Proceedings of ASME 2014 Dynamic Systems and Control (DSC) ConferenceDSCC 2014

October 22-24, 2014, San Antonio, USA

DSCC2014-6099

DESIGN, MODELING AND CONTROL OF A LAB-SCALE COMPRESSED AIRENERGY STORAGE (CAES) TESTBED FOR WIND TURBINES WITH SIMULATED

INPUT AND DEMAND POWER VARIABILITY

Farzad A. Shirazi, Pieter Gagnon, Perry Y. Li and James D. Van de VenMechanical Engineering Department

University of MinnesotaMinneapolis, MN 55455

Email: fshirazi@umn.edu

ABSTRACTA Compressed Air Energy Storage (CAES) test-bed has

been developed to experimentally demonstrate the energy stor-age concept proposed in [1] for offshore wind turbines. Thedesign of the testbed has been adapted to the available air com-pression/expansion technology. The main components of thesystem consist of an open accumulator, a hydraulic pump-motor, air compressor/expander, an electrical generator andload, a differential gearbox and a hydraulic control valve.These components are first characterized and then a dynamicmodel of the system has been developed. The objective isto regulate the output current/voltage of the generator whilemaintaining a constant accumulator pressure in the presenceof input and demand power variations in the system. Thisis achieved by Proportional-Integrator (PI) control of pump-motor displacement and field current of the generator. The sta-bility of these controllers has been proved using an energy-based Lyapunov function. Experimental results for storageand regeneration modes have been presented showing excellentperformance of the system in response to power disturbances.

NOMENCLATUREBc/e Compressor/expander damping (N.m/rad/s)Dc/e Compressor/expander displacement (cc/rev)Dpm Pump-motor displacement (cc/rev)Gg Gear ratio between generator and differentialJg Generator inertia (kg.m2)Jpm Pump-motor inertia (kg.m2)Jc Compressor inertia (kg.m2)IF Generator field current (A)

IL Generator current (A)P Accumulator pressure (Pa)P0 Ambient pressure (Pa)qsp Hydraulic supply flow rate (m3/s)R Specific gas constant (J/kgK)Ra Generator armature resistance (Ω)RL Electrical load (Ω)Tc/e Compressor/expander torque (N.m)Tg Generator torque (N.m)Tpm Pump-motor torque (N.m)T0 Ambient temperature (K)Va Air volume (m3)VL Generator voltage (V)ρ Air density at room temperature (kg/m3)ηg Generator efficiencyηpm Pump-motor mechanical efficiencyηvpm Pump-motor volumetric efficiencyϕ(IF) Generator magnetic flux function (Tm2)ωc/e Compressor speed (rad/s)ωg Generator speed (rad/s)ωpm Pump-motor speed (rad/s)

1 IntroductionLarge-scale cost-effective energy storage technologies

have recently attracted significant attention from researchers[2]. This is mainly due to higher penetration of renewablesources of energy in the electric grid. The intermittent andtransient nature of these energy sources can result in instabil-ity of power systems. Therefore, energy storage systems are

1 Copyright c⃝ 2014 by ASME

ideal suppliers of ancillary services to provide regulations inthe electric grid. Compressed air energy storage (CAES) sys-tems have offered competitive features compared to other en-ergy storage approaches. Large power capacity, high cycle lifeand fast response time have made CAES systems practical forenergy storage purpose. One important feature of energy stor-age systems is their round-trip efficiency. The round-trip effi-ciency of conventional CAES systems is usually less than 50%,which is lower than batteries and hydro-pump storage systems.Therefore, increasing the round-trip efficiency of CAES sys-tems will make them even more attractive in the market.

A novel CAES system for offshore wind turbines has re-cently been proposed in [1, 3]. A schematic of the proposedarchitecture is shown in Figure 1. In this design, the mechani-cal power from the wind turbine rotor is converted to hydraulicpower before conversion to electric power. The system enjoysfrom a hydraulic transmission for the turbine with high speedvariability. The excess energy from the wind turbine is storedin the storage vessel prior to electricity generation, while thegenerator power is maintained at the desired value demandedby the electric grid. This allows the electrical components to bedownsized. In particular, downsizing the generator will conse-quently improve the capacity factor of the system defined basedon the generator size. Additionally, the generator speed will beregulated at a constant value that eliminates the need for costlypower electronics [4].

