BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Q&A
DTSTART;VALUE=DATE-TIME:20210513T100500Z
DTEND;VALUE=DATE-TIME:20210513T103500Z
DTSTAMP;VALUE=DATE-TIME:20210513T221423Z
UID:indico-contribution-1772-19432@conferences.iaea.org
DESCRIPTION:https://conferences.iaea.org/event/214/contributions/19432/
LOCATION:Virtual Event
URL:https://conferences.iaea.org/event/214/contributions/19432/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Exploring the physics of a high-performance H-mode with small ELMs
and zero gas puffing in JET-ILW
DTSTART;VALUE=DATE-TIME:20210513T094800Z
DTEND;VALUE=DATE-TIME:20210513T100500Z
DTSTAMP;VALUE=DATE-TIME:20210513T221423Z
UID:indico-contribution-1772-17071@conferences.iaea.org
DESCRIPTION:Speakers: Elena De La Luna (Laboratorio Nacional de Fusión\,
CIEMAT\, 28040\, Madrid\, Spain)\nRecent experiments in JET-ILW have been
successfully exploring a high-performance H-mode scenario with no gas dosi
ng at low $q_{95}$ ($I_p=3$ MA\, $B_t$=2.8 T\, $q_{95}=$ 3.2) and low tria
ngularity\, with peak neutron rates reaching values of 3.6$\\times 10^{16}
$ s$^{-1}$. This was enabled by operation at very low gas fueling\, which
is challenging in JET with the metal wall due the need to control the W in
flux into the core region. By starting H-mode operation at high density\,
applying a high level of gas injection early during the NBI heating phase
to avoid ELM-free phases\, it was possible to reduce the gas puffing to ve
ry low levels ($\\approx 10^{21}$e/s)\, achieving a high performance\, low
density regime (called no-gas regime in the rest of the text) with averag
ed $n_e\\approx3\\times 10^{19}$ m$^{-3}$ and Greenwald fraction of ≈0.3
5\, amongst the lowest ever achieved in JET-ILW. Operation at such low den
sities allows decoupling ions from electrons\, resulting in higher $T_i/T_
e$ than what obtained in conventional ELMy H-modes at higher densities and
similar heating power.\nOne of the best examples of this no-gas scenario
(#94900) is shown in Fig. 1\, compared to the reference ELMy H-mode discha
rge (#94777) at similar heating power. Both discharges are heated by 20 MW
of neutral beam injection (NBI) and up to 4 MW of ion cyclotron resonance
heating (ICRH). In discharge #94900 the gas puffing is switched off at 8.
5 s\, resulting in a strong decrease in edge density (from 7$\\times 10^{1
9}$ m$^{-3}$ to 2$\\times 10^{19}$ m$^{-3}$) and a significant increase in
the density profile peaking. This is accompanied by an increase in pedest
al temperatures ($T_{e\,ped}\\approx$1.5 keV\, $T_{i\,ped}\\approx$2 keV)\
, enhanced toroidal rotation ($v_{tor\,0}\\approx$450 km/s)\, improved cor
e ion confinement ($T_{i\,0}\\approx$15 keV\, $T_{i\,0}\\approx 2\\times T
_{e\,0}$) and ELMs substantially smaller and faster than those of the refe
rence with gas fuelling ($f_{ELM}\\approx$60 and 25 Hz respectively). In t
he no-gas phase\, the ion temperature\, stored energy and neutron rate con
tinuously increase until the appearance of a core MHD mode ($n$=4) trigger
ed by a sawtooth crash (see sharp drop in the neutron rate at 10.5 s\, Fig
. 1(e))\, suggesting the performance is limited by MHD rather than transpo
rt. It must be noted that this MHD event does not lead to a disruption\, t
he discharge survives and is landed safely. Density and radiation remain e
ssentially constant after the gas puff is switched off\, indicating partic
le transport is fast enough to provide adequate density and impurity contr
ol. This behavior differs from what observed in the hot-ion H-mode develop
ed in JET-C(1) where density and radiated power increased constantly durin
g long ELM free periods\, eventually leading to a radiation collapse and b
ack transition to L-mode.\nDespite the absence of gas puffing and strong e
lectron density peaking of the no-gas scenario\, which typically lead resp
ectively to an increase in W source and to strong inward impurity convecti
on\, central impurity accumulation does not take place\, the temperature o
f the outer target and the total radiated power are comparable to the refe
rence discharge. Both discharges also show very similar 2D radiation patte
rns\, with strong localization on the LFS midplane at $\\psi_N$>0.8 and ve
ry low central values. Due to the much lower electron density at the pedes
tal top\, the no-gas discharge reaches similar radiated power to the refer
ence with a factor 4 increase in mid-Z (Ni/Fe/Cr/Cu) and high-Z (W) impuri
ty concentrations\, which in turn are the cause for the increased LOS-inte
grated $Z_{eff}$ measurement (Fig. 1(c)). Due to the strong localization o
f these impurities on the LFS midplane\, their increased concentration doe
s not affect the plasma center where $Z_{eff}$ remains < 1.4 as in the ref
erence\, so core dilution is also kept under control.\nThe no-gas scenario
exhibits remarkably good absolute and normalized performance\, albeit tra
nsient\, reaching peak values of $H_{98}\\approx$1.4\, $\\beta_N\\approx$2
.2\, $W_{MHD}\\approx$9 MJ\, and this is achieved at much lower collisonal
ity ($\\nu_{e\,ped}^*$< 0.1\, close to ITER values) than the reference ELM
y H-mode at higher density. The electron pedestal pressure is also slightl
y smaller than the reference\, albeit at lower density and higher temperat
ure\, but there is a significant increase in core electron and ion pressur
es\, resulting in a 35% increase in global energy confinement at the maxim
um stored energy. Improved transport driven by the increase in sheared $E\
\times B$ flow [2] is thought to contribute to both the strong peaking of
