- Email: [email protected]
Disruptions in TFTR
773
Journal of Nuclear Materials 176 & 177 (1990) 773-778 North-Holland
Disruptions
in TFTR
A.C. Janos, E. Fredrickson, M. Corneliussen, Nagayama, D.K. Owens and M. Ulrickson Princeton Plasma Physics Laboratory,
Princeton
K.M. McGuire, M. Bell, A. Cavallo, Y.
University, P.O. Box 451, Princeton, NJ 08543, USA
This paper addresses the processes during disruptions of plasmas in the Tokamak Fusion Test Reactor (TFTR), including precursor phenomena, and the resulting effects which are of concern in future devices, such as heat loads and runaways. The pre-disruptive behavior varies for disruptions due to operation at different operating limits such as (a) bigb density, (b) low qa, and (c) high poloidal beta j3,. Power loading and runaway electron impact on the fist wall of a tokamak are important issues in present research devices and in the design of future devices such as CIT and ITER. In particular, disruptions can deposit large amounts of power in local&d areas which can then damage the wall or adversely affect operation. The time scales and spatial profiles of heat deposition on the fist wall have been studied with a poloidal array of infrared detectors and an extensive array of thermocouples providing coverage of the entire axi-symmetric, primary power-handling limiter in TFTR the bumper limiter. While there were often common hot spots, different types of dismptions caused different heating patterns and heat loads on the bumper limiter.
1. Introduction The reduction and amelioration of disruptions are important issues for present research devices and for the design of future devices such as CIT and ITER [l].
Disruptions can deposit large amounts of power in localized areas which can damage the wall or adversely affect operation. Disruptions are more likely to occur near one of the operating limits (low qa, high density, or high j3). While operation away from these limits can minimke the likelihood of a disruption, there is the chance of a disruption due to a wide variety of problems such as walI fragments falling into the plasma (possibly not unlike density-limit disruptions). It would be valuable to predict the onset of disruptions and also to characterize how a plasma disrupts onto the wall so that steps can be taken to minimk the damage. Disruptions in TFTR vary considerably in detail but have some common features. All disruptions exhibit a rapid (< 100 as) drop in the electron temperature, often known as the thermal quench, in which the plasma temperature decreases every-where within the plasma. For those disruptions which take longer to lose plasma stored energy, each disruption actually consists of several discrete, fast events. Disruptions begin with the thermal quench (transport) phase. A fast positive current spike is concurrent with or follows the thermal quench. A slow current decay then occurs. The time for
decay (tens of msec) is considerably longer than the thermal quench. The rate of current decay increases nearly linearly with current, suggesting that there is a characteristic decay time. Bursts of hard X-rays are typically observed during the current quench. During these bursts, the current decay seldom hesitates but instead continues falling linearly or exponentially to zero. the current
2. Precursors The pre-disruptive behavior varies for the different types of disruptions. Many disruptions occurred without observed precursors. Disruptions more often had modes present during or after the transport phase than before the transport phase. High- j3 disruptions (fig. 1) typically have either no precursor or a very rapidly growing (m, n) = (1, 1) mode (growth time on the order of one msec). The disruption then consists of one very fast thermal quench, followed by the current quench. The thermal quench can occur in less than one millisecond. There was no clear evidence that the disruptive instability began at some radius and spread, as might be expected if low m tearing modes were growing and creating an ever expanding ring of ergodic field lines. Rather, the whole profile began to decay at the same
0022-3115/90/$03.50 0 1990 - Elsevier Science Publishers B.V. (North-Holland)
Soft
x-ray
Mirnov Coil
3.915
3.920
Time (seconds)
E