In this CAES system, air is comressed/expanded in a liq-uid piston compressor/expander (F) and high pressure (∼20-30MPa) compressed air is stored in a dual chamber storagevessel called an “open accumulator” (E) with both liquid andcompressed air. Since the air compressor/expander (C/E) is thepathway for the majority of the stored energy, it is critical thatit be efficient and sufficiently powerful. This is challengingbecause compressing/expanding air 200-300 times heats/coolsthe air greatly, resulting in poor efficiency, unless the process issufficiently slow which reduces power [5]. There is therefore atrade-off between efficiency and power. Additionally, the pres-sure is kept constant inside the open accumulator resulting ina more efficient performance of the system components whilethe accumulator acts as a damper in response to high frequencyfluctuations in the input power (wind turbulence) [4].

A variable displacement hydraulic pump (B) is directly (nogearbox) attached to the wind turbine (A) in the nacelle con-verting wind power to hydraulic power. At the ground level,there is a tandem connection of a variable displacement hy-draulic pump/motor (C), a near-isothermal liquid piston aircompressor/expander (F) and a fixed speed induction gener-ator (G) on a single shaft, all driven by the pump (B). Theliquid piston air compressor/expander, the main unit for stor-ing/regenerating energy, consists of a compression/expansionchamber filled with some porous material (F1) and a liquidpiston pump/motor (F2) [6]. The porous material is used inaddition to an optimized compression/expansion trajectory [7]and water spray to enhance the heat transfer inside the chamberto improve its thermal efficiency which has a significant effect

FIGURE 1. Schematic of the CAES system integrated with a windturbine

on the overall efficiency of the storage system.This paper discusses the design, modeling, control and ex-

perimental results of a 1kW CAES test-bed built at the FluidPower Control Laboratory of the University of Minnesota. Theobjective of the project is to develop the first lab-scale proto-type of the CAES system proposed in [1] and investigate theperformance of the system under input and demand power dis-turbances. The system pressure and the generator output cur-rent/voltage will be regulated at constant values in the pres-ence of variable input power simulating the wind power, andvariable electrical loads simulating the electric grid changingpower demand. The rest of the paper is organized as follows.Section 2 describes the lab-scale testbed design and its dif-ferences with the original large-scale configuration. Dynamicmodeling of the testbed is presented in Section 3. The con-troller design and the Lyapunov stability of the closed-loopsystem have been discussed in Sections 4 and 5. Section 6presents the closed-loop experimental results and Section 7concludes the paper.

2 System DescriptionThe testbed consists of the following main components:

a variable displacement hydraulic pump-motor, a differentialgearbox, a wound-field electrical generator, an electrical load,a pneumatic compressor/expander (C/E), an open accumula-tor and a hydraulic control valve. The specifications of thesystem components are listed in Table 1. Although most ofthe components have high power capabilities, the commercialcompressor used has a maximum power output of 1kW of com-pressed air. Therefore, the performance of the rest of the com-ponents was scaled to this magnitude, resulting in a nominal1kW testbed. A 10 gal accumulator has been used with a pre-charge pressure of 300 psi. The input power to the system issupplied from a hydraulic power unit at 2600 psi. The work-ing principle of the system is simple; when the input poweris higher than the demand the system works in energy storagemode and when input power is less than demand the system

2 Copyright c⃝ 2014 by ASME

TABLE 1. Specifications of CAES testbed componentsComponent Type and company Specifications

