Coupled hydromechanical analysis of an underground compressed air energy storage facility in sandstone

Coupled hydromechanical analysis of an underground compressedair energy storage facility in sandstone

M. SANCHEZ*, A. SHASTRI* and T. M. H. LE{

The underground storage of compressed air is a favourable and low-cost option for balancing off-peak electricity demands. Large volumes of air can be compressed and stored in the ground duringlow-demand electricity periods, and this pressurised air can then be released to generate electricityduring high-demand times. Excavated caverns (generally in salt rock), depleted gas fields andabandoned/excavated mines are the main options typically considered in the design of acompressed air energy storage (CAES) facility. The storage of compressed air in aquifers is a newalternative that has been studied recently. The natural formations around a CAES facility aresubjected to significant perturbations in hydraulic and mechanical fields. This paper presents ananalysis of a CAES project in an aquifer using a fully coupled hydromechanical framework with theability to consider the main physical phenomena that control the behaviour of this kind of system. Thecase study is based on an actual planned CAES project in Iowa, USA. Particular attention was paid tochanges in the hydraulic and mechanical fields induced by the operational phases anticipated in aCAES project and to the impact of porosity heterogeneities on system performance. The analysisconfirmed that the site is not suitable for a CAES plant.

KEYWORDS: finite-element modelling; partial saturation; pore pressures; repeated loading

ICE Publishing: all rights reserved

NOTATION

b body forces vectorD elastic tensor (two-dimensional axisymmetric case)D9a mechanical dispersion tensor phase a

Dia dispersion tensor of species i in phase a

Dim dispersion coefficient corresponding to species i

E Young’s modulusf i sink/source term of species iH Henry’s constantI identity matrixi superscript identifying species air (a), water (w) or

solid (s)ic non-advective heat flux

i ia non-advective mass flux (i 5 w, a; a5l, g)

j ia total mass flux respect to a fixed system

k permeability tensork0 reference permeability (for w0)krl relative permeabilityMa air molecular massMW water molecular massn relative permeability parameterP0 retention curve parameterqa volumetric flux of phase respect to the solidPa phase pressurePc capillary pressure suction (pg2pl)Pv vapour partial pressureR gas constantSa volumetric fraction of pore volume occupied by a

phaseT0 reference temperature

a subscript identifying phases liquid (l) or gas (g)e strain vector {er,ey,eh,cxy}T

hai fraction mass (species/phase) per unit volume phase

l0 retention curve parameterma dynamic viscosity of phase an Poisson’s coefficientra mass of phase a per unit of volume of phase as total stress vector {sr,sy,sh,dxy}T

s9 effective stress vector (st2Ipl)t tortuosityw porosityw0 reference porosityva

i mass fraction of species i in phase a

INTRODUCTIONCompressed air energy storage (CAES) technology allowsfor the economical storage of energy during periods of lowdemand to meet energy requirements at periods of higherdemand. In the CAES concept, electricity is used to compressair and store it in a geological formation. Energy is recoveredafterwards by decompressing the air (from the undergroundstorage), burning it (generally with the aid of a small amountof natural gas) and expanding the products of the combus-tion using a turbine (Succar & Williams, 2008). Thistechnology has a number of advantages – among others, it is

N one of the few systems suitable for a long-term storageof electrical energy

N a low-cost technology in which pre-existing formations(e.g. salt caverns or abandoned mines) can be reused;furthermore, gas turbines are relatively cheap

N environmentally friendly with low greenhouse gasproduction (due to low-fuel consumption gas turbines)

N an ideal partner for balancing off-peak wind energyproviding a reliable base load power.This last aspect is particularly relevant considering that

the energy generated from wind farms is dependent on

Manuscript received 8 September 2013; first decision 27September 2013; accepted 5 May 2014.Published online at www.geotechniqueletters.com on 30 June2014.

