Multidisciplinary Advanced Engineering Consultants and Software Developers
Pioneers in Engineering Automation and Digitisation
Case Studies
By selecting these case studies, we aim to teach from the lessons learned
Ammonia Filling Station
Bow-Tie Risk Assessment
Overview
​
Bow-Tie Risk Assessment is a robust methodology used to visualize and analyze Major Accident Hazards (MAH) and ensure that risks are managed effectively.
Recently, a Bow-Tie workshop was conducted to identify and assess MAH risks related to an ammonia truck loading operation, incorporating preventive and mitigative barriers for key risk scenarios.
The assessment highlighted the importance of Safety Critical Elements (SCEs) and their role in maintaining operational safety.
​
​
Methodology
​
The assessment was carried out using BowTie XP software, with a focus on developing a structured framework for MAH scenarios. Key steps included identifying threats, top events, consequences, and barriers. The analysis differentiated between engineering and administrative risk reduction measures, categorized into prevention, detection, control, mitigation, and emergency response.
​
The workshop also reviewed escalation factors - situations that could compromise barrier effectiveness - and proposed controls to prevent their impact. This structured approach ensured that risks were reduced to As Low As Reasonably Practicable (ALARP). Figure 1 below shows a generic bowtie diagram illustrating the methodology.
​
Figure 1: A standard Bow-Tie diagram illustrating the relationship between threats, barriers, and consequences in risk management
Findings
​
The analysis identified several MAH scenarios for ammonia loading, including process fires, explosions, toxic releases, and unignited ingress of hazardous gases into manned areas.
For each scenario, a comprehensive list of 18 SCEs was compiled, addressing components such as containment systems, emergency shutdown mechanisms, fire protection systems, and evacuation procedures (figure 2).
​
Key outputs included Bow-Tie diagrams that provide a clear visual representation of risk pathways, highlighting the effectiveness of barriers and identifying areas for improvement (figure 3).
Figure 2: A flowchart depicting the systematic process for identifying and classifying Safety Critical Elements (SCEs) during a Bow-Tie analysis
Figure 3: Visual representation of risk pathways and barrier effectiveness generated using BowTie XP software during the workshop
Applications
​​
Bow-Tie Risk Assessment serves as a vital tool in safety management, offering clear insights into risk pathways and the effectiveness of mitigation strategies.
By integrating this approach, organizations can enhance their readiness to handle complex safety challenges and foster a culture of proactive risk management.
Battery Energy Storage System (BESS)
Overview: Battery Energy Storage System (BESS)
​
A proposed Battery Energy Storage System (BESS) in Europe is composed of racks of modules of Li-ion batteries, which like all Li-ion systems are occasionally prone to failure. A system of cells can fail in any of several ways resulting from a multitude of causes, including overcharging, off design operation, or high charge-discharge cycling rates, in addition to factors related to ageing.
Once the cycle of thermal runaway is established, little can be done to prevent escalation. Underlying failure causes can include, age, wear, random failures, as well as, infant mortality. This study examined the thermal management of the BESS units and a range of consequences of thermal runaway.
​
Release rate calculations for small, medium and large rupture releases of battery components (which releases toxic and flammable gases) were performed. The immediate leak rate, and the subsequent release rate at different time intervals, were posited for a range of scenarios.
The releases were categorised into various sizes, locations, and in the case of external releases, various wind speeds and directions were also taken into account. Pressure build up relief options were also be considered; i.e., the time dependent variance of the release was accounted for.
​
​
Pool Fires and Jet Fires: Modelling Releases
​
The scenarios involving pool fires and jet fires were modelled using exploCFD and FLUENT. For jet fires, a jet of fuel (both gas and atomised liquid) were released from a release location and immediately ignited.
​
A jet, in some cases, was found to impinge on other BESS units and cause several other releases and jet fires, pool fires and/or BLEVE events. Figure 1 below shows atypical small jet fire scenario modelled using exploCFD.
Figure 1: Jet fire modelling
Advanced Explosion Modelling
​
All explosion scenarios were modelled using exploCFD. Advanced Analysis Australia is recognised as having one of the most advanced methodologies in the world for modelling explosions (see figure 2).
