16 Sept’ 20

Impact Analysis of Faults in an EV

Aseem Routray
CTO Organization
KPIT Technologies Limited

1
 EV Model

HVAC Motor
-3 Phase I/M
FEAD 6 1 -Thermal Ports
Regen Optimization -Fault Inclusion
Electrical

Motor Controller
-Inverter
Distribution Box -Controls
DC-DC converters- Dynamics
EV Controls
Contactors-
5 Model 2 -Pre-charge circuitry
-Loss estimation(Datasheet)
Fault Inclusion- -Heat Sinking
-Fault Inclusion

Battery Pack
Vehicle Dynamics Thermal -Battery Modules
Brake Force Distribution- -BMS - LMU and CMU
Longitudinal Dynamics- 4 3 -Communication bus
Maximum allowed Regeneration- -Battery Thermal Management

©2020, KPIT Technologies Limited. All Rights Reserved 2

Framework:
Vehicle FEAD System
Dynamics with HVAC

Individual
Motor Drive

©2020, KPIT Technologies Limited. All Rights Reserved 3

 Individual drive system view Control Strategy

Induction Motor
Distribution Box

Battery
Pack
Inverter with Pre-charge logic

©2020, KPIT Technologies Limited. All Rights Reserved 4

 Battery pack(Exploded View)

Battery Pack

BTMS
Battery Subsystem
SOC Estimation Block

8 x 3 Battery Module

Communication bus
from each Module BMS
Central Management Unit

Individual Battery blocks duplicated

to replicate a battery module
with thermal ports and
passive balancing interfaced with BMS

©2020, KPIT Technologies Limited. All Rights Reserved 5

 Base Libraries
BMS
Central
BMS Local Management
Management Unit
Units
Fuse

LiFePO4 Single Cell Model Relays and Contactors

DC-DC Converter
BLDC Fan
For Motor
Controller Cooling

©2020, KPIT Technologies Limited. All Rights Reserved 6

Test Case: (Vehicle tested for Delhi Drive Cycle)
The rotor speed is seen to track the drive cycle reference

DDC
reference

Rotor Speed

©2020, KPIT Technologies Limited. All Rights Reserved 7

 Case: Motor Air Cooling model Validation

#The data was taken for an entire drive phase

• When tested with available drive data, natural air cooling models were used to depict the
convergence and a maximum tolerance of 9 K was observed in the motor temperature data
generated by simulation.

©2020, KPIT Technologies Limited. All Rights Reserved 8

 Case: Battery Liquid Cooling Model Validation

Battery
Charging Battery
phase@0.6 C Discharging
phase@0.6C

• When validated with available test data, the convergence was seen and a tolerance of 1 K was
observed in the battery temperature data generated by simulation.
• A battery module was used for the experiment.

©2020, KPIT Technologies Limited. All Rights Reserved 9

 Case: HVAC(Heating Ventilation and Air Conditioning)
Model Validation

Soaking Drive phase

phase

• Validated with available data, convergence was seen with a tolerance of 3.5 K was observed in the
cabin temperature data generated by simulation.

©2020, KPIT Technologies Limited. All Rights Reserved 10

 Fault Injection

Motor Faults Inverter Faults Contactor Faults Wire Faults

• Phase to Ground Short • Open Circuit in one IGBT • De-latch fault • Open Circuit
• Double Phase Short Circuit • Open circuit in two IGBTs • Latch Fault • Short Circuit
of same/different leg
• Double Phase to Ground • Chattering Fault • Chattering of connection
• Three Phase Short Circuit • IGBT Short circuit fault
Thermal Faults (integration in Progress)
• Three Phase to Ground • Jitter Fault in one/ both
IGBTs of the same leg • Motor Over temperature
• Single Phase Open Fault
• Short Circuit across DC • Battery Over temperature
• Single Phasing followed by Link through a resistor
another open Phase • IGBT Over temperature

©2020, KPIT Technologies Limited. All Rights Reserved 11

Test Case: ( Motor Single-Phasing Fault)
• The average torque remains the same and the same power is delivered by the remaining two stator
phases. So, value of current in each phase increases and the third phase, being open, carries zero current.
Phase 1 Current

Phase 2 Current

Phase 3 Current

Fault
inserted
at t = 4 s

Fault injection

©2020, KPIT Technologies Limited. All Rights Reserved 12

Test Case: (Contactor Chattering Fault)
• The relay/Contactor shows a repeated ON-OFF behavior due to the chattering fault injected at t=4.5s. This leads to a
pattern showing torque production upto reference torque i.e, 0.4 p.u. and consequent reduction of torque till zero.

