3 TEST SET UP, TESTING EQUIPMENT AND TEST METHOD
3.3 TEST PROCEDURE
3.3.3 Testing scheme
The testing schemes were developed from discussions amongst DMT, DUH, D. Colbourne and IEC SC61C WG4.
In table 1) the different variables or parameters of both test matrix are summarized with the used values or descriptions.
Remark: *) In parentheses are numbers of fans used for testing Table 1) Variables of the test matrix
Between the tests the tent was ventilated (contaminated air extracted) for air ex-change. The extraction was monitored and stopped at a R290 LEL level of ≤ 1 %.
The individual test matrix for the AHT top CU cabinet and the Carrier base mounted CU cabinet measurements are shown in Annex 1 and 2.
To avoid any explosion within testing several general precautions and countermeasures were made. Base of these precautions and countermeasures was an expert’s report to explosion prevention (Report 20658672: Explosionsschutzkonzept gemäß § 6 (9) 2. Gef-StoffV für einen Versuchstand zur Untersuchung von Leckageszenarien an Kühlgerä-ten). This expert report was made by the specialists of the DMT department “Fire and explosion protection”.
This expert’s report included:
Review and Assessment of Material Properties
Measures to prevent hazardous explosive atmospheres
Precautions against ignition
Measures for reduction of explosion effects to a safe level
Organizational explosion protection measures
Examples of precautions and countermeasures for explosion protection:
Several thermal conductivity measurement sensors were placed around the tent to monitor a possible leakage of R290 in the surrounding. Alert level were indi-cated automatically in the control room.
No persons stayed in the test hall during the measurements.
The control office was located one floor below the test hall.
Access of unauthorized and untrained persons was prohibited.
The released amount of R290 was controlled continuously by online monitoring the concentration in the tent by the test personnel.
Manual closing of the R290 supply and start of the extraction of R290 from the tent after fault-related test stop or completion of test from the control room.
Avoidance of electrostatic charges within testing (no opening of the zipper doors before air extraction after completion of test).
Use of fans for the multideck cabinets were built classified as „II 2 G Ex d e ib IIB T3 Gb“.
A risk assessment of the manufacturer of the fan was made for this leakage test.
People involved in testing were instructed how to behave according the explosion protection rules listed in expert report.
4 Examples of test results
All test measurement results from the different sensors were collected in excel sheets. The R290 concentrations were transformed to %-values of the according lower explosion level (LFL) from R290 (propane: 2.1 % v/v; NFPA). For each test run these transformed data of all sensors were written for the same time scale in one excel-sheet. All raw data were given from DMT to the consulter of DUH for further data analysis.
4.1 Example results of test run 1
Before a test run was started all parameters were adjusted as described in the example from the following table 2 (extract of the test matrix of the Carrier cabinet with base CU as shown annex 2).
Test no Room (m2) Release
mass (g) Cabinet Cabinet
po-sition Leak location Condenser airflow
Table 1) Extract test 1 & 2 of the test matrix of the Carrier cabinet with base CU
With these settings test 1 (and 2) were performed. Sensors were positioned (see fig. 2) and numbered as shown in figure 3.
Figure 3) Numbering of the sensor positions
The according data for the first 3 min at measurement point 1 (MP 1) are listed in tab. 2. At MP 1 (0.5 m distance to the cabinet) IR sensor 1 was detecting the R290 concentration in parallel with a thermal conductivity sensor (FTC5).
Table 2) Extract measurement results of test 1 (Carrier cabinet with base CU)
The resulting curve of the concentration (expressed in percent of the LFL) against time for IR sensor 1 is plotted in figure 4a. The plot of the comparable curve from thermal conductivity sensor FTC5 is shown in fig. 4b.
