DIVERTOR INVESTIGATIONS ON NSTX-U LEADING TO FNSF

Mike Kotschenreuther

Brent Covele

Swadesh Mahajan

Prashant Valanju

Jonathan Roeltgen

Zhong-Ping Chen

The University of Texas at Austin

Institute for Fusion Studies

ADVANCED DIVERTOR INVESTIGATIONS ON NSTX-U

• Overarching goals: experimentally investigate advanced

divertors on NSTX, to be able to create a validated ability to

design an FNSF divertor

• Well known challenges of the “standard divertor” (SD) for STs:

→Low wetted area due to low R (especially with high triangularity d) => high heat flux

→Low connection length (“line length”); in standard models, this leads to high target

plate temperature, difficulty radiating/detaching, and high erosion

• These challenges must be overcome by appropriate poloidal

field geometries in the ST divertor – “advanced divertors”

→Examine these in conjunction with Li targets – another attractive solution

→XD, Snowflake are our primary advanced divertor geometries of interest for NSTX

now

DIVERTOR PERFOMANCE- EXPERIMENTAL STUDIES AND COMPARISONS

• Let us, for example, label different geometries by

→DI (Flaring versus converging- relative to standard divertor)

→Flux expansion at the plate FPLATE

(NOTE: flux expansion and flaring (DI) are unrelated quantities, measuring different

aspects of the divertor flux surface shape)

→L|| (“connection length” or “line length”- important in standard models)

→LDLEG (Length of the divertor leg – in poloidal plane)

• What is needed is to compare the divertor properties of

geometries with different DI, LDLEG, and L||

• Past prediction of UT group: flux surface shape (i.e DI) is

important to detachment performance (not just flux exp. Fplate)

DETACHMENT CAUSED “INSTABILITY” IN STANDARD DIVERTOR (HEURISTIC EXPLANATION)

• As the cold detachment front moves upstream, the interaction area with neutrals increases

• Hence, the atomic dissipation increases, which causes the temperature to get colder

• This positive feedback moves the cold front from the divertor to core X-point

• This is the edge of the H-mode- if the core boundary is cooled enough, H-mode barrier is degraded

ENABLING “STABLE” FULL DETACHMENT: X-DIVERTOR PRESCRIPTION

stays

• Create a 2nd x-point in the SOL downstream, increasing field line flaring at the target plates

• Change the feedback the detachment front “sees” if it migrates upstream

• Also, line length decreases greatly as the front moves away from the plate, adding further “stabilizing” feedback against front migration

• With flared field lines, more of the radiating region stays away from the H-mode barrier- confinement degradation is avoided

DIVERTOR PERFOMANCE- DETACHMENT EFFECTS ON CONFINEMENT

EXPERIMENTAL STUDIES AND COMPARISONS

• University of Texas: construct equilibria using CORSICA to enable

experimental tests (in close cooperation with NSTX-U)

• Use SOLPS to both help design experiments and interpret them

• Experimentally test the effect of detachment on confinement for

different geometries (different DI, FPLATE, L||, and LDLEG - and other

parameters that may be of interest)

→Which geometries allow the highest detachment with the highest confinement (e.g.

pedestal pressure)

→Which geometries allow detachment at the lowest density (for the highest current drive

efficiency)

→Can this be achieved for low recycling Li operation?

→What is the effect of impurity seeding?

Recent XD experiments on DIII-D

• Recent XD experiment on DIII-D (D2 puffing only):

Higher flaring (DI) allowed higher detachment/lower heat

flux with much higher confinement (pedestal pressure)

• Comparison of NSTX and DIIID is very valuable- allows

elucidation of effects of important variables

→STs naturally have much lower L|| than normal A: testing similar

geometries on NSTX and DIIID allows a large variation in their

values to elucidate what is important

• On NSTX we would like to reproduce/improve upon the

DIIID results

Scan of D2 puff rates in XD and SD in DIII-D: Differences in heat flux (shown vs corresponding pedestal pressure for given puff rate)

