Date post: | 17-Jan-2016 |
Category: |
Documents |
Upload: | abraham-price |
View: | 217 times |
Download: | 0 times |
11 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
A Statistical Analysis of A Statistical Analysis of
Root Causes of Disruptions at JETRoot Causes of Disruptions at JET
Peter de Vries, Mike Johnson, Barry Alper
and JET EFDA Collaborators
IEA Disruption Workshop (W70)
Culham 8 October 2009
22 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
MotivationMotivation A disruption of a Tokamak discharge often induces large
forces on the surrounding structure and large heat loads on in-vessel components. It is therefore important to prevent or mitigate these events, especially in large devices as ITER
– Hence: Study the causes and consequences of disruptions
In order to find ways to prevent or mitigate disruptions a detailed picture of all the precursors or causes is necessary.
– What are the main causes of disruptions in JET?– What are the characteristics and ‘precursor signs’?
This information may help us to work out better ways to avoid or mitigate disruptions in JET and possibly ITER.
33 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Statistics: Disruption rateStatistics: Disruption rate Disruption rate1: the fraction of discharges that disrupt?
– Variation but also signs of a downward trend:
Period All (Uninten.)
<1987: 24% (22%)
1987-1992: 21% (19%)
1992-1996: 31% (28%)
1996-1998: 21% (19%)
1998-2001: 19% (17%)
2001-2004: 12% (8%)
2004-2007: 8% (6%)
2008-2009: 5.5% (3.4%)
[1] P.C. de Vries, Nucl. Fusion 49 (2009) 055011
44 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Statistics: PTN triggersStatistics: PTN triggers A simple protection system (PTN) at JET is able to detect
disruptive plasmas, terminate the pulse, mitigating Forces1. – 49% of all unintentional disruptions with a warning time of 200ms or more
– 66% of all unintentional disruptions with a warning time of 30ms or more
– 25% of all unintentional disruptions is not detected at all
[1] P.C. de Vries, Nucl. Fusion 49 (2009) 055011
55 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Causes of disruptionsCauses of disruptions Many causes of disruptions have been studied for many
years and the basic classes are ‘known’. However, these studies often focus on special examples or the detailed physics behind the disruptions and do not often deal with more complex cases or those with technical causes.
At JET a dedicated disruption database exist:– Record of all disruptions since 1985 but no information on causes
– A detailed analysis of the causes of all disruptions with Ip>1MA between 2000 and 2007 (= 1707 cases) has been carried out.
– With the aim of building a realistic/complete picture of chain of events leading to disruptions at JET
66 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
What has been done?What has been done? All 1707 disruptions have been analysed manually by:
– Checking the session leader pulse schedule (XPSEDIT).– Reading main session and pulse information:
• DCO display overview for a session overview• Session Leader JOTTER comments (session + pulse)• Other technical information, meeting minutes, etc.
– Checking if all auxiliary heating systems worked as requested.– Analysing plasma control systems:
• PPCC and shape control (XLOC, EFIT, KL1 video, …)• Density, Gas and other RT control
– Doing basic signal analysis: • MHD: n=1, n=2 and locked mode, Radiation and impurity levels, …
– Carrying out more detailed analysis:• Detailed MHD analysis (Kink modes, NTMs,…), MARFEs, etc.
Each step towards the disruption labelled…
77 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
What labels?
Physics problems
Technical problems
88 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Classification of disruptionsClassification of disruptions Although in some cases a clear unambiguous cause can be
identified, the process that leads to a disruption is often:– Complex, with several problems occurring at the same time– Disruption happen very fast and are not always well diagnosed
Thus the analysis and the resulting classification may be subjective, depending on who did it and which diagnostics were available.
– But, this analysis is done for many disruptions, and we can build a statistical picture.
Disruptions can be classified according to the root cause or the specific path or chain of events that lead to the disruption.
99 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
What can be studied?What can be studied? Chain of events that leads to the disruption.
– Example: SC WAL IMP RC MHD ML DISR– A flow chart of all events can be built
Statistics of causes and links between certain events.– How often is the root cause a human mistake?– How often does an SC error cause a RC?– What are the main ‘chain of events’ or classes that occur in JET?
