SAE INTERNATIONAL ADDITIONAL INFORMATION FROM THE SAE CRP

Transcription

1 SAE INTERNATIONAL ADDITIONAL INFORMATION FROM THE SAE CRP

2 Discussion on KBA test results and data

3 KBA test result summary Background: Four vehicles were tested to help understand the general safety of yf. The final report detailed how the specific test conditions were carried out. General test condition: 40 kph, 40% off-set, hot/wet test, leakage varied in three test stages: Stage 1: leakage through parts as damaged in crash Stage 2: damaged but not leaking parts were manually made leaking Stage 3: modifications to add nozzle, change direction of flow, adding plates and higher temps Stage 1/2 - leakage (stage1)? - Ignition/HF? Stage 3 - Type of manipulation - Ignition? - HF under hood? - HF in cabin? Vehicle 1 Vehicle 2 Vehicle 3 Vehicle 4 yes no yes no no no KBA conclusion: no general product safety issue Nothing reported Lines turned Plate added Lines turned Plate added no no no no no no no KBA recommendation: further investigation (3 min TWA<HBL, peak above with added plate) no no Fish-mouth opening in HP line Yes (2 of 3 tests) yes (under hood) KBA test does not show general safety issue for yf. Ignition only happens with special leakage opening/line turning. HF emissions with plate are not a risk for passengers or first responders. no

4 While the CRP is having difficulty understanding the representativeness of the approach used in Stage 3. There is general agreement with the KBA test results and data!

5 Degrees C KBA data does not conflict with (700 C) minimum exhaust temperature for ignition Exhaust temperature at the time of refrigerant release Vehicle 1 Ignition Vehicle 2 Vehicle 3 Vehicle V07-3V03-1V06-2V01-1 V16-3V13-2V14-3 V15-3I V17-3V22-3V21-3V12-2V04-1 V18-3I V02-1V05-2V08-2V09-2V10-3V11-3V19-3V20-3 SAE CRP and KBA data support that 700 o C is the minimum exhaust surface temperature for ignition and that good vehicle design can prevent ignition at much higher temperatures

6 PPM of HF KBA data supports position that HF generated underhood under expected releases (phase 1 and 2) are not an issue Less than 1 PPM peak values generated underhood V01-1 V02-1 V03-1 V04-1 V05-2 V06-2 V07-3 V08-2 V09-2 V10-3 V11-3 V12-2 V13-2 V14-3 KBA data supports that levels of HF generated through expected refrigerant release in a collision (Stage 1 and 2) are inconsequential and do not represent any danger of injury

7 PPM 6 KBA data supports that HF generation from thermal breakdown of the two refrigerants are similar and inconsequential R-1234yf Peak value (not TWA) R-134a Peak value (not TWA) 1 0 V17-3 V21-3 KBA data supports that R-134a and R-1234yf have similar HF generation from thermal breakdown (exclusive of ignition)

8 KBA data that R-1234yf ignition under hood could produce HF in significant quantities is not new information. While refrigerant ignitions are considered extremely unlikely the CRP report states that under hood ignitions of R- 1234yf could produce HF levels under hood above the health based limits of 95 ppm. The report also stated that the risk of exposure associated with this type of event is in the range of 3 x per vehicle operating hour. KBA data that HF levels under hood could be significant if a refrigerant ignition occurs agrees with the original CRP report and was shown to be a very low risk.

9 KBA data supports position that HF exposure inside the cabin is not concerning KBA data supports position that HF levels in the cabin are not concerning

10 KBA data supports that condenser breach has a significant effect on preventing Ignition Test Exhaust Temp at release Location of breach Ignition V condenser no V condenser no V condenser + line (fish mouth) no V15-3I line (fish mouth) Yes V condenser + line (fish mouth) no V line (fish mouth) no V18-3I line (fish mouth) Yes KBA data supports position that ignitions are much less likely when the condenser is breached. The data supports position that leak location and direction are a significant factor.

