ML20063P345
| ML20063P345 | |
| Person / Time | |
|---|---|
| Site: | Clinch River |
| Issue date: | 10/07/1982 |
| From: | Jerrica Johnson, Matlin E, Witherspoon J ROCKWELL INTERNATIONAL CORP. |
| To: | |
| Shared Package | |
| ML20063P343 | List: |
| References | |
| NUDOCS 8210130315 | |
| Download: ML20063P345 (15) | |
Text
.-
LIQUID METAL-TO-GAS LEAK DETECTION INSTRUMENTS
- E. Matlin, J. E. Witherspoon, J. L. Johnson Rockwell International, Energy Systeers Group Canoga Park, CA 4
ABSTRACT It is desirable for liquid-metal-cooled reactors that small liquid metal-to-gas leaks De relianly detected. Testing has been performed on a number of detec-tion systems to evaluate their sensitivity, response time, and performance charac.
teristics. This testing has been scheduled in three phases. The first phase was 1
aimed at screening out the least suitable detectors and optimizing the perfomance of tne most promising.
In the second phase, candidates were tested in a 1500 ft3
" walk-in" type enclosure in whicn leaks were simulated on 24-in, and 3-in. piping.
In the third pnase of testing, selected type detectors were tested in the 1500-ft3 enclosure with Clinch River Breeder Reactor Plant (CR8AP) pipe insula-tion configurations and detector tubing configuration with cell gas recirculation simulated. Endurance testing of detection equipment was also performed as part of-this effort.
4 Test results have shown that aerosol-type detectors will reliably detect leaks as small as a few grams per hour when sampling pipe insulation annu11. Sensitivity is improved wnen the leasing liquid metal temperature is above 5000F. When cer-tain conditions are met, direct aerosol sampling of a cell will provide adequate sensitivity to detect small leaks. Leaks greater than several hundred kg/h can ne detected in minutes wnen annuli or small call atmospheres are sampled. Aerosol-type detectors appear to be relatively insensitive to typical reactor environmental pollutants.
INTRODUCTION The use of liquid metals (sodium or a sodium-potassium alloy) as heat transfer fluids in nuclear power plants necessitates a means of detecting leaks of these fluids from pipes and vessels. It is often desirable that even very small leaks, sometimes referred to as " weeping" leaks, be detected in relatively short times.
Tnis is especially true for leaks wnich occur where the pipe or vessel surface is exposed to an air atmosphere because corrosion by sodium reaction products at high temperature can cause the leak size to grow. Larger leaks will increase the clean-up and repair time. For leaks in inerted cells, corrosion proceeds very slowly.
However, for the main primary heat transfer piping and components usually contained in inerted cells for loop-type reactors, it is prudent to detect even small leaks, l
low corrosion rates notwithstanding.
i l
Historically, the approach to leak detection in liquid-metal-cooled reactors l
has been to attempt to detect only relatively large flowing leaks. The method most i
commonly used depends upon the electrical conductance of liquid metals and utilizes either probe or cable-type detectors Installed below pipe and equipment. The probe-type detector consists of a rigid cable in which the insulated conductors are exposed at one end. This end is installed vertically just underneath a hole pro-vided in the inner sheath of the pipe or vessel insulation. Detection requires that
- This work was performed under DOE Contract EY-76-C-03-0824 1
1
~
(8210130315 821007 l
(A__
page 1
\\
the leaking liquid metal flow from the leek site to the insulation sheath, down the sheath to the hole, and onto the exposed conductors, where it causes an electrical short. The cable detector is made up of an assembly of two or more conductors insu-lated by glass braid or ceramic beads. The cable is mounted horizontally below the pipe or vessel. When a leak occurs, the flowing liquid metal impregnates the insula-tion, causing the conductors to be shorted. These types of detectors have on occa-sion detected large leaks but have been shown to be unreliable, particularly for detecting small leaks. Also, f aise alarms have taken place due to damage during construction or later due to the entrance of moisture.
Because of-the need in liquid metal fast breeder reactors (LMF8Rs) for sensi-tive, reliable leak detectors, an extensive development program was started in the early 1970s. The initial phase of this provam was the testing of candidate detec-tion systems in a small test chamber. Based on the results of these " screening" tests, the best performing systems were tested in a large " walk-in" enclosure using more prototypical pipe and detector configurations over the range of expected LMFBR environments. These tests are referred to as the " mockup" tests. With the know-ledge gained from the mockup tests, a series of verification tests was performed using expected CRBRP detector and pipe insulation configurations and cell gas recir.
culation rates. Long-term endurance tests were also cnmoleted. The results of the mockup and verification tests are presented in this paper.
