Verification of leakage through the side reflector graphite of the PBMR reactor

REFLECTOR GRAPHITE OF THE PBMR

REACTOR

MARl US VAN WYK

M. ENG. (MECHANICAL)

DISSERTATION SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE

MAGISTER ENGENERIAE (MECHANICAL ENGINEERING) SCHOOL OF MECHANICAL AND MATERIAL ENGINEERING

AT THE

NORTH-WEST UNIVERSITY, POTCHEFSTROOM

PROMOTER: D.L.W. KRUEGER

POTCHEFSTROOM

- - '" _{.-. -. -.. -.- -}

--ABSTRACT

As a result of manufacturing and temperature constraints of the reactor core components, leakage flow is an inevitable, and generally undesirable, occurrence within the PBMR reactor. Leakage flow occurs between the narrow gaps of the graphite blocks within the Core Structures as a result of the large pressure gradient over the pebble bed.

The PBMR utilizes computational fluid dynamics (CFD) codes for the simulation of flow and heat transfer through the reactor. Due to hardware limitations, it is not yet possible to model the leakage paths between the graphite blocks of the reactor CFD model in detail since, in some locations the leakage paths are in the order of 175-micron in width and would require a very fine mesh structure. It is therefore required to simplify some of the more complex leakage flow paths with the use of a porous medium sub-model.

In order to calibrate the porous medium sub-model to produce similar flow resistance as the detail leakage path, it is necessary to separately model the complex leakage path in detail, using CFD to determine the actual flow resistance characteristics as function of leak flow rate and helium density. There was a wide spread in the calculated Reynolds numbers throughout the flow path of the detail leakage paths, and it was uncertain whether the leakage flow was laminar, in the transition zone or turbulent. This raised uncertainty with regards to the accuracy of the CFD models of the detail leakage paths.

An experiment was devised that contained all the flow phenomena of the actual detail leakage paths within the reactor, and was used to validate the numerical CFD modelling of the helium flow through the side reflector leakage paths. Three leakage gap sizes, 175, 280 and 380-micron were experimentally tested. The experiments were simulated with CFD and it was found that there was a

good

correlation between the laminar CFD results in both the 175 and 280-micron gap sizes.

It was finally concluded that the detail leakage path CFD models were correctly modelled as laminar.

Pageii

--Lekvloei is 'n onafwendbare, en in die algemeen ongewenste, gebeurlikheid in die Kieselbed Kern Reaktor (KKR). Lekvloei is 'n gevolg weens beperkinge in die vervaardiging, asook die temperatuur limiete van die reaktor kern komponente. Lekvloei vind plaas tussen die nou gapings van die grafiet blokke binne in die kern struktuur as gevolg van die groot druk gradient oor die kieselbed.

Die KKR maak gebruik van berekenings vloei meganika (BVM) rekenaarpakkette vir die simulasie van helium vloei asook hitte oordrag binne in die reaktor. Weens beperkinge in hardeware is dit nog nie moontlik om die lekvloeipaaie tussen die grafietblokke in die reaktormodel in detail te modelleer nie. Die lekvloeipaaie is in sekere posisies in die orde van ongeveer 175 mikron in wydte, en sal noodgedwonge 'n baie fyn roosterstruktuur verys. Dit is gevolglik nodig om die lekvloeipaaie in die reaktormodel te vereenvoudig met 'n porieuse submodel.

Dit is dus noodsaaklik om die detaillekvloeipaaie in die reaktor d.m.v 'n sub BVM model te simuleer. So kan die porieuse sub model in die reaktor gekalibreer word om dieselfde vloeiweerstand in die porieuse model teweeg te bring as funksie van die helium vloeitempo en digtheid. Daar was 'n groot variasie in die Reynolds getalle bereken deur die vloeipad van die detail lekvloeipaaie en dit was nie duidelik of die vloei laminer, turbulent of in die transisie gebied was nie. Daar was gevolglik onsekerheid aangaande die BVM resultate van die detaillekvloeipaaie.

'n Eksperiment was ontwerp wat al die vloeiverskynsels bevat het wat voorkom in die detail lekvloeipaaie in die reaktor. Die doelwit van die eksperiment was om die numeriese BVM modulering van die helium vloei deur die reaktor kantreflektor lekvloeipaaie te valideer. Drie verskillende lekvloeipad wydtes, 175, 280 en 380 mikron, was eksperimenteel ondersoek. Die eksperimentele opstelling was met BVM gesimuleer en daar was gevind dat daar 'n goeie korrelasie was tussen die laminere BVM en eksperimentele resultate in die 175 en 280 mikron gaping wydtes.

Die uiteindelike gevolgtrekking was dat die detail lekvloeipad BVM modelle korrek gesimuleer was as laminer.

CONTENTS

TABLE OF CONTENTS

ABSTRACT II

UITTREKSEL III

LIST OF FIGURES VI

LIST OF TABLES VIII

TABLE OF ABBREVIATIONS AND ACRONYMS IX

1. INTRODUCTION 1

1.1 BACKROUND TO THE PBMR PROJECT 1

1.2 HOW THE PBMR REACTOR UNIT WORKS 3

1.3 CFD MODELLING OF THE PBMR REACTOR 5

1.4 OBJECTIVES AND OVERVIEW OF THIS REPORT 6

2. SIDE REFLECTOR LEAKAGE CFD SIMULATIONS 8

2.1 SIDE REFLECTOR LEAKAGE 8

2.2 SEALING KEY CFD SIMULATIONS & GEOMETRy 12

2.3 SIMULATION METHODOLOGy 15

2.4 SIMULATION RESULTS AND UNCERTAINTy 16

3. EXPERIMENTAL METHODOLOGy 23 3.1 EXPERIMENTAL PROPOSAL 23 3.2 CONCEPTUAL DESiGN 24 3.3 EXPERIMENTAL SET-UP 34 3.4 EXPERIMENTAL RESULTS

..

37 3.5 UNCERTAINTY ANALYSIS. 46 4. LITERATU RE STUDY 48 4.1 INTRODUCTION ... 48

4.2 FLOW BETWEEN TWO INFINITE PARALLEL PLATES 48

4.3 FLOW THOUGH A RECTANGULAR .CHANNEL 51

4.4 FLOW LOSS THROUGH A SUDDEN EXPANSION OR CONTRACTION 52

5. CFD SIMULATION OF EXPERIMENT 54

5.1 OVERVIEW OF DETAIL GAP SIMULATION PROCEDURE 54

5.2 GAP GEOMETRY AND NUMERICAL MESH 54

5.3 BOUNDARY CONDITIONS & DIFFERENCING SCHEME 58

5.4 ANALYSIS AND RESULTS 59

5.5 CONVERGENCE 69

6. CONCLUSIONS 71

6.1 EXPERIMENTAL TESTS 71

6.2 CFD SIMULATIONS OF EXPERIMENT 71

6.3 DETAIL LEAKAGE CFD SIMULATIONS 72

7. RECOMMENDATIONS 73

7.1 EXPERIMENTAL TESTS 73

7.2 CFD SIMULATIONS OF EXPERIMENT 73

7.3 DETAIL LEAKAGE CFD SIMULATIONS 73

8. REFERENCES 74

APPENDICES 77

PROCESSED EXPERIMENTAL DATA 77

EXPERIMENTAL PROCEDURE.. ... 79

MASS BALANCE CALIBRATION

81

Pagev

--CONTENTS

LIST OF FIGURES

Figure 1: Basic Fuel Element Design Of The PBMR Reactor [2]. 1

Figure 2: PBMR Fuel Elements Or "Pebbles" [2]. 2

Figure 3: Main Power System Of The PBMR [4]. 2

Figure 4: Layout Of The MPS [1]. 3

Figure 5: Simplified Helium Flow Path Within The PBMR Reactor [6]. 4

Figure 6: Typical Leakage Flow Path With Sealing Key. 5

Figure 7: Leakage Flow Paths In-Between The Graphite Blocks Within The PBMR Reactor [7]. 8 Figure 8: Primary, Secondary Coolant Flow And Leakage Flows Through The Reactor [8]. 9

Figure 9: Location Of Typical Sealing Keys. 10

Figure 10: Leak Path Between Adjacent Graphite Blocks With Sealing Key. 11

Figure 11: Typical Mesh Structure Of Sealing Key Model [7]. 12

Figure 12: Porous Key Model Mesh Structure [7]. 12

Figure 13: Helium Flow Path Of The Reactor CFD Model [7]. 13

Figure 14: General Three-Dimensional Key And Helium Flow Path Geometry [10]. 14 Figure 15: Plan View Of Helium Leak Flow Path Between Adjacent Graphite Blocks [10] 14 Figure 16: Different Sealing Key Positions Simulated In The CFD Analysis. 15 Figure 17: Pressure Loss In Detail Key Model Compared With Its Equivalent Porous Model (5kPa Test

Case). 16

Figure 18: Pressure Loss Mainly Due 175-Micron Gaps (Centralised Position) 17 Figure 19: Pressure Loss Through Porous Model Of Inner Side Reflector Key (Centralized Location).

