A simulation software tool development for

adiabatic CAES with thermal storage

Dr Xing Luo

School of Engineering, University of Warwick

28th February 2018

Outline of the Presentation

1. Introduction

2. Mathematical modelling of key components

3. Application case study

4. Data driven model - machine learning

• Adiabatic Compressed Air Energy Storage (CAES) with Thermal Energy Storage (TES)

is a complex system with subsystem coupling, components interactions and parameter

sensitive.

• Cycle efficiency depends on whole system dynamic behaviours.

• Dynamic modelling of complete systems is essential to provide support for feasibility

studies of EES applications, system optimisation and control strategy development,

and management of grid integration.

Demand for a whole system dynamic modelling and simulation

tool development

1. Introduction

Electric Machine

CAES storage reservoir

Atmospheric air inlet

Heat exchangers

TES hot storage

TES cold storageExhaust

HP & LP turbinesLP & HP Compressors

Control sysExternal grid

Power conditioning

Air pipeline with direction

TES working medium flow

Shaft of compressors/turbines

Electricity connection

Heat exchangers

A Simulink based tool for 1D dynamic modelling & control CAES-TES is developed.

Areas: mathematics, thermodynamics, heat transfer, mechanical & electrical engineering.

Features: model based design, signal I/O connection, case studies, complied and

protected, design & test documentation, initial server test for public release.

Models for thermodynamic analysis(steady state):Turbine/expander (Isentropic) Turbine/expander (Polytropic), …...

, …...

CAES-TES tool library

Compressors

Compressed airreserviors

Heat storage reservoirs

Heat exchangers

Electrical power systems

Auxiliary components

Turbines andExpanders

Models for thermodynamic analysis(steady state):Compressor (Isentropic) Compressor (Polytropic), …...

Models for dynamic modelling & control:Positive displacement machines Turbomachinery…...

Models for dynamic modelling & control:Positive displacement machines Turbomachinery

…...

Electrical load

…...

Electrical generators/motors AC Synchronous machine

AC Asynchronous machine

DC generator/motor machine

…...Overall CAES/TES

examples

Controllers for CAES/TES systems

2nd level subdirectory examples:

3rd level subdirectory examples:

1st level subdirectory:

1

2

3

4

5

6

7 8 9

Models for dynamic modelling & control:Time-varying dynamic models (considering temperature, pressure variations with time)

Models for thermodynamic analysis(steady state):adiabatic/isothermal/isobaric compressed air storage reserviors

Scroll-type expander

piston type reciprocating expander

Radial-inflow turbine

…...

Models for thermodynamic analysis(steady state):the Number of Transfer Units (NTU) method, …...

Models for dynamic modelling & control:Time-varying dynamic models (considering temperature, pressure variations with time)

10

12

14

16

18

11

13

15

17

19

20

Measurements

Pneumatic actuators

1. Introduction

Introduction mathematical modelling module simulation model (inputs, outputs, parameters) Code explanation corresponding to modelling equations operation example

Version no. and developer(s) Design documentation:

1. Introduction

https://estoolbox.org/index.php/download

1. Introduction

]~

2

1~

2

1

3

1~2

sin~~

cos[

2

0

2

0

2

3222

0

0000

00

2

_

sss

sss

ssss

sssscs

rrkr

kkrk kkr

kyryrkr

kxrxkkrzV

]12~2[, 0

2

_ ikrrziV ssssss

pn

i

sssscstotalses iVVVV1

___ ),(2)(

Fixed scroll

Moving scroll

Fixed scroll

s

)],,(,,,[1

____3___33 essscstotalssVessscssCS

s

s PPPτKPPPSKJ

10

,,0,,

,,,

3

3________

___33

ssSss

Sessscstotalssessscstotals

essscssCS

andFsign

FPPPFandPPPF

PPPSK

0

scroll-type air motor:

2.Mathematical modelling of key components

3-Phase to dq axis

Conversion(stator

voltage)

3-Phase to dq axis

Conversion(stator

voltage)

3-Phase to dq axis

Conversion(rotor

voltage)

3-Phase to dq axis

Conversion(rotor

voltage)

Asynchronous MotorModel

Asynchronous MotorModel

dq axis to 3 phase

Conversion(stator

current)

dq axis to 3 phase

Conversion(stator

current)

dq axis to 3 phase

Conversion(rotor

current)

dq axis to 3 phase

Conversion(rotor

current)

Uas

Ubs

Ucs

Uar

Ubr

Ucr

T_mechanicalT_electromagnetic

Speed

Ias

Ibs

Ics

Iar

Ibr

Icr

Uds

Uqs

Udr

Uqr

Uds

Uqs

Ids

Iqs

Idr

Iqr

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-100

0

100

200

300

400

Time (s)

Ro

tor

sp

ee

d (

rad

/s)

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5-100

-50

0

50

100

Time (s)

Sta

tor

cu

rre

nt (p

ha

se

A)

q axis

d axis

Induction motor:

2.Mathematical modelling of key components

a whole CAES system dynamic model:

Multi-stage compressor modules

Valve controller Adiabatic tank

Regulator controller Scroll type expander

Ideal voltage source

(grid)

Asynchronous motor

PMSG

Safety controller

Electrical load

Case studies are implemented on:

1. multi-stage compression/expansion

2. multi-stage heat exchange

3. connecting with grid simulation system

4. dynamic controlfor starting up process

5. Machine learning for fast simulation (HIL)

https://www.youtube.com/watch?v=Eo9Yy_llFNY

3. Application case study

4. Data driven model - machine learning

Problem: If use numerical models for all components, when more components are connected to the system simulation, the simulation speed is getting slower and slower

Motivation

with TES

4. Data driven model - machine learning

Open-loop NN

(for training use)

Close-loop NN

(for further training and

approximation approach)

Nonlinear autoregressive with exogenous inputs (NARX) neural

network (NN) model

NARX model open-loop

equation:

𝑦(𝑡) = 𝐹[𝑦(𝑡 − 1), 𝑦(𝑡 − 2), . . . ,𝑦(𝑡 − 𝑛𝑦),

         𝑢(𝑡 − 𝑑), 𝑢(𝑡 − 𝑑 − 1), . . . 𝑢(𝑡 − 𝑑 − 𝑛𝑢)

NARX model close-loop

equation:

𝑦(𝑡) = 𝐹[𝑦 (𝑡 − 1), 𝑦 (𝑡 − 2), . . . , 𝑦 (𝑡 − 𝑛𝑦),   

𝑢(𝑡 − 𝑑), 𝑢(𝑡 − 𝑑 − 1), . . . , 𝑢(𝑡 − 𝑑 − 𝑛𝑢)

4. Data driven model - machine learning

NN

Numerical model: accurate but slow NARX NN model: fast but less accurate

NN

:subsystems for a numerical model

mixed

NARX NN-numerical

mixed model: fast model

with accuracy

Air tank model

Air tank model

Thank you! PACSRLab