Sub-cooled nucleate boiling flow cooling experiment in a small rectangular channel

(1)

Sub-cooled nucleate boiling flow cooling experiment in a

small rectangular channel

Lway Al-Maeeni

lbam@kth.se

Master of science thesis

Department of Physics

Division of Nuclear Reactor Technology

Royal Institute of Technology (KTH)

Stockholm, Sweden,

(2)

TRITA-FYS 2015:49

ISSN 0280-316X

(3)

Abstract

This master thesis project is an experimental attempt to analyse the behaviour of the sub-cooled boiling regime. A small rectangular channel is constructed based on certain parameters, as the hydrodynamic entrance length, to get firstly a fully developed flow, heat flux needed to be deliv-ered to the system to get a bubbly flow, and other important material specifications as thermal conductivity and temperature limitations of the components utilized. Besides the construction of the channel a water loop is designed with a proper micropump to get a sufficiently stable and low flow rate in the channel. Heat flux is delivered to the laminar flow (Re≈320) through heat-ing cartridges inserted into the aluminium wall providheat-ing a maximum power of around 2800 W. Temperature measurements are done for fixed inlet temperatures and different wall temperatures; this measurements together with the input potential to the cartridges and total heating area are later on used to calculate the heat flux which is plotted against wall temperature which follows the expected results. A high speed camera is also used to have a visualization of the bubbles created in the channel.

(4)

(5)

Acknowledgements

Without doubt a master thesis marks an important stage in the academical life of a student. This stage is the last effort to move on from a student life to the industry or perhaps to pursuit higher research ambitions. For some of them these 6 months’ project becomes a nightmare, a stressful period of distress and sleepless nights. Perhaps some are still wandering around not knowing what they are doing or where to go. Maybe this one of the darkest periods in their life.

Fortunately this is not my reality. Since I was offered to apply as a diploma worker at ABB Corporate Research by my Professor Henryk Anglart at KTH I knew something great was awaiting. As I started, a bit nervous perhaps, not really knowing what I was doing for some days, the people at the company did not hesitate to show their support at the very beginning, which was a great strengthening factor that made me confident to accomplish the goals of my work. Never did I see angry faces nor reluctant attitudes when I asked for help. This was a very positive and enriching life experience.

In my project, since the aim was to build an experimental set up, many people were involved in its construction. And I dare to say that workers from literally all levels were part of it. From professors and post docs to people like the cleaning lady. Engineers, plumbers, electricians, secretaries, everyone supported in one way or another.

Beside to thank my supervisors Henryk Anglart (KTH), Tor Laneryd and Rebei Bel Fdhila (ABB CRC) for their support and guidance, I would like to thank Mohamed Ali Rahmani (Dal´ı), who is the heat transfer and flow lab manager for his constant supervising and advising, Sead Travancic for helping me out and showing me how things work, Lukas Migas for his expertise in electrical

engineering, the electricians Mats and Gugge, the plumbers ˚Ake and Anders, the carpenter Bosse,

to the experienced people working at the labs Gerhard and Bertil, to EMAUS mekaniska who manufactured important parts of the experimental set up, to all the collegues that showed so much support, especially from the other diploma workers. The list of people I would like to thank is long.. May they all find guidance and clarity in their lives.

I want as well to thank my family, each and everyone of them, wherever they are in the World, for their support and patience, and for letting me fight my own battles. Life is short and useless if it is not used to achieve greatness, so let’s be positive and fight for greatness!

(6)

(7)

Chapter 1 Introduction

As the search for more compact devices continues, the cooling of these systems has become a major area of research. Be it in power electronic devices or in the nuclear industry, the cooling phenomena has a great impact on the efficiency and safety parameters of these systems. Since it is known that the phase change of water from liquid to vapour requires some energy, the heat removal from heated walls will be greater than through solely flowing liquids or gases. Therefore it is highly interesting to investigate the cooling features of the sub-cooled nucleate boiling flow regime in narrow channels with geometries similar to those used for the cooling of transformers for example. Among power electronic devices the cooling through water is essential since the temperature can

reach around 120o_{C and heat fluxes between 5-20W/cm}2_{. Heat removal by two phase flow cooling}

is more effective due to the high heat transfer coefficient of the phase change.

In the nuclear industry the two phase cooling through narrow channels is also present. For instance the Jules Horowitz reactor (JHR) being constructed in France has a particular fuel arrangement in the reactor core. The JHR is a 100M W materials testing reactor that is intended to replace the OSIRIS test reactor in the CEA dependencies in Saclay, France. Among the objectives of the JHR is to produce and supply radioactive isotopes for medical treatments as well as research and development of fuel materials for the nuclear industry. [1][2]

The fuel element in the reactor core of the JHR is composed by concentric fuel plates that are 1.37mm thick and separated by hydraulic gaps of 1.95mm thick held by stiffeners as seen in Figure 1.1b. In case of a pressure drop due to loss of flow or any instability in the reactor core, one of the most important safety mechanisms is heat removal and therefore the importance of understanding the flow behaviour [1]. Optimally a sub-cooled nucleate boiling two phase flow is able to remove more heat from the heated surfaces but would be difficult to control if it goes further to other flow regimes with higher void coefficients and reaches the critical heat flux. At the point of critical heat flux the flow is mainly steam and from that point on, if the wall temperature increases, the heat flux will start to decrease and thereafter the whole system temperature will increase provoking infrastructure damages. Later on in this report there will be an explanation on this phenomena that is summarized in the boiling Nukiyama curve shown in Figure 2.1.

(10)

1.1. OBJECTIVES 2

(a) Cross section of reactor core (b) Concentric fuel plates with coolant gap of around 2mm

Figure 1.1: A total of 37 fuel elements consisting of concentrical fuel plates with gaps for coolant of around 2 mm. [2]

1.1 Objectives

For this master thesis project there are several objectives that were specified from the beginning of the work and even secondary objectives that were specified either at the beginning or added when their realization seemed plausible through the working progress. As for the main objective is to construct a functional experimental set up for a small rectangular channel, where experiments may be performed for different controlled input parameters as for example wall temperature, inlet tem-perature, flow rate, etcetera. The secondary objectives are to perform temperature measurements at various positions for the wall heat flux and even the water temperatures, and visualization of the flow phases with a high speed camera. The idea of the visualization is to capture the moment of the formation of the bubbles at their onset of nucleate boiling points and focus on the sub-cooled nucleate boiling phase of the flow.

1.2 Scope

This work does not include any computational fluid dynamic calculations (CFD) nor any simulation of any of the phenomenon present in the experiment and its set up. The focus of the work is solely the construction of the experimental set up.

(11)

Chapter 2 Theoretical background

2.1 Sub-cooled nucleate boiling and bubble formation

The boiling process in a heated channel is considered complex because it undergoes phase changes and enthalpy increase. Different flow stages are formed depending on the heat flux to the fluid. A scheme of this boiling process where the heat flux delivered to the flow is depicted in Figure 2.1 as a function of the temperature.

The area of focus in this project is the sub-cooled nucleate boiling region (between point A and B) i.e. the area where the first bubbles are formed and yet not merged with others forming a slug flow. This is the very first stage of the boiling process and presents a very inhomogeneous void fraction distribution [9].

At the beginning of the sub-cooled nucleate boiling, bubbles are formed at nucleation sites when the wall temperature reaches temperatures equal or above the flowing fluid’s saturation point. The nucleation sites are defined as minuscule imperfections or cavities in the heating wall surface where bubbles are formed. A bubble is evaporated liquid that detaches the wall when reaching a critical size. This bubble will deliver heat to the bulk when collapsing (condensing), increasing the flow temperature.