the density profile and the improved performance. Additional effects assoc
iated with the large population of fast ions present in these plasmas\, as
shown in [3]\, might also play a role in the overall improved thermal tra
nsport. The added impurity control is provided by the increased ion temper
ature screening enhanced by the extreme toroidal rotation [4].\nAn especia
lly interesting feature of the no-gas regime is the marked reduction in EL
M size compared to conventional ELMy H-mode plasmas. With the decrease in
edge density the type I ELMs are replaced by very small ELMs at a much hig
her frequency. ELM size increases as the pedestal pressure increases\, but
remains significantly smaller than those obtained in the reference pulse
at higher density and similar heating power. We note that in those conditi
ons the link between ELM size and pedestal collisionality and/or edge dens
ity typically found for Type I ELMs is lost [5]. Both the electron density
and temperature pedestals become wider\, and there is a substantial reduc
tion in the maximum $\\nabla n_e$ as the edge density decreases\, resultin
g in a lower maximum $\\nabla P_e$. Pedestal stability analysis indicates
that the edge operating point is below the peeling-ballooning boundary\, w
hich might explain the absence of large type I ELMs. The underlying physic
s mechanisms responsible of the onset of these small ELMs are still a matt
er of ongoing investigation. \nThe new no-gas H-mode regime recently demon
strated in JET-ILW provides a valuable opportunity to study the confinemen
t properties and ELM dynamics of high temperature plasmas with temperature
and density profiles substantially different from those obtained in the c
onventional scenarios. With the aim of improving our understanding and inc
reasing the accuracy of extrapolations for ITER\, this scenario allows val
idating existing transport models and investigating the role of different
physics mechanism involved in the observed improved energy confinement and
impurity control.\n![Comparison of a no-gas H-mode discharge (#94900) wit
h a type I ELMy H-mode reference (#94777) with gas dosing during the main
heating phase ($Ip$=3 MA\, $q_{95}$=3.2\, low $\\delta$\, $P_{NBI}$=22 MW\
, $P_{ICRH}$=4 MW)][1]\n\n**ACKNOWLEDGEMENTS**\nThis work has been carried
out within the framework of the EUROfusion Consortium and has received fu
nding from the EURATOM research program 2014-2018 under grant agreement No
. 633053. The views and opinions expressed herein do not necessarily refle
ct those of the European Commission. This research was supported in part b
y grant FIS2017-85252-R of the Spanish Research Agency\, including ERDF-Eu
ropean Union funding.\n**References**\n 1. The JET team\, Plasma Phys. Con
trol. Fusion 37 (1995)\, p. A35\n 2. J. Garcia et al.\, Plasma Phys. and C
ontrol. Fusion 61 (2019) 104002\n 3. J. Garcia et al.\, Nucl. Fusion 55 (2
015) 053007\n 4. C. Angioni et al.\, Physics of Plasmas 22 (2015) 055902\n
5. A. Loarte et al.\, Plasma Phys. and Control. Fusion 45 (2003) 1549-156
9\n\n [1]: https://i.postimg.cc/pTgkr6z1/94900-94777-v4.png\n\nhttps://co
nferences.iaea.org/event/214/contributions/17071/
LOCATION:Virtual Event
URL:https://conferences.iaea.org/event/214/contributions/17071/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Experimental investigation and gyrokinetic simulations of multi-sc
ale electron heat transport in JET\, AUG and TCV
DTSTART;VALUE=DATE-TIME:20210513T093100Z
DTEND;VALUE=DATE-TIME:20210513T094800Z
DTSTAMP;VALUE=DATE-TIME:20210513T221423Z
UID:indico-contribution-1772-17070@conferences.iaea.org
DESCRIPTION:Speakers: Alberto Mariani (Istituto per la Scienza e la Tecnol
ogia dei Plasmi\, CNR\, Milano (Italy))\nCore transport in present tokamak
s is mostly ascribed to micro-turbulence driven by the non-linear saturati
on of ion-scale ITG-TEM [1] instabilities ($k_\\theta\\rho_i\\le1$\, where
$k_\\theta$ is the poloidal wave number and $\\rho_i$ the ion Larmor rad
ius). It has been shown that electron-scale ETGs [2] ($k_\\theta\\rho_e\\l
e1$) can also impact the heat transport\, also exchanging energy with ITG-
TEM turbulence by multi-scale coupling [3-9]. This topic of investigation
gains a particular relevance due to its potential impact on devices like I
TER\, dominated by electron heating. ETG modes have been shown to play a r
ole in plasmas with mixed ion and electron heating\, since a proper balanc
e of ion heating\, decreasing the ETG threshold in $T_e$ gradient (which i
ncreases with increasing $T_e/T_i$ [10])\, and electron heating (pushing $
T_e$ gradient towards threshold while increasing the threshold due to $T_e
/T_i$ increase)\, could destabilize them. Also all mechanisms that stabili
ze ITGs\, such $E\\times B$ or fast ions from neutral beams (NBI) and/or i
on cyclotron resonance heating (ICRH)\, due to multiscale interactions ope
n a window favourable for ETG destabilization. \n\nThe response of the $T_
e$ profiles to the applied heating can be experimentally investigated by p
erforming normalized electron heat flux scans and/or RF power modulation a
nalysis. The two methods can be used in conjunction to extract information
on the dependence of the gyro-Bohm normalized electron heat flux $q_{egB}
$ on the normalised $T_e$ logarithmic gradient $R/L_{Te}$\, yielding exper
imental values for the threshold $R/L_{Te\,crit}$ for the onset of turbule
nt transport and for the ‘electron stiffness’ $\\partial q_{egB} /\\pa
rtial R/L_{Te}$. The experimental results can be compared with the output
of gyrokinetic (GK) simulations\, which infer both $R/L_{Te\,crit}$ (fast
