Progress in understanding Intermittency in Magnetically Confined Plasma Devices Ghassan Y. Antar University of California San Diego Acknowledgments G. Counsell (MAST, UK) P. Devynck (Tore Supra, France) S. I. Krasheninnikov (UCSD, USA) B. LaBombard (Alcator C-MOD, USA)
MOTIVATION It is very important to understand the scrape-off layer in L-mode - First wall deterioration via sputtering due to the plasma flux. - Impurities that come from the wall finds its way back into the plasma core and deteriorate the confinement. Turbulence in the core can be described as diffusive???? 1) Diffusive Transport: Small radial velocity with respect to the sound speed Turbulent transport in the SOL has 2 components: 2) Convective Transport: Significant radial velocity up to the order of the sound speed (1/10th) 7/28/2006 G. Y. Antar 2
Things to worry about in ITER-like magnetic fusion devices: - In H-mode: ELMs and the associated turbulence (diffusive and convective). - In L-mode: turbulence and convective transport. Data taken on the MAST tokamak Comparing ELMs and turbulence reveals that at the mid-plane the two phenomenon leads to approximately the same particle flux to the walls but the frequency of the turbulent structures are ~2000 more frequent!!!! Therefore it is crucial to fully understand the turbulent structures in the SOL. 7/28/2006 G. Y. Antar 3
First measurements of Intermittent Bursts is in the SOL of CALTECH First-time intermittency is investigated is in the SOL of ADITYA S. Zweben and R. Gould Nuclear Fusion, 23, 1625 (1983) Jha et al. Phys. Rev. Lett., 69, 1374 (1992) 7/28/2006 G. Y. Antar 4
OUTLINE Part I: The origin of intermittency is convective radial transport by avaloids. Part II: The origin of convective radial transport is a saturation of an edge instability. Part III: Universality is deduced by comparing the PISCES linear plasma device, the MAST spherical tokamak, the Alcator C-MOD tokamak and the Tore Supra tokamak. Part IV: The poloidal distribution of avaloids in a tokamak. 7/28/2006 G. Y. Antar 5
Part I: Intermittent Structures are Convective in Nature PISCES-A Linear Plasma Device Plasma Parameters: n e ~ 5x10 17 m -3, T e ~ 15 ev, B = 0.12-0.2 T Plasma radius = 2.5 cm, length = 1 m Vessel radius = 10 cm 7/28/2006 G. Y. Antar 6
I sat decrease and change in nature: In the core fluctuations about an average value In the SOL fluctuations above the average value G. Y. Antar, et al., Phys. Rev. Lett. 87, 065001 (2001). core SOL Wall Isat [a.u.] Momentarily plasma current in the far SOL can be as high as the average current inside the main plasma column r [cm] 7/28/2006 G. Y. Antar 7
Conditional Averaging reveals that Bursts 1- do not conserve mass 2- they have an asymmetric Shape 3- Their electric field is positive meaning outward propagation Duration of the burst Radial velocity V r =E θ /B 7/28/2006 G. Y. Antar 8
The bursts radial velocity is far from being negligible reaching 1/10 th of the sound speed. The radial scale of the bursts is about 1 cm The radial scale of the bursts [mm] Intermittent Bursts Contribute to about 40% of the total radial transport while occupying only 20% of the total duration of the signals The radial velocity of the bursts [m/s] core SOL r [cm] AVALOIDS are defined as large-scale concentration of density with high radial velocity encountered intermittently in the scrape-off layer 7/28/2006 G. Y. Antar 9
Conclusion of Part I: Avaloids are the source of Intermittency observed in magnetically confined devices plasma exists far from the main column Complex dynamics in the SOL High Radial Velocity Large scales Significant contribution to radial transport. 7/28/2006 G. Y. Antar 10
Part II: The origin of intermittency in the SOL investigated on PISCES G. Y. Antar, Phys. Plasmas, 10, 3629 (2003) Starting point of the investigation: The PDF of the time between bursts has 1. A central frequency 2. A long tail It is not clear whether the underlying is coherent or incoherent. (the time between two bursts) 7/28/2006 G. Y. Antar 11
In the PISCES SOL, the PDF is independent of r. Inside the main plasma column, The PDF changes shape. This is caused by the decrease of number of events recorded above the chosen threshold equal to three times the standard deviation 7/28/2006 G. Y. Antar 12
In the SOL the increase of the number of events increases monotonously. Inside the main plasma the increase is abrupt where N b goes from 0 to 200 in 1 cm. PDF( T) in the SOL and inside the main plasma column are now similar suggesting a relation between the two regions. 7/28/2006 G. Y. Antar 13
