RAMS 2013 paper by Golnaz Sanaie and Fred Schenkelberg
Three-phase inverters are physically large, complex and expensive elements of major solar power generation systems. The inverter converts DC power created by the photovoltaic (PV) panels to AC power suitable for adding to the power grid.
The inverters’ reliability testing is a complex task and relies on reliability block diagrams (RBD), vendor and field data, plus selecting accelerated life tests (ALT) based on critical elements of the product.
This paper illustrates a case study that developed an RBD, used field and vendor data, and includes the design and use of two ALTs. The result is a working framework or model that provides a reasonable estimate of the expected lifetime performance of the inverter. While any project similar to this, is always a work in progress, the examination of the decisions and inputs for the model proves valuable for the continued improvement of the model and resulting life predictions. This project provides an excellent real life example of reliability estimation having multitude of constraints including sample size, test duration, and field data and thus having to rely on all sources of available data starting from field and vendor data to theoretical component reliability calculations, ALT plan execution, failure analysis, and finally summarizing the results using RBD to estimate product expected lifetime. At the time of writing this paper, based on completion of system level ALT, an availability of 99.97% is valid over a 10 year period according to Ontario weather as the main installation base. This will be revisited once subsystem ALT is completed.
2. Schneider Electric – the global specialist
in energy management
Balanced geographies – FY 2011sales
billion € sales Western
North Europe
(last twelve months) 32%
America
23%
Asia
Pacific
Rest of 27%
of sales in new economies World
(last twelve months) 18%
Diversified end markets – FY 2011 sales
people in 100+ countries
Utilities & Infrastructure 24%
Industrial & machines 22%
Data centres 16%
Non-residential buildings 29%
Residential 9%
of sales devoted to R&D
2
4. Estimating Solar Inverter Useful Life
Why?
Increased confidence in expected product life cycle cost leading to
lower risk on extended service contracts.
Higher confidence in predicted availability over time
How?
1. Establish reliability block diagram (RBD) and calculate theoretical
Availability
2. Review field data and determine aging trends
3. Select critical parts for ALT based on reviewing: DFMEA, component
derating, reliability predictions, and field data
4. Design ALT plans
5. Based on ALT results update RBD and estimate availability at 10 years
Note: This study was done on GT500MV as pilot project
4
5. Content
Establish reliability block diagram (RBD) and calculate theoretical
1
Availability
2 Review field data
3 Select critical parts for ALT
4 Design ALT plans
5 Summary & conclusion
5
6. System Reliability Block Diagram
Select critical parts for detailed reliability calculation based on:
Component derating calculations
Design analysis with engineering
Reviewing vendor data
Reviewing DFMEA
Critical blocks: Fans, AC/DC caps, IGBT’s
6
7. Detailed reliability calculations
t
( )
R(t ) e
●Inductor cooling fans:
L10@Ttest of 40C is 59,300 hours
Measured fan temp @45C =66C
L10 @Tuse of 66C is 34,652 hours
T T t
-> η = 73,367 hours, β= 3 ( test use )
AF 1.5 10
ln( R (t ))
R(10 years) = 94%
(Similar calculations for power bridge /circulating fans)
●For DC bus capacitors: Ea 1
T T
1
EM
7
L LB e
k U M
-> η = 292,000 hours, β= 7 E
U
(Similar calculation for AC capacitors)
●For IGBT’s:
L1 10 11.359 (T C) 3.81
-> η = 2,116,824 hours, β= 2.3
7
8. System Availability
Calculate Availability based on:
a) Detailed Weibull reliability calculation for critical parts
b) Exponential assumption for all other parts based on their MTBF
c) Average Mean-Time-To-Repair for each blocks (field data)
t
A(t ) R(t ) R(t u )m(u )du
0
Availability in 10 Years = 99.97%
8
9. Content
Establish reliability block diagram (RBD) and calculate theoretical
1
Availability
2 Review field data
3 Select critical parts for ALT
4 Design ALT plans
5 Summary & conclusion
9
10. Review Field Data
(Based on 85 samples and 9 months field data)
Non-repairable systems analysis using Weibull++:
● Time to failure analysis resulted in β <1
● Time to repair analysis indicated improving trend
Repairable systems: time between failures increasing
Failure Rate
MCF vs. Time
Therefore, no aging trend is found
Time
10
11. Content
Establish reliability block diagram (RBD) and calculate theoretical
1
Availability
2 Review field data
3 Select critical parts for ALT
4 Design ALT plans
5 Summary & conclusion
11
12. Select Critical Parts for ALT
Selection was made based on:
Detailed reliability calculations
Component derating calculations
Design analysis with engineering
Reviewing vendor data,
Reviewing DFMEA
IGBT and DC Buss capacitors were identified as main drivers of
product aging
12
13. Content
Establish reliability block diagram (RBD) and calculate theoretical
1
Availability
2 Review field data
3 Select critical parts for ALT
4 Design ALT plans
5 Summary & conclusion
13
14. Design ALTs for Subsystems and System
Inverter bridge ALT based on IGBT aging mechanism (temperature
cycling is the driving factor for aging)
Nu Tu 3.831
Coffin-Manson equation AF ( )
Nt Tt
● Ontario was chosen as representative environment:
Environment ΔTc at 480Vdc AF B=-3.831 Test cycles for 10 years
Ontario ΔTc 90th percentile, 69.6°C 34 108
Ontario ΔTc 50th percentile, 63.5°C 48.6 75
System ALT testing based on DC buss capacitor aging mechanism
Peck’s equation RHu 3.0 Ea 1 1
AF ( ) exp ( )
● Ontario weather data: RHt k Tu Tt
Ontario Weather AF Test hours for 10 years
90th percentile, 30°C/90% RH 19 4,660
50th percentile, 15°C/70% RH 241 364
14
15. ALT Plan Execution
Test setup for Bridge ALT Bridge shown inside HALT
chamber
Test setup for System ALT Inverter System shown
inside the walk-in chamber
15
16. ALT Reliability Statements
Confidence level:
ln(1 C )
n
ln( R)
Sample Limitation: 2 samples of system and 6 samples of Inverter
bridge. Therefore, ALT plans are stated as:
Testing 6 Bridges within a test chamber set to cycle from -40°C to 90°C for 108 cycles
simulates 10 years of temperature cycling at Ontario’s 90th percentile weather
conditions, which if the unit operates without failure demonstrates 80% reliability with
74% confidence.
Testing 2 samples of Inverter Systems at 60°C and 85%RH for 326 hours simulates 10
years (8 hours per day) usage at Ontario’s nominal weather conditions, which if the unit
operates without failure would demonstrate a 80% system reliability with 36%
confidence.
Note: this project is focused on providing the approach for life prediction and low
confidence level is due to sample size
16
17. Content
Establish reliability block diagram (RBD) and calculate theoretical
1
Availability
2 Review field data
3 Select critical parts for ALT
4 Design ALT plans
5 Summary & conclusion
17
18. Summary and Conclusion
Based on completion of system level ALT, an availability of 99.97% is
valid over a 10 year period according to Southern Ontario weather as
the representative environment with less than 36% confidence.
This project provides a real life example of using ALT to:
Estimate long term product availability
Gain assurance in product life cycle cost
Reduce risk on extended service contracts
Increase customer satisfaction
18