Fault
inserted at
t = 4.5 s

Developed
Torque
Load
Torque

Fault injection

©2020, KPIT Technologies Limited. All Rights Reserved 13

Wire Faults :
Wire faults are the faults arising out of a loose or broken electrical
connection which can be categorized into chatter, open and short
circuit at either terminal
User has the flexibility to insert a selected type of wire fault at
defined instants to check effect of faults at various instants
during the operation of the drive

A total of 74 wire fault cases can be permuted based

on the nature of the wire fault to be inserted.

©2020, KPIT Technologies Limited. All Rights Reserved 14

Test Case: (Open circuit Wire fault in Battery supply B+ )
• The battery supply gets disconnected due to fault in wire and the stator currents are seen to decay gradually. The stator
currents after the open circuit fault insertion is supplied by the dc link capacitor.

Phase 1 Current
•

Phase 2 Current
•

Phase 3 Current
•

Fault inserted at
t = 2.8 s

Fault injection

©2020, KPIT Technologies Limited. All Rights Reserved 15

Test Case: (chatter Wire fault in contactor excitation coil)
• Chatter wire fault was injected into the relay excitation coil which is having no direct effect on the power system. The
chatter lasting for 0.5s disrupts the drive behavior. The chatter results in failing of excitation, thereby preventing any
electromechanical action. The drive restores its normal working condition after the fault is removed.
Phase 1 Current
•

Phase 2 Current
•

Phase 3 Current
•

Fault
inserted
between
Normal working Fault injection Restoring Phase t=2 s
(after t = 2.5 s) and
t=2.5 s

©2020, KPIT Technologies Limited. All Rights Reserved 16

 Factors Levels
Type of Wire fault 74
Wirefault Injection Time instance of 3
Arriving at an optimum number of cases injection
Wire selected for ~100
• Pairwise interactions with three Injection
factors:
74C x 3C1 ,3C1 x 100C ,100C1 x 74C
1 1 1
• The maximum number of rows
Study Pairwise
to be considered is 100C1 x 74C1 •Levels of each factor were Interactions •Domain based expertise to
determined be used to reduce the
• All other pairwise interactions are •Total Number of Exhaustive
cases to be determined from
•Applying Taguchi Method to
number of rows
•Final screening will have
get maximum number of test
included within the above number of the levels of each factor
cases considering all
reduced number of injection
cases for output validation
pairwise interactions
cases •Maximum number of rows is
given by the maximum pairs
from the levels of factor
• Final number of cases to be screened Exhaustive pairs
to be done based on domain expertise number of Final Screening
and probability of occurrence of the Wirefault cases
fault based on its physical location or
proximity to certain subsystems

©2020, KPIT Technologies Limited. All Rights Reserved 17

Wire Fault Injection Algorithm
Algorithm architected to run a regression for every wire fault in every subsystem
DIAGNOSTICS/
REGRESSION
ANOMALY DETECTION
FOR FAULT
DATA
GENERATION
Proceed Simulate
Detect Anomaly
to next and log results
with help of
Subsystem/Fault in CSV format
transition matrix

Add the wire Determine the

START Identify Wires Disconnect FINISH
fault block optimum number
For connections
of states showing
Fault Insertion to add a type of
expectancy of
wire fault
each transition
using wire fault library
during runtime

©2020, KPIT Technologies Limited. All Rights Reserved 18

Highlights

Importance Experience and Challenges Benefits

Need to achieve low simulation

Virtual Failure Modes and Effect Minimization of time and
time for executing an iterative
Analysis(FMEA) with occurrence resources used by the
simulation process for different
detection and severity analysis Manufacturers
test cases without compromising
Minimizing the occurrence of much on fidelity
A better sense of trust and
damages and mitigate the Need for extensive data gathering increased belief in the brand
severity of the faults on physical for trying different frameworks amidst the Users
system for fault diagnostics

Makes way for Future Diagnostics

and further Prognostics It presents a perfect blend of
model-based designing and data
sciences for fault diagnostics
Ensures a safer and reliable
mobility experience

9/14/2020 19
Thank You

Let’s reimagine mobility

20