Fuel Coolant Interaction of Liquid Tin.

Fuel Coolant Interaction of Liquid Tin

Albin Cassirer & Daniel Gullberg albinca@kth.se, dgul@kth.se

SA104X Degree Project in Engineering Physics, First Level Supervisor: Weimin Ma

Department of Physics School of Engineering Sciences Royal Institute of Technology (KTH)

Stockholm, Sweden, 2014

May 21, 2014

Abstract

In the case of a severe accident in a nuclear power plant the reactor may heat up, melt and mix with fuel material to form a substance called corium. In today's nuclear power plants the primary strategy to cool the corium in the event of a severe accident is to ood the ex-vessel cavity with water. The reactions which occur when the liquid metal comes in contact with the water, known as fuel coolant interaction (FCI), can be violent and in the worst case scenario lead to containment failure. In the MISTEE laboratory at KTH, small scale FCI experiments are con- ducted. This thesis explores how dierent temperatures of liquid tin and water aects the presence of steam explosion. Higher melt superheat and lower water temperature was found to increase the likelihood of steam explosions. Furthermore, a phenomenon was observed, hereby referred to as immediate steam explosion, where the melt exploded immediately upon contact with water. All previous research found states that steam explosion only occurs in the later stages of FCI, thus the results are con- tradictory.

The thesis also includes research on jet breakup in the initial phase

of FCI and how it is aected by melt velocity, diameter and temperature

as well as water temperature. The experiments performed did not yield

data which could be analyzed so no conclusions could be drawn.

Sammanfattning

Om en allvarlig olycka inträar på ett kärnkraftverk kan reaktorn vär- mas upp, smälta och bilda en smälta som innehåller bränsleämnen. I dagens kärnkraftverk är den primära strategin för att kyla reaktormateri- alet vid en eventuell allvarlig olycka att fylla utrymmet utanför reaktorn med vatten. De reaktioner som uppstår när den smälta metallen kommer i kontakt med vattnet (FCI) kan vara mycket våldsamma och i värsta fall leda till skador på skyddsväggarna. Vid MISTEE på KTH forskar man på reaktionerna i liten skala för att få en ökad förståelse för processerna i FCI. Denna avhandling undersöker sambandet mellan temperaturen på smältan och vattnet och förekomsten av steam explosion. Högre smält temperatur och lägre vattentemperatur visade sig öka sannolikheten för steam explosions. Vidare, ett fenomen observerades, som kommer att ref- ereras till som omedelbar steam explosion, där smältan exploderade direkt vid kontakt med vattnet. All tidigare forskning som hittades pekar på att steam explosions endast sker i senare skeden av FCI.

Avhandlingen inkluderar även forskning på uppbrytning av en met-

all stråle i den initiala fasen av FCI och hur den påverkas av strålens

hastighet, diameter och temperatur samt vattnets temperatur. De ut-

förda experimenten resulterade inte i data som kunde analyseras och inga

slutsatser kunde dras.

