDW- Continuously Rotating Detonation Waves
13. DLP- Digital Light Projection
14. UV- Ultra-Violet
15. CFD- Computational Fluid Dynamics
16. STEP- Standard for the Exchange of Product Data
17. SST- Shear Stress Transport

3
Table of Contents
1. Introduction .................................................................................................................... 7

1.1 Scope of Study............................................................................................................. 8

2. Background ..................................................................................................................... 9

2.1 Tesla valve introduction and applications......................................................................... 9

2.1.1 Cooling...........................................................................................................10

2.1.2 Fluid mixing.................................................................................................... 11

2.1.3 Flow suppression ............................................................................................. 11

2.2 Design optimizations ...................................................................................................12

2.3 Ceramics as an engineering material ..............................................................................17

2.4 CerAM VPP ...............................................................................................................17

3. Methodology...................................................................................................................20

3.1 Design .......................................................................................................................20

3.2 CFD simulations .........................................................................................................22

3.3 3D printing.................................................................................................................30

3.4 Practical evaluation .....................................................................................................34

4. Results and discussion .....................................................................................................36

4.1 Printed designs............................................................................................................36

4.1.1 CerAM VPP ....................................................................................................36

4.1.2 SLA 3D printing ..............................................................................................36

4.2 CFD simulation results.................................................................................................37

4.3 Practical evaluation results ...........................................................................................38

4.4 Discussion ..................................................................................................................39

5. Conclusions ....................................................................................................................43

6. Future work ...................................................................................................................44

7. Acknowledgements .........................................................................................................45

8. Appendices .....................................................................................................................49

4
 List of Figures
1. Original design of the Tesla valve[7]. ............................................................................................ 9

2. Forward and reverse fluid flow configurations[8]. ........................................................................... 9

3. Tesla valve-type channels for cooling lithium-ion batteries[12]. ...................................................... 10

4. Different configurations of flow channels used for cooling of lithium-ion batteries[12]. ..................... 10

5. Visualization of microfluid mixing abilities of Tesla valves in forward and reverse fluid flow
configurations[18]. ................................................................................................................... 11

6. CRDW propagation in an RDE[21]. ............................................................................................ 11

7. Tesla valve intake in an RDE for suppression of pressure feedback and combustion product backflow[26].
................................................................................................................................................ 12

8. Results of design optimizations to produce high pressure drop in a Tesla valve[27]. ........................... 13

9. Left: Single-stage rectangular cross-section Tesla valve. Right: Multi-stage circular cross-section Tesla
valve[27]. ................................................................................................................................. 13

10. Left: The Tesla loop (flow from top) merging with the main channel at a near perpendicular angle. Right:
The Tesla loop merging with the main channel at a higher feedback angle[28]. ................................. 14

11. Pressure difference in the forward and reverse flow configurations for the optimal and reference Tesla
valves[28]. ............................................................................................................................... 14

12. Tesla valves with increasing number of Tesla loops, creating Tesla valves with sequentially higher
stages[29]. ............................................................................................................................... 15

13. Difference in diodicities with different stages of Tesla valves and Reynolds numbers[29]. ................... 15

14. Control regions identified for optimization of a Tesla valve to achieve high diodicities[9]. .................. 16

15. Control regions optimized to achieve high diodicities in the Tesla valve[9]. ...................................... 16

16. CAD model of a spiral 3D Tesla valve[30]. ................................................................................... 16

17. Effect of bore size on final dimensions on parts produced through CerAM VPP. ................................ 18

18. Front view of the six prototypes selected for CFD and practical testing. a) Prototype I. b) Prototype II. c)
Prototype III. d) Prototype IV. e) Prototype V. f) Prototype VI ........................................................... 20

19. Left: Top view of Tesla valve main channel and Tesla loop. Right: Sectional view of Tesla valve indicating
pipe diameter (d1 ) and pipe thickness (t). ...................................................................................... 21

20. Fluid region of Tesla valve prototype I, chosen for grid independence studies .................................... 23

21. Mesh generated for grid independence studies. a) Tetrahedral mesh, isometric view. b) Tetrahedral mesh,
face view. c) Polyhedral mesh, isometric view. d) Polyhedral mesh, face view. e) Polyhedral mesh, close-
up view. .................................................................................................................................... 24

5
22. Inflation layers and their impact on accuracy of results. Left: Lack of inflation layers. Right: Presence of
inflation layers. ......................................................................................................................... 24

23. “Law of the wall” graph generated by plotting experimental velocity values against distance of mesh
elements from boundary wall. ...................................................................................................... 25

24. Values of y plus generated based on the input inflation layer settings. ............................................... 26

25. Report definition plots used to determine convergence of a CFD solution. Left: Report definition plot of
inlet pressure. Right: Report definition plot of outlet pressure. ......................................................... 29

26. Views of the theorized optimal Tesla valve. a) and b) Isometric views. c) Top view. d) Right side view. e)
Front view. ................................................................................................................................ 30

27. Lithoz CeraFab 8500 3D printer. ................................................................................................. 31

28. Slicer view for 3D printing parts. a), b) and c) Slicer view for CerAM VPP 3D printing. d) Slicer view for
SLA 3D printing. ........................................................................................................................ 32

29. Temperature profiles for pre-conditioning, debinding and sintering. ................................................. 33

30. Practical setup for practical evaluation of Tesla valve prototypes. Left: The entire practical setup. Right:
Zoomed-in view of the Tesla valve prototype in action. .................................................................... 35

31. Different views of a Tesla valve prototype successfully printed using CerAM VPP. .............................. 36

32. The six Tesla valve prototypes 3D printed using SLA. Left: Tesla valve prototypes I, IV, V and VI. Right:
Tesla valve prototypes II and III. .................................................................................................. 37

33. Variation in diodicity values as measured by CFD simulations for different Tesla valve prototypes. ...... 38

34. Diodicity values as evaluated practically for different Tesla valve prototypes. .................................... 38

35. Comparison between diodicity results obtained from CFD simulations and practical evaluations. ........ 39

List of Tables
1. Grid independence study results for different mesh setups. .............................................................. 27

2. Calculations for inlet velocity of the Tesla valve, for CFD simulations. ............................................. 28

3. Process parameters for CerAM VPP and SLA 3D printing. ............................................................. 31

4. Thermal cycles for pre-conditioning, debinding and sintering for CerAM VPP 3D printing.................. 34

5. CFD simulation results for all Tesla valve prototypes. .................................................................... 37

6. Practical evaluation results for all Tesla valve prototypes. .............................................................. 38

7. Comparison between diodicity results obtained from CFD simulations and practical evaluations. ....... 39

6
1. Introduction
A common method of controlling the direction of flow of fluids in pipe constructs is to employ valves, which
are capable of opening and closing through mechanical actuation, to allow or prevent fluid from flowing in a
preferred direction, to vary the amount of fluid flowing in a channel, to alter fluid pressure and velocity or to
enable mixing of two dissimilar fluids[1]. While these valves are effective at controlling fluid flow, the usage
of mechanical motion for actuation has certain disadvantages. For instance, at elevated fluid temperatures,
joints involved in the mechanical motion for actuation may degrade due to lowered fatigue resistance faster
as compared to joints that are not exposed to high temperatures during actuation[2]. Further, frequent actuation
of these joints can also result in frictional wear and tear of the joints, leading to eventual failure of the valve
entirely. Additionally, constant use of mechanical valves can also result in cavitation effects, further shortening
the life of valves that operate on mechanical motion[3].

To address these issues, a type of valve that does not involve mechanical motion for actuation may be
preferable to valves that require mechanical motion for actuation. The absence of moving parts results in
mechanical reliability under extreme conditions (such as elevated operating temperatures and pressures) and
a valve that lasts longer as compared to valves that require mechanical motion for actuation[4]. One such valve
is the Tesla valve, a type of NMP valve developed by Nikola Tesla in the year 1916.

A Tesla valve operates by manipulating the geometry of fluid paths such that a pressure differential is created
across the valve, thereby controlling the direction of fluid flow. This valve does not comprise any moving
parts and displays high reliability at extreme conditions as a consequence of having no moving parts[4]. This
thesis builds on previous designs of this valve, improving performance through design optimizations. Fluid
flow simulations and practical evaluations confirm the effectiveness of the design optimizations, through the
iterative improvement of a metric associated with performance of the valve, called the diodicity of the valve.

Diodicity of a valve is described as the ratio of reverse pressure difference to the forward pressure difference.
The reverse pressure difference is the difference in static pressures between the outlet and inlet of the valve
when it is desired that the fluid flows through the pipe in the impeded and restricted configuration. Similarly,
forward pressure difference is the difference in static pressures between the outlet and inlet of the valve when
it is desired that the fluid flows through the pipe in the unimpeded and unrestricted configuration. It is therefore
desired that the diodicity of the valve be greater than 1, preferably significantly greater than 1 such that the
reverse pressure difference is maximized. While it is possible to manipulate the geometry of the pipe in a way
to obtain a particular diodicity value for a given fluid flow scenario, this thesis aims to maximize the diodicity
for the chosen fluid flow scenario.

AM offers unique design freedoms that were not available previously from conventional manufacturing
processes, such as lathe and milling operations[5]. This has allowed for the manufacture of complex
geometries, such as Tesla valves with intricate designs and resolutions. Due to the freedom of design, it is now
possible to manufacture Tesla valves with increased diodicity, owing to the fact that design optimizations are
no longer restricted by conventional manufacturing processes. Recesses, swept holes and other features
impossible to manufacture conventionally can now be manufactured without issue. SLA is one such AM
process and is chosen as the AM process for the manufacture of Tesla valves designed in this thesis.

In SLA, a photosensitive resin is polymerized by sequential and programmed light projection. The light
polymerizes the resin based on a .STL file uploaded to the 3D printer, where the .STL file itself is based on a
CAD model of the part to be manufactured. It is possible to add ceramic particles to the resin, such that the
monomeric resin cures around the ceramic particles, essentially locking the ceramic particles in place. After
the part has been printed, it can then be post-processed (fully curing the resin, support structure removal and
washing in an IPA bath), after which the part is debinded and sintered to remove the polymerized resin, as
well as to consolidate the ceramic particles together to form a homogeneous and mechanically stable part.
This results in a part that is completely ceramic, in which the ceramic particles have fused together to form
the final part[6]. This process of using SLA with ceramic particles is called CerAM VPP and will be referred
to as such throughout the rest of the thesis document. Throughout this document, the term 3D printing will be
used to describe the process of additive manufacturing, although it must be noted that in certain cases the two
are distinct from one another. In such cases, the distinction will be made between the two terms, but otherwise
the term 3D printing will be used to refer to a generic additive manufacturing process.

7
1.1. Scope of Study
Although design optimization of Tesla valves has been performed in previous literature, this has been
done for 2D versions of valve. A comprehensive design optimization for 3D versions of the valve remains
yet to be carried out. Therefore, the scope of this thesis is to perform design optimizations for a 3D version
of the Tesla valve and inspect the effects of the design optimizations on diodicity of the valve. Through
the design optimizations and proof of manufacturability through CerAM VPP, the optimized Tesla valve
can potentially find wider industrial applicability, in domains such as aerospace and automobile industries.

8
2. Background
2.1. Tesla valve introduction and applications
A Tesla valve is a valve that comprises no moving parts i.e., it is simply composed of a path through which
fluid can flow[7]. Figure 1 below from US patent document US1329559A illustrates the first Tesla valve
ever designed, by Nikola Tesla himself in the year 1916.

Figure 1: Original design of the Tesla valve[7].

In this version of the Tesla valve, the fluid domain comprises two openings, which are the inlets and
outlets of the Tesla valve, a design commonly seen in Tesla valves designed since. A Tesla valve can be
thought of as a main channel plus additional “Tesla loops” attached to the main channel. In figure 1, the
main channel is illustrated by the long-dashed line (labelled 7) originating from the rightmost opening
(labelled 5) and moving right to left across the interior of the valve. Tesla loops can then be seen to be the
bulbous channels (labelled 2) that are designed to prevent fluid flow within the fluid path when fluid is
flowing from left to right. This is also defined as reverse fluid flow in the context of a Tesla valve, and
the opening labelled 4 is the inlet while the opening labelled 5 is the outlet for this configuration.
Contrastingly, fluid flow is said to be forward when the fluid is flowing from right to left i.e., from the
opening labelled 5 to the opening labelled 4.

It can then be understood that in the reverse fluid flow configuration, the fluid enters the fluid domain
from opening 4, and splits into two streams: one stream moving up into the Tesla loop, and the other
continuing through the main channel[8]. Due to the curve of the Tesla loop, the fluid in this region follows
the circular path of the Tesla loop and eventually collides with the fluid moving in the main channel,
thereby creating turbulence and resistance to flow for the fluid. One such area of turbulence is indirectly
labelled 7. Similarly, in the forward fluid flow configuration, the fluid enters the fluid domain from
opening 5 and continues along what we have termed the main channel. There may be slight splitting of
the fluid into the main channel and Tesla loop, but the fluid in the forward fluid flow configuration remains
largely present in the main channel and flows through the valve unimpeded. Figure 2 helps better illustrate
reverse and forward fluid flow configurations.

Thus, we can see that Tesla valves are characterized by no moving parts, and a significantly higher
resistance to flow in one configuration (reverse fluid flow) than the other (forward fluid flow). This
resistance to flow in one configuration is what defines this mechanical construct as a valve. However, it
must be noted that since a Tesla
valve has essentially no areas where
the fluid is stopped completely, the
valve cannot technically offer
complete fluid blockage in a given
direction. Under steady state
Figure unavailable due to copyright issues. Find conditions, Tesla valves (and indeed
referred figure at [8] as figure 1. all NMP type valves) provide mildly
effective resistance in the reverse
fluid flow configuration, whereas
providing net flow unidirectionally if
the flow is oscillatory[9]. Therefore,
NMP type valves are best used for
Figure 2: Forward and reverse fluid flow flows which are oscillatory in nature,
configurations[8]. (such as in a pulsed detonation

9
application, in a jet engine) as is discussed in later sections.

As a mechanical valve, Tesla valves have been used for more than just creating a pressure differential for
a fluid flow. A few applications for Tesla valves are as discussed below:

2.1.1. Cooling

Cooling is a phenomenon that

is understood as a reduction
in temperature of any
material. Cooling finds
certain applications, such as
in battery systems where it is
required to keep the
temperature of the battery
assembly at a certain range
(between 20 and 40˚C[10]) to
ensure the safety, stability,
performance, and endurance
of the battery system[11]. It is Figure 3: Tesla valve-type channels for cooling lithium-ion
therefore critical that batteries[12].
operating temperatures of a
battery system be maintained at certain temperature ranges, for which a multitude of methods
for thermal management have been proposed, one of which is utilizing Tesla valve-type
channels for cooling lithium-ion batteries[12].

Figure 3 illustrates one example of how Tesla valve channels could be used to carry relatively
cooler medium through neighboring sources of heat to reduce temperature locally cool the
system[12]. The fluid was passed through three different configurations, a zig-zag channel in
the shape of the letter “Z”, Tesla valve in the forward flow direction and in the reverse flow
direction, both of which can be seen in figures 2 and 4.

