Navigating ammonia risks:
Improved modelling for a decarbonized future
Outline
1 Introduction
2 Ammonia pipeline releases
3 Ammonia dispersion behaviour
4 Ammonium nitrate: Solid explosion modelling
5 Risk modelling for an ammonia pipeline
2 DNV ©
Outline
1 Introduction
2 Ammonia pipeline releases
3 Ammonia dispersion behaviour
4 Ammonium nitrate: Solid explosion modelling
5 Risk modelling for an ammonia pipeline
3 DNV ©
Ammonia applications
Traditional use
• Ammonia has been used in industrial scale for more than a 100 years
• About 80% of ammonia produced used as fertilizer in agriculture
• Many other uses: refrigerant gas, cleaning solutions, manufacture of
plastics, explosives, pesticides and other chemicals
Key material in the energy transition
• Energy storage and hydrogen carrier
• Zero-carbon fuel
4 DNV ©
Ammonia hazards
Toxicity
• The primary hazard of concern is usually toxicity
• American Industrial Hygiene Association guidelines for 1 hour exposure:
• “Minor irritation”” (ERPG-1): 25 ppm
• “Serious toxicity” (ERPG-2): 150 ppm
• “Potentially lethal” (ERPG-3): 1500 ppm
• Odor detection threshold ~17 ppm
Flammability
• Flammable hazards much less concern than toxic effects
• Ammonia is hard to ignite
• Flammable hazard in confined areas
5 DNV ©
Ammonia safety considerations
• Increased use of ammonia is predicted
• New applications, potentially in closer proximity to populations
• Two primary set of liquid ammonia conditions: cold and pressurized
• Cold ammonia:
• Transport by ship, storage in tanks, terminal pipelines
and loading operations
• Temperature just below atmospheric boiling point -33°C
• Atmospheric pressure or slightly above
• Pressurized ammonia:
• Transport by rail/trucks, long-distance pipelines
• Pressure well above saturation pressure (about 8.5 bar at 20°C)
https://www.mdpi.com/1996-1073/16/10/4019
6 DNV ©
Safety insights through software solutions
30+ 1000+ Segments:
years Customers and Across assets: Upstream, midstream and World leading risk
3000+ active licenses Offshore and onshore downstream, decarbonization products
(CCUS, hydrogen and ammonia)
PhastTM SafetiTM KFXTM EXSIM
• Phast, the most widely • Leading Quantitative • A leading computational
used software for Risk Assessment fluid dynamics (CFD) • Well-known and
consequence analysis (QRA) tool simulator for fire extensively validated
• 900 Phast client • 340 Safeti client • Unique capabilities in CFD gas explosion
companies with 3000+ companies with 900+ simulating the behaviour of simulation software tool
active licenses active licenses ammonia, CO2, hydrogen
Phast-CFD Easy-to-use Computational Fluid
Dynamics modelling within Phast
7 DNV © powered by KFX
Phast modelling
Weather Toxic
LOC scenario Discharge Dispersion Radiation
Explosion
8 DNV ©
Outline
1 Introduction
2 Ammonia pipeline releases
3 Ammonia dispersion behaviour
4 Ammonium nitrate: Solid explosion modelling
5 Risk modelling for an ammonia pipeline
9 DNV ©
Phast pipeline releases: overview
• Phast pipeline release modelling
• Fluid phases: liquids, vapour, super-critical
• Accident scenario: full-bore rupture and partial
breaches of a long pipe
• Transient modelling
• Accounts for valve action and finite pumped
throughput
• Pipelines can be buried - crater effects
10 DNV ©
Phast pipeline releases: inputs and results
Auto-sectioning results Time varying discharge results
Depressurisation results
(based on values, pressure drop, (flow rate, atmospheric
(fractional loss, etc.)