Contents

1 Introduction 5

1.1 Fundamental physics of steam explosions . . . . 5

1.2 Molten fuel coolant interaction: A nuclear safety concern . . . . 6

1.3 Previous research . . . . 6

1.4 Motivation of research . . . . 7

1.5 Scope of research . . . . 7

1.6 Objective of research . . . . 7

2 Experimental method 8 2.1 Overall description of experimental approach . . . . 8

2.2 MISTEE . . . . 8

2.2.1 Melt generator . . . . 9

2.2.2 Test chamber and visualization system . . . . 9

2.3 Experimental procedure . . . 10

2.4 Image processing . . . 12

3 Results 15 3.1 Jet breakup . . . 15

3.2 Steam explosion map . . . 15

3.2.1 Immediate steam explosion . . . 16

4 Discussion and analysis 18 4.1 Jet breakup . . . 18

4.2 Steam explosion map . . . 19

4.2.1 Immediate steam explosion . . . 20

4.3 Sources of error . . . 20

4.3.1 Mass of charge . . . 20

4.3.2 Temperatures of water and melt . . . 21

4.3.3 Quantities obtained from image processing . . . 21

4.3.4 Other possible sources of errors . . . 21

4.4 Conclusions and suggestions for further research . . . 22

5 References 23 A Failed experiments 24 A.1 The use of better simulant materials . . . 24

A.2 Formation of jets . . . 24

A.3 Failure of various equipment . . . 24

List of Figures

1 Phases of steam explosion. From left to right, pre-mixing, trig-

gering, propagation, expansion. . . . 5

2 Schematics of the experimental setup . . . . 8

3 Schematics of melt generator . . . . 9

4 Schematics of test chamber and visualization system . . . . 9

5 Background removal. From left to right: raw image, gray scale version, background removed . . . 12

6 Steam explosion map including results from 3.1 . . . 16

7 Immediate steam explosion. Each frame is captured 2.5 ms after the previous . . . 16

8 Immediate steam explosion map . . . 17

9 Jet break up. Each frame is captured 20 ms after the previous . 19 List of Tables 1 Results of jet break up experiments . . . 15

2 Results of steam explosion experiments . . . 15

List of Algorithms 1 Psedocode for calculating average backgrounds . . . 13

2 Pseudocode of background removal in gray scale. The process is the same for each RGB channel . . . 14

Nomenclature

ρ

Density of jet ρ

Density of coolant D Diameter of jet

u Velocity of jet

L Jet breakup length

g Gravitational acceleration

T

Temperature of coolant

T

Temperature of jet

M

Mass of jet

1 Introduction

In the introduction a brief description is given on the physical phenomena stud- ied in the thesis and the relevance of this research to the eld of nuclear safety is motivated. Some previous research is listed and nally the scope and objective of the report is claried.

1.1 Fundamental physics of steam explosions

Steam explosions are a violent boiling or ashing of a volatile liquid, e.g water, caused by swift heat transfer between a high temperature liquid, e.g molten material and a volatile liquid. This phenomenon sometimes occur in fuel coolant interactions (FCI), which is the interaction between a superheated liquid and a coolant. FCI is pertinent for industries where hot liquids mixes with cool volatile liquids, e.g, paper, steel and aluminum casting and nuclear energy. It can also be observed in nature for example when lava from a volcano mixes with sea water [2].

In steam explosions a volatile liquid is superheated by a superheated melt, causing an explosive vaporization which fragments the molten material into ne particles which increases the heat transfer area, increasing the pace of steam generation. This is turn produces a shock wave of hydrodynamic loadings to the surroundings [2].

Figure 1: Phases of steam explosion. From left to right, pre-mixing, triggering, propagation, expansion.

Steam explosions can conceptually be described by four phases; pre-mixing,

triggering, propagation and expansion, see gure 1. The pre-mixing phase typ-

ically consists of a gravity driven jet of melt penetrating the surface of a water

pool. The extreme heat of the melt causes the water closest to the melt to im-

mediately vaporize and form a layer of steam surrounding the melt. The layer

of steam isolates the materials, reducing the heat transfer. Dierences in veloci-

ties, densities and the development of steam causes the jet to fragment into ner

particles. In the second phase, a trigger, either spontaneous or due to external

forces, breaks the vapor lm of the particle, allowing the coolant to once again

come in direct contact with the melt. Because the melt has been thermally

isolated, a large discrepancy in temperature is still present and the amount of

steam generated will be immense. In the next phase, the propagation phase,

the rapid expansion due to steam development causes a shock wave. The wave

destabilizes nearby vapor lms which in turn can trigger other particles, making the reaction propagate throughout the melt. In the last phase, expansion, ther- mal energy is converted into mechanical energy which creates hydrodynamic loadings to the surroundings [2].This will create an impulse, increasing with larger amounts of steam created, which in turn depends on temperature dif- ference and the amount of melt involved. The impulse can get big enough to threaten the security of reactor containment walls, in the case of nuclear power plants [1].