Air compressor MCH6/EM 2.2kW, 3200 psi

Coltri Americas

Pump-motor AA10VG18EP11/10R 36kW, 4350 psi

Rexroth Bosch 18cc/rev

Generator 0220-4686-01 Cummins 4kW, 3600 rpm

Flow control KVFBA5210 5gpm at

servo-valve Sauer Danfoss ∆P=1000 psi

Accumulator A3-50-E-10G-BN-M-20 10 gal, 5000 psi

Eaton precharged at 300 psi

is in energy regeneration mode. Figure 2 shows a schematicof the CAES test-bed. As explained before, in the original de-sign the variable displacement pump-motors and the generatorare all connected to a single shaft. Here, a fixed displacementcompressor (C) is chosen due to limited options for variabledisplacement air compressors available in the market. There-fore, a differential gearbox was used to regain the degree of thefreedom that was lost by having a fixed displacement compres-sor. In this case, the motor/pump, compressor and the genera-tor (G) are connected to the differential as shown in Figure 2.Therefore, the components do not run at a same speed and thegenerator speed will not be constant. However, the torques areassumed to be at the same value (considering gear ratios) dic-tated by the C/E. The hardware of the CAES testbed is shownin Figure 3.

A hydraulic control valve is used to change the inputpower into the system simulating the variability of wind power.The demand power is determined by resistive loads connectedto the generator. The system inputs are the pump-motor dis-placement and the generator field current. A high pressureair compressor is used for the storage mode and an air motorfor the regeneration mode. The hydraulic pump-motor motordrives the compressor and the generator in the storage mode,and pumps the fluid to the accumulator in the regenerationmode to keep the pressure constant.

In storage mode, the excess power is used to compress airup to 2400 psi, which is then fed into the open accumulator.The system pressure is kept constant by regulating the rate oilflows out of the accumulator. In the regeneration mode, com-pressed air is released from the accumulator to run an air mo-tor (AM). Due to limitations in commercial pneumatics it wasnecessary to reduce the pressure from 2400 psi to a workingpressure of 100 psi to run the air motor. In order to compen-sate for the compressor and air motor losses, a hydraulic gearmotor (GM) and an Electro Proportional Flow Control (EPFC)valve have been added to the system. The compensation forcompressor is linear with the compressor speed while the airmotor compensation is based on the calculated power for airexpansion from 2400 psi to ambient pressure. A clutch bearingis installed such that the compressor will not rotate when the

FIGURE 2. Schematic of the CAES test-bed

FIGURE 3. CAES testbed hardware

system is in the regeneration mode. From here on, the com-bination of the compressor, air motor, gear motor and EPFCvalve will be called an “air compressor/expander” as shown inFigure 2.

It should be noted that a liquid piston C/E [6] is being de-veloped in the group at the University of Minnesota that will re-place the “Air C/E” in the next generation of the system bring-ing the testbed closer to the original design. This will removethe need for a differential and as a result the speed of the gen-erator can be regulated at the grid frequency while providingthe demanded power. A comparison of the CAES test-bed andoriginal system design is summarized in Table 2.

3 Copyright c⃝ 2014 by ASME

TABLE 2. Comparison between CAES test-bed and original system design features

Feature Test-bed Original Design

Power 1kW 5MW

Input and demand Hydraulic supply unit at 2600 psi with a Wind power, Grid

powers control valve, Resistive loads

Transmission Differential gearbox-Multiple shafts-fixed torque Single shaft-fixed speed

Air compressor Fixed displacement Liquid piston compressor/expander

(motor) compressor and motor with variable displacement

Control inputs Pump/motor displacement and Pump/motor and liquid piston

generator field current displacements

Control objectives Constant accumulator pressure Constant accumulator pressure

and generator current/voltage and generator speed

3 Dynamic Equations of TestbedThe speed dynamics in the differential gear can be ob-

tained as follows.!

Jg +Jc4 G2

gJpm

4 GgJpm

4 Gg Jc +Jpm

4

"#ωg

ωc/e

$=

!−Tg +

Gg2 Tpm

Tpm2 −Tc/e

"(1)

where, the pump-motor speed can be obtained from the dif-ferential speed kinematics ωpm =

ωc/e+Ggωg2 when other two

speeds are known. The C/E speed ωc/e is positive in compres-sion and negative in expansion mode. Assuming an isothermalprocess, the C/E torque can be written as a function of its up-stream pressure and speed as follows.