LEAK OETECTION METHODS TESTED The " screening" tests evaluated the following type detectors: (1) contact-and cable-type conductivity, (2) aerosol transport, (3) product of combustion, (4) chem-ical, and (5) gas analysis.
l The principle of operation of the conductivity detectors has been described previously. An illustration of the probe-type detector tested is shown in Fig-urc la. For these tests, in addition to glass braid insulat.ed cable, a configura-4 tion as shown in Figure Ib was tested. This type detector employs metal-sheathed, magnesium. oxide-insulated, two-conductor cable. The sheath is milled out in selec-ted locations to expose the conductors.
Leaking liquid metal forms an electrical bridge between the conductors and the sheath at these locations.
Two' types of aerosol detectors were tested. These sense the aerosol generated l
by the reaction of sodium vapor with oxygen in the presence of at least small quan-tities of moisture. The plugging filter aerosol detector (PFAD), Figure Ic,
(
embodies a differential pressure sensor connected across a filter. Gas from the leak site is drawn through the filter. Upon a liquid metal leak, the sodium aerosol carried by the gas collects on the filter and the pressure drop increases. This pressure drop is then monitored and provides an alam when it approaches a preset limit.
In addition, chemical analysis of the residue on the filter provides confir-mation of the liquid metal release.
A second type of aerosol detector, referred to as a sodium ionization detector (SID), Figure Id, uses a heated filament to ionize particles in the gas that flows through it.
The particles are collected by an electrode to provide a measure of ion current. By coeration of the filament at the proper temperature, only sodium, or a species with a lower ionization potential, will become charged with the resultant ions collected.
l
____ _ page 2
4 The product of comoustion detector tested was a coannercially available unit wnica operates on an ionization principle; invisible or visible combustion products
~
wnien enter the detector's outer chamoer are ionized oy a small radiation source and disturb the salance between tnis channer and the inner reference chamber. This imoalance generates tne output signal.
A pH detector was evaluated during the initial phase of testing. The gas to be sampled for sodium specie was buobled tnrough water and the resulting change in pH was measured.
Gas analysis refers to the measurement of the oxygen or hydrogen concentration oefore, during, and af ter a simulated liquid metal leak.
The first series of tests, the screening tests, demonstrated that the most promising type of detectors from the standpoint of sensitivity and adaptability for reliable LMFBR application were the aerosol types. The hydrogen and smoke detectors showed some promise; conductivity held out some possioility for use in special dpplications. Therefore, these four detection approaches were selected for the mockup test series.
MOCKUP TESTS Test Chamber The test chamber Figure 2, for the mockup tests consisted of an approximately 1500 ftJ steel-lined rectangualar enclosure, which could be sealed. For testing in air, moisture could be introduced and the concentration controlled. For nitrogen testing, noth motsture content and oxygen content could be controlled. A conerger-cial smoke detector was installed near the enclosure ceiling. Cell atmosphere gas sampling was provided for monitoring for oxygen, hydrogen, and moisture and for col-lecting sodium on a filter for analysis.
Test Articles Severr.1 test articles were used in the mockup test series.
In the first series of tests, the test article consisted of a section of 24-in. stainless steel pipe approximately 10 ft long.
Insulation consisted of an inner stainless steel sheath.
5.5 in, of calcium-silicate insulation, and a stainless steel outer sheath. The diameter of the inner insulation sheath was such as to provide a 1-in. annulus catween the pipe and the insulation. The pipe section was heated internally and was mounted horizontally as shown in Figure 2.
A tube installed in the top of the annu-lus was connected to one port of a 7-port manifold, the exhaust of which was passed tnrough a'SID and PFAD in parallel and thence back into the enclosure using a vacuum The other ports of the manifold were left open to simulate the presence of pump.
otner sections of pipe sampled by the one SID or PFAD.
A gas inlet port was pro-video at a point underneath tne pipe, upstream of the sampling port.
