Structural Mechanics Modeling For GTRF

R. Ghelichi and D. Parks

MIT, Dept. of Mechanical Engineering

April, 2015

Introduction

Fretting wear removes material from the cladding surface and can eventuallyperforate the cladding (fuel failure).

(Kim, et al., NED, 2008) (Kim, et al., NED, 2009)

Introduction

How best to document and demon-strate the emerging capabilities ofadvanced system modeling?Key Structural Drivers of GTRF

1. Flow-induced vibration

strength

o source; stiffness

2. In-service relaxation of

clamping force facilitates slip;

gap opening; impact/sliding

o Thermal & irradiationcreep

o Irradiation growth

Approach

Full rod structuraldynamic simulation:

Input Stiffness of the springs

and Dimples

CFD pressure loadhistory

Gap size (irradiationcreep + wear) and/orpreloads;

Rod conditions: creepdown

Grid condition: growth

pellet/clad contacts;gaps

Output History of contact

forces, slip distances,work rates, etc.

Single cell grid, 3Dgrid/rod contactinteraction andwearOutput

History of contactforces, slipdistances, workrates.

Enables updating ofsurface geometrybased on evolvingscar profiles thatmodifies the gaps.

Input Gap size (irradiation

creep + wear)and/or preloads

Initial wear depths

Time variation ofrod moment and/orrotation; transversedisplacement orforce

Spring/Dimple Normal Stiffness

Loadings and Boundary conditions:

Surfaces at A are fixed in all direction as they are connected to theother plates

Surfaces at B are fixed in all directions;

Displacement applied on RP for each side of dimples or spring.

Spring/Dimple Normal StiffnessD

imp

leS

prin

g

Without Interaction With Interaction

Spring/Dimple Normal Stiffness

Table : Calculated stiffness in different sets of simulation

Spring Dimple1 Dimple21T 2T 1T 2T 1T 2T

K(N/mm) 60 86 235 292 412 486

Spring/Dimple Normal Stiffness

Spring/Dimple Normal Stiffness

int1.aviMedia File (video/avi)

Spring/Dimple Normal Stiffness

Table : Calculated stiffness forFyUy

M2 M1 M31T 2T 1T 2T 1T 2T

K(N/mm) 59 87 213 269 388 465

Beam Model Simulation

Beam Model:

o Element: B31

o Contact throughconnector elements

Beam Model Simulation

A Python code controls the whole simulation. Starting from thefollowing inputs:

Dynamic Implicit Simulation

Reliable results Expensive Hard to control the time step

Modal Simulation

Very Fast Results converge to Implicit

dynamics

Fix time step

CFD Pressure load

{qx(z, t)qy(z, t)

}=

20

P (, z, t)

{cos()sin()

}Rd

{Fx(t)Fy(t)

}=

L10

{qx(z, t)qy(z, t)

}dz

CFD Pressure loadDiscrete Nodal ForceConsider a given uniform-values

{HxHy

}=

{

00

}if : f < 1 or f > 2{

AxAy

}if : 1 f 2

By generating a random phase (f) on H(f):{Fx(f)

Fy(f)

}=

{HxHy

}exp i(f)

The fluid load P (f) is discretized with equally-spaced increments f and thenby keeping the real part of Inverse Fourier Transformed of P (f):{

Fx(t)Fy(t)

}=

{AxAy

}f

N1j=1

cos((fj) 2fit)

Ax(y) =

2 RMS2x(y)f(2 1)

Wear Calculation

Qt = FN

Wdg =N

i=1Qti

Archard Law:

Vdebris = KWdg

K is the wear coeffi-cient.

More accurate calculation for dissipatedfrictional work can be done based on thefollowing graph:

S. Fouvry et al,, Wear, Volume 185, 1995, P. 35-46

Wear Calculation

Find the worst grids/contact location based on the wear friction work:In the beam simulation there are 6 (dimples/springs) 7 (number of spans)connectors to be checked. Each of them has its own normal and tangentialdisplacement (based on the position of the spring or dimple with respect to thebeam).

The dashline shows the position of the gap in this grid.

Wear Calculation

By extracting this information, it is possible to reduce considerably the time ofthe simulation.

Dissipated work: Wdg =Ni=1 FNii

Preliminary results

The maximum value of Wdg for Gap=10m is mostly in the range of:

Wdg = (1 105) (1 104)(J)

The simulation time is 5s so:

Wdg =dWdgdt

= (2 106) (2 105)(Watt)

Zirconium wear coefficient K has been estimated to 50 200 1015Pa1:

V = K Wdg = 1 1019 4 1018m3

s= 8200

m3

day 326400m

3

day

Considering the contact area of 1 3(mm) it will be about at maximum in theorder of:

0.11m

day

Yetisir and Fisher, Nuc. Eng., Volume 176, 1995, P. 261-271

3D-Model

800000 elements

3D-Model

800000 elements

Submodeling

90000 elements( 10% of the full model)

Drive relativemotion frommodel beamhistories

Compute 2Dcontact and slipcontours

Preliminary ResultsA simple comparison:

Kim et al(2008)

Future steps

In-depth study of modal analysis simulations for beam Surface modification in 3D-Model (wear scars) Calculate the new values for gaps as an input for Beam

simulations; account for thermal creep, irradiation in cladding& grids (with INL and UMichigan)

Multi-based modeling with coupled grid compliance Improved CFD connections to turbulent loading of rods

En Passant Cracks