Details on specific classes of disruptions– How fast is the chain of events after an NTM is triggered?– What is the operational space in which a class occurs?– Can we improve the detection?
1010 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Chain of Events (Unintentional)Chain of Events (Unintentional)
MAR
LOQ
IMP
RC
MHD
ML
VDE
GIM
Flow Chart of 417 Intentional JET Disruptions from 2000 to 2007
NTM
VSK
IMC
HUM
NC
1111 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
MAR
IMC
UFO
LHC
ICH
HD
ROT
VST
LOQ
GWLNPK
HUM
AUX
IP
RECMSH
QED
PRPKNK
2ST
NC
IMP
HL
RC
MHD
ML
VDESTOP
WAL
ITB
LON
DIV
NBI
SC
GIM
RCY
Flow Chart of 1283 Unintentional JET Disruptions from 2000 to 2007
NTM
1212 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Flow Chart of 1283 Unintentional JET Disruptions from 2000 to 2007
MAR
IMC
UFO
LHC
ICH
HD
ROT
VST
LOQ
GWLNPK
HUM
AUX
IP
RECMSH
QED
PRPKNK
2ST
NC
IMP
HL
RC
MHD
ML
VDESTOP
WAL
ITB
LON
DIV
NBI
SC
GIM
RCY
NTM
Disruption Process
1313 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Flow Chart of 1283 Unintentional JET Disruptions from 2000 to 2007
MAR
IMC
UFO
LHC
ICH
HD
ROT
VST
LOQ
GWLNPK
HUM
AUX
IP
RECMSH
QED
PRPKNK
2ST
NC
IMP
HL
RC
MHD
ML
VDESTOP
WAL
ITB
LON
DIV
NBI
SC
GIM
RCY
NTM
Disruption Process
1414 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
1
10
100
1000D
IAV
SK
LIB
MA
RR
CY
SL
VD
EV
ST
LO
QM
LP
SA
BL
PE
LP
RO
RT
CA
UX
RC
,E
FC
MA
GN
ON
PD
VV
SV
SA
QE
DU
FO
ELM
NB
IIC
HR
MP
MH
DIM
PG
WM
SH
IPW
AL
IMC
LH
CLO
ND
IVIT
BN
CS
CH
DH
UM
NT
M
Root Cause
Eve
nts
Statistics of Root Causes Statistics of Root Causes
Main root causes of Unintentional Disruptions (67%):
1. Neo-classical Tearing Modes (NTM) 17%
2. Human error (HUM) 8.3%
3. High density operation (HD) 6.5%
4. Shape control problems (SC) 6.4%
5. Density control problems (NC) 6.2%
6. Internal Transport Barrier (ITB) 5.9%
7. No divertor cryo-pumping (DIV) 5.1%
8. Low density error field mode (LON) 4.0%
9. Lower Hybrid Current Drive (LHC) 3.9%
10. Impurity control problems (IMC) 3.4%
1515 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Statistics of Root CausesStatistics of Root Causes No or unclear root cause: 3%
– For 5.5% the analysis is uncertain
Human error: 8.3%
Real time control problem: 16.5%– SC, NC, IMC, RTC
Wall and impurity issue: 7.1%– WAL, IMP, UFO, RCY,MAR
Technical root cause: 19.3%– DIV, LHC, NBI, ICH, RMP (STOP)
Physics root cause: 45.2%– NTM, ITB, RC, LOQ, QED, GWL/HD, …
1616 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
CharacterisationCharacterisation Disruption classes defined either by root cause or the chain
of events.– Characterise the precursors for each class
• Best to detect the root or otherwise next steps in the chain of events
• What are the typical warning times from for example PTN
– Characterise the consequences (forces/current quench, heat loads)
Find methods to prevent or mitigate specific classes– Focus attention on those which are most relevant– What is the best action to be taken to mitigate the effects?– Which precursors could be used to detect specific classes?