The position where the first bubble is formed, in point A in Figure 2.1, is called the onset of nucleate boiling (ONB). ONB determines the end of the single-phase forced-convection of the liquid and the beginning of the two-phase flow in the channel. In Figure 2.2 the wall and liquid temperature distribution is shown for different stages in nucleate flow boiling and their distance notations from the literature [8]. A more detailed discussion on the needed distance to attain sub-cooled bubbly flow is done in section 2.5.

2.2 Hydrodynamic entrance length, L

e

When water is introduced into the channel it will take some distance to stabilize the flow and attain a well distributed velocity profile, i.e. a hydraulically fully developed flow. This is important to happen before it reaches the testing section where there will be heat flux delivered to the flow. The hydraulically fully developed flow can be defined as the flow when the velocity pattern is constant

(12)

2.2. HYDRODYNAMIC ENTRANCE LENGTH, LE 4 log∆Tsup logq00 A B C D

Figure 2.1: Scheme of the Nukiyama boiling curve

through the channel along the flowing axis and the radial, this means the velocity derivative is more or less equal to zero[11]. At this point the velocity profile is parabolic for laminar flow, and flatter in case of turbulent flow [15]. The distance it takes to stabilize the flow and attain a

hydraulically fully developed flow is called hydrodynamic entrance length, Le. A scheme of the

hydrodynamic development of a certain mass flow W is shown in Figure 2.3.

In Figure 2.3 the parameter δ denotes the limit to the inviscid flow in the hydrodynamic entrance region. When the hydraulically fully developed flow is reached at point C in the figure, δ becomes equal to the radius of the channel (in case of a tube formed channel and laminar flow). The velocity profile after this point becomes parabolic and with a zero velocity derivative. The hydrodynamic

entrance length is then the distance between point A and C and is denoted Le. This length can be

estimated according to different mathematical models depending on the type of flow (laminar or turbulent), the channel geometry, viscosity and flow rate. An educated guess for this length could be done by applying, in these models, the geometry and parameters expected to be used in the experiment. For this purpose, basic hydrodynamic equations are sufficient.

The hydraulic diameter is per definition:

DH =

4 · A

Pw

(2.1)

Where A is the cross section area and Pw is the wetted perimeter. This together with the material

density (ρ), the fluid velocity (v) and the material dynamic viscosity (µ) can be used for the Reynolds number calculation, which is defined as:

ReDH = ρ ·

v

µ· DH =

v

ν · DH (2.2)

Where ν is the kinetic viscosity and is defined as:

ν = µ

(13)

5 2.2. HYDRODYNAMIC ENTRANCE LENGTH, LE

Figure 2.2: Temperature distribution in flow boiling (ONB denoted as zON B)[8]

r x δ δ A W B C

Figure 2.3: Schematic layout of the hydraulically fully developed flow

Dynamic and/or kinetic viscosities and other hydraulic properties of water are obtained from MATLAB’s XSteam tables. Table 2.1 shows some of the viscosity values of liquid water at different temperatures.

Table 2.1: Dynamic and kinetic viscosities of water at different temperatures

T [o_C] _{µ [10}−6_· kg m·s] ν [10 −6_· m2 s ] 20 1002.7 1.003 40 653.0 0.658 60 466.4 0.474 80 354.4 0.365 100 282.2 0.297

According to the literature [15] the correlation for the estimation of hydrodynamic entrance length for laminar flow is:

Le

D

lam

(14)

2.2. HYDRODYNAMIC ENTRANCE LENGTH, LE 6

This correlation is valid for laminar flow (Re ≈ 2300) in a tube, where the water enters with uniform velocity (see Figure 2.3). As for the turbulent flow correlation, although there is no real general expression, the equation could be approximated to:

10 . Le

D

turb

. 60 (2.5)

Articles discussing the issue of the entrance length for different cross sectional geometries relevant to this work have been published. In Plessis and Collins’ article [11] for example, the apparent friction factor-Reynolds number product is used to find a more accurate correlation for different types of short straight ducts. According to the article, citing the work of Shah and London [12], the product is given by following relation:

f Re = √3.44

x+ (2.6)

Here x+ _{is the dimensionless axial distance. The friction factor is depending on the mean}

ve-locity gradient at the wall of the channel. The apparent friction factor is the average friction between inlet and the channel and is affected by certain asymptotic conditions for the flow in the radial coordinate until the hydrodynamic entrance length. There is a more developed ex-planation on the mathematical analysis of these asymptotes affecting the relation in the work of Plessis and Collins [11]:

fapp= 1 δx Z δx 0 f dx (2.7)

For a rectangular channel the correlation is depending on the ratio between the longer side and the shorter α∗ = a/b, where a and b denote the longer and shorter side respectively. Tables are given for different values, some of them are summarised in Table 2.2.

Table 2.2: Values for f Re and Lefor different α∗-values, were 1 corresponds to a square

and 0 to parallel plates

α∗ f Re L+ e 1.00 14.23 0.059 0.50 15.55 0.049 0.25 18.23 0.0356 0.10 21.17 0.0264 0.05 22.47 0.0234 0.02 23.36 0.0217 0.00 24 0.0205

If the channel is very wide and small, the hydrodynamic entrance length could be approximated

to parallel plates, which corresponds to α∗ = 0 in Table 2.2, i.e. Le = 0.0205m. In later sections

the calculations pertaining hydraulic entrance length of the experimental set up will be done and compared with the results found in the literature.

(15)

7 2.3. HEAT TRANSFER

2.3 Heat transfer

Heat is a form of energy that seeks to attain equilibrium when two media with different temper-atures meet. The hotter media will transfer its energy to the media with less energy. In fact when talking about heat what is actually meant is this transferring process and therefore it is to be mentioned in this report as heat transfer, and not only heat, for the sake of accuracy. Heat transfer is defined as follows [15]:

”Heat transfer (or heat) is thermal energy in transit due to a temperature difference”. The physical mechanisms of heat transfer are mainly three: conduction, convection and radiation. Physically they are different heat transfer processes, but all of them follow the same principle, transferring the higher molecular energy in the direction of decreasing temperatures. In an exper-imental set up some of the mechanisms may be more dominant than the others since the elements used en their surfaces of contact between the parts may be different in nature.