linear runs)\, and the dependence of the saturated heat flux on $R/L_{Te}$
(more costly nonlinear runs). Resolving both ion and electron scales (i.e
. performing nonlinear multi-scale simulations) is computationally very de
manding and just became possible in the last years.\n\nIn order to access
a broad range of parameters\, a great effort is actually devoted to analys
e different machines\, comparing experimental and numerical results\, with
in the framework of EUROfusion and of the ITPA Transport & Confinement gro
up. In this paper\, the analysis of plasmas of three different tokamaks\,
i.e. the Joint European Torus (JET\, at Culham\, UK)\, ASDEX Upgrade (AUG\
, at Garching\, DE) and the Tokamak à Configuration Variable (TCV\, at La
usanne\, CH)\, is presented. Dedicated plasma discharges have been analyse
d experimentally and modelled numerically\, by means of GK codes (GENE [11
] and GKW [12]) and reduced quasi-linear models (TGLF [13] and QuaLiKiz [1
4]). The results of the different tokamaks concur to make a general pictur
e indicating that ETGs could also be important for electron heat transport
in fusion relevant conditions\, in particular when $T_e\\sim T_i$ with co
nsistent fast ion density.\n\nTCV is equipped with an NBI system\, that al
lows the plasma to achieve $T_e\\sim T_i$ in conjunction with high $R/L_{T
e}$ (due to ECRH)\, allowing to access parameters compatible with ETGs. Tw
o dedicated L-mode discharges\, with $B_0=1.41$ T\, $I_p=170$ kA have been
performed with a different proportion of deposited ECRH power on- vs off-
axis to perform a heat flux scan. Each pulse presented different phases co
rresponding to a different proportion of NBI/ECRH power to vary $T_e/T_i$\
, with ECRH both steady and modulated to allow a perturbative analysis. B
oth the experimental analysis and GK modelling (linear multi-scale and non
linear ion-scale simulations) tend to indicate a possible role of ETGs at
mid-radius when both ECRH and NBI are injected simultaneously\, and at a l
arger toroidal radius $ρ_{tor}=0.7$ also when only ECRH is injected. In t
he former case\, the main mechanism which explains the failure of ion scal
es alone to explain the experimental fluxes\, is the stabilisation of ion-
scales by the fast ions that are produced by the NBI. These results\, publ
ished in [6]\, provide hints of a contribution of ETGs to electron heat tr
ansport in TCV plasmas.\n\nExperiments on the AUG tokamak to study electro
n heat transport [5] have produced H-mode discharges with $B_0=2.5$ T\, $I
_p=0.8$ MA\, injecting 2.5 MW of ECRH (steady and modulated to perform the
perturbative analysis) and 5 MW of NBI in order to have $T_e\\sim T_i$. D
ifferent discharges had different proportions of ECRH power deposition on-
vs off-axis\, in order to obtain the heat flux scan. At mid-radius\, both
the electron heat pulse diffusivity $χ^{HP}$ (from perturbative analysis
) and $q_{egB}$ (from steady state scan)\, indicate strong turbulence lev
els above $R/L_{Te}\\sim$ 6-7\, leading to a moderate/high electron stiffn
ess\, consistent with the possible presence of ETGs. Both GENE and GKW lin
ear-gyrokinetic simulations predict a role for ETGs for $R/L_{Te}>6$\, bas
ed on an effective model for nonlinear turbulence saturation [15]. Both io
n-scale and multi-scale simulations are being performed in order to analys
e the most representative AUG pulse\, to test the existence of an ETG ‘w
all’ limiting the achievable $R/L_{Te}$. The preliminary multi-scale res
ults are in agreement with TGLF in indicating a >30% contribution of ETGs
to the electron heat flux for the cases close to threshold\, setting high
electron stiffness above it (see figure 1-2).\n\nFollowing early results p
ointing to an important role of ETGs in JET [4]\, very recently dedicated
sessions on ETGs have been performed at JET. Both L- and H-mode plasmas ha
ve been obtained\, with $B_T=3.3$ T\, $I_p=2$ MA\, injecting 0-20 MW of NB
I and up to 6 MW of ICRH (H minority\, mainly heating electrons)\, achievi
ng heat flux scans for a range of $T_e/T_i$ values. The preliminary analys
is of the experimental data indicates that JET results are very similar to
the AUG ones\, with a strong increase of the electron stiffness for $R/L_
{Te}>6$ (see figure 3). L-mode cases\, in particular\, allow to obtain suf
ficiently large values of $q_{egB}$ at large $R/L_{Te}>6$\, giving the hin
t of a possible ETG ‘wall’. In parallel\, high performance hybrid disc
harges are analyzed in order to study the ETG impact on these scenarios. B
oth sets of data are being modelled by means of single scale GK simulation
s and reduced models\, in order to set the basis for heavier multi-scale G
K simulations.\n\n\nThis work has been carried out within the framework of
the EUROfusion Consortium and has received funding from the Euratom resea
rch and training programme 2014– 2018 and 2019– 2020 under grant agree
ment No. 633053. The views and opinions expressed herein do not necessaril
y reflect those of the European Commission. This work was also conducted u
nder the auspices of the ITPA Topical Group on Transport and Confinement.
We acknowledge the CINECA award under the ISCRA initiative\, for the avail
ability of high performance computing resources and support.\n\n![$q_{e\,i
\,gB}$ vs $R/L_{Te\,i}$ for AUG at mid-radius\, comparing exp. with GENE i
on/multi-scale and TGLF.][f1]\n\n![GENE ion/multi-scale $q_{eGB}$ spectra.
][f2]\n\n![Experimental $q_{egB}$ vs R/LTe (JET).][f3]\n \n[1] W. Horton\,
Rev. Mod. Phys. 71\, 735 (1999)\n[2] W. Dorland et al.\, Phys. Rev. Lett.