1717 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
MAR
IMC
UFO
LHC
ICH
HD
VST
LOQ
GWLNPK
HUM
AUX
IP
RECMSH
QED
PRPKNK
2ST
NC
IMP
HL
RC
MHD
ML
VDESTOP
WAL
ITB
LON
DIV
NBI
SC
GIM
RCY
Flow Chart of 1283 Unintentional JET Disruptions from 2000 to 2007
NTM ROT
1818 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
NTM triggered disruptionsNTM triggered disruptions Of all disruptions with as root cause and NTM
– 51% disrupted due to pure mode locking• Pure ICRH discharges with large sawteeth, NBI discharges at high N
– 34% disrupted as a consequence of the termination sequence• Mainly wall interaction leading to low q or high radiation
• Looks bad but the termination sequence will have a mitigating effect
– 10% disrupted due to a fast shut-down of VS (FRFA temperature)• This problem has been solved at JET
– 2% disrupted due to density pump-out and an error field-mode
Detection: – These disruptions have a clear precursor that can be detected well
in advance of the disruption– 34-44% disrupted during the fast-stop or emergency termination.
1919 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
0.0
0.2
0.4
0.6
0.8
1.0
0.01 0.10 1.00 10.00tDISRUPTION - tPRECURSOR
Acc
umul
ated
fra
ctio
n
NTM triggered disruptionsNTM triggered disruptions Warning time from precursors
– n=1 or n=2 MHD mode, Locked Mode (ML)– A larger fraction can be detected compared to average disruption statistics
and also earlier: 90% more than 1s before tDISR
– Improve detection and mitigation technique for these modes at JET
ALL ONLY NTM
PTN
ML
NTM
Trigger
2020 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
MAR
IMC
UFO
LHC
ICH
HD
ROT
VST
LOQ
GWLNPK
HUM
AUX
IP
RECMSH
QED
PRPKNK
2ST
NC
IMP
HL
RC
MHD
ML
VDESTOP
WAL
ITB
LON
DIV
NBI
SC
GIM
RCY
Flow Chart of 1283 Unintentional JET Disruptions from 2000 to 2007
NTM
2121 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Statistics on Human ErrorsStatistics on Human Errors Breakdown of human mistakes
– Density control: 35%– Impurity control: 26%– Too low density: 11%– Shape controller request: 11%– Too low q: 6%– NBI timing: 3%– …
Who is to blame– Session Leader: 89%– Gas Matrix set-up: 9%– Engineer-in-Charge: 1% – Laser ablation: 1%
Prevention methods– Better preparation, improved interfaces and operation procedures, ...
2222 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
MAR
IMC
UFO
LHC
ICH
HD
ROT
VST
LOQ
GWLNPK
HUM
AUX
IP
RECMSH
QED
PRPKNK
2ST
NC
IMP
HL
RC
MHD
ML
VDESTOP
WAL
ITB
LON
DIV
NBI
SC
GIM
RCY
Flow Chart of 1283 Unintentional JET Disruptions from 2000 to 2007
NTM
2323 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
0.0
0.2
0.4
0.6
0.8
1.0
0.01 0.10 1.00 10.00
tDISRUPTION-tPRECURSOR
Acc
um
ula
ted
Fra
ctio
n
Disruptions at Greenwald limitDisruptions at Greenwald limit Not in the top 10 of causes at JET
– Characterised by a H to L back transition prior to disruption• ELMs stop, density drops but also a drop in edge temperature
– No clear mode lock, well before the disruption– Detection of H-L transition + new ideas for preventive actions
ALL ONLY GWL
HL
PTN
2424 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
MAR
IMC
UFO
LHC
ICH
HD
ROT
VST
LOQ
GWLNPK
HUM
AUX
IP
RECMSH
QED
PRPKNK
2ST
NC
IMP
HL
RC
MHD
ML
VDESTOP
WAL
ITB
LON
DIV
NBI
SC
GIM
RCY
Flow Chart of 1283 Unintentional JET Disruptions from 2000 to 2007
NTM
2525 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Overview Top 10Overview Top 10 Classed by Root Cause %
unint. <F/Ip
2> (MN/MA2)
tPTN (s)
% tPTN >0.04s
Improve prevention or mitigation at JET
1 NTM 17 0.12 0.4-1 86 Earlier detection of NTM