The leak was simulated by loading sodium in a storage tank and controlling the cover gas pressure and the heating rate so that a controlled leak rate could te estaaltsned througn expanston of the sodium. The sodium injection line initially i
entered above the top of the pipe so that the sodium dripped on the pipe. This was later changed so that the injection line penetrated the pipe from the inside.
causing the sootum to flow out through a machineo note in the pipe to more closely '
simulate an actual leak.
i
The test article was changed somewhat for the 24-in. pipe natural convection tests. The 24-in insulated pipe was retained, but as shown in Figure 3, the annu-lus sampling line was removed and replaced with an upper vent tube. The aerosols generated by the leak were thus transferred only Oy natural convection to the test enclosure. The cell atmosphere was sampled for aerosol content by externally located PFAQs and SIDs as described previously.
The last test series utilized a 3-in. diameter test article designed to simu-late auxiliary system piping in an LMFBR. The test module was an 80-in. long stain-less steel pipe which was heated by a tubular heater banded to the outside of the pipe.. Thermal insulation, 4.5 in, thick, was mountec on stainless steel foil, which was wrapped around the pipe. A stainless steel foil cover surrounded the entire assemoly (pipe and insulation). For the last series of 3-in. pipe natural convec-tion tests, the above configuration was changed somewnat to provide a more complete annulus between the pipe and insulation,.as shown in Figure 4, for the reasons dis-cussed below.
TEST RESULTS 24-in. Pipe Annulus Sampling The test conditions for the 24-in. pipe annulus sempling tests are shown in Table It the results of these tests are given in Table II.
It is seen that by sampling the annulus between the pipe and insulation, very small leaks (0.4 to 100 g/h) can be detected by either SIDs or PFADs. The response time will vary from a few minutes to days, depending on the conditions, particularly liquid metal temperature, with the SID generally responding f aster. Calculation of the aerosol concentration in the gas en$ering the SIDs or PFADs indicates the detec-tion threshold to be approximately 10-lig/cmJ. Test 2 is of interest in uncer-standing the dependence of aerosol generation on ten'perature.
In this test, the 0
sodium temperature was varied from 400 to 700 F.
It was seen that at approxi-nately 6000F, a substantial change in aerosol concentration occurred, probably due to the melting of sodium hydroxide (NaCH).
The response of the conductivity-type detectors we's not consist t; where leak indication was obtained, it was generally obtained for the larger leak, high-temperature conditions.
24-in. Natural Convection, Cell Sampling Natural convection tests were run to detennine if adequate leak sensitivity could be ootained by sampling cell atmospheres in lieu of annulus sangling or as a backup to annulus sampling.
If effective, substantial cost savings would result from elimination of the tubing and manifolds, cell penetrations, etc. Table III l
gives the t(st conditions. Results of the test are shown in Table IV. Note that in I
the 1500-ftJ enclosure, leaks of 2 g/h or more were detected for all test conditions. This was true even for Test 10 where the vent tubes were oeliberataly l
l l
l Page 4 l.-
= _ _
sealed. To apply these data to larger, typical LMFSR cells, it is necessary to take into consideration cell recirculation. Calculations were performeo, assuming that the recirculation system removes a f raction of the aerosol from the recirculated gas prior to returning it to the cell. Based on the calculations, liquid metal leaks of 3
re into 1-in. annu11 can be gtecteg by cell sampling in cells up
.300 g/cm or to 100,000 ft In these calculations, 10*
g/cm is used as tne threshold of detection.
3-in. Natural Convection Cell Sampling For small auxiliary piping, annulus monitoring, though effective, would be very costly and in many cases not justified by the consequences of a leak. Therefore, all small piping testing was oriented toward verifying the practicality of-utilizing cell atmosphere monitoring. The test conditions for these tests are shown in Table V; the results are presented in Table VI.
In the first two tests, 16 and 17, the inner insulation sheath was wrapped relatively t1ghtly around the pipe and heaters, providing an irregular annulus, with no space where the sheath contacted the pipe. As seen in Table VI, effective detection did not take place. Calculation of the aerosol concentration in the cell indicated that the levels were below the i
octection threshold for aerosol detectors. Examination of the test article follow-Ing these tests revealed that most of the leaked sodium had not reacted and was con-tained by the inner metal sheath. Accordingly, in Tests 18 through 21, the insula-tion design was altered to provide a complete annulus around the pipe as shown in Figure 4.