1.2 Molten fuel coolant interaction: A nuclear safety con- cern

In the case of a severe accident in a nuclear power plant, parts of the core can melt and blend with the fuel and form a mix called corium. The corium is extremely hot and if not cooled down to avoid a catastrophe. The most common method to accomplish this is to ood the ex-vessel cavity with water with the intention that the corium will be cooled naturally by the water. Apart from melt coolability problems, FCI can contain steam explosions which poses a threat to the containment integrity [1].

The pre-mixing phase is of great importance to FCIs as it will determine the overall coolability of the melt and therefore also the risk of containment failure. If the jet does not break up into ner particles a solid debris bed can form on the bottom of the water pool which might not be coolable through natural circulation of water. This can result in the corium reacting with the concrete containment and in the worst case scenario lead to containment failure and release of radioactive substances [1]. Studies have been performed on the formation of a debris bed at the bottom of the tank and it has been concluded that the ratio between jet size and water depth is important to achieve a coolable bed [5].

1.3 Previous research

Extensive research has been conducted regarding FCI, both in small and large scale. The dierence between the two is the amount of melt used. When using a small amount the reactions can more easily be studied, while a large mass allows the reactions to reach a steady- or quasi-steady-state which better simulates the case of a severe accident. In this section a brief summary of some of the major

ndings made in previous research will be presented.

As previously mentioned, 1.1, FCI can be divided into four relatively distinct

phases; pre-mixing, triggering, propagation and expansion. For the triggering

and propagation phases some contradictory results have been found. Some

controlled large scale experiments displayed diculties in triggering parts of the

melt, which hinders propagation. Other experiments had no such problems in

achieving propagation through the melt [2]. Because of the risks associated with

steam explosions and the fact that several test suggests that triggering can be

caused by a wide variety of reasons, the probability of triggering is set to 100

% when modeling FCIs [4].

Various studies have been conducted on the pre-mixing phase of FCI, espe- cially how a jet breaks up. One parameter which has been studied is how far the jet travels before breaking up, known as the jet breakup length. At the moment no one model is able to consistently predict the jet breakup length in all cases, so two widely used correlations is presented in equation (1) and equation (2).

One was conceived by Saito et al[5]:

L = 2.1 · u · s

ρ

· D

ρ

· g (1)

Epstein et al uses[5]:

L =

√ 3

2 · D ·  1 + ρ

ρ



· r ρ

ρ

(2)

As can be observed from the formulae, the jet breakup length depends on the ratio of the densities between the jet and the coolant. It also depends on jet diameter and velocity as well as the gravity constant.

1.4 Motivation of research

As previously mentioned FCIs occur in a large variety of natural and industrial environments. A greater understanding of FCIs can help mitigate severe ac- cidents in nuclear power plants, which has motivated lot of research currently conducted. At KTH research is being performed continuously, so the facilities required were already in place. For this thesis, experiments with tin were per- formed, and although tin is not a perfect simulant to corium, it still provides valuable data.

1.5 Scope of research

The thesis focuses on jet breakup and steam explosion for the FCI with liquid tin. The original scope were to perform experiments and with corium simulants and compare the results with the existing theories presented in 1.3. However, due to technical diculties and decient quality in the gathered data, only analysis related to the existence of steam explosion for liquid tin FCI could be performed.

1.6 Objective of research

The goal of this report was not to develop new physical models for FCI and

steam explosion. Rather, qualitative comparisons will be made between existing

theory and the results obtained in the experiments performed and suggestions

for further research will be made.

2 Experimental method

This section contains information about the MISTEE facility used to perform the experiments as well descriptions of the experimental setup and procedures.

2.1 Overall description of experimental approach

The purpose with the experiments was to create an environment where the pre- mixing phase of FCI could be studied. Tin (Sn) was used for the melt, mainly because of its low melting point and low safety requirements.