18

Figure 20: Sections Where Reynolds Numbers Were Calculated. 19

Figure 21: Wall Function y+ Values Of Inner Side Reflector Key Model (5kpa Pressure Differential). 20 Figure 22: Wall Function y+ Values Of Inner Side Reflector Key Model (250kpa Pressure Differentia/).

21

Figure 23: Wall Function y+ Values Below 30 21

Figure 24: Expected Situation: Laminar Modelling Under Predicts Leakage Flow Rate As Opposed To

Turbulence Modelling. 22

Figure 25: Conceptual Experimental Model That Closely Replicates The Actual Key Set-up Within The

Reactor. 25

Figure 26: Horizontal Cut Away Section Of Figure 25 Portraying The Helium Flow Around The Key.. 26 Figure 27: Three Dimensional Cut Away Of The Helium Flow Path In The Experiment 27 Figure 28: Simplified Test Arrangement Entailing Helium Flowing Through A Thin Slit. 28

Figure 29: Shank Arrangement Manufactured Entirely On A Lathe. 29

Figure 30: Experimental Rig For Instrumentation And Proof Of Concept [17]. 30

Figure 31: Final Rig For Flow Rate Measurement [17]. 30

Figure 32: Velocity Variation Of Radial Flow Path Compared With Constant Area Flow. 31 Figure 33: Friction Loss Of Radial Flow Path Compared With Constant Area Flow. 33 Figure 34: Pressure Recovery Of Radial Flow Path Compared With Constant Area Flow 33 Pagevi

-Figure 35: Test Rig. 34

Figure 36: Mass Balance Rig [19]. 35

Figure 37: Complete Experimental Set-up. 36

Figure 38: Test Piece Geometry [19]. 37

Figure 39: Typical Raw Dataset [19]. 38

Figure 40: Typical Processed Dataset [19]. 39

Figure 41: Typical Mass Calibration Curve [19]. 39

Figure 42: Upper Flange With O-Ring Groove [19]. 41

Figure 43: Lower Flange With Proud Face [19] 41

Figure 44: Density Variation For The 175-Micron Channel Height. 43 Figure 45: Density Variation For The 280-Micron Channel Height. 45 Figure 46: Density Variation For The 380-Micron Channel Height. 45

Figure 47: Flow Between Two Infinite Parallel Plates. 48

Figure 48: Rectangular Cross Section For Which Fully Developed Flow Solution Is Known. 51 Figure 49: Schematic Of Sections With Sudden Expansion and Sudden Contraction (excerpted from:

Flow Resistance: A Design Guide For Engineers, I.E. Idelchik, Erwin Fried [29]) 52

Figure 50: Cut Away of Test Piece. 55

Figure 51: Helium Flow Cavity within Test Piece. 55

Figure 52: Two Dimensional Mesh Structure of Helium Flow Path in Experiment 56

Figure 53: Refined Mesh Structures.. 57

Figure 54: Refinement 4 Mesh Structure at the Gap Exit. 57

Figure 55: Inlet, Outlet and Wall Boundaries of CFD Model. 58

Figure 56: Symmetry Boundaries. ... 59

Figure 57: Static Pressure Variation For Mesh Sensitivity Analysis (175-Micron, 250kPa Case). 63 Figure 58: Static Pressure Variation For The 175-Micron Gap Height (Laminar And Turbulent Values).

65 Figure 59: Static Pressure Variation For The 175-Micron Gap Height (Laminar And Calculated

Values) 65

Figure 60: Static Pressure Variation for the 280-Micron Gap Height (Laminar and Turbulent). 67 Figure 61: Static Pressure Variation for the 280-Micron Gap Height (Laminar and Calculated Values).

67 Figure 62: Static Pressure Variation Through 380-Micron Gap Height 69 Figure 63: Static Pressure Convergence Of 250kPa Laminar Simulations (175-Micron Gap). 70

Page vii

---CONTENTS

LIST OF TABLES

Table 1: Sealing Key Model Dimensions [10] 14

Table 2: Centralized Location, Turbulent & Laminar. 17

Table 3: Sealing Key Model Reynolds ..Numbers 18

Table 4: Actual Operating Conditions Of Key System In Reactor. 23

Table 5: Schedule Of Experiments With Helium [19]. 37

Table 6: Average Mass Flow Rates for Helium [19]. 42

Table 7: Instrumentation Accuracy [19]. 47

Table 8: Values of

S

valid for Re < 10000 [29]. 53

Table 9: Calculated Pressure Loss Through Gap Excluding Entrance And Exit Losses. 59 Table 10: Calculated Pressure Loss Though Test Piece (175-Micron Gap) 60 Table 11: Calculated Pressure Loss Though Test Piece (280-Micron Gap) 61 Table 12: Calculated Pressure Loss Though Test Piece (380-Micron Gap). 62

Table 13: Static Pressure Loss For Mesh Sensitivity Analysis. 62

Table 14: 175 Micron Gap Height Laminar Results. 64

Table 15: 280 Micron Gap Height Laminar and Turbulent Results. 66

Table 16: 380 Micron Gap Height Laminar and Turbulent Results. 68

Table 17: Test Data (175-Micron) [19]. 77

Table 18: Test Data (280-Micron) [19]. 78

Table 19: Test Data (380-Micron) [19]. 79

TABLE OF ABBREVIATIONS AND ACRONYMS Pageix - --Abbreviation or Definition Acronym

CCS _{Core Conditioning System}

CFD _{Computational Fluid Dynamics}

FRU Flow Research Unit

HPC _{High Pressure Compressor}

HPT _{High Pressure Turbine}

HVAC Heating, Ventilation and Air-conditioning

LPC _{Low Pressure Compressor}

MCR _{Maximum Continuous Rating}

MPS _{Main Power System}

P Pressure

PBMR Pebble Bed Modular Reactor

PCU Power Conversion Unit

PPB _{Primary Pressure Boundary}

PTG Power Turbine Generator

QAP _{Quality Assurance Procedure} RCS _{Reactivity Control System}

RCSS Reactivity Control and Shutdown System

RU Reactor Unit

-- __u..._ _...-.--.

INTRODUCTION

1. INTRODUCTION

1.1 BACKROUND TO THE PBMR PROJECT

In the following sections of this chapter a brief background of the PBMR reactor shall be presented to familiarize the reader with the main workings and terminology of the PBMR reactor. Thereafter, focus will be shift to the helium leakage flow within the graphite core structures of the PBMR reactor and its associated modelling intricacies. Apart from the obvious advantages that the Pebble Bed Modular Reactor (PBMR) concept has over alternative power generation technologies in the market today, the fundamental differences are rooted in its nuclear fuel structure and direct helium Brayton cycle.

The original concept was born in the 1950s out of an ingenious idea by Dr. Rudolf Schulten who can undoubtedly be regarded as the father of the pebble bed reactor [1]. His idea was to produce a safer form of nuclear fuel to be used in a new high temperature reactor. The fuel would be, and is still today in PBMR guise, in the form of a hard billiard-ball-like sphere, which encapsulates silicon carbide-coated uranium granules as can be seen in Figure 1.

Fuel element design for PBMR

Diameter 60 mm

Fuel sphere

.t

smm graphite layer

.

Coaled po rticles imbedded in graphitematrix

~. Pyralytic carbon

.;.

