NetCAM HPIC 2012

Report
LA-UR-12-24875
Canberra NetCAM, Dynamic
Radiation Source and CAM Alarm
Modeling
James T. Voss
Jonathan A. Hudston
Tom McLean
RP-2 Group
Los Alamos National Laboratory
Los Alamos, NM, 87545
Presented at 2012 HPIC Meeting,
UNM-LA, Los Alamos, NM
September 24-26 2012
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 1
Introduction: Outline

Introduction
•
•
•

Evaluation recap
Evaluation update
Suggested areas for future improvement
Current NetCAM performance
•
Alarm algorithms and set points
—
•

Alarm modeling
Dynamic radiation source testing
Conclusions
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 2
Introduction

Canberra NetCAM evaluation began 9/2008 at LANL
•
•
Selected as candidate for continuous air monitor at the RLUOB facility
Perceived advantages of NetCAM dongle over ASM1000
—
—
—

Immediate problems found with:
•
•
•
•
•

Cost ($3.5 K cheaper than ASM1000)
Networking capability (built-in web browser)
Peak-shape fitting algorithm included
Hardware
Firmware
User interface
Intra and Inter-communications
Documentation incomplete
Spent next 3.5 years resolving these issues
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 3
Introduction

NetCAM dongle
•
Up to 8 CAM heads can be connected
—
•
•
•
•

but 1:1 configuration selected for RLUOB
RS-232 output to PC ( terminal emulator) console program
RJ-45 ethernet connections (unit has built-in web browser)
Remote monitoring using RadHawk (RadNet-compliant) listener
Has wireless capability too ( not used at LANL)
AS1700 CAM head
•
•
•
•
1700 mm2 PIPs detector
Efficiency of ~32% for electroplated distributed 239Pu source
Flow rates ~ 2cfm
Original firmware: version 1.10 (now have 2.4)
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 4
Canberra NetCAM


Panel PC functions as local display (runs embedded Win XP)
Dongle configuration

2 RJ-45 ports

RS-485 (to CAM head)

RS-232 for console connection
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 5
Recent issues and resolutions

Power supplies for Panel PC and NetCAM dongle not UL-listed
•
•

Sigma-based DAC-h alarm limit not correctly calculated
•

Also leakage voltage of >30v AC measured on dongle
Resolved using quality power supplies
Issue fixed by Canberra
Acute false alarm rate abnormally high
•
•
Issue identified through modeling of NetCAM performance (discussed later)
Issue fixed by Canberra
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 6
Canberra NetCAM

Extensive list of required fixes satisfactorily completed earlier this year
•
•
•

Acceptance test passed 7/2012
•
•
•

Now offers reliable, robust operation
Able to automatically reboot to restore normal operation
Couples low detection limits with low false alarm probability
Alarm response tests (acute and chronic)
Performance tests
Reliability tests
54 NetCAM units delivered to RLUOB facility in 8/2012
•
Additional 13 units purchased as spares
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 7
Future NetCAM improvements

Revise calculation of sigma-based DAC-h alarm limit i.e. :
•
•

Modify automatic energy calibration scheme
•

Currently too restrictive and unable to locate or track 7.69 MeV peak
Modify performance test algorithm
•

Net TRU counts = Gross counts – sum of tail contributions
Variance in Net counts = Gross counts + sum of tail contributions
Currently takes >7 minutes whereas ASM1000 took ~ 2 minutes
Allow user to select chronic analysis update frequency
•
Currently fixed at 4 minute intervals
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 8
NetCAM alarm algorithms: acute alarm

Acute alarm solely resides with the Alpha-Sentry CAM head
•
Based on a user-set count interval (6 - 60 seconds)
•
Counts in TRU region (2.8 - 5.8 MeV by default) and Rn region (5.8 - 6.0
MeV by default) summed
•
Alarm sounds if following conditions satisfied
- the number of counts per channel in the TRU ROI is twice that of the Rn ROI
- the number of TRU ROI counts exceeds the user-set minimum
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 9
NetCAM acute alarm set points

Traditional LANL acute alarm set points;
•
•
•

12 second count time
80 or more TRU ROI counts required to generate an alarm
Default ROI boundaries used
Experience has shown that these settings adequately prevent false
alarms but are they optimal ?
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 10
Acute alarm optimization: Spreadsheet analysis tool
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 11
Acute alarm optimization: Spreadsheet analysis tool
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 12
Acute alarm optimization: Spreadsheet analysis tool
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 13
Acute alarm optimization: Spreadsheet analysis tool
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 14
Acute alarm: Calculated 239Pu DAC-h activity at TRU count rates
corresponding to 1 false alarm per year per 60 NetCAMs
Count time (s)
Minimum
TRU counts
Average
cpm
Average
239Pu DAC-h*
6
16
160
14
18
24
80
7.0
30
29
58
5.1
* Assumptions: 2 cfm, 30% detection efficiency, DAC factor = 5E-12 μCi/cm3 and energy
calibration is correct
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 15
Acute alarm: Calculated 239Pu DAC-h activity corresponding to
detection probabilities of 50% and 95% per count interval
Detection Prob. = 50%
Detection Prob. = 95%
t(s)
TRU
counts
cpm
DAC-h
TRU
counts
cpm
DAC-h
6
28
280
25
69
690
61
18
97
323
28
166
553
49
30
166
332
29
248
496
44
* Assumptions: 2 cfm, 30% detection efficiency, DAC factor = 5E-12 μCi/cm3
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 16
Modeling of NetCAM alarm response