Behavior of N b as function of the threshold suggests that the signal inside the main plasma column is dominated by a coherent oscillation Intermittent signal N b ξ=0.5 I th /σ N b ξ=1 A Gaussian signal I th /σ A coherent oscillation N b I th /σ I th =2σ 7/28/2006 G. Y. Antar 14
Inside the main plasma column, a coherent oscillation exists in the form of wave packets in agreement with the behavior of N b dependence on I th /σ The jsat signal @ r=2 cm is filtered around 70 khz 7/28/2006 G. Y. Antar 15
Perturbations in the far SOL are correlated to that inside the main plasma column The vertical probe position is altered step by step For each step of the vertical probe, the horizontal probe position is reciprocated all the way until the plasma core. 7/28/2006 G. Y. Antar 16
The cross-conditional averaging technique confirms the cross-correlation results: The SOL bursts are correlated to the main plasma column fluctuations. CA RF (r F =2cm) For r R = 11 and r F = 2 cm: No correlation is reported because the signal for r>10 cm is essentially white noise. For r R = 2 cm and r F = 2 cm: the cross-conditional averaging is similar to the auto-conditional averaging reflecting global modes in the plasma. CA RF (r F =2cm) For r R = 7 cm and r F = 2 cm: The cross-conditional averaging clearly shows that a burst in the SOL is correlated to a wavepacket in the main plasma column. 7/28/2006 G. Y. Antar 17
Using probes on PISCES linear plasma device, bursts in the far SOL of PISCES are correlated to oscillations inside the main plasma column. The fact that bursts look intermittent and not quasi-periodic may be caused by the different paths they have as they interact with the turbulent background that also modifies their scale-lengths and velocities.
Imaging avaloids on CSDX Linear Plasma Device to complement the PISCES results RF Source Langmuir probe Phantom V7 camera 1 µs exposure time 10 µs between frames 1 mm resolution Pump Transparent End-plate Gas: Argon CSDX parameter Typical value Gas pressure 0.5 - few mtorr Electron Temperature ~ 2.7 ev Ion Temperature ~ 0.2 ev (est.) Electron Density Few x 10 12 cm -3 Power (Helicon 13.56 Mz source) 1500W (typically) Magnetic Field Strength Up to ~ 1000 G 7/28/2006 G. Y. Antar 19
Good agreement between the images profiles and fluctuations and those done using Langmuir probe 7/28/2006 G. Y. Antar 20
The system transits from low to high mode number fluctuations in time and can remain in one of the modes for relatively long time [ No evidence of zonal flows were observed in the movies.] 9 th order polynomial fit of a poloidal cut inside the main plasma 7/28/2006 G. Y. Antar 21
Rigid body rotation is observed in CSDX with no significant shear For r>32 mm the velocity goes to 0 resulting from the edge boundary layer 7/28/2006 G. Y. Antar 22
Using Imaging we are able to observe the growth of avaloids, their scale lengths, velocities and to which modes they are correlated to inside the plasma as it was suggested by the experiment on PISCES. Note that no detachment of the avaloid, hence, it is not a blob but rather has a finger-like shape 7/28/2006 G. Y. Antar 23
From an average movie imaging half the main plasma column and the SOL, the convective structures properties are: Life-time ~ 60 µs; L r,avaloid ~ 5 cm; L θ,avaloid ~ 2 cm; V r,avaloid ~10 5 cm/s V θ, main plasma ~ 4.5x10 4 cm/s V θ,avaloids ~ 4.5x10 4 cm/s 7/28/2006 G. Y. Antar 24
Avaloids on CSDX are observed to be associated to low frequency modes that are occurring inside the main plasma column Poloidal mode number determined by a 9 th order polynomial fit as a function of time. Poloidal wave-number spectrum as a function of time. A poloidal cut in the SOL of CSDX showing large fluctuations for small poloidal mode numbers 7/28/2006 G. Y. Antar 25
CA I is the conditionally averaged movie It reveals that the onset of avaloids is associated with the non-linear evolution of the poloidal number m=1 instability. 7/28/2006 G. Y. Antar 26
CONCLUSION of Part II The study of avaloids onset on PISCES indicated strong correlation between the far SOL and the oscillation inside the main plasma column. We concluded that intermittency results from a saturation of an instability at the edge of the plasma. On CSDX we are able to observe the instability onset by using fast imaging. Avaloids are observed to be associated to the low poloidal numbers (m=1). Imaging confirmed that avaloids are emitted from the main plasma edge and offered a new image to avaloids as structures that are elongated in the poloidal plane with a finger-like rather than blobby shape. 7/28/2006 G. Y. Antar 27