($

Snican carbid. barrier coa6ng

Inner pyralytic corban

Half section . Porous carbon bull.r

Diameter0,92 mm IIIC::JJ.

Cooted particle

-Diameter 0,5 mm Uraniumdioxide

Fuel

Figure 1: Basic Fuel Element Design Of The PBMR Reactor [2].

This encapsulation or rather, "containment" function of each fuel particle is one of the fundamental design differences and advantage between current generation reactors and high temperature gas-cooled reactors with coated particle fuel. The inherent Page1

design of these fuel particles, coupled with the advanced design of the PBMR reactor, prevents a major or severe loss of containment. Figure 2 gives an interesting physical perspective of the PBMR fuel element.

Figure

2: PBMR Fuel Elements Or "Pebbles" [2].

The PBMR reactor is a 400MW helium cooled, graphite moderated high temperature reactor (HTR). The PBMR reactor unit (RU), which forms part of the main power system (MPS), shown in Figure 3, consists of a 1250-ton vertical steel pressure vessel (RPV), 6 m in diameter and about 30.17 meters high [3]. It is lined with a 1-meter thick layer of graphite bricks, which serves as an outer reflector and a passive heat transfer medium. The graphite brick lining is drilled with vertical holes to house the control elements.

MAIN

POWER SYSTEM

Recuperator

Pre-cooler

Figure 3: Main Power System Of The PBMR [4].

---

----INTRODUCTION

Helium is used as the coolant and energy transfer medium, to drive a closed cycle gas turbine and generator system shown in Figure 4. The reactor core, when fully laden, contains approximately 456000 fuel spheres or pebbles [5]. The geometry of the fuel region is annular and located around a central graphite column, which serves as an additional nuclear reflector.

MPS layout

r

ReIIeloI~esse~

j

~~e vesselrondifioning~~'em

l

I

COle(ondifioningsystem

I

I

GenerDlOr

I

I

~werturbine

]

[Re<uperolor1

f"

Figure 4: Layout Of The MPS [1].

1.2 HOW THE PBMR REACTOR UNIT WORKS

Consider Figure 5, which present a simplified two-dimensional cake slice layout of the RU main helium flow path. The helium enters the RU at point A named the inlet plenum, at a pressure of approximately 9Mpa, 500°C, where after it flows upwards in the riser channels, denoted by flow path number 1 in Figure 5, through the inlet slots towards point B which is located directly above the annular fuel zone marked in red. From point B the helium flow is split up into three distinct paths, namely:

.

The helium flow path between points Band C via path 3 though the actual pebble bed where the helium is heated to approximately

900°C,

which is the prime Page3

INTRODUCTION

function of the RU. From here the helium exits the fuel zone through outlet slots machined into the graphite bricks below the core (between points C and D) into the outlet plenum shown at point D.

. The helium flow path between points Band E via path 4 in Figure 5, named the annulus. This path serves to equalize the pressure between the outer side of the side reflector graphite blocks and the core. This helium path is naturally formed by the necessary gap that exists between the core barrel (CB) and the SR as a consequence of the difference in expansion coefficients of the CB steel and the SR graphite. The function of the CB that encloses the graphite SR is mainly to provide support to the annular graphite brick structure of the SR. The SR graphite blocks are partly prevented to move outward in a radial direction under the mass of the pebbles within the core due to the equalization pressure present in the annulus. The shown path between points E and D, that is between the bottom of the annulus and the outlet is a simplified presentation of the helium that leaks through the gaps between the graphite blocks from the annulus to the lower part of the fuel zone, which is at a lower pressure. This leakage present within the SR will be discussed in more detail in the next chapter.

. The helium flow path between points Band F along path 5: This helium path serves to cool the centre reflector graphite.

Fuel zone filled with pebbles (red colored area)

1

-

Flow in risers

2

-

Inlet I Outlet leakage flow

3

-

Average flow through fuel zone

4

-

Flowdown annulus

(between side reflector & barrel)

5

-

Centre column cooling

6 -Control rod cooling

uti

Figure 5: Simplified Helium Flow

Path Within The PBMR Reactor [6].

INTRODUCTION

Note the leakage path between the inlet plenum and the outlet plenum. Once again this leakage path is a simplified presentation of helium that leaks from the inlet plenum at higher pressure through the gaps between the graphite blocks to the outlet plenum at lower pressure.

The helium exits the RU through the outlet where after the helium flows to the power conversion unit (PCU) where the heat energy is transformed into electrical energy. The cycle is then repeated.

1.3 CFD MODELLING OF THE PBMR REACTOR

PBMR utilizes computational fluid dynamics (CFD) codes for the simulation of the flow and heat transfer within the reactor. The reactor is modelled in great geometrical detail resulting in a rather large model. Due to computer hardware limitations it is not yet possible to model the more complex leakage paths between the graphite blocks in detail.

To complicate matters further, the side reflector leakage paths contain special sealing mechanisms referred to as keys, designed to minimize the leakage as shown in Figure 6.

typicalsquareshapedouter sidereflectorkey

innersidereflectorblock

Figure 6: Typical Leakage

Flow

Path With Sealing

Key.

The Helium flows around very narrow gaps

- in the order of 175-micron- aroundthe

sealing key described in detail in the next chapter. Modelling of the helium flow through these sealing mechanisms therefore require a very fine numerical mesh structure which cannot be effectively incorporated within the relatively course reactor CFD model.

It is therefore required to simplify the complicated key leakage flow paths with the use of a porous medium model within the reactor CFD model. The porous medium modelling technique essentially simplifies the geometry of the key, but retains the flow resistance property and position of the key. This requires the actual sealing key arrangement to be separately modelled in detail in order to determine the flow resistance characteristics as function of leak flow rate and helium density. The simplified porous medium model within the reactor CFD model is then calibrated to produce similar results.

The detail key CFD simulations are modelled as either laminar or turbulent due to simulation code turbulence model limitations. Also, there was a large scatter in the calculated Reynolds numbers within the actual leakage key arrangement in it was uncertain whether the helium flow was in the laminar, transition or turbulent regime or a combination thereof. This raised uncertainty with regards to the accuracy of the detailed sealing key simulations performed. It is therefore questionable whether the leakage flow rate through the reactor side reflector was under or over predicted and could have far reaching implications.

The implication of over or under predicting the leakage flow rate can be summarized as follows: Over prediction of the leakage flow rate implies less helium flow through the fuel core that will result in an over prediction of the maximum fuel temperature. Under predicting the leakage flow rate causes the exact opposite to happen.

Due to this uncertainty present within the reactor CFD model it was decided to device an experiment that contains all the flow phenomena of the actual sealing key arrangement within the reactor that can be used to validate the numerical CFD modelling of the helium leakage flow through the side reflector leakage paths.

1.4 OBJECTIVES AND OVERVIEW OF THIS REPORT

From the above background it can be seen that the leakage path through the side reflector of the PBMR reactor needs to be simulated with good accuracy. In order to

INTRODUCTION

calibrate a simplified model, using porous media, against a detailed model it is required to establish the accuracy of the detailed model. In order to achieve this, it was proposed to carry out specifically designed physical model tests to compare measured leak flow results against a detailed CFD model of the same test configuration.

This report will set out to describe the experimental set-up and compare the experimental and numerical CFD results obtained. Based on the outcome of this comparison it is then possible to deduce whether the leakage flow in the PBMR reactor was under or over predicted and also to formulate a simulation methodology for the detail key CFD models.

In the next chapter (Chapter 2) the CFD simulations of the detail leakage key set-up within the reactor will be discussed together with the inherent uncertainty contained within the simulations.

The experimental proposal and conceptual design of the experimental will be discussed in chapter 3 together with the experimental set-up and testing. In the latter part of chapter 3 the experimental results will be thoroughly discussed.

Chapter 4 shall present the theory of flow between flat plates and channel sections. This theory will be utilized in chapter 5 to calculate the pressure loss through the gap of the experimental setup.

Finally, in chapter 5, the experimental results will be compared with the obtained CFD and calculated results.

SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

2. SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

2.1 SIDE REFLECTOR LEAKAGE

Leakage

flow

is an inevitable and generally undesirable occurrence within the PBMR reactor as a result of manufacturing and temperature constraints of reactor core components. Leakage flow occurs between the gaps of the graphite blocks within the Core Structures (CS) due to the large pressure gradient over the pebble bed (approximately 280kPa). Figure 7 gives a physical representation of some of the graphite blocks located between the fuel zone and the CB.

typical leakage flow path between adjacent graphite blocks

outer side reflector graphite block inner side reflector

graphite block 1 I?; 10 ,;;:::: .... ,<1> 1"5

,0

Figure 7: Leakage Flow Paths In-Between The Graphite Blocks Within The PBMR

Reactor [7].

The function of the graphite blocks is four fold [3]:

.

Firstly it acts as a container that has to maintain its shape to form the core of the

reactor.

SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

It also acts as a thermal shield between the fuel core and the metallic core structures i.e. the CB.

. The graphite structure is also used to channel the primary helium coolant from the inlet plenum (see Figure 8) through the core and then to channel it to the outlet plenum from where it is collected in the outlet plenum and flows to the high-pressure turbine.

.

. And finally it acts as a gamma and neutron radiation shield by limiting the loss of radiation from the core.

Figure 8 shows a simplified flow diagram showing the primary and secondary coolant flows as well as the various leakages flows.

CR coolant

ow

.

Primary cooling flow

. Secondary cooling flow

.

Leakage flow

Leakage hrough top reflector

Inlet slots Top inlet plenum

Control rod coolant flow

Riser channels Bottom inlet plenum Inlet pipe

Outlet slots

J---outlet

pipe

Outlet manifold

Figure 8: Primary, Secondary Coolant Flow And Leakage Flows Through The

Reactor [8].

Helium

returning from the PCU flows through the inlet pipe of the reactor unit into the bottom inlet plenum from where it flows via the riser channels up into the top inlet plenum. From the top inlet plenum the helium flows through the inlet slots into the core. A small percentage of the total helium flow leaks into the leakage paths shown in Page9

SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

orange in Figure 8 (see Figure 7 for more detail of the actual leakage path).

At the maximum mass flow rate, the large pressure differential across the pebbles results in gas leaking through the gaps in the side reflector therefore bypassing the core by flowing down the annulus between the graphite and core barrel. This flow then re-enters the core lower down thus bypassing a significant section of the core.

This phenomenon has the undesirable effect of raising the maximum fuel temperature since more leakage flow implies less helium flow through the core. It is therefore of the utmost importance to increase the leakage path flow resistance in order to maximise the helium flow rate passing through the core.

In an attempt to minimize leakage through the leakage paths, a sealing mechanism, referred to as a sealing key, is used. A sealing key is a robust, rectangular shaped, graphite-sealing device as shown in Figure 9, designed to minimize leakage flow by simply increasing the flow resistance between the inside and outside of the side reflector graphite blocks.

Graphite block Sealing keys placed_{on top of each other}

... Je E m ~ (,) E E .... x .... Detail of typical sealing key

Figure 9: Location Of Typical Sealing Keys.

Figure 10 presents a plan view of a typical leakage path. The sealing key is forced

flush against the graphite blocks due to the pressure differential over the key,

Page10

SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

effectively sealing the leakage path.

-loose chamfered square key

Key forced flush against graphite blocks due to pressure difference low pressure

side

Helium flow path between adjacent graphite blocks

high pressure side

Figure 10: leak Path Between Adjacent Graphite Blocks With Sealing Key.

Leakage flowinfluences the following:

.

Core temperature: Keys are used in the leakage paths between the graphite blocks to restrict the radial leakage flows and thus force more flow through the

pebble bed resulting in a lower

fuel temperature. Poor key sealing efficiency significantly increases the core bypass flow, which reduces the amount of coolant gas passing through the core to cool down the pebbles. The bypass flow runs though the annulus and heats up the core barrel, which in turns heats up the reactor pressure vessel.

. The pressure loss through the core: In order to reduce the core temperatures, it is required that the keys seal well to force more coolant flow through the core. This however increases the pressure differential across the core.

. The pressure differential across the side reflector graphite blocks: The pressure differential across the side reflector is directly related to the pressure differential across the core. Both these values will be significantly influenced by the efficiency of the sealing keys.

. The temperature gradient over the side reflector graphite blocks: Graphite temperatures are of significant importance since the life of the graphite is dependant on the temperature and radiation dose that the graphite is exposed to over the life of the block. Furthermore, temperature gradients across the blocks induce thermal stresses, which may result in the failing of graphite blocks.

It is therefore imperative to have confidence in the prediction of CFD for the simulation of the leakage paths.

SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

2.2 SEALING KEY CFD SIMULATIONS & GEOMETRY

Detailed simulation of keys within the reactor model is computationally expensive because of the more detailed and very fine mesh required. As a consequence each key is modelled as a porous medium within the detail reactor model, which essentially simplifies the geometry of the key, but retains the flow resistance property and position of the key.

Compare the detailed mesh structure in Figure 11 showing the flow path around a typical detail key model with its much simpler porous model equivalent in Figure 12. The mesh density in this instance was reduced by a factor of 50.

Figure 11: Typical Mesh Structure Of Sealing Key Model [7].

Porosity defined for red coloured cells

Figure 12: Porous Key Model Mesh Structure

[7].

The location of all the different porous key models in the helium flow path of the detail reactor model is shown in Figure 13.

--.---SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

Outer side reflector keys

Inner side reflector keys

Inner bottom reflector keys (250 mm)

Outer bottom reflector keys

Inner bottom reflector keys (450mm)

Inlet plenum sealing keys

Outer bottom reflector keys

Figure 13: Helium Flow Path Of The Reactor CFD Model

[7].

Table 1 presents a list of all the different key model configurations with their respective dimensions that were analysed. All dimensional variables used are explained in Figure 14 and 15.

HELIUM FLOW PATH AROUND KEY

J.,5

[7

1x1 mm chamfer

Figure

14: General Three-Dimensional Key And Helium Flow Path Geometry [10].

80.01

Loose chamfered square key Helium flow path

between adjacent

graphite blocks B

()

1X1 mm CHAMFER

79.66

Figure 15: Plan View Of Helium leak Flow Path Between Adjacent Graphite

Blocks [10].

Table 1: Sealing Key Model Dimensions [10].

Page14

--- ---

---Key A (mm) B C Key Chamfer

(mm) (mm) height Centralized location

Outer side reflector 0.175 0.175 7 450 10x15° Inner side reflector 0.35 0.35 4 250 10x15° Outer bottom reflector 0.175 0.175 7 450 1x1 Inner bottom reflector 0.175 0.175 4 250 1x1 Inner bottom reflector 0.175 0.175 7 450 1x1 Inlet plenum sealing 0.175 0.175 7 450 1x1

-. ._ ..u.. _...

SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

2.3 SIMULATION METHODOLOGY

In theory, as explained in Figure 10, the sealing key should be forced flush against the graphite blocks due to the expected pressure difference across the leakage path. However, in reality there will always be some leakage present due to geometrical inaccuracies and consequently all the key models were simulated either with the key placed in the centre of the leakage path (centralized location) or flush against the graphite blocks (flush location) yielding two extreme conditions as can be seen in Figure 16. Even with perfect sealing the chamfer that is machined at the end of each key, to prevent the formation of cracks within the graphite structure of the key, accounts for leakage flow through the reactor side reflector.

Centralized location Flush location

Figure 16: Different Sealing Key Positions Simulated In The CFD Analysis.

Various

simulations were carried out on different types of keys found in the reactor in order to determine the flow resistance characteristics as a function of leak flow rate and Helium density. The simplified porous medium model (Figure 12) was then calibrated to produce a similar flow resistance.

All the key models were simulated at three different pressure gradients of 5kPa,

Page15

-- - -

-Key A (mm) B C Key Chamfer

(mm) (mm) height Flush location

Outer side reflector 0.35 0 7 450 10x15° Inner side reflector 0.7 0 4 250 10x15° Outer bottom reflector 0.35 0 7 450 1x1 Inner bottom reflector 0.35 0 4 250 1x1 Inner bottom reflector 0.35 0 7 450 1x1

100kPa and 250kPa respectively [7]. These pressure values were chosen to be within the expected operating envelope of the keys found in the reactor [8]. The least square method was then used to fit a quadratic polynomial through the obtained data sets of pressure as a function of measured superficial velocity.