Fig. 1. (a) A typical high /3 disruption with an (m, n) = (1, 1) precursor. (b) Analysis of the (m, n) = (1, 1) mode with the soft X-ray and Mirnov diagnostics. The mode has one phase inversion at the plasma center, consistent with an internal kink structure. This mode is coupled out to the plasma edge as an (m, n) = (6, 1) magnetic mode, strongly distorted due to high lambda and toroidal effects. time (on a 100 gs timescale), as if transport had suddenly deteriorated over the whole plasma volume. In many cases, whatever instabi~ty was enhancing the transport would subside and the plasma would attempt to reheat for a moment. In most cases, though, the plasma suffered further disruptions until the discharge terminated. In high &lisruptions, there is often a large, short burst of broadband emission at the second harmonic electron cyclotron frequencies which may occur up to 0.5 ms before the electron energy transport phase. High density disruptions (fig. 2) also typically show no precursor activity, of either (2, 1) or (3, 1) tearing mode nature, prior to the first partial thermal quench; this is in contrast to the JET observations [2]. However, they typically dump the stored energy in a series of partial thermal quench events during which low m MHD activity [(m, n) = (3, 1) or (2, I)] is often observed. These high density disruptions are characteristi-
cally radiation collapse disruptions since the radiated power fraction often reaches or exceeds 100%. More typically, they exhibit a collapse of the plasma edge temperature. Low qa disruptions (fig. 3) are more similar to high-j3 disruptions than to high density disruptions. Occasionally, the disruption is preceded by a flattening of the temperature profile near the q = 1 surface or less often by a slow cooling of the plasma edge. The transition to the actual disruption is discontinuous, so it is not clear what relation, if any, these observations have to the disruption mechanism.
3. Power iouding on the bumper limiter The plasma-facing wall in the TFTR is covered in large part by a bumper limiter [3] as shown in fig. 4. This limiter extends the full 360* toroidally inside the
115
A.C. Janos et al. / Disruptions in TFTR
(a)
I
I
(b) w
m/n=3/1
3
d XC Irn -Rn
I
---YWI
270
250
25 10
z .-
c ___.-.-
j
Neutral Beam Power / .-
. . . .._.__
_ __.__...
-.
_.’
_.
_
\‘_ 230
___.....’
Radiated Power y
a
210 2.90
2.95
3.00 Timer(sec)
3.072
3.074
3.076
3.078
3.080
3.082
Time (set)
current, loop voltage, electron temperature and radiated power for a high density disruption, nR/B - 3.5 for of electron temperature during the transport phase of the disruption. Note the step like decay with reheat disruptions. The top trace is the integrated Mimov signal showing a large (m/n) = (3/S) mode during the decay.
Fig. 2. (a) The plasma
q(a) = 3.3. (b) Contours between
(4
2.or
04 290
z
280
8
I
-10
4
S I m k
I
I
0
3 5 2 270
\
.-S
2
z” 1:
I
260
I
250 _._._......__. .. t Radiated Power
a 0 3.30
3.35
3.40 Time (set)
240
I
3.45
3.50
3.440
3.442
3.444
3.446
3.448
3.450
Time (set)
Fig. 3. (a) Low qa ( - 2.5) disruption. (b) Low qa discharge with a coherent oscillation near the q =1 surface. - 5 ms before the final disruption. No other precursor was observed.
A flat spot develops
A.C. Jams
et al. / Disruptions
in TFTR
Heat deposition on the limiter, as determined by temperature rises measured by fast IR detectors, begins during the thermal quench and may extend throughout the current decay phase. A current spike may occur at or after the start of heat deposition on the limiter. On a fast time scale, the heat deposition is extremely non-uniform. Measurements of the 2-D distribution of power loading on the entire TFTR bumper limiter, as a result of disruptions, can now be made for each discharge [4]. TFTR has the unique feature that the bumper limiter is extensively instrumented with thermocouples (fig. 5). In each bay, thermocouples are located at 0 O, k 15 o and f 45 o with respect to the midplane. The difference in energy in the tiles before and after each shot is calculated, based on the temperature-dependent heat capacity, to yield power loading and estimates of total energy deposited. Based on fast magnetic (toroidal average) data and soft X-ray imaging data, all (types of) disruptions moved inward in major radius and therefore would be expected to heat primarily the bumper limiter. Hot spots were observed. Very localized hot spots indicated that a hot spot could be confined to one bay. Often, the hot spots extended across several bays. Heat deposition distribution varies depending on the type of disruption. In particular, heating is minimal and very uniform for high-density disruptions, compared to other types of disruptions.
