Development of A Generalized Integral Jet Model Duijm

Development of A Generalized Integral Jet Model Duijm, N. J. 1, 2 ; Markert, F. 4 (presenter); Keßler A. 3 Dept. of Management Engineering, Technical University of Denmark DTU, Produktionstorvet 424, 2800 Kgs. Lyngby, Denmark, nidu@dtu. dk 2 Nicestsolution, Baunegaardsvej 16, 4040 Jyllinge, Denmark, nicestsolution@duijm. dk 3 Fraunhofer Institute for Chemical Technology, Joseph-von-Fraunhofer Str. 7, 76327 Pfinztal, Germany, armin. kessler@ict. fraunhofer. de 4 Dept. of Civil Engineering, Technical University of Denmark DTU, Brovej 118, 2800 Kgs. Lyngby, Denmark, fram@byg. dtu. dk 1

Background • Reliable predictions of accident scenarios Ø improved smaller uncertainty margins Ø save society for substantial costs. • Example transient releases: – Releases from high-pressure pipe networks, • high initial release rate • rapidly decreasing • Need to know: – What is the maximum extent of such a release? 2 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Background – The alternative to “integral models” • fully 3 -dimensional CFD models. – routinely used in risk analysis: – But: » Limitations in computational effort » Limitations in efforts needed to interpret the results • CFD calculations for risk assessment – cover limited number of release scenarios » typically some 10 up to 50 scenarios – Provides uncertainty: » are worst case scenarios adequately covered? – we have not yet seen these models routinely applied on transient release scenarios. 3 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Integral models • Integral type models have been developed since about 1970. – to describe stationary plumes and jets in cross-flows (wind) • These models are widely used for risk analysis to describe the consequences of many different scenarios. • Similar models have been developed for stationary jet fires • These models are the “back bone” of nowadays commercial hazard consequence assessment software, such as PHAST (by DNV GL), and EFFECTS (by TNO). 4 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Integral models • These models are not suited to handle transient releases, as e. g. : – Releases from pressurised equipment – In case of gas ignition • A second model is needed to describe the rapid combustion of the flammable part of the plume (flash fire) • A third model for characterizing the remaining jet fire. 5 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Objective • To describe the first steps of the development of an integral-type model – describing the transient development and decay of a jet of flammable gas – Intending to transfer the stationary models to a fully transient model – Predicting the maximum extension of short-duration, high pressure jets. • Experimental support: – conducting a set of transient ignited and unignited spontaneous releases at initial pressures between 25 bar and 400 bar. In the following some first experimental results and theoretical considerations are discussed in the development of integral models describing transient behaviour. 6 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Problem Consequence assessment of releases of pressurized gases needs models to describe the extent of the jets to the limits of the hazardous concentrations. • The models may use: – simple correlations: – 1 -dimensional integral models: for simple axisymmetric jets for jets in a crossflow or with buoyancy • These models describe stationary jets • Transient releases usually are modelled by a sequence of semi-stationary jets. 7 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Modelling transient releases • For releases with fast dynamics, it can be questioned whether this approach is always adequate? Answering: the following theories are considered and compared to experimental data • Integral models – Stationary – Top hat models • CFD model • Measurements of high pressure jets 8 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Traditional self-similar jet model • 9 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Traditional self-similar jet model • 10 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Integral transient model • The stationary integral model – forward integration of ”slices” ds along the centreline s of the jet – ds can be chosen to be constant Ø Eulerian approach • The transient model – the ”slices” are transferred into ”puffs” – Puffs are followed during their lifetime – integrating over timesteps dt Ø Lagrangian approach 11 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Integral transient model • The transient model – additional descriptions needed for the adjacent puffs: • to ensure they remain connected • They interact together as a single (transient) jet. • to consider the development of length of the puffs (in addition to the radial growth) • using simple continuity: – decreasing centre-velocity of the (stationary) jet with distance, • the puffs become shorter and expand • the radial growth of the (stationary) jet – is 50% due to radial expansion of the air in the jet and – Is 50% due to the entrained air from outside. 12 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Integral transient model • In a transient jet, the front works against the still air ahead: – This has to be “made-up” by (dynamic) pressure from behind the jet • Such pressure also creates (extra) radial expansion – radial expansion • The radial velocity represents a momentum, – i. e. it requires a radial pressure gradient to set the fluid in radial motion. • The transient jet is thus the result of – the balance between transverse pressure differences • (determining the deceleration of each “puff”) and – radial expansion due to the radial pressure gradients • (determining the length of each puff) 13 DTU Civil Engineering, Technical University of Denmark 11 September 2017

