IEEE Power Electronics Magazine Compendium - March 2018 - 116

lifetime model parameters and an
IGBT electrical parameter. By adding
the variance to the parameters for reliability analysis, the lifetime of components is no longer a constant value but
is distributed in a certain range. The
details of the procedure and the case
study can be found in [34].

Mission
Profile

Input

PV Inverter
(with Specific Topology and
Control Scheme)
Mission Profile
Translation
ElectroThermal Tj, ∆Tj
Model of
IGBTs

Lifetime Distribution with
a Required Confidence
Level (e.g., 90%)

Lifetime
Lifetime Consumption MonteCarlo
Model of
Analysis
IGBTs

Multicomponents reliability
assessment of converter system

After the reliability (or the curve of
probabilities of surviving until a specific running time) of individual power
electronics components under the
Parameter Variations
given mission profiles and converter
Design Iterations to Select Proper IGBT Modules for a Given
design is generated, the overall reliDesign Specification
ability of the whole converter system
can be decided depending on the conFIG 17 A Monte-Carlo-based method for lifetime prediction of IGBT bond wires [34].
nection logics of the components. For
the N components connected in series
logic, the overall reliability RN_series(t) of the system can be
To take into account the prior variances, a Monte-Carlobased method for the lifetime prediction of an IGBT module
calculated as
bond wire is proposed in [34]. The flowchart of the method
N
is shown in Figure 17, which is studied based on a photovolR N_series (t) = % R x (t),
(6)
x=1
taic (PV) inverter application. The IGBT bond-wire lifetime
model proposed in [35] is applied
where Rx(t), which is a time-dependent variable, represents the reliability of an individual power electronics
b2
b b
m t on
component under the given mission profiles and conN f = ATT hj c T
I Vb Db ,
(7)
j.min
verter design.
For the N components connected in parallel logic, the
where ∆Tj is the junction temperature fluctuation, Tj,min is
overall reliability RN_parallel(t) of the system can be calcuthe minimum junction temperature, ton is the heating time of
the power cycling, V is the blocking voltage of the chip, D is
lated as
the bond wire diameter, I is the current per wire, and A,
N
b 1, b 2, b 3, b 4, b 5, and b 6 are the constant parameters as disR N_parallel (t) = 1 - % 61 - R x (t)@ .
(7)
x=1
cussed in [35]. Figure 18 shows the considered variances in
2

1) Probability
Distribution Function

4

5

Output

6

2) Sensitivity Study of Individual Parameter Variation

pdf
pdf

pdf

β1

pdf

β1

β2

3) Variation of All Parameters
pdf

Damage

pdf

T j, min

pdf

Vce, On

β2

Damage

β1 T j, min
β2 ∆ T j

Damage
pdf

∆Tj

90% CI Lifetime

Damage

FIG 18 A step-by-step Monte-Carlo analysis of an IGBT module based on (7). b1 and b2: fitting coefficients of the applied lifetime
model; ∆Tj,min: minimum junction temperature; ∆Tj : junction temperature fluctuation; Vce,on: collector-emitter voltage; pdf: probability distribution function [34].

116

IEEE PowEr ElEctronIcs MagazInE

z	 December 2016
Table of Contents for the Digital Edition of IEEE Power Electronics Magazine Compendium - March 2018