It was observed that

compared to the Z-type flow
channel, the forward Tesla
valve-type flow channel
exhibited an increase in
cooling efficiency of 4.5%,
whereas the reverse Tesla
valve-type flow channel
exhibited an increase in
cooling efficiency of 17.6%. Figure 4: Different configurations of flow channels used for
It can be seen in figure 4 that cooling of lithium-ion batteries[12] (figure reused with permission
flow in the reverse Tesla from authors).
valve-type channel is
significantly more turbulent than flow in the other channels, which grants insight on why this
type of flow channel displays higher cooling efficiencies. The relation between turbulence and
cooling efficiency is well-understood, with research showing that turbulent flow constantly
mixes flow near boundary walls with flow away from boundary walls, essentially mixing flow
within the channel and enhancing heat transfer coefficient of the fluid, leading to higher cooling
efficiency[13]. Further, it is known that for a uniform and laminar flow, the cooling performance
and efficiency gradually reduces along the length of the channel[12], due to the thermal
homogenization of the fluid. In this situation, boundary layer fluid is still the same temperature
of the fluid that is supposed to be cooled down, and the inner mainstream liquid heats up as
well, reducing the temperature gradient in the fluid channel and driving down the cooling
efficiency.

10
 It can then be understood why Tesla valves in the reverse fluid flow configuration would be
efficient in picking up and carrying away heat from boundary walls of the flow channel.

2.1.2. Fluid mixing

Fluid mixing involves introducing two or more homogenous fluid stream inputs, and then
combining them in a finite space to produce a homogenous fluid stream output which is a
mixture of the input streams. This can be achieved through multiple methods[14], including
using magnetic fields[14], [15], acoustic fields[14], [16] and electric fields[14][17].
Microfluidic mixing is mixing of fluids as described above, but at the microscopic scale, and is
used in fields ranging from chemistry, drug development, biology and material science[14].

Figure 5: Visualization of microfluid mixing abilities of Tesla valves in forward and

reverse fluid flow configurations[18] (figure reused with permission from authors).
Microfluidic mixing takes advantage of the above-mentioned ability of Tesla valves to create
turbulence from a laminar flow when the Tesla valve is used in the reverse fluid flow
configuration[18]. As seen in figure 5, when the fluid flows in the forward flow configuration
(labelled “a”), the two fluid streams (blue and green) remain largely unmixed and heterogenous,
but when the fluid flows in the reverse fluid flow configuration (labelled “b”), the two separate
fluid streams begin mixing from the third Tesla loop itself, and the flow becomes completely
homogenous by the time the flow reaches the outlet of the Tesla valve. It can be observed that
the homogeneity of the fluid streams is achieved before the fluid reaches the outlet of the Tesla
valve, implying that the Tesla valve could be shortened and still retain its ability to mix the two
fluid streams. It is also interesting to observe that the Tesla valve design in figure 5 is nearly
identical to the very first Tesla valve design seen in figure 1, with no design optimizations
whatsoever, some of which can be seen in figure 4. It can then be reasoned that with design
optimizations, the Tesla valve in figure 5 could theoretically perform heterogenization of the
fluid streams much before the third Tesla loop, increasing the mixing efficiency of the Tesla
valve.

2.1.3. Flow suppression

Another application for Tesla valves is for

pressure feedback reduction and
suppression in RDEs[19]. An RDE operates
by continuous detonation of a fuel (often a
mixture of a primary fuel with an oxidizer),
in a precise and controlled manner[20]. In
an RDE, the combustion propagates by way
of CRDWs in the combustion chamber. As
can be seen from figure 6, fuel is injected
through micro-channels situated around the
top flat ring of the cylindrical combustion
chamber. The fuel is then detonated at a
Figure 6: CRDW propagation in an RDE[21] (figure micro-channel, forming a detonation wave.
reused with permission from authors).
Next, the process is repeated at a

11
 consecutive micro-channel, leading to a detonation wave that travels outwards again,
connecting with the detonation wave from the previous detonation. This process continues in a
rotating manner, by fuel injection through each of the micro-channels, forming a CRDW,
highlighted in red in figure 6[21].

Detonation has a higher thermal efficiency than conventional rocket engines that operate on the
principle of deflagration[21], [22][23], which is when the combustion wave from the
combustion of the fuel travels at subsonic speeds. In detonation, by contrast, the combustion
wave from the flue combustion travels at supersonic speeds. In detonation, therefore, the
combustion takes place much more rapidly, leading to relatively more complete combustion,
lower thermal losses to the ambient environment, higher specific impulse and higher pressure
gradients within the combustion chamber, all of which contribute to higher thermal efficiencies
of a detonation-type engine in comparison to a deflagration-type engine.

However, detonation also produces significant pressure feedback due to the violent nature of
fuel combustion, which threatens to compromise compressor efficiency[24], induce instabilities
related to a “pop-out mechanism”, wherein undesirable deflagration occurs in the RDE, and
self-ignition of the air-fuel mixture ahead of the planned detonation, thereby reducing operating
efficiency of the RDE[25]. Another common issue with RDEs is combustion product
backflow[26], wherein the products of combustion or combustion front itself travels in a
direction that is reverse of the desired direction of fuel injection, further reducing the thermal
efficiency and creating combustion instabilities in the RDE.

For the suppression of

pressure feedback and
combustion product
backflow, a Tesla valve
intake structure was
introduced into a channel
through which enriched
air is pushed into to the
combustion chamber, as
seen in figure 7. The blue
Figure 7: Tesla valve intake in an RDE for suppression of pressure
feedback and combustion product backflow[26] (figure reused with line arrow represents
permission from authors). pressure backflow and
combustion product
backflow, which travels through a Tesla valve in the reverse fluid flow configuration, which
suppresses the above-mentioned flows. The green arrow represents the direction that enriched
air flows into the combustion chamber, which travels through a Tesla valve in the forward flow
configuration, thereby continuing unimpeded into the combustion chamber. This flow is
compared to a simple convergent-divergent intake structure, also seen in figure 7.

It is observed that the Tesla valve intake structure significantly reduces pressure feedback
(minimum pressure feedback observed is approximately 6.5%) as compared to a regular
convergent-divergent intake structure (minimum feedback pressure observed is approximately
30%), thereby solidifying the role of Tesla valve structures in backflow prevention and
suppression of fluids.

2.2. Design Optimizations

This thesis project explores ways to increase the diodicity of ceramic Tesla valves by exploiting the
design freedom of additive manufacturing, by employing design optimizations for a 3D version of
the Tesla valve. The diodicity refers to the ratio of reverse pressure difference to forward pressure
difference, and therefore design optimizations that can increase the reverse pressure difference
across a Tesla valve without significantly increasing the forward pressure difference are chosen and
included in the final design of the Tesla valve.

12
In their paper titled “Tesla valve for Hydrogen Decompression: Fluid Dynamic Analysis”[27],
Bäckman et al. investigate the role of certain chosen design parameters for the increase of pressure
drop across a Tesla valve. A high pressure drop valve is required in applications such as storage of
hydrogen gas for hydrogen fuel cells, where a high pressure drop at the outlet of the Tesla valve is
needed to decompress hydrogen gas, which is then supplied to the fuel cell for use. Three geometric
parameters are chosen and investigated, namely the cross-section of the pipes used in the Tesla valve,
the valve-to-valve distance for a multi-stage Tesla valve, and inner radius of the Tesla loop in a Tesla
valve.

As seen in figure 8, it was observed that a circular cross-section of the pipes in the Tesla valve, with
a large valve-to-valve distance and a small inner curve radius all contributed to a higher pressure
drop across the Tesla valve. We observe that the highest pressure drop of 1862 Pascals is associated
with a multi-stage model labelled “CMS70”, the “C” referencing circular cross-section, “MS”
referencing that the Tesla valve is multi-staged i.e., it has more than one Tesla loop, and the “70”
referencing that the valve-to-valve distance is 70 mm. Figure 9 shows the difference between circular
and rectangular cross-sectional Tesla valves, while also showing the difference between a single-
stage and multi-stage Tesla valve. It is to be noted that a multi-stage Tesla valve is defined as having
three Tesla loops, or as being a Tesla valve of three stages.

Figure 8: Results of design optimizations to produce high pressure drop in a

Tesla valve[27] (table reused with permission from authors).

Figure 9: Left: Single-stage rectangular cross-section Tesla valve. Right: Multi-stage

circular cross-section Tesla valve[27] (figure reused with permission from authors).

Various design parameters may be chosen for optimization of the design of a Tesla valve, one such
parameter being the “feedback angle” of the Tesla valve[28], which is the angle at which the outlet

13
of the Tesla loop of the Tesla valves curves back into the main channel of the Tesla valve. Optimizing
this particular parameter allows for a more head-on interaction of the flow developed in the Tesla
loop with flow developed in the main channel, thereby increasing the pressure developed at the inlet
of a Tesla valve used in the reverse fluid flow configuration.

This is clearly seen in figure 10, which illustrates velocity flow lines within Tesla valves with two
different feedback angles, with one Tesla valve having a feedback angle steeper than the other. The
fluid flow in both images is in the reverse fluid flow configuration, with the fluid flowing from the
right side of the Tesla valve to the left.

Figure unavailable due to copyright issues. Find referred figure at [28] as figure 3 and
figure 4.

Figure 10: Left: The Tesla loop (flow from top) merging with the main channel at a near perpendicular
angle. Right: The Tesla loop merging with the main channel at a higher feedback angle[28].

Further, from figure 11, the effect of increasing the feedback angle in an optimally designed Tesla
valve is reflected in the pressure difference, compared to the pressure difference in a reference Tesla
valve. It is observed that the reverse flow pressure difference in the optimized Tesla valve
(represented by square icons, labelled as opt Δpr) is higher as compared to the reverse flow pressure
difference in the reference Tesla valve (represented by circular icons, labelled as ref Δpr). The
forward flow pressure difference for both the reference and optimally designed Tesla valves appears
to be identical, indicating that the design optimization of increasing the feedback angle of the Tesla
loop does not affect the forward flow pressure difference significantly.

It has also been observed that an

increase in stages -the number of Tesla
loops present in one combined Tesla
valve- also correlates to an increase in
the diodicity of the Tesla valve[29].
This can be attributed to a combined
increase in pressure difference in the Figure unavailable due to copyright issues.
reverse fluid flow configuration of Find referred figure at [28] as figure 6.
each Tesla loop. The fluid flowing
within the Tesla valve in the reverse
fluid flow configuration increases the
pressure at the inlet of the Tesla loop,
with each consecutive Tesla loop
sequentially increasing the pressure Figure 11: Pressure difference in the forward and reverse flow
difference at the inlet of the Tesla configurations for the optimal and reference Tesla valves[28].
loop. The effectiveness of this design
optimization has also been briefly explored in the work of Bäckman et al., although the scope of
their work did not involve understanding the effect of different number of Tesla valve stages, but
rather just the difference between a single-stage Tesla valve and a Tesla valve with three stages[27].
Figure 12 illustrates how Tesla loops can be sequentially combined to create Tesla valves of
increasing stages[29].

14
In figure 12, the first Tesla valve
(labelled “a”) comprises a single
stage, whereas the fourth Tesla
valve (labelled “d”) comprises four
Tesla loops, and thereby comprises
four stages. Also seen are the two
and three stage Tesla valves, all of
which are used for simulating fluid
flow to verify the effect of Tesla
valve stages on the diodicity of
Tesla valves. Taking the example
of the single stage Tesla valve
(labelled “a”), forward fluid flow
configuration is when fluid is
flowing from the bottom left
hemispherical plenum, through the
pipe channels and into the
Figure 12: Tesla valves with increasing number of Tesla loops,
hemispherical plenum located at creating Tesla valves with sequentially higher stages[29] (figure
the top left. In a multi-stage Tesla reused with permission from authors).
valve, the addition of more stages
essentially results in the outlet flow of the first Tesla loop becoming the inlet flow for the next Tesla
loop, with sequential Tesla loops compounding the restrictive nature of the Tesla valve to increase
performance of the Tesla valve in the reverse fluid flow configuration. In the reverse fluid flow
configuration, an assumption can be made that the pressure at the outlet of the first Tesla loop is
significantly lower than the pressure at the inlet of the first Tesla loop. The pressure at the outlet of
the first Tesla loop is then the pressure at the inlet of the second Tesla loop, and due to the pressure-
drop inducing characteristics of the Tesla valve, the pressure sequentially drops as the fluid flows
from the first stage of the Tesla valve to the last, thereby resulting in an overall pressure drop that is
significantly higher in a multi-stage Tesla valve, than what would be observed for a single-stage
Tesla valve with comparable fluid flow volume.

CFD simulations were carried out for the four different Tesla valves at varying Reynolds numbers
(Re), to accurately capture the effect of stages of a Tesla valve for different fluid flow scenarios and
regimes. The results of this testing are shown in figure 13, and it can be observed that the highest
diodicity (2.643) is seen in a multi-stage Tesla valve with four stages, carrying fluid with a Reynolds
number of two hundred.

Figure 13: Difference in diodicities with different stages of Tesla valves and
Reynolds numbers[29] (table reused with permission from authors).

Another method of design optimization of a Tesla valve for high diodicity has been provided by
Bardell et al[9]. The Tesla loop portion of a Tesla valve is identified to be the region where most of
the “diodic” action occurs in a Tesla valve, and therefore the design optimization of this region
(labelled 3, 8 and 9) leads to improved diodicity of the Tesla valve, the basic outline of which can
be seen in figure 14. The other control regions are found to have negligible contributions to the
diodic action of the Tesla valve, with their contributions to the diodicity of the Tesla valve
approaching unity. In fact, it is observed that control regions labelled 1, 7, 2 and 6 have an overall
diodicity of lower than 1.1.

Based on the effect of the identified control regions of the Tesla valve on diodicity, a new design
was generated to optimize the diodicity of the Tesla valve. From figure 15, it can be seen that control

15
regions 3, 8 and 9 have been
significantly altered to increase fluid
velocity through these control
regions, which in turn increases the
momentum with which the fluid
flowing through the Tesla loop (and
thereby through the control regions)
impinges on fluid flowing in the main
channel (labelled as control region 4),
resulting in a higher pressure
generated at the inlet of the Tesla
valve in the reverse fluid flow
configuration. Design optimizations Figure 14: Control regions identified for optimization of a
of the Tesla valve were validated with Tesla valve to achieve high diodicities[9] (figure reused with
both numerical and simulation permission from authors).
methods to reveal that the diodicity of
the Tesla valve increased from 1.27 for the conventional Tesla valve, to 1.74 for the design optimized
Tesla valve.

Additionally, studies to investigate

design optimizations for 3D Tesla
valves have also been performed. Sano
has demonstrated that additive
manufacturing can indeed be used to
fabricate complex 3D Tesla valves
which incorporate design optimizations
of previous literature[30].