Since the studied reactions transpires in a very brief time span it is best captured using high-speed cameras and X-ray imaging, since it will yield good data without having the images obscured by melt diusion. The two capturing methods together provide all necessary information to analyze the fragmentation of the jet.

Unfortunately the series of experiments carried out was unable to utilize X- ray radiography due to technical diculties and safety concerns and therefore solely relied on high-speed photography. In addition to the visual imaging sys- tems, K-type thermocouples were used to gather temperature measurements of the melt and water.

2.2 MISTEE

Melt generator

Laser Sensor

Camera

Light

Figure 2: Schematics of the experimental setup

The MISTEE facility (Micro Interactions in Steam Explosion Experiments) is a laboratory at the Royal Institute of Technology constructed with the purpose of studying the fundamental processes of FCI. The facility is limited to small scale experiments and is located inside a 700mm thick concrete containment (4mx4mx4m) which shields the surroundings from the harmful eects of the X-ray.

The MISTEE facility is equipped with the so called SHARP imaging system

(Simultaneous High-speed visual Acquisition of x-ray Radiography and Photog-

raphy), which provides complete visual information. As mentioned in section 2.1

only the high-speed photography was used for this research. The setup consists of two main parts, the melt generator and test chamber. A schematic of the setup is presented in gure 2.

2.2.1 Melt generator

Inductive coil Crucible Pneumatic

release system Plug

Ceramic Housing

K-Type Thermocouple

Figure 3: Schematics of melt generator

The purpose of the melt generator is to heat the material to a desired tem- perature and release it into the water tank in a controlled manner. The melt generator uses induction to induce heat into the crucible. The induction is generated through a water cooled copper coil which wraps around the crucible.

Insulation is stued between the coil, the crucible and the housing. The insula- tion serves two purposes, minimizing the temperature in the coil and minimizing the heat loss in the crucible. The temperature of the melt is measured using a K-type thermocouple. When the desired temperature is reached, the melt is released using the pneumatic trigger system which raises the plug blocking the hole at the bottom of the crucible.

2.2.2 Test chamber and visualization system

Laser Sensor

Camera

Light K-Type

Thermocouple

Figure 4: Schematics of test chamber and visualization system

The melt falls into water lled plexiglass tank when released from the crucible.

Between the tank and the melt generator a laser is aimed towards a sensor, see

gure 2. On its way towards the tank, the melt will break the beam which triggers the high-speed camera to start capturing.

A powerful light source is directed towards the test chamber to satisfy the large light requirements of the camera. In addition to the imaging system, the test chamber is equipped with a K-type thermocouple which can measure the water temperature.

All experiments presented in this thesis used 2000 frames per second and a resolution of 288 × 888 px.

2.3 Experimental procedure

The same setup was used in both types of experiments, but required dier- ent levels of precision. The jet breakup experiments were more sensitive to disturbances, as those processes had to be studied in more detail than steam explosions, which have a binary outcome.

The next page depicts a step by step list describing the process of setting

up and executing a single experiment. The process has to be repeated for each

individual test.

1. Assemble the melt generator

(a) Insert induction coil into housing (b) Insert crucible into coil

(c) Isolate with ber materials

(d) Put K-type thermocouple into crucible 2. Align melt generator

(a) Insert phantom through crucible

(b) Position furnace and coil so the phantom breaks the laser (c) Verify that the crucible is horizontally aligned

3. Set up camera and lighting (a) Turn on camera and lights

(b) Position the light to light up the test chamber as evenly as possible (c) Focus the camera using the phantom and verify that the lighting is

satisfying

(d) Save image of phantom for reference of relation between pixels and distance

4. Set up plug and release system

(a) Put the plug in the crucible and check for leaking using water (b) Turn on air pressure supply to pneumatic system

(c) Verify functionality of the release system using water 5. Turn on the water supply for the melt generator cooling system 6. Turn on the melt generator

7. Prepare the charge of material

8. Add the charge to the crucible when the temperature is above the melting point

9. Heat water to a temperature slightly above the desired test temperature 10. Stabilize the melt temperature

11. Verify that the charge is completely melted and homogeneous 12. Release the melt using the pneumatic release system

13. Download the images from the camera to the PC

2.4 Image processing

The raw images from the experiments are dicult for a human and impossible for a computer to analyze without pre-processing. The research carried out in this thesis does not require much computer analysis but instead relies on manual analysis. However, in order to properly analyze the data it is necessary to rst improve its quality.

