Title: Fusion Reactor Steel
1Fusion Reactor Steel
- Sonny Martin Tevis Jacobs
- Yucheng Zhang Jiawen Chen
2- Fusion
- Neural networks
- Models
- Predictions and conclusions
3Fusion
4Design parameters
- Neutron damage 100-150 dpa
- Transmutation helium
- Up to 700ºC
5Irradiation Effects
- Irradiation hardening
- Irradiation creep
- Activation
- Swelling
6Irradiation Hardening
- Dislocation loops
- Condensation of defects
- Helium bubbles
7Why model?
- Suitable reactor does not exist
- Experiment would be costly
- Help design ITER
- Because were doing a modelling course!
8It cost an estimated 8 million to produce the
current data (116 million, 14 million,11
million)from FISSION reactor.
9- Fusion
- Neural networks
- Models
- Predictions and conclusions
10Neuron
Activation Function (tanh)
input
output
input
output
Biological Neuron
Digital Neuron
11Input Layer
Hidden Layer
Output Layer
12fi tanh ( Swij(1)xj ?i(1) )
y Swi(2)fi ?(2)
13OR logic
Value Input Input Output
Value X1 X2 Y
Value 0 0 0
Value 0 1 1
Value 1 0 1
Value 1 1 1
14Train by adjusting weights
Y X1? W1 X2 ? W2W3
W11, W21, W30
b
a
d
c
15(No Transcript)
16(No Transcript)
17y
x
18- Fusion
- Neural networks
- Models
- Predictions and conclusions
19Total Elongation
20Adding known science
- He/dpa
- exp-1/(Irradiation T)
- exp-1/(Test T)
- 1/L
211/L
??y 2 ??y2, dislocations ??y2, bubbles
22(No Transcript)
23(No Transcript)
24(No Transcript)
25Yield Strength
26- Fusion
- Neural networks
- Models
- Predictions and conclusions
27Yield Strength Predictions
28Uniform Elongation
29Uniform Elongation Predictions
30Measured Uniform Elongation Data
31We were not able to create a reasonable model for
uniform elongation
32- For the first time, predictions have been made of
suitability of austenitic steel for fusion
33Total Elongation Predictions
34Measured Total Elongation Data
35- For austenitic stainless steels irradiated to
doses consistent with fusion, ductility is likely
to be unacceptably small
36Yield Strength Predictions
37- Irradiation hardening makes the steel brittle
38- BCC steel might be better
39Future work
- Verify experimentally (in 15 years)
- Investigate discrepancy in uniform ductility
- Record comprehensive data when doing experiment
40Acknowledgements
- Professor Harry Bhadeshia
- Richard Kemp
- The noodle lady in Market Square