Use of Dispersion Modeling Software In Ammonia Refrigeration Facility Design By: Martin L. Timm, PE Corporate Process Safety Manager For the UW-Madison IRC R&T Forum, May 8-9, 2013
Introduction My IIAR paper (March 2013) documents a case study of a hypothetical machinery room along with indoor and atmospheric dispersion model results for various design options. This presentation will address: Building Use Emergency Ventilation Considerations Modeling Considerations Indoor Results Outdoor Results Conclusions This is an overview. Please see the paper for additional details. Available at IIAR.org May 1, 2013
Building Use: CO 2 Production Facility Carbon Dioxide raw gas is purified and liquefied to make beverage-grade CO 2 Machinery room is much like any ammonia refrigeration machinery room The 2-stage ammonia refrigeration process uses oilflooded screw compressors, shell and tube heat exchangers, evaporative condensers, etc. The refrigerant charge (inventory) 15,000 lbs. (6,804 kg) ammonia Subsequent work has identified opportunities to reduce the ammonia charge to below 10,000 lbs Results of the original work are still valid and instructive
Ventilation decisions include: the design volume flow rate of air the number and type of exhaust fans their location the orientation and velocity of the air discharge provisions for makeup air to replace the exhausted air whether machinery room pressure must be maintained negative relative to other adjacent spaces location, function, and type of refrigerant leak detectors, location and function of ventilation controls, design basis for heating equipment to maintain a minimum room temperature in cold climates This paper concentrated on the items in yellow
Sidewall mounted fan Fan Types Considered in This Case Study Roof-mounted upblast fan Roof-mounted upblast fan with entrainment (multiple flow impeller to induce (mix) additional outdoor air with the fan discharge
Case Study Machinery Room Dimensions and Volume Room contains: Ammonia compression equipment CO 2 compression equipment Purification vessels, filters, etc. Heat exchangers and low-side vessels Outdoors: Evaporative condensers High pressure receiver
Flow Rate Comparison With and Without 1:1 Entrainment
Cases Considered for leak modeling: lb/min kg/min 1 0.45 10 4.5 100 45 500 227 Corresponds to approximate hole size at 180 PSIG: 0.09 to 1.9 (2 mm to 48 mm) for saturated ammonia vapor 0.02 to 0.55 (0.6 mm to 14 mm) for saturated ammonia liquid The 500 lb/min case was chosen as an upper limit because 1,000 lb/min would have resulted in a leak rate that exceeded the ventilation fan capacity for those fans designed for 2.8 ACH. It should not be assumed that this leak rate is credible for all or most facilities during their lifetimes. The consideration of probability of leaks of various sizes is beyond the scope of this paper.
Discharge Conditions Two ammonia state conditions were chosen for the modeling: +77 F (25 C) superheated vapor at atmospheric pressure (indoor releases) -28 F (-33.3 C) two-phase 0.12 liquid fraction by weight (indoor and outdoor releases) Leak Duration was specified at 100 minutes Note: Case Study Inventory of 15,000 lbs. would be exhausted in 30 minutes at the flow rate of 500 lb/min
UDM Plume Geometry for Outdoor Continuous Release Warm gas will continuously rise Cold aerosol may be dense gas and slump to ground Lift-off can occur once gas warms IRC R&T Forum May 8-9 10
Discharge of Warm Gaseous Ammonia At Four Different Locations, 30 ACH @ 3000 fpm Downwind Extent of 200 PPM at Leak Rate of 500 lb/min
Discharge of Cold Two-Phase Ammonia At Four Different Locations, 30 ACH @ 3000 fpm Downwind Extent of 200 PPM at Leak Rate of 500 lb/min
Discharge of Cold Two-Phase Ammonia Roof fan w/o entrainment 30 ACH Downwind Extent of 200 PPM at Leak Rate of 500 lb/min
Discharge of Cold Two-Phase Ammonia Roof fan with entrainment 30 ACH Downwind Extent of 200 PPM
Sample Dispersion Visualization Import of Concentration Contours into 3D CAD Program Dispersion of Stack Exhaust Reaching an Air Inlet IRC R&T Forum May 8-9 15
Conclusions (1 of 3): Dispersion modeling is useful for understanding tradeoffs in air volume, discharge velocity, fan type, orientation, and location The existing IIAR recommendation for upblast fans with a discharge velocity of at least 2,500 ft/min (762 meters/min) is justified Leaks that result in the creation of a cold aerosol outdoors near ground level are expected to have the potential for reaching the longest downwind distance to concentrations of concern Outdoor downwind concentration can be potentially reduced by placing the potential leak source inside a wellventilated IIAR-compliant machinery room
