GA-A22065

DIVERTOR PARTICLE EXHAUST AND WALL INVENTORY ON DIII-D

by R. MAINGI, G.L. JACKSON, M.R. WADE, M.A. MAHDAVI,

P.K. MIODUSZEWSKI, G. HAAS, M.J. SCHAFFER, J.T. HOGAN, and C.C. KLEPPER

JUNE 1995

MASTER

.

GENE- ATOMfCS

DISCLAIMER

Portions of this document may be illegible electronic image products. Images are produced from the best available original document.

G A-A22065

DIVERTOR PARTICLE EXHAUST AND WALL INVENTORY ON DIN-D

by R. MAINGI; G.L. JACKSON, M.R. WADE: M.A. MAHDAVI,

P.K. MIODUSZEWSKI,~ G. HAAS! M.J. SCHAFFER, J.T. HOGAN? and C.C. KLEPPERt

This is a preprint of a paper presented at the 22nd European Conference on Controlled Fusion and Plasma Physics, June 3-7,1995, in Bournemouth, United Kingdom, and to be published in the H?OC€€D/NGS

*Oak Ridge Associated Universities.

+Oak Ridge National Laboratory. $Max Planck lnstitut for Plasmaphysik

Work supported by U.S. Department of Energy

under Contract Nos. DE-AC03-89ER51114 and DE-AC05-850R21400

GA PROJECT 3466 JUNE 1995

DIVERTOR PARTICLE EXHAUST AND WALL INVENTORY ON DIII-D

R. Main@.* G.L. Jackson, M.R. Wade,? M.A. Mahdavi, P.K. Mioduszewski,?

General Atomics, P.O. Box 85608, San Diego, CA 92186-9784

G. Haas,# M.J. Schaffer, J.T. Hogan,? C.C. Klepper?

Oak Ridge Associated Universities * f Oak Ridge National Laboratory

#Max Planck Institut for Plasmaphysik, Garching

Many tokamaks achieve optimum plasma performance by achieving low recycling;

various wall conditioning techniques including helium glow discharge cleaning (HeGDC) are

routinely applied to help achieve low recycling. Many of these techniques allow strong, transient wall pumping, but they may not be effective for long-pulse tokamaks, such as the

International Thermonuclear Experimental Reactor (ITER), the Tokamak Physics Experi- ment (TPX), Tore Supra Continu, and JTdOSU. Continuous particle exhaust using an in-situ pumping scheme may be effective for wall inventory control in such devices. Recent particle

balance experiments on the Tore Supra and DIII-D tokamaks demonstrated that the wall particle inventory could be reduced during a given discharge by use of continuous particle exhaust [ 1,2]. In this paper we report the first results of wall inventory control and good performance with the in-situ DIII-D cryopump, replacing the HeGDC normally applied between discharges.

To examine the global particle balance and the role of graphite walls, a series of discharges without HeGDC (following reference discharges with HeGDC) was executed on the DIII-D tokamak. After the inter-shot HeGDC was terminated, the wall inventory

gradually increased. Discharges were conducted with the cryopump off until burn-through

problems were encountered because the neutral pressure became too high during the current rampup phase, despite the elimination of the prefill gas puff. At this point, the divertor

cryopump was activated and discharges were executed in an effort to reduce the net wall inventory. These discharges were conducted in lower single-null divertor configuration and had the following plasma parameters: Ip = 1.5 MA, Bt = 2.1 T, pNB1 = 6.3 MW (ELMy, H-mode confinement discharges).

The DIII-D advanced divertor includes a toroidally symmetric biasing ring and baffle, which create a pumping plenum for the in-vessel helium cryogenic condensation pump [3]. The cryopump particle exhaust rate was optimized by placement of the outer divertor strike point near the entrance to the pump plenum [1,4]. Several diagnostics provided measurements of the neutral pressure in this plenum, thereby allowing for multiple estimates of the exhaust flux. In this paper, data is presented from both a fast time-response (T - 2 ms) neutral

pressure gauge [5] and a slower time-response, magnetically shielded capacitance manometer (T - 150 IIIS).

1

The wall particle loading rate during discharges is estimated as the difference between the measured input gas sources and sinks:

where S E I = energetic beam particle fueling,

Ss? =

dN e - -

cold particle fueling from gas in beam lines,

= gas puff fueling, %Uff

- - dlv, - ' neutral gas buildup rate,

neutral loss rate due to plasma formation, dt

dt sclyo = cryopump exhaust rate = R(P0) * Po

R(P0) = measured cryopumping speed,

PO = pump plenum neutral pressure,

Swall = wall pump rate.

The net wall loading over the duration of the discharge is readily obtained by integration

of the wall loading rate, Swall, defied in Eq. (1). The largest neutral particle source is the cold gas input during the current rampup phase and the largest sink is the cryopump exhaust (when activated). In the absence of strong divertor particle exhaust, the wall inventory showed a net increase of 150 torr-1 by the end of the reference discharge(#83742). With continuous particle exhaust, the net wall inventory was reduced at the end of the first active

cryopump discharge (#83757) by about 200 torr-1. The particle balance described above was applied to each discharge of the experiment. The amount of gas exhaust between discharges was small (-10-20 torr-1) during non-disruptive terminations [6] but was included in the

analysis. However, one disruptive termination of the plasma resulted in the evolution and pumpout of 420 torr-1 gas from the wall. In Fig. 1, the computed net wall loading increased after the reference discharge. Toward the end of the wall loading phase (discharges*#83751- #83755), the programmed main plasma density setpoint was increased to reduce the number of discharges required to arrive at the full wall capacity. Hence, the net wall loading per discharge increased during these discharges. The cryopump was activated after discharge

#83756; the subsequent dischaiges displayed a reduction in net wall inventory (wall

unloading phase). This reduction occurred using either the ion gauge or capacitance

manometer in the cryopump exhaust rate calculation. By the end of the wall unloading phase, the net wall depletion during the discharges approached zero, suggesting that an equilibrium wall inventory value was being reached.