11 KBA conclusions on safety of the 4 vehicles The results of the experiments have shown insufficient evidence to support the suspicion that occurrence of a serious threat within the meaning of the Product Safety Act in the vehicle types tested. The KBA testing confirms that there are no safety concerns with these vehicles using R-1234yf

12 Summary 1.) KBA data supports that the minimum exhaust surface temperature for refrigerant ignition is 700 o C. 2.) KBA data supports that levels of HF generated through expected refrigerant releases poses no safety concern. 3.) KBA data supports that R-134a and R-1234yf have similar HF generation from thermal breakdown. 4.) KBA data supports that HF levels under hood, in the rare event of ignition, could be above the health based limits (covered in CRP very small risk). 5.) KBA data supports that the HF levels measured inside the vehicle are not concerning. 6.) KBA data supports position that ignitions are much less likely when the condenser is breached. 7.) The KBA testing confirms that there are no safety concerns with the vehicles tested using R-1234yf. The KBA test results do not conflict with and in fact support the SAE CRP results

13 Back up

14 Some discussion on the simplicity of the CRP fault trees

15 Data Support Values for FT Input Probabilities Exposure Vehicle Fire Propagation Conditions present for refrigerant ignition? Vehicle fire? Exposure to vehicle fire? Conditions present for propagation? Occupant unable to exit vehicle? Published traffic safety data OEM data used: - CAE, crash results (fluid distribution) - GIDAS database (fluid leakage) Ignition Collision occurs? Flammable concentration near hot surface? AC system leak occurs? Hot surface at relevant temp? Flammable concentration at hot surface? OEM data used: - Establish correct temperature (700ºC) - Determine likelihood temp occurs OEM data used - GIDAS database OEM data used - CAE, crash results Mitigating factors present? OEM data (quench, fan, steam): - Effectiveness of mitigation Fault tree analysis supported by extensive data set

16 After temperature and concentration limits are met There are only two significant mitigating factors for ignition (cooling fan operation and release of coolant) There are only two significant mitigating factors for fire propagation (air currents from the dynamics of the crash and quenching)

17 Reference for 700 O C as the minimum exhaust surface temperature required for ignition

18 R-1234yf R-1234yf + PAG oil R-134a + PAG oil Ignition Temperature (degc) No ignitions Results Summary of Flammability of Refrigerant/oil <Graphical> G H C No ignitions I A D E F B Chart represents > 800 refrigerant releases 300 = Ignition Observed = Mixed Behavior = No Ignition Lab Tests Veh #1 Veh Veh Veh Veh Veh #2 #3 #4 #5 #6 Vehicle Tests (R-1234yf + PAG oil) KBA Tests Note: - Each data point represents a separate release on the subject vehicle - hollow data points represent non-ignition releases All data (including KBA) support use of 700 C as minimum relevant temperature. No data supports use of auto-ignition temperature (405 C) as being relevant

19 Support for 700 C as the minimum exhaust surface temperature for ignition There has not been one well documented source that has come forward with any data to suggest that they have measured refrigerant ignitions below 700 o C. All car manufacturers are held accountable for design related vehicle fires and vehicle safety issues if there were concerns below 700 o C wouldn t it be likely that more than one manufacturer would be raising an issue? The SAE CRP considered (in the sensitivity analysis) lower temperatures ignitions that would relate to 10% vehicle usage over the minimum exhaust surface temperatures. The risk related to this estimation was 3 x per vehicle operating hour. More than one OEM tested the Daimler B-Class. During release testing, ignitions were observed propagating from the back side of the turbo surface. Additional measurements showed that the hottest spot on the exhaust surface was the back side of the Turbo which was 25 to 30 deg C higher than the front side where Daimler s witness testing was measuring.