SDXD

High DI: Large difference in PEDESTAL PRESSURE for low heat flux

SDXD

Lower DI: Smaller difference in PEDESTAL PRESSURE for low heat flux

High DI – SD comparison Lower DI – SD comparison

XD SDSDXD

Large difference in PEDESTAL PRESSURE for high DI Compared to SD detachment

Less difference in PEDESTAL PRESSURE for low DI (Divertor Thompson-confirms w less scatter)

High DI – SD comparison Lower DI – SD comparison

Scan of D2 puff rates in XD and SD in DIII-D: Diffs in detachment (probe jsat)(shown vs corresponding pedestal pressure for given puff)

Important aspects of DIIID results to inform NSTX-U experiments

• The high DI cases that showed strong detachment with high confinement

have:

→ LESS flux expansion than the low DI (!!)

→ Hence: flux surface flaring as parameterized by DI appears important to obtaining

detachment with good H-mode confinement; NOT flux expansion

→ The high DI case has only slightly higher L|| (~20%) than low DI – possible indication L|| may not

the crucial variable

→ So, can STs with much less L|| get the same advantages from XDs as normal A?

• UT- construct equilibrium on NSTX-U to vary different quantities (DI, Fplate,

LDLEG, and L||) to test significance experimentally

• Also, NSTX experiments should consider non-geometrical variables (not yet

tested on DIII-D)

→Puffing location

→Seeded impurities

→Li vs B

NSTX-UPGRADE

• These are our (Brent’s) initial attempts

• Brent has been interacting with Stefan Gerhardt and has made considerable progress in conforming to NSTX-U machine constraints

• Also would like to work with Egemen Kolemen in future DI ≈ best DIII-D cases

Use SOLPS to help interpret results

• Give simulation results as to the relative

importance of L|| ,LDLEG , FPLATE

• Help to design NSTX experiments and interpret NSTX results

SOL MHD stability: ballooning ??• NSXT has higher SOL ideal MHD ballooning parameter a than DIIID (on the

outboard mid-plane)

→Ideal/resistive MHD could play a role in increasing turbulence at midplane OR

near the plate during detachment

→Divertor geometry can also effect stability

• This could play an important role in NSTX

• With SOL width ~ 1/Bpol, this effect should get much stronger on the way from NSTX -> FNSF

• Not part of the original grant, but we are interested in developing developing a (1D) SOL ideal/resistive ballooning code

• This should require only modest effort but may give large impact, if

instabilities are found to be strong

→Could be especially important for accurately projecting from NSTX to FNSF

• Feedback on the desirability of this from NSTX team???

CONCLUSIONS• It is important to experimentally test a wide range of divertor

configurations on NSTX-U

• Of particular importance: to demonstrate cases with strongly detached outer divertor legs along with excellent H-mode confinement→ Including X-divertor configurations with longer divertor legs and the 2nd X-pt near the

plate

→ Correlate SOL flux surface shape parameters with the ability to attain divertor detachment (as strong as possible) together with excellent H-mode confinement

• We will/are working to develop MHD equilibria to allow experimental testing of this on NSTX-U

• We will also employ SOLPS in the future to help in experimental design and interpretation of results

ITER X-Divertors: (a) The ITER standard divertor baseline equilibrium. (b) The original 2004 version of the ITER XD with special PF coils near the targets. (c) A 2013 XD with outer flux expansion optimized for an incident B field at 1 degree at the outer target. (d) A 2013 XD with maximal outer flux expansion within coil current limits and with 15 MA of plasma current. (e) A 2013 XD with maximal outer flux expansion within coil current limits and with 14 MA of plasma current.

DI=1.88 DI=1.64DI=1.05 DI=1.74 DI=2.04

XD cases within ITER coil current limits & with baseline hardware An XD on ITER:

Plasma and divertor parameters for the ITER X-Divertor equilibria. The minor radius was reduced in the 2013 XD plasmas to maintain 15 cm of clearance from the first wall. Confinement times are computed using the ITER98(y,2) scaling, with a density 80% of the Greenwald limit [21] and an assumed heating power of 120MW.