Various design parameters of Tesla

valves were chosen as variables to be
optimized through design of
Figure 15: Control regions optimized to achieve high experiments. The overall design
diodicities in the Tesla valve[9] (figure reused with permission presents a spiral pipe structure, with
from authors). Tesla loops emerging on the outer
periphery of the spiral pipe structure as
seen in figure 16, thereby resulting in a structure that can only be best described as a 3D Tesla valve.
Parameters to be optimized included the following, in order of their effect on diodicity of the 3D
Tesla valve:

a. Spiral radius, which is defined as the radius of

the main channel of the Tesla valve.

b. Feedback angle, which is defined as the same

angle as previously defined, the angle at which the
outlet of the Tesla loop of the Tesla valves curves
back into the main channel of the Tesla valve.

c. Angle of approach, which is defined as the angle

at which the inlet of the Tesla loop connects to the
main channel of the Tesla valve.
Figure 16: CAD model of a spiral 3D Tesla
valve[30] (figure reused with permission from d. Spiral height, which is defined as the pitch of the
authors).
spiral profile of the Tesla valve.

e. Aspect ratio of the Tesla valve, which is defined as a geometric ratio between the radius of the
Tesla loop and diameter of the main channel of the Tesla valve.

16
 The interplay of these parameters has also been explored, but the main focus is on the individual
effect of each of these parameters, as all combinations of the interplay between these parameters has
been observed to have a lower effect on the diodicity of the Tesla valve when compared to the
parameters by themselves. The 3D Tesla valve was manufactured using stereolithography, and
printed in a polymeric resin at first, and then later with a ceramic resin to produce a final part made
from ceramic. The reasons for printing in ceramic, as well as details of the exact procedure of
printing have been discussed in later sections. With the Tesla valve design seen in figure 16, a
diodicity of 1.74 was achieved.

2.3. Ceramics as an engineering material

Ceramics are an umbrella term for materials that include clay, porcelain, earthenware, nitrides, and
carbides, to name a few. Historically speaking, ceramics have been used by human civilizations for
close to 26, 000 years, for applications that include artistry and pottery. They can be considered to
be one of the most important materials to have been discovered in the early stages of human
civilization development and have paved the way for advancing human civilization by allowing for
creation of structures such as huts and houses, while also significantly increasing the quality of life
for human beings by providing a material through the use of which organic matter can be stored and
transported. In the present day, ceramics find use in applications that range from use in refractory
material, wear-resistant coatings, utensils, engine parts, ball bearings, biological implants, with
newer applications for ceramics being discovered regularly.

As an engineering material, ceramics can often be found in applications that require chemical
inertness[31]. Constituent atoms and molecules of ceramics are bonded ionically, meaning that the
outer valence shell of ceramic atoms and molecules are filled entirely, granting the ceramic particles
chemical inertness. Examples of such ceramics are oxides of metals such as aluminium, zirconium,
titanium and silicon[32]. It is well understood that metals such as the aforementioned are commonly
found in nature in their oxide forms and are inert in nature owing to the natural tendency of reactive
elements such as metals to form compounds that have low energy states in nature. Ceramics that
form oxides for example, are therefore understood to be chemically inert, and find use in applications
such as apparatuses for chemical reactions and processes.

Wear resistance is also a commonly observed attribute of ceramics[33]. For instance, alumina oxide
ceramics are found to have excellent wear resistance, owing to the unique chemical structure of the
ceramic. This type of ceramic comprises a crystalline microstructure that is highly uniform, with a
miniscule amount of glassy phase being present. Due to the uniformity in the crystal structure of the
ceramic, deformation through cracks is highly unlikely under working conditions, and this lends
itself to the wear resistance of the ceramic itself. Such ceramics are often used in situations where
high friction between two surfaces is expected and is undesirable, such as coatings for subtractive
manufacturing tools (drill bits, cutting tools, lathe tools) and ball bearings. Similar coatings are also
used to mitigate frictional losses for applications such as turbine blades, pumps, pipes, and chutes.

With relevance to this thesis report, ceramics are also frequently used for thermal insulation[34],
owing to their unique property of ionic bonding as discussed before. Heat transfer through
conduction in the solid state can be attributed to two main mechanisms. Heat is transferred in the
solid state firstly through vibrations of constituent atoms or molecules of the solid-state material, or
through movement of free electrons of the constituent atoms or molecules[35]. Ceramics negate the
heat transfer through conduction through the movement of free electrons, as ceramic materials do
not comprise any free electrons due to ionic bonds present in the ceramic material. Ceramics also
are relatively inert, and therefore do not decompose or deteriorate with an increase in temperature.
For these reasons, ceramics are often used for withstanding high temperatures in refractories, space
shuttle exterior plates, cookware and high temperature combustion engines.

2.4. CerAM VPP

CerAM VPP is an abbreviation for ceramic vat photopolymerization, which is an additive
manufacturing method used to create parts and components in ceramic material. Additive
manufacturing enables manufacturing of parts through layer-wise deposition of material over pre-

17
existing layers. In particular, CerAM VPP utilizes the property of photopolymer resins to polymerize
in the presence of light activation. Since ceramics cannot directly be 3D printed (unlike polymers
and metals), photosensitive resin monomer particles are polymerized around suspended ceramic
particles, thereby locking them in place. To polymerize the photopolymer-ceramic suspension (called
ceramic slurry), light activation in the form of ultraviolet light, lasers or photonic light can be used.
The printed part, referred to as a green part, is a construct of photopolymer polymerized around
ceramic particles, from which the polymer must be removed, and the ceramic particles condensed to
form a final brown part. Exact details of CerAM VPP are presented in the following text.

Although many variations exist amongst additive manufacturing processes, the overall structure of
additive manufacturing a part is as given below, and in the context of CerAM VPP [36][37]:

1. Generating a CAD model of the part:

This step involves generation of a design of the part to be manufactured. This step is common
across all additive manufacturing processes and is indeed required for CerAM VPP as well. In
the case of CerAM VPP, it must also be noted that bores of certain dimensions may not be printed
as designed, due to issues regarding resin clogging bores that are too small. CerAM VPP also
inherently has trouble maintaining dimensional accuracy of the bores, as is seen in figure 17.
The smaller the bore, the more likely it is to be printed with higher dimensional inaccuracy
(assuming the bores are printed
laterally in the z-direction). Further,
CerAM VPP involves post-processing
which causes shrinkage in the final
part, details of which are mentioned in
the post-processing section of this
report. This shrinkage is crucial when
designing the part, as the final
dimensions of the printed part will be
significantly different from the CAD
Figure 17: Effect of bore size on final dimensions on
design of the same part if no size parts produced through CerAM VPP (figure reused with
compensation for the part is permission from authors).
performed.

2. Conversion of the CAD model to .STL format:

This step is also common across all additive manufacturing processes. Conversion to . STL
format is usually a step handled by the CAD software, and generally converts a CAD file from
the native file format of the CAD software to a format that is recognized by additive
manufacturing machines (or 3D printers). Conversion to .STL format is relevant for CerAM VPP
as well.

3. Pre-processing of the print:

Once the .STL file has been prepared, a few pre-processing steps need to be performed before
the part can be manufactured. Firstly, the .STL file needs to be transferred to the 3D printer. This
step quite simply involves uploading the .STL file into the 3D printer and slicing the model.
Slicing is a step which converts the 3D model into a collection of layers, which form the final
part when printed in succession. Each layer (or slice) in this way builds the part, and each layer
is printed in a line-wise manner, through a set of G-codes. G-codes are essentially a set of
instructions that inform the 3D printer as to where the print nozzle needs to move, and
consequently print. In this way, G-codes can be understood as a series of directions which the
print nozzle takes in order to print out each layer of the part to be printed.
Additionally, care must be taken to ensure that the print is of appropriate dimensions such that
it fits within the build volume of the 3D printer, ensuring the 3D printer has enough raw material,
ensuring that the material that the part is printed in is the correct material, and other checks on
the printer’s operability. The design of the part to be printed must be re-checked to ensure that it
prints as planned.

18
 Once it has been ensured that the 3D printer will print as planned, pre-processing on the end of
the part to be printed itself is performed. Support structure generation, layer height thickness,
overhang angles, infill densities and build plate adhesion strategies are amongst the most
commonly adjusted parameters when printing a part. In CerAM VPP, this includes ensuring that
the part to be printed is attached to the build plate by way of rafts or platforms, which are each
directly printed on the build plate.

4. Build:
Once the above steps are completed, the print is started. There is usually a minimal level of
supervision or intervention required at this step. During this step, the model is printed as per the
sliced layers. However, in CerAM VPP the layers are not printed in a line-wise manner as
CerAM VPP operates on the principle of DLP printing, which means the layers of the part are
printed an entire layer at a time. The light source is illuminated onto a reflective surface, which
then projects the sliced layer onto the ceramic slurry, thereby photopolymerizing it and
producing the green part.

5. Post-processing of the print:

Once the part has been printed, post-processing steps such as removal of the part from the build
plate, removal of support structures, manipulating surface finish and applying surface coatings
can be performed. Post-processing steps can also include coating the part to achieve desirable
mechanical and aesthetic properties. This step (apart from assembly and installation of parts) is
the last step in the 3D printing process for any given printed part.
For parts printed using CerAM VPP and other SLA processes, additional post-processing is
required. The part fresh from the 3D printer still contains unpolymerized monomer, so the
printed part is treated to fully polymerize the part. This is done by exposing the printed part to
UV radiation for a certain period, and cleaning unpolymerized resin off the surface of the printed
part using IPA. This amount of post-processing would be enough for a part printed using SLA,
but CerAM VPP involves further post-processing to produce the final part. In addition to the
above post-processing steps, the part (green part) must have the polymerized resin removed,
which is done through a debinding step in which the part is exposed to high temperatures (at
temperatures of up to 600˚C)[37] so as to evaporate the polymerized resin. This results in a part
which is a loose collection of ceramic particles with little to no polymer left. After this, the part
must be sintered at high temperatures and elevated pressures to consolidate the loosely held
ceramic particles. After debinding and sintering, the final part is ready for use, although it must
be kept in mind that the final part will have shrunk significantly as compared to the designed
part.

6. Application:
Once the part has been appropriately printed and post-processed, it is now ready for use, and can
be employed to serve the purpose for which it was designed.

While expansive, the study and design optimization of Tesla valves has not yet been fully understood
and realized for 3D Tesla valves, with preliminary studies showing great promise for valves of such
type. The remarkable design freedoms offered by additive manufacturing make for Tesla valves that
can be 3D and high complex, with additive manufacturing coming into previously untapped
manufacturing potential. Tesla valves find applications in demanding environments, such as in the
aerospace and medical industries, and it is therefore of great importance to design and manufacture
3D Tesla valves with high diodicities.

19
3. Methodology
The methodology section of the project report details the exact implementation and thought process that
has gone into realizing the goals of the thesis project. The following sections address the implementation
in detail.

3.1. Design
With the previous design optimizations studied, six Tesla valve prototypes were designed, as seen in
figure 18. The prototypes were created also keeping in mind the physical limitations of the CerAM
VPP setup, which are discussed later. The prototypes were made primarily in SOLIDWORKS 2022,
with some initial iterations of the prototypes made in Fusion 360 and SOLIDWORKS, as is seen in
Appendix A. Design optimizations from previous literature, are as listed below with the
implementations:

1. Circular cross-section of the Tesla valve, from Bäckman et al[27].

2. Spiral 3D Tesla valve design, from Sano[30].
3. Increased feedback angle at the outlet of the Tesla loop, from Gamboa et al[28].
4. Wide Tesla loop inlet, from Bardell et al[9].
5. Narrow Tesla loop outlet, from Bardell et al[9].
6. Multi-stage Tesla valve design, from Bäckman et al [27]., Mohammadzadeh et al[29]., Sano[27],
[29], [30].

a) b) c)

d) e) f)

Figure 18: Front view of the six prototypes selected for CFD and practical testing. a)
Prototype I. b) Prototype II. c) Prototype III. d) Prototype IV. e) Prototype V. f) Prototype VI.

• Prototype I: Spiral Tesla valve with single Tesla loop, uniform main spiral thickness and
uniform inlet and outlet of Tesla loop.
• Prototype II: Spiral Tesla valve with single Tesla loop, uniform main spiral thickness, wide
Tesla loop inlet and narrow Tesla loop outlet.
• Prototype III: Spiral Tesla valve with single Tesla loop, non-uniform main spiral thickness,
wide Tesla loop inlet and narrow Tesla loop outlet.
• Prototype IV: Same as prototype I, but with two Tesla loops instead of one.
• Prototype V: Same as prototype II, but with two Tesla loops instead of one.
• Prototype VI: Same as prototype III, but with two Tesla loops instead of one.

The dimensions for the Tesla valve prototypes were chosen firstly based on the printing setup, and
secondly on the testing rig for practical evaluation of the Tesla valve prototypes. With printing setup
in mind, the peripheral dimensions of the Tesla valve prototypes were designed to not exceed 55
millimeters. The actual x-y print area of the 3D printer is 102 millimeters by 64 millimeters, but to
compensate for the shrinkage of the ceramic print during debinding and sintering, the prototypes to
be printed need to be scaled up in the slicing software. Once the oversized prototype is printed, the
debinding and sintering reduces the size of the oversized prototype to a size that is nearly identical
to the original CAD model, thereby resulting in a dimensionally accurate print with a dimensional
accuracy within ±0.1 millimeters. Due to the scaling up of the original CAD model that is done in
the slicing software (and therefore will be 3D printed), the effective area within which the original
CAD model can be designed reduces. With this constraint, the effective x-y build area was roughly

20
55 millimeters in both the x and y directions. The original CAD models were designed keeping this
in mind and can be seen in figure 19. It must be noted that despite the size constraints exclusive to
CerAM VPP (or rather, the 3D printer being utilized for printing, which in this case is the CeraFab
8500), the actual optimized Tesla valve can have peripheral dimensions greater than 55 millimeters
in both the x and y directions.

The main channel diameter is set at 20 millimeters (the calculations and reasoning behind choosing
this value for the main channel diameter is explained shortly), and the diameter of the Tesla loop for
all the prototypes is set at 0.7 times the main channel diameter based on CFD simulation tests run
for different Tesla main channel diameter multipliers, the results of which are outlined in Appendix
B. Similar tests were conducted for determining the optimum main channel diameter, although the
main channel diameter (and subsequently the Tesla loop channel diameter) were significantly based
on the build plate dimensions of the 3D printer as well. Two Tesla loops per main spiral were chosen
because more Tesla loops (3 or more) would congest the main channel and interfere with the
operation of the Tesla valve. For calculation of the maximum diameter of the main channel and Tesla
loop diameter, a simplified 2D diagram with calculations was implemented, as seen in figure 19.

r1

R r t
Figure 19: Left: Top view of Tesla valve main channel
and Tesla loop. Right: Sectional view of Tesla valve
indicating pipe diameter (d1) and pipe thickness (t).

The Tesla valve designed is larger in the horizontal direction, than in the vertical. Therefore,
horizontal dimension calculations were carried out to decide the main channel diameter. The limiting
equation governing the horizontal dimension of the Tesla valve is given in equation (3.1).

𝑅 + 2𝑟 + 𝑟1 + 𝑡 ≤
55 (3.1)
2

Where,
• R is the radius of the main channel, in 𝑚𝑖𝑙𝑙𝑖𝑚𝑒𝑡𝑒𝑟𝑠.
• r is the radius of the Tesla loop, in 𝑚𝑖𝑙𝑙𝑖𝑚𝑒𝑡𝑒𝑟𝑠.
• r1 is the inner radius of the Tesla valve, in 𝑚𝑖𝑙𝑙𝑖𝑚𝑒𝑡𝑒𝑟𝑠.
• t is the thickness of the Tesla valve, in 𝑚𝑖𝑙𝑙𝑖𝑚𝑒𝑡𝑒𝑟𝑠.