pipeline length, etc.) expansion, duration)
Release
Pressure Front Pressure Front
Pipeline data
Process conditions Pipeline Failure data Valves
(length, thickness,
(material, temperature, (location, breach size, (type, location, closing
diameter, roughness,
pressure, pumped inflow) release direction) time, PoF)
elevation)
11 DNV ©
Phast pipeline releases: recent improvements
• Phast/Safeti 9.1: extended Pipebreak-II model with
Ammonia liquid density predictions
significant improvements for pipeline releases 1.2
Soave-Redlich-Kwong (SRK) Equation of State
1.15
• Initial rapid depressurization regime
Peng-Robinson EOS (Phast 9.0 and earlier)
Liquid density ratio [Predicted/NIST]
1.1 DIPPR and SRK hybrid offset method (Phast 9.1 and later)
• Subcooled liquids including drainage
1.05
• Small holes (aperture ratio below 20%) 0.95
0.9
• New method for liquid density predictions 0.85
• Improved validation against experiments
0.8
0.75
0.7
0 50 100 150 200 250 300 350 400 450 500
Impact Pressure [bar]
• More accurate liquid pipeline releases ➔ avoid workarounds
• Avoid underprediction of fluid inventories
12 DNV ©
Phast pipeline releases: ammonia case study
Pressure-liquified ammonia pipeline to industrial area Temperature-liquified ammonia export pipeline to jetty
• 5.5 km long, 20” diameter, buried • 3.2 km long, 600 mm diameter, buried
• Upstream conditions: 50 barg, 25°C • Upstream conditions: 10 barg, -34°C
• Pumped flow: 277 kg/s • Pumped flow: 386 kg/s
Full-bore rupture
of loading arm
13 DNV ©
Full-bore rupture ammonia loading arm: flow rate
Rapid depressurisation regime
Isolation valves closing at 60 s
(20 m from rupture location)
Drainage regime
Steady-state regime (pumped flow)
14 DNV ©
Outline
1 Introduction
2 Ammonia pipeline releases
3 Ammonia dispersion behaviour
4 Ammonium nitrate: Solid explosion modelling
5 Risk modelling for an ammonia pipeline
https://www.ntsb.gov/investigations/AccidentReports/Reports/PAB0702.pdf
15 DNV ©
Phast Dispersion Model
Unified Dispersion Model (UDM)
• Extensive, continuous verification and
validation in past 35+ years
• Effects considered in dispersion calculations:
• Buoyancy or lack of buoyancy
• Aerosol formation, rainout and pool vaporization
• Touchdown, lift-off, and capping at the mixing layer
• Air entrainment and cloud spreading
• Phast-CFD and KFX can also be used to model
ammonia dispersion
• Effects where terrain or geometry are important
16 DNV ©
Reaction of Ammonia with Water
• The model developed by Raj and Reid (1978)
is used to describe the dissolution of ammonia
in water, and the heat produced.
• This model is only applicable for ammonia
dissolution in water and not appropriate for any
other material.
• The dissolution of ammonia with water causes
the pool to warm up.
• The current SafeAm JIP is developing new
models
• The reaction of ammonia with water vapour in
air in not modelled in Phast.
17 DNV ©
Validation: Desert Tortoise & FLADIS
• Pressure-liquified ammonia releases
• Ambient temperature, high pressure (~8 barg) liquified NH3
• Little or no rainout or lift-off predicted. Good validation results
• Desert Tortoise (top right):
• August/September 1983, Nevada
• 60-80 kg/s
• Dispersion measurements at 100m, 800m downwind (1m height)
• FLADIS:
• 1994-1996, Sweden
• 0.3 – 0.4 kg/s
• Dispersion array at 20m (0.1m height), 70m (0.5m) and 238m
(1.5m)
18 DNV ©
Red Squirrel Tests (DNV Spadeadam)
• RS1
• Low pressure spills of refrigerated (-33 deg C) ammonia into a bund
at very low pressure
• Consider empty (concrete base) and water-filled tests
• RS2
• High pressure, ambient temperature liquid ammonia releases
• Horizontal orientation
• RS3
• Low to high pressure, cold (-30 deg C) liquid ammonia releases →
to understand transition from pool to 2-phase flow
• Downwards into concrete bund
DNV ©
Red Squirrel: Pool mass validation
• Compare pool mass to load cell data on land
• Excellent for 1st hour
• Good alignment of peak mass (90% LF for 1C)
• After 1 hour for 1C & 1D, we over-estimate vaporisation rate (pool dry-out)
• Experiments on water consistent with Phast predictions for mass dissolved
20 DNV ©
Toxic Modelling
• Probit
• Dangerous Dose
• Concentration thresholds
21 DNV ©
Key factors for ammonia dispersion
• Fluid state (vapour, sub-cooled, pressure liquified). Cases where near 100%
rainout expected (e.g. subcooled) best modelled as impinged or at zero height
• Buried vs overland pipelines. Buried pipes have reduced velocity and initial
dilution
• Time-varying options. Subcooled cases close to a valve, using averaging over
valve closure time can be best.