Fig. 4. Schematic of the interior features of TFTR, including the bumper limiter.
vacuum vessel and f60” poloidally with respect to the midplane. It is the primary power-handling limiter. It is divided into 20 bays toroidally and is comprised of 1920 carbon tiles.
BAY
BCDEFGHI
J
E” P=z
1989
i?:: “0
L
MNOPOASTA
t t
t 2
K
ii
6
t B
5
w
::
::
3
& i
6
4
a
“2 cd ::
E 2 D 5.
5 e
$
I
2
Fig. 5. Schematic of the bumper limiter unfolded to fit a rectangle with coordinates of “bay” and “poloidal angle” with respect to the midplane. Locations of the 100 thermocouples are indicated by solid square dots. Also indicated are features which are used to help understand the observed heating patterns.
ABCDEFGHIJKLMNOPPRST BAY
-5
Temperature
Change
(“C)
*’
Fig. 6. Increase in temperature of the bumper limiter during an ohmic discharge (no neutral beams) with a disruption at 1.15 MA during ramp-down of the plasma current.
Between
disruptions
of the same
type,
there
is a
significant amount of variation in the heat patterns while there are some axon patterns. The similarities
-5
Temperature
with some specific known mechanical feature on the bumper limiter such as high spots or spots next to recessed areas.
are often hot spots associated
Change
(93
”
Fig. 7. Increase in temperature of the bumper limiter during an ohmic discharge (no neutral beams) with a high-density disruption at
1.8 MA.
778
A. C. Janos et al.
Hot spots (often near the midplane) often occur with adjacent cool spots on each side of them within the same bay. This may be due to distortion of the bay (the most precise alignment is done along the midplane) or due to shielding of those spots or increased cooling due to material generated from the hot spots. An example of a discharge with a disruption is shown in fig. 6. The 1.8 MA current flat-top discharge disrupted during the ramp-down of the plasma current at 1.15 MA. There are numerous localized hot spots near the midplane which correlate with mechanical features (fig. 5). Based on the ohmic input power and radiated power, the input energy to the bumper limiter was 5 MJ up to just prior to the disruption, and 8 MJ total, If the heating were uniform, these energies would correspond to temperature changes of 5 and 8°C respectively. In general, high-density disruptions (fig. 7) have relatively little impact on the heating of the bumper limiter. The heating is very uniform (very few hot spots) and the temperature rise is small. The plasma current at disruption is 1.8 MA. Compared with the previous disruption, the peak bumper limiter temperature increase is much lower (15°C compared to 25°C) and the heat distribution is much more uniform. The cool spots above and below the bay I hot spot are obvious. The low power loading of these high-density disruptions is consistent with the fact that much of the energy is radiated away. During the last part of the discharge in fig. 7, the radiated power fraction, Prad/Pheat, is 100%.
’ Disruptions
in TFTR
4. Statistics of disruptions
In the first years of operation, - 20% of TFTR plasmas disrupted during the programmed flattop period. This number has declined in recent years to - 5% or less. This is due in part to increased experience in operation, improvements in control, and a shift from high current, L-mode discharges to low current Supershot ones. While there is no direct dependance of disruption frequency on plasma current, disruptions are more frequent at low qa.
Acknowledgements
Work supported AC02-76-CH03073.
by US. DOE
Contract
No.
DE-
References VI Iter-Related Physics R and D Needs 1991/92 (and Beyond), ITER-IL-PH-16-O-18 (March 1990). 121 J.A. Wesson et al., Nucl. Fusion 29 (1989) 641. [31 G.W. Barnes et al., in: Proc. of the 13th Symp. on Fusion Energy, Knoxville, TN., October, 1989, to be published. 141 A. Janos, et al., Rev. Sci. Instr. 61 (1990) 2973.