We know that r is 0.7 times the value of R, and we know that r 1 is 2 millimeters. We also know that
the thickness of the Tesla valve wall is 1 millimeter. These values were chosen based on the printing
setup, and not for technical or performance issues pertaining to the Tesla valve itself. Therefore,
equation (3.1) then becomes:
𝑅 + 2(0.7𝑅) + 2 + 1 ≤ 27.5 (3.2)
Solving for R, we get a value of 10.208 millimeters, which has been rounded down to 10 millimeters
for the sake of simplicity. Therefore, if the radius of the main channel is 10 millimeters, the diameter
of the main channel is 20 millimeters, and the diameter of the Tesla loop is 14 millimeters. It must
be noted, the simplified diagram seen in figure 19 represents a Tesla valve comprising a Tesla loop
positioned at the extreme periphery of the main channel, whereas the actual optimized Tesla valve
comprises a Tesla loop which sits slightly further away from this extreme periphery, as seen in figure
19. Therefore, a margin of around 1 millimeter can be assumed to exist within the x-y build volume,
guaranteeing that the optimized Tesla valve can be built within the build volume comfortably.

21
3.2. CFD Simulations
CFD simulations were performed in order to better understand the effects of design optimizations on
the Tesla valve prototypes. CFD simulations are essentially carried out to digitally replicate working
conditions of fluid flow, such that the problem tackled by the CFD simulations can generate results
identical to those obtained by practical tests. For the purposes of this thesis project, CFD simulations
were performed by first generating design prototypes to be tested on CAD software such as Fusion
360 and SOLIDWORKS, exporting the designs as a STEP file, importing the .STEP file into the
chosen CFD simulation software (ANSYS 2023 R2 Student), and then performing the CFD
simulations based on preferred mesh geometry and boundary conditions.

After the .STEP file is imported, meshing of the prototypes is required to solve the fluid dynamic
problem. Meshing is performed in order to break down the complex problem of fluid flow from a
general continuous Navier-Stokes equation into relatively simple algebraic equations, which together
represent the fluid flow of the problem. Since we are using water as the primary fluid for testing
within the Tesla valve, the most relevant Navier-Stokes equation for solving this problem is the
Navier-Stokes equation for incompressible fluid flow for Newtonian fluids. Water is an
incompressible fluid[38], as during fluid flow, water may display density changes, but these changes
are negligible and do not affect the result. Further, water follows mass conservation of flow, meaning
that the difference between amount of mass entering and leaving the control volume must be zero,
therefore no mass is stored by the control volume[39].

For incompressible Newtonian fluid flow, two general Navier-Stoker equations are used. The first
equation is the conservation of mass[40] equation, given in equation (3.3), and the second equation
is the momentum equation[41], given in equation (3.4).

∇ ∙ (𝜌𝒗) = 0 (3.3)
Where,
• ∇ represents divergence, which is a measure of imbalance between the input and output
values to a vector field. A high divergence indicates that the difference between input and
output vectors at a particular point tends to move away from zero.
• 𝜌 is the density of the fluid, in 𝑘𝑔/𝑚3.
• 𝒗 is the velocity field of the fluid particles, in 𝑚/s.

𝜕𝒗 (3.4)
𝜌( + 𝒗 ∙ ∇𝒗) = −∇𝑝 + 𝜇∇2 𝒗 + 𝜌𝒈
𝜕𝑡
Where,
• 𝜌 is the density of the fluid, in 𝑘𝑔/𝑚3.
𝜕𝒗
• is the rate of change of velocity with respect to time, in 𝑚/𝑠 2.
𝜕𝑡
• 𝑝 is the pressure of the fluid, in 𝑃𝑎.
• 𝜇 is the dynamic viscosity of the fluid, in 𝑃𝑎. 𝑠.
• 𝒈 is the acceleration vector due to gravity (body force), in 𝑚/𝑠 2.

Navier-Stokes equations were provided by French engineer Claude-Louis Navier and Irish
mathematician George Gabriel Stokes. These are a set of partial differential equations that are used
to describe various physical phenomena such as compressible and incompressible fluid flow,
conservation of momentum and conservation of mass. Certain assumptions are made to better fit the
Navier-Stokes equations for each of these applications (including assumptions such as density
change is negligible, lack of inviscid fluid, linear stress in material and isotropic fluid).

As mentioned before, meshing essentially enables the conversion of the above-mentioned complex
Navier-Stokes equations into simple algebraic equations that are more solvable by CFD solvers, that
can be applied to individual mesh elements, the solutions of which are then combined into a matrix
and solved. It stands to reason then that the more mesh elements are present, the better the solution

22
becomes in terms of emulating real-life scenarios. However, more mesh elements come at a
computational cost, where higher number of mesh elements means the CFD solver software needs
to allocate more memory to solve the problem, as well as take more time to arrive at a solution. To
determine if a mesh is sufficiently accurate to solve a particular fluid dynamics problem, grid
independence studies are conducted. A grid independence study compares a particular metric (in this
case diodicity) and how it changes depending on the number of mesh elements.

For the problem presented in this thesis report, a

grid independence study was conducted on Tesla
valve prototype I, a relatively simple version of the
Tesla valve, seen in figure 20. Note that the CFD
simulations were all run for the fluid region within
the Tesla valve prototypes, and the fluid region
seen in figure 20 is representative of one such
region. The chosen Tesla valve prototype for grid
independence studies was Tesla valve prototype I,
due to its relatively simple design, and subsequent
Figure 20: Fluid region of Tesla valve prototype I, easier and quicker CFD calculation. The mesh
chosen for grid independence studies.
generated for this prototype was uniform and had
low aspect ratios amongst the mesh elements across all chosen mesh setups. Further, the mesh was
visually inspected, and was found to have no zones of localized mesh element clusters, which can
potentially cause solutions to not converge accurately. The mesh for the medium mesh element setup
is as seen in figure 21. It must also be mentioned that the mesh generated comprised inflation layers.

Fluids in real-world settings have no velocity at the boundary of the container they are moving
through, and in CFD simulations this phenomenon is commonly referred to as the no-slip condition
(as the fluid does not slip at the boundary). CFD simulation solutions are calculated at the nodes of
mesh elements, and the solutions are then interpolated linearly between the centroids of these mesh
elements. This means, the variation of the solution between the mesh elements is linear, and not
quadratic (or in fact of a higher order). In fluid flow problems, the velocity varies from 0% developed
(stationary) at the boundary layer, to 99% developed (nearly full velocity) at the mainstream region
of the flow (usually the central region of any fluid domain, where the velocity is maximum), and this
velocity profile is essential to be captured by the CFD solution. In figure 22, the curved black line
represents the velocity profile curve along the axial direction of flow, the fluid flow direction is from
the left side to the right. The start of the velocity profile curve at the bottom represents the velocity
of the fluid when it is contact with the boundary wall and is therefore considered stationary. The end
of the curve at the top represents the fluid flow velocity when the velocity within the fluid region is
at its maximum, or when it is fully developed. In the left diagram in figure 22, the black rectangles
represent a sectional view of the mesh elements, with the black rectangles towards the bottom of the
images being closer to the boundary wall (the lowest rectangle is the mesh element in contact with
the boundary of the fluid domain) and the blue dots represent the centroids of the mesh elements,
where the calculated solution is interpolated. It can be seen from the right image in figure 22 that the
rectangles vary in thickness, with mesh elements closer to the boundary wall being thinner. The

23
thinner rectangles are inflation layers and are generated to capture a more accurate profile of the
fluid flow behavior closer to the walls.

Figure 21: Mesh generated for grid independence studies. a) Tetrahedral

mesh, isometric view. b) Tetrahedral mesh, face view. c) Polyhedral mesh,
isometric view. d) Polyhedral mesh, face view. e) Polyhedral mesh, close-up
view.

As can be seen from figure 22, the solution (represented by red lines) varies between the two mesh
element setups. In the left image of figure 22, the red lines appear more visible and further away
from the actual velocity profile, whereas the red lines appear to be nearly invisible and much closer
to the actual velocity profile in the right image. As the number of mesh elements increases, the
accuracy of the solution increases through this way, as more mesh elements can discretize the
problem statement to a higher degree, leading to more solution points for the same solution. More
mesh elements therefore increase the accuracy of the solution and are good tools to implement to
Velocity
Mesh profile of
elements actual flow
(black
curve)

Velocity
profile of
simulated
flow (red
curve)
Figure 22: Inflation layers and their impact on accuracy of results. Left:
Lack of inflation layers. Right: Presence of inflation layers.

better capture the flow profile of areas of interest, which may be boundary walls and intersection
points of fluid flow channels. This explanation also holds true for number of mesh elements in

24
general, with more mesh elements resulting in a higher degree of accuracy of the final solution. This
does, however, come at the computational and time cost as mentioned before.

Another reason to include inflation layers in the mesh, is to optimize the value of y plus of mesh
elements present at the outermost periphery of the fluid domain of Tesla valve prototype chosen for
CFD simulations. Y plus is essentially a dimensionless quantity determined by the CFD solver when
CFD simulations are run[42]. As seen from earlier sections, the velocity gradient close to the
boundary wall of the fluid flow region is an area of interest, and usually requires a very fine mesh to
accurately be depicted. To determine whether a mesh element is fine enough to capture the velocity
profile at the boundary of the fluid flow region, it was required to develop a spatial orientation system
that could measure the distance of the centroid of the mesh element closest to the boundary of the
fluid flow region, therefore
determining if the mesh element
is small enough. This system
could have been Cartesian with x,
y, and z components, but a radial
system with just one value
(distance of mesh element
centroid from boundary wall)
proved to be simpler, intuitive,
and accurate. This distance is then
used to calculate the y plus value
of the mesh element according to
equation (3.5), and to determine if
the mesh element is fine enough
to capture the behavior of fluid
Figure 23: “Law of the wall” graph generated by plotting flow at the boundary of the
experimental velocity values against distance of mesh elements from region.
boundary wall[42] (figure reused with permission from authors).

𝜌𝑦𝑢𝜏
𝑦+ = (3.5)
𝜇

Where,

• 𝑦 + is the value of y plus.

• 𝜌 is the density of the fluid, in 𝑘𝑔/𝑚3.
• 𝑦 is the distance between the boundary wall and centroid of mesh element, in 𝑚𝑖𝑙𝑙𝑖𝑚𝑒𝑡𝑒𝑟𝑠.
• 𝑢 𝜏 is the friction velocity, in 𝑚/𝑠.
• 𝜇 is the dynamic viscosity of the fluid, in 𝑃𝑎. 𝑠.

Figure 23 illustrates a relationship seen between the values of y plus and velocity, both values
obtained for the same mesh element. This graph tells us that for a certain value of y plus of our mesh
element, we can expect a particular value of velocity developed. For higher values of y plus, our
expected velocity increases as the larger mesh elements are unable to capture the laminar behavior
of the fluid flow seen close the boundary wall (higher y plus value correlates positively with larger
mesh elements). Through experimental analysis of fluid flow through pipes, it has been determined
that close to the boundary wall, regardless of mainstream flow of the pipe, the flow is laminar. Using
the previously derived y plus value to characterize this, a mesh element having a y plus value of
lower than 5 can accurately capture this profile, whereas a mesh element having a y plus value of
higher than 30 usually is not able to capture the profile accurately. Consequently, a y plus value of
lower than 5 is usually desirable when deciding mesh properties of a fluid flow region.

However, y plus is a function calculated by the CFD solver and is based on multiple variables. These
variables are not available to a user of the CFD simulation software, and therefore the value of y plus
is not one that can be calculated before the CFD simulation is run, but only after a calculation has
occurred. At high y plus values, the CFD solver references the “law of the wall” graph to generate
an accurate velocity profile even though the y plus value is high, and theoretically inaccurate. Since

25
y plus values are only calculated for the boundary mesh elements, inflation layers must be designed
properly to ensure that the y plus values of the boundary mesh elements are lower than 3, and are
ideally as small as possible. Based on trial and error, the following calculations were used to
determine the inflation layer settings to obtain an appropriate inflation layer thickness with a low
enough y plus value as well.

To calculate the boundary layer thickness, values of Reynolds number and pipe diameter are needed.
Equation (3.6) describes this relationship:

𝑝𝑖𝑝𝑒 𝑟𝑎𝑑𝑖𝑢𝑠
𝛿 = 0.37 × (3.6)
𝑅𝑒0 .2

Where,

• 𝛿 is the boundary layer thickness, in 𝑚𝑖𝑙𝑙𝑖𝑚𝑒𝑡𝑒𝑟𝑠.

With a pipe radius of 2 millimeters and a Reynolds number of 3200 (calculation for this is elaborated
upon in later sections), a boundary layer thickness of 0.1473 millimeters was calculated. Next, the
inflation layer thickness calculations were performed to calculate the first layer thickness, growth
rate and number of layers. The total inflation layer thickness was to exceed the boundary layer
thickness, such that the inflation layers span the entire thickness of the boundary layer, and are
therefore able to accurately map the behavior of flow parameters in the boundary layer.

To calculate the inflation layer thickness, the number of layers, growth rate and first layer thickness
all need to be assumed and fine-tuned to arrive at a value that is larger than the boundary layer
thickness. The inflation layer thickness can then be calculated from equation (3.7):

1−𝑟𝑛
𝐿𝑡 = 𝐹1 × (3.7)
1−𝑟

Where,

• 𝐿𝑡 is the inflation layer thickness, in 𝑚𝑖𝑙𝑙𝑖𝑚𝑒𝑡𝑒𝑟𝑠.

• 𝐹1 is the first layer thickness, in 𝑚𝑖𝑙𝑙𝑖𝑚𝑒𝑡𝑒𝑟𝑠
• 𝑛 is the number of layers.
• 𝑟 is the growth rate.

It is found that if first layer thickness is 0.01 millimeters, number of layers is 10 and growth rate is
1.2, the inflation layer thickness sums up to 0.2596 millimeters, which sufficiently covers up the
boundary layer thickness of 0.1473 millimeters. The perpendicular cross-section of these inflation
layers is as seen in figure 24, where the visible geometry is the inlet of the Tesla valve prototype for

Figure 24: Values of y plus generated based on the input

inflation layer settings.

26
which grid independence studies are carried out. The maximum value of y plus is observed to be
1.35, indicating that the inflation layers will accurately capture the velocity profile of fluid flow
within the Tesla valve prototype.

Once the mesh has been generated in the meshing window of ANSYS, the mesh (tetrahedral in
nature) is converted to a polyhedral mesh in the setup window of ANSYS. The polyhedral nature of
the mesh is also visible in figure 24, in the left image. This is done so that the mesh elements are
more ordered in the direction of fluid flow. Polyhedral mesh elements align themselves in the
direction of fluid flow, and this is advantageous as having mesh elements aligned with the direction
of fluid flow generate solutions between the aligned mesh elements, and therefore do not have
perturbations, or fluid flowing in unexpected directions due to misaligned tetrahedral mesh elements.
Tetrahedral mesh elements, while simple to implement and easy to execute solutions for, tend to be
aligned in random directions, which means that the generated solutions interpolated between the
centroids of the tetrahedral mesh elements connect between the misaligned mesh elements. This is
not an accurate representation of fluid flow, which tends to (for the most part) flow in linear and
straight paths without unexpected deviations or turns. Once the mesh elements are finally generated
and converted to polyhedral mesh elements, the solution is run based on viscosity models, type of
fluid, boundary conditions, solution methods, initialization methods and convergence conditions.
Each of these is described in greater detail in later parts of this thesis report.