The initial step is to remove the background in order to increase the visibility of the moving elements, i.e the melt. The process of removing background is most eectively done in a 1D space, i.e gray scale. The raw data is captured in RGB-format, and in order to avoid data losses all processing is done for each color channel separately as well as for a gray scale version using all three channels in combination.

Figure 5: Background removal. From left to right: raw image, gray scale version, background removed

To remove the background from the images the background rst have to be

approximated. This is achieved by manually selecting a subset of the images in

which the melt has not yet entered the eld of view of the camera. An average

value for each pixel is calculated and the process is repeated for each color

channel and for the gray scale version. Algorithm 1 depicts psuedocode for the

process.

Algorithm 1 Psedocode for calculating average backgrounds gray_bg_sum_ := [ 0 . . . 0 ]

rgb_bg_sum := [ [ 0 . . . 0 ] , [ 0 . . . 0 ] , [ 0 . . . 0 ] ] f o r image in background_images :

// Convert to gray s c a l e

bw_image := convert_to_bw ( image ) // Add value to sum

gray_bg_sum := bw_image + gray_bg_sum // Add value to sum o f each channel

f o r i , channel in enumerate ( image ) :

rgb_bg_sum [ i ] := rgb_bg_sum [ i ] + channel // Divide with number o f images f o r average

gray_avg_bg := gray_bg_sum / length ( background_images ) rgb_avg_bg := rgb_bg_sum / length ( background_images )

The background is removed from each channel by diving each pixel value

by the average background value. Pixels are typically saved as a three item

tuple with one byte per item. The possible values for each channel therefore

span [0, 255]. When divided each pixel value by its corresponding background

value, a value around 1 is obtained. In order to be used with standard tools,

each pixel value have to be scaled with a factor around 255 and truncated to a

byte. Algorithm 2 depicts psuedocode of the background removal when the a

background reference is already obtained and gure 5 displays the steps visually.

Algorithm 2 Pseudocode of background removal in gray scale. The process is the same for each RGB channel

// P revious ly c a l c u l a t e d background values gray_avg_bg

f o r image in experiment_images :

bw_image := convert_to_bw ( image )

// Divide each p i x e l value with avg value bw_image := bw_image / gray_avg_bg

// Scale p i x e l values bw_image := 255*bw_image

// Truncate to i n t e g e r between 0 and 255 bw_image := truncate ( round (bw_image ) ) // Save image to hard d r i v e

save (bw_image)

3 Results

3.1 Jet breakup

The results from the jet breakup experiments are presented in 1. Steam explo- sion is referred to as SE and immediate steam explosion as ISE.

# T

[ °C] T

[

C] M

[g] F all height [mm] J et SE ISE

1 52 275 425 180 Yes No No

2 40 277 493 180 Yes Yes Yes

3 65 287 502 180 Yes No No

4 80 275 500 180 Yes No No

5 53 351 499 180 Yes Yes No

6 48 361 501 180 No Yes No

7 63 381 300 180 No Yes No

8 72 370 500 180 No Yes Yes

9 67 310 498 180 Yes No No

10 70 297 301 330 Yes No No

11 N/A 399 150 330 No No No

12 65 304 300 330 No Yes No

Table 1: Results of jet break up experiments

3.2 Steam explosion map

A total of eight successful experiments were carried out with the sole purpose of developing a map for the relation between melt and water temperature and the presence of steam explosions. The results are presented in table 2. A complete map is displayed in gure 6 using data from table 2 and table 1. Steam explosion is referred to as SE and immediate steam explosion as ISE.