With the modified insulation design, otitection was achieved with the SID and PFAD detectors in all cases. Detection times variec from a few minutes to approxi-mately I day. Similar results wert achieved by the h drogen detector.
i 3 g/h). Detection f
In Test 21, a relatively large leak was simulated (243 x 10 time was a few minutes for both SIDs and PFAQs, indicating these detectors respond rapidly to large leaks. Rapid detection was also achieved by the hydrogen detector.
3 enclosure As with the 24-in. pipe natural convection testing, the 1500-f t data must be converted to expected results for larger cells with recirculating gas flows. ~
VERIFICATION TESTING Verif cation testing of SIDs and PFADs has also been completed. The first mode of testi g utilized the 1500-f t3 enclosure to repeat tests previously performed on 24-and J-in. piping but with several major changes. First, the atmosphere wthin the enclosure was recirculated at an equivalent rate of 105/ min, simulating typical CRBRP cells. Second, the piping insulation as well as heater and thersnoccuple and support penetrations were designed to match the planned CRBRP design. Finally, the annulus sampling manifold, including total length of tubing from the pipe to the detecters, was-designed to duplicate the present CR8RP design.
The test conditions and results are shown in Table VII.
It is seen in Tests 1 through a, where the detection is based on PFAD monitoring of annu11, detection of the 100 g/h leak took place in less than 7 h over the temperature range of 655 to 1010or for both nitrogen and air environments.
In this temperature range, response in nitrogen was superior to that in air For direct cell monitoring where SIDs were employed, the response time was com-parable-(in some cases faster), with the maximum time being 3.5 h.
It must be kept ra5..
in sind that the response time for direct cell monitoring is a function of cell volume, circulation rata, and greater or lesser opportunity for settling and plate-Thus, on many plant cells, direct cell monitoring can be expected to be less out.
effective than annulus monitoring.
The second mode of verification testing is the long-term operation of SIDs and PFADs in an LMFBR-type environment.
In previous testing, the various detectors were operated only during the performance period of each test.
It is desirable to obtain data on the endurance capability of the detectors and their susceptibility to mal-function due to impurities in the sample gas nct related to sodium species.
Accordingly, arrangements were made to install sampling lines in the upper plena of an evaporator and a superheater, part of the steam generation system for EBR-II. These sampling lines are connected to aerosol detectors; SIDs and PFADs whien are continuously operated.
Testing was started on April 4, 1977. and was completed in Aagust 1982. The folicwing has been observed.
1)
PFAD filter membranes need not be changed more frequently than every 8 weeks to avoid significant DP cell pressure ouildup due to the col-1ection of atmospheric dust.
2)
The electrical calibration of the PFADs has been extremely constant.
and total recalibration and maintenance time has been insignificant.
3)
SID maintenance has also been low and has been essentially limited to filament replacement required every 6 to 10 months.
4)
Activities of maintenance personnel (cigarette smoke, increased dust generation) were apparent from the SID output trace.
SUlmARY AND CONCLUSIONS r
Liquid metal-to-gas leak detection testing has indicated that conductivity type l
leak detectors (probe and cable) have a usefulness limited to large leaks or leaks in a confined space. Annulus-type aerosol sampling, utilizing either SIDs or PFADs, provides a sensitive method of detecting leaks in both air and nitrogen down to the with enhanced sensitivity response time when'the liquid tem-range of several g/h.0 perature is above 500 F.
Response time will vary depending upon liquid metal tem-Aerosol sampling of cells is an effective method of leak perature, leak rate, etc.
detection for moderata-size cells even with relatively high recirculation rates such as 105/ min. Relatively large leaks (200 kg/h) can be detected by aerosol sampling of annulf or small cells in minutes. Long-term endurance testing hes shown both SIDs and PFADs to be relatively insensitive to typicel reactor environmental pollu-tants and otherwise quite reliable.
00010/ nth /edi 4
page 6 1--- - -.
,e T All!.10 1 MOCliU15 IEAK L)CT1?CTION T!*ST CONDITIONS
- s=._..
Total Secli.un g.,g 7,, n g,,
I.s nk Selluns Tent M ',' " I"
atu ro 7;,n, Reina rks 11a to Leaked Oxygen (vppin) g.py (g/h)
(g) h)
No.