Once the solution is run, the mesh elements are revisited and the inlet and outlets are reversed, to
change the fluid flow configuration from reverse to forward. Once the solutions have been run,
pressure values at points at the inlet and outlet are measured and are used for diodicity calculations.
Table 1 illustrates three different mesh element setups for the grid independence studies, with the
high mesh element setup comprising 368746 tetrahedral mesh elements and 950746 polyhedral mesh
elements after mesh conversion, medium mesh element setup comprising 326100 tetrahedral mesh
elements and 842587 polyhedral mesh elements and low mesh element setup comprising 301964
tetrahedral mesh elements and 727453 polyhedral mesh elements. The different mesh elements were
obtained after changing the maximum mesh element size in the ANSYS meshing window. This
metric determines the maximum length of every mesh element, and reducing this results in an
increase in the total number of mesh elements of the mesh, as the volume for which the mesh is
generated remains the same.
Table 1: Grid independence study results for different mesh setups.

Setup Initial elements Final elements Diodicity Difference (%)

(tetrahedral) (polyhedral)
High 368746 950746 1.355 -
Medium 326100 842587 1.357 -0.18%
Low 301964 727453 1.347 0.79%
It can be seen that with an increase in mesh elements, the diodicity appears to increase as well, with
the highest diodicity being seen in the high mesh element setup. However, the increase is negligible,
with a change in diodicity between the different setups not exceeding 1%, meaning that the change
is statistically insignificant. Due to this, it can be concluded that a polyhedral mesh element number
of around 850000 is sufficient to accurately capture the diodicities of the tested Tesla valves. Another
limitation that drives the mesh element number is that the student license of ANSYS 2023 R2 only
allows for mesh elements lower than 1 million, and therefore a mesh element number of 950746 (as
seen in the high mesh element number setup) is already quite close to the limit of the CFD software.
From the grid independence studies, it was concluded that a mesh element number of 850000 to
950000 would be used for testing Tesla valve designs.

The process of meshing has been extensively discussed, as has the justification for type of element
and number of mesh elements and inflation layers. In the setup window of ANSYS, further settings
are manipulated to arrive at the final solution for measuring diodicities in Tesla valve prototypes.
Firstly, the viscosity model is chosen. For the purposes of this thesis project, the SST k-omega model
has been chosen. While it has been suggested that the k-kL-omega model better predicts fluid flow
behavior[43] more accurately than other models, this has only been shown to be true up to Reynolds

27
number values of 1500, with no proof of accuracy of this model beyond this value of Reynolds
number. Previous literature indicates that diodicity increases with an increase in Reynolds
number[29], for which reason the testing of Tesla valve prototypes outlined in this thesis report has
been with the highest possible Reynolds number, a number that far exceeds 1500. Therefore, due to
a lack of evidence, the default viscosity model of SST k-omega has been chosen for testing of the
Tesla valve prototypes.

Since practical evaluation of the Tesla valve prototypes is most easily possible with water as the
fluid, water is chosen as the testing fluid for the CFD simulations as well. As for the boundary
conditions, the inlet boundary condition was set to a certain velocity value, and the outlet was kept
at constant pressure. The velocity value at the inlet of the Tesla valve prototypes was calculated by
taking into account the cross-sectional area of the Tesla valve prototypes, as well as the maximum
permissible flow rate of the practical evaluation setup for the prototypes. This was done to
corroborate the CFD simulation solutions and results with practical evaluation conducted for the
same boundary conditions. A maximum flow rate of 6 liters per minute is achievable by the practical
evaluation setup, and therefore the inlet fluid velocity is set to 0.8 meters per second. This was done
by first calculating the inlet velocity based on the maximum flow rate and cross-sectional area of the
Tesla valve as given in equation (3.8), and then revising and rounding off the velocity. Exact
calculations to obtain the inlet velocity are performed using equation (3.9), with results tabulated in
table 2. In table 2, D visco is the dynamic viscosity of the fluid (water), pipe ID is the inner diameter
of the Tesla valve, which is then used to calculate Vmax, the maximum fluid velocity calculated with
the cross-sectional area of the Tesla valve and maximum flow rate of the practical evaluation setup,
Remax is the maximum Reynolds number calculated based on equation (3.9), and Vmax (rev) is the
revised maximum velocity.

𝑄= 𝐴×𝑈 (3.8)

𝜌𝑈𝐿
𝑅𝑒 =
𝜇
(3.9)

Where,
• 𝑄 is the flow rate of the fluid, in 𝑚3 /𝑠 or 𝑙𝑖𝑡𝑒𝑟𝑠/s.
• 𝐴 is the cross-sectional area of the pipe, in 𝑚2.
• 𝑈 is the fluid flow velocity, in 𝑚/s.
• 𝑅𝑒 is the Reynolds number of the fluid flow.
• 𝜌 is the density of the fluid, in 𝑘𝑔/𝑚3.
• 𝐿 is the characteristic linear dimension of the fluid flow channel, in this case the inner
diameter of the Tesla valve, in 𝑚𝑒𝑡𝑒𝑟𝑠.
• 𝜇 is the dynamic viscosity of the fluid, in 𝑃𝑎. 𝑠.
Table 2: Calculations for inlet velocity of the Tesla
valve, for CFD simulations.
Density 𝜌 998.2 𝑘𝑔/𝑚3
D visco 𝜇 0.001003 𝑃𝑎. 𝑠
Pipe ID 0.004 𝑚𝑒𝑡𝑒𝑟𝑠
Pipe area 0.0000126 𝑚2
Vmax 0.7957747 𝑚/s
Remax 3167.865687
Remax (rev) 3200
Vmax (rev) 0.8 𝑚/s

Further, turbulence intensity percentage was calculated based on equation 3.10[44] , and a value of
5.841% was obtained.
1 (3.10)
𝐼 = 0.16 × (𝑅𝑒)−8 × 100
Where,

28
 • I is the turbulence intensity in %, with a low turbulence intensity indicating low turbulence
in the fluid flow.
• Re is the Reynolds number of the flow.

At the outlet, the default outlet boundary conditions were maintained (pressure outlet, constant
pressure). Once the boundary conditions were set, the solutions methods were chosen; the pressure-
velocity coupling scheme was set to Coupled and under spatial discretization, the gradient was set
to Least Squares Cell Based, pressure was set to Second Order[45], momentum was set to Second
Order Upwind, turbulent kinetic intensity was set to Second Order Upwind and specific dissipation
rate was also set to Second Order Upwind.

After setting up the boundary conditions and solution methods, report definitions were created to
monitor pressure values at the inlet and outlet of the Tesla valve prototypes. This was done not to
measure the pressure values and therefore the diodicity, but to determine when the solution is
converged. In CFD simulations, the first iterations of the solution tend to be incorrect and unusable,
as the solution has not yet reached a final and stable solution. To achieve this, the CFD solver must
iteratively solve the problem statement over and over again, until the solution has stabilized. To
monitor this, report definitions at the inlet and outlet were created, which reported how the inlet and
outlet pressures varied with iterations. Once the pressure values at the inlet and outlet stabilized and
remained constant with more and more iterations, the solution can be called converged. To start the

Figure 25: Report definition plots used to determine convergence of a CFD solution.
Left: Report definition plot of inlet pressure. Left: Report definition plot of outlet
pressure.
solution, the above steps were performed, after which the solution was initialized and calculated for
a total of 2000 iterations. An example of such convergence is illustrated in figure 25. From figure
25, we can deduce that the solution converged after around 200 iterations, as after 200 iterations the
solution did not change much, the inlet pressure report definition remained largely at one value of
average facet of pressure.

The Tesla valve prototypes chosen for CFD simulations were identical to the Tesla valve prototypes
3D printed for practical evaluation. However, since the Tesla valve prototypes comprised an

29
 increasing number of design optimizations, Tesla valve prototype VI was theorized to be the most
optimized and is as seen in figure 26.

a)
b)

c) d)

e)

Figure 26: Views of the theorized optimal Tesla valve. a) and b)

Isometric views. c) Top view. d) Right side view. e) Front view.

For each of the six Tesla valve prototypes, a mesh was generated such that the polyhedral mesh
element number (the number of mesh elements) was as close to 850000 as possible. Next, the CFD
simulations were run for each of the six prototypes, with 3 separate CFD simulations for each
prototype. Each of these three CFD simulations had slightly different meshes, and they differed in
the number of mesh elements each had. The maximum mesh size was changed in each of the CFD
simulation setups by 0.01 millimeters, such that the meshes for each set of CFD simulations per
prototype were slightly different, but still contained a similar number of mesh elements. This was
done to verify that the number of mesh elements did not change the overall output of the CFD
simulations, and also to verify that the CFD simulations were accurate and consistent, with each
simulation providing roughly the same output despite having different meshes.

The steps discussed in this section thus outline the general workflow of performing CFD simulations
for replicating practical fluid flow in a Tesla valve.

3.3. 3D printing
The printer used to 3D print the one Tesla valve prototype was CeraFab 8500 DLP 3D printer,
designed and produced by Lithoz, Austria, and is as shown in figure 27. As outlined in section 2.4,
the CeraFab 3D printer operates on the principle of CerAM VPP and produces green parts that
comprise ceramic particles held in a matrix of polymerized resin. After the green parts are printed,
the parts are subject to two post-processing steps. These steps are debinding and sintering, but a third
post-processing step may also be included, called pre-conditioning. Pre-conditioning involves
heating up the green part to a slightly higher temperature than room temperature such that uncured
resin on the surface of the green part demonstrates lower viscosity, and therefore is more easily
cleanable by compressed air or manual wiping.

30
CerAM VPP was used to determine if it was possible to produce ceramic Tesla valves through this
particular additive manufacturing process. However, since
CerAM VPP utilizes post-processing steps of debinding and
sintering, both of which can take up significant resources in terms
of machinery, operational costs, time and expertise, it was decided
that all prototypes of the Tesla valve would not be produced
through CerAM VPP. CerAM VPP is a technology just used to
demonstrate that the final optimized design of the Tesla valve can
indeed be 3D printed and post-processed in ceramic, but the
testing of all iterations of designs that led up to the design of the
final optimized Tesla valve do not need to be produced through
CerAM VPP. For this reason, one prototype would be produced
using CerAM VPP, but all six Tesla valve prototypes to be
practically evaluated would be produced using SLA 3D printing Figure 27: Lithoz CeraFab
8500 3D printer (figure
to conserve time and resources. reused with permission from
Lithoz GmbH).
The process parameters used to print one Tesla valve prototype
through CerAM VPP and process parameters for printing six
Tesla valve prototypes for practical evaluation through SLA are as given in table 3.

Table 3: Process parameters for CerAM VPP and

SLA 3D printing.

Printing CerAM VPP SLA

method
Printer CeraFab 8500 Prusa SL1
Slicer Lithoz in-built slicer PrusaSlicer 2.6.1
Pixel size (x-y 40 microns 47 microns
plane)
Material Lithalox 350 Prusament Resin
(alumina) Tough Transparent
Red
Layer height 0.025 millimeters 0.05 millimeters
Shrinkage 1.215 (x-y plane), None
compensation 1.275 (z-direction)
Post- Lithasol 20 Washing with IPA,
processing cleaning, drying, drying, UV curing
pre-conditioning,
debinding, sintering

Another reason for using SLA to print and test the Tesla valve prototypes instead of CerAM VPP is
that the build volume of the CeraFab 8500 3D printer is significantly smaller than the build volume
of the Prusa SL1, and therefore it would not be possible to print multiple prototypes through CerAM
VPP. Batch production of Tesla valve prototypes would become impossible through CerAM VPP but
can be possible through SLA due to the larger build volume and fewer resources and time to post-

31
process green parts. Figure 28 illustrates this difference in build volumes and shows the slicer view
for both CerAM VPP and SLA 3D printing.

b)
a)

d)

c)

Figure 28: Slicer view for 3D printing parts. a), b) and c) Slicer view for CerAM
VPP 3D printing. d) Slicer view for SLA 3D printing.

Once the green parts are printed, post-processing in the form of curing, unwanted resin removal and
drying takes place. For CerAM VPP, the pre-conditioning step gets rid of some unwanted resin
sticking to the surfaces of the green part, while compressed air and Lithasol 20 (a solvent for uncured
Lithalox resin) eliminate all traces of unwanted resin. Pre-conditioning involves heating up the green
part to temperatures as high as 120 ˚C, for an average cycle time of 72 hours. This step ensures that
all the unwanted resin on the surface of the green part is removed, as well as performing minor
thermal curing of the green part before it is debinded and sintered. After pre-conditioning, the green
part is subject to debinding, which involves heating the green part to temperatures as high as 1100
˚C to remove all the polymerized resin from the green part, resulting in a part that comprises only
ceramic particles loosely connected to each other. Debinding takes an additional 72 hours to
complete, after which sintering is performed to bind the ceramic particles finally thermally to each
other, producing a final part that has no unwanted resin or polymerized resin, and has fully fused
ceramic particles, much like a ceramic part that is conventionally manufactured. The thermal cycles
for pre-conditioning, debinding and sintering are as shown in table 4, and temperature profiles of the
three processes are as seen in figure 29.

32
 Temperature profiles for pre-conditioning, debinding and
sintering
1800.0
1600.0
1400.0
Temperature (˚ C)

1200.0
1000.0
800.0
600.0
400.0
200.0
0.0
0.0 500.0 1000.0 1500.0 2000.0 2500.0 3000.0 3500.0 4000.0 4500.0 5000.0
Time (minutes)

Pre-conditioning Debinding Sintering

Figure 29: Temperature profiles for pre-conditioning, debinding and

sintering.

33
 Table 4: Thermal cycles for pre-conditioning,
debinding and sintering for CerAM VPP 3D printing.

Process Preconditioning Debinding Sintering

Time (in Temperature Time (in Temperature Time (in Temperature
minutes) (in ˚C) minutes) (in ˚C) minutes) (in ˚C)
0 25 0 25 0 25

250 75 190 120 60 200

430 75 250 120 660 600
505 90 350 130 1020 1150

805 90 530 130 1590 1650

905 110 930 170 1710 1650
1625 110 1110 170 2250 1200
1675 120 1610 220 2878 50

3955 120 1850 220

4315 30 2000 250
2300 250

2675 325
2975 325
3185 430

3305 430
3975 1100
4331 30
Total time: 71.917 hours Total time: 72.183 hours Total time: 47.967 hours

While the processes of pre-conditioning, debinding and sintering are applicable for CerAM VPP, the
same cannot be said for SLA 3D printing on the Prusa SL1. Once the print has been completed on
the Prusa SL1, a majority of the work associated with obtaining the final part has been completed
already. After the print has finished, the print is washed in IPA to remove any uncured resin sticking
to the surface of the cured and freshly printed part. This step can take between 10 to 30 minutes,
followed by a final curing post-processing step. The curing post-processing step includes heating up
the cleaned part and exposing this part to UV radiation, to fully cure any unpolymerized resin present
within the printed part. After these post-processing steps, the final part is produced and is ready for
use.