Hour Duration Mass flow controller
10:50:06 00:00:06 0 0,1 0,1 0
10:50:11 00:00:11 0 0,1 0,1 0
10:50:16 00:00:16 0 0,1 0 0
10:50:21 00:00:21 0 0,1 0,1 0
10:50:26 00:00:26 0 0,1 -0,1 0
10:50:31 00:00:31 15 0,1 0 0
10:50:36 00:00:36 15 14,8 0 0
10:50:41 00:00:41 15 15,1 0,1 0
10:50:46 00:00:46 15 15,2 0,1 0
10:50:51 00:00:51 15 15,2 0 0
10:50:56 00:00:56 15 15 0 0
10:51:01 00:01:01 15 15 0,1 1
10:51:06 00:01:06 15 15 0 3
10:51:11 00:01:11 15 15,1 -0,1 6
10:51:16 00:01:16 15 15 0,9 10
10:51:21 00:01:21 15 15,1 3,4 13
10:51:26 00:01:26 15 15 5,4 18
10:51:31 00:01:31 15 15,2 7,1 26
10:51:36 00:01:36 15 15,3 9,4 31
10:51:41 00:01:41 15 15,1 12,7 30
10:51:46 00:01:46 15 15,2 15,3 30
10:51:51 00:01:51 15 15 18,3 33
10:51:56 00:01:56 15 15,1 22,4 35
10:52:01 00:02:01 15 15,1 26 36
10:52:06 00:02:06 15 15,1 28 38
10:52:11 00:02:11 15 15 30 39
10:52:16 00:02:16 15 15,1 31,3 38
10:52:21 00:02:21 15 15,1 33,5 40
10:52:26 00:02:26 15 15 34,7 41
10:52:31 00:02:31 15 15,1 35,4 43
10:52:36 00:02:36 15 15,1 35,8 43
10:52:41 00:02:41 15 15,1 36,8 43
10:52:46 00:02:46 15 15 37,5 41
10:52:51 00:02:51 15 15 39,7 40
10:52:56 00:02:56 15 14,8 40 42
10:53:01 00:03:01 15 15,1 41,4 43
Figure 4) Concentration time curve of the leakage simulation test 1 (IR sensor 1)
Figure 4b) Concentration time curve of the leakage simulation test 1 (FTC sensor 5)
As can be seen from tab. 2 after 31 s the mass flow controller was switched to 15 l/min re-lease of R290 (V set [l/min]). After the next time step of 5 s it was indicated that this flow was realized (V as-is [l/min] with a value of 14.8 l/min). After further 25 s R290 is indicated from
the thermal conductivity detector FTC5 (00:01:01) with a rough value of 1 % LFL. After fur-ther 15 s IR sensor 1 is indicating R290, too (00:01:16 with 0.9 % LFL).
Comparing the concentration values of both sensor types (fig. 4a & b) shows advantages and disadvantages of both sensor types. The thermal conductivity detector reacts faster to a change of the gas concentration. The IR sensor shows a delayed increase and decrease due to a membrane around the measurement cell which is necessary for explosion protection of this sensor. An advantage of the IR sensor is the higher sensitivity (smother curve) and lower detection limit. Another disadvantage is a technical set for an upper detection limit at beneath 100 % LFL. The FTC sensor is not restricted to the 100 % LFL level and was therefore, used at critical measurement points as sensor type to measure concentration far above the 100 % LFL level.
After 5 min release time the R290 release in the cabinet was stopped. 10 min after starting the test the air of the tent was extracted and exhausted to the outside of the test hall. This can be seen in fig. 5 for the concentration time curves of all sensors. 13 min after starting test 1 all measurements were stopped.
Figure 5) Concentration time curves of the leakage simulation test 1 (all sensors)
Especially for lower concentrations (below 30 % LFL) the IR sensor is more precise than the FTC.
To examine the risk potential by potential leakages in a cabinet in closed rooms the meas-ured concentration times curves in dependence of the location in the tent are very valuable.
4.2 Repeatability of the R290 concentration measurements
To examine the repeatability of this test method several tests under same conditions were repeated. For this examination in the 20 m² tent a diffusor was placed and R290 released.
Pictures of this diffusor are shown in pic. 10a & b and for the test setup in pic. 10c & d.