85\, 5579 (2000)\n[3] N.T. Howard et al.\, Nucl. Fusion 56\, 014004 (2016
)\n[4] N. Bonanomi et al. Nucl. Fusion 58\, 124003 (2018)\n[5] F. Ryter et
al.\, Nucl. Fusion 59\, 096052 (2019)\n[6] A. Mariani et al.\, Nucl. Fusi
on 59\, 126017 (2019)\n[7] S. Maeyama et al.\, Phys. Rev. Lett. 114\, 2550
02 (2015)\n[8] A. Marinoni et al.\, Nucl. Fusion 57\, 126014 (2017) \n\n[9
] C. Holland et al.\, Nucl. Fusion 57\, 066043 (2017)\n[10] F. Jenko et al
.\, Phys. Plasmas 8\, 4096 (2001)\n[11] F. Jenko et al.\, Phys. Plasmas 7\
, 1904 (2000)\n[12] A. Peeters et al.\, Comput. Phys. Commun. 180\, 2650 (
2009)\n[13] G.M. Staebler et al.\, Phys. Plasmas 23\, 062518 (2016)\n[14]
J. Citrin et al.\, Plasma Phys. Control. Fusion 59\, 124005 (2017)\n[15] G
.M. Staebler et al.\, Nucl. Fusion 57\, 066046 (2017)\n \n\n\n [f1]: htt
ps://www.dropbox.com/s/smf1h6o8qe7xjrt/figure1.jpg?raw=1\n [f2]: https://
www.dropbox.com/s/8kzvtsueeu3gpfe/figure2.jpg?raw=1\n [f3]: https://www.d
ropbox.com/s/leisivovq69sqv1/figure3.jpg?raw=1\n\nhttps://conferences.iaea
.org/event/214/contributions/17070/
LOCATION:Virtual Event
URL:https://conferences.iaea.org/event/214/contributions/17070/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Improved prediction scheme for turbulent transport by combining ma
chine learning and first-principle simulation
DTSTART;VALUE=DATE-TIME:20210513T091400Z
DTEND;VALUE=DATE-TIME:20210513T093100Z
DTSTAMP;VALUE=DATE-TIME:20210513T221423Z
UID:indico-contribution-1772-17068@conferences.iaea.org
DESCRIPTION:Speakers: Masanori Nunami (National Institute for Fusion Scien
ce)\n**Outline** A novel scheme to predict the plasma turbulent transport
is developed by combining the machine learning technique and the first-pr
inciple gyrokinetic simulations. The machine learning technique is applied
to find the relevant input parameters of the nonlinear gyrokinetic simula
tions which should be performed and to optimize the reduced transport mode
l. The developed scheme can drastically reduce the computational costs to
perform the quantitative predictions of the plasma profiles and the turbul
ent transport levels. Utilizing the scheme\, the quantitative predictions
for the turbulent transport can be realized by only one-time first-princip
le simulation for each radial position. \n\n**Conventional transport predi
ction schemes** The first-principle simulation based on the gyrokinetics
is powerful and reliable way to predict the turbulent transport or the pla
sma profiles. Indeed\, it has been possible to validate the gyrokinetic si
mulations against the experimental observations within the experimental er
rors [1]. For the predictions\, there are two main schemes. One is perform
ing many first-principle simulations\, which is called *flux-matching tech
nique* [2]. The other is employing the reduced transport model\, which is
constructed by results of many gyrokinetic simulations. In the former sche
me\, by performing gyrokinetic simulations with changing the input paramet
ers for the plasma profiles\, we can find the transport fluxes that agree
with the observations quantitatively. However\, to obtain the matched tran
sport fluxes\, numerous gyrokinetic runs should be demanded. For example\,
even in the case of the single-species plasma\, the conventional flux-mat
ching demands 5 times or more runs for each radial position. Therefore\, i
n particular\, for the case of multi-species plasmas\, we must perform a h
uge number of the simulations in the multi-dimensional parameter space whi
ch is exponentially expanding with increasing the number of the species. I
n the latter scheme\, on the other hand\, the reduced model enables us to
obtain the turbulent transport fluxes without additional nonlinear gyrokin
etic runs. However\, since such reduced models are constructed by the limi
ted parameters of the plasma profiles and include certain prediction error
s\, it is hard to reproduce the nonlinear simulation results precisely. In
this work\, by combining the first-principle simulations and the machine
learning techniques via the reduced model\, we reduce the number of the gy
rokinetic runs and realize more reliable and efficient predictions.\n\n**A
new transport prediction scheme** We develop a new transport prediction
scheme\, which consists of three parts as shown in Fig.1. Here\, we consi
der the ion heat transport as an example. To reduce the number of the firs
t-principle simulations\, we have to find the relevant input parameters of
the simulations to realize the resultant transport fluxes which is close
to the experimental results. First\, in the developed scheme\, for finding
the input parameters of the simulation\, we employ the reduced model for
the ion heat diffusivity [3]\, which is constructed under the adiabatic el
ectron assumption. The model includes the turbulent contribution $\\cal L$
and the zonal-flow contribution $\\tau_{\\rm ZF}$ as $\\chi_{\\rm i}^{\\r
m model}/\\chi_{\\rm i}^{\\rm GB} = A_1 {\\cal L}^{\\alpha_0} / (A_2 + \\t
au_{\\rm ZF}/{\\cal L}^{1/2})$. Here\, ${\\cal L} \\equiv a(\\rho) [R/L_{T
{\\rm i}} - \\beta_0 R/L_{T{\\rm i}}^{\\rm cr}]$ with the critical tempera
ture gradient $R/L_{T{\\rm i}}^{\\rm cr}(\\rho)$\, $\\tau_{\\rm ZF}$ is th
e zonal-flow decay time\, and $A_1\, A_2$\, $\\alpha_0$\, and $\\beta_0$ a
re constant numbers determined in the model. Using the machine learning nu
merical library [4] for the initial model\, we find the initial guess of t
he temperature gradient which realizes the transport flux $Q_{\\rm i} = -n
_{\\rm i} \\chi_{\\rm i} \\nabla T_{\\rm i}$ that agrees with the observat
ions in the target plasma. Second\, using the guessed parameter\, one tria
l run of the first-principle simulation is performed for each radial posit
ion. Third\, using the results of the trial run for each radial position\,
the machine learning is performed again to optimize the reduced model by
tuning the parameters with $\\alpha_0 \\to \\alpha$ and $\\beta_0 \\to \\b
eta$ in the initial model. Then we can obtain the optimized transport mode
l $\\chi_{\\rm i}^{\\rm opt}$ suitable for the target plasma by only one g