Figure 2: Overpressures from a developing internal explosion; 0.31 seconds after ignition
Explosions and Overpressure
​
In general, explosions are caused by the delayed ignition of flammable gas clouds; i.e., the air and the fuel are pre-mixed prior to ignition. Explosions can be divided into two categories, supersonic, which are called detonations, and subsonic, which are called deflagrations.
The weakest deflagrations are sometimes called flash fires. The strongest detonations can cause fatalities, structural impairment of buildings on or offsite, where damage can extend for several kilometers offsite.
The overpressure produced is a function of several parameters, including the fuel types and mixture ratios, the air fuel ratio, and the confinement and congestion.
​
The key outputs from explosion simulations are pressures and impulses. Overpressure can sometimes be categorised into static and dynamic, and these are sometimes output separately but, in all cases, it is the sum of the two (called the stagnation pressure) that must be designed for.
Figure 3: Effects of large explosion on adjacent containers and structures are common outputs of explosion study.
Dust Explosion
Accident Investigation
Overview: Explosive Event in a Plastic Factory
​
This accident investigation revolved around a violent explosion that occurred in a plastic factory.
The accident caused one fatality and two serious injuries. The explosion was so strong that it completely decimated the building housing the plant and shattered windows several kilometers away. The accident lead to the hearing of a court case.
​
Potential Scenarios: High Probability
​
The public prosecutor at the time alleged that a gas cylinder in the plant was responsible for the massive explosion. However, explosion modelling guided and overseen by Dr. Madhat Abdel-jawad from Advanced Analysis, showed that even all of the gas in the cylinder in question would not be sufficient to generate sufficient overpressure to cause the amount of damage caused by the accident.
Dust lift rate due to gas explosion in a fume hood calculated using exploCFD
Numerical Modelling
​
Instead, the modelling showed that a small gas explosion in a fume hood was sufficient to rupture the relatively weak 30” piping downstream of the suction hood and agitate and ignite an appreciable amount of epoxy dust, generating a dust explosion powerful enough to destroy the building.
Dispersion, Fire and Explosion Consequence Analysis
Probabilistic Consequence Analysis
​
A probabilistic analysis includes consequence analysis for dispersion, fires and explosions.
Release frequencies based on wind rose data and taking into account the inventories, multiple release rates, release directions, and locations for all the inventories on site were calculated and ascribed to the consequence from each release.
​
Consequence Analysis: Dispersion, Fire, Explosion
​
The study included toxic and flammable dispersion modelling, as well as, fire and explosion modelling. Computational Fluid Dynamics (CFD) analysis was carried out on releases of Ammonia, Methane and Hydrogen, and other gases, including CO.
On this particular site, the most severe consequences were found to arise from toxic releases. Of importance, was the potential effect on the surrounding area, which includes other industrial sites, as well as, an encroaching residential area. A 10-5 gas cloud superimposed onto a Google maps image was generated.
Consequence Analysis Capability
​
The flammable gas clouds resulting from the dispersion analysis were simulated for delayed ignition and explosion overpressures, for critical locations and structures for the site. Also modelled, were fires and other explosions arising from the Boilers on site.
Flammable Gas Cloud on exploCFD
The study was used as a basis for the siting of a new laboratory, admin and workshop buildings. The study was also used to evaluate the escape routes and muster areas for each scenario.
Advanced Analysis has a highly advanced capability for carrying out consequence and risk analysis for toxic and flammable releases, as well as, fires and explosions.
Gas Explosion on Site using exploCFD
Modelling the Dispersion
of the Products of
TiCl4 Hydrolysis
Overview
​
Advanced Analysis Australia carried out a study focusing on the dispersion of products resulting from the hydrolysis of Titanium Tetrachloride (TiCl4).
​
Through comprehensive computational fluid dynamics (CFD) modelling, various spill configurations were examined, each contained within a bund structure.
​
This study provided a crucial order of magnitude assessment of hydrochloric acid and titanium oxychloride dispersion under representative conditions, offering a valuable foundation for further quantitative risk analysis (QRA) considerations.
​
​
Consequence Analysis: Gas Cloud Sizes and Exclusion Zones in Different Bund Configurations
​
While it was found that the flow regimes near the spill surface were different, there was little difference between the gas cloud sizes at the lower concentrations for both bund configurations but there was significant difference at the higher concentrations particularly at 150ppm and 500ppm.
The region of exclusion for most concentrations was found to be between 350-400m at these conditions and the volume examined could sustain these conditions for several days.