2.3.1 By conduction

Conduction is often referred to as molecular activity where energy transfer occur due the particle interaction in form of lattice vibrations. An important characteristic is that there is no flow of material, no bulk motion. An example of conduction is a tea spoon submerged in hot water, the heat will then transfer from the highly energetic water (high temperature) to the spoon, which is initially cold. Conduction heat transfer follows Fourier’s law, for a temperature distribution T (r, t) [8]

q”(r, t) = −λ · ∇T (r, t) (2.8)

Here q00 is the heat transfer measured in W/m2_{, which is proportional to the temperature gradient.}

The constant λ is a transport property that depends on the material where the heat is transported, and it is called thermal conductivity. In Figure 2.4 the elements of the before mentioned equation can be found. The heat transfer in direction x affects a material with thermal conductivity λ and

width L. The heat will go from the side with higher heat (T1) to the side with lesser (T2), i.e. the

side with lower temperature.

q_x00= −λ · ∆T

x ⇒ −λ ·

T2− T1

L (2.9)

2.3.2 By convection

Convection type of heat transfer is the transportation of energy involving bulk in motion. Typically it is about a heated wall delivering heat to a flow of gases or liquids, as the case is in a boiling water reactor core, where the heat from the fuel elements delivers energy through the cladding to the coolant that takes it away. The same phenomena is found usually in heat sinkers used in power electronic devices. Convective heat transfer can be subdivided into natural and forced convection for both laminar and turbulent flows. The natural convection is when the flow is driven by buoyancy forces, while forced convection is driven by other forces like a circulation pump. As explained by Anglart [8], the heat transfer follows Newton’s equation of cooling (see equation 2.10),

(16)

2.3. HEAT TRANSFER 8 T x T(x) T1 T2 L q_x00

Figure 2.4: Conduction, energy is conducted to areas with less heat

transfer coefficient h. The value of the heat transfer coefficient depends on the flow conditions and the natural properties of the fluid. As the heat transfer may vary on the heating surface the value of heat transfer coefficient is a sum of several local heat transfer values over that surface, as explained in equation 2.11, where F is the surface.

q00 y u(y) y Tf Tw T (y)

Figure 2.5: Convection heat transfer diagram

q00= h · (Tw− Tf) (2.10) h = 1 F Z F hlocdF (2.11)

In Figure 2.5 a scheme over the boundary layer development is shown for convective heat transfer. The velocity distribution u(x) goes in direction x. Heat flux from the wall makes the temperature

(17)

9 2.4. EXAMPLES OF SMALL CHANNEL EXPERIMENTAL SET UPS

2.3.3 By radiation

Heat transfer through radiation is also a present phenomena in the experimental set up, and will be regarded as heat losses since the energy is escaping the system. Radiation heat transfer is the emission of energy by a hot element. Usually the the heat transfer through radiation goes together with convective heat transfer, since the media where the energy is transferred to is rarely only vacuum. In case of vacuum, then the portion of heat transfer is solely radiative, otherwise the presence of gas contribute to the convection. [15] An easy example of radiation heat transfer is the warming effect from a fire pit a cold night outdoors. The surface that is emitting heat to the surrounding has an emissive power which has a maximum value following the Stefan-Boltzman law

giving the rate of energy per unit area, Eb. This law follows equation 2.12 which is dependent on

the surface temperature Tw to the power of four, and multiplied by the Stefan-Boltzmann constant

σ (5.67 · 10−8W/m2_K4_{). [15] A so called black body emits all the energy, but it is not often the}

case in real life materials. Therefore a factor ε is added to the Stefan-Boltzmann equation. This factor is a radiative property specific for the surface material and has a varying value between 0 and 1.

E = εEb = εσTw4 (2.12)

Another factor that affects the thermal energy of the material is the absorbed quantity of irradiation G from another heat source. This absorbed portion is denoted here as α and varies between 0 and 1, giving the relation for the total absorbed irradiation as:

Gabs = αG (2.13)

Assuming Tw as the wall temperature and Tsur the surroundings temperature different from each

other then the irradiation can be approximated to a blackbody at Tsur. This leads to G = σTsur4

giving the final relation for heat flux through radiation:

q_rad00 = q A = εEb(Tw) − αG = εσ(T 4 w − T 4 sur) (2.14)

2.4 Examples of small channel experimental set ups

There are many publications about experiments on two phase flow cooling in small channels. In this section some few of these experiments are reviewed to see the different solutions for the construction of functional experimental set ups. Investigating these experiments is a fundamental part of this work since they give examples and ideas of instruments and materials and are actually later on used in this master thesis project for the construction of the set up.

Peng & Wang (1993)[3]

This experiment consisted on a 2mm thick and stainless steel plate with three micro-channels with cross section 0.6 x 0.7mm. The analysed test area is shown in Figure 2.6a. The sub-cooled inlet

(18)

2.4. EXAMPLES OF SMALL CHANNEL EXPERIMENTAL SET UPS 10

temperature varied between 40 and 70oC. The liquid velocities in these channels were adjusted

between 1.5 to 4m/s. The plate was heated with an electrical current transformer matched with a SCR voltage regulator as shown in Figure 2.6b. The heat flux was not homogeneously distributed along the channels.

(a) Test section (b) Schematic of circuit

Figure 2.6: Set up arrangement of the Peng and Wang experiment

Figure 2.7: Result of Peng and Wang experiment

Several measurements were performed on this experiment. The heat flux could be measured by the total input power applied in means of voltage and current. Temperature measurements for the liquid along the channels were done with the help of thermocouples mounted on the back of the plate, and in the inlet and outlet of the channel. From these measurements different heat transfer correlation parameters were calculated and compared in graphs (see Figure 2.7)

Something particular for this experiment is that although all parameters for a fully developed boiling were met, no bubbles were formed in the micro-channels. According to the article this phenomena could be explained by the idea that there is a minimum bulk size necessary for the bubble growth in the liquid.

(19)

Qu & Mudawar (2003) [4]

A study was performed on a set-up with twenty one parallel micro-channels each with dimensions 0.231 x 0.713mm in a copper test block. The length of the channels was approximately 50mm.

The inlet water temperature varied between 30 and 60o_{C. The inlet velocity varied between}

0.14 and 0.40m/s. As for the heating there were 12 cartridge heaters, as shown in Figure 2.8a, embedded in the copper powered by a 0 to 110 VAC variac. The channels had an upstream and a downstream plenum to ensure the flow distribution. The whole heating arrangement and location of the elements is shown in Figure 2.8b.

(a) Heater arrangement (b) Heating section

Figure 2.8: Qu and Mudawar experiment arrangement

Measurements were done for several mass velocities of two different inlet temperatures, 30 and

60o_{. The flow rate of the water was quantified by two rotameters where the water passed through.}

Temperature were measured with thermocouples along the channels. The electrical power applied to the system was measured with a wattmeter. The heat flux, calculated with the readings of the thermocouples and power supplied and plotted against the difference between wall and inlet temperature (see Figure 2.9).

Liu & Garimella (2007) [5]

In an improved experiment from the same university as Qu and Mudawar’s article, twenty-five micro-channels were cut into the top face of a copper test block. The dimensions of the channels were 0.275 x 0.636mm and 0.406 x 1.063mm and they were 25.4mm long. The inlet temperature for

(20)

2.4. EXAMPLES OF SMALL CHANNEL EXPERIMENTAL SET UPS 12

Figure 2.9: Heatflux as a function of temperature increase in the channel.

heating system used consisted of eight cartridge heaters embedded in the copper block providing a maximum power input of 200W each, controlled digitally by a regulated dc power source and its arrangement can be seen in Figure 2.10a and 2.10b.

(a) Heater arrangement (b) Heating section

Figure 2.10: Liu and Garimella experiment arrangement

Flow rates in different ranges were measured with two turbine flow meters installed in parallel. Several Copper-Constantan thermocouples (also called T-type) from 36-gauge were inserted just under the micro-channels, few centimetres above the base. Extrapolating the readings from these thermocouples the wall temperature could be known. Thermocouples were also used for the inlet and outlet temperature and more thermocouples were inserted in the copper block to measure the average heat flux. The heat flux was regulated and compared with the wall temperature for several

(21)

Figure 2.11: Heatflux as a function of temperature

values (see Figure 2.11, reaching a maximum heat flux of 128.8W/cm 2.

