3. Real-time Adaptive Control Engineering Lab
Simulation and
Validation
Model
Development
Control design
Optimization
Data
Hardware info
Requirements
Constraints
Transient
profiles
Hardware/configuration
recommendations
Subsystem
specifications
Sensitivity analysis
results
Control strategy
Control-oriented
“grey-box” model
3
8. Integrated Power Systems
Power systems that combine multiple
power sources/loads through synergetic
integration
Examples of integrated power systems
– Hybrid vehicles
– All‐electric ships
– SOFC/GT system
8
www.defenseindustrydaily.com
www.techjournal.org
www.americanhistory.si.edu
9. Characteristics of IPS
Multiple and heterogeneous power/heat plants involved
High efficiency and (intended for) self‐sustaining
Close thermal, chemical, mechanical and electrical
couplings
More complex and challenging tasks for control,
optimization and integration
– High efficiency system often operates on or close to the
boundary of admissible state and input sets
– Mobile requirements require fast load following capability
and sufficient power reserve and safety margin
9
11. Power Management for IPS
Coordinate multiple, heterogeneous power plants
(including energy storage devices)
Manage transient operations
Assure safe operation in case of component and
subsystem failure
Facilitate effective system reconfiguration
Achieve optimal performance in terms of power quality
and system operation efficiency
11
12. Optimization in Power Management of IPS
Optimization: A natural formalism for power
management of IPS
– Assure optimal performance during normal operation
– Guarantee effective reconfiguration in case of failures
– Enforce hard and soft constraints
Challenges:
– Computationally intensive (nonlinear dynamics, long
time horizon, mixed form of models)
– Real‐time performance requirements
12
13. Our Approach
Algorithm development
– Integrated perturbation analysis and sequential
quadratic programming (IPA‐SQP) to speed up
optimization
– Sensitivity function approach to explore multiple
time‐scales in IPS
– Incremental reference governor to reduce problem
complexity
Algorithm evaluation and validation
– RT‐Lab for algorithm development and evaluation
13
15. Case Study: IPS for All-Electric Ships
Main Subsystems:
1. PGM: Power Generation Module
• PGM1:Gas turbine
• PGM2:Fuel cell
2. EPM: Electric Propulsion Module
3. ESM: Energy Storage Module
4. PCM: Power Conversion Module
• PCM1: DC/DC
• PCM2: DC/AC
• PCM3: DC/DC
• PCM4: AC/DC & DC/DC
• PCM5: DC/AC
• PCM6: DC/AC
Main features of IPS:
1. Redundant power sources
2. Reconfigurable power flow path
15
16. Case Study: IPS for All-Electric Ships
System representation of a shipboard integrated power
system
DC Hybrid Power System (DHPS)
1. Multiple power sources
2. Multiple power converters
3. Energy storage bank
16
16
17. Test-bed Setup
• Power converter
1. Unidirectional DC/DC (1, 2)
2. Bidirectional DC/DC (3)
• RT‐LAB with 4 targets
• Power supply (1, 2)
• Electronics load (1, 2)
• Energy storage bank 17
18. Explore Time-scale Separation
Main Subsystems:
PGM: Power Generation Module
PGM1:Gas turbine
PGM2:Fuel cell
(~seconds-minutes)
EPM: Electric Propulsion Module
(~ms - seconds)
ZEDS:
Power Conversion Module
(s – ms)
Vital loads
Non-vital loads
DC STBD bus
DC Port Bus
Zone1
PCM1
NV
load
PCM3
Vital
load
PCM2
NV
load
PCM6
Vital
load
PCM5
MEPM
MEPM
Zone2
PCM1
PCM1
NV
load
PCM3
Vital
load
PCM2
NV
load
PCM6
Vital
load
PCM5
PGM2PCM-4
PCM1
PCM-4
AC 4160V/60HZ
DC 600V
PORT 1100VDC
STBD 900VDC
PORT 900VDC
PGM: power generation module
PCM: power conversion module
EPM: electric propulsion module
PGM1
STBD 1100VDC
18
19. Use RT-Lab for Algorithm Development
Explore the trade‐off between
optimality and computational
efficiency
Level 1: static optimization (all
dynamics are considered
infinitely fast)
Level 2: ignore fast dynamics
Level 3: consider both slow and
fast time dynamics
– Calculate the corrections to the
Level 2 solution
CostJ Time required to solve for u
L1
L2
L3
Opt
Opt
Performance
loss due to non
real-time
19
21. Simulation and Validation
With the time‐scale separation
– Better tracking
– The timeliness of the optimal
solution proved to be critical
Power demand
Solution with
Time scale
separation
Solution
without Time
scale
separation
21