yrokinetic simulation at each radial position.\n\n![A schematic of the dev
eloped prediction scheme by combining the machine learning and the first-p
rinciple simulations.][6]\n\nFigure 2(a) shows the comparison of the tempe
rature gradient dependences of the ion heat diffusivities obtained by the
initial reduced model and the optimized model for the ITG turbulent transp
ort in the high-$T_{\\rm i}$ plasma in the LHD [5]. Compared with the init
ial model\, the optimized model with $\\alpha/\\alpha_0=0.89$ and $\\beta/
\\beta_0=0.92$ quite agree with the nonlinear runs which are never used fo
r the construction of the optimized model. In addition\, it is confirmed t
hat the optimized model can reproduce the weakening of the profile stiffne
ss due to the nonlinear effects. Furthermore\, as shown in Fig.2(b) for th
e convergence checks of the developed scheme\, one trial run is enough to
construct the optimized model for each radial position because the machine
learning via the initial model can guess the relevant temperature gradien
t which is close to the guess from the nonlinear runs\, independently. The
refore\, the scheme can reduce the number of the first-principle simulatio
n runs to only once for each radial position. The reduction of the computa
tion time will be more remarkable in the multi-species case.\n\n![(a) The
temperature gradient dependences of the ion heat diffusivities obtained fr
om the initial model (dotted curve) and the optimized model (dashed curve)
at $\\rho=0.65$ in the high-$T_{\\rm i}$ LHD plasma. (b) The convergence
checks of the developed scheme for the iterations of the first-principle s
imulations. In (a)\, the diamond represents the result of the one trial ru
n\, and the circles are the nonlinear simulation results.][7]\n\nUsing the
optimized model\, we can obtain the guesses for $R/L_{T {\\rm i}}$ and pr
edictions for $\\chi_{\\rm i}$ as shown in Fig.3. Although the temperature
gradients guessed by the initial and the optimized models are not so diff
erent from each other\, the optimized model can reproduce the turbulent di
ffusivities by the nonlinear simulations better than the initial model. At
least in this application\, we can obtain the transport levels which quan
titatively agree with the nonlinear runs by using the developed scheme wit
h performing just one-time first-principle simulation for each radial posi
tion. Since the developed scheme enables us to thoroughly reduce the numbe
r of the first-principle simulations\, the scheme can be applied to the in
tegrated transport code for quick and precise predictions of the plasma pr
ofiles under the operation scenarios with drastically saved computational
resources.\n\n![Radial profiles of (a) the guesses of the ion temperature
gradients by the initial model (triangles)\, the optimized model (diamonds
)\, and the nonlinear simulations (circles)\, and (b) the predictions of t
he ion heat diffusivities at the guessed temperature gradients in the init
ial model (triangles)\, the optimized model (diamonds)\, and the nonlinear
simulations (circles) for the LHD plasma.][8]\n\n[1] M. Nunami\, *et al.*
\, Phys. Plasmas **25**\, 082504 (2018). \n[2] T. G$\\ddot{\\rm o}$rler\,
*et al.*\, Phys. Plasmas **21**\, 122307 (2014). \n[3] S. Toda\, *et al.*\
, J. Phys. Conf. Ser. **561**\, 012020 (2014).\n[4] M. Abadi\, *et al.*\,
arXiv:1603.04467 (2016). \n[5] Y. Takeiri\, *et al.*\, Nucl. Fusion **57**
\, 102023 (2017).\n\n\n [6]: https://workshop.nifs.ac.jp/fec2020/image/10
0-Nunami-image-fig_1.png\n [7]: https://workshop.nifs.ac.jp/fec2020/image
/100-Nunami-image-fig_2.png\n [8]: https://workshop.nifs.ac.jp/fec2020/im
age/100-Nunami-image-fig_3.png\n\nhttps://conferences.iaea.org/event/214/c
ontributions/17068/
LOCATION:Virtual Event
URL:https://conferences.iaea.org/event/214/contributions/17068/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Predict First: flux-driven multi-channel integrated modelling over
multiple confinement times with the gyrokinetic turbulent transport model
QuaLiKiz
DTSTART;VALUE=DATE-TIME:20210513T085700Z
DTEND;VALUE=DATE-TIME:20210513T091400Z
DTSTAMP;VALUE=DATE-TIME:20210513T221423Z
UID:indico-contribution-1772-17069@conferences.iaea.org
DESCRIPTION:Speakers: Jonathan Citrin (FOM DIFFER - Dutch Institute for Fu
ndamental Energy Research)\nAn accurate and predictive model for turbulent
transport fluxes driven by microinstabilities is a vital component of fir
st-principle-based tokamak plasma simulation. However\, tokamak scenario p
rediction over energy confinement timescales is not routinely feasible by
direct numerical simulation with nonlinear gyrokinetic codes. Reduced orde
r modelling with quasilinear turbulent transport models provides significa
nt computational speedup\, and is justified in many regimes. The justifica
tion of the quasilinear approximation for transport driving spatial scales
is a consequence of the underlying structure of tokamak microturbulence\,
and is validated by comparison to nonlinear simulations. This approach ha
s emerged as a successful tool for prediction of core tokamak plasma profi
les. We focus on significant progress in the quasilinear gyrokinetic trans
port model QuaLiKiz [1\,2]\, and its application within flux driven integr
ated tokamak simulation suites.\n\nTo model 1s of JET plasma on order of 2
4 hours with 10 CPUs\, QuaLiKiz employs an approximated solution of the mo
de structures to significantly speed up the computation time compared to f
ull linear gyrokinetic solvers. Additional approximations include maintain
ing shifted-circle $(\\hat{s}-\\alpha)$ geometry\, and the electrostatic l
imit. These approximations\, together with optimisation of the dispersion
relation solution algorithm within integrated modelling applications\, lea
ds to flux calculations $10^{6-7}$ faster than local nonlinear gyrokinetic
simulations. This allows tractable simulation of flux-driven dynamic prof
ile evolution over multiple confinement times including all transport chan
nels: ion and electron heat\, main particles\, impurities\, and momentum.