The article explains how the heat losses and measurement uncertainties were estimated. A heat loss of 2 up to 12 % depending on the heat flux and the flow rate. Measurement errors were

estimated for the different thermocouples to 0.3 oC. For the flow meter there was an uncertainty

of 2.4 % and heat flux measurement error was in the range of 5-9.4%.

Wu & Cheng (2007) [6]

30mm long micro-channels with trapezoidal cross section with a hydraulic diameter of 0.186mm

was heated by a silicon wafer. The inlet temperature was set to 35o_.

Figure 2.12: Wu & Cheng’s experimental set up heated by silicon wafer.

Fischer et. al (2012) [7]

Another article that may not directly have to do with the purpose of this project has been per-formed by Fischer. The issue of interest is the experimental set up and the solution provided for the heating of the water. In the experiment many features of the before mentioned articles have been used to achieve the heating of the fluid. Here heater cartridges, with a maximum power of 100 W each, were inserted in a copper block which had a surface in direct contact with the water.

(22)

2.5. CREATING THE BOILING 14 Table 2.3: A summarised version on the relevant parameters of the revised experiments

X-Section Length

(mm)

Heating material

Heating Heat flux

(W/cm2)

Tin (oC) v (m/s)

Peng & Wang 0.6 x 0.7 60 Stainless

steel El. Trans-former w/ SCR volt-age regula-tor 6E9 40 - 70 1.5 - 4

Qu & Mudawar 0.23 x 0.71 50 Copper 12

car-tridges

30 - 60 0.14

-0.4

Liu & Garimella 0.28 x 0.64

0.40 x 1.10

25.4 Copper Cartridges

1600 W

129 66 - 95

-Wu & Cheng 0.20 x 0.15 30 Silicon Silicon

wafer

3.54 35

-The copper block with the inserted heating elements were thermally insulated with Teflon. This differs with the other experiments that used some ceramic insulation. The aim of the insulation is to reduce the radial thermal losses through the surfaces that are not interesting to heat up. The heat flux was calculated extrapolating linearly with the readings of two thermocouples located 15.5 and 30.5mm from the heater surface. This is explained later on in the section about measurement acquisition.

For these experiments the authors almost always decided plot the heat flux delivered to the bulk against the wall temperature or the wall super heat. This is one of the results that is expected to be achieved at the end of the construction of the experimental set up.

2.5 Creating the boiling

From the previous section on experiments performed for the investigation of the two phase flow in small channels, some interesting ideas are to be investigated. For instance the type of heating element is one of the most important parts of the project. In order to achieve the sufficient heat to boil the flowing water in the channel, the right type of heater with the sufficient heat flux is to be installed. There are several alternatives for the heating of the water flow as investigated in the literature review of other experiments. To understand thoroughly the practical the usage of these alternatives, different heater types are tested and their maximum heat flux calculated by simply dividing the effect over the heating area:

q00= Q

A (2.15)

Cartridge heaters

A first testing was done heating up static water up to boiling point, using an aluminium block with two embedded cartridge heaters. Each cartridge can give a maximum effect of 400W . The

(23)

15 2.5. CREATING THE BOILING effect from the heating element was controlled by a transformer and measured with the help of a voltmeter. The heating element of dimensions 81 x 48 x 14mm was submerged in a water container containing approximately 500ml of water. The ends where the cartridges were inserted were sealed with epoxy glue. The temperature was measured with a T-type thermocouple in direct contact with the wall of the heating element, in this way the temperature of the water directly after the wall was measured. By knowing the resistance of each cartridge, the current was calculated with Ohm’s law and the effect from the heating element was calculated by:

Q = U · I (2.16)

here U is the potential and I is the current. By applying maximum effect (for a potential almost equal to 230V ) the boiling point was reached after approximately 20 seconds with a starting

temperature of 65o_{C. This gave a heat flux from the element of around 2W/cm}2_{. This heat flux}

could be increased if the heating area is reduced in a way that only one side of the element is in direct contact with the water insulating the other sides with for instance ceramic. Assuming the contact area with water is reduced to only one surface (by insulation) the heat flux could reach

4.9W/cm2 _{or even higher values in the order of 10W/cm}2 _{as in other similar experiments. Another}

improvement is that the aluminium could be replaced by copper, since copper has higher thermal conductivity than aluminium (401W/mK and 250W/mK respectively).

From the obtained results of this test, more accurate assumptions can be made to estimate the distance needed for bubbling of the water flow at different inlet temperatures. Assuming a constant

surface heat flux coming from the heater, i.e. q00constant, using the fluid temperature change ∆T =

Tpoint− Tin as a variable in Newton’s law of cooling, following relation can be derived [15]:

dq = ˙m · cp· dT Z x 0 dq = q00· (P · x) ˙ m · cp· (Toutbulk − T bulk in ) = q 00_{· A · x} ⇒ x = m · c˙ p· (T bulk out − Tinbulk) q00_{· P} (2.17)

Plotting shown in Figure 2.13a gives the inlet temperature needed for different values of the

distance from the starting point of the heating. A reasonable inlet temperature is around 60o_{C or}

higher to keep the heating surface as short as possible. At 60oC the distance needed is 0.17m, and

with a proper warming bath the inlet temperature could easily be increased to 80oC (taking into

consideration the possible heat losses in the loop) which needs less than 0.10m. Another important parameter is the power needed to heat up the flow to boiling temperature. Assuming heat transfer energy for static water valid, then the power will be equal to the flow times the heat capacity times the temperature difference which will give the results shown in Figure 2.13b. For instance

for an inlet temperature around 80oC the power that the system needs is around 0.8 - 1kW . This

(24)

2.6. BODY STRUCTURE MATERIAL 16 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 20 30 40 50 60 70 80 90 100

Inlet temperature (in degrees)

Distance from inlet (m)

(a) 20 30 40 50 60 70 80 90 100 0 500 1000 1500 2000 2500 3000 3500

Inlet temperature (degrees)

Power needed (W)

(b)

Figure 2.13: Required distance to heat water to saturation temperature for different inlet temperatures and power

Foil heaters

A second test for the heater was done using foil heaters manufactured by CALESCO [13] with dimensions 100 x 200mm. This kind of heating elements are of significantly less power. The maximum potential is 24V and every foil has a resistance of 1.8Ω, which means a current of around 13A, reaching a maximum effect of 320W for every foil. The heat flux reached is then

1.07W/cm2_{. To improve the efficiency of the heating up the foils could be placed in direct contact}

with water. After comparing the distance from the beginning of the heating element for both experiments in Figure 2.14, it is easily deduced that the cartridge heater is more effective and the needed distance for reaching the bubbly regime is less. For instance for inlet temperatures of 60

-80o_{C the distance needed is between 0.5 and 8cm, in contrast with the values for the foil heater}

that go above 0.5m!

2.6 Body structure material

Another aspect that is important to investigate for the building of the channel is the material of the structure on which the heaters and the transparent material for visualization are going to be installed. In order to choose the right material there are several properties that have to be taken into account. It is important to check for instance the thermal conductivities, heat deflection temperatures (or melting point for metals) and their handleability during its construction. Some few materials have been chosen, by their availability on the local markets, to be compared. These materials are: Aluminium, copper, stainless steel, fibreglass, polycarbonate and plexiglass (acrylic glass).