3.4. Practical evaluation

Practical evaluation of the Tesla valve prototypes is done to ascertain how the Tesla valve prototypes
perform, whether they are able to behave as pressure and fluid flow diodes and which design
optimizations have the most effect on the diodicity of the Tesla valves. For practical evaluation, the
same prototypes that were designed for CFD simulations as seen in figure 18 were 3D printed with
SLA and implemented in a testing rig. The testing rig seen in figure 30 comprises two FDAD 33
high pressure sensors (labelled “a” and “b” in figure 30) that transmit their pressure output to an
Almemo 2680 data logger (labelled “c” in figure 30). The pressure sensors are capable of measuring
relative pressure between the limits of 0 and 1 bar, with a resolution of 0.0001 bar. The data logger

34
 d
d

a b c
Figure 30: Practical setup for practical evaluation of Tesla valve prototypes.
Left: The entire practical setup. Right: Zoomed-in view of the Tesla valve
prototype in action.
is capable of displaying three readings simultaneously, which for instance could be two pressure
readings from the pressure sensors, and their difference. Further, the fluid is pumped throughout the
system with the help of two small motors situated between the pressure sensors and the data logger
in figure 30. The outlet of the system is on the leftmost edge of figure 30, the orange bucket, whereas
the inlet is on the rightmost edge of figure 30, the blue channel. Finally, the Tesla valve prototype
(seen in bright red color and labelled “d”) is placed between two pipes that are connected to the
pressure sensors through T-pipe connectors.

The inner diameter of the pipes is 4 millimeters, which matches the inner diameter of the Tesla valve
prototypes. The two motors are capable of pumping fluid in two directions, which is useful in this
case because the fluid flow needs to be reversed to measure the diodicity of the Tesla valves. The
measurement of pressure at the inlet and outlet of the Tesla valve is therefore performed by the
pressure sensors when fluid is flowing within the system by the actuation of the two motors, and
these readings are measured and recorded by the data logger. Once the readings have been measured,
the direction of fluid flow is reversed and the same process occurs again, to measure the pressure
difference in the opposite direction. Both these readings are then used to measure the diodicity of the
Tesla valve. As mentioned before, the maximum flow rate provided by the two motors is 6 liters per
minute. The theoretical calculated velocity of fluid flow within the pipe was calculated to be 0.795
m/s, and it was practically found that the velocity within the system was 0.874 m/s. This discrepancy
can be explained by the fact that in theoretical calculations, we assume that velocity is constant
across a cross-sectional plane of the pipe. However, in practical situations, the velocity of fluid flow
is laminar at areas approaching the boundary walls of the pipe, and static where the fluid flow touches
the boundary. This results in a net reduction of cross-sectional area in which the flow is mainstream
and turbulent, and this reduction in area results in a higher net mainstream velocity, as described by
the equation of continuity, equation (3.11)[46]:

𝐴1𝑉1 = 𝐴2 𝑉2 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 (3.11)

Where,

• 𝐴1 is the cross-sectional area at point 1, in 𝑚2.

• 𝑉1 is the fluid velocity at point 1, in 𝑚/𝑠.
• 𝐴2 is the cross-sectional area at point 2, in 𝑚2.
• 𝑉2 is the fluid velocity at point 2, in 𝑚/𝑠.

35
4. Results and discussion
4.1. Printed designs
It was previously decided that the 3D printing of the Tesla valves to be practically evaluated would
first be printed using the SLA technique, and one prototype would be 3D printed using the CerAM
VPP technique to prove the feasibility of using this technique to 3D print 3D Tesla valves. The results
of 3D printing both of these versions of the 3D Tesla valve are as given in the next sections.

4.1.1. CerAM VPP

Figure 31: Different views of a Tesla valve prototype

successfully printed using CerAM VPP.

Through CerAM VPP and post-processing steps outlined in previous sections (pre-
conditioning, debinding, sintering), it is seen that the 3D printing of one prototype of the Tesla
valve was successful, and the print has functional structural integrity with no major cracks or
deformations as seen in figure 31. As mentioned in previous sections, the layer height to print
the ceramic parts is 0.025 millimeters, with an x-y build space resolution of 0.04 millimeters.

It must be noted that support structures were required to firstly manage the printing of the
overhang regions, and secondly to anchor the printed prototype onto the build plate. The base
of the ceramic print as seen in figure 28 was removed, leaving the final printed part with just
the Tesla valve itself, and vertical support structures that do not affect the functioning of the
Tesla valve prototype in any way. It must also be noted that while it is possible to use the
ceramic 3D printed prototype seen in figure 31 for practical evaluation of the Tesla valve, it is
not strictly necessary to. The prototype printed through CerAM VPP was composed of alumina
particles, and therefore the surface finish of the part was roughly 0.9 microns, with a relative
density of around 98.4%[47].

4.1.2. SLA 3D printing

36
 SLA 3D printing for prototyping the six Tesla valves proved to be successful, with the printed
prototypes being printed as shown in figure 32. It can be seen that the Prusa SL1 was successfully
able to 3D print the prototypes, and these prototypes had no visible defects despite having geometry
such as long bores, overhangs, thin sections and small inner diameters. No support structures were
used internally to produce
these prototypes, although
there was a need to have
support structures on the
outside of the print to
anchor the printed
prototype to the build plate
of the 3D printer, as seen in
figure 28d. The 3D printing
was completely successful,
and it was possible to use
these prototypes to
practically evaluate their
performance using the
Figure 32: The six Tesla valve prototypes 3D printed using SLA. Left: Tesla testing rig as shown in
valve prototypes I, IV, V and VI. Right: Tesla valve prototypes II and III. figure 30.

One possible reason for not

observing any visible defects is because the resin used was not a transparent resin, which has
been known to cause defects due to unwanted scattering of light[30]. The chosen red resin is
particularly good for visualization of internal recesses and defects, but also performs well such
that these defects do not occur in the first place.

4.2. CFD simulation results

The CFD simulations were performed for the six Tesla valve prototypes based on the procedure
outlined in section 3.3 of this thesis report. The exact pressure values obtained while performing
CFD simulations of each of the Tesla valve prototypes is provided in Appendix C, and the diodicity
results are as given in table 5.

Table 5: CFD simulation results for all Tesla valve

prototypes.

Prototype Prototype Prototype Prototype Prototype Prototype Prototype

number I II III IV V VI
Diodicity 1.248 1.198 1.129 1.313 1.219 1.306

A graphical interpretation of the CFD simulation results is as seen in figure 33.

37
 Variation in diodicity values as measured by CFD
simulations
1.35
1.313 1.306
1.3
1.248

Diodicity
1.25
1.219
1.198
1.2

1.15 1.129

1.1
0 1 2 3 4 5 6 7
Tesla valve prototype number

Figure 33: Variation in diodicity values as measured by CFD simulations for

different Tesla valve prototypes.

4.3. Practical evaluation results

Practical evaluations were carried out for the six Tesla valve prototypes based on the procedure
outlined in section 3.4 of this thesis report. The exact pressure values obtained while practically
evaluating each of the Tesla valve prototypes is provided in Appendix D, and the diodicity results
are as given in table 6.
Table 6: Practical evaluation results for all Tesla
valve prototypes.

Prototype Prototype Prototype Prototype Prototype Prototype Prototype

number I II III IV V VI
Diodicity 1.723 1.166 0.851 1.285 1.209 0.966

A graphical interpretation of the CFD simulation results is as seen in figure 34.

Diodicity values evaluated practically

2
1.723
1.8
1.6
1.4 1.285
1.166 1.209
1.2
Diodicity

0.966
1 0.851
0.8
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7
Tesla valve prototype number

Figure 34: Diodicity values as evaluated practically for different Tesla valve
prototypes.

38
 CFD simulations and practical evaluations both offer a comprehensive view on how diodicity is
affected by changing geometrical constructs and constraints. Table 7 compares the diodicities
obtained through the two testing methods, and figure 35 represents a graphical comparison between
CFD simulations and practical evaluation results for the Tesla valve prototype diodicities.
Table 7: Comparison between diodicity results obtained
from CFD simulations and practical evaluations.

Prototype Diodicity through CFD Diodicity through practical

number simulations evaluation

I 1.248 1.723
II 1.198 1.166
III 1.129 0.851

IV 1.313 1.285
V 1.219 1.209
VI 1.306 0.966

Comparison between results obtained from CFD

simulations and practical evaluations
2
1.8
1.6
1.4
Diodicity

1.2
1
0.8
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7
Tesla valve prototype number

CFD simulations Practical evaluation

Figure 35: Comparison between diodicity results obtained from CFD

simulations and practical evaluations.

4.4. Discussion
From the above testing methods and previous literature, we can observe that 3D printing has come
to the point where additively manufacturing complex geometry such as 3D Tesla valves is possible
in materials as diverse as polymers and ceramics. With high printing resolution, flexibility in design
and ease of prototyping numerous iterations of designs, it can safely be said that additive
manufacturing is not just an emergent technology, but an effective and highly relevant method for
production of prototypes and final parts. This is true despite the extensive post-processing associated
with CerAM VPP, as turnaround times for prototypes can still be in the context of days, which is still
quicker than traditional manufacturing methods that are focused on mass production. This thesis
project highlights one use case of additive manufacturing, and the use case has been implemented
successfully.

39
As for the performance of the Tesla valves themselves, certain trends and observations become
apparent after taking a closer look at the results. For starters, we observe that the diodicity of 3D
Tesla valves is not as high as expected from previous literature. Certain design optimizations in
previous literature highlight 2D Tesla valves with diodicities much higher than the ones obtained
through this thesis project, which seems counterintuitive at first. It must be noted however, that
certain design optimizations were either incorporated into all the designs or were not incorporated at
all. For instance, increasing the feedback angle at outlet of Tesla loop, circular cross-section of Tesla
valve and spiral design (to produce a 3D Tesla valve, the primary goal of the thesis project) were all
design optimizations that have made their way into every single Tesla valve prototype, and therefore
their effect on the increase of diodicity of the Tesla valve is not quantified. Other optimizations as
outlined in appendices A and B have also made their presence felt in the final prototypes of the tested
Tesla valve prototypes.

Discussions related to diodicities of the final prototypes of the Tesla valves are only restricted to
results obtained by practical evaluation of the Tesla valve prototypes, for reasons given in earlier
sections. It is observed that amongst the single loop and double looped prototypes, the Tesla valve
prototypes that comprise variable main channel thickness perform the worst. From forward and
reverse pressure data present in appendix D, it is clear that the main channel thickness had a very
direct and detrimental impact on the diodicity of the Tesla valve prototypes. Tesla valve Prototype
VI with 2 points where the main channel was narrowly designed showcased pressure difference
values a magnitude higher than Tesla valve prototypes with a constant main channel thickness. It
appears that altering the main channel thickness and reducing it induces the well-understood Venturi
effect, causing high localized fluid velocity and low pressure, but creating high pressure through
fluid which has not entered the narrow region of the main channel. The high pressure can be
explained by a resistance to fluid flow caused by the narrow main channel region. This high pressure
is seen in both the forward and reverse fluid flow configurations and serves little purpose than to
increase pressure in both orientations. It can then be concluded that main channel thickness should
be kept uniform in as many cases and instances as possible when considering the diodicity
optimization of 3D Tesla valves.

While this pressure increase in both forward and reverse fluid flow configurations is understandable,
it is observed that the diodicities of Tesla valve prototypes III and VI both dip below 1, indicating
that the forward pressure difference is more than the reverse pressure difference. One explanation
for this could be that at the narrow section in the main channel, the flow is highly turbulent as a result
of high localized velocity, which causes fluid particles to enter the Tesla loop even in the forward
fluid flow configuration. This is indeed seen in vector lines of velocity flow present in appendix C,
and confirms that fluid flow, which should ideally stay only in the main channel of the Tesla valve
when in the forward fluid flow configuration, technically flows into the Tesla loop. It is then easy to
understand the reasons for this increase in pressure difference in the forward fluid flow configuration;
when the fluid flows through the narrow section in the main channel, a highly localized increase in
velocity and drop in pressure occurs, which increases the pressure upstream and downstream in the
entire Tesla valve. Further, the fluid enters the Tesla loop, and this fluid then flows through the Tesla
loop and reconverges with the fluid that was flowing through the main channel, and this convergence
of flow is what causes yet another pressure increase. Ironically, this exact method of pressure
increase is required in the reverse fluid flow configuration, which is being seen in the forward fluid
flow configuration. These two pressure increase steps contribute to the unexpectedly high pressure
difference in the forward fluid flow configuration, whereas only one of these pressure increase steps
are present in the reverse fluid flow configuration through the same Tesla loop prototype.

In the reverse fluid flow configuration, the fluid flow splits into two at the inlet of the Tesla loop,
and the two flows reconverge at the area where the main channel is narrow. However, since the fluid
has enough space to flow through (by means of two channels, the main channel and Tesla loop, recall
that the Tesla loop inlet has been widened as well), the first mechanism of pressure increase does not
occur as it does in the forward fluid flow configuration. Only one mechanism of pressure increase is
observed in the reverse fluid flow configuration, which is the primary method of pressure increase
in Tesla valves. Due to this discrepancy, in Tesla valve prototypes with a main channel whose

40
thickness reduces, forward pressure difference can be expected to exceed reverse pressure difference,
resulting in a diodicity lower than 1.

Following this train of thought, it can be understood why Tesla valve Prototype I was the most
successful. Constant main channel thickness and no narrowing of the Tesla loop leads to effortless
flow in the forward fluid flow configuration, with relatively restricted fluid flow in the reverse fluid
flow configuration, which is precisely how Tesla valves operate as pressure diodes. The impact of
narrowing sections of the Tesla valve is so severe in fact, that Tesla valve Prototype IV (two Tesla
loops, no narrowing whatsoever) boasts a higher diodicity than Tesla valve Prototype II (single Tesla
loop, narrowing of main channel only), when the overall trend clearly shows that multiple loops fare
significantly worse than single loop designs.

Additionally, minor losses must also be touched upon to understand why the designed Tesla valves
did not perform as well as expected. In fluid dynamics, minor losses are losses in pressure head
experienced when a pipe undergoes one or more of the following geometric changes:

1. Sudden expansion or contraction in the pipe

2. Gradual expansion or contraction in the pipe
3. Bend in the pipe, or change in direction of fluid flow
4. Fluid flow passing through entrances and exits
5. Obstructions in the fluid flow path

These can all be understood as hindrances to fluid flow, and hindrances to the inertia carried by
flowing fluid. Naturally, when fluid begins to carry momentum and inertia, any change to this inertia
causes pressure to rise upstream. In the case of the designed Tesla valves of this report, the spiral
design of the Tesla valve constantly changes the direction of the fluid flow, forcing fluid particles to
alter their velocity vector at all points in the Tesla valve. This drastically increases the minor losses
encountered in the Tesla valve, and results in significant pressure increases at the inlet of whichever
configuration is being used regardless of if it is the forward or reverse fluid flow configuration.
Pressure increases from minor losses in this manner are roughly constant between the two
configurations, but since the value remains constant, it constitutes a much larger percentage of the
forward pressure difference, as the forward pressure difference is much lower than the reverse
pressure difference. This therefore means that the diodicity of the spiral design Tesla valve suffers a
lot, as a consequence of minor losses in both the forward and reverse fluid flow configurations.