*Zachry Department of Civil Engineering, Texas A&M University,College Station, TX, USA{SINTEF Building and Infrastructure, Trondheim, Norway

Sanchez M. et al. (2014) Geotechnique Letters 4, 157–164, http://dx.doi.org/10.1680/geolett.13.00068

157

Offprint provided courtesy of www.icevirtuallibrary.com Author copy for personal use, not for distribution

weather conditions and is therefore quite difficult toanticipate.

Historically, the CAES design has been based on excavatedcaverns or abandoned mines in salt rock or hard rockformations (e.g. Kim et al., 2012). Two commercial-scaleCAES projects are operating (Succar & Williams, 2008) –caverns in a salt dome in Huntorf, Germany (effective since1978) and a plant also built in salt rock in Alabama, USA(operating since 1991). CAES design in porous/permeablerocks is a relatively new alternative in which the compressed airis not stored in a tank-like mining solution, but is directlyinjected into the aquifer. Figure 1 shows a schematicrepresentation of this system with its main components –wind farm, gas turbine (pumps and other devices), injectionwell(s), aquifer, air bubble, cap rock and the electrical network.

CAES in aquifers is known to have the following severaladvantages.

N It is economical when compared with other solutions(e.g. the cost of production of a CAES plant in salt rockis <2$/kWh, in hard rock <30 $/kWh and porous rock<0?11 $/kWh (Succar & Williams, 2008)).

N The cost of expanding storage capacity is relatively lowbecause it simply requires extra wells.

N The widespread availability of porous rock; this isparticularly relevant in countries such as the USA whereregions with wind resources overlap with areas favour-able for the injection of compressed air into porousrocks (Succar & Williams, 2008).However, the design of this kind of facility is complex. A

number of factors have to be considered in the design. Forexample, the maximum air pressure (Pa) must notcompromise the geo-mechanical stability of the naturalrock, the entrapment condition of the aquifer is key forsystem feasibility (the presence of cracks/discontinuitiescould hamper the storage capacity) and rock heterogene-ities may also impact the air migration.

This paper reports on a technical feasibility study of anactual CAES project in porous rock using a truly coupledmulti-phase framework.

CASE STUDYThe case study is located in Dallas, Iowa, a site pre-selectedby the Iowa Stored Energy Plant Agency for developing aCAES unit. An initial feasibility study based on data from

the Redfield natural gas storage plant (located in theproximity of the Dallas Center) showed potential to couplethis plant with wind farms (Moridis et al., 2007). The projectwas abandoned when a subsequent site investigation revealedthat the site was inappropriate for a CAES plant (Heathet al., 2013). Some of the issues associated with this site are asfollows: small reservoir volume, heterogeneity of thesandstone formation and relatively low permeability. Basedon information gathered from field and laboratory investiga-tions (Heath et al., 2013), a coupled analysis of this case wascarried out, looking at perturbation in the hydromechanicalfields induced by operation of the CAES. The impact of rockheterogeneities on the volume of air injected, reservoirpressure and water production was also studied.

Figure 2 shows the distribution of materials at differentdepths, including the Eau Claire formation, Mount Simonsandstone and the Red-clastics formation, together withestimated values of permeability and porosity (Heath et al.,2013). The first formation (the cap rock) is a dolomite andthe second is the aquifer with four subdivisions (A–D),followed by very low permeability bedrock below theaquifer. A strong heterogeneity characterises soil propertiesfrom subdivisions C and D.

Figure 3 presents the experimental data used to determinethe parameters of the adopted hydraulic constitutive laws,including the intrinsic permeability model, the waterretention curve and the relative permeability law. Table 1lists the adopted constitutive equations. It was assumed thatthe sandstone behaves elastically. The mechanical para-meters were obtained from Dewers et al. (2014). Additionalinformation about the selected site can be found elsewhere(e.g. THG, 2011; Heath et al., 2013; Dewers et al., 2014).