# T

[°C] T

[°C] M

[°C] F allheight[mm] Jet SE ISE

1 36 309 102 180 Yes Yes No

2 32 304 100 180 Yes Yes No

3 53 263 100 180 No No No

4 52 268 101 180 No Yes No

5 51 290 100 180 Yes Yes Yes

6 52 292 100 180 Yes Yes Yes

7 73 262 100 180 Yes No No

8 73 287 101 180 Yes No No

Table 2: Results of steam explosion experiments

0 10 20 30 40 50 60 70 80 90 100 0

20 40 60 80 100 120 140 160 180

Coolant temperature [°C]

Melt superheat [°C]

Total steam explosion map

Jet breakup study, no steam explosion Jet breakup study, steam explosion

Steam explosion map study, no steam explosion Steam explosion map study, steam explosion

Figure 6: Steam explosion map including results from 3.1 3.2.1 Immediate steam explosion

Figure 7: Immediate steam explosion. Each frame is captured 2.5 ms after the previous

Out of the 11 tests which resulted in steam explosion 4 exploded immediately

as the melt hit the surface. In gure 7 the phenomenon is displayed in a series

of frames captured in experiment #2 from table 1. The subset of immediate

steam explosions are presented in gure 8.

0 10 20 30 40 50 60 70 80 90 100 0

20 40 60 80 100 120 140 160 180

Coolant temperature [°C]

Melt superheat [°C]

Immediate and not immediate steam explosion

Jet breakup study, no steam explosion Steam explosion study, no steam explosion Jet breakup study, not immediate steam explosion Jet breakup study, immediate steam explosion

Steam explosion map study, not immediate steam explosion Steam explosion map study, immediate steam explosion

Figure 8: Immediate steam explosion map

4 Discussion and analysis

4.1 Jet breakup

The primary goal with the analysis of the jet breakup experiments was to try to nd and verify the aforementioned relations between the breakup length and various parameters 1.3. However, the data was found to be unsatisfactory for such a comparison. This section will explain why the analysis failed in this particular instance.

According to equation (1), the jet breakup length, diameter and velocity should be extracted from the data. The analysis was unsuccessful in retrieving either of these, and a discussion of why that was the case can be found in this section.

The jet velocity needs to be calculated just before the jet penetrates the water, which was impossible in this case, because in the images nothing can be seen above the water. This can be observed in gure 9 and was most likely caused by insucient lighting. Originally, two lamps were used to light up the test chamber, and were congured to light the tank both below and above the water surface. Partway into the jet break up experiments one of the lamps broke, and from that point and forward only one lamp was used. This lamp was aimed straight into the water, since this is the most important area, and because of this, the area above the water is unobservable. The only other possible implementation is to calculate the velocity during the rst moments after the melt has entered the water. However, this would not give an accurate measurement since a lot of energy and velocity is lost in the moment of impact.

Calculation of the jet diameter faces similar problems as the one for jet velocity. The diameter should preferably be calculated above the water surface as the impact alters its dynamics. When in the water the jet's diameter diers between dierent points, gure 9, and is further obstructed by vapor in some experiments.

The jet breakup length is not a trivial task to calculate. As can be seen in

gure 9 it is unclear exactly where the jet is broken up. A major problem is

that jet breakup occurs in a variety of fashions. Furthermore, a signicant part

of the images were obscured, making the analysis even more dicult. Because

of the aforementioned reasons and the fact that other parts of the analysis also

failed the jet breakup length was never found and no comparisons with theory

could be made.

Figure 9: Jet break up. Each frame is captured 20 ms after the previous

4.2 Steam explosion map

In this section a discussion is held on how the results from the steam explosion experiments can be interpreted. The conclusion drawn is that both the temper- ature of the water and the melt aect the presence of steam explosions. Higher melt superheat and lower water temperature increases the likelihood of steam explosions. In section 4.2.1 the phenomenon of immediate steam explosion will be discussed.