Journal of Nuclear Materials 176 & 177 (1990) 773-778 North-Holland
Disruptions
in TFTR
A.C. Janos, E. Fredrickson, M. Corneliussen, Nagayama, D.K. Owens and M. Ulrickson Princeton Plasma Physics Laboratory,
Princeton
K.M. McGuire, M. Bell, A. Cavallo, Y.
University, P.O. Box 451, Princeton, NJ 08543, USA
This paper addresses the processes during disruptions of plasmas in the Tokamak Fusion Test Reactor (TFTR), including precursor phenomena, and the resulting effects which are of concern in future devices, such as heat loads and runaways. The pre-disruptive behavior varies for disruptions due to operation at different operating limits such as (a) bigb density, (b) low qa, and (c) high poloidal beta j3,. Power loading and runaway electron impact on the fist wall of a tokamak are important issues in present research devices and in the design of future devices such as CIT and ITER. In particular, disruptions can deposit large amounts of power in local&d areas which can then damage the wall or adversely affect operation. The time scales and spatial profiles of heat deposition on the fist wall have been studied with a poloidal array of infrared detectors and an extensive array of thermocouples providing coverage of the entire axi-symmetric, primary power-handling limiter in TFTR the bumper limiter. While there were often common hot spots, different types of dismptions caused different heating patterns and heat loads on the bumper limiter.
1. Introduction The reduction and amelioration of disruptions are important issues for present research devices and for the design of future devices such as CIT and ITER [l].
Disruptions can deposit large amounts of power in localized areas which can damage the wall or adversely affect operation. Disruptions are more likely to occur near one of the operating limits (low qa, high density, or high j3). While operation away from these limits can minimke the likelihood of a disruption, there is the chance of a disruption due to a wide variety of problems such as walI fragments falling into the plasma (possibly not unlike density-limit disruptions). It would be valuable to predict the onset of disruptions and also to characterize how a plasma disrupts onto the wall so that steps can be taken to minimk the damage. Disruptions in TFTR vary considerably in detail but have some common features. All disruptions exhibit a rapid (< 100 as) drop in the electron temperature, often known as the thermal quench, in which the plasma temperature decreases every-where within the plasma. For those disruptions which take longer to lose plasma stored energy, each disruption actually consists of several discrete, fast events. Disruptions begin with the thermal quench (transport) phase. A fast positive current spike is concurrent with or follows the thermal quench. A slow current decay then occurs. The time for
decay (tens of msec) is considerably longer than the thermal quench. The rate of current decay increases nearly linearly with current, suggesting that there is a characteristic decay time. Bursts of hard X-rays are typically observed during the current quench. During these bursts, the current decay seldom hesitates but instead continues falling linearly or exponentially to zero. the current
2. Precursors The pre-disruptive behavior varies for the different types of disruptions. Many disruptions occurred without observed precursors. Disruptions more often had modes present during or after the transport phase than before the transport phase. High- j3 disruptions (fig. 1) typically have either no precursor or a very rapidly growing (m, n) = (1, 1) mode (growth time on the order of one msec). The disruption then consists of one very fast thermal quench, followed by the current quench. The thermal quench can occur in less than one millisecond. There was no clear evidence that the disruptive instability began at some radius and spread, as might be expected if low m tearing modes were growing and creating an ever expanding ring of ergodic field lines. Rather, the whole profile began to decay at the same
0022-3115/90/$03.50 0 1990 - Elsevier Science Publishers B.V. (North-Holland)
Soft
x-ray
Mirnov Coil
3.915
3.920
Time (seconds)
E
