1. Impurity transport characterisation JET operational scenariosin
JET operational scenarios
C. Giroud, C. Angioni, H. Nordman, D. Strintzi, M. Barruzzo, M. Brix, P. Burratti, C. Challis, P. de Vries, J. Hobirk, E. Joffrin, X. Litaudon, A. Meigs, J. Mailloux, M. O'Mullane, V. Naulin, P. Mantica, M. Valisa.
Experimental data collected from 3 dedicated sessions, and parasitic experiments to scenario development sessions in 2007 and 2009.

2. Motivation
Transport of impurities was probed over a wide range of scenarios
Assess where the transport of impurity is neoclassical ? Is there a radial dependence ? An impurity dependence ?
How does Dimp/ceff vary ?
Characterisation of the impurity peaking in JET scenarios
Detailed comparison of impurity transport of L-mode discharges with theoretical models (covered in Emiliano’s talk)
No pedestal impurity transport in this work
Experimental data collected from 3 dedicated sessions, and parasitic experiments to scenario development sessions in 2007 and 2009.

3. Database of impurity transport
ne(r/a~0.2)
<ne> r/a<1.0
In green: discharges discussed in talk
Reduce error estimation on DZ and VZ by testing the analysis method over a wide range of experimental conditions.
NBI heated discharges are investigated and no Internal Transport Barrier discharges are presented. (data available but not presented here)
MHD has been observed (NTMs) in all discharges, apart from L-mode discharges.
Discharges with weakly ST or ST free ( q0 uncertainty ~ 0.2)

4. Impurity diffusion: neoclassical?
Symbol
open neoclassical
full measured
NBI heated discharges only
At r>0.6, impurity diffusion is 2 order of magnitude above neoclassical value for all probed scenarios.
Region of impurity diffusion reduction varies between scenarios and can be very central r/a<0.2 to r/a<0.4.
At r>0.6, diffusion of impurity is Z independent.
 Look into more details in scenario dependence

5. Impurity diffusion: neoclassical?Ar Ni
Dimp/Dneo
L-mode:
reduced diffusion within r/a<0.2, Dimp/Dneo~10 for all Z
H-mode: ( q0~2, H98(y,2)~0.6!)
reduced diffusion within r/a<0.2, Dimp~Dneo but is not characteristic behaviour, and likely to be influenced by presence of MHD.

6. Impurity diffusion: neoclassical?Ar Ni
Dimp/Dneo
r/a
r/a
r/a
L-mode:
reduced diffusion within r/a<0.2, Dimp/Dneo~10 for all Z
H-mode: ( q0~2, H98(y,2)~0.6)
reduced diffusion within r/a<0.2
Dimp~Dneo but not characteristic behaviour MHD presents.
Hybrid 2007: ( q0~1, H98(y,2)~ , 3/2NTM)
reduced diffusion within r/a<0.4, Dimp~10 Dneo
at r/a~0.6, Dimp~40-50 Dneo
-Add error bars

7. Impurity diffusion: neoclassical?Ar Ni
Dimp/Dneo
2/1 NTM
r/a
r/a
r/a
L-mode:
reduced diffusion within r/a<0.2, Dimp/Dneo~10 for all Z
H-mode: ( q0~2, H98(y,2)~0.6)
reduced diffusion within r/a<0.2
Dimp~Dneo but not characteristic behaviour MHD presents.
Hybrid 2007: ( q0~1, H98(y,2)~ , 3/2NTM)
reduced diffusion within r/a<0.4, Dimp~10 Dneo
at r/a~0.6, Dimp~40-50 Dneo
Steady-state 2007
reduced diffusion within r/a<0.4, Dimp~ 2-3 Dneo (Ar/Ne)
at r/a~0.6, Dimp~20-40xDneo.
Bt=1.7 Ip=1.2MA
Bt=1.7 Ip=1.2MA

8. Impurity diffusion: neoclassical?Ar Ni
3/2 NTM
Dimp/Dneo
3/2 NTM
r/a
r/a
r/a
L-mode:
reduced diffusion within r/a<0.2, Dimp/Dneo~10 for all Z
H-mode: ( q0~2, H98(y,2)~0.6)
reduced diffusion within r/a<0.2
Dimp~Dneo but not characteristic behaviour MHD presents.
Hybrid 2007: ( q0~1, H98(y,2)~ , 3/2NTM)
reduced diffusion within r/a<0.4, Dimp~10 Dneo
at r/a~0.6, Dimp~40-50 Dneo
Steady-state 2007
reduced diffusion within r/a<0.4, Dimp~ 2-3 Dneo (Ar/Ne)
at r/a~0.6, Dimp~20-40xDneo.
Hybrid 2009 H98(y,2)~1.37, 4/3 or 5/4 NTM
reduced diffusion within r/a<0.4, but variation for shot without 3/2 NTM
r/a<0.2 Dimp~ 2-4 Dneo (Ar/Ne/Ni)
r/a~0.35 Dimp~4-10 Dneo (Ar/Ne)

9. Impurity diffusion
Neoclassical diffusion: mostly in high confinement hybrid discharges within r/a<0.2 where Dimp~2-4 Dneo for all impurity.
Region of reduced diffusion can vary from r/a<0.2 ( L-mode, “H-mode”) to r/a<0.4 for hybrid, SS.
The diffusion is mainly due to turbulence across the plasma radius
how does Dimp compared to ceff?