2 Human error 8.3 0.20 0.04-0.1 69 Improve preparation & procedures
3 Power switch-off at high density 6.5 0.31 0.01-0.02 20 Improve scenario design & detect too high density / radiation
4 Shape control error 6.4 0.18 0.2-0.4 50 Improve shape control & detect high recycling
5 Density control error 6.2 0.22 0.04-0.1 42 Improve density control & detect high radiation
6 Strong ITB 5.9 0.21 <0.001 5.2 Detect strong ITBs
7 No divertor cro-pumping 5.1 0.21 0.1-0.2 71 Detect high density / radiation
8 Low density error field mode 4.0 0.12 0.4-1 90 Improve scenario design
9 LHCD arc 3.9 0.13 0.1-0.2 83 Detect Iron influx / high radiation
10 Impurity control error 3.4 0.19 0.02-0.04 29 Detect high radiation / Impurities
Greenwald limit 2.6 0.37 <0.001 13 Detect HL transition / high radiation
Intentionally kicked VDE - 0.40 <0.001 3 -
2626 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Overview Top 10Overview Top 10 Classed by Root Cause %
unint. <F/Ip
2> (MN/MA2)
tPTN (s)
% tPTN >0.04s
Improve prevention or mitigation at JET
1 NTM 17 0.12 0.4-1 86 Earlier detection of NTM
2 Human error 8.3 0.20 0.04-0.1 69 Improve preparation & procedures
3 Power switch-off at high density 6.5 0.31 0.01-0.02 20 Improve scenario design & detect too high density / radiation
4 Shape control error 6.4 0.18 0.2-0.4 50 Improve shape control & detect high recycling
5 Density control error 6.2 0.22 0.04-0.1 42 Improve density control & detect high radiation
6 Strong ITB 5.9 0.21 <0.001 5.2 Detect strong ITBs
7 No divertor cro-pumping 5.1 0.21 0.1-0.2 71 Detect high density / radiation
8 Low density error field mode 4.0 0.12 0.4-1 90 Improve scenario design
9 LHCD arc 3.9 0.13 0.1-0.2 83 Detect Iron influx / high radiation
10 Impurity control error 3.4 0.19 0.02-0.04 29 Detect high radiation / Impurities
Greenwald limit 2.6 0.37 <0.001 13 Detect HL transition / high radiation
Intentionally kicked VDE - 0.40 <0.001 3 -
2727 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Overview Top 10Overview Top 10 Classed by Root Cause %
unint. <F/Ip
2> (MN/MA2)
tPTN (s)
% tPTN >0.04s
Improve prevention or mitigation at JET
1 NTM 17 0.12 0.4-1 86 Earlier detection of NTM
2 Human error 8.3 0.20 0.04-0.1 69 Improve preparation & procedures
3 Power switch-off at high density 6.5 0.31 0.01-0.02 20 Improve scenario design & detect too high density / radiation
4 Shape control error 6.4 0.18 0.2-0.4 50 Improve shape control & detect high recycling
5 Density control error 6.2 0.22 0.04-0.1 42 Improve density control & detect high radiation
6 Strong ITB 5.9 0.21 <0.001 5.2 Detect strong ITBs
7 No divertor cro-pumping 5.1 0.21 0.1-0.2 71 Detect high density / radiation
8 Low density error field mode 4.0 0.12 0.4-1 90 Improve scenario design
9 LHCD arc 3.9 0.13 0.1-0.2 83 Detect Iron influx / high radiation
10 Impurity control error 3.4 0.19 0.02-0.04 29 Detect high radiation / Impurities
Greenwald limit 2.6 0.37 <0.001 13 Detect HL transition / high radiation
Intentionally kicked VDE - 0.40 <0.001 3 -
2828 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Current quench timeCurrent quench time Statistics on the current quench time1,2.
0
2
4
6
8
10
12
14
16
18
20
0 200 400 600 800 1000 1200
Ip/S (kA/m2)
t 20-8
0/S
(m
s/m
2 )
0.0
0.1
0.2
0.3
0.4
0.5
0 2 4 6 8 10 12 14 16 18 20
t20-80/S (ms/m2)
Per
cen
tag
e Overall statistics (All disruptions)
[1] J. Wesley, FEC (2006 IAEA) IT/P1-21
[2] V. Riccardo, Plasma Phys. Control. Fusion 47 (2005) 117–129
2929 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Current quench timeCurrent quench time Statistics on the current quench time1,2.