Pic. 10a) Diffusor setup: bottom of the bucket with holes for equal gas release
Pic. 10b) Diffusor setup: bucket with cotton as diffusor material; in the centre a smaller jar in which the tube to the R290 pipe is connected
Pic. 10c) Diffusor hanging at a stand Pic. 10d) Diffusor located at the narrow wall of the 20 m² room; sensors placed according scheme fig. 6
The scheme of the sensor positioning for this diffusor test is shown below in figure 6.
Figure 6) Scheme of the diffusor test setup
The according measurement schedule is shown in table. 3.
Table 3) Test schedule of the repeatability test with the diffusor
The data for all 4 test runs for one measurement point or location are shown in comparison in the following figures 7a to 7g for several measurement points (MP).
Sensors from MP1 to MP10 are type infrared (IR) and MP 11 & 12 thermal conductivity (FTC).
Differences between the scatter of the curves of the four repeated runs may derive from the sensor type and distance from the diffusor (release point), position and height in the room.
To examine this for all sensors at the different measurement points the average at a time point where roughly a 50 % LFL value occurs were selected. For this moment beneath the average the standard deviation and the relative standard deviation were calculated (Tab. 4).
Test no Room (m2) Release m ass
(g) Unit Unit base
install height Unit position
-1 20 1500 Diffusor 1.8 m Broad w all,
m iddle
Test no Leak location Unit airflow Louvre Mass flow
(g/m in) Rem arks
-1 Diffusor Off - 60 Repeatablity: 4
x (-1/1 to -1/4)
Figure 7a) Four concentration-time curves of the diffusor test at MP 1 (0.5 m from diffusor)
Figure 7b) Four concentration-time curves of the diffusor test at MP 5 (1 m from diffusor)
Figure 7c) Four concentration-time curves of the diffusor test at MP 7 (2 m from diffusor)
Figure 7d) Four concentration-time curves of the diffusor test at MP 9 (4.5 m from diffusor)
Figure 7e) Four concentration-time curves of the diffusor test at MP 10 (5 m from diffusor)
Figure 7f) Four concentration-time curves of the diffusor test at MP 11 (2.5 m from diffusor) Start air
ex-haust of tent
Figure 7g) Four concentration-time curves of the diffusor test at MP 12 (2.5 m from diffusor)
Table 4) Comparison of the scatter (RSD) between different measurement points
A comparison of the data in tab. 4 gives no clear answers. It seems to be that the sensors positioned nearer to the diffusor (MP 5 & 1) have less scatter than for MP 9 & 10. This be-comes clearer comparing the curves (7a & b with 7d & e). MP 7 shows in tab. 4 a quite high RSD (20 %). But this is caused by a strong deviation of the curve from run 3 (MP 7-3) in fig.
7c. After reaching values of > 60 % LFL after 3 min the scatter is quite low between the dif-ferent curves.
The different concentration-time curves from MP 11 (FTC sensor) are showing a higher scat-ter below 80 % LFL (tab. 4 23 % RSD afscat-ter 15 min) but are approaching quite similar values after 20 min. Only MP 12 which is positioned 2 m above floor level shows especially after 20 min an increasing scatter. This is probably caused by fluctuating R290 levels in that height.
After 30 min the R290 loaded air was extracted from the tent.
Measurement
Point Sensor type Distance to release point [m] / Height [m]
In the figures 7a to 7e all IR sensors are showing a mixture between a flat line and a “squig-gle” reaching the 100 % LFL level. This irregular behaviour of the IR gas sensors during the R290 measurement has a technical reason.
As this IR sensor is designed to be used as a stationary sensor for gas monitoring it would not regularly face gas concentrations higher than 100%. But the very high resolution and pre-cision beneath the ATEX certificate were the reasons to use this detector. Due to an alarm-control it will set its analogue output to a value below 0% (e.g. to switch on an alarm horn).
To prevent this behaviour DMT reprogramed the measurement software, so values below 0 % appear as 100 % to keeps the concentration at the upper limit. Because the sensor is quite slow it is very difficult for the software to clearly identify this case, so a “jumping” or
“squiggle” appears at concentrations near 100%.