Two distinct regions can be observed, gure 6, one with steam explosion

and one without. Both higher melt superheat and lower water temperature

was found to cause steam explosion. The exact boundary between the two

domains could not be determined with the limited data and its proneness to

errors, section 4.3. Furthermore, steam explosion might also depend on other

properties than temperature.

It should be noted that the melt superheat is more interesting than the absolute temperature in the case of steam explosions. Some trial experiments performed, section ŸA, with materials of temperatures over 1000 degrees but with low superheat did not display steam explosion. Superheat corresponds to how much thermal energy the melt can emit before solidifying. The amount of heat the melt can emit determines the amount of steam that can potentially be generated.

4.2.1 Immediate steam explosion

Earlier theory suggests that steam explosion only occurs after the pre-mixing phase of FCI is nished [3]. As observed in the experiments, steam explosions sometimes occur immediately as the melt hits the water, gure 7. The exact relation between water sub-cooling, melt superheat and immediate steam explo- sions could not be found given the limited amount of data. However, gure 8 suggests that immediate steam explosion is constrained to the region between steam explosions occurring or not.

One might speculate why the phenomenon of immediate steam explosion occurs and how it could potentially be utilized. Since the steam explosions happen immediately at the surface, its propagation will be limited as there is little to no melt for the explosion to propagate to. This could result in a series of smaller impulses instead of a few strong ones. Furthermore, the phenomenon could possibly be explained due to the fact that low dierence in temperature between melt and water reduces the pace of steam generation. This in turn could lead to the protecting layer of vapor never being fully evolved allowing for free heat transfer and a never-ceasing generation of steam.

4.3 Sources of error

Both when performing the experiments and when doing the analysis there were unfortunately several possible sources of error. Fortunately, the results related to steam explosions are less aected. In this section possible measurement errors are discussed. The discussion is twofold; both how large certain errors might be and how it can aect the results. In the discussion, certain errors are dicult to estimate and in these cases a worst case scenario has been assumed and as a consequence, the errors might be slightly exaggerated.

4.3.1 Mass of charge

The mass of the charge released in each experiment was weighted with a scale

with a precision of 10

g . However, the error was much larger for several rea-

sons. The primary reason was that the charge was put into the crucible manually

and often some of the charge was lost, approximately a couple of grams. Orig-

inally the plan was to weigh all the equipment between experiments to more

reliably tell how much of the charge was lost in each experiment. However, this

idea was discarded mainly because the mass is not relevant for the analysis. The only relevancy of mass size is whether or not the FCI will reach a steady- or quasi-steady-state, but that was not of interest in this thesis.

4.3.2 Temperatures of water and melt

The temperatures were measured using a K-type thermocouple with an error smaller than 1

C . Unfortunately, other aspects contributed to a larger error.

The water and melt temperatures could not be measured simultaneously be- cause the laboratory was only equipped to read one thermocouple at a time. In practice, this meant that the temperature of the water was measured about a minute prior to the release of the melt charge. The water temperature dropped slowly, resulting in a temperature decrease of at most 1−2

C . Another possible source of error is that the water and melt might not have been homogeneous when the temperatures were measured, since no stirring was present. Consider- ing this, a reasonable estimation of the total errors is 5

C for the water and 3

C for the melt. The eect of errors in temperatures is important for this thesis.

While temperature does not show up in the jet breakup analysis it is crucial for the steam explosion map.

4.3.3 Quantities obtained from image processing

The last few quantities used in the analysis were obtained by the mean of im- age processing and analysis on the experiment data. In this category falls jet diameter and velocity. The method used for measuring distances in the pic- tures, which both quantities depend on, was to rst calibrate the measurements using pictures of an object with known size and extracting a relation between distance and number of pixels. This calibration was made accurately, the size of the known object was measured several times with precision and the con- version to pixels was likely also correct. Nonetheless a small and neglectable error existed, but since this particular part of the analysis failed, this section is primarily of academic interest.