From the obtained coefficients of the quadratic polynomial, M = XIV2+ x2V , that is, XI and x2' the porous medium permeability coefficients, in STAR-CD by Computational Dynamics (CD) [11], could then be calculated as a = XI/ dl and f3 = x2/ dl, where dl represents the porous medium length in the direction of the measured superficial velocity.

2.4 SIMULATION RESULTS AND UNCERTAINTY

Figure 17 presents a contour plot of total pressure through the helium flow path of the inner side reflector detail key CFD model compared to its equivalent calibrated porous medium model. 5000. 4688. 4375. 4083. 3750. 3438. 3125. 2813. 2500. 2188. 1875. 1563. 1250. 937.5 625.0 312.5 0.0000

Detail key CFD model _{Equivalent porous} medium model TOTALPRESSURE

RELATIVE PA

Figure

17: Pressure Loss In Detail Key Model Compared With Its Equivalent Porous Model (5kPa Test Case).

The pressure loss through the detail key model was found to be largely due to the flow resistance through the 175 micron gaps (key in centralised position), denoted as gap A and B in Figure 17 and reproduced in Figure 18 for clarity.

Page16

---SIDE REFLECTOR LEAKAGE CFD SIMULATIONS TOTAL PRESSURE RELATI VE PA 5000. 4639. 4676. 4516. 4355. 4194. 4033. 3672. 3711. 3549. 3366. 3227. 3066. 2905. 2743. 2562. 2421. Pressure drop in gap A approximately 2500 Pa Gap A

Figure 18: Pressure Loss Mainly Due 175-Micron Gaps (Centralised Position).

TOTAL PRESSURE RELATIVE PA 2535. 2367. 2239. 2091. 1944. 1796. 1646. 1500. 1352. 1205. 1057. 909.0 761.2 613.4 465.6 317.6 170.0 Pressure drop in gap B approximately 2500 Pa

The obtained permeability coefficients are presented in Table 2 for the turbulent and laminar centrally located detail key model simulations. It is evident that the flow resistance is higher in the turbulent cases compared to all the laminar cases.

Table 2: Centralized Location, Turbulent & Laminar.

Figure 19 presents a plot of the pressure loss through the calibrated inner side reflector porous model for both the turbulent and laminar cases. It is clear that the leakage flow would be under estimated in the turbulent case relative to the laminar case under the same pressure loss.

--- - -

-Page17

Turbulent Laminar Key type

Alpha Beta Ipha Beta

Outer side reflector 56495 40208 27786 91170

Inner side reflector 3746 5683 2215 12403

Outer bottom reflector 104130 91964 50723 153392 Inner bottom reflector

140701 82995 55956 232920 (250)

Inner bottom reflector

104130 91964 50723 153392 (450)

10 15 20 Superficialvelocity(mls)

25

Figure 19: Pressure Loss Through Porous Model Of Inner Side Reflector

Key

(Centralized Location).

Normally the choice between laminar and turbulent CFD results would be a rather straightforward decision based on the Reynolds number of the particular set-up. Table 3 presents Reynolds numbers of the detail key models simulated. The Reynolds numbers were calculated at three different sections as shown in Figure 20.

It is evident, particularly in the 175-micron gap (section B) that the Reynolds number appears to be in the transition zone with a typical Reynolds number spread of 300

-5000.

At section C, which represents the chamfer part of the key arrangement, the Reynolds number varied from 2158 to 94851. It therefore seems possible that the flow in the leakage path could be partly laminar, especially through the gap section and partly turbulent through the chamfer

part.

Table 3: Sealing Key Model Reynolds Numbers.

--Page18

:m 250 200 .. a. C. II> II> .2 150 GI :; II> II> GI It 100 50 00 5 Pressure Reynolds

Key Model number

loss (Pa)

Section A Section B Section C

987 1220 447 3521 Outer side 99770 10448 3827 30162 reflector 249442 18043 6609 52085 987 3872 1473 6117 Inner side 99820 24217 9213 38265 reflector 249417 40670 15472 64264 Outer 998 690 312 5081

SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

SectionA

SectionB

Figure 20: Sections Where Reynolds Numbers Were Calculated.

All the turbulent cases were simulated in Star-CD (by Computational Dynamics (CD)) with the standard k - & high Reynolds number turbulence model [11]. A two-layer turbulence model was at the time considered as to computationally expensive since a two-layer model requires a much finer mesh structure. The standard k - & high Reynolds number turbulence model was not the best turbulence model to use since the Reynolds numbers calculated through the 175-micron gap were in some cases in the low Reynolds number regime.

More specifically,with k

-

&

high Reynolds number model the boundary layer is

approximated with wall functions applied within the first cell layer adjacent to the wall of the numerical mesh. In order to accurately predict the correct boundary layer thickness certain criteria has to be satisfied. For example the wall function y+ values are required

to be within a range of 30

-

500 [13][14]. Figure 21 presents a plot of y+ values for the 5kPa inner side reflector key arrangement.

Page19 - --Reynolds number bottom _{99946 7318 3308 53862} reflector 249717 12888 5825 94851 Inner 5000 528 239 2158 bottom 99988 6197 2811 25339 reflector (250) 249947 11599 5262 47426

Front view Back view VPLUS LOCALMX- 119.4 LOCAL MN- 0.0000 119.4 112.0 104.5 97.04 89.58 82.11 74.65 67.18 59.72 52.25 44.79 37.32 29.86 22.39 14.93 7.465 0.0000

Figure 21: Wall Function y+ Values Of Inner Side Reflector Key Model (5kpa

Pressure Differential).

It is evident that the wall function y+ values through the 175-micron gap sections are below the minimum specified required value of 30. Within a large portion of the helium flow through the chamfer the wall function y+ values are within the acceptable limit. Unsuccessful attempts were made to increase the cell height of the wall cells in order to increase the wall function y+ values within the desired range since it was found that the opposing wall cells of the gap would be overlapping each other. This would mean that the entire gap flow path would be modelled as a boundary layer region. The only alternative is to model the flow with a two-layer turbulence model, which does not rely on wall functions to model the boundary layer, but numerically solves the boundary layer and therefore the finer mesh required in the wall region.

In the 250kPa inner side reflector key CFD model the wall function y+ values were within the specified limit (Figure 22) except for a very small area as shown in Figure 23. For this particular case the calculated Reynolds number range was 15472

- 64264.

The gap Reynolds number in this case was higher than the other cases by a factor of 2 because the gap size was 350-micron instead of the usual 175-micron.

Page 20

--SIDE REFLECTOR LEAKAGE CFD SIMULATIONS

Frontview Backview VPLUS

LOCALMX- 784.8 LOCALMN- 0.0000 784.6 735.6 666.7 637.7 566.6 539.6 490.5 441.5 392.4 343.4 294.3 245.3 196.2 147.2 96.10 49.05 0.0000

Figure 22: Wall Function y+ Values Of Inner Side Reflector Key Model (250kpa

Pressure Differential).

YPLUS LOCAL MX- 29.95 LOCAL MN- 11.10 29.95 26.76 27.60 26.42 25.24 24.06 22.66 21.70 20.53 19.35 16.17 16.99 15.61 14.63 13.46 12.26 11.10

Figure 23: Wall Function y+ Values Below

30.

From the above discussion it is clear that there are ample reason to question the validity of the detail key simulations performed. Due to hardware constraints the detail keys could only be simulated as either laminar or turbulent. It was therefore necessary to simulate the detail key model by first assuming that the flow around the actual key was in the well ordered or laminar regime and thereafter the simulations were repeated with a turbulent modelling approach where it was assumed that the flow was in the fully turbulent region.

Page21

--The fundamental problem was to decide which of the two modelling approaches, either laminar or turbulent, would be the more accurate choice. It was not known whether the leakage flow in the reactor side reflector would be under or over estimated and since leakage flow influences the maximum fuel temperature this posed to be a major CFD modelling problem which could only be resolved with experimental testing.