Part III: Universality of avaloids by Comparising Tore Supra, Alcator C-MOD, MAST and PISCES G. Y. Antar, et al., Phys. Plasmas 10, 419 (2003). Tore Supra tokamak Alcator C-MOD tokamak MAST Spherical tokamak PISCES a = 76 cm, R = 2.32 m B T = 3.5 T, Ip =1 MA limiter machine a = 21 cm, R = 70 cm B T = 5.3 T I p ~ 0.8 MA divertor machine a = 52 cm, R = 73 cm B T = 0.6 T, I p =700 ka First wall far from the LCFS n e ~ 10 17 m -3, T e ~ 10 ev, B = 0.12-0.24 T Plasma radius = 2.5 cm Vessel radius = 10 cm 7/28/2006 G. Y. Antar 28
Plasma Far from the Last Closed Flux Surface in PISCES and MAST exists in form of intermittent bursts PISCES MAST 7/28/2006 G. Y. Antar 29
Similarity of the PDF of I sat fluctuations Gaussian for negative fluctuations Strongly Skewed for positive fluctuations Similarity of the power spectra of I sat One scaling region approximately the same scaling exponent - 1.6 Only large scales survive 7/28/2006 G. Y. Antar 30
Similarity of the avaloid temporal signature Non-conservation of mass Asymmetric shape (like ELM s or saw teeth) 7/28/2006 G. Y. Antar 31
The distribution of the time between two bursts is similar in the four devices: Approximately the same central frequency 200 khz Skewed form of the PDF showing higher probability for longer waiting times than from a Gaussian distribution No relation is detected between the waiting time and the amplitude of the bursts on the four devices. Another proof that SOC is not the process underlying the physics of avaloids. In addition, these structures exist where the profile is FLAT. 7/28/2006 G. Y. Antar 32
Conclusion of Part III The universality of intermittent convective transport was demonstrated by a detailed comparison, using 9 different statistical properties, of four different plasma devices with four different probe architectures. 7/28/2006 G. Y. Antar 33
Part IV: In Tokamaks, What is the poloidal distribution of avaloids? G. Y. Antar, et al., Phys. Plasmas 12, 023506 (2005). Diagnostics used on MAST: 12 target probes (TP) set to ion saturation current and acquired at 1 MHz. The reciprocating probe (RP) with measurement of the radial turbulent flux acquired at 1 MHz. Fast Imaging area Reciprocating probe Different Target probes used 7/28/2006 G. Y. Antar 34
As in PISCES the SOL of MAST is intermittent and plasma is recorded up to 30 cm away from the LCFS 7/28/2006 G. Y. Antar 35 r [m]
Unlike the mid-plane SOL, at the target intermittent structures seem not present 7/28/2006 G. Y. Antar 36
No intermittency recorded on the high field side. Even though intermittent structures are not clearly identified in the SOL at the target, the skewness and flatness factors increase with the distance to the strike point in agreement with the mid-plane SOL behavior. 7/28/2006 G. Y. Antar 37
Comparison of I sat taken at the mid-plane and at the target plates Mid-plane Target Higher particle fluxes at the mid-plane than at the targets. Higher turbulence levels at the mid-plane than at the targets. Turbulent fluctuations are more intermittent at the mid-plane than at the targets but the same behavior is detected. 7/28/2006 G. Y. Antar 38
Avaloids effect the whole low-field SOL appearing as localized perturbation in the poloidal plane similar to what was found on CSDX A poloidal cut around the LCFS in time Avaloids ELMs 7/28/2006 G. Y. Antar 39
Conclusion of Part IV Avaloids are poloidally asymmetric. They are affected by the magnetic configuration. Their intensity is very small at the target plates due to field lines expansion which transforms a localized structure at the mid-plane into an elongated one at the target. 7/28/2006 G. Y. Antar 40
Recent Progress in SOL understanding Intermittency in the SOL of magnetic fusion devices is caused by large-scale structures with large radial velocities that we call avaloids. Avaloids result from the non-linear evolution of a low poloidal number edge instability. Avaloids like normal turbulence are affected by the magnetic geometry and shear. Universality: Avaloids have the same properties on different magnetic fusion devices. 7/28/2006 G. Y. Antar 41
Publications G. Antar, et al., Turbulence Intermittency and Bursts Properties in the Scrape-off layer of the Tore Supra Tokamak, Phys. Plasmas 8, 1612 (2001). G. Y. Antar, et al., Experimental Evidence of Convective Transport in Fusion Devices, Phys. Rev. Lett. 87, 065001 (2001). G. Y. Antar, et al., Universality of intermittent convective transport in the scrape-off layer of magnetically confined devices, Phys. Plasmas 10, 419 (2003). G. Y. Antar, On the Origin of Intermittency in the scrape-off layer of linear magnetically confined devices, 10, 3629 (2003) 7/28/2006 G. Y. Antar 42