Aerosol Sampling ISOLDE facility aerosol results Prepared by

Aerosol Sampling ISOLDE facility aerosol results Prepared by: Enrico Da Riva, Gennaro Bozza 12 -04 -2012 June 8, 2021 G. Bozza, E. Da Riva 1

Remaining simulations No. Simulation Status 1 Sampling Tube, 45 Degrees Cut, operation (3 m/s) Done 2 Sampling Tube, 45 Degrees Cut, flush (8 m/s) Done 3 Sampling Tube, coaxial, operation (3 m/s) Done 4 Sampling Tube, coaxial, flush (8 m/s) Done 5 Sampling Tube, shrouded, operation (3 m/s) Not done 6 Sampling Tube, shrouded, flush (8 m/s) Not done 7 Bend, 45 degrees and R b=1 Done 8 Bend, 45 degrees and R b=3 Done 9 Bend, 90 degrees and R b=1 Done 10 Bend, 90 degrees and R b=3 Done 11 Straight pipe, 10 m long, D = 1. 12 m 12 Final Enlargement, 7 cm long 13 Complete Isolde Ventilation duct June 8, 2021 Almost done First conclusions available Done G. Bozza, E. Da Riva 2

Table of Contents 1. Transport efficiency (90° bend) 2. ISOLDE (stack + sampling pipe) June 8, 2021 G. Bozza, E. Da Riva 3

1. Transport efficiency (90° bend) June 8, 2021 G. Bozza, E. Da Riva 4

Geometry • D = 10 cm, flow rate = 20 m 3 h-1 • Different bending radius (R/D = 1, R/D = 3) • All possible orientations June 8, 2021 G. Bozza, E. Da Riva 5

Settings sensitivity Modifying Drag law, Saffman’s force, Step Length Factor, Integration scheme Analytic Default settings June 8, 2021 G. Bozza, E. Da Riva All other settings 6

Star. CCM+ vs. Fluent results for the 90° bend 0. 1 μm y R/D=1 x R/D=3 June 8, 2021 G. Bozza, E. Da Riva 7

Star. CCM+ vs. Fluent results for the 90° bend 1 μm y x June 8, 2021 G. Bozza, E. Da Riva 8

Star. CCM+ vs. Fluent results for the 90° bend 10 μm No-gravity & long stretch vertical y x long stretch horizontal June 8, 2021 G. Bozza, E. Da Riva 9

Trapped particle distribution (example) dp=10 μm, June 8, 2021 G. Bozza, E. Da Riva 10

Simulation vs. correlation Long stretch horizontal (R/D=1) Transport Efficiency Simulation Star. CCM Simulation Fluent Correlation Particle 0. 1μm 95% 87% 100% Particle 1μm 40% 86% 99% Particle 10μm 5% 28% 33% • This table shows that Fluent results are much closer to the correlation than Star. CCM+ results. • The 15% of efficiency underestimation of Fluent, compared to the correlation, is explainable in the following way: Fluent overestimates the turbulent inertial deposition compared to the correlation. June 8, 2021 G. Bozza, E. Da Riva 11

Turbophoresis Star. CCM+ Inlet Downstream of the elbow Fluent Almost the same (very uniform) distribution at every cross section Unlike STARCCM+, in Fluent the effects of the turbophoresis are not relevant. June 8, 2021 G. Bozza, E. Da Riva 12

Conclusions 1/2 Star. CCM+ vs Fluent Turbulent inertial deposition: negligible in Star. CCM and Correlation Relevant in Fluent (15% with L ~ 100 D) Gravity: Transport efficiency Gravity deposition is the dominant phenomenon Relevant influence of the elbow but not of the bending radius of the elbow. Negligible influence of the gravity June 8, 2021 G. Bozza, E. Da Riva 13

Conclusions 2/2 Some rough numbers D = 10 cm, flow rate = 20 m 3 h-1 Transport Losses % Straight stretch L=10 m Elbow Turbulent deposition Gravitational deposition TOTAL 0. 1 μm ~5% ~10% ~0% ~15% 10 μm horizontal ~10% ~35% ~25% ~70% 10 μm vertical ~10% ~35% ~0% ~45% June 8, 2021 G. Bozza, E. Da Riva 14

2. ISOLDE (stack + sampling pipe) June 8, 2021 G. Bozza, E. Da Riva 15

ISOLDE Stack (main pipe) Machine Mode: Qmain = 7500 m 3/h Flush Mode: June 8, 2021 Sampling pipe Qmain = 15000 m 3/h G. Bozza, E. Da Riva 16

Tables Entering the inlet 2 After the inlet 3 1 Before the Sampling point Machine Mode: Qmain = 7500 m 3/h 0. 1μm 1 - Particles before the sampling point Nst 1μm 10μm 1257262 1257108 1255855 4191 4190 4186 4758 4730 4766 3 - Particles after the inlet, outside of the stack 3419 Nis 3662 3971 4 - Particles reaching the instrumentation Nout 1336 1324 375 Aspiration efficiency: Ni/Ns 114% 113% 114% 2 - Particles entering the sampling inlet Ni Transport efficiency: Nout/Ni 28% 8% Sampling efficiency = Asp eff * Trans eff 32% 9% June 8, 2021 G. Bozza, E. Da Riva To instrumentation 4 17

Eddies (sampling inlet) Strong recirculation zone which might negatively affect the measurement What to do: avoid this geometry, and use a recommended one (coaxial or shrouded) June 8, 2021 G. Bozza, E. Da Riva 18

Eddies (enlargment) The diverging angle of the final enlargement is too large. Therefore, a separation of the boundary layer occurs and the result is a recirculation zone which might negatively affect the sampling efficiency. What to do: modify the geometry, and possibly eliminate the enlargement. June 8, 2021 G. Bozza, E. Da Riva 19

Conclusions • Simulations for the machine case (7500 m 3 h-1 in the stack, 25 m 3 h-1 in the sampling) were run. • Generally, we can distinguish two groups of particles with different behaviour: • 0. 1 ÷ 1 μm → ~ 30% Sampling Efficiency • 10 μm → ~ 10% Sampling Efficiency • Recirculation zones at the inlet and at the final enlargement which could negatively affect the measurement. • Next steps: simulations for the flush mode, simulation for the only stack. June 8, 2021 G. Bozza, E. Da Riva 20

What can be already recommended: multiple shrouded probes for sampling • Because it is necessary to design sampling systems to operate under both normal and accident conditions (when proportionately higher concentrations of large particles could be present), it is customary to sample isokinetically. Under the ANSI standard, for large ducts it is recommended that rakes of isokinetic probes be used to span the duct cross section, ostensibly to collect representative samples. (A Predictive Model for aerosol transmission through a shrouded Probe, HONGRUI GONG, SUMIT CHANDRA, ANDREW R. MCFARLAND, AND N. K. ANAND). June 8, 2021 G. Bozza, E. Da Riva 21