Fitting transport models to 14MeV neutron camera data

11
Fitting transport models to 14MeV neutron camera data D C McDonald, K D Zastrow and I Voitsekhovitch

description

Fitting transport models to 14MeV neutron camera data. D C McDonald, K D Zastrow and I Voitsekhovitch. KN3 data for T puff. Puff diffuses to core. puff seen by KS3. KN3 D-T neutron flux (vert.). Inner channels see puff later than outer ones. Time [s]. Methods for fitting transport models. - PowerPoint PPT Presentation

Transcript of Fitting transport models to 14MeV neutron camera data

Page 1: Fitting transport models to 14MeV neutron camera data

Fitting transport models to 14MeV neutron camera data

D C McDonald, K D Zastrow and I Voitsekhovitch

Page 2: Fitting transport models to 14MeV neutron camera data

Time [s]

KN

3 D

-T n

eutr

on f

lux

(ver

t.)

KN3 data for T puff

Inner channels see puff later than outer ones

puff seen by KS3

Puff diffuses to core

Page 3: Fitting transport models to 14MeV neutron camera data

• TRANSP models the expected 14MeV neutrons from given diffusion and pinch profiles

– Time consuming to optimise the profiles– not clear how to discuss errors

Methods for fitting transport models

choose a D(r) and v(r)

run TRANSP to produce nT(r,t) and RKN3

i(t)

run SANCO to produce nT(r,t) then calculate RKN3

i(t)

choose initial D(r) and v(r)

calculate 2 and choose new D(r) and v(r)

• Automatically finds an optimised parameterised solution with correlated errors

• Initial version only treats neutron reactivity in 1d

UTC

TRANSP

Page 4: Fitting transport models to 14MeV neutron camera data

stronger inboard emissivity in core

similar inboard/outboard

emissivity at r/a0.6

TRANSP fast ion density r and dependencies artificially plotted on concentric ellipses

• Beam-plasma neutron emissivities are not constant on a flux surface

• For JET beam trajectories fast ion birth orbits can be divided into radial 3 zones:

– passing orbits in the plasma core

– trapped orbits at mid-radius with inboard banana tips

– trapped orbits in the plasma periphery with outboard banana tips

• This results in a poloidal dependence of the parallel velocity and hence fast particle density / fusion emissivity (also volume and field line angle)

C. Challis

Poloidal asymmetry in 14MeV emission

Page 5: Fitting transport models to 14MeV neutron camera data

Poloidal asymmetry in 14MeV emission

• The result of the asymmetry is that UTC cannot match the

• This is a clear sign that the 1d model really is inadequate

• Some method is required to include the poloidal asymmetry in UTC

Page 6: Fitting transport models to 14MeV neutron camera data

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Minor radiusT

riti

um

de

ns

ity

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Minor radiusT

riti

um

de

ns

ity

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Minor radiusT

riti

um

de

ns

ity

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Minor radiusT

riti

um

de

ns

ity

Linearised neutron emission

• For trace tritium DT neutron emission is the sum of the emission from individual elements of the T profile

• If we could take the TRANSP predicted signals from each element, we would completely describe the 2d neutron profile

• PPPL has a long term (~6 months) plan to output this data

• We can do it with 1 TRANSP run for each element

Page 7: Fitting transport models to 14MeV neutron camera data

TRDT PPF (61097 UID=dmcd)

DTrr: rr=01,..., 19

The simulated KN3 signal for each unit density profile element (channel, t) (cts m/s)

RHOT: radial position of each element in root normalised toroidal flux (r)

RTOT: total 14MeV count rate for each unit density profile element (r,t) (cts m3/s)

SVrr: simulated volume fuelling

SWrr: simulated wall fuelling

The TRDT DDA and how to use it

Using the PPF

• Form DT(channel,r,t) and RTOT(r,t)

• Put your T density profile onto the RHOT grid nT(r,t)

• Sum over channels to get predicted 14 MeV signals

Rtot(t) = r RTOT(r,t) nT(r,t)

KN3(channel,t) =

r DT(channel,r,t) nT(r,t)

Page 8: Fitting transport models to 14MeV neutron camera data

t = 23.35s

Consistency of method

• First cross-check is against a TRANSP run with a full nT(r,t) profile

• The plots show that the matrix method agrees with the full run for both

– total 14 MeV neutrons

– individual cameras

• Small disparity in the central cameras is largely due to noise from TRANSP Monte-Carlo simulation

TRANSP

matrix

Page 9: Fitting transport models to 14MeV neutron camera data

Poor match on outer channelsMuch better match on outer channels

UTC with 1d reactivity modelUTC with 2d TRANSP reactivity• The initial UTC runs, with 1d

reactivity, could not match the inner and outer vertical KN3 cameras together

• With the matrix method we get much better agreement on the KN3 vertical cameras

• The 2’s are still too big on some channels, which is believed to be due to effects from...

– sawteeth

– ELMs

– neutral particles

Effect of method on UTC

Fit to vertical 14 MeV cameras

Page 10: Fitting transport models to 14MeV neutron camera data

Plan of action

•Method requires a TRANSP runs for all tritium puff shots, so these need to be requested

•We propose that these runs are started now and that they are carried through to a UTC analysis

•Timing:

•Initial TRANSP run and validation~ 3 days

•20 basis function runs and TRDT ~ 1 day

•Basic UTC analysis ~ 1/2 day

All done

~half done

20 shots selected for EPS

Page 11: Fitting transport models to 14MeV neutron camera data

Summary

•TRANSP fitting of D(r) and v(r) to neutron data is slow, tedious and does not result in a measure of significance for the results

•UTC resolves these problems, but its original 1d reactivity model couldn’t reproduce the asymmetric neutron emission

•Solution is to use UTC with a 2d description of the neutron model passed from TRANSP as a matrix

•Greatly reduces asymmetry of UTC fit to neutron data

•Do require TRANSP runs for all tritium puff shots

•Further problems: saw teeth/ELMs transport, the effect of CX with neutral particles