An explanation as to why multiple loop 3D Tesla valves perform poorly could be multifold. In a
single Tesla loop configuration, the reverse fluid flow configuration sees a lot of resistance to flow,
and the reverse fluid flow configuration sees relatively lower resistance to flow (but still sees a
certain amount, owing to the non-uniform nature of a Tesla valve due to Tesla loop presence). For a
single loop design, this difference in resistance is stark because the sole Tesla loop offers a massive
amount of resistance to fluid flow in the reverse fluid flow configuration, as all the fluid in the Tesla
valve is being manipulated immediately by the Tesla loop. This huge resistance far outweighs the
resistance observed by the Tesla valve in the forward fluid flow configuration. However, in a Tesla
valve with multiple Tesla loops, the story is slightly different. After the fluid has passed the first
Tesla loop, the flow moves forward with significantly reduced momentum (due to a reduction in
fluid velocity, which comes at the price of increased fluid pressure) into the second Tesla loop. It can
be appreciated that the effect of the first Tesla loop would be significantly more than the effect of the
second Tesla loop, considering that the first Tesla loop manipulates faster and more energetic fluid
as compared to the second Tesla loop. Therefore, the total pressure difference would be contributed
by the first Tesla loop in a majority, and the second Tesla loop in a minority. This means that the
pressure difference from two Tesla loops is not the mathematical multiplication of the effect of a
single Tesla loop, multiplied by two, the effect of more and more Tesla loops results in diminishing
returns for pressure difference, and therefore diodicity.

Now when we consider the forward fluid flow configuration, the above scenario looks quite different.
Fluid flowing in the forward fluid flow configuration does not meet much resistance when flowing,
only a marginal decrease in fluid velocity when crossing the area comprising the outlet of the Tesla
loop. This then means that the majority of fluid still carries forward enough momentum after the first

41
Tesla loop, such that the second Tesla loop contributes an equal amount of flow resistance as the first
Tesla loop (the fluid flow remains roughly the same through both sections). This results in the total
pressure difference in the forward fluid flow configuration being the mathematical multiplication of
the pressure increase seen at the first Tesla loop by the number of loops present, which in our case is
two. In the reverse fluid flow configuration, had the total pressure drop been the mathematical
multiplication of the pressure increase seen at the first Tesla loop, the diodicity of the Tesla valve
amongst single loop and multiple loop Tesla valves would have been identical. However, in the
reverse fluid flow configuration, an increase in the number of loops results in slightly increased
pressure difference, but a disproportionately higher pressure difference in the forward fluid flow
configuration (the fluid flow remains largely unchanged in the forward fluid flow configuration),
which explains why multiple loops don’t increase the diodicity of 3D Tesla valves, but rather can
contribute to a decrease.

Overall, the maximum diodicity observed through practical evaluation is 1.723, which is close to the
value obtained by Sano [30] , with a value of 1.74. The designs bear resemblance with each other in
that they both comprise a single Tesla loop, with no narrowing of the main channel or Tesla loop
sections of the Tesla valve, adding further credibility to the results of this thesis report. It should also
be noted that CFD simulations were not able to accurately predict the trend or actual values of a
majority of testing scenarios of the Tesla valve prototypes. This may be due to unreliable software
algorithms, incorrect setup of problem statements or inaccurate meshing settings, or other pertinent
reasons. It must therefore be said that in evaluation of components such as 3D Tesla valves, it is best
to also attach a section of practical evaluation, as is done in this thesis report.

42
5. Conclusions
This thesis project aimed to firstly evaluate if additive manufacturing can manufacture complex geometries
that may be required to manufacture to attain high diodicity Tesla valves, and secondly to investigate different
3-dimensional geometry for the diodicity enhancement of Tesla valves. Two similar additive manufacturing
technologies were used to produce Tesla valve prototypes, the technologies being SLA and CerAM VPP. SLA
3D printing was used to produce initial prototypes quickly and cheaply, whereas CerAM VPP was used to
demonstrate that CerAM VPP can indeed produce fully functional ceramic parts produced through additive
manufacturing, and post-processed through different heat treatment processes particular to parts made out of
ceramic. Secondly, the project aimed to incorporate various design optimizations into a 3D Tesla valve to
increase the diodicity and performance of the Tesla valve. CFD simulations and practical evaluations were
conducted to assess the effectiveness of the design optimizations, and water was chosen as the working fluid
in the Tesla valves.

Through the work conducted in the course of this thesis project, it can be concluded that:

1. Complex 3D Tesla valves can indeed be manufactured through AM, more particularly through CerAM
VPP and SLA 3D printing, with no discernable flaws or issues in the final finished product. With the
printing parameters outlined in previous sections, it is possible to manufacture 3D Tesla valves in an
appreciable time frame, with materials and 3D printers available in the present day.
2. Design optimizations from 2D Tesla valves to 3D Tesla valves do not always translate well, and certain
design optimizations from the 2D version of an optimized Tesla valve resulted in an overall decrease in
the diodicity and performance of 3D versions of the Tesla valve. This decrease in performance can be
attributed to the fact that a 3D Tesla valve involves more complex design, and therefore encounters a
higher number of pressure and velocity perturbations, the likes of which are not usually seen in 2D Tesla
valves.
3. Despite certain design optimizations hindering the performance of the Tesla valves, iterations of the 3D
Tesla valve have been successfully manufactured and tested. CFD simulations and practical evaluations
both reveal that incorporating certain design optimizations result in a Tesla valve which has approximately
the same diodicity as the first spiral design Tesla valve, while design optimizations such as narrowing and
widening of the Tesla loop in a 3D Tesla valve generally has disastrous consequences for the diodicity of
the Tesla valve. Indeed, diodicities of less than 1 have been observed in Tesla valves comprising the above-
mentioned design optimization.

43
6. Future work
For future work on this project, the following topics could be worthwhile starting points for new areas of
research:

• Investigation of how different fluids interact and behave within the optimized Tesla valve, and
whether different Tesla valves offer better diodicities for different operating fluids.
• Testing of different dimensions of Tesla valves, such that the dimensions that could lead to optimal
diodicity of the Tesla valve are not restricted by the size of the 3D printer (or any other manufacturing
machine) that manufactures the Tesla valve.
• Investigation of the effects of multiple Tesla loops for each spiral turn of the Tesla valve, by fitting
more Tesla loops onto the main spiral.
• Generation of optimized Tesla valve designs by computational and generative fluid flow, extending
past conventional pipe and tubular designs for Tesla valves.
• Practical evaluation of Tesla valves in an oscillatory flow testing setup, which best replicates not only
the intended application for Tesla valves according to this thesis report, but also replicates the ideal
conditions for maximum performance of a Tesla valve.
• Inclusion of more metrics (such as alteration of inlet and outlet dimensions of Tesla loops,

44
7. Acknowledgements
I would like to express my deepest thanks to Uppsala University, and their faculty at the department of
Materials Science and Engineering, for the vast knowledge and experience I have amassed during my time at
Uppsala University. Uppsala University has also provided me with the opportunity to complete my thesis
project, and master’s degree in additive manufacturing to a degree that I am professionally satisfied with,
something that I am very grateful for. I would also like to thank Fraunhofer Institut für Keramische
Technologien und Systeme, for enabling this thesis project, and for providing me with all the resources and
knowledge I needed at any point of time during my thesis project.

In particular, I would like to extend my thanks to Urban Wiklund and Francesco D’Elia for their continuous
support and approval at major steps of the thesis project, and their guidance and advice for me when I was
struggling with my thesis project.

My deepest thanks also go out to the fine folk over at the Fraunhofer Institut für Keramische Technologien
und Systeme, with whom I have carried out the work required for this thesis project. Uwe Scheithauer has
been an invaluable resource to me, as have his team consisting of bright minds such as Stephan Holtzhausen,
Martin Propst, Eric Schwarzer-Fischer and Nick Alsdorf, each of whom have assisted me in this thesis project
in their own unique ways and contributions, without which this thesis project would have been impossible to
execute.

Lastly, I extend my thanks to my friends who supported me during the thesis project, at times when I needed
help the most. Without their support, it is difficult to imagine the completion of this thesis project.

45
References
[1] Kyle Daniels and Erik Tiemensma, ‘Process and control valves • News • Innovations • Fluid Handling
Pro’, Process and Control Valves. Accessed: May 20, 2023. [Online]. Available:
https://fluidhandlingpro.com/fluid-process-technology/fluid-process-control-valves/

[2] L. Witek, ‘Failure and thermo-mechanical stress analysis of the exhaust valve of diesel engine’, Eng Fail
Anal, vol. 66, pp. 154–165, Aug. 2016, doi: 10.1016/J.ENGFAILANAL.2016.04.022.

[3] B. Liu, J. Zhao, and J. Qian, ‘Numerical analysis of cavitation erosion and particle erosion in butterfly
valve’, Eng Fail Anal, vol. 80, pp. 312–324, Oct. 2017, doi: 10.1016/J.ENGFAILANAL.2017.06.045.

[4] Z. jiang Jin, Z. xin Gao, M. rui Chen, and J. yuan Qian, ‘Parametric study on Tesla valve with reverse flow
for hydrogen decompression’, Int J Hydrogen Energy, vol. 43, no. 18, pp. 8888–8896, May 2018, doi:
10.1016/J.IJHYDENE.2018.03.014.

[5] S. Yang, Y. Tang, and Y. F. Zhao, ‘A new part consolidation method to embrace the design freedom of
additive manufacturing’, J Manuf Process, vol. 20, pp. 444–449, Oct. 2015, doi:
10.1016/J.JMAPRO.2015.06.024.

[6] J. W. Halloran, ‘Ceramic Stereolithography: Additive Manufacturing for Ceramics by

Photopolymerization’, Annu Rev Mater Res, vol. 46, pp. 19–40, Jul. 2016, doi: 10.1146/ANNUREV-
MATSCI-070115-031841.

[7] Nikola Tesla, ‘Nikola Tesla – Valvular Conduit (Patent)’, Unites States Patent Office, pp. 01–16, Feb.
1920.

[8] A. R. Keller, J. Otomize, A. P. Nair, N. Q. Minesi, and M. Spearrin, ‘High-diodicity impinging injector
design for rocket propulsion enabled by additive manufacturing’, 2022, doi: 10.2514/6.2022-1265ï.

[9] R. Louis Bardell, ‘The Diodicity Mechanism of Tesla-Type No-Moving-Parts Valves’, Washington, 2000.

[10] Q. Wang, B. Jiang, B. Li, and Y. Yan, ‘A critical review of thermal management models and solutions of
lithium-ion batteries for the development of pure electric vehicles’, Renewable and Sustainable Energy
Reviews, vol. 64, pp. 106–128, Oct. 2016, doi: 10.1016/J.RSER.2016.05.033.

[11] J. Lin, H. N. Chu, K. Thu, M. Wojtala, F. Gao, and K. J. Chua, ‘Novel battery thermal management via
scalable dew-point evaporative cooling’, Energy Convers Manag, vol. 283, p. 116948, May 2023, doi:
10.1016/J.ENCONMAN.2023.116948.

[12] Y. Lu et al., ‘Performance optimisation of Tesla valve-type channel for cooling lithium-ion batteries’, Appl
Therm Eng, vol. 212, p. 118583, Jul. 2022, doi: 10.1016/J.APPLTHERMALENG.2022.118583.

[13] S.-F. Zheng et al., ‘Performance evaluation with turbulent flow and heat transfer characteristics in
rectangular cooling channels with various novel hierarchical rib schemes’, Int J Heat Mass Transf, vol.
214, p. 124459, Nov. 2023, doi: 10.1016/j.ijheatmasstransfer.2023.124459.

[14] Z. Li, B. Zhang, D. Dang, X. Yang, W. Yang, and W. Liang, ‘A review of microfluidic-based mixing
methods’, Sensors and Actuators A: Physical, vol. 344. Elsevier B.V., Sep. 01, 2022. doi:
10.1016/j.sna.2022.113757.

[15] J. E. Martin, L. Shea-Rohwer, and K. J. Solis, ‘Strong intrinsic mixing in vortex magnetic fields’, Phys
Rev E, vol. 80, no. 1, p. 16312, Jul. 2009, doi: 10.1103/PhysRevE.80.016312.

[16] P.-H. Huang et al., ‘An acoustofluidic micromixer based on oscillating sidewall sharp-edges’, Lab Chip,
vol. 13, no. 19, pp. 3847–3852, 2013, doi: 10.1039/C3LC50568E.

[17] K. Zhang, Y. Ren, L. Hou, X. Feng, X. Chen, and H. Jiang, ‘An efficient micromixer actuated by induced-
charge electroosmosis using asymmetrical floating electrodes’, Microfluid Nanofluidics, vol. 22, no. 11, p.
130, 2018, doi: 10.1007/s10404-018-2153-2.

46
[18] Q. M. Nguyen, J. Abouezzi, and L. Ristroph, ‘Early turbulence and pulsatile flows enhance diodicity of
Tesla’s macrofluidic valve’, Nature Communications 2021 12:1, vol. 12, no. 1, pp. 1–11, May 2021, doi:
10.1038/s41467-021-23009-y.

[19] X. Yang et al., ‘Suppression of pressure feedback of the rotating detonation combustor by a Tesla inlet
configuration’, Appl Therm Eng, vol. 216, Nov. 2022, doi: 10.1016/j.applthermaleng.2022.119123.

[20] Q. Wang and S. Zhang, ‘Thrust characteristics research on continuous rotating detonation engine’, Case
Studies in Thermal Engineering, vol. 47, p. 103127, Jul. 2023, doi: 10.1016/J.CSITE.2023.103127.

[21] W. Lin, J. Zhou, S. Liu, Z. Lin, and F. Zhuang, ‘Experimental study on propagation mode of H2/Air
continuously rotating detonation wave’, Int J Hydrogen Energy, vol. 40, no. 4, pp. 1980–1993, Jan. 2015,
doi: 10.1016/J.IJHYDENE.2014.11.119.

[22] Z. Xia, H. Ma, Y. He, G. Ge, and C. Zhou, ‘Visual experimental research on the propagation instabilities
in a plane-radial rotating detonation engine’, Aerosp Sci Technol, vol. 122, Mar. 2022, doi:
10.1016/j.ast.2022.107335.

[23] J. Zhou et al., ‘Investigation of pressure gain characteristics of RDE with Tesla valve inlet scheme’, Exp
Therm Fluid Sci, vol. 146, Aug. 2023, doi: 10.1016/j.expthermflusci.2023.110909.

[24] Z. Ji, B. Zhang, H. Zhang, B. Wang, and C. Wang, ‘Reduction of feedback pressure perturbation for
rotating detonation combustors’, Aerosp Sci Technol, vol. 126, Jul. 2022, doi: 10.1016/j.ast.2022.107635.

[25] Y. Zhong, D. Jin, Y. Wu, and X. Chen, ‘Investigation of rotating detonation wave fueled by “ethylene-
acetylene-hydrogen” mixture’, Int J Hydrogen Energy, vol. 43, no. 31, pp. 14787–14797, Aug. 2018, doi:
10.1016/j.ijhydene.2018.05.174.