NUMERICAL APPROACHThe fully coupled program CODE_BRIGHT (Olivellaet al., 1996) was adopted in this study to perform thenumerical analyses. The framework was formulated using amulti-phase/multi-species approach. The liquid (l) phasemay contain water (w) and dissolved air (a), and the gas (g)

Wind farm CAES power plant

Electricity to grid

Caprock

Aquifer

Bedrock

Airbubble

Fig. 1. Schematic representation of the main components of aCAES facility in an aquifer, including the air bubble associatedwith the concept

865

870

875

880

885

890

Dep

th: m

EauClaire

formation

MtSimon

A

B

C

D

Redclastics

895

900

905

910

915

92010–20 10–1410-16

Permeability: m2 Porosity10–18 10–120.05 0.200.150.10 0.25

Fig. 2. Distribution of the materials, porosity and permeabilityat different depths at the Dallas site in Iowa, USA, including thecap rock (Eau Claire dolomite), the aquifer (Mount Simonsandstone) and bedrock (Red-clastics formation) (adapted fromHeath et al. (2013))

158 Sanchez, Shastri and Le


phase is a mixture of dry air and vapour. The approach iscomposed of balance equations, constitutive equations andequilibrium restrictions. The equation associated with thewater mass balance is

LLt

(hwl Slwzhw

g Sgw)z+(jwl zjw

g )~f w (1)

The equation associated with the air mass balance isderived in a similar way as

LLt

(hal Slwzha

gSgw)z+(jal zjag)~f a (2)

The balance of momentum for the porous medium reducesto the equilibrium equation

+szb~0 (3)

The formulation also includes the transport of reactivespecies (e.g. Guimaraes et al., 2006).

NUMERICAL ANALYSES AND DISCUSSIONA two-dimensional axisymmetric section was adopted tomodel the 30 m thick aquifer with four layers A–D (Fig. 2).Figure 4(a) shows the finite-element mesh (1048 elements)used in the analysis. A denser mesh was adopted within200 m from the well to properly capture the high hydraulicgradients anticipated in this region. Figure 4(b) presentsthe adopted geostatic initial total stresses.

A critical operational aspect of a CAES facility in porousrock is the development of the initial air bubble, generatedby the air injection that simultaneously displaces thenatural water. It is recommended that a large air bubblebe developed in an aquifer to operate the CAES withinjection/production cycles without significant changes inreservoir pressure. To develop the initial air bubble, air wasinjected at a constant pressure of 12 MPa for 1 year.Figure 4(c) illustrates the applied air pressure at theborehole position (i.e. throughout the whole aquiferthickness), together with the initial liquid pressure (Pl)field. Based on field data (Heath et al., 2013), a hydrostatic

10–12

10–13

10–14

10–15

10–16

10–17

10–18

0 0.200.15Porosity

(a) (b)

TestEquation (4)P

erm

eabi

lity:

m2

0.100.05 0.25

0.6

0.4

Cap

illar

y pr

essu

re: M

Pa

0.2

0.8

0

0 0.80.60.4Degree of saturation: %

0.2 1.0

TestEquation (8)

0

(c)

0.8

0.6

Air

rela

tive

perm

eabl

ity

0.4

0.2

1.0

0 0.80.60.4Degree of air saturation

0.2 1.0

TestEquation (6)

0

(d)

0.8

0.6

Wat

er re

lativ

e pe

rmea

blity

0.4

0.2

1.0

0 0.80.60.4Degree of air saturation

0.2 1.0

TestEquation (5)

Fig. 3. Main constitutive laws for the hydraulic problem. (a) Intrinsic permeability (only tests identified as horizontal by Heath et al.(2013) were considered). (b) Water retention behaviour (tests from Heath et al. (2013)). (c) Relative permeability law for gas.(c) Relative permeability law for water (tests from Heath et al. (2013)). The symbols correspond to experimental data and the curvesto model fitting. The equations referred to in each plot are given in Table 1

Coupled hydromechanical analysis of an underground compressed air energy storage facility in sandstone 159


initial Pl distribution was adopted. Impermeable top andbottom boundaries were assumed. After the period ofconstant air pressure (i.e. air bubble formation), thecompression/decompression cycles were simulated for 3years by imposing an injection/withdrawal air flow rate (asexplained later). Figure 4(d) shows the injected/producedair flow rate alongside the distribution of air saturation atthe end of the air bubble formation.