Fig. 1. (a) A typical high /3 disruption with an (m, n) = (1, 1) precursor. (b) Analysis of the (m, n) = (1, 1) mode with the soft X-ray and Mirnov diagnostics. The mode has one phase inversion at the plasma center, consistent with an internal kink structure. This mode is coupled out to the plasma edge as an (m, n) = (6, 1) magnetic mode, strongly distorted due to high lambda and toroidal effects. time (on a 100 gs timescale), as if transport had suddenly deteriorated over the whole plasma volume. In many cases, whatever instabi~ty was enhancing the transport would subside and the plasma would attempt to reheat for a moment. In most cases, though, the plasma suffered further disruptions until the discharge terminated. In high &lisruptions, there is often a large, short burst of broadband emission at the second harmonic electron cyclotron frequencies which may occur up to 0.5 ms before the electron energy transport phase. High density disruptions (fig. 2) also typically show no precursor activity, of either (2, 1) or (3, 1) tearing mode nature, prior to the first partial thermal quench; this is in contrast to the JET observations [2]. However, they typically dump the stored energy in a series of partial thermal quench events during which low m MHD activity [(m, n) = (3, 1) or (2, I)] is often observed. These high density disruptions are characteristi-
cally radiation collapse disruptions since the radiated power fraction often reaches or exceeds 100%. More typically, they exhibit a collapse of the plasma edge temperature. Low qa disruptions (fig. 3) are more similar to high-j3 disruptions than to high density disruptions. Occasionally, the disruption is preceded by a flattening of the temperature profile near the q = 1 surface or less often by a slow cooling of the plasma edge. The transition to the actual disruption is discontinuous, so it is not clear what relation, if any, these observations have to the disruption mechanism.
3. Power iouding on the bumper limiter The plasma-facing wall in the TFTR is covered in large part by a bumper limiter [3] as shown in fig. 4. This limiter extends the full 360* toroidally inside the
115
A.C. Janos et al. / Disruptions in TFTR
(a)
I
I
(b) w
m/n=3/1
3
d XC Irn -Rn
I
---YWI
270
250
25 10
z .-
c ___.-.-
j
Neutral Beam Power / .-
. . . .._.__
_ __.__...
-.
_.’
_.
_
\‘_ 230
___.....’
Radiated Power y
a
210 2.90
2.95
3.00 Timer(sec)
3.072
3.074
3.076
3.078
3.080
3.082
Time (set)
current, loop voltage, electron temperature and radiated power for a high density disruption, nR/B - 3.5 for of electron temperature during the transport phase of the disruption. Note the step like decay with reheat disruptions. The top trace is the integrated Mimov signal showing a large (m/n) = (3/S) mode during the decay.
Fig. 2. (a) The plasma
q(a) = 3.3. (b) Contours between
(4
2.or
04 290
z
280
8
I
-10
4
S I m k
I
I
0
3 5 2 270
\
.-S
2
z” 1:
I
260
I
250 _._._......__. .. t Radiated Power
a 0 3.30
3.35
3.40 Time (set)
240
I
3.45
3.50
3.440
3.442
3.444
3.446
3.448
3.450
Time (set)
Fig. 3. (a) Low qa ( - 2.5) disruption. (b) Low qa discharge with a coherent oscillation near the q =1 surface. - 5 ms before the final disruption. No other precursor was observed.
A flat spot develops
A.C. Jams
et al. / Disruptions
in TFTR
Heat deposition on the limiter, as determined by temperature rises measured by fast IR detectors, begins during the thermal quench and may extend throughout the current decay phase. A current spike may occur at or after the start of heat deposition on the limiter. On a fast time scale, the heat deposition is extremely non-uniform. Measurements of the 2-D distribution of power loading on the entire TFTR bumper limiter, as a result of disruptions, can now be made for each discharge [4]. TFTR has the unique feature that the bumper limiter is extensively instrumented with thermocouples (fig. 5). In each bay, thermocouples are located at 0 O, k 15 o and f 45 o with respect to the midplane. The difference in energy in the tiles before and after each shot is calculated, based on the temperature-dependent heat capacity, to yield power loading and estimates of total energy deposited. Based on fast magnetic (toroidal average) data and soft X-ray imaging data, all (types of) disruptions moved inward in major radius and therefore would be expected to heat primarily the bumper limiter. Hot spots were observed. Very localized hot spots indicated that a hot spot could be confined to one bay. Often, the hot spots extended across several bays. Heat deposition distribution varies depending on the type of disruption. In particular, heating is minimal and very uniform for high-density disruptions, compared to other types of disruptions.