[26] X. Yang, F. Song, Y. Wu, J. Zhou, Z. Yang, and Y. Kou, ‘Experimental study on tesla valve and bypass
manifold to suppress feedback of rotating detonation engine fuel by kerosene’, Acta Astronaut, vol. 211,
pp. 755–763, Oct. 2023, doi: 10.1016/j.actaastro.2023.07.018.

[27] E. Bäckman and M. Willén, ‘Tesla Valve for Hydrogen Decompression: Fluid Dynamic Analysis’,
Stockholm, Jun. 2019.

[28] A. R. Gamboa, C. J. Morris, and F. K. Forster, ‘Improvements in fixed-valve micropump performance

through shape optimization of valves’, Journal of Fluids Engineering, Transactions of the ASME, vol. 127,
no. 2, pp. 339–346, Mar. 2005, doi: 10.1115/1.1891151.

[29] K. Mohammadzadeh, E. M. Kolahdouz, E. Shirani, and M. B. Shafii, ‘Numerical investigation on the

effect of the size and number of stages on the tesla microvalve efficiency’, Journal of Mechanics, vol. 29,
no. 3, pp. 527–534, Dec. 2012, doi: 10.1017/jmech.2013.29.
[30] Lion Sano, ‘Parameteruntersuchung dreidimensionaler, mittels CerAM VPP gefertigter Teslaventile’,
Dresden, Mar. 2022.

[31] Robert W Messler, ‘CERAMIC AND GLASS ATTACHMENT SCHEMES AND ATTACHMENTS’, in
Integral Mechanical Attachment, Elsevier, 2006, pp. 239–277. doi: 10.1016/B978-075067965-7/50027-4.

[32] E. S. Mirdamadi, M. Haghbin Nazarpak, and M. Solati-Hashjin, ‘Metal oxide-based ceramics’, in

Structural Biomaterials: Properties, Characteristics, and Selection, Elsevier, 2021, pp. 301–331. doi:
10.1016/B978-0-12-818831-6.00012-4.

[33] E. Medvedovski, ‘Wear-resistant engineering ceramics’, Wear, vol. 249, no. 9, pp. 821–828, Sep. 2001,
doi: 10.1016/S0043-1648(01)00820-1.

[34] K. Zhao, F. Ye, L. Cheng, J. Yang, and X. Chen, ‘An overview of ultra-high temperature ceramic for
thermal insulation: Structure and composition design with thermal conductivity regulation’, J Eur Ceram
Soc, Jul. 2023, doi: 10.1016/j.jeurceramsoc.2023.07.046.

[35] R. W. Serth, ‘HEAT CONDUCTION’, Process Heat Transfer, pp. 1–41, Jan. 2007, doi: 10.1016/B978-
012373588-1/50004-8.

47
[36] Ian Gibson, David Rosen, and Brent Stucker, ‘Additive Manufacturing Technologies’, Dec. 2010.

[37] E. Schwarzer-Fischer et al., ‘CerAMfacturing of silicon nitride by using lithography-based ceramic vat
photopolymerization (CerAM VPP)’, J Eur Ceram Soc, vol. 43, no. 2, pp. 321–331, Feb. 2023, doi:
10.1016/j.jeurceramsoc.2022.10.011.

[38] B. Cushman-Roisin and J. M. Beckers, ‘Equations of Fluid Motion’, International Geophysics, vol. 101,
pp. 77–97, Jan. 2011, doi: 10.1016/B978-0-12-088759-0.00003-1.

[39] Y. Bai and Q. Bai, ‘Hydraulics’, Subsea Engineering Handbook, pp. 315–361, Jan. 2019, doi:
10.1016/B978-0-12-812622-6.00013-0.

[40] K. Van Canneyt and P. Verdonck, ‘Mechanics of Biofluids in Living Body’, in Comprehensive Biomedical
Physics, vol. 10, Elsevier, 2014, pp. 39–53. doi: 10.1016/B978-0-444-53632-7.01003-0.

[41] M. Wu, J. K. Huusom, K. V. Gernaey, and U. Krühne, ‘Modelling and simulation of a U-loop Reactor for
Single Cell Protein Production’, in Computer Aided Chemical Engineering, vol. 38, Elsevier B.V., 2016,
pp. 1287–1292. doi: 10.1016/B978-0-444-63428-3.50219-8.

[42] Christopher E. Brennen, ‘An Internet Book on Fluid Dynamics’.

[43] S. M. Thompson, T. Jamal, B. J. Paudel, and D. K. Walters, ‘Transitional and turbulent flow modeling in
a tesla valve’, in ASME International Mechanical Engineering Congress and Exposition, Proceedings
(IMECE), American Society of Mechanical Engineers (ASME), 2013. doi: 10.1115/IMECE2013-65526.

[44] ‘ANSYS FLUENT 12.0 User’s Guide - 7.3.2 Using Flow Boundary Conditions’. Accessed: Sep. 11, 2023.
[Online]. Available: https://www.afs.enea.it/project/neptunius/docs/fluent/html/ug/node238.htm

[45] ‘ANSYS FLUENT 12.0 Theory Guide - 18.4.1 Discretization of the Momentum Equation’. Accessed: Sep.
11, 2023. [Online]. Available:
https://www.afs.enea.it/project/neptunius/docs/fluent/html/th/node371.htm#sec-solve-pinterp-theory

[46] N. CRACIUNOIU and B. O. CIOCIRLAN, ‘Fluid Dynamics’, Mechanical Engineer’s Handbook, pp.
559–610, Jan. 2001, doi: 10.1016/B978-012471370-3/50009-7.

[47] Lithoz GmbH, ‘Material Overview LCM Technology’, 2023. [Online]. Available: www.lithoz.com

48
 8. Appendices
Table of Contents
1. Appendix A ................................................................................................................................ 49
2. Appendix B ................................................................................................................................ 51
3. Appendix C ................................................................................................................................ 55
4. Appendix D ................................................................................................................................ 63

49
Appendix A
This section explores the influence of increasing feedback angle, as per design optimizations by Gamboa et
al[28]. Each design comprises a different feedback angle, which is determined by an offset value that the Tesla
loop sits at from the main channel. Therefore, the metric driving the increase in feedback angle is an offset
distance between the Tesla loop and main channel of the Tesla valve.

Figure I: Reverse (left) and forward (right) flow configuration pressure plots for 0
millimeters offset value, corresponding to least feedback angle.

Figure II: Reverse (left) and forward (right) flow configuration pressure plots for 5
millimeters offset value.

Figure III: Reverse (left) and forward (right) flow configuration pressure plots for
10 millimeters offset value.

Figure IV: Reverse (left) and forward (right) flow configuration pressure plots for
15 millimeters offset value.

50
 Figure V: Reverse (left) and forward (right) flow configuration pressure plots for 20
millimeters offset value.

Table I: Diodicity change with change in Tesla loop offset values.

Setup Reverse pressure Forward pressure

Diodicity
difference (Pa) difference (Pa)
0 millimeters offset 453.97 199.57 2.27
5 millimeters offset 510.87 176.21 2.89
10 millimeters offset 577.82 188.24 3.06
15 millimeters offset 549.84 189.41 2.90
20 millimeters offset 549.84 189.25 2.90

Offset value versus diodicity

3.5
3.06
2.89 2.9 2.9
3

2.5 2.27
Diodicity

1.5

0.5

0
0 mm 5 mm 10 mm 15 mm 20 mm
Offset value

Graph I: Diodicity change with change in Tesla loop offset values.

51
Appendix B
This section explores the change in diodicity of a spiral Tesla valve design with a change in the ratio between
main channel diameter and Tesla loop diameter.

Figure I: Reverse (left) and forward (right) flow configuration pressure plots for a
ratio of 0.3 between the main channel and Tesla loop diameters.

Figure II: Reverse (left) and forward (right) flow configuration pressure plots for a
ratio of 0.4 between the main channel and Tesla loop diameters.

Figure III: Reverse (left) and forward (right) flow configuration pressure plots for a
ratio of 0.5 between the main channel and Tesla loop diameters.

52
Figure IV: Reverse (left) and forward (right) flow configuration pressure plots for a
ratio of 0.6 between the main channel and Tesla loop diameters.

Figure V: Reverse (left) and forward (right) flow configuration pressure plots for a
ratio of 0.7 between the main channel and Tesla loop diameters.

Figure VI: Reverse (left) and forward (right) flow configuration pressure plots for a
ratio of 0.8 between the main channel and Tesla loop diameters.

53
 Figure VII: Reverse (left) and forward (right) flow configuration pressure plots for a
ratio of 0.9 between the main channel and Tesla loop diameters.

Figure VIII: Reverse (left) and forward (right) flow configuration pressure plots for a
ratio of 1 between the main channel and Tesla loop diameters.

Table I: Diodicity change with change in main channel and

Tesla loop diameter ratios.

Loop Reverse pressure difference Forward pressure difference

Diodicity
multiplier (Pa) (Pa)
0.3 372.33 315.09 1.18
0.4 375.48 284.46 1.32
0.5 421.34 307.38 1.37
0.6 376.33 284.61 1.32
0.7 405.91 284.28 1.43
0.8 403.99 284.26 1.42
0.9 372.09 284.25 1.31
1 253.81 188.32 1.34

54
 Main channel and Tesla loop diameter ratio
versus diodicity
1.6 1.43 1.42
1.32 1.37 1.32 1.31 1.35
1.4 1.18
1.2
1
Diodicity

0.8
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1 1.2
Main channel and Tesla loop diameter ratio

Graph I: Diodicity change with change in main channel and Tesla loop
diameter ratios.

55
Appendix C
This section outlines the test results of the CFD simulations run, to calculate the diodicity of the Tesla valve
prototypes that were later 3D printed and practically evaluated. Given below are the pressure map and velocity
vectors for the entire Tesla valve, and for the areas where the Tesla loop interacts with the main channel. These
plots are followed by quantified results for pressure differences seen in the Tesla valve prototypes in the
forward and reverse fluid flow configurations, along with calculated diodicity values for all the prototypes.

Figure I: Pressure and velocity vector profiles of Tesla valve Prototype I in the reverse
fluid flow configuration.

Figure II: Pressure and velocity vector profiles of Tesla valve Prototype I in the
forward fluid flow configuration.

56
Figure III: Pressure and velocity vector profiles of Tesla valve Prototype II in the
reverse fluid flow configuration.

Figure IV: Pressure and velocity vector profiles of Tesla valve Prototype II in the
forward fluid flow configuration.

57
Figure V: Pressure and velocity vector profiles of Tesla valve Prototype III in the
reverse fluid flow configuration.

Figure VI: Pressure and velocity vector profiles of Tesla valve Prototype III in the
forward fluid flow configuration.

58
 Figure VII: Pressure and velocity vector profiles of Tesla valve Prototype IV in the
reverse fluid flow configuration.

Figure VIII: Pressure and velocity vector profiles of Tesla valve Prototype IV in the
forward fluid flow configuration.

59
Figure IX: Pressure and velocity vector profiles of Tesla valve Prototype V in the
reverse fluid flow configuration.

Figure X: Pressure and velocity vector profiles of Tesla valve Prototype V in the
forward fluid flow configuration.

60
Figure XI: Pressure and velocity vector profiles of Tesla valve Prototype VI in the
reverse fluid flow configuration.

Figure XII: Pressure and velocity vector profiles of Tesla valve Prototype VI in the
forward fluid flow configuration.

61
 Table I: CFD simulation results for Tesla valve
prototype I.

Simulation run number 1 2 3

Number of mesh elements

846098 884179 815480
(polyhedral)
Reverse pressure difference (Pa) 904.213 903.438 897.153
Forward pressure difference (Pa) 722.657 725.498 719.483

Diodicity 1.251 1.245 1.247

The average diodicity of Tesla valve prototype I through CFD simulations is thus 1.248.

Table II: CFD simulation results for Tesla valve

prototype II.

Simulation run number 1 2 3

Number of mesh elements

876021 919008 851981
(polyhedral)
Reverse pressure difference (Pa) 898.602 919.896 923.606
Forward pressure difference (Pa) 775.252 781.513 733.85

Diodicity 1.159 1.177 1.259

The average diodicity of Tesla valve prototype II through CFD simulations is thus 1.198.

Table III: CFD simulation results for Tesla valve

prototype III.

Simulation run number 1 2 3

Number of mesh elements
850815 890488 820196
(polyhedral)

Reverse pressure difference (Pa) 10048.08 9957.557 9909.372

Forward pressure difference (Pa) 8802.131 8870.291 8825.386
Diodicity 1.142 1.123 1.123

The average diodicity of Tesla valve prototype III through CFD simulations is thus 1.129.

62
 Table IV: CFD simulation results for Tesla valve
prototype IV.

Simulation run number 1 2 3

Number of mesh elements
949469 984969 912339
(polyhedral)

Reverse pressure difference (Pa) 974.694 978.348 974.363

Forward pressure difference (Pa) 742.502 745.97 740.581
Diodicity 1.313 1.311 1.316

The average diodicity of Tesla valve prototype IV through CFD simulations is thus 1.313.

Table V: CFD simulation results for Tesla valve

prototype V.

Simulation run number 1 2 3

Number of mesh elements
897505 929968 863604
(polyhedral)
Reverse pressure difference (Pa) 969.688 1000.4 989.052

Forward pressure difference (Pa) 806.504 813.076 808.399

Diodicity 1.202 1.23 1.223

The average diodicity of Tesla valve prototype V through CFD simulations is thus 1.219.

Table VI: CFD simulation results for Tesla valve

prototype VI.

Simulation run number 1 2 3

Number of mesh elements

870457 896918 851981
(polyhedral)
Reverse pressure difference (Pa) 19719.36 19900.53 19001.5
Forward pressure difference (Pa) 14920 14933.53 15035.58

Diodicity 1.321 1.333 1.264

The average diodicity of Tesla valve prototype VI through CFD simulations is thus 1.306.

63
Appendix D
This section outlines the test results of the practical evaluations, to determine the diodicity of the Tesla valve
prototypes that were first simulated. Given below are pressure and pressure difference values for each of Tesla
valve prototypes.

Table I: Practical evaluation results Table II: Practical evaluation results

for Tesla valve prototype I. for Tesla valve prototype II.

Reverse pressure 1616 Reverse pressure 1014

difference (Pa) difference (Pa)
Forward pressure 938 Forward pressure 870
difference (Pa) difference (Pa)

Diodicity 1.723 Diodicity 1.166

Table III: Practical evaluation results Table IV: Practical evaluation results
for Tesla valve prototype III. for Tesla valve prototype IV.

Reverse pressure 5262 Reverse pressure 1558

difference (Pa) difference (Pa)
Forward pressure 6186 Forward pressure 1212
difference (Pa) difference (Pa)

Diodicity 0.851 Diodicity 1.285

Table V: Practical evaluation results Table VI: Practical evaluation results

for Tesla valve prototype V. for Tesla valve prototype VI.

Reverse pressure 1582 Reverse pressure 15792

difference (Pa) difference (Pa)
Forward pressure 1308 Forward pressure 16344
difference (Pa) difference (Pa)

Diodicity 1.209 Diodicity 0.966

64