A typical daily variation of energy demand is presentedin Fig. 5(a) and Fig. 5(b) shows the adopted pressure cyclethat attempts to use off-peak energy to supply energy athigh-demand times. The daily variation of air compression/decompression was simulated by injecting air at a constantrate for 12 h and then withdrawing air at a constant ratefor another 12 h. An air injection rate of 261023 kg/s wasapplied to achieve air pressures of about 10 MPa duringmaximum compression. The same air flow rate was appliedduring production. Minimum air pressures of around8?2 MPa were computed during decompression. Theadopted boundary conditions led to rather balancedpressure cycles around the hydrostatic pressure.

The results of the analysis considering a constantporosity distribution are presented first (i.e. w50?13).Figure 6 shows the evolution of air pressure Pa for bothstages (i.e. air bubble formation and injection/withdrawal

cycles). A detailed cyclic variation of Pa between 720 and740 d is presented in the inset of this figure. The computedPa decreases just after the bubble formation period andthen oscillates around the hydrostatic Pl.

Figure 7 shows the contour plots of air pressure Pa

(Fig. 7(a), air saturation (Fig. 7(b)), horizontal total stress(Fig. 7(c)), vertical total stress (Fig. 7 (d)) and porosity(Fig. 7(e)). The results are presented at three differentinjection periods – the end of bubble formation, 18 monthsand 36 months. Contour plots at the end of injection andwithdrawal stages are shown for each of these periods. Atthe end of the constant-Pa period, the bubble evolves toonly about 30 m from the injection well (Fig. 7(a)). This isquite a small volume of air for this kind of project. This canbe attributed to the unfavourable low permeability of thisaquifer for a CAES facility. The degree of saturation (andthe associated air bubble size) does not change significantlyduring the compression/decompression cycles (Fig. 7(b)).The development of a non-uniform bubble is related tobuoyancy effects included in Darcy’s equation (Table 1).Note that, to facilitate the air inflow/outflow in the aquifer,irreducible water saturation (i.e. Srl<0?073 for this rock(Heath et al., 2013)) should prevail in the air bubble to leadto maximum air permeability. However, in this analysis Srl

was only observed in a quite tiny volume around the well.

Table 1. Summary of constitutive laws and equilibrium restrictions

Variable Equation Parameter relationships Parameters

Constitutive equations (a5l or g; i 5 w or a)

Darcy’s law Liquid and gasadvective flux qa~{k

kra

ma

(+Pa{rag) k~k0w3

(1{w)2

(1{w0)2

w30

Ik056?5 610214 m2 a

w050?13

(4)

Relativepermeability

Liquid relativepermeability

krl~(Se)1=2½1{(1{(Se)1=l)�2Srl&0:073a

l50?65a (5)

Relativepermeability

Gas relative per-meability

krg~ASleg

l52?3A50?8a

(6)

Fick’s law Vapour and airnon-advective flux

iia~{Di

a+via~

Dia~ wraSatDi

mIzraD’a� � Dw

m~d(273:15zT)2:3

Pg

d~5:9|10{12

t51 (7)

Retention curve Phase degree ofsaturation Se~ 1z

Pc

P0

� �1=(1{l0)" #{l0

Se~Sl{Slr

Sls{Slr

Sl~1{Sg

l050?34P050?0047MPaa

(8)

Mechanicalconstitutivemodel

Linear elasticstress

C~(1{2n)=2

A~E=(1zn)(1{2n):s’~D

::eD~

1{n n n 0

n 1{n n 0

n n 1{n 0

0 0 0 C

2664

3775

E510 GPan50?3b

(9)

Phase density Gas density rg~hagzhw

g Olivellaet al. (1996)

(10)

Liquid density rl~1002:6exp½4:5|10{4(Pl{0:1){3:4|10{4T � T545uC (11)

Phase viscosity Liquid viscosityml~2:1|10{12 exp

1808:5

273:15zT

� �(12)