4.3 Presentation of all test results
In annex 1 the measurement matrix for the Top CU multideck cabinet is shown. The accord-ing concentration time curves of all room positions summarized for one test run in a graph are shown for all these test runs in annex 3 (figure 8a from test no. 17 to fig. 8ab from test no. 47).
The comparable graphical presentation of all test results as concentration time curves for the Base CU cabinet (matrix from annex 2) are shown in annex 4 (figure 9a from test no. 1 to fig.
9v from test no. 50).
5 Summary
In this report the testing conditions for simulated leakage tests with multideck cabinets are described.
All tests were performed under strict safety regulations and precautions because R290 is a highly flammable refrigerant. Despite the limitations caused thereby and especially in view of the sensor systems which can be used as online monitoring system, the resulting data could be used as expected for a subsequent risk assessment of cabinets for usage with several hundreds of gram of R290.
Essen, June 21, 2016
_______________________________ _________________________________
Dr. D. Renschen S. Roeser
(Head Product Assessment Refrigeration & Air Quality) (Technician Product Assessment Air Quality)
Annex 1a
Measurement Matrix Top CU (A)
Test noRoom (m2)Release mass (g)CabinetCabinet positionLeak locationCondenser airflowEvaporator airflowDoorsKick- platesCU coverRoof coverMass flow (g/min)Release time (s)
Measurem ent time [min]Remark 17OffOffOpenOnOnOff30 18OffOffOpenOffOnOff30 19OffOffOpenOn*OnOff10- 20OffOffClosedOn*OnOff10- 21OffOffClosedOn*OnOff30The two doors in the middle opened 1min after leakage; Fan started 4:30 min later 22OffOnClosedOn*OnOff30- 23OffOnOpenOn*OnOff30- 24OnOffOpenOn*OnOff30Effect of cond airflow on leak from evap 25On (x1)OffClosedOn*OnOff6018,0Fan #3 on (this cabinet has 6 condenser fans) 26On (x 2)OffClosedOn*OnOff6019,0Fans #2 + #5 on 27On (x 3)OffClosedOn*OnOff6020,0Fans #1 + #4 + #6 on On (x 4)OffClosedOn*OnOff60- (Was not done because previous test Cf < 12 g/m3) 28OffOffClosedOn*OnOff105060- 29OffOffClosedOn*OnOff3016,730- 30OffOffClosedOn*OnOff608,318- 31OffOffClosedOn*OffOff30- 32OffOffClosedOn*OffOn30- 33OffOffClosedOn*OnOn30- 34OffOffClosedOn*OnOff30Add polystyrene panel to CU corner to divert 35OffOffClosedOn*OnOff30- 37Condenser return bendsOffOffClosedOn*OnOff3030- 36Room centreCU housingOffOffClosedOn*OnOff30Move to centre of room (see test with top CU C)
16,7 20500top CU (A)
Broad side, centre
Evaporator return bends CU housing
8,3 16,7
25,0
30,0
Matrix Top CU Determine kick-plate case with the highest Cf 5060 16,730,0
Annex 1b
Measurement Matrix Top CU (A)
Test noRoom (m2)Release mass (g)CabinetCabinet positionLeak locationCondenser airflowEvaporator airflowDoorsKick- platesCU coverRoof coverMass flow (g/min)Release time (s)
