1. Comparison of modelling approaches for CFD simulations of high pressure H2 releases
E. Papanikolaou, D. Baraldi
Joint Research Centre - Institute for Energy and Transport

2. Outline Notional nozzle concept
Notional nozzle approaches investigated
Experimental description
Simulations set up
Results
Conclusions
Ongoing work

3. Notional nozzle concept
The jet flow of a high pressure unintended release is complex: a supersonic region in the near field and a subsonic region downstream. Numerical modelling of the near field is quite demanding
Need for reasonable computer run-times, for engineering applications
Replacement of the actual release nozzle by a notional, occupying an area with the same flow rate as the actual one
Does not necessarily exist in a physical sense
Assumptions made:
atmospheric pressure
uniform velocity profile
Under-expanded jet release
Level 1: “reservoir” conditions
Level 2: real orifice
Level 3: notional nozzle location

4. Notional nozzle approaches investigated
Birch et al. (1984)
Conservation of mass, isentropic change, ideal gas, temperature at notional nozzle equal to atmospheric
Birch et al. (1987)
Conservation of mass, conservation of momentum, isentropic change, ideal gas, temperature at notional nozzle equal to atmospheric
Ewan and Moodie (1986)
Conservation of mass, isentropic change, ideal gas, temperature at notional nozzle equal to the one at the actual release nozzle
Schefer et al. (2007)
Conservation of mass, conservation of momentum, isentropic change, real gas (Abel-Noble equation of state), temperature at notional nozzle equal to atmospheric
Notional nozzle approaches that take into account the conservation of energy have been also proposed such as Yücel & Ötügen (2002), Molkov et al. (2009) and Xiao et al. (2011)

5. Experimental description
High momentum horizontal (free-surface) H2 experiments in the HYKA test facility of the Institute for Nuclear and Energy Technologies of FZK
Selected experiment for investigation: H2 release from 1 mm diameter at stagnation pressure of 98.1 bar and stagnation temperature of 14.5 ºC
H2 concentration and flow velocity were measured on the jet symmetry axis at 1.5 m and 2.25 m from the release

6. Simulations set up: conditions at the actual and notional nozzle
Pressure (kPa)
Density (kg/m3)
Temperature
(K)
Sonic Velocity
(m/s)
Ideal gas*
5165
5.25
239
1178
Real gas* (Abel-Noble)
4932
4.89
235
1216
Table 1: Conditions at the release (actual nozzle)
Approach
Diameter (10-3 m)
Area
(10-5 m2)
Density (kg/m3)
Temperature (K)
Velocity (m/s)
Birch (1984)*
7.14
4.00
0.085
288
1293
Birch (1987)*
5.78
2.62
1972
Ewan&Moodie*
6.81
3.64
0.103
239
1178
Schefer
5.86
2.69
2029
Table 2: Conditions at the notional nozzle
* A discharge coefficient equal to 0.91 was used to match the calculated mass flow rate to the experimental

7. Simulations set up: dispersion calculations
ANSYS-CFX version 12.1
4 notional nozzle approaches
“real” pipe with diameter equal to the experimental (D=1mm) and length equal to 10D to evaluate the level of accuracy with a coarse mesh (shock region not resolved)
Equations of mass, momentum and energy (total enthalpy) conservation
4 commonly used turbulence models: standard k-, SST, RNG k- and baseline k- (BSL)
All simulations run as transient cases for 5 s
High resolution scheme for discretization of advection terms
2nd Order Backward Euler for discretization of transient terms
Ideal gas law for the notional nozzle cases, Redlich Kwong equation for the “real” pipe cases
Inlet with boundary conditions for velocity and temperature from the notional nozzle approaches or a given mass flowrate (equal to the experimental) for the “real” pipe cases.
All simulations had a common incoming level of turbulence (intensity of 5%) assigned to the inlet