It was however expected that laminar modelling would under predict the leakage flow rate as apposed to turbulence modelling for a fixed pressure differential P across the reactor side reflector as shown in Figure

24.

-laminar CFDmodel

-

real~y

-turbulent CFDmodel

mt m ml

Massflowrate

Figure 24: Expected Situation: laminar Modelling Under Predicts leakage Flow

Rate As Opposed To Turbulence Modelling.

Another unknown present with the detail key simulations is the effect of surface roughness. It is uncertain how the surface roughness would affect the outcome of the results and furthermore how to correlate the surface roughness used in the CFD with that of the actual manufactured material.

Page22

---EXPERIMENTAL METHODOLOGY

3. EXPERIMENTAL METHODOLOGY

3.1 EXPERIMENTAL PROPOSAL

The Flow Research Unit (FRU) [9] of the University of the Witwatersrand was contracted to do a detail experimental design, taking cognisance of the proposed conceptual experiments and experimental test specification put forth by the PBMR. They were also contracted to carry out all experimental work. A design specification report [10] was prepared (by the author of this document) for the FRU where various conceptual experimental designs were presented (conceptual design by author of this document).

In this chapter, various initial experimental concepts together with their shortcomings will be discussed. The final experimental design shall be elaborated on in the latter part of this chapter.

The original test specification report [10] presented to the FRU was only meant to serve as a guideline with regards to geometrical dimensions, operating conditions and even materials as a consequence of financial limitations imposed on the experiment by the PBMR. Originally the experiment had to be designed according to the actual flow conditions and geometry of the current key models. Table 4 presents the operating conditions of the keys within the reactor specified to the FRU.

Table 4: Actual Operating Conditions Of Key System In Reactor.

It was however quickly realised that the cost implications of having to execute the experiment at such a high temperature of 773.15K would be impractical and financially out of reach. It was therefore decided to perform the experimental testing at room temperature. Exact kinematic similarity could therefore not be achieved within the

Page 23

- - _{-- -}

---Operating temperature 773.15 K

Operating pressure 8.93Mpa

Reynolds number range 100

-

60000 Pressure loss range OkPa

-

250kPa

Mass flux range 6.676E-5 kg/s to 1.092E-2 kg/s

limited scale of the experiment, however; exact scaling was not required in a qualitative validation such as this [15].

The mass flow rate as a function of pressure differential was required to be extracted from the experiment. This relationship had to, originally, be determined for the following parameters:

· The exact state of the gas entering the gap.

· Material type. Specifically with steel and graphite. There are many areas within the PBMR where it is necessary to accurately predict the flow using CFD within thin steel gaps, instead of graphite gaps.

· Geometry. The actual key and gap geometry had to be replicated within the experiment as far as practically achievable.

· Surface roughness. It was anticipated that there would be a discrepancy between the expected experimental results and the CFD simulations as a consequence of the way surface roughness parameters are specified within the CFD model. It was therefore recommended that a polished and perhaps a sandblasted surface finish be tested.

It was however argued that if the experimental results could be accurately simulated for the steel case there should be no reason why similar success could not be achieved for the graphite case and therefore, in the light of practical and cost limitations it was decided to dismiss the graphite material testing.

Stringent quality control was imposed on the FRU by the PBMR. The FRU had to comply with the ISO 9000 calibration and measurement standards as specified in the PBMR Calibration and Measurement System Analysis Quality Assurance Procedure [16].

3.2 CONCEPTUAL DESIGN

The possibility of having an experiment that closely replicates the actual key arrangement within the reactor side reflector was initially investigated. It was soon realized that the biggest constraint with such an arrangement would be manufacturability. The shortest keys found within the reactor are 250 mm in height -see Table 1. The smallest gap size that exists between the key and the surrounding

Page24

---EXPERIMENTAL METHODOLOGY

graphite blocks is 175 micron. It would therefore be required that within the actual experiment this small gap be maintained throughout the height of the key which is from a manufacturing standpoint very challenging.

Another constraint is the fact that the gap sizes within the experiment were required to be measured precisely and up to now no practical way of achieving this with the above arrangement could be found. This would present a major problem if the CFD simulations did not compared with the experimental results. Would the discrepancy be due to poor geometrical measurement or a poor CFD model?

Figure 25 presents a conceptual design of the above-mentioned arrangement where the actual key within the reactor side reflector is replicated closely. The set-up consists of two inlets on either side of the pressure vessel. The pressure vessel, together with the two end caps with inlets is divided into three compartments that are separated by a top floor and a bottom floor.

Inlet End cap Steel block End cap Pressure vessel _Outlet Inlet

Figure

25: Conceptual Experimental Model That Closely Replicates The Actual Key Set-up Within The Reactor.

Page25

-The middle compartment contains the actual key with two adjacent steel blocks instead of the graphite blocks found within the reactor. Helium enters the middle compartment through the shown top and bottom floor inlets that can be seen in more detail in Figure 26 and 27. From there, the helium enters the gap formed by the adjacent steel blocks that surrounds the key, which is placed in the middle of the pressure vessel. The helium flows around the key and enters a small chamber on exiting the gap from where it finally exits through a small round outlet drilled through the pressure vessel.

A closer scrutiny of Figure 26 reveals another problem with this design. The inlet and outlet chamber are separated by the adjacent steel blocks, which apart from the gap, must be completely sealed from each other. Once assembled, there would be no practical method to establish whether any unwanted leakage occurs between the aforementioned inlet and outlet chamber, which are at different pressures during operation. Leakage between these two chambers could be minimised if the pressure vessel and steel blocks were manufactured as one component with the aid of wire erosion. This manufacturing method proved to be very costly. Welding will cause warping of the pressure vessel and steel blocks and was therefore not an option. In light of this discussion it was decided to abandon this concept.

Bottom floo~

inlet

Locating holes

Figure 26: Horizontal Cut Away Section Of Figure 25 Portraying The Helium Flow

Around The Key.

EXPERIMENTAL METHODOLOGY

Key

End cap

Inlet

Figure 27: Three Dimensional Cut Away Of The Helium Flow Path In The

Experiment.

Since it would not have been possible to design a replica of the key way in which the mass flow through the experiment could be attributed to the flow through the gap alone, the numerical validation requirement was reconsidered. It was decided to design a simplified experiment that would contain the major flow phenomenon found within the flow field of the actual key arrangement. Emphasis would be placed on measurability and adjustability of the leakage gaps within the experiment. The ability to precisely measure the gap size during experimental testing would essentially reduce the experimental error. It is important to realize that the fundamental aim and objective of the test facility is not to replicate the key geometry but rather to investigate a given flow phenomena and compare the results with those obtained from CFD simulations.

Figure 28 shows a typical example of such a simplified test arrangement. By replacing the test section with different test specimens a variety of configurations could be tested at low cost.

Flow measurement done in thin outlet pipe

Figure 28: Simplified Test Arrangement Entailing Helium Flowing Through A Thin

Slit.

This concept was however not adopted for various reasons, the first being manufacturing constraints. The test section had to be manufactured in two halfs and then bolted together which would moreover be difficult to seal properly. Another problem was the inlet helium gas jetting on the middle section of the gap. This will cause a non-uniform flow distribution through the gap, which creates another undesirable unknown.

A promising concept was the shank arrangement shown in Figure 29. The set-up consists of a cylindrically shaped shank with a tapered end section and a head section. The pressure vessel is machined with a tapered hole at one end through which the tapered shank fits. The cylindrical shank is then precisely centred within the pressure vessel by fastening the tapered shank tightly into the tapered hole with the aid of a nut at the far end of the tapered shank.

Helium enters the inlet chamber of the pressure vessel through the inlet where after it flows through the gap formed by the outer surface of the shank head and the wall of the inner surface of the pressure vessel. On exiting the gap it flows into the outlet chamber and exits through the outlet machined into the pressure vessel wall.

The hart of this arrangement is the fact that every component is manufactured on a Lathe and therefore a much higher machining accuracy could be achieved than with milling at much lower costs. The exact inlet diameter of the pressure vessel and outer diameter of the shank head at room temperature are therefore known quantities so that the exact gap size is practically known beforehand. Much more uniform surface's Page28

EXPERIMENTAL METHODOLOGY

could be achieved on a Lathe than with a milling manufacturing process.