20 Reference data with regard to coolant release as a mitigating factor for ignition.

21 Effect of coolant release Several vehicles were tested by CRP OEMs with and without coolant release. One vehicle with a transverse engine, turbocharged, front take down vehicle that was tuned for ignition (specific nozzle, reduced charge, restricted flow, preheated refrigerant and optimized release location) showed five for five ignitions with release. When coolant was also released (2 liters at the radiator end tank) there were 0 for 5 ignitions observed. One vehicle with a longitudinal engine, turbocharged with the A/C on the same side as the exhaust take down (specifically tuned for ignitions - 5 for 5 tests) showed a 50% reduction in ignition when coolant (2.0 L) was released in 8 individual tests. The Daimler B class tested by more than one OEM showed that coolant release increased the exhaust temperature required for ignition by 30 to 45 degrees K when.25l to 3L of coolant were released. When ignitions were observed with coolant release they tended to be brief, self extinguishing events (due to quenching by the water).

22 Coolant Release Impact When a radiator breaches, coolant is released underhood Plot shows 7.75 vol-% (stoich.) Dry air (79 vol-% N 2 & 21 vol-% O 2 ) With a radiator breach containing 50:50 mix C 2 H 6 O 2 and H 2 O Key Facts/Observations 15 vol-% O 2 needed for ignition US Bureau of Mines Bulletin 627 O 2 < 15% when coolant > ~70 C cannot support ignition Coolant temperature always > ~70 C when exhaust surface > 700 C This analysis very conservative Not consider heat capacity of water vapor and aerosol Heat sink to pull heat from ignition kernel of weak flame Calculations suggest significant mitigation whenever coolant breach occurs

23 Reference material that hood buckling decreases the likelihood of ignition.

24 Effect of hood buckling Several vehicles were tested with and without hood buckling in an attempt to understand the effect. One transverse turbocharged front take down vehicle that was tuned for ignition (specific nozzle, reduced charge, restricted flow, preheated refrigerant and optimized release location) showed five for five ignitions with release. When the hood was partially opened 1 for 5 ignitions were observed. A vehicle with a longitudinal engine, turbocharged with the A/C on the same side as the exhaust take down (specifically tuned for ignitions - 5 for 5 tests) showed a 66% reduction in ignitions when the hood was opened. The Daimler B class tested by more than one OEM showed that partially opening the hood increased the exhaust surface temperature required for ignition by 20 to 40 degrees K.

25 Using conservative fault tree analysis as a prediction tool

26 Observation with regard to using fault trees to make predictions Fault trees are designed to assess risk and to understand design sensitivities In many cases, fault tree analysis does state that the risk of an adverse event is X. It simply states that the risk can be no higher than X Predictions made using fault trees do not require absolute knowledge of the probabilities of all events - they often rely on using conservative data to help assess the risk - which typically leads to an overestimation of risk. The CRP fault trees make many conservative assumptions which prevents them from being used to make accurate predictions based on the quoted risk.

27 Observation with regard to using fault trees to make predictions Some of the conservative assumptions in the CRP fault trees: 100% of vehicles will be able to achieve over 700 o C exhaust temperatures. Most diesels, non turbocharged and larger displacement V engines will not reach these temperatures (the CRP fault trees assume all vehicles will). There is much data that shows that not all vehicles designs will have ignition when there is sufficient concentration of refrigerant near a 700 o C exhaust surface (the CRP fault trees assume that all vehicles will have an ignition) The CRP fault trees assume that all vehicles have restrictive heat shields that trap refrigerant near the hottest part of the exhaust system (not all vehicles do). The CRP fault trees assume that 65% of mid speed collisions (15 kph to 50 kph) will have a breach in the A/C system (there was only 50% at 40 kph in the KBA testing)

28 Observation with regard to using fault trees to make predictions The CRP fault tree work does not state that the probability of having an exposure to a vehicle fire based on the use of R-1234yf is 3 x 10-12, but rather that the probability can be no higher than that. Given the conservative analysis done the actual probably would certainly be much lower than that reported by the CRP.

29 Concerns with other fault tree assessments