Using CFD to Analyze Gas Detector Placement in Process Facilities Presented by: Scott G. Davis, Olav R. Hansen, Filippo Gavelli and Are Bratteteig GexCon
Outline Background CFD based dispersion study Gas Detector System Examples Danger clouds Optimize small leak rates
Barriers 3 Chain of Causes! Barriers Consequence Development! Threat 1! Consequence 1! Threat 2! Threat 3! Top Event! Consequence 2! Consequence 3! Threat 4! Threat 5 Consequence 4! Probability Reducing Measures Consequence Reducing Measures
4 Mitigation Measures Triggered q Activation of q Emergency Shutdown (ESD) or Blowdown isolating the inventory and limiting the size of the release q Ignition source control q Minimize the likelihood of ignition of flammable gases or liquids due to loss of containment (large leak rates) q Shutdown of ventilation in HVAC inlets q Activation of fire water pumps and deluge q Activation of PA/alarms to alert personnel
Gas Detection System 5 Detector types q Infrared point q Superior to catalytic detection q Cannot detect hydrogen q Infrared line q Better coverage than point detectors q Prone to misalignment or beam blockage q Acoustic q Large coverage q False alarms (pneumatic systems) q Cannot detect low velocity leaks (pools)
Gas Detection System 6 Design q Number of detectors q Regulations vs. company practice q Proportional to module volume q Layout q Equally spaced vs. staggered (cloud size) q Clustering around leak sources (not recommended) q Distribution according to ventilation patterns q Number of line vs. point detectors
Gas Detection System 7 Example of clustering detectors around leak source Gas detectors Gas leak, not detected
Gas Detection System Design q q Set points and voting q 10%-25% low, 30%-60% high q 1-2 LELm line q 2ooN (2 out of N detectors)/3oon Measures to be taken 8
Complexities involved in GDS 9 q Facility specific gas dispersion q Potential inventories q Types of release q High/low momentum jets, flashing, liquid spills, lighter or heavier than air gases q Air movement and ventilation patterns (open, confined, dead zones) q Topography q Design Basis q Prescriptive and qualitative assessment conservative q Performance based - CFD q Cases that are virtually impossible to anticipate ANSI/ISA-RP12.13.02 Recommended Practice for the Installation, Operation, and Maintenance of Combustible Detection Instruments
CFD - FLACS Allows for 3D representation of geometry and detectors Model the ventilation, dispersion and explosion potential
Detection Criteria Leak rate and cloud size important Cloud is what is detected and affects risk (ignition)
Detection Criteria Recall need to detect the cloud not the leak Site dependent
Detection Criteria Correlation to compare inhomogeneous cloud to dangerous gas cloud (if ignited unacceptable consequences) Mapping to determine when dispersed clouds are dangerous
Detection Criteria Comparison of detectable cloud size vs. hazard potential
GDS Design Parameters Unrealistic and not impossible to detect all leaks Small leaks can lead to very small clouds < 1m3 Dangerous leaks typically a relaxed requirement Cloud size is generally quite large ALARP optimize for small, higher frequency leaks For example 0.1 kg/s False Alarms raise detection level/voting? Redundancy Caution against optimizing detectors around chosen scenarios
GDS Design Parameters Time to detection Important parameter to evaluate effectiveness Relative to the leak rate Many reasonably designed systems will detect clouds that can lead to escalation or unacceptable consequences Minimizing time to detection can play crucial role in initiating mitigation measures (ignition source, ESD or deluge)
Explosion Risk during leak 1 7
Confirmed Gas Detection 1 8 Some explosion risk prior to gas detection
Deluge activated 1 9 Additional risk accumulation until deluge activated (full coverage)
CFD based dispersion study Geometry model Important to predict flow phenomena in complex geometry
CFD based dispersion study GDS will be compared to dispersion results Vary wind direction and speed, leak characteristics Typically 100-600 scenarios are performed per area Gas density is very important need to simulate both lighter and heavier than air gases Record data at over 1000 sensor locations (line as well)
CFD based dispersion study Detection of Dangerous clouds Larger leak rates relaxed requirement Clouds quite large Limited options with respect to mitigation measures Optimization of GDS against small leaks Increased benefit of initiating mitigation measures High probability events (0.1-0.3 kg/s leak rates) Typically never reach dangerous sizes
CFD based dispersion study
CFD based dispersion study
Dangerous Clouds P max = 1.0 barg P max = 1.3 barg* P max = 0.6 barg Leak *Design Accidental Load = 1.2 barg Stoichiometric Cloud Leak
Layout Comparison Ensure all dangerous clouds are detected
False Alarms Numerous shutdowns costing $4-8 million Performance going from 1x60%LEL to 2x20%LEL
Minimum Number of Detectors Maximize probability of detection medium/large leaks
Optimize Small Leaks 0.1 kg/s Optimize probability of detection for layouts
Optimize Small Leaks 0.1 kg/s Compare detected vs. max potential cloud size
Summary CFD allows one to quantity the detection of GDS to potential hazard Time to detection and probability of detection are typically reported Relative to leak rate Couple the results to measures that can be initiated by detection Improving a GDS by a few seconds may not be important for ESD (~ 30 seconds) But could be very important for shutting down ignition sources Limited scenarios are performed do not get trapped into optimizing sensors around chosen leaks sources Requires the input of multiple disciplines CFD is not stand alone but provides firm and repeatable base upon which complex reasoning for GDS is based