4.3.4 Other possible sources of errors

Apart from the possible errors discussed earlier there are several other sources which need to be examined. Some physical quantities were taken from litera- ture. Temperature dependencies were not taken into consideration, which could potentially yield a very small error and neglectable error, but the analysis never reached the point where these quantities would be used.

Finally, even when studying the binary outcome of steam explosions it is not

a 100 % certain endeavor to decide if it occurred or not. In most cases, it is

easily observed whether or not a steam explosion occurred, but in some cases

it was not. In some experiments, the melt would make a slight expansion just

after entering the water which closely assembled a steam explosion. Other cases

displayed a similar expansion further down in the water tank. Moreover, the

camera's eld of view did not cover the whole tank and a steam explosion at the bottom of the tank could not be detected. Considering this, there might exist data points which were misjudged, although it is unlikely. This could possibly explain the deviant measuring points in the steam explosion maps. The usage of X-ray images would likely have simplied this part of the analysis.

4.4 Conclusions and suggestions for further research

In this thesis FCI experiments with tin were performed with the intention of studying jet breakup and steam explosions. Analysis of the jet breakup could not be performed due to lack of data, but the steam explosion experiments yielded some interesting results. It was found that higher melt superheat and lower water temperature caused steam explosions and that in the region between having steam explosions or not, immediate steam explosions could occur. In this event the melt would explode immediately upon contact with the water, while earlier theory states that steam explosions occur later in FCIs.

The authors would like to suggest a few areas in which FCI and steam

explosion research could be continued. Most of these regards immediate steam

explosion. Most importantly, the experiments should be recreated, preferably

with high temperature melts. Secondly, the impulse generated by an immediate

steam explosion should be compared to that of an ordinary steam explosion to

determine whether it is less dangerous. Finally the exact boundary between

having steam explosions or not should be determined and whether or not it

depends on other parameters than temperatures.

5 References References

[1] Qazilbash Hira Ginopoulou Christina, Manickam Louis. Reactor cavity

ooding in swedish bwrs. May 2012.

[2] Roberta Concilio Hansson. An Experimental Study on the Dynamics of Melt- Water Micro-Interactions in a Vapor Explosion. Licentiate thesis, Royal Institute of Technology, 2007.

[3] H. Haraldsson. Breakup of jet and drops during premixing phase of fuel coolant interactions, 2009.

[4] Daniel Magallon. Steam explosion in light water reactors. January 2006.

[5] K. Moriyama J.Sugimoto N. Yamano, Y.Maruyama. Technical note on ex-

vessel melt debris coolability and steam explosions. Technical report, OECD

Nuclear Energy Agency, 1996.

A Failed experiments

When performing the experiments a large number did not yield any data which could be used for analysis. Other experiments were performed with the purpose of testing the equipment and setup. Some of the diculties encountered is described in this section.

A.1 The use of better simulant materials

The experiments were initially performed using materials with a higher melting temperature which better simulates corium found in a reactor vessel. The ma- terials in question were W O

CaO , W O

Bi

O

and W O

ZrO

. The problems encountered were mainly associated with heating the charge to suciently high temperatures reliably. Temperatures of about 1300-1400 degrees were needed and often when approaching this temperature the furnace would fail and the experiments would have to been restarted. One possibility might have been in- sucient insulation used, but it is also inherently dicult to reach such extreme temperatures with the relatively simple setup used for the experiments in this thesis. Because of this sometimes the charge would not even melt, a problem mainly encountered with W O

CaO .

A.2 Formation of jets

To obtain proper results from the experiments the melt needed to form a jet when being poured into the water. In many cases, multiple droplets formed instead of a jet, which renders the experiment useless. It was not until a custom made crucible in metal with a sloping funnel was received that jets were obtained in the experiments.

A.3 Failure of various equipment

While the two aforementioned reasons were by far the ones which lead to the

most failed experiments, some equipment have malfunctioned seemingly at ran-

dom causing failed experiments. Examples of equipment which have malfunc-

tioned are cameras not being triggered to start recording, the automatic release

system failing and leaking of the charge.