Nut Tapered Shank Outlet chamber Outlet Shank Head Pressure Vessel Inlet

Figure 29: Shank Arrangement Manufactured Entirely On A Lathe.

The pressure within the outlet chamber and end cap chamber is equalized with the aid of small holes not shown in Figure 29 that connect the two chambers. The static pressure reading instruments could therefore be located in the end cap, which would result in a more accurate static pressure reading. Also, by equalizing the pressure between the two chambers will prevent leakage of helium through the tapered shank. Unfortunately, the shank arrangement was born after the FRU have started manufacturing of the final chosen concept discussed in the next paragraph.

The FRU proposed two experiments consisting of an initial pilot rig shown in Figure 30 and a final test shown in Figure 31. According to the FRU the pilot rig was intended for use in the validation and calibration of the experimental equipment and as part of a repeatability exercise [17]. The final rig would be used in the actual construction of flow rate versus pressure drop curves.

The pilot rig consists of two circular steel flanges in which plenums have been milled as

Page29

--shown in Figure 30. An inlet and outlet were drilled into the two plenums of the bottom flange. When assembled the inlet and outlet plenums are connected with the actual gap milled into each flange face between the plenums. The gap is 60 mm in width and 80mm long. Helium enters the inlet plenum through the inlet where after it flows through the gap and exits into the outlet plenum and then through the outlet to the atmosphere.

t

_SECTION

PARTB

Figure 30: Experimental Rig For Instrumentation And Proof Of Concept [17].

The final test rig shown in Figure 31 has an axis symmetric shape with a radial and vertical gap formed by the upper and lower inserts.

<;b240

Upper insert

holder Outlet

Upper insert

Figure 31: Final Rig For Flow Rate Measurement [17].

EXPERIMENTAL METHODOLOGY

Helium enters the set-up through the inlet and flows towards point B in the figure. From there the helium expands through the radial gap to point C where after it changes direction to flow upwards in the vertical gap to exit into the outlet chamber and then through the outlet to the atmosphere.

For a constant mass flow rate through the radial channel, the flow velocity is a function of the disk radius as a result of conservation of mass and can be expressed as the

recursivefunction

vr(n+l)

=

vr(n)(r(n)

I r(n + 1»),where

v(I)

is the velocityat radius r(1)

at the inletof the disc at pointB in Figure31 and r(n + 1)> r(n),n EN. Let

vave

bethe

average velocity through an arbitrary constant flow area gap equal in thickness to the disk flow gap of the final rig. Then, suppose that vave is the average velocity at the

average radius of the disk, then applying the recursive velocity function it can be shown that v(1)= 2vave1(1+ reI) I r(k» where

r(k)

is the outer radius of the disk. By applying the recursive velocity equation, the velocity of the disk can now be directly compared with constant area flow as shown in Figure 32. In the actual key set-up the flow area through the gap remains constant in the flow direction and therefore the velocity remains constant through the gap.

-

Diskflow

.

Diskflow

-

Constant flow area

. Constantflowarea

0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04

Distance (m)

Figure 32: Velocity Variation Of Radial Flow Path Compared With Constant Area

Flow.

It can also be shown that the pressure loss due to friction per unit distance through the radial flow path is not constant as would be the case in the actual key set-up found in the reactor, by applying the recursive Bernoulli equation [18]

across the disk flow path. The friction coefficient, c, were specified as a constant loss per unit length across the disk flow path. Figure 33 presents a graph of the friction loss through the disk compared with the loss through a constant flow area. The pressure

recovery part

(p / 2)(vr(n)

2

-

Vr(n+l)2)

of the recursive Bernoulli equation is shown in

Figure 34 for the flow through the radial gap of the final rig.

The proposed final experimental rig therefore contains flow phenomena not found within the actual key set-up. This would introduce uncertainty into the CFD models of the final rig in the sense that, if the CFD results would not correlate with the experimental results what could it be ascribed to? Would it be the inability of the CFD model to predict flow through fine gaps, or the inability to deal with the radial diffusion? It was therefore decided to abandon the final rig concept and to do experimental testing based on the pilot rig arrangement.

Page32

---EXPERIMENTAL METHODOLOGY ... (U

~

c::: .8 13 ...

'-o

-

_Q)

::3

-a

(fI (fI E !!!p ::3 (fI (fI Q) a: I I I . I I I I . I I I I I I I I I I I I I I I I I I I I I I I . I I I I I I I - .. .. .. .. .. .. .. .. .. .. .. .. -.- .. .. .. .. .. .. .. .. ..i .. .. .. .. .. .. .. .. ..i .. I I I I . I I I I I . . I I I I I I I I I . I I I I . I I I I I I I I I I . . . I I I I I I I I . I I . I I . . -, ~ r , ~ r---I I I I . I I I I I I . I I I I I I I . . I I I I . I . . I I I . . I I I . . . I I I I I I I I I I I I I I __1_.. __ __L __ J __ __1_.. L.. __.. : _: :_ : C

-

Constantflow area

.

Constant

flow area

-

Disk flow

.

Disk flow , ,, ,, , I

,---

_I I I , , , , , , I

-- -- ..,. -- -- .-

-- -- I

I I I I I I I I I I I I . I . I . I . I I . . I I I I I

-r ----

,

r

--..

I I . I I . . . I I I I I . I I , , , o , ,

:- ---:

c

0.005

0.01

0.015 0.02 0.025 Distance (m) 0.03 0.035 0.04

Figure 33: Friction Loss Of Radial Flow Path Compared With Constant Area Flow.

>-. ... Q) > o (.J Q) ... I I I I B I I . I I I I I I I I I I I I I I I I I I I . I I I I I ,

---

r

---,-I . 1 1 I . 1 I 1 I 1 1 . . I 1 I . 1 1 1 1 . I 1 1 1 1 I I I I 1 . I I I I 1 1 _'" __ - - __ __ - _1_- __ _ ___ __I. ___ _ _ '" _ _ _ __ _1__ __ _ _ I. ___ _ ____ 1 I 1 I 1 1 1 1 I 1 1 I 1 1 I I I I 1 1 I I . 1 1 I I . . I 1 1 1 . . 1 1 1 I . I I I 1 . . I . I 1 I I I I - - -.- - - -." - - - -,- - - -r - - - , - - - -.- - - - ...- - - - -r -I I I I I I I I I 1 I I I I I I I 1 1 I 1 . 1 I I 1 . 1 1 I 1 I 1 I 1 1 . . 1 I . ~ I "' ~ I._---____ . . I . 1 I 1 I I 1 I 1 I . 1 I 1 I . 1 I 1 I . 1 . . . . 1 1 . I . 1 1 1 . I I . I . I I - -1-- -- - - r- --, - - - -- -.-- - -- -- - --r - -- -- ---1 . I . I 1 . I I I I . I I . . . 1 . I I , , , , ,

-

Constantflowarea . Constant flowarea

-

Disk flow

.

Disk flow Q) ... ::3 (fI (fI Q)

a:

C

...

0.04

, ,, "'-, , I I I I I I I ,---I I I I I I I 0.005 0.01 0.015 0.02 0.025 Distance (m) 0.03 0.035

Figure 34: Pressure Recovery Of Radial Flow Path Compared With Constant Area

Flow.

3.3 EXPERIMENTAL SET-UP

Figure 35 presents a schematic of the experimental test rig. The helium source is a 20Mpa pressurized gas cylinder connected to a pressure regulator in order to maintain a constant predefined set stagnation pressure of approximately 3Mpa, which is one-third of the actual operating pressure in the reactor. This was necessary in order to obtain a reasonable steady state time of approximately 20 seconds where after the pressure in the helium bottle would fall below 3Mpa.

The static pressure reading was taken at the end of the inlet line from the bottle on entry to the test piece inlet plenum with a static pressure transducer. From the inlet plenum the helium flows through the thin gap and exits into the outlet plenum and from there out to the atmosphere via the outlet line, which was connected to a control gate valve. The function of the control gate valve was to manually set a predefined static pressure drop through the gap by altering the mass flow rate.