10. Dimp vs ceff Dimp ceff difficult to determine accurately in r/a~0.15
At r/a~0.6, on average Dimp/ceff ~1. Some doubt on discharges with Dimp/ceff~0.4.
At r/a~0.35, Dimp/ceff ~0.4 for hybrid low d (no 3/2NTM), Dimp/ceff~1. for L-mode and H-mode.
Similar to finding in DIII-D where DHe/ceff ~0.5 in core of VH mode

11. Influence of neoclassical pinch ?
r/a~0.15 Ne Ar Ni
Due to MHD (mainly NTM at mid-radius), a comparison between neoclassical value and determined convection is only meaningful in the plasma core r/a<0.2.
In the core (r/a~0.15), Vmeas is mostly of the same sign as Vneo, and vary on average from neoclassical level to within a factor ~4. Vmeas is of the same order as Vneo.
We do not observe the simple rule V=Vneo for high Z.

12. Global impurity peaking
Steady-state profile are calculated from the determined transport coefficient with UTC-SANCO assuming a constant source.
C-profile always less peaked than the electron density and is hollow for electron peaking below 1.3.
Ne peaking is similar to C but can become more peaked for the electron density above 1.3.
Ar and Ni are generally more peaked than the electron with a discrepancy that seems to increase at high electron density peaking.
 what plasma region determine the global peaking ?

13. Global impurity peaking
The Z dependence of global peaking is the result of impurity peaking in central region (r/a<0.4).

14. Imp. convection and global peaking
In the core (r/a~0.15), Vmeas is mostly of the same sign as Vneo, and vary on average from neoclassical level to within a factor ~4.
On average, magnitude of RV is greater than D. As Z increases, magnitude of RV increases in respect to D  V plays an important role in central peaking (can be influenced by Vneo).
The Z dependence of global peaking is the result of impurity peaking in central region (r/a< ).
 Comparison experimental results of impurity transport with predictions made with Weiland models and linear GK GS2 for L-mode discharges (no MHD).

15. Conclusion
The transport of impurity is rarely neoclassical even in the central region (r/a<0.2)
Dimp~10x Dneo.
one exception: some low d hybrid discharge where Dimp~2-4 Dneo.
Dimp/ceff:
at r/a~0.6, Dimp/ceff~1
at r/a~0.4, Dimp/ceff~0.4 for hybrid discharges at low d
Convection (r/a<0.2)
Vmeas is on average of the same sign as Vneo, and comparable in magnitude.
magnitude of RV is greater than D (more so high Z)  V plays important role in peaking.
Observed Z dependence of global peaking is the result of impurity peaking in central region. (results from a neo and anomalous contributions).
Anomalous transport predictions (L-mode) Linear GK and fluid model are in good agreement with experimental impurity peaking with exception of C : this remains an open question ?

16. What to address in C28C29 ?
No dedicated experiment on testing exp/theory in C28 C29:
 knowledge of impurity transport in support to scenario development mainly
Type of discharges to be
Ohmic
ST L-mode
ST H-mode 2.5MA
Hybrid high d/low d at 1.7T/1.5MA and then increase Ip
What problem could be encountered from W point of view ?
Ohmic: not much expected ?
ST L-mode: no problem expected ?
ST H-mode: pedestal (elms) + ST activity + W influx ?
Mainly pedestal, source and MHD activity not so much addressing transport due to turbulence.
More effect of MHD to be expected than shown in this talk !

17. What to address in C28C29 ? Tools
At some stage during the development need additional information to go forward
Ar/Ne, Ni injection maybe Af: probe the transport to have a better handle on the problem
how often can we realistically be expected to do that ? How useful would it be ?
Tools
SANCO with W or STRAHL ?
Reliable W atomic data : how long would that need ?
present W data
future development of W data
ionisation, recombinaison, SXR, bolometry
cW at edge and core output in PPF
influx output in PPF

18. What to address in C30b ?
Dedicated experiment on experiment / theory impurity transport:
He, Be, B: low Z end to be tested
high Z end with W
effect of rotation and ICRH

20. Impurity peaking in central region
Symbol
open V neoclassical
full measured Ne Ar Ni
On average, magnitude of RV is greater than D
As Z increases, magnitude of RV increases in respect to D.
Convection plays an important role in the central peaking (and as Z increases). Neoclassical convection is comparable to Vmeas and can have an influence.

21. Anomalous transport: D and V
Dimp
Vimp
Anomalous transport calculated with Weiland fluid models for ITG/TE modes.(Nordman, EPS 2010)
Predicted Dimp and Vimp are the same for Ne, Ar, Ni.
At mid-radius, the predicted D and V are in reasonable agreement for Ne, Ar and Ni.
Dmeas and Vmeas in core not predicted by Weiland model as modes are found stable within r/a<0.4.
Dmeas/Chi_eff~ / Weiland model~  GS2 comparison on-going.
Nordman: chi_i_eff that he is showing not chi_i

22. Anomalous transport: peaking
Additional prediction of anomalous peaking with linear GK GS2 (Angioni, Strintzi).
Agreement between Linear GK code and Weiland model prediction on peaking
Within errors, the predicted peaking is in agreement with experiment apart from Carbon.
Carbon is predicted peaked whereas it is measured flat or hollow in the experiments in L-modes at mid-radius  thermodiffusion higher than predicted in codes.
(Giroud IAEA 2006)