Gas viscositymg~1:48|10{12 exp

(273:15zT)1=2

1z½119=(273:15zT)�

!(13)

Equilibrium restrictions

Henry’s law Air dissolved massfraction

hal ~vl

arl~Pa

H

Ma

Mwrl

(14)

Psychometriclaw

Water vapourdissolved massfraction

hwg ~(hw

g )0 exp{(pg{pl)Mw

R(273:15zT)rl

� �(hw

g )0~MwPv(T)

R(273:15zT)

(15)

aParameters estimated from data published by Heath et al. (2013)bParameters estimated from data published by Dewers et al. (2014)



The operation of the CAES also changes the stress andporosity fields (Figs 7(c)–7(e)). The perturbations are morenoticeable near the injection zone. Nevertheless, they areinsignificant due to the high stiffness of the host rock.

One of the issues to thwart this project was theheterogeneous permeability of the reservoir, in particularin layers C and D (Heath et al., 2013). The available fieldinformation is limited, but it is still useful to explore theimpact of random distributions of porosity on CAESperformance. Two plausible scenarios corresponding to

Excess energy

Midday

Midday

Midnight

Midnight

Peak energy

Pow

er d

eman

d

supplyP

eak load

10.0

9.8

9.6

9.4

9.2

9.0

Gas

pre

ssur

e: M

Pa

8.8

8.6

8.4

8.2

Intermediate

Base loadsupply

load supply

Time of day(a)

Time of day(b)

Fig. 5. (a) Typical daily variation of energy demand (modifiedafter Kushnir et al. (2012)). (b) Adopted daily pressure cycle inthe numerical modelling

16

12

8

Initial airbubbleformation

Compression/decompressioncycles

Air

pres

sure

: MP

a

0

0 1500

Time: d

135012001050900

720

10

9

8

11

7730 740

750300 600450150

4

Fig. 6. Evolution of air pressure for the uniform distribution ofporosity case. The plot presents both the constant air pressureimposed during the first year of operation (associated with theair bubble formation) and the computed air pressure during thesubsequent 3 years of plant operation with compression/decompression cycles. The inset shows the variation of airpressure between 720 and 740 d

30 m200 m

(a)

(c)

(d)

(b)

1000 m

Vertical stress 24 MPaSyy-stress: MPa

24.6724.6024.5224.4524.3724.3024.2224.1524.0724.00

Liquid pressure: MPa

Air saturation1.000.890.780.670.560.440.330.220.110.00

9.309.279.239.209.179.139.109.079.039.00

Initial vertical stress

Initial liquid pressure

1000 m

1000 m

1000 m

Imposed air pressure during air bubble formation 12 MPa

Imposed cyclic air flowrate 2.0×10–3 kg/s

Air saturation at end of air bubble formation

Fig. 4. (a) Geometry and adopted finite-element mesh. (b) Initial stresses and mechanical boundary conditions. (c) Initial liquidpressure and imposed air pressure during the air bubble formation stage. (d) Air saturation of the end of air bubble formation andimposed cyclic air flow rate during operation of the CAES plant



short and long correlation lengths were explored. Togenerate random porosity fields, the procedure proposedby Le et al. (2011, 2013) was adopted. This methodcombines local average subdivisions (Fenton & Griffiths,2008) with a Markovian correlation function. Porosity isassumed to follow a log-normal distribution with constantmean equal to 0?13, standard deviation of 0?075 andcorrelation length of 3–100 m. The influence of porosity onpermeability is accounted for by equation (4) in Table 1.