Measurem ent time [min]Remark 38OffOffOpenOn*OnOff10100120- 39OffOffOpenOn*OnOff3050- 40OffOnOpenOn*OnOff3050- 41On (x3)OffClosedOn*OnOff6045(i) Start with max no fans from 20 m2 room. (Fans #1 + #4 + #6 on) 42On (x 4)OffClosedOn*OnOff6030(ii) If Cf < 12 g/m3, then do test with one fewer fan. But if Cf > 12 g/m3, then do test with one more fan. (Fans #1 + #3 + #4 + #6 on) 43OffOffClosedOn*OnOff10100110- 44OffOffClosedOn*OnOff3033,350- 45OffOffClosedOn*OnOff6016,731- 46Evaporator return bendsOffOffOpenOn*OnOff30- 47Condenser return bendsOffOffClosedOn*OnOff30-
bendsOffOffClosedOn*OnOff30-Matrix Top CU 75033,338
1000 top CU (A)Broad side, centre
Evaporator return bends33,3 Condenser return bends
16,7 40
Annex 2
Measurement Matrix Base CU (C)
Test noRoom (m2)
Remark 1Evap return bendsOffOffNoneNoneNoneNone30Overall baseline 2Cond return bendsOffOffNoneNoneNoneNone30Overall baseline 3OffOffNoneNoneNoneNone103039- 4OffOffNoneNoneNoneNone3020- 5OffOnNoneNoneNoneNone3022- 6aOn (x1)OffNoneNoneNoneNone60 Fan #3 on 6bOffOffNoneNoneNoneNone60Fans off 7On (x 2)OffNoneNoneNoneNone60Fasn #3 + #2 on 8On (x 3)OffNoneNoneNoneNone60Fans #1 + #3 + #4 on; Only do if previous test Cf > 15 g/m3 9aOffOffNoneNoneNoneNone30Cardboard box on top of cabinet to simulate CU 9bOffOffNoneNoneNoneNone30Move rear from wall 3 × min distance (30 cm) 9cRoom centreOffOffNoneNoneNoneNone30Move to centre of room 10OffOffNoneNoneNoneNone105060- 11OffOffNoneNoneNoneNone3030,0- 12OffOnNoneNoneNoneNone30- 13On (x1)OffNoneNoneNoneNone60Fan #3 on (If Cf > 12 g/m3, then do test with one more fan. (I.e., max 2 tests)) 14On (x 2)OffNoneNoneNoneNone60- 15300OffOffNoneNoneNoneNone3010- 16750OffOffNoneNoneNoneNone3025- 482050OffOffOpenNone--301,710 49750OffOffOpenNone--6012,525 50100OffOffOpenNone--601,71214 10 Condenser return bends5
10150Base CU (C)
Narrow end, centre5 10300base CU (C)
Narrow end, centre
Matrix Base CU Condenser return bends8,3 20,0 Evaporator return bends
Annex 3
Graphical presentation of all tests made with the Top CU cabinet
Figure 8a – Test CabA17
Figure 8b – Test CabA18
Figure 8c – Test CabA19
Figure 8d – Test CabA20
Figure 8e – Test CabA21
Figure 8f – Test CabA22
Figure 8g – Test CabA23
Figure 8h – Test CabA24
Figure 8i – Test CabA25
Figure 8j – Test CabA26
Figure 8k – Test CabA27
Figure 8l – Test CabA28
Figure 8m – Test CabA29
Figure 8n – Test CabA30
Figure 8o – Test CabA31
Figure 8p – Test CabA32
Figure 8q – Test CabA33
Figure 8r – Test CabA37
Figure 8s – Test CabA38
Figure 8t – Test CabA39
Figure 8u – Test CabA40
Figure 8v – Test CabA41
Figure 8w – Test CabA42
Figure 8x – Test CabA43
Figure 8y – Test CabA44
Figure 8z – Test CabA45
Figure 8aa – Test CabA46
Figure 8ab – Test CabA47
Annex 4
Graphical presentation of all tests made with the Base CU cabinet
Figure 9a – Test CabB1
Figure 9b – Test CabB2
Figure 9c – Test CabB2b
Figure 9d – Test CabB3
Figure 9e – Test CabB4
Figure 9f – Test CabB5
Figure 9g – Test CabB6
Figure 9h – Test CabB6b
Figure 9i – Test CabB7
Figure 9j – Test CabB8
Figure 9k – Test CabB9
Figure 9l – Test CabB9b
Figure 9m – Test CabB9c
Figure 9n – Test CabB10
Figure 9o – Test CabB11
Figure 9p – Test CabB12
Figure 9q – Test CabB13
Figure 9r – Test CabB15
Figure 9s – Test CabB16
Figure 9t – Test CabB48
Figure 9u – Test CabB49
Figure 9v – Test CabB50