Tc/e =P0Dc/e

2π

#ln%

PP0

&− P−P0

P

$+Bc/eωc/e (2)

where, the upstream pressure is regulated at a constant valueof 2400 psi. The speed-dependency of the C/E torque is due tofriction losses. The generator under study is an AC wound-fieldgenerator where its output voltage is a function of the genera-tor speed and field current. The function ϕ(IF) is nonlinearlydependant on IF and denotes the magnetic flux in the genera-tor windings. The generator voltage, current and torque can bedetermined from the following equations.

VL =RLϕ(IF)ωg

R, IL =

VL

RL=

ϕ(IF)ωg

R, Tg =

ϕ(IF)IL

ηg(3)

where, R = RL +Ra. The flow rate and torque of the pump-motor are obtained as follows.

qpm =Dpmωpm

2πη pvpm

, Tpm =−DpmPη p

pm

2π , p =

'1 if Motor−1 if Pump

(4)

The pressure dynamics of the air inside the accumulator canbe derived from the ideal gas law and an isothermal processassumption. The air volume dynamics can be determined fromthe mass continuity by subtracting flows going in and out ofthe accumulator.

VaP =ρRT0Dc/eωc

2π +P%−

Dpmωpm

2πη pvpm

+qsp

&(5)

Va =Dpmωpm

2πη pvpm

−qsp (6)

where, qsp = kv(

Psp −P and kv indicates the hydraulic valveopening and Psp is the hydraulic supply pressure at 2600 psi.

4 PI Controller DesignIn this section, first a transformation will be used to decou-

ple the speed dynamics of the generator and C/E in Eqn. (1).

Let’s assume#

ωgωc/e

$= S

#ωgω⊥

g

$, where S =

#1 0a 1

$is a coor-

dinate transformation and a is a constant to be determined. ω⊥g

is a decoupled state perpendicular to ωg. By substituting thetransformed coordinates in Eqn. (1) and defining the inertia

matrix as J =

!Jg +

Jc4 G2

gJpm

4 GgJpm

4 Gg Jc +Jpm

4

"=

#J11 J12J12 J22

$, the trans-

formed inertia matrix will be obtained as,

J = ST JS =

#J11 +2aJ12 +a2J22 J12 +aJ22

J12 +aJ22 J22

$=

#J1 00 J2

$(7)

By making the non-diagonal terms equal to zero the two dif-ferential equations will be decoupled. Therefore, we havea = −J12

J22=

JpmGgJpm+4Jc

and the decoupled dynamics of speeds can

4 Copyright c⃝ 2014 by ASME

be written as follows.

J1ωg =a+Gg

2Tpm −Tg −aTc/e (8)

J2ω⊥g =

Tpm

2−Tc/e (9)

By taking the derivative of the generator voltage in Eqn. (3)and substituting ω∗

g =RV ∗

LRLϕ(I∗F )

, the following can be derived.

J1˙VL =

J1

R

!Rϕ∗ ˙IFV ∗

Lϕ∗ +

RLϕ∗

J1J1 ˙ωg −

RLRRL

ϕ∗V ∗L

"(10)

where ϕ∗ = ϕ(I∗F) and ϕ∗ = ϕ(I∗F). From now on, the * and˜ terms denote the equilibrium point and error values, respec-tively. By substituting the generator speed dynamics from Eqn.(8) in Eqn. (10), considering ϕ = ϕ∗+ ϕ∗IF , VL =V ∗

L +VL andassuming −ϕ∗V ∗

LRLηg

+(a+Gg)

4πη ppm

D1P− aTc/e = 0 in the equilibriumpoint, the following can be obtained for the generator voltagedynamics.

J1˙VL =

J1ϕ∗V ∗L

ϕ∗˙IF − (ϕ∗)2

RLRηgVL −

ϕ∗ϕ∗V ∗L

RLRηgIF

− ϕ∗ϕ∗

RLRηgIFVL +

(a+Gg)

4πη ppm

D2P+J1V ∗

L

RRL (11)

Here, we have considered a PI control law for the pump-motordisplacement as Dpm = D1 +D2 = KI

)Pdt +KPP. The inte-

gral term D1 will satisfy the equilibrium point condition whileD2 is a term proportional to pressure error. More discussionon displacement control will be presented in the next section.Now, let’s neglect second order terms in the above equationand rewrite it in a linear matrix from. We will consider higherorder terms later in the Lyapunov stability analysis.