​
The analysis could easily be extended to other configurations, spill regimes and, inclusion of groundwater and it could also be formulated within QRA.
Modelling Conditions
​
The following representative conditions were selected for modelling:
Wind speed: 1.5 m/s
Relative humidity: 70%
Area of Bund: 55 m2
Volume of spill (TiCl4): 14500L
Chemical Kinetics and Difusion
​
The overall key chemical reaction for hydrolysis was modelled as:
TiCl4(l) + 3H2O(l,v) => TiO2.H2O.3HCl(s)+ HCl(g)
(Kapias, Griffiths J Hazardous Materials A119(2005),41-52)
which incorporated several reactions producing hydrochloric acid and titanium oxychloride.
The Damköhler number was calculated to be large and, therefore, equilibrium reaction rates were used.
The water vapour was carried to the TiCl4 surface via the humidity in the air. The procedure involved the minimisation of the Gibbs free energy. The gas-through gas diffusion was modelled using the Stefan-Maxwell diffusion model.
Results
​
To obtain indicative results, two bund configurations were considered. One where the spill level was flush with the top bund walls and another where the spill level sat below the bund walls by 4.87m (figs. 1a and b, respectively). Hexahedral grids were used in all simulations (fig. 2). Mesh independence was verified.
Figure 2: Hexahedral grids
A region of 400 m behind the tank was monitored for species concentrations. The figures below (fig. 3 and fig. 4) show the gas cloud locations for concentrations between 3 ppm and 500 ppm for both configurations. It could be seen that there was little difference between the gas cloud sizes at the lower concentrations for both bund configurations but a significant difference at the higher concentrations particularly at 150 ppm and 500 ppm.
​
In general, a distance of between 350-400 m of clearance was required to remain clear of all the concentrations with the exception of 500 ppm for the walled configuration. The mass rate of depletion of the TiCl4 was calculated by integrating the Ti and Cl mass flow in products across the liquid surface. This was calculated at 10.6 g/s which means that a volume of 14500 l could sustain these gas cloud concentrations for several days at these conditions.
​
While it was found that the flow regimes near the spill surface are different, there is little difference between the gas cloud sizes at the lower concentrations for both bund configurations but a significant difference at the higher concentrations particularly at 150 ppm and 500 ppm. The region of exclusion for most concentrations is found to be between 350-400 m at these conditions and the volume examined can sustain these conditions for several days. The analysis can easily be extended to other configurations, spill regimes and, inclusion of groundwater and it can also be formulated within QRA.
Figure 3: Gas cloud 1
Figure 4: Gas cloud 2
Crude Oil
Explosion and Fire
Accident Investigation
Overview
​
An explosion occurred in a tank used for storing crude oil, resulting in 1 fatality and over 12 serious injuries.
The explosion happened in a corroded dewatering tank with an oil inlet, oil outlet, and water outlet. After isolating and draining the tank, four workers entered to perform hot work.
​
However, the crude outlet had an unusual design with 2 dead legs below the drainage level, each containing 1500 kg of crude that was not visible.
​
Root Cause Summary
​
During hot work, an experienced worker cutting into one of the deadlegs with an oxyacetylene torch ignited residue inside the deadleg. The flame traveled through the deadleg, causing the crude in it to ignite and create a small pool fire within the deadleg.
​
The flame then reached the blocked off common main, where volatile substances and methane produced by bacteria were present. This led to a gas explosion, pushing crude out of the deadleg, and generating an oil mist explosion in the tank. The tank's small openings were insufficient to vent the high pressure.
Schematic drawing of deadleg
Composition of the Explosion
​
The determination of the explosion's composition was rendered unattainable; however, it may have stemmed from one or more of the following factors:
a) Emission of volatiles originating from the crude oil,
b) Influence of bacterial activity linked to weathering processes,
c) Liberation of volatiles resulting from initial combustion.
​
​
Ignition of the Cavity
​
Despite its lean and feeble nature, the explosion within the confined cavity space can attain pressures of up to 0.4 bar, exerting its force over a considerable duration. Consequently, one must contemplate the reaction of the slug under such circumstances.
Pressure development in confined cavity: up to 0.4 bar exerted over time
Reaction of the Oil Slug
​
CFD simulations were used to model the behavior of the crude oil under pressure. The analysis revealed distinctive shapes of the oil mist cloud imprinted on the tank's inside walls. Even crude oil, when atomized into a mist, can cause a powerful explosion.
