J.Th

1
Acknowledgement Research is subsidized by
International Atomic Energy Agency in Vienna
(No. 11557 and by Ministry of Education of
Czech Rep. No. J04/98 212200008
Verification of CFD models
by Stimulus Response method

• J.Thýn, M. Nový, P. Houdek, G. Borroto Portela)
  and R.Žitný
• Department of Process Engineering
• CTU in Prague, Faculty of Mechanical Engineering,
  
• ) ISCTN Avenida Salvador Allende, La Habana,
  Cuba
• E-mail thyn_at_fsid.cvut.cz

1. CASE STUDY - CONTINUOUS OHMIC HEATER Direct
ohmic heater with three parallel
channels Reynolds number in central
channel 1300, in lateral channel 1600,
inlet/outlet pipes 3000.
Is a flow regime
in a part of apparatus
2. EXPERIMENTS Radiotracer Tc99m was injected
into inlet pipe. Local responses were detected
by collimated scintillation detectors D1, D2, D3
and D4 (see Fig). RTD (residence time
distribution) of the heater was determined by
stimulus response method using the conductivity
probes C1 and C2 measuring KCl concentration at
inlet and outlet
LAMINAR OR
TURBULENT?
3. CFD MODEL and RESPONSE EVALUATION CFD model
(Fluent 5.2, laminar, k-?, RNG and low Reynolds
flow models) predicts distribution of tracer
c(t,x,y,z) in time t. While the RTD is given
directly by Fluent, the theoretical responses of
detectors D1,2,3 and 4 must be calculated
according to
where D(x,y,z) is the response of collimated
detector to the unite activity situated in x,y,z
(PSR-Point Source Response). The function
D(x,y,z) can be evaluated theoretically (using
view factor/ Monte Carlo method) or by
interpolation of values D(xk,yk,zk) determined
experimentally (the locations xk,yk,zk of unit
activity tracer have nothing to do with position
of CFD nodes) lk is distance between
(x,y,z) and (xk,yk,zk) points.
Do not look
4. COMPARISON of EXPERIMENT WITH CFD PREDICTION
for an
answer in a
CFD CODE.
Local responses of collimated detectors
Experimental response (1), RNG k-? model (2)
and laminar flow (3)
Comparison of RTD CFD prediction and
conductivity method
Try

• 5. CONCLUSIONS
• New method of CFD postprocessing enables
  prediction of collimated detector responses to a
  gamma radiation tracer distribution.
• A case study predicts only small difference
  between the laminar and turbulent flow model with
  small advantage for RNG k-? . An effect of real
  injection, inaccuracy of collimation algorithms
  as well as effects of operational parameter
  ambiguities are to be analysed in future.

EXPERIMENTS