FORTRAN program written to simulate NetCAM performance
•
Code samples background and TRU spectral distributions specified by user
•
Respective total count rates independently set by user
•
Poisson stats used for number of bkg. and TRU counts and associated energies
per 6 second update frequency
•
Both contributions are summed to form an integrated spectrum
•
Performs acute and chronic alarm (Valley mode) analysis under conditions
specified by user
—
analysis frequency, ROI settings, cycle time, alarm set points, etc …..
—
valley (tail-fitting) mode used for chronic analysis
•
Both true and blind man’s differential approaches are considered
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 17
Acute alarm modeling vs spreadsheet predictions
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 18
Acute alarm: Calculated average time-to-alarm as function of
average 239Pu DAC-h activity
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 19
NetCAM chronic alarm algorithms

Blindman’s differential approach used for NetCAM chronic analysis
•


Spectrum refreshed at end of each count cycle
Valley mode
•
Sequential exponential tail-fitting and subtraction of tail counts
•
Net counts in TRU ROI used to determine activity
—
recent improvements avoid non-physical net TRU cpm results
•
Uncertainty calculation incorrectly implemented by Canberra
—
grossly overestimates uncertainty in net counts
—
compensates by using a relatively small kσ factor
•
Alarm sounds when the fixed DAC-h limit and sigma-based limit are exceeded
—
an analysis every 4 minutes and at end of count cycle
Peaks mode
•
Not seriously considered as default analysis mode after some early problems

UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 20
Chronic alarm modeling
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 21
NetCAM alarm modeling: conclusions

Code appears to emulate NetCAM behaviour well

Predictions are dependent on background spectrum and count rate

Current number of TRU counts required for an acute alarm appears
to be too conservative

Valley analysis mode capable of 239Pu detection limits of 2 DAC-h
with negligible false alarm rates based on available Rn/Tn
background data
•
•
Count cycle times of about 12 minutes appear optimal
Alarm response time can be as good or even better than true differential approach
if NetCAM algorithm allowed freedom to analyze data more frequently
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 22
Dynamic radiation source

Problem:
•

Evaluation of CAM heads (sensitivity, time-to-alarm)
—
Currently dependent on radioactive aerosols
—
Time intensive, expensive and requires specialized facility
Solution:
•
Dynamic Radiation Source (DRS)
—
Mimics the challenge of plutonium aerosol detection
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 23
Production DRS: Overhead view
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 24
Introduction: Advantages of DRS

Provides non-specialized in-house testing

Low cost (~2K) versus ~10K per aerosol test

Multiple test scenarios with various CAMs

Reproducibility

Supports iterative development of CAM analysis algorithms

No contamination issues

Rn/Tn background spectrum also present
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 25
DRS: Alpha Sentry/ASM1000 count rate variation
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 26
DRS: Alpha Sentry / NetCAM dongle test data
15 minutes
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 27
DRS: Alpha Sentry / NetCAM dongle test data
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 28
DRS: Alpha Sentry / NetCAM dongle test data
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 29
DRS: Alpha Sentry / NetCAM dongle test data
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 30
DRS: Alpha Sentry / NetCAM dongle test data
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 31
Result summary: Average time to alarm (2 DAC-h limit)
CAM
Analysis
mode
Cycle
time
(min.)
Average time to
alarm (min.)
Std. dev.
(min.)
AS-1700R / NetCAM
Valley
2
10
2
AS-1700R / NetCAM
Valley
9
11
2
AS-1700R / NetCAM
Valley
17
10
2
AS-1700R / NetCAM
Peaks
2
8
3
AS-1700R / NetCAM
Peaks
9
9
3
AS-1700R / NetCAM
Peaks
17
9
3
AS-1700R / ASM1000
Valley
15
15
0
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 32
Conclusions

Canberra NetCAM now capable of providing reliable operation and
protecting workers
•
•
•
low alarm limits coupled with low false alarm probability
optimized alarm set points can be calculated using modeling
example of an ultimately successful collaboration between vendor and customer

Further beneficial improvements to NetCAM are readily achievable

DRS shown to be a useful tool in evaluating CAM chronic alarm
algorithms
•
Empirical data lends support to the modeling predictions.
UNCLASSIFIED
Operated by Los Alamos National Security, LLC for the U.S. Department of Energy’s NNSA
Slide 33

similar documents