Fig. 4. Schematic of the interior features of TFTR, including the bumper limiter.
vacuum vessel and f60” poloidally with respect to the midplane. It is the primary power-handling limiter. It is divided into 20 bays toroidally and is comprised of 1920 carbon tiles.
BAY
BCDEFGHI
J
E” P=z
1989
i?:: “0
L
MNOPOASTA
t t
t 2
K
ii
6
t B
5
w
::
::
3
& i
6
4
a
“2 cd ::
E 2 D 5.
5 e
$
I
2
Fig. 5. Schematic of the bumper limiter unfolded to fit a rectangle with coordinates of “bay” and “poloidal angle” with respect to the midplane. Locations of the 100 thermocouples are indicated by solid square dots. Also indicated are features which are used to help understand the observed heating patterns.
ABCDEFGHIJKLMNOPPRST BAY
-5
Temperature
Change
(“C)
*’
Fig. 6. Increase in temperature of the bumper limiter during an ohmic discharge (no neutral beams) with a disruption at 1.15 MA during ramp-down of the plasma current.
Between
disruptions
of the same
type,
there
is a
significant amount of variation in the heat patterns while there are some axon patterns. The similarities
-5
Temperature
with some specific known mechanical feature on the bumper limiter such as high spots or spots next to recessed areas.
are often hot spots associated
Change
(93
”
Fig. 7. Increase in temperature of the bumper limiter during an ohmic discharge (no neutral beams) with a high-density disruption at
1.8 MA.
778
A. C. Janos et al.
Hot spots (often near the midplane) often occur with adjacent cool spots on each side of them within the same bay. This may be due to distortion of the bay (the most precise alignment is done along the midplane) or due to shielding of those spots or increased cooling due to material generated from the hot spots. An example of a discharge with a disruption is shown in fig. 6. The 1.8 MA current flat-top discharge disrupted during the ramp-down of the plasma current at 1.15 MA. There are numerous localized hot spots near the midplane which correlate with mechanical features (fig. 5). Based on the ohmic input power and radiated power, the input energy to the bumper limiter was 5 MJ up to just prior to the disruption, and 8 MJ total, If the heating were uniform, these energies would correspond to temperature changes of 5 and 8°C respectively. In general, high-density disruptions (fig. 7) have relatively little impact on the heating of the bumper limiter. The heating is very uniform (very few hot spots) and the temperature rise is small. The plasma current at disruption is 1.8 MA. Compared with the previous disruption, the peak bumper limiter temperature increase is much lower (15°C compared to 25°C) and the heat distribution is much more uniform. The cool spots above and below the bay I hot spot are obvious. The low power loading of these high-density disruptions is consistent with the fact that much of the energy is radiated away. During the last part of the discharge in fig. 7, the radiated power fraction, Prad/Pheat, is 100%.
’ Disruptions
in TFTR
4. Statistics of disruptions
In the first years of operation, - 20% of TFTR plasmas disrupted during the programmed flattop period. This number has declined in recent years to - 5% or less. This is due in part to increased experience in operation, improvements in control, and a shift from high current, L-mode discharges to low current Supershot ones. While there is no direct dependance of disruption frequency on plasma current, disruptions are more frequent at low qa.
Acknowledgements
Work supported AC02-76-CH03073.
by US. DOE
Contract
No.
DE-
References VI Iter-Related Physics R and D Needs 1991/92 (and Beyond), ITER-IL-PH-16-O-18 (March 1990). 121 J.A. Wesson et al., Nucl. Fusion 29 (1989) 641. [31 G.W. Barnes et al., in: Proc. of the 13th Symp. on Fusion Energy, Knoxville, TN., October, 1989, to be published. 141 A. Janos, et al., Rev. Sci. Instr. 61 (1990) 2973.