!˙IF

J1˙VL

"=

!0 0

−ϕ∗ϕ∗V ∗L

RLRηg

−(ϕ∗)2

RLRηg

"#IFVL

$

+

!1

J1ϕ∗V ∗L

ϕ∗

"˙IF +

!0

J1V ∗L

R

"RL (12)

The idea is to select u = ˙IF such that both error terms will bevanishing exponentially. By considering a PI control law as˙IF = λp

˙VL+λIVL, the closed-loop Acl matrix of the system willbe obtained as follows.

!˙IF˙VL

"=

! aλpJ1−cλp

λI +λp(b+cλI)

J1−cλpa

J1−cλpb+cλI

J1−cλp

"#IFVL

$(13)

where, a= −ϕ∗ϕ∗V ∗L

RLRηg, b= −(ϕ∗)2

RLRηgand c= J1ϕ∗V ∗

Lϕ∗ . It can be easily

shown that the eigenvalues of the closed-loop Acl matrix willbe negative if λp <

ϕ∗RLϕ∗V ∗

Land λI <

(ϕ∗)2

J1RLRηgλp +

(ϕ∗)3

J1Rηgϕ∗V ∗L

. Bythis choice of gains the local stability of the linearized systemaround the equilibrium points is guaranteed.

5 Lyapunov Stability Analysis for PI Controllers5.1 Lyapunov Function for Field Current Control

Considering the closed-loop matrix Acl , the correspond-ing positive definite Lyapunov matrix P can be obtained fromPAcl +AT

clP = −αI for α > 0 that satisfies the exponentialstability of the generator error system in Eqn. (13). Let’s as-

sume that P =

#p11 p12p12 p22

$, therefore the Lyapunov function

and its derivative will be ΓVL =12 p11 I2

F + p12VLIF +12 p22V 2

L ≥ 0and ΓVL =−α

2 I2F − α

2 V 2L ≤ 0, respectively. The equality to zero

holds when ΓVL = ΓVL = 0.If the PI controller gains λp and λI satisfy the aforemen-

tioned upper bounds, then the positive definiteness of the ma-trix P will be guaranteed which means p22 > 0, p11 > 0 andp11 p22 − p2

12 > 0.

5.2 Lyapunov Stability Anaylsis for Nonlinear Sys-tem and PI Controllers

The stability of the generator voltage linear error systemwas proved in the previous section. In this section, the higherorder terms in voltage dynamics as well as the coupling be-tween voltage and pressure dynamics will be considered andanalyzed. Let’s define the Lyapunov function for the system asthe summation of two energy-related Lyapunov functions forthe generator voltage and accumulator pressure.

Γ = ΓVL +ΓP =12

p11 I2F + p12VLIF +

12

p22V 2L

+ P0Va

#PP∗ ln

%PP∗

&− P

P∗

$+

β2

D22 (14)

After some algebra, the derivative of this function will be ob-tained as follows.

Γ = −I2F

*α2− c0 p12VL

+−V 2

L

*α2− c0 p22 IF

+(15)

− c1

!−p12 IF − p22

RLVL +

ωpP0

2πP∗η pvpm

,1

P∗ +βK2

pP∗

Va

-"KpP2

where, c0 =ϕ∗ϕ∗

J1RRLηg> 0 and c1 =

(a+Gg)

4πη ppm

> 0. By appropriateselection of α , β and Kp the higher-order and coupling termscan be dominated and the exponential stability of the systemis guaranteed. For the generator current control all the abovediscussion holds as well only by substituting IL = VL

RL.