Figure 8 shows the evolution of air pressure at twodifferent porosity distributions for 4 years of the analysisand Fig. 9 presents the numerical results in terms of airpressure and air saturation for the two porosity fields. Forthe long correlation length, the presence of low-porosityzones dominates the performance of the system, delayingthe air flux (e.g. compare the air bubble formation inFigs 7(b) and 9(d). As for the short correlation length, themodel predicts the presence of undesirable fingering effects,leading to air flow pathways that do not fill the entirestructure (Fig. 9(b)). The impact of heterogeneities is alsoevident when looking at the cumulative injected air duringthe first year of injection (Fig. 10(a)). The air injected in theshort correlation length case is about three times higherthan that computed in the uniform porosity case, indicating

Initial air bubble (at endof constant-pressure period)

75 m

30 m

Cycle 18 months 36 months

Air pressure: MPa

Air saturation

SXX-stress: MPa

Syy-stress: MPa

Porosity

12.00

1.00

17.0016.6716.3316.0015.6715.3315.0014.6714.3314.00

25.6725.3325.0024.6724.3324.0023.6723.33

26.00

23.00

0.13050.13030.13020.13000.12980.12970.12950.12930.12920.1290

0.890.780.670.560.440.330.220.110.00

10.679.338.006.675.334.002.671.330.00

C

C

C

C

C

D

D

D

D

D

75 m

30 m

75 m

30 m

75 m

30 m

75 m

30 m

75 m

1000 m

30 m

Domain used in contour plots

(a)

(b)

(c)

(d)

(e)

(f)

Fig. 7. Contour plots of (a) air pressure, (b) air saturation, (c) horizontal stress, (d) vertical stress and (e) porosity for the case withuniform porosity distribution. (f) Schematic representation of the region analysed including the initial domain (75 m630 m) used inthe contour plots presented in (a)–(e). C and D refer to compression and decompression respectively

16

12

Air

pres

sure

: MP

a

8

4

0

0 135012001050900750

Time: d

600

Initial airbubbleformation

Compression/decompressioncycles

Porosity distribution:short correlation lengthPorosity distribution:long correlation length

450300150 1500

Fig. 8. Evolution of air pressure for the random distributionporosity cases (i.e. for long and short correlation lengthanalyses). The plot shows both the constant air pressureimposed during the first year of operation (associated with airbubble formation) and the computed air pressures during thesubsequent 3 years of plant operation with the associatedcompression/decompression cycles


https://www.researchgate.net/publication/237089707_Stochastic_analysis_of_unsaturated_seepage_through_randomly_heterogeneous_earth_embankments?el=1_x_8&enrichId=rgreq-eae6efb9-c16e-4b53-87b3-935f5867ffdd&enrichSource=Y292ZXJQYWdlOzI2NDE1ODU1OTtBUzoxMzk4ODg1MzYwNjgwOTZAMTQxMDM2MzQzNDg3MQ==


the presence of air pathways. Water production is anotherkey aspect to be studied in a CAES project. It has to bevery low to prevent problems with the gas turbine (Succar& Williams, 2008). The water produced in the shortcorrelation analysis was the highest (Fig. 10(b)), but thepredicted amount of water produced is small for this kindof project.

A study of the literature suggests that there is excellentpotential for developing CAES storage in aquifers (e.g.Succar & Williams, 2008). A lesson learnt from this particularcase study is that CAES projects require an exhaustive geo-characterisation of the proposed site alongside intensivenumerical analyses. Encouraging feasibility studies based ona site investigation very close (<9 km) to the target CAESlocation were refuted when specific site information, and the

associated modelling, demonstrated that the location wasunsuitable for a CAES plant in the aquifer.

Aspects that need to be studied in detail for this kind ofproject include the rock permeability and its retentionproperties. A low permeability impedes the formation (in aconvenient time) of a suitable air bubble required for airstorage. Aquifer hydraulic properties also control pressurefluctuations during compression/decompression cycles thatare critical for CAES projects in aquifers. For example,system operations may be hampered due to a drop indecompression pressures below the operational range ofthe turbine. At the other extreme, high pressures duringcompression may affect the geo-mechanical integrity of theaquifer. A good understanding of the mechanical beha-viour of the host rock and cap rock, together with coupled