QuaLiKiz is open source and available at www.qualikiz.com. \n\nIn this con
tribution\, we will summarize the justification of the quasilinear approxi
mation [3\,4]\, sketch the basis of the QuaLiKiz transport model and its v
alidity in comparison to nonlinear simulations\, and illustrate validation
of the model against experimental measurements at JET through flux-driven
simulations within the JINTRAC integrated modelling suite [5\,6]\, see fi
gure 1 for an example. This capability 1) enhances the interpretation of p
resent-day experiments\, 2) enables “Predict First” simulations to aid
with experimental optimization\, and 3) allows theory-based extrapolation
to future machine performance\, at least with respect to core turbulence
physics. While we focus here on JINTRAC simulations\, QuaLiKiz is also cou
pled to the ASTRA [7\,8]\, CRONOS [9] and ETS [10] integrated modelling co
des. \n\nRecent QuaLiKiz applications within integrated modelling include:
W-accumulation interpretation and optimization\, where the QuaLiKiz predi
ction of background kinetic profiles is critical for setting the neoclassi
cal heavy impurity transport level [11-13]\; modelling of multiple-isotope
experiments at JET\, where fast isotope mixing in the Ion Temperature Gra
dient (ITG) regime is crucial for experimental interpretation and has impo
rtant implications for potential scenarios in JET DT\, as well as for reac
tor burn control [14]\; development of Uncertainty Quantification methods
using Gaussian Process Regression to enhance statistical rigour in model v
alidation\, providing avenues for error propagation within QuaLiKiz simula
tions in integrated modelling [15]\; predictive modelling for ITER scenari
os\, which predict the target Q∼10 when using a theory-based pedestal bo
undary condition [16]\; and predictive modelling for DTT scenarios [17].\n
\nBeyond standard application within integrated modelling\, QuaLiKiz has b
een leveraged for the development of realtime calculation capability for s
cenario optimization and realtime-oriented applications. This is based on
machine learning methods\, where a large database of pre-calculated QuaLiK
iz runs is used to train feedforward neural networks to accurately reprodu
ce model predictions. The neural network transport model provides a furthe
r 6 orders of magnitude speedup\, 1 trillion times faster than the anchori
ng nonlinear simulations [18]. By coupling to the RAPTOR [19] control-orie
nted fast tokamak simulator\, realtime-capable transport predictions are p
ossible. This opens up a plethora of possibilities and innovation in realt
ime controller design and validation\, scenario preparation\, and discharg
e optimization.\n\nWhile QuaLiKiz has had significant predictive success\,
continuously challenging and improving the model is a crucial component f
or instilling validity in wide parameter space. Beyond its role in experim
ental interpretation and prediction\, reduced models such as QuaLiKiz are
a key player in the multi-fidelity model hierarchy due to its feasibility
for systematic comparison with experiments and identifying trends in model
validation. This spurs further research\, also incorporating higher fidel
ity linear and nonlinear models\, ultimately improving our understanding o
f core tokamak turbulence physics.\n\nWe thus conclude with an overview of
recent work dedicated to testing and improving the underlying QuaLiKiz as
sumptions. This includes: modification of the collisionality model\, criti
cal for obtaining the correct parameter dependencies of Trapped Electron M
odes (TEM)\; validating the QuaLiKiz Electron Temperature Gradient (ETG) m
odel versus multi-scale nonlinear GENE simulations\; testing validity of Q
uaLiKiz towards the L-mode edge\, where the standard ITG/TEM/ETG paradigm
breaks down at high collisionality\, due to the onset of modes with a drif
t-resistive nature\, currently out of QuaLiKiz scope\; testing the impact
of s-α geometry on the turbulence regime\, compared to full geometry\, pa
rticularly at more outer radii where shaping effects are more prominent. F
uture work will extend QuaLiKiz to electromagnetic regimes. \n\n![Multi-ch
annel predictive modelling with JINTRAC-QuaLiKiz of JET discharge #91227\,
corresponding to stationary state following simulation over 10 energy con
finement times. Experimental profiles were fitted with Gaussian Process Re
gression corresponding to data averaging over the time window t=8.2-8.5s.
Core boundary condition was set at $\\rho=0.8$. From Ref [14].][f1]\n\n\n*
*References**\n\n[ 1] J. Citrin et al.\, Plasma Phys. Control. Fusion 59 1
24005 (2017)\, and http://qualikiz.com\n[2] C. Bourdelle et al.\, Plasma P
hys. Control. Fusion 58 014036 (2016)\n[3] A Casati et al\,. Nucl. Fusion
49 085012 (2009)\n[4] J Citrin et al.\, Phys. Plasmas 19 062305 (2012)\n[5
] G. Cenacchi and A. Taroni\, JET-IR \, 84 (1988)\, eNEA-RT-TIB–88-5\n[6
] M. Romanelli et al.\, Plasma and Fusion Research 9 3403023 (2014)\n[7] G
. V. Pereverzev et al.\, IPP Report 5/42 (August 1991)\n[8] E. Fable et al
.\, Plasma Phys. Control. Fusion 55 124028 (2013)\n[9] J.F. Artaud et al.\
, Nucl. Fusion 50 043001 (2010)\n[10] D. Kalupin et al.\, Nucl. Fusion 53
123007 (2013)\n[11] S Breton et al.\, Nucl. Fusion 58 96003 (2018)\n[12] F
. Casson et al.\, submitted to Nucl. Fusion\n[13] O Linder et al.\, Nucl.