• Substrate and dissolution (land / water). Dissolution removes mass
permanently, but accelerates vaporisation of the rest
• Surface roughness. Toxic far field results mean SR is a key input
• Averaging time. Far field concentrations reduced, and widths increased
22 DNV ©
Example study
23 DNV ©
Outline
1 Introduction
Insert screenshot from
GIS case study here…
2 Ammonia pipeline releases
3 Ammonia dispersion behaviour
4 Ammonium nitrate: solid explosion modelling
5 Risk modelling for an ammonia pipeline
24 DNV ©
Solid explosions modelling in Phast and Safeti
• Suitable for industries handling high
explosives (e.g., mining, construction,
chemical processing)
• Designed for compliance with Kingery &
Bulmash correlation (Lees, 2012) and GB/T
37243-2019 Chinese standard
• Calculates: Overpressure, Impulse, Duration
and Arrival Time
• There are 4 preloaded solid materials in 9.1
25 DNV ©
Ammonium Nitrate Explosion
Input Value
Material Ammonium Nitrate
Flammable mass 10 tonnes
Event location Warehouse
Event frequency (/yr) 1e-5
26 DNV ©
Outline
1 Introduction
2 Ammonia pipeline releases
3 Ammonia dispersion behaviour
4 Ammonium nitrate: Solid explosion modelling
5 Risk modelling for an ammonia pipeline
27 DNV ©
Risk Modelling in Safeti (Overview)
• To determine risk, the following can be defined in Safeti:
Risk Ranking
Points
Individual
Risk
Weather Population
Frequency
Toxic Societal
Risk
Equipment Discharge Dispersion Fires Event Tree Impact
Exceedance
Ignition Explosion Human Data
Probability Vulnerability
Effect
Ignition Congested Building Hazard Contours
Sources Regions Vulnerability Effect Level
28 DNV ©
Ammonia CFD Effects Modelling in Phast-CFD/Safeti
Webinar 06/May/2025: Ammonia Pipeline Storage and Offloading Case Study
29 DNV ©
DNV’s Commitment to Best Modelling Practice
KFX-Ammonia R&D JIP (2022-2024)
• Equinor, Vår Energi, Horisont Energi
• Objective to improve KFX with respect
to advanced ammonia safety analyses
for realistic conditions;
• Extend learnings to Phast-CFD/Safeti
model development
• Focus on modelling of spray release
and dispersion of ammonia;
ammonia-water interaction with
absorption of ammonia gas in water
droplets.
• Provides solid basis for safe scaling
of ammonia infrastructure
Software tools 2 Joint industry projects 3 Physical experimentation
1 and model validation
30 DNV ©
KFX – Unique features
• Spray models (Lagrangian and multi-phase)
• NH3 specific: model can handle absorption of
ammonia vapour in water droplets
• CO2 specific: model can handle sublimation (solid
dry-ice CO2 particles turning to gas) and
condensation (CO2 vapour to liquid)
• Water mitigation (deluge)
• Shallow layer pool model – tracking of liquid
spread on terrain
• Vegetation modelling
• Real fluid equations: handle liquid, multiphase
and supercritical releases
31 DNV ©
Summary
32 DNV ©
Summary
• Phast, KFX (CFD) and Safeti enable navigating Ammonia risks effectively
via:
• Providing simple to detailed consequence and risk modelling solutions
for numerous practical applications relevant to Ammonia process safety
and risk-based design decision making
• Specification of background data to a great level of detail for specific
consequence/risk modelling analysis
• Great control of input/output data to fully understand the
consequence/risk picture for effective RRM
• Consequence modelling tools: Phast and KFX handle key physical
phenomena such as:
• Ammonia reaction with water, effect of deluge systems
• Solids explosion
• Craters
• Time-varying releases
• Topography
• Vegetation
• JIP between DNV and partners contribute to developing models for our
tools and facilitating best practise for safe Ammonia economy
33 DNV ©
Thank you
For further enquiry, demo or quote, please contact digital@dnv.com
www.dnv.com
34 DNV ©