For this experimental set-up the temperatures are expected to reach a maximum of 120-130oC,

(which is a common limit for power electronic devices) therefore it is important that the heat deflection temperature (or melting point) of the structure to be much higher. For aluminium and stainless steel the melting point is way too high to become a concern for the experiment

(25)

17 2.6. BODY STRUCTURE MATERIAL 0 0.5 1 1.5 2 20 30 40 50 60 70 80 90 100

Inlet temperature (in degrees)

Distance from inlet (m)

Cartridge Foil

Figure 2.14: Comparation of the inlet temperatures as a function of the distance for both heating element types

(above 600oC). On the other hand the fibreglass and plexiglass present very low heat deflection

temperatures around 70 degrees Celsius, which will cause problems when in contact with the heated surfaces. As for the polycarbonate it has a convenient 140 degrees Celsius heat deflection temperature, which is almost above the limit of the expected temperatures.

At the same time the heating elements are to be installed into structure which has to have a high thermal conductivity in order to heat the flow. Among the above mentioned materials copper and aluminium are of very high thermal conductivities which are 406 and 205W/mK, respectively ,at room temperature. This value is considered extremely higher compared to thermal conductivities of stainless steel (16W/mK), plexiglass (0.2W/mK), polycarbonate (0.19W/mK) and fibreglass (0.04W/mK)[14]. Therefore the choice should be between copper and aluminium. Since aluminium is of more common usage in power electronic devices it will therefore the material to be used in contact with the flow and where the heating elements will be inserted.

As for the window for the channel analysis, a sufficient transparency of the material have to

be ensured. Since it is not interesting to heat this part in particular, a material with lower

thermal conductivity is needed. There will be a contact area between the aluminium and the window material, and since the temperature of the aluminium is going to reach relatively high temperatures, polycarbonate is the only choice left.

Table 2.4: Material specification

Material Thermal conductivity [W/mK] HDT [oC]

Aluminium 205 660

Fibreglass 0.04 70 - 100

Plexiglas 0.2 76 - 79

Polycarbonate 0.19 140

(26)

(27)

Chapter 3 Method

3.1 Experimental set-up

The experimental set-up is divided into two main sections: the construction of the narrow channel and the rest of the water loop. Through the following subsections a more elaborated explanation can be found on the different components to be utilized in the experiment. Besides the differ-ent compondiffer-ents, also a discussion regarding the measuremdiffer-ent techniques and tools utilized for measurement of temperature, void fraction, etcetera. can also be found.

3.1.1 Channel model specifications

The general idea of the experiment is to analyse the heat flux of a heated channel. This channel is a narrow channel of about 3 - 5 millimetres deep and 100 millimetres wide. Through this cross section deionised water will flow to remove the heat coming from the walls. As it is of interest for this project the flow regime being tested is the sub-cooled region, with heat flux varying between

1 - 20W/cm2_{. The length of the channel is sufficient to attain a hydraulically fully developed flow}

in order to have a constant velocity profile and the length needed for the water to reach saturated temperature of the the liquid is also considered.

In order to attain a hydraulically fully developed flow the water has to have a constant velocity

profile, and this happens at the hydrodynamic entrance length (Le). This Le can be estimated

theoretically using the correlation in equation 2.4 corresponding to a laminar flow. There are some assumptions that may not be accurate for this project’s channel configuration, like the assumption of a cylindrical channel, but is considered sufficient for an estimation. For the given dimensions of the hydraulic diameter (see equation 2.1) the entrance length is calculated to around 93mm. This means that the channel should have at least 100mm after the water inlet to stabilise the flow and attain an almost constant velocity profile.

Since the assumptions for this estimation are not the optimal ones, a closer investigation and comparison based on the literature is performed in section 2.2. The channel configuration used in this project is a very small and wide channel that can be approximated to a channel with two

parallel plates. This corresponds to α∗ = 0 in Table 2.2 which gives a sufficient Le of 20.5 mm.

(28)

3.1. EXPERIMENTAL SET-UP 20 Having this lengths in mind the flow would need between 20 to 100mm for stabilizing before heat flux is introduced to the testing section.

3.1.2 Heaters

Another length regarded for the construction of the channel is the length it takes for flowing water to reach the sub-cooled boiling region. For this purpose the heat flux or even the power coming from the heating elements will affect the dimensions of this distance. As discussed earlier in this report, the cartridge heaters are more effective to deliver heat to the system in shorter time and distance. As a rough estimation of this length it was assumed that the water’s saturation temperature at normal pressure was sufficient as an upper limit to reach for the water temperature increase. Graphs on these estimations are shown in Figure 2.13 where the inlet temperature of the water is plotted against the distance to reach boiling point, and the power needed is plotted against the inlet temperature. As it is shown in Figure 2.13b the power needed in the system is in the order of 0.5 - 3.5 kW which makes around 5 to 7 cartridges sufficient if the water is preheated to reach an adequate inlet temperature.

The cartridges chosen are able to give an effect of maximum 400W each, which corresponds to

a maximum radial heat flux of 24W/cm2_{. If the channel is to be heated from one side then 5}

-7 cartridges will give 2-2.8kW which is sufficient for inlet temperatures over 40oC. According to

the graphs a distance between 100 and 250 mm is enough for the heating part of the channel. To make the experimental set up more practical the distance should be as short as possible which is achieved by increasing the effect of the cartridges through potential regulation.

Due to the relatively high effect of all the cartridges together (around 2.8kW ) the current needed reaches 12A according to equation 2.16 (relatively high current), and therefore they are connected to a 20A auto-transformer (see Figure 3.1). This auto-transformer is then connected to a three phase five cable connection (AC 230V and 32A output) utilizing only one phase and neutral. In Figure 3.2 the electrical set up to which the cartridges are connected is shown. The ”load” represents all the connected cartridges. A Fluke 179 multimeter is used to ensure a known amount of potential being delivered from the auto-transformer when regulating. As for a model with both sides delivering heat to the channel, the amount of cartridges are doubled and the effect needed will be half of the one side heating model, if the heat is well distributed through the aluminium.

When running the experiment it is important to know the effect being delivered to the channel when controlling the input voltage from the auto-transformer. The cartridges are installed in parallel and each cartridge has a resistance of around 132-135Ω. By increasing the potential over the total resistance (around 20Ω for seven cartridges) both the current and the power are increased as shown in Figure 3.3. As discussed in the section about creating the boiling in Figure 2.13, the power needed for a bulk with a volume flow rate of around one litre per minute and inlet temperature

of 60oC to heat the water to saturation temperature is almost 1500W . According to the relation

(29)

21 3.1. EXPERIMENTAL SET-UP

Figure 3.1: Auto-transformer

Figure 3.2: Schematic layout of the cartridges electrical supply

3.1.3 Channel model

Two different models were designed for the small narrow channel. The fist model was heated on one side having a transparent wall on the other for visualization purposes. Transparency through the long narrow sides of 3mm for both models was a requirement, to easily check the bubble formation from a side view. The main differences between the models are the total heat transfer to the bulk and the visualization. To get a more practical design both models are manufactured in a way that makes it easy to replace every part of the set up. As it is shown in Figure 3.4 the aluminium block that has the inlet and outlet apertures to the channel is exactly the same for both models. All drawings are performed with the modelling software SolidWorks. The holes where the heating cartridges are installed are shown in more detail in detail F in Figure 3.4. All other dimensions are as well specified.