Initial air bubble (at endof constant-pressure period) Cycle 18 months 36 months

Air pressure: MPa

10.679.338.006.675.334.002.671.33

12.00

0.00

1.000.890.780.670.560.440.330.220.110.00

12.0010.679.338.006.675.334.002.671.330.00

1.000.890.780.670.560.440.330.220.110.00

0.250.220.200.170.150.120.100.070.050.02

Air saturation

Air saturation

Porosity

Air pressure: MPa

C

D

C

D

C

D

C

D

1000 m

75 m

30 m

Domain used in contour plots

200 m

200 m

75 m

30 m

(b)

(c)

(a)

(d)

(e)

(f)

(g)

75 m

30 m

75 m

30 m

75 m

30 m

Fig. 9. Results related to random distribution of porosity. Contour plots of (a) air pressure for short correlation length, (b) airsaturation for short correlation length, (c) air pressure for long correlation length, (d) air saturation for long correlation length,(e) porosity distribution for short correlation length and (f) porosity distribution for long correlation length. (g) Schematicrepresentation of the region analysed including the initial domain (75 m630 m) used in the contour plots presented in (a), (b), (c) and(d). C and D refer to compression and decompression respectively



hydromechanical analyses, are critical for evaluating(among others) borehole stability during daily cycles andthe mechanical integrity of both the aquifer and cap rockduring air injection/production cycles.

CONCLUSIONSA coupled hydromechanical study was performed to analysethe behaviour of a prospective CAES in an aquifer. Themodel was based on field and laboratory information of anactual site considered for a CAES project in Iowa, USA. Theanalysis confirmed that the site is not adequate for a CAESplant. The relatively low permeability of the natural rockwould prevent the development of a large air bubblenecessary to maintain a stable volume of air for the turbo-generator. Furthermore, the heterogeneous character of theaquifer (in particular layers C and D) could cause a fingeringeffect of the injected air and the development of air flowpathways. Additional field data and more comprehensivestochastic analysis are necessary to confirm the impact ofheterogeneities on the CAES performance.

ACKNOWLEDGEMENTSThe authors gratefully acknowledge Dr Stephen Bauer andDr Thomas Dewers from Sandia National Laboratories fortheir technical assistance and useful discussions. Financialsupport from the Zachry Department of Civil Engineering,Texas A&M University is also greatly appreciated.

REFERENCESDewers, T., Newell, P., Broome, S., Heath, J. & Bauer, S. (2014).

Geomechanical behavior of Cambrian Mount Simon sand-stone reservoir lithofacies, Iowa Shelf, USA. Int. J. GreenhouseGas Cont. 21, 33–48.

Fenton, G. A. & Griffiths, D. V. (2008). Risk assessment ingeotechnical engineering. Hoboken, NJ: Wiley.

Guimaraes, L., Gens, A., Sanchez, M. & Olivella, S. (2006). THM

and reactive transport analysis of expansive clay barrier inradioactive waste isolation. Comm. Num. Methods Engng 22,No. 8, 849–859.

Heath, J., Bauer, S. J., Broome, S. T., Dewers, T. & Rodriguez,M. (2013). Petrologic and petrophysical evaluation of the DallasCenter Structure, Iowa, for compressed air energy storage in theMount Simon sandstone. Albuquerque, NM: Sandia NationalLaboratories.

Kim, H., Rutqvist, J., Ryu, D., Choi, B., Sunwoo, C. & Song, W.(2012). Exploring the concept of compressed air energy storage(CAES) in lined rock caverns at shallow depth: a modelingstudy of air tightness and energy balance. Appl. Energy 92, No.0, 653–667

Kushnir, R., Dayan, A. & Ullmann, A. (2012). Temperature andpressure variations within compressed air energy storagecaverns. Int. J. Heat Mass Transfer 55, No. 21–22, 5616–5630.

Le, T., Gallipoli, D., Sanchez, M. & Wheeler, J. (2011). Stochasticanalysis of unsaturated seepage through randomly hetero-geneous earth embankments. Int. J. Num. Anal. MethodsGeomech. 36, No. 8, 1056–1076.

Le, T., Gallipoli, D., Sanchez, M. & Wheeler, J. (