Both exhaust flux estimates indicate that the net wall inventory approached or went

below the initial value at the end of the discharge sequence to within the error bars of the estimates. The difference between the ion gauge and capacitance manometer estimates may be explained by the presence of hydrocarbons in the pump plenum, which would cause the

2

ion gauge to indicate a higher neutral pressure for a given particle flux because of the

increased ionization effi- ciency for hydrocarbons as

compared with deuterium

1500 m Cryopump On

a3735 a3745 a3755 a3775 Discharge Number

Fig. 1. Cumulative wall loading vs. discharge number. deduced from ionization gauge (closed) and capacitance manometer (open).

molecules. Hence, the analy- sis indicates that the wall in-

ventory level with divertor pumping was effectively re-

stored to or reduced below

the value prior to the termin- ation of the HeGDC sessions.

The conclusion that the wall inventory was first increased and then reduced is supported by other data. On DIII-D the gas input required to reach a prescribed ohmic target density

during the plasma startup phase is determined by the digital feedback system and affected by

wall conditions. If it is assumed that the wall outgassing rate increases with the wall inven- tory, then the external gas input required to reach the target density is expected to increase as the wall inventory decreases. As displayed in Fig. 2, the reqked gas did decrease during the

phase without HeGDC, i.e. as the wall inventory was increased during the wall loading phase.

The gas directly pumped by the cryopump has been subtracted when the croypump was on.

Correspondingly, the required gas increased during the.phase of discharges with the active

cryopump, i.e. as the wall inventory was reduced during the wall unloading phase. The data in Fig. 2 also suggest that the wall was nearly saturated at the end of the wall loading phase because the gas input required during the rampup phase approached- the plasma inventory

(20 torr-e), indicating a very high fueling efficiency (2 60%). It is worth noting that excellent

density control was re-established within a few discharges after the cryopump was activated.

During, the wall loading phase of the discharge se- quence, the plasma stored en-

ergy in the ELMy phase of the first discharge without preced-

ing HeGDC was reduced by -15% as compared with the reference discharges (Fig. 3).

120 n 2 100 s t: 80

60 a

I " " " . " I 'I

- -

- i

- a 40 - a n m off

Subsequent discharges exhib- a3735 83745 a3755 a3765 a3775

ited roughly the same stored Discharge Number

energy, i.e. the stored energy Fig. 2. Gas input required to achieve the ohmic density vs. discharge number.

did not degrade as the net wall

3

loading increased. The plasma stored energy on the first two

discharges with the cryopump on (#83757, #83758) was restored to the value of the ref-

erence discharges; the follow- ing discharges (#83761-

#83767) showed a modest

decline in the stored energy because cryopump operation

1.4 7 1.3 z 1.2 6 1.1 f 1 w' 0.9 2 0.8 Q 0.7

0.6

m

83

I ' " " ' " * I ' " " 1 1 ' ' I ' " ' 1 ' " '

He Glow He Glow Off

Cryopump On 2-9 . * . . . .

83765 83775 I. ,l,m,,, ,, . I . , , , , ( ) ) , . . I . . ,

He Glow On

735 83745 83755 Discharge Number

Fig. 3. Plasma stored energy vs. discharge number.

reduced the line-average plasma density. This result is expected because of previous work [7] indicating that the thermal confinement time has a weak density dependence ( z,h = E:18).

It is evident from the data and analysis that the wall particle inventory can be controlled by continuous divertor cryopump operation, even in the absence of inter-shot HeGDC. Coupled with the observation that density control is maintained in this mode of operation, the data suggest that a pumping, Le. low recycling, wall can be maintained for a substantial

period of time. This result is particularly attractive from the standpoint that next generation devices will not be able to easily turn off super-conducting magnet coils in order to perform HeGDC for particle control in ELMy H-mode discharges, i.e. long-pulse particle control and reasonably high stored energy can be obtained by divertor pumping alone. It cannot be

inferred from these data, however, that HeGDC is not needed for peak plasma performance in ELM-free plasmas. That experiment awaits installation of new hardware which will allow efficient pumping of high-triangularity VH-mode discharges.

This is a report of work supported by the U.S. Department of Energy under Contract

Nos. DE-AC03-89ER5 11 14 and DE-AC05-850R21400. The first author is supported by a U.S. DOE Magnetic Fusion Energy Postdoctoral Fellowship administered by Oak Ridge Associated Universities.

151

Mahdavi M A et al., Proc. of 8th European Physical Society Conf. 26-30 July 1993,

Part II, p. 647

Mioduszewski P K et al., to be published in the Proc. of the'1994 Plasma-Surface Interactions Conference, Mito, Japan, 23-27 May 1994 Smith J P et al., Fusion Technol. 21 1638 (1992)

Maingi R et al., to be published in the Proc. of the 1994 Plasma-Surface Interactions

Conference, Mito, Japan, 23-27 May 1994 Haas J et al., J. Nucl. Mater 121 151 (1984); KZepper C C et al., J. Vac. Sci. Technol. A

11 446 (1993) Maingi R et al., "Control of Wall Particle Inventory with Divertor Pumping on DIII-D," General Atomics Report GA-A22014, April (1995), submitted to Nucl. Fusion

Schissel D P et al., Nucl. Fusion 34 1401 (1994)

4