The transparent material used for visualizing for the narrow channel is polycarbonate. In order to have a very narrow channel of 3mm depth, a frame of this material is constructed to be placed between the aluminium base plate and the frontal window (in the case of the one side heated model), or between the aluminium base plate and the other aluminium block (in the case of the

(30)

3.1. EXPERIMENTAL SET-UP 22 0 50 100 150 200 250 0 2 4 6 8 10 12

Current throu the system [A]

Potential [V] 0 50 100 150 200 250 0 500 1000 1500 2000 2500 3000

Effect produced by cartridges [W]

Potential [V]

Figure 3.3: Current developed and produced power

two side heated model). The function of this 3mm thick polycarbonate is to mark the depth of the channel. Figures 3.5a and 3.5b are the drawings in SolidWorks of both models, the one side heated and the two side heated.

Attached to the base plate, two funnels are designed and installed at the inlet and outlet aper-tures. To seal and avoid leakages, these funnels are screwed with Loctite 5922 as gasket between the aluminium block and the funnels. The design and dimensions of the funnels is shown in Figure 3.6.

3.1.4 Optimizing the heating

The inlet temperature is fixed to an optimal value of 60o_{C, in that way the sub-cooled bubbly}

regime can be created in the channel much easier and it is not too high to let the cartridges heat up the water (higher heat flux). The water has to be then preheated in a heating water bath that warms up the water up to the requested temperature and delivers more heating to the liquid if the water temperature in the tank sinks. A six-litres heating bath is included to the water loop. The

purchased bath is a N¨uve NB20.

Another way in which it is expected to optimize the heating area to the upper part of the channel closer to the bulk is to install an insulation block on the back side of the base plate (i.e. the aluminium block). On this insulation block paths are drilled to pass the thermocouples intended to measure the wall temperature at different heights of the channel close to the cartridges. This insulation block is shown in Figure 3.8. On Figure 3.8a the path for the thermocouples is clearly seen for the passage of the thermocouples installed on the back side of the aluminium block (see Figure 3.8b) for wall temperature measurement.

The heat produced by the cartridges is spread in the aluminium block heating all the surfaces of the channel. In order to concentrate the heating to come out only from the wall in contact with the water the whole channel structure is covered by an insulating material. This material is of extremely low thermal conductivity and it must stand the temperatures employed in the experiment. Among several materials available the one chosen is the UPM S2, which has a thermal

(31)

Figure 3.4: Drawings of the base plate with all dimensions (in mm)

conductivity of 0.35W/mK and a maximum temperature limit before deformation of 155o_{C. The}

whole structure is installed into a block of this material. In Figure 3.9 the manufacturing of the insulation is performed. In this way the heat losses through radiation are drastically reduced. Holes were drilled in this insulation block to allow the free passage of the cables of cartridges and thermocouples. Another hole is drilled to aloud air suction (see Figure 3.9b) to remove unexpected and undesired bubbles from the upper plenum of the channel (the space before the outlet).

3.1.5 Attaining the right flow rate

Since the requested flow velocity is decided to be few centimetres per second, the circulation has to be adjusted by a pump with certain factors, like a slow flow rate, a high temperature limit and a stable constant flow. There are several approaches how to calculate these factors. As for the calculation of the flow rate one could investigate the mass flux which is calculated by the fluid density ρ at a certain temperature multiplied by the flow velocity v. The mass flow rate is this mass flux multiplied by the cross sectional area A of the channel (see equation 3.2).

˙ m = lim ∆t→0 ∆m ∆t = dm dt (3.1) ˙ m = ρ · ˙V = ρ · v · A (3.2)

A better way to determine the flow requirements of the pump, for further comparison to the differ-ent products on the market, is by knowing the volume flow rate (the amount of litres per second or

(32)

3.1. EXPERIMENTAL SET-UP 24

(a) One side heated (b) Two side heated

Figure 3.5: Drawings of the channel models. Only the one side heated model is con-structed for this master thesis project.

slow velocity and the cross sectional area would be at least around 6ml/s, which corresponds to 360ml/min. Another factor to also have in mind for the choice of pump is temperature limitations of the pump since preheated water is going to pass through the pump into the channel (water

temperature over 40o_{C and under 80}o_C).

Among the available pumps that satisfies both the volume flow rate and the temperature exposure is an I-Drive Micropump (see Figure 3.10). This micropump is able to deliver sufficiently slow flow velocities and keeps a steady pace at relatively high temperatures. According to the specifications of this pump, and since there will be no significant effect from any pressure contribution other than the normal ambient pressure (which is around 1bar), it will work properly. From the technical data for the Micropump shown in Figure 3.11 it is understood that for this specific Micropump model, in order to achieve the right flow, the revolutions per minuter for the I-Drive needed are between 500-1150RP M . This will give a volume flow rate between 600-1500ml/min, which is sufficiently low and suitable for the experiment. [23]

The pump requires a main input power of 24V DC with a maximum current of 2.9A and as for the control signal in is a 0 to 5V DC input. Depending on this control input the flow velocity will vary, the more control voltage is introduced to the system the faster the flow is. The actual performance of the pump may vary from the one shown in the pumps technical data sheet and therefore it was tested in order to have an idea of its actual functionality. For different input voltages the time it takes to fill a 400ml mug was measured and the values are shown in Table 3.1.

From the test performed on the actual volume flow rate, it is clear that the input voltage should be around 1.0V to attain the expected velocity. With a fixed flow rate, it is easier to analyse the creation of bubbles depending on the heat flux, i.e. the effect being delivered to the system from the heating elements.

(33)

Figure 3.6: Funnel model in SolidWorks. Dimensions in mm.

Table 3.1: Volume flow rates and flow velocities for different control voltages

U[V] V [l/min]˙ v [cm/s] Re 0.5 0.33 1.11 331.2 1.0 1.09 3.64 1083.9 1.5 1.71 5.71 1703.3 2.0 2.67 8.89 2649.6 2.5 3.43 11.43 3406.6 3.0 4.0 13.33 3974.4 4.5 6.0 20.0 5961.6 5.0 6.86 22.86 6813.3

3.1.6 Experimental scheme

To summarize the experimental set-up section, the complete loop of the experiment consists of a heating bath (water tank), a micropump, the testing channel and a heat exchanger (see Figure 3.12). The heating bath is used to provide the desired inlet water temperature. The micropump regulates the flow rate of the water and with it both laminar and turbulent flow can be achieved. The narrow channel will be heated up by its cartridges increasing the temperature of the water provoking the formation of bubbles. After passing through the channel, the water temperature at the outlet will be increased several degrees compared to the inlet temperature, it is therefore necessary to use a heat exchanger to reduce the temperature down to a temperature close to the one in the water tank. This is important because otherwise the water temperature of the tank will be increased not allowing the desired inlet temperature to be stable.

(34)

3.2. MEASUREMENT ACQUISITION 26

Figure 3.7: Heating bath N¨uve NB20

(a) Path in insulation block (b) Thermpocouple installation

Figure 3.8: Insulation block located on the back side of the base plate to focus the heat to the bulk

3.2 Measurement acquisition

3.2.1 Wall super heat, ∆T

sup

The heat transfer coefficient increases significantly when the boiling process starts. This boiling heat transfer allows huge amount of heat flux to be taken from heated surfaces to the bulk through forced convection. Due to this high heat transfer coefficient this phenomena becomes interesting for its many practical applications involving heat development in electronic devices and other areas where cooling is needed, like in the nuclear industry.