Fusion 59 016003 (2019)\n[14] M. Marin et al.\, Nucl. Fusion 60 046007 (2
020)\; and this conference\n[15] A. Ho et al.\, Nucl. Fusion 59 056007 (2
019)\n[16] P. Mantica et al.\, Plasma Phys. Control. Fusion 62 014021 (202
0)\n[17] I. Casiraghi. P. Mantica et al.\, this conference\n[18] K.L. van
de Plassche et al.\, Physics of Plasmas 27\, 022310 (2020) \; and this con
ference\n[19] F. Felici et al.\, Plasma Phys. Control. Fusion 54 025002 (2
012)\n\n\n [f1]: https://gitlab.com/qualikiz-group/QuaLiKiz-documents/-/r
aw/master/conferences/conference_images/citrin_IAEA_2020.jpg\n\nhttps://co
nferences.iaea.org/event/214/contributions/17069/
LOCATION:Virtual Event
URL:https://conferences.iaea.org/event/214/contributions/17069/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Strong reversal of simple isotope scaling laws in tokamak edge tur
bulence
DTSTART;VALUE=DATE-TIME:20210513T084000Z
DTEND;VALUE=DATE-TIME:20210513T085700Z
DTSTAMP;VALUE=DATE-TIME:20210513T221423Z
UID:indico-contribution-1772-17067@conferences.iaea.org
DESCRIPTION:Speakers: Emily Ann Belli (General Atomics)\nThe role of the n
onadiabatic electron drive in regulating the isotope mass scaling of gyrok
inetic turbulence is assessed in the transition from ion-dominated core tr
ansport regimes to electron-dominated edge transport regimes. The scaling
of the plasma energy confinement time with hydrogenic isotope mass is of
critical importance\, as most tokamaks operate with deuterium (D) as the m
ain ion species\, while ITER calls for dominant hydrogen (H) operation in
the first phase\, transitioning to 50:50 deuterium-tritium (DT) fuel compo
sition at reactor-level operation. Experimental observations often show c
onfinement *improving* with increasing ion mass {1}. Simple gyroBohm-scali
ng theoretical arguments (that ignore electron dynamics)\, however\, predi
ct that the turbulent ion energy flux scales with the square root of the i
on mass\, with the implication that the global confinement *degrades* with
increasing ion mass. Using nonlinear gyrokinetic simulations of DIII-D\,
we illustrate a remarkable transition in the turbulent isotope scaling to
wards the plasma L-mode edge. The transition is controlled by finite elect
ron-to-ion mass-ratio dependence of the nonadiabatic electron response\, d
ominantly generated by the parallel motion\, which represents a correction
to bounce-averaging of the electrons. The nonadiabatic electron drive str
ongly regulates the turbulence levels and plays a key role in altering --
and in the case of the DIII-D edge\, *reversing* -- the simple gyroBohm sc
aling rule. The finite electron-mass correction is larger for light ions a
nd increases with increasing $q$ so that\, while it is weak in the core\,
it dominates the mass scaling in the edge. Overall\, these results may hav
e favorable implications for global energy confinement and for the power t
hreshold for the L-mode to H-mode transition in a reactor like ITER from H
to D to DT\, consistent with recent experimental observations comparing h
ydrogen and deuterium plasmas {2}.\n\n**Theoretical basis for gyrokinetic
isotope scaling**\n\nThe ion gyrokinetic equation together with the assump
tion of purely adiabatic electrons describes ion energy fluxes $Q_a$ that
exhibit *simple gyroBohm scaling*:\n\n$\\qquad Q_a = C_0 \\\, Q_{\\rm GBa}
\\quad\\text{where}\\quad Q_{\\rm GBa} = Q_{\\rm GBD} \\sqrt{m_a/m_{\\rm
D}} \\\; .$\n\nHere\, the subscript $a$ is the species index\, $Q_{\\rm GB
D} \\doteq n_e T_e c_{sD} \\rho_{*D}^2$ is the deuterium gyroBohm energy f
lux\, $c_{sD}=\\sqrt{T_e/m_D}$ is the deuterium sound speed and $\\rho_{*D
}=(c_{sD}/\\Omega_{D\,{\\rm unit}})/ a$ is the normalized deuterium ion-so
und gyroradius. Because $C_0$ is species-independent\, we must always obs
erve $Q_{\\rm H} < Q_{\\rm D} < Q_{\\rm T}$\, i.e. heavier isotopes should
give rise to confinement degradation. When kinetic electron dynamics are
fully retained\, however\, we expect the more complicated *true gyroBohm
scaling*:\n\n$\\qquad Q_a = C\\left(m_e/m_a\\right) Q_{GBa} \\\; \,$\n\nth
at contains an additional electron-to-ion mass-ratio dependence. We obser
ve that the $m_e/m_a$ mass dependence of $C$ typically opposes the simple
gyroBohm mass dependence\, and can in some cases dominate and reverse the
gyroBohm dependence so that $Q_{\\rm H} > Q_{\\rm D} > Q_{\\rm T}$.\n\n**R
eversal of simple gyroBohm scaling in electron-transport dominated edge re
gimes**\n\nUsing CGYRO {3} we gauge the influence of kinetic electrons on
the isotope scaling of energy flux in the transition from ion-dominated (c
ore) transport regimes to electron-dominated (edge-typical) transport regi
mes. Simulation parameters are based on DIII-D #173147 at t=1705ms\, an o
hmically-heated L-mode discharge. Fig. [1] shows that CGYRO matches the t
otal (e+i) experimental power-balance flux in both the ion-dominated core
and the electron-dominated edge ($r/a \\ge 0.9$)\, where $Q_e \\sim 1.5 \\
\, Q_i$ and TGLF underestimates the edge electron transport. The dominant
linear mode in the core is ion-temperature-gradient (ITG) driven\, wherea