A differential pressure transducer was utilized to measure the difference in static pressure across the gap between the inlet and outlet plenums. Temperature readings were measured with a thermocouple. The tip of the sensor was positioned approximately in the centre of the inlet plenum. Output from the transducers was conditioned as needed and then passed to a digital oscilloscope.

Inlet plenum Gap Outlet plenum Static pressure transducer Control gate valve Helium gas bottle

Figure

35: Test Rig.

The mass flow rate measurement was the most difficult aspect of the experimental set-up since the expected mass flow rate were in the order of 1 to 10 grams per second

Page34

--EXPERIMENTAL METHODOLOGY

through the gap. The only practical method to measure such small flow rates was to actually weight the amount of helium that escaped the helium gas cylinder with time. The entire test rig together with the helium gas bottle was placed on a counter-balanced scale so that the change in mass due to venting of the gas through the test piece was measured directly with a load cell.

Figure 36 presents a schematic of the complete experimental set-up, consisting of the mass balance rig and the test rig. The mass balance rig consists of a balance beam that can pivot around the shown knife-edge. At the right end of the balance beam hangs the actual test rig bolted onto a steel mounting frame or cage with the helium gas bottle placed in a horizontal position at the bottom of the cage. The balance beam together with the test rig and gas bottle is balanced with a counter balance weight that hangs via a chain on the left end of the balance beam.

The combined mass of the gas bottle, test piece, mounting frame and transducers is of the order of 150

-

180 kg. By using a counterbalance arm meant that the load cell required for the mass measurement could have a much smaller range and thus could meet resolution requirements. The load cell was aligned with the mass centre of the mounting frame so that there was a one to one correlation between the change in mass due to venting of the gas and the change in mass measured by the load cell. Because there is no rigid linkage between the load cell and the counterbalance arm the load cell could only measure a decrease in mass.

BAlANCE WEIGHT .+1+'

~fl

_,

I

,

I

DlFF PRESS TIWISOUCER STATICPRESS TRANSOUCER View A-A ..

Figure 36: Mass Balance

Rig [19].

The entire balance arm with test rig and counter weight balances on a tool steel knife-edge during operation so that there is virtually no rotational resistance on the balance Page35

-arm and consequently, the rig responded very quickly to changes in mass. The FRU have found that the system was sensitive to the point that it registered the footfalls of the operators moving around the equipment as well as vibrations of the building.

Figure 37 shows a photograph of the complete experimental set-up. The actual test piece is shown in more detail in Figure 38. The test piece consists of two 280mm diameter mild steel flanges. Each flange contains two half-plenums of 40 x 70mm and 25mm deep that were milled into to the flange with an end mill. When assembled the two plenums are separated by a 60mm wide, 80mm long channel of 380-micron deep. The total maximum gap height of the channel is the sum of a 35-micron rebate grinded into the top flange and a 345-micron rebate grinded into the lower flange.

Figure 37: Complete Experimental Set-up.

Experimental testing was therefore first carried out with the maximum 380-micron gap height followed by the 280-micron and then the smallest 175-micron gap height. This was achieved by successively grinding down the face of the lower flange, thereby decreasing the gap height. The advantage of this technique was the fact that the actual surface finishing of the channel was maintained for the three different gap heights. The surface finish was therefore eliminated as a variable when comparing different experimental results.

Page36

-EXPERIMENTAL METHODOLOGY PRESSURE TRANSDUCER

-_

~280 '" '"

-j

OUTLET iNLET

Figure

38: Test Piece Geometry [19].

3.4 EXPERIMENTAL RESULTS

3.4.1 Data Processing

Table 5 presents the schedule of tests conducted with helium. Each test was repeated three times at the same pressure drop.

Table 5: Schedule Of Experiments With Helium [19].

Figure 39 presents a typical raw data set displaying the voltage outputs for the static pressure reading, temperature reading and differential pressure reading.

Page37

- _{-- -}

--Nominal pressure drop (kPa) 380-Micron 280-Micron 175-Micron

5 Test 28

-

30 Test 31

-

34 Test 51,54,55

75 Test 24

-

27 Test 35 - 37 Test 56 -59

150 No test data Test 38

-

41 Test 49,50,52,53

Pbmr32a

Differential Pressure Transducer

.0.50 Temperatur, .1.00 > .125 ':i .e- .1.50_::s o .1.i5 .2.25 Time (5) . 00

Static Pressure Transducer

Figure 39: Typical Raw Dataset [19].

Note the oscillations in the load cell reading, which according to the FRU can be attributed to the pendulum motion of the test piece and counterweight. The FRU has gone to considerable lengths to reduce the effect of the swinging masses. The data acquisition system acquired data at a higher sampling rate than what was required due to the minimum setting on the scope. Three-point averaging was therefore applied to the data that smoothed and reduced the higher frequency noise in the signal. The differential pressure and static pressure transducer outputs were constant and did not vary for the duration of the tests.

Data processing involved the uploading of the measured data into Microsoft Excel where the data was first truncated by discarding all data before a trigger point, which was specified as the point at which the static pressure in the inlet plenum was constant. The truncated data before the specified trigger point contained the transient effect of the opening of the control gate valve and was of no value. According to the FRU, the data was also truncated at the point where the mass plot diverged from linear where necessary, as it indicated non-steady flow or the termination of the test by closing the outlet valve. Figure 40 illustrates a set of fully processed data. Since the differential and static pressure transducer outputs were constant they were not included in the processed test data.

Page38 --- - _{- --}

-J

Probe _.._-- ..--- -- --...-...-...-...."... ...---- -- . -."..".---... I"""""'" i'\oad Cell

L-0 10 :!) 30 40 50 60 70 80

EXPERIMENTAL METHODOLOGY

Differential P = 5 kPa Static P = 3.21 MPa

Longer testing time with 5 kPa drop

0.25 Pbmr32a

~

Mass

-

Temp. Y.2.2039E-03x + 12.029E.03

~.

09888 25 0.20 20_ o CI CD ~ 15 CD_.. = .. I'!! !. 10 E ~ _ 0.15

~

II) II) 1"1I 0.10

:e

0.05 0.00 10 20 30 40 50 Time (s) 60 70 o 80

Figure 40: Typical Processed Dataset [19].

Before each set of tests, a calibration of the mass readings was performed. This was done by successively taking the voltage reading of different sets of calibrated weights placed on the mass balance rig. A kgNolt calibration factor was then determined from this data acquired for each test session. A typical best-fit mass calibration curve is shown in Figure 41.

Actual Mass vs change in voltage

5 . CallbnItionMasses -U".lfftt 6 2 y.'.40662E+OOx R> .9.99602E.01 o o 0.5 1.5 2 2.5 Mass (kg) 3 3.5 4 4.5

Figure 41: Typical Mass Calibration Curve [19].

Temperature readings with their corresponding voltages were acquired at the start and end of each experiment by using the digital readout of the thermocouple. A similar

Page39

/

./

/

/

./

/

./ ,/' I

:£

/'

procedure was followed as with the mass calibration factor, to determine a calibration factor for the temperature readings. A three-point average was applied to the mass balance data since the oscilloscope had a minimum sampling rate of 5kHz. The acquired voltage data of the oscilloscope was then converted to a mass reading and a linear best fit applied. The slope of the best fit line was the mass flow rate through the test piece, while the y intercept of the line was the zero offset of the mass balance. A similar procedure was applied to the temperature readings where the data was multiplied by a calibration factor.

3.4.2 Geometry

Geometrical measurements were taken with calibrated callipers and dial gauges. The surface roughness of the test piece gap was measured using a Talysurf.

As mentioned before the total gap height of the test piece is the combination of the depths of the rebates in each flange. The rebate depth in each flange was measured with a dial gauge. The sum of the two measured values yielded the total gap height and was verified by placing plastic gauges within the test piece. bolting up the test piece and then re-opening the test piece. The plastic gauges were then compared to their corresponding calibration charts.

According the FRU all the plastic gauges agreed with the total dial gage reading and there was no indication of significant variation of channel height along the length and width of the gap measured within the resolution of the calibration chart. Three plastic gauges were placed along the length of the gap at equal spacing. Geometrical measurements are presented in Figure 42 and Figure 43.

Page40