Generally, to investigate the boiling heat transfer from a heated wall to a flow, the heat flux values are plotted against the so called wall superheat. The definition of the wall superheat is the difference between the wall temperature and the fluid’s (or bulk) saturation temperature. This relation has already been used in equation 2.10 in the section about heat transfer by convection of this report. It is called superheat because during the sub cooled regime the wall temperature is

(35)

27 3.2. MEASUREMENT ACQUISITION

(a) UPM S2 being manufactured (b) Hole for air suction

Figure 3.9: The whole channel structure installed into a block of UPM S2

Figure 3.10: I-Drive MICROPUMP GJ-N27

higher than the liquid bulk temperature.

∆Tsup= Twall− Tsat (3.3)

The saturation temperatures of liquids at different conditions can be found in tables, not at all difficult to find. The more problematic issue will be to get a more accurate measurement for the wall temperature since a thermocouple installed in the heated surface in touch with the bulk will affect the flow. To avoid installing thermocouples affecting the bulk, other solutions are found in for instance the experiment of Fischer [7], mentioned in an earlier section, where the wall temperature is measured by extrapolating linearly from two different thermocouples installed at two different distances from the surface, as shown in Table 3.2, since the heat conduction from the heating element to the heated block was proven to be one dimensional. Following Fischer’s description the calculation of the wall temperature becomes:

Twall = T1−

s1

s2− s1

(36)

Figure 3.11: Volume flow rates for the different revolutions as a function of the differ-ential pressure. [23]

Table 3.2: Fischer’s experiment thermocouple position

Thermcouple No. Distance from heater surface s [mm]

1 15.5

2 30.5

Inspired by the experiment of Fischer et. al. thermocouples were installed inside the wall instead of on the wall surface affecting the bulk flow. Holes were drilled close to the cartridges from the back of the wall until around 1mm from the wall surface which is in contact with the water. In this holes thermocouples were installed, as seen in Figure 3.8b. Using a linear extrapolation technique like the one used in the previous mentioned experiment will give an accurate wall temperature just below the surface of the channel. This was considered unnecessary for the purpose of this master thesis project, since installing thermocouples so close to the surface will give a sufficiently high accuracy only depending on the quality and the type of the thermocouple. The thermocouple type used in this project is a T-type thermocouple, also called copper versus constantan thermocouple. The reason why this type was chosen is because of the temperature range in which it can be used

(-200o_{C to +350}o_{C) and its quite good accuracy of 0.5}o_{C (in this temperature range), since our}

system is expected to reach wall temperatures of not more than 120oC.

A total of five thermocouples were installed in the channel: one at the inlet (tc1), three at the heated section (tc2, tc3 and tc4) and at the outlet (tc5). With tc1 and tc5 the temperature difference between inlet and outlet can be measured, and with tc2, tc3 and tc4 the wall temperatures. In Figure 3.13 the positions in the channel are specified and marked with red color.

(37)

29 3.2. MEASUREMENT ACQUISITION

Figure 3.12: The water loop

(38)

3.2.2 High speed visualization

For recording the bubble formation in form of images a high speed camera is used. A fast frame capturing will help do discover sudden changes in the flow when submitted to heat. For instance, the formation of bubbles at the ONB positions could take a fraction of a second as well as the collapsing of the bubble when condensed in the flow. The flow phases are then easily recognized. Therefore a camera with sufficient amount of frames per second with a relatively good image definition is used. Among the available cameras there was the MotionPro X4 shown in Figure 3.14, with the corresponding software Motion Studio. This camera can take five thousand frames per second at full resolution.[17] For the purposes of this work around two to three thousand frames per second over a time lapse of some few seconds are sufficient to analyse the regime changes in the flow.

Figure 3.14: MotionPro X4 high speed camera

3.2.3 Methods of void fraction measurement

An important parameter and indicator of the flow mechanisms in the channel that was considered for measurement in this project is the void fraction. It is of special importance since what is being controlled in this experiment is the appearance of bubbles in the sub-cooled bubbly flow regime and it is a parameter that determines the characteristics of the two-phase flow. The void fraction is defined depending on the type of measurement that is requested. There is a volumetric void fraction, area ratio, length and even time ratio. However, and despite the different definitions of the void fraction there are several suitable techniques to measure it.

Various void fraction measurement techniques have been discussed in previous years. Many of these techniques have been studied by Hewitt already in 1978 [16], and among them there is the quick closing valves technique, radiation attenuation, x-ray, conductivity and impedance method, among others. From these techniques the most reliable is the quick closing valve technique which is usually used for calibration of other void fraction meters. This technique is perhaps the simplest way to identify the volumetric ratio between gas and liquid and consist mainly of two valves that close simultaneously trapping a portion of the flow. The water is then drained to know exactly the amount of liquid and gas in the decided volume in between the valves when enclosed.[19]

The quick closing valves method, despite the fact of its reliability, it only works good with bigger channels, since a small leak in a small channel will mean a big deviation from the real void value. Another drawback of this technique is that when the valves close this has to happen simultaneously, otherwise it may lead to the previous discussed measurement deviation. Even another aspect that is important to think about is that when closing the valves the whole flow in the system has to

(39)

31 3.2. MEASUREMENT ACQUISITION stop or somehow divert the flow from the channel to another, otherwise the liquid will be pumped to the system but the flow loop is closed. Therefore the quick closing valves is not a appropriate solution for a narrow rectangular channel.

3.2.4 Impedance method for void fraction measurement

The method that was planned to be used in this project is the impedance method. The impedance method uses a technique that is based on the measurement of conductivity of a medium to get a global value of the void fraction. Liquid and gas phases of water have different conductivities and relative permittivities and therefore the vapour built in the flow will have lower conductivity, ergo higher resistivity according to following equations:

σ = 1

ρ (3.5)

ρ = R · A

` (3.6)

From the previous equations σ denotes the conductivity, which is inverse proportional to the resistivity ρ. The ρ is the fraction between the product of the resistance R and the cross sectional area A, and the length ` as shown in Figure 3.15. The resistance is different for different materials and therefore the conductivity of water is different from that of vapour. In this way, and with a proper calibration, the void fraction can be measured. [19]

A

`

Figure 3.15: Resistive material and the dimension for resistivity calculation

Although the theory is exactly the same there are different techniques for the probes measuring the conductivity in water channels. The difference are often related to the physical limitations of the experimental set up, or even the precision wanted in the measurement. The probe system could be based on coaxial, parallel flat plates, wire grid and wall flush mounted circular arcs electrodes. Among the experiments explained in articles about void fraction measurement there are some of special interest for the aim of this project and the feasibility of using these techniques in the experiment will be analysed.

In the experiment of Ito et al.[20] a micro wire-mesh sensor was constructed for void measurement in a rectangular narrow channel. Unlike the usual wire grid the micro wire-mesh reduces the intrusive effect in the flow which is important for mini or micro scale channels. For each crossing point of transmitter and receiver in the grid the local electrical conductivity is measured. When the results of many points are put together the void fraction profile of the measured section can be obtained. The article also explains that in order to know the void fraction at one crossing point

(40)

3.2. MEASUREMENT ACQUISITION 32 the conductivity of the liquid (L) has to be known for calculation, in this way the conductivity of the gas (G) is not needed. If i is the sample number and j and k represent the indices in the grid, then the following relation will give a local void fraction value:

εij,k = σL− σij,k σL− σG = 1 − σij,k σL (3.7)

Other suitable probe system for this project using the wall flush mounted circular ring electrodes technique, which is used in the work of Andreussi et al. [22]. The rings are of stainless steel and are located surrounding the channel in a way that the flow passes through the rings. The conductivity is then measured between these two rings. The obtained measurement will give an idea of the global void fraction and not the local which may be of importance when having a relatively wide rectangular channel, to better see the development of bubbles. The material of the rings may differ from experiment to experiment. For instance Portillo et al. [18] use a ring type sensor made of copper instead of stainless steel.