CFD To obtain a better understanding of the processes in the radial direction, some CFD calculations where performed using Open. Foam with the standard k- model. 14 DTU Civil Engineering, Technical University of Denmark 11 September 2017

CFD Flow direction at 0. 2 s after start of the release. 15 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Fraunhofer ICT experimental setup 16 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Experimental setup Open air experimental plant building of Fraunhofer ICT to be used for the high pressure storage tube (l. ) and the concrete wall, here with an ignited jet in front of it (r. ) 17 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Experimental setup Schematic (l. ) and rendered (r. ) cross cut of the release section with conical reduction to 10 mm dia. of the release pipe where the rupture disk is inserted. 18 DTU Civil Engineering, Technical University of Denmark Pressure decay curves inside the high-pressure storage tube for sensor P 1 in black color at the upstream end of the tube and sensor P 2 in red color at the downstream end of the tube close to the reduction to the release pipe. 11 September 2017

Measurements In total 20 experiments were carried out: • initial pressures ranging from 25 bar to 400 bar ( four pressure steps) • with and without ignition and • two experiments for each case for reproducibility purposes. Nominal pressure [bar] Activation pressure [bar] t 90 -10 Initial pressure decay [bar/s] Averaged pressure decay [bar/s] Initial gas temperatu re [°C] 25 28. 2 ± 0. 9 0. 225 ±. 000 310. 9 ± 10 95. 2 ± 3 12. 8 ± 0. 1 100 105. 6 ± 2. 9 0. 218 ±. 000 1350 ± 37 376 ± 10 12. 3 ± 0. 3 200 222. 6 ± 5. 8 0. 200 ±. 004 3206 ± 97 837 ± 25 11. 4 ± 0. 4 400 409. 2 ± 12. 1 0. 180 ±. 003 6739 ± 233 1703 ± 62 9. 1 ± 1. 2 [s] Measured release conditions 19 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Selected experiments • Experiment H 05 and H 06 Flow rate and notional properties calculated from release conditions for selected experiments. Experiment Activation pressure (bar) Initial flow rate (kg/s) Initial gas temperature (K) Notional nozzle diameter (mm) Notional jet velocity (m/s) H 05 421. 52 1. 602 282. 19 105. 2 2161 H 06 421. 41 1. 583 283. 01 105. 5 2158 20 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Results Stationary theory "Top-hat" model non-dimensional distance s/do 160 21 140 120 100 80 60 40 20 0 0 500 1000 DTU Civil Engineering, Technical University of Denmark 1500 2000 non-dimensional time tuo/do 2500 3000 3500 11 September 2017

Conclusion • The objective – to develop an integral-type model describing the path and spreading of flammable gases in the environment. – to predict the development and path of jets and plumes from transient releases – The concept for this transient model is based on the existing integral type models for stationary releases, transformed in a Lagrangian framework of connected “puffs”. – In order to describe the interaction of the adjacent puffs, • the differences in dynamic pressure will be used – to balance between the de-/acceleration of the whole puff and the radial expansion of the puff. • Experimental data and results from CFD simulations will be necessary to quantify the empirical parameters that govern this balance 22 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Conclusion • The introduction of “pressure” in this way allows for future expansions of the model, viz. : – A better description of the interaction of the jet or plume with obstacles such as walls, pipes and process vessels. – The transient behaviour on ignition from an un-ignited plume through flash fire to a jet fire, • viz the model would provide a generalized framework capable of describing unignited jets; ignited jets (jet flames); and the transition on ignition, i. e. a flash fire. • The main application of such model – in the field of risk analysis of installations handling hazardous materials, as • offshore installations, • process industry • installations in a hydrogen-fuelled transport infrastructure. 23 DTU Civil Engineering, Technical University of Denmark 11 September 2017

Thank you for listening Questions to corresponding author: • Nijs Jan Duijm nidu@dtu. dk • Or • Frank Markert fram@byg. dtu. dk 24 DTU Civil Engineering, Technical University of Denmark 11 September 2017