5 Copyright c⃝ 2014 by ASME

6 Closed-loop Control Implementation and Experi-mental ResultsFor the system data acquisition and control, an Advantech

PCI 1716-AE DAQ board has been used. Real-Time WindowsTarget (RTWT) has been employed to communicate with theboard and interface with Simulink/MATLAB. The measure-ments are processed in the Simulink model and the controllersare implemented based on the filtered feedbacks from the ac-cumulator pressure and generator rms current/voltage. Thegenerator AC current is measured by a CSLT Honeywell hall-effect current sensor. The generator voltage is first converted toa low voltage signal using a transformer and a voltage dividerand then measured by the board. The generator field currentand pump-motor displacement are commanded from computerusing current driver circuits. The pump-motor and accumulatorflow rates and directions are obtained from AW-Lake gear flowmeters and sensors. The accumulator pressure is measured by aMLH03KPSB01A Honeywell pressure transducer. The pump-motor speed is measured by a HB6M-1024-750 US-Digital op-tical encoder and the generator speed is obtained from the ACvoltage signal.

The control objective is to regulate the accumulator pres-sure at 2400 psi meanwhile providing a constant generator out-put in the presence of variations in input and/or demand powerin the system. In order to cover a wider range of demand powerlow and high power regions are considered for the generatorcontrol. In the low power region, the current is regulated at aconstant value and power increases by increasing the resistiveload. In the high power region the voltage is kept at a constantvalue and power increases as the resistive load drops. Figure 4shows how voltage, current and power change vs. each otherin low and high power regions of operation. A single constantvoltage region cannot be obtained in the system since in thelow power region the output current must be low. This requiresa relatively low torque on the generator which is not possibleto attain because of the minimum torque dictated on the systemby the compressor.

Figure 5 illustrates the closed-loop response of the systemwhen the input power to the system is changing from 2500W to3200W, and the generator current is regulated at 12A and theaccumulator pressure at 2400 psi. The electrical load is con-stant at 7Ω corresponding to the corner point power of 1008Win Figure 4. The generator output power is determined fromPg =VLIL =

V 2L

RL. The pump-motor displacement, the generator

field current and the input power are also depicted. Figure 6shows the closed-loop performance of the system in the pres-ence of variable electrical loads from 660W to 1600W. Theinput power is fixed at 3200W. The input power is calculatedfrom Pin = qsp∆Ppm, where ∆Ppm is the pressure drop acrossthe hydraulic pump-motor. The generator current is firstly reg-ulated at 12A corresponding to the low power region. In thehigh power region, the voltage is regulated at 84V and thesteady-state currents will match with the values shown in Fig-ure 4. The closed-loop system response in the regenerationmode in the presence of input and demand power variations is

11 12 13 14 15 16 17 18 19 2030

40

50

60

70

80

90

IL (A)

V L(V)

400 600 800 1000 1200 1400 160030405060708090

Power (W)

V L(V)

400 600 800 1000 1200 1400 160012

14

16

18

20

Power (W)

I L(A)

High Power Region

Low Power Region

RL=4.3ΩRL=5.9Ω

FIGURE 4. Generator low and high power operating regions

100 150 200 250 300 3502380240024202440

t(s)

Air p

ress

ure

(psi

)

Actual PressureDesired Pressure

100 150 200 250 300 3508

1012141618

t(s)

Cur

rent

RM

S (A

)

Actual CurrentDesired Current

100 150 200 250 300 3506

7

8

9

t(s)

Dis

plac

emen

t (cc

/rev)

100 150 200 250 300 3501.5

2

2.5

3

t(s)

Fiel

d cu

rrent

(A)

100 150 200 250 300 350240026002800300032003400

t(s)

Inpu

t pow

er (W

)

FIGURE 5. Storage mode closed-loop response in the presence ofinput power variability

shown in Figure 7. In this experiment, the regeneration poweris approximately constant at 300W. The difference between thesum of the input and regeneration powers and demand poweris wasted through the inefficiencies in the system. In summary,the closed-loop performance objectives are met in both storageand regeneration modes in the presence of input and demandpower disturbances in the system. However, it should be noted

6 Copyright c⃝ 2014 by ASME

100 150 200 250 300 350

2400

2450

t(s)

Air p

ress

ure

(psi

)

Actual PressureDesired Pressure

100 150 200 250 300 3506

8

10

t(s)

Dis

plac

emen

t (c

c/re

v)

100 150 200 250 300 3501.5

2

2.5

3

t(s)