As for this experiment, and since the geometry of the channel is not suitable for the types of electrodes presented in other articles, the measuring could be done with a system created in the lab, with the expertise of laboratory engineers at ABB Corporate Research. Close to the outlet stainless steel wires could be installed and inserted with a Teflon insulation to avoid the direct contact with the aluminium, as shown in Figure 3.16. The aluminium would be used as ground. Then an AC voltage source with a frequency of at least 50Hz would be installed to avoid any possible galvanic effect. In order to capture a signal difference, what is measured is the potential fall over a resistance of 1kΩ.

This void fraction measurement technique is not installed in the currently constructed experimental set up. Although the very well elaborated explanation, it could be installed in future adjustments of the experiment.

Figure 3.16: Scheme on the void fraction measurement method to be tested in the channel

(41)

33 3.3. RUNNING THE EXPERIMENT!

3.3 Running the experiment!

There are several steps for running the experiment. Following is an example procedure for running the experiment. It is worth mentioning that the input parameters can be regulated to satisfy the purpose of the experimentation.

(a) Turn on the water bath and set the temperature to 60oC and wait until the water heats up

to that temperature.

(b) Start the pump and set it at the decided velocity of 1litre/min (which corresponds to an input control voltage of 1V ).

(c) Take away the excess of air in the channel with the valve above it. (d) Turn on the acquisition systems and start the corresponding softwares.

(e) Turn on the camera. (f) Start measuring.

(g) Connect the heating elements and increase the input voltage to the cartridges until reaching 150-175V .

After some minute first bubbles will appear on the wall surface at the nucleation sites. Later on the number of bubbles will increase and some of them will detach flowing along the bulk. Without increasing the power of the heating elements the sub-cooled boiling regime will stabilize. The parameters that are easier to change and analyse their effect on the flow are both the flow rate and the heat flux. One important result that is often present in articles about small rectangular channels in general is the plot of the heat flux as a function of the wall temperature or wall super heat. For different wall temperatures the heat fluxes are calculated according to theory previously explained.

A more complete manual of usage of the experiment is added at the end of the report as an appendix.

(42)

(43)

Chapter 4 Measurements and results

The temperature measurement

The temperature measurements obtained for the different power values supplied by the heating elements to the channel are summarized in Table 4.1. For these measured values the corresponding heat flux for every input voltage to the cartridges is calculated using equations 2.15 and 2.16,

through the wall area which is 100 times 400mm2, assuming negligible heat losses through radiation

and conduction to other materials in the channel. The plotting of the heat flux as a function of the wall temperature is shown in Figure 4.1.

The inlet temperature to the channel (Tin named tc1 in Table 4.1) is relatively fixed to 60oC.

The temperature difference between inlet and outlet is shown in Figure 4.2. In order to make the heat flux measurements for a fixed inlet temperature, the heat exchanger is used and regulated

manually. As seen from Figure 4.2 the outlet can reach almost 30oC higher temperature than the

inlet and the water returning to the tank will therefore increase the temperature of the water in the loop. With the help of a thermocouple installed directly before the water tank and after the heat exchanger the temperature is controlled to be reduced to the desired inlet temperature by the heat exchanger. In that way the water don’t need to be heated by the tank any more, it will only need to be kept constant by it. After adjusting this temperature, some few minutes were required between the measuring points to have a stabilized temperature distribution in the system.

This procedure was performed for different inlet temperatures. In Figure 4.3 it can be seen that the higher the inlet temperature is, the bigger the slope becomes when increasing the wall temperature. The area represented by the results corresponds to heat fluxes between point A and C in the Nukiyama boiling curve shown in the theoretical background of this report (see Figure 2.1). Another interesting result is the temperature profile in the system when the channel is subjected

to a constant heating of around 1.2kW corresponding to a heat flux of 2.9W/cm2_{. The idea is}

to see how the heat in the walls and the inlet/outlet temperatures increase and later on stabilize. These measured temperatures are plotted against time in Figure 4.4.

(44)

36 Table 4.1: Temperature reading for different input voltages

Voltage[V] tc1 [o_C] _{tc2 [}o_C] _{tc3 [}o_C] _{tc4 [}o_C] _{tc5 [}o_C] 99 59.87 77.20 81.21 82.54 65.16 100.6 60.18 77.71 81.42 82.84 64.26 110.1 60.09 80.44 84.72 86.57 64.68 120 59.96 83.55 88.73 91.01 67.04 130 60.17 86.71 92.05 94.34 69.61 140 60.24 90.21 96.03 98.75 71.70 149.5 59.92 96.77 101.32 104.60 73.17 160.9 60.39 98.11 104.19 109.30 76.36 170.9 60.71 102.41 110.68 111.09 77.91 180.2 60.59 106.97 112.79 112.21 79.75 191 60.54 109.97 114.56 113.61 82.89 201.1 60.69 113.65 115.68 114.86 85.26 210.1 60.96 115.56 116.23 115.78 87.69 The visualization

The visualization of the sub-cooled bubbly flow is performed for 3000f rames/s which makes it possible to measure the bubbles’ time from production at the nucleation points to collapse or detachment from the wall. Among many features that can be analysed with the high speed visualization camera is the time of bubble formation, velocity profile of the bubbles, size, nucleation sites, etcetera. Some samples were filmed at different heights of the channel, with a heat flux able to attain fully developed boiling. The different flow regimes in the channel could clearly be recognized as seen in Figure 4.5 along the heated wall.

The images shown in Figure 4.5 where marked at the top with frame number, time, the amount of frames per second being taken and the image resolution. In Figure 4.5a the various nucleation sites can be recognized. At this position the flow regime is in the sub-cooled nucleate boiling region. In Figure 4.5b the bubbles start to merge forming a slug flow, entering the saturated nucleate boiling region. Figure 4.5c shows a fully developed slug flow where the vapour content is increasing, and there is less water in contact with the wall. As for Figure 4.5d a big bubble is covering most of the channel cross section, hence the flow regime could almost be considered as annular flow.

(45)

37 75 80 85 90 95 100 105 110 115 120 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6

Measured wall temperature, T

w [

o_C]

Heat flux for different wall temperatures [W/cm2]

Thermocouple 2 Thermocouple 3 Thermocouple 4

Figure 4.1: Heat flux as a function of the wall temperature, for the three thermocouples

installed in the wall at fixed inlet temperature (Tin = 60oC).

Power [kW] 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 55 60 65 70 75 80 85

90 Inlet and outlet temperatures [

o_C]

Inlet Temperature Outlet Temperature

(46)

38

Mean wall temperature, Tmean

w [ o_C] 60 70 80 90 100 110 120 Heat flux, q" [W/cm 2] 1 2 3 4 5 6 7 T in=40 o_C T in=60 o_C T_in=80oC T_in=90oC

Figure 4.3: Heat flux against wall temperature for different inlet temperatures

0 50 100 150 200 250 300 350 400 450 500 50 60 70 80 90 100 110 Temperature profile Time [s] Thermocouple 1 Thermocouple 2 Thermocouple 3 Thermocouple 4 Thermocouple 5

(47)

39

(a) ONB (b) Beginning of slug flow

(c) Slug flow (d) Almost anular flow

(48)

Sub-cooled nucleate boiling flow cooling experiment in a small rectangular channel