– ITB, kicked VDEs and other clean disruptions have a fast quench
0
2
4
6
8
10
12
14
16
18
20
0 200 400 600 800 1000 1200
Ip/S (kA/m2)
t 20-8
0/S
(m
s/m
2 )
0.0
0.1
0.2
0.3
0.4
0.5
0 2 4 6 8 10 12 14 16 18 20
t20-80/S (ms/m2)
Per
cen
tag
e
Intentional VDEs
Overall statistics (All disruptions)
Disruption due to strong ITBs
[1] J. Wesley, FEC (2006 IAEA) IT/P1-21
[2] V. Riccardo, Plasma Phys. Control. Fusion 47 (2005) 117–129
3030 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Current quench timeCurrent quench time Statistics on the current quench time1,2.
– Other classes show significantly slower quench times
0
2
4
6
8
10
12
14
16
18
20
0 200 400 600 800 1000 1200
Ip/S (kA/m2)
t 20-8
0/S
(m
s/m
2 )
0.0
0.1
0.2
0.3
0.4
0.5
0 2 4 6 8 10 12 14 16 18 20
t20-80/S (ms/m2)
Per
cen
tag
e Overall statistics (All disruptions)
Disruptions due to impurity control errors
Current rise and low li disruptions
[1] J. Wesley, FEC (2006 IAEA) IT/P1-21
[2] V. Riccardo, Plasma Phys. Control. Fusion 47 (2005) 117–129
3131 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Relation other devices and ITERRelation other devices and ITER The statistics of causes and classes at other devices and
ITER may be different because:– JET differs technically from these devices or ITER
• Some problems are typical to JET (e.g. the VS n=2 interference)
• Failure rates differ for VS or density control systems or other subsystems such as auxiliary heating
– JET does not operate in the exact same operational domain as other devices and especially ITER
• Operation at the Greenwald density and with high radiation fractions
• Operation at high N , with large Sawteeth prone to trigger NTMs.
What will change when JET starts operating with the new ITER-like wall (i.e. W divertor + Be limiters)?
Thus, an comparison study with other devices would be very interesting.
3232 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Disruption Prevention and MitigationDisruption Prevention and Mitigation Focus on most relevant causes of disruptions but look into
realistic chain of events– How do we handle the large fraction of disruptions caused by power
switch-off and density/impurity control problems?
Characterisation of all disruption classes may lead to better overall understanding and more focussed prevention and mitigation methods
How to tackle disruption prevention/mitigation?– Limit the chance of mistakes by the operator (human errors)– Prevent disruptions by scenario development– Be aware that disruptions may be caused by simultaneous problems– Be aware of the failure rate of subsystems– Built a robust detection system with tailored response/actuators
3333 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Reserve SlidesReserve Slides
3434 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
0.0
0.2
0.4
0.6
0.8
1.0
0.01 0.10 1.00 10.00tDISRUPTION - tPRECURSOR
Acc
umul
ated
fra
ctio
n
ITB triggered disruptionsITB triggered disruptions Very fast ITB triggered disruptions show no warning
– Of the 76 cases only 7 tripped the termination network PTN– Often due to a pressure driven internal kink mode– At JET prevented by limiting the ‘strength of the ITB’
ALL ONLY ITBs
PTN
PTN
3535 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Normalised ForcesNormalised Forces The normalised forces differ per root-cause of class of
disruption – Can be attributed to the triggering of the disruption avoidance
scheme (PTN) at JET by the respective precursors
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.0 0.2 0.4 0.6 0.8 1.0Normalised Force F/Ip
2 [MN/MA2]
%
All disruptionsKicked VDE's onlyImpurity control errorsToo strong ITBLow density locked modesCaused by NTMs
78% are VDEs
3636 Root Causes of Disruptions at JET – IEA Disruption Workshop 2009 – Peter de Vries
Example of labellingExample of labelling This process has its limitations as some cause of events are rather complex. Example #69617 (predominant ICRH)
– 1) NTM ML RMP SC WAL IMP RC– 2) NTM ML RMP SC LOQ
NTM
ML triggered fast-stop t=52.199s
Sawtooth crash
Locking
52.1s 52.2s 52.3s 52.6s 52.7s