1 1%
1,000 980 100 10 1,000 Forced C onvection 2
1%
1,000 400 100 10 1,000 Forced 700 C onvection 3
21%
30,000 640 100 6
600 Forced 920 Conve ction 4
1%
1,000 980
- 0. 4 250 100 Torced C onvection 5
21%
30,000 350 2
200 400 Forced 640 Convecdon 6
21%
30,000 640 2
100 200 Poiced -
~
Convection 20 J,000 Forced 2
7 17,0 1,000 400 100 g
C onvcetion 8
190 1,000 500 2
50 100 Forced 2
C onve ction' I
I l
l l
l l
Page 7
TABLE 11 MOCKUP LEAK DETECTION TEST RESULTS
=
To st No.
Paramete r 1,
2 3
4 5
6 7
8 DP 59 min 26 h 6h 6h 97 h 14 h NR *I NR "I I
I
~
54 min 10 min 33 min 52 min DP (C ell)
Ih 26 h 6h 2h 47 h 11 h NR "I NR *I I
I (Annu:us) 16 min 58 min 8 min 31 min 25 min 25 min SID 7 min 30 h 4h 3 min 1 min 1 min 13 h NR 38 min 47 min (Annulus)
Cable Ih NR 6h NU 104 h 18 min NR NR f;
y (Annulus) 10 min 6 min 13 min C ontact NR!
NR 19 h NR NR NR 8h NR-
/
38 min 4 min (Annulus) liydrogen NU 11 h 5 hr ifR 23 h 20 min 3h 10 h 45 min 4 min 32 min 42 min (Cell) l Smoke ht NU 6h NU 31 h NR NR 8
39 min 4 min 48 min l
(Celli (al Sodium was found present in filter pads, 130 pg.and 110 pg for Test 7 collected in 5 h 22 min; 130 pg and 100 pg for Test 8 coliccted in 50 h.
hv - Not U sed NR - No Re sponse DP - Differential Pressure e
TAIALIC !!!
NATU RAL CONVECTION TEST PARAh1ETERS sodium Total Tcut oxyyen hiois tu re Tempe r-
,,,, g g,,-
Leak 8
- I,*
- No.
($ )
(vppm)
N " l"
. rime Sli.e T ik
- 4. F) g j3 )
(h) g;,,)
9 9("I 1
1,000 980 100 20 350 24 10 *I 1
1,000 980 100 20 625 24 I
11 1
1,000 980 2
100 175 24 12 1
1,000 640 2
100 200 24 13 21 15,000 900 2
100 175 24 14 21 15,000 640 2
100 180 24 15 21 15,000 400 2
100 155 24 (a) Vent tubes were scaled I
TABLEIV NATURAL CONVECTION TEST RESULTS Differential Sodium Pre s sure Ionization Detector Cell De tector Test No.
Floor ceiling Floor Ceiling 2
Area Area Area Area 9
3.9 h 3.9 h 0.25 h 0.25 h 0.4 h NR *I 2.8 h I
10 4.5 h 7.2 h 0.78 h 0.67 h 0.1 h NR 1.3 h 11 1.0 h 0.7 h
< 0.1 h
<0,1 h
< 0.1 h NR NP.
12 12.1 h 12.4 h 1.2h NU NR NR NR 13 11.9 h 14.6 h 0.88 h 0.95 h NR NR NR 14 8.0 h 12.9 h 0.025 h 0.016 h 121.1 h NR NR 15 53.8 h 44.2 h 0.01 h 0.01'h NR NR NR 11R
<0.5 h
< 0.8 h
<1 min
<1 min NR NR NR 13X Z3.6 h 25.4 h 15.2 h 15.2 h 15.7 h NR 0.58 h (a) NU - Not used l
NR - No Responne f
e
T A111 4 Y SM AI,1 PIPING TEST ISAttAMETElts S "I "
- Total -
p;{
1.cak
.g. m p r -
Jcak T.wtisun c
Te st Oxygen M uisturc 3;
, g,,
Rate Leaked N o.