s in the edge an electron temperature gradient-driven trapped electron mod
e (TEM) dominates. Fig. [2] compares the simulated ion energy flux for de
uterium versus hydrogen versus 50:50 DT as the main ion species (with all
other experimental parameters fixed). In the ITG-dominated regime\, $Q_{\
\rm H} \\sim Q_{\\rm D} \\sim Q_{\\rm DT}$\, meaning simple gyroBohm scali
ng is broken. This is dominantly due to electron collisions\, which more
strongly stabilize heavier species {4}\, and weakly to the ${\\mathbf E} \
\times {\\mathbf B}$ flow shear {5}. However\, a near gyroBohm scaling ca
n be recovered by scaling the electron collision rate and the ${\\mathbf E
} \\times {\\mathbf B}$ shearing rate with the main ion thermal speed. In
contrast\, in the TEM-dominated edge regime\, a strong *reversal* from th
e gyroBohm scaling is found\, with $Q_{\\rm H} \\gg Q_{\\rm D} \\gg Q_{\\r
m DT}$. This implies that hydrogen confinement relative to deuterium is e
xpected to be significantly worse than expected by the simple gyroBohm mas
s scaling. We demonstrate that this reversal is due to the nonadiabatic e
lectron drive from the kinetic electron parallel response which acts to en
hance the TEM turbulence for light ions.\n\n![Total energy flux $(Q_e + Q_
D)$ comparing CGYRO and TGLF with experimental DIII-D power balance.][1]\n
\n![CGYRO energy flux for DIII-D #173147 comparing DT\, D\, and H showing
a strong\, favorable reversal from simple gyroBohm scaling in the edge.][2
]\n\n**Key role of the nonadiabatic electron response**\n\nThe reversal fr
om gyroBohm scaling in the electron transport-dominated edge is controlled
by finite electron-to-ion mass-ratio dependence of the nonadiabatic elect
ron response {6}. At fixed plasma gradients\, this nonadiabatic effect is
strongly enhanced at increased $q$\, as shown in Fig. [3]\, and thus domi
nates in the plasma edge. This is consistent with the $q$-dependence of th
e electron parallel timescale relative to the ion drift timescale: $({\\rm
v}_i/a) \\tau_e \\sim q (R_0/a) {\\rm v}_i/{\\rm v}_e \\sim q (R_0/a) \\s
qrt{m_e/m_i}$. For massless electrons\, $\\tau_e \\rightarrow 0$\, such th
at the passing nonadiabatic distribution vanishes and the trapped distribu
tion is bounce-averaged and independent of mass ratio. At finite electron
mass\, the nonadiabatic correction increases with $q$ and decreases with
$m_i$. Thus\, for light species like hydrogen\, deviation from the bounce
-averaged limit is larger than for deuterium. Electron collisions provide
a secondary mass-ratio correction to the flux scaling\, so to recover simp
le gyroBohm scaling at low $q$ it is also necessary to reduce the collisio
n frequency to eliminate $m_e$-dependence of the trapped-passing boundary
layer width (shown in Fig. [3] for $q<2$ when ${\\bar \\nu}_e \\rightarrow
0$).\n\n![Comparison of D and H energy flux for DIII-D #173147 at $r/a=0.
9$\, showing reversal from gyroBohm scaling at large $q$.][3]\n\n**Implica
tions for global confinement and the L-H threshold in a reactor**\n\nFor a
ssessing the isotope scaling of global energy confinement in a reactor lik
e ITER\, it is essential to properly treat the precise nonadiabatic electr
on dynamics. Fluid or even bounce-averaged electron models are unlikely t
o recover the correct ion-mass scaling. For a full transport analysis\, ad
ditional influences (e.g. impurities\, heating\, MHD) beyond the scope of
this work must also be considered. However\, plasma confinement is known
to be sensitive to edge conditions. Tokamak L-mode edge conditions typica
lly lead to electron transport-dominated turbulence regimes such as studie
d here\, for which the nonadiabatic electron drive is enhanced\, resulting
in a favorable reversal of the simple gyroBohm scaling with ion mass from
H to D to DT. This has implications for lowering the power threshold for
the L-mode to H-mode transition in a reactor like ITER and could trend th
e theoretical turbulent-based global energy confinement isotope scaling to
ward agreement with experimental observations. \n\nThis work was funded by
US DOE Grants DE-FG02-95ER54309 and DE-FC02-06ER54873.\n\n{1} ITER Physic
s Basis Editors\, *Nucl. Fusion* **39**\, 2175 (1999).\n{2} C.F. Maggi et
al.\, *Plasma Phys. Control. Fusion* **60**\, 014045 (2018).\n{3} M. Nakat
a et al.\, *Phys. Rev. Lett.* **118**\, 165002 (2017). \n{4} J. Garcia et
al.\, *Nucl. Fusion* **57**\, 014007 (2017).\n{5} J. Candy\, E.A. Belli\,
and R. Bravenec\, *J. Comput. Phys.* **324**\, 73 (2016).\n{6} E.A. Belli\
, J. Candy\, and R.E. Waltz\, *Phys. Plasmas* **26**\, 082305 (2019).\n\n\
n [1]: https://fusion.gat.com/conference/event/104/attachments/161/1489/B
elli.Emily.IAEA2020.Fig3.jpg\n [2]: https://fusion.gat.com/conference/eve
nt/104/attachments/161/1488/Belli.Emily.IAEA2020.Fig2.jpg\n [3]: https://
fusion.gat.com/conference/event/104/attachments/161/1487/Belli.Emily.IAEA2
020.Fig1.jpg\n\nhttps://conferences.iaea.org/event/214/contributions/17067
/
LOCATION:Virtual Event
URL:https://conferences.iaea.org/event/214/contributions/17067/
END:VEVENT
END:VCALENDAR