8. Simulations set up: dispersion calculations
Computational domain: 15 m × 10 m × 10 m
Minimum and maximum timestep was 10-8 s and 10-3 s for the notional nozzle cases and 10-8 s and 10-4 for the “real” pipe cases
Unstructured mesh: nodes for notional nozzle cases and nodes for “real” pipe cases
Solution domain for all cases

9. Results: H2 concentration at 1.5 m and 2.25 m from the release
MEV: Mean Experimental Value
STD: Standard Deviation
30% - 50% over or under prediction of MEV
Comparison between approaches
In general, Birch 1987, Schefer and “real” pipe cases perform better
Comparison between turbulence models
General tendency for higher predictions with k- , followed by RNG k- , BSL and lastly SST

10. Results: Flow velocity at 1.5 m and 2.25 m from the release
Comparison between approaches
Majority of predictions lie within the 30% over/under prediction of MEV
Highest values predicted by approaches with highest release velocity (Schefer, Birch 1987)
Comparison between turbulence models
General tendency for higher predictions with k- , followed by either RNG k-  or BSL and lastly SST

11. Results: Contour plots of H2 concentration – Birch 1984 & Birch 1987
Case: Birch84 – BSL
H2 mass in flammable cloud: g
Flammable volume: m3
Case: Birch87 – BSL
H2 mass in flammable cloud: g
Flammable volume: m3
Case: Birch84 – k-
H2 mass in flammable cloud: 1.51 g
Flammable volume: m3
Case: Birch87 – k-
H2 mass in flammable cloud: 0.73 g
Flammable volume: m3
0.5 m
Case: Birch84 – RNG
H2 mass in flammable cloud: g
Flammable volume: m3
Case: Birch87 – RNG
H2 mass in flammable cloud: g
Flammable volume: m3
Case: Birch84 – SST
H2 mass in flammable cloud: g
Flammable volume: m3
Case: Birch87 – SST
H2 mass in flammable cloud: g
Flammable volume: m3

12. Results: Contour plots of H2 concentration – Schefer and Ewan
Case: Schefer – BSL
H2 mass in flammable cloud: g
Flammable volume: 0.11 m3
Case: Ewan – BSL
H2 mass in flammable cloud: g
Flammable volume: 0.21 m3
Case: Schefer – k-
H2 mass in flammable cloud: g
Flammable volume: 0.16 m3
Case: Ewan – k-e
H2 mass in flammable cloud: g
Flammable volume: 0.34 m3
Case: Schefer – RNG
H2 mass in flammable cloud: g
Flammable volume: m3
Case: Ewan – RNG
H2 mass in flammable cloud: g
Flammable volume: 0.23 m3
Case: Schefer – SST
H2 mass in flammable cloud: g
Flammable volume: m3
Case: Ewan – SST
H2 mass in flammable cloud: g
Flammable volume: 0.18 m3

13. Conclusions
The conclusions are relevant to the selected experiment, the available data and the simulations’ set up. To make general comments, more experimental conditions should be investigated
Birch 1987 and Schefer approaches produce more accurate results for the H2 concentration
Including the conservation of momentum in the approach increases the accuracy of the results
The coarse mesh of the “real” pipe cases produced accurate enough results for both H2 concentration and flow velocities on the symmetry axis
Further investigation is necessary on both axial and radial distances from the release

14. Ongoing work Assessment of approaches with:
different initial conditions (stagnation properties, release diameter)
both axial and radial experimental measurements of H2 concentration and flow velocity
Effect of:
Grid resolution. Grid independence for notional nozzle approaches.
Turbulence intensity at the source (5%, 10%).
Discretization schemes

15. Thank you for your attention

16. Results: Contour plots of H2 concentration – Schefer and “real” pipe
H2 mass in flammable cloud: g
Case: Schefer – BSL
Flammable volume: 0.11 m3
Case: “real” pipe – BSL
H2 mass in flammable cloud: g
Flammable volume: m3