(%)
(vppm) 4 7)
(g/h) gg)
(h)
(in. )
16 1
1,000 500 2
100
-200 3
17 21 15,000 650 2
100
~200 3
18 1
1,000 980 100 4.17
-400 3
19 1
1,000 650 100 10
~1,000 3
20 21 15,000 650 100 10
-1,000 3
21 1
1,000 40 243,000 C.033 -A,000 3
TABLE V1 SM ALL PIPING TEST RES111.TS Diffe rential Sodium Pres sure ionization Detector Cell De tec tor Tet Smoke 1tyr1rogen Floor Ceiling Floor Cetling N '*
Arca Area Area A rara I
I 16 NR NR 116 h NR "I NU "I 12,7 h 17 NR NR NR NU NU NR 18 2.6 h 2.5 h 0.5 h 0.5 h NR 1.0 h 19 27.4 it 14.0 h 3.2 h 5.7 h NR 4.0 h-20 25.9 h "I 17.6 h 0.65 h 0.60 h NR 0.72 h I
< 0.73 thI 0.06 h 0.008 h
< 0.03 h
<0.0s h 21 0.008 h (a) A ead leak was found on the sasopling line of the !)iffe rentist L
Preneure (DP) dets,: tor which prch+nted thu D13 acrone the (ilter from rising.
(b) Recorder pen stuck, exact time of re e.ponna cannot lee datu rminud.
(c) NU - Not tased NR - No rouponse FO 3 10
TABLE Vit VERIFICATION TEST. TEST CONDITIONS AND StNIARY 0F RESUl.TS Test Number 1
2 3
4 5
.6 7
8 A1N N
"2 Air Air N
N Air Air 2
2 2
l Detection Time 1.3tl 0.CN 7.0H 6. 6 11 3.5H 1.2H 0.8H 2.OH i3 Detector Annulus Annulus Annulus Annulus Cell Cell Cell Cell l'S PFAO PFAD PFAD PFAD 510 SID SID 510 SodIunandPige 995 730 936 650 500 650 500 650 w
l.
Temperature ( F)
Nominal l
Actual, Pipe Tempera-1000/1010 735/735 937/951 655/670 510/525 655/655 515/565 655/680 l
I ture l F)
(Start / Detect)
I i
i 1
gg geog ATN 1885ukA1H3N SMfAfN N
ExpuSED C-1 e
SMEAT Ms0 INSULATION h
a hesc 8NSULAT1080 EseO SEAL ti.CASLtDETecTOR
- a. emost MTECTOR GASFLOW pmEgrusts P#tSSVns Tmaae80gCE#
SODiueA AND SOctutA 8PFl433VR E DeF F EfitN TIALI R8ACTtON#GQQuCT 3:40eCoTOR OffEC704
- AGROSOL WNC
- 11003U48 SON L
pasetsumE
-. / sFFEnf NTIAL eo.Aaa4NT PSL O.OscAvOn
<m sum.,
J..-
Nr
,0 l
1
,u s:
eu.eN, a
, m.
f AEADOUT
_~[
[.
I FROed i
-e-x 07 p SfLAMENT COLLECTOA -
PLATE F LDw CONTMOL Sv$7 EM FtLTER Ss4S VOQ AGE qF V
as. SQOlusa noaseFATION OCTEC10R 85801 s.PFAO DEigCtem eus Figure 1.
Typus of Lozak Detectors page 12
e de SMOKE DETECTOR MANIFOLD 500iUM
,5TORAGE TANS e
il ll 11 m f1 f1
'~
OUMMY YUSE5 IS[c' sto T, soccuM AkNULUS INJGCTION [ptPE ggggggg 3Aup tig
]
stock tNsuLATIoM
{
s-
_g ung 7/W//////////////)(/n/'
?cTsciom,J
/,
-/
u s
(
/
auMes\\-
'/
MOCxuevrst Antictr
/
/
I gas
/j nsTunu u=e
=7:V//M]////////////
""\\YETSCTOR CAeLE AS CONTACT CELL--
j DETECTOn tNLET OCTECTOR SAWLING WNE OXYGEN METER MYOROGEN METEn M015TURE ME7ER CAs se ttump LINE
- pgu,
}
sas+2 Figure 2.
Walk-In Test Chamber page 13
l
/ E[,ont f es
\\n r/}-
fy i
_s
.g vawv / %
\\ wwutus N msutArion A
92843 Figure 3.
Natural Convection l
Test Article G
I l
O
4 w
A l
edbnj
- o
-a
~
f,;5
>5 i:st.r 5-sS A Isa!!
c I
nb i,
m _-
4 6
IM e
0
)
=
m I
k..
it Ig
};
E
' i $
2,-
>o
.5 5*
- i R
g.:
B-a AU E
sa SS It j
2 m
"lu as as b5 G
u ks a
wt QE b
q E
yC 4
a wz w
zyd
., 823 2
B irs 2 --<
e.
e e
g l
7Y3rEL313L
.. -