Fiel

d cu

rrent

(A)

100 150 200 250 300 350500800

110014001700

t(s)

Gen

erat

or p

ower

(W)

100 150 200 250 300 35010

15

20

t(s)

Cur

rent

RM

S (A

)

Actual CurrentDesired Current

FIGURE 6. Storage mode closed-loop response in the presence ofdemand power variability

that in both storage and regeneration modes there exist infeasi-ble operating regions due to mechanical limitations in the sys-tem. For example, in high input powers and low power regionof the generator the compressor speed can exceed the allow-able speed of 2250rpm or the inefficiency of the generator inlow field currents impedes high electrical power outputs in lowinput powers.

7 ConclusionsThis paper presented a CAES testbed designed and built

based on the concept proposed in [1]. The system designwas adapted to the available compression/expansion technol-ogy. The system architecture and dynamic model were dis-cussed. PI controllers were considered to command the pump-motor displacement and the generator field current to regu-late the open accumulator pressure and the generator outputcurrent/voltage in the presence of input and demand powervariability. The Lyapunov stability of the closed-loop systemwith PI controllers around equilibrium points was analyzed andproved. The closed-loop experimental results for different fea-sible operating regions of the system were presented showingexcellent performance of the prototype in response to powerdisturbances in both storage and regeneration modes. The nextgeneration of the system is being developed in the group usinga liquid piston compressor/expander and a variable displace-ment water pump to solve the practical issues discussed in thepaper.

50 100 150 200 250 300 350

2400

2450

t(s)

Air p

ress

ure

(psi

)

50 100 150 200 250 300 35010

15

20

t(s)

Cur

rent

RM

S (A

)

50 100 150 200 250 300 35012

14

16

18

t(s)

Dis

plac

emen

t (c

c/re

v)

50 100 150 200 250 300 3501.5

2

2.5

3

t(s)

Fiel

d cu

rrent

(A)

50 100 150 200 250 300 3501000

2000

3000

t(s)

Inpu

t pow

er (W

)

50 100 150 200 250 300 350500

1000

t(s)Gen

erat

or p

ower

(W)

Actual PressureDesired Pressure

Actual CurrentDesired Current

FIGURE 7. Regeneration mode closed-loop response in the pres-ence of input and demand power variability

REFERENCES[1] P. Y. Li, E. Loth, T. W. Simon, J. D. Van de Ven and S.

E. Crane, “Compressed Air Energy Storage for OffshoreWind Turbines,” in Proc. International Fluid Power Exhi-bition (IFPE), Las Vegas, USA, 2011.

[2] J. Twidell and G. Gaudiosi, Offshore Wind Power, Multi-Science Publishing Co. Ltd, Essex, UK, 2009.

[3] M. Saadat and P. Y. Li, “Modeling and Control of a NovelCompressed Air Energy Storage System for Offshore WindTurbine,” in Proc. American Control Conference, pp. 3032-3037, Montreal, Canada, 2012.

[4] M. Saadat, F. A. Shirazi and P. Y. Li, “Nonlinear Con-troller Design with Bandwidth Consideration for a NovelCompressed Air Energy Storage System,” in Proc. ASMEDynamic Systems and Control Conference, Stanford, USA,2013.

[5] M. Nakhamkin, E. Swensen, R. Schainker and R. Pol-lak, “Compressed Air Energy Storage: Survey of AdvancedCAES Development,” in Proc. ASME International PowerGeneration Conference, pp. 1-8, San Diego, USA, 1991.

[6] J. D. Van de Ven and P. Y. Li, “Liquid Piston Gas Com-pression,” Applied Energy, vol. 86, Issue 10, pp. 2183-2191,2009.

[7] F. A. Shirazi, M. Saadat, B. Yan, P. Y. Li and T. W. Si-mon,“Iterative Optimal Control of a Near Isothermal LiquidPiston Air Compressor in a Compressed Air Energy StorageSystem,” in Proc. American Control Conference, pp 2940-2945, Washington, DC, 2013.

7 Copyright c⃝ 2014 by ASME

View publication statsView publication stats