CFD simulations modelling crude oil behavior under pressure
FEA Modelling
​
The tank's roof was designed to be frangible, but only 2 out of 8 specified conditions specified in API 650 for a tank with a frangible roof were met during construction. As a result, instead of breaking off, the tank's floor sheared off. This caused the tank, along with its concrete ring foundation weighing 500,000 kg, to lift off the ground.
​
FEA simulations demonstrated the stress concentrations that caused the tank's floor to shear off.
Importantly, the simulations revealed that only 0.05 bar of pressure was required for the tank to lift off the ground.
FEA simulations: stress concentrations and tank lift-off at 0.05 bar
Explosion Accident Investigation:
A Case where Energy Efficiency
Caused an Explosion
Overview: Explosive Event in an Industrial Rotary Oven
​
Advanced Analysis completed an accident investigation where a series of powerful explosions were recorded in a Zinc oxide production facility. The explosions occurred in one of the rotating ovens which are designed to produce Zinc Oxide from Zinc.
The event happened without any personnel being in the vicinity, and therefore there were no fatalities or injuries. More concerning, there were anecdotal reports that this type of event was common since a relocation of the facility to a new plant.
This accident investigation was carried out with the purpose of preventing a similar event from recurring in the future. The investigation isolated a most probable cause for this entire class of accidents.
The heat generated in the oven was supplemented with exhaust gases from a separate furnace. The purpose of supplementing gas was to reduce the thermal load required to be produced by the rotating oven. This made the process more energy efficient, which in turn reduced environmental footprint of the process, and also lowered production cost.
​
​
Potential Scenarios: High Probability
​
The furnace exhaust gases were composed almost entirely of water and carbon dioxide. These gases were mixed with fresh air, causing a drop in the temperature of the exhaust gases. The new location of the plant was in an area with a wet climate, and on the day of the accident, it was a rainy day with an outside relative humidity of near 100%.
​
This means that the drop in temperature of exhaust gases due to mixing them with near saturated air resulted in some condensation of water. As the mixed gases traveled in the air supply pipes and ducts, which were not insulated, further heat loss occurred, resulting in further condensation. The liquid water collected in low points, which arose from slight misalignments in the air supply piping.
Schematic drawing of industrial rotary oven air suply
Water in Molten Zinc
Schematic representation of the process following the deposition of water into the rotating oven.
Numerical Modelling
​
At some point, some of this stagnating water collected in the pipeline dropped into the Molten Zinc oven.
The water was immediately superheated resulting in its expansion several times an order of magnitude.
​
This caused a pressure rise in the oven and suddenly displaced the Molten Zinc causing its atomisation. This resulted in an increased reaction rate, the Zinc droplets oxidised rapidly in a mechanism similar to a dust explosion.
Expansion of water droplet in molten Zinc. Multiphase simulation in Ansys.
Numerical Modelling
It was recommended to put in a water trap in the air supply pipeline, heat air supply with the furnace exhaust gases via a heat exchanger, or remove the exhaust gases supply. The solution ensured the accidents never occurred again.
Mining Safety:
Blast Analysis Optimization
Overview
​
Continuing an extensive history of conducting fire and explosion consequence analysis studies, as well as, accident investigations in mining and metal processing operations for various sites and clients, Advanced Analysis has concluded a new study focusing on assessing the consequences of drill and blast operations on crucial structures.
In this study conducted at a nickel mine for a prominent mining technical service provider, Advanced Analysis engineers calculated the time history of overpressure and impulse acting on a structure located at the end of a ventilation shaft. This was due to a charge detonated in a cavern at the opposite end of the shaft.
Sequenced ANFO explosion below an underground ventilation shaft
Analysis of Firing Sequence Effects
​
exploCFD was employed to conduct a detailed examination of the effects of different firing sequences. This analysis involved comparing the outcomes of simultaneous detonations to those of sequential blasts, with each blast having a 40-millisecond delay between them.
​
The analysis concluded that using a reduced amount of explosives effectively maintains overpressures below a pre-specified critical threshold. This highlights a safe and efficient approach, emphasizing the potential of such analyses to enhance safety practices within the mining industry.
Shock waves from multiple firings propagating in a tunnel.