Conclusions (2 of 3): For large leaks in small rooms, a ventilation rate of 30 ACH in the machinery room provides a better assurance than lower ventilation rates of staying below the LFL Note: Large and small are relative terms! Lower ventilation flow may be effective at staying below flammable limits if: Total ammonia inventory in the system is low enough, or potential leak size is minimized through an effective mechanical integrity program the inventory can be reliably isolated into smaller volumes when an alarm is triggered, to prevent all of the inventory from being released by a single leak
Conclusions (3 of 3): Design factors that can potentially improve outdoor dispersion effectiveness include increasing discharge velocity, increasing discharge height, or use of fan types that entrain additional outdoor air for pre-dilution air at the fan outlet There may be benefit to IIAR and other industry participants considering explicitly allowing the use of an engineering analysis to justify lower ventilation rates when room volume is relatively large and ammonia inventory is relatively small If relying on isolation systems to justify lower ventilation rates in new construction, consideration should be given to evaluating the reliability of the detection and isolation systems with a quantitative or semi-quantitative tool such as layer-of-protection analysis (LOPA)
Closing Remarks: The information presented for this case study is essentially a consequence analysis for various design options Owners and operators may find this type of information useful to support decisions, but other information will likely be required such as local regulatory requirements, proximity of sensitive public receptors, etc. A consequence analysis does not consider the credibility or probability of the events When an even more refined understanding of the resulting risks is desired, probability may be evaluated as well, using a tools such as LOPA or Quantitative Risk Analysis (QRA), which build on the consequence analysis
Additional Material IRC R&T Forum May 8-9 20
A Key First Decision Equipment location Indoors? Outdoors? Impacts: Combination? First cost Operating cost Equipment life expectancy Maintenance Reliability Consequences of leaks and how to respond
In this case study, the climate dictated putting compression equipment and heat exchangers indoors Many end users choose to put at least some of the ammonia equipment indoors. So ventilation design becomes relevant to achieve a safe, cost effective installation Moisture in CO2 raw gas must be removed; must keep some vessels, heat exchangers, and condensate drains above freezing
Required Ventilation Calculated from ASHRAE Standard 15
Required Ventilation ASHRAE 15 Vs. 12 ACH and 30 ACH This table shows air flow from room Table on next slide shows air discharged when dilution air at 1:1 ratio is entrained and added to the room air
Indoor Dispersion Modeling 1. UDM not suitable for estimating the change in ammonia concentration in the room over time 2. For long duration steady leaks, could calculate leaving concentration after a long time from air flow and leak rate, independent of room size 3. For short duration and/or to account for room volume, use equation like: 4. PHAST implementation of this algorithm is INBU
Note on Indoor Modeling Method assumes instantaneous uniform mixing within the room In reality, mixing will be non-uniform Any leak will have a zone close to the release point that goes through the UFL and LFL Indoor modeling is still useful for understanding trends IRC R&T Forum May 8-9 26
Indoor Discharge of Warm Gaseous Ammonia Indoor Conc. @ Leak rate of 10 lb/min (4.5 kg/min)
Indoor Discharge of Warm Gaseous Ammonia Indoor Conc. @ Leak rate of 500 lb/min (225 kg/min)
Indoor Discharge of Warm Gaseous Ammonia Indoor Conc. @ Leak rate of 100 lb/min (45 kg/min)
Outdoor Dispersion Modeling Many programs available from various software vendors One class of programs uses Uniform Dispersion Model (UDM) for outdoor dispersion, solving a set of differential equations In this study PHAST 6.7 was used Simplifying assumptions: one exhaust fan flat terrain no influence from surrounding structures CFD programs have more extensive computational capabilities, but are labor and cost intensive to model even simple arrangements; beyond the scope of this case study
Discharge of Cold Two-Phase Ammonia Roof fan w/o entrainment 12 ACH Downwind Extent of 200 PPM at Leak Rate of 500 lb/min
Discharge of Cold Two-Phase Ammonia Roof fan with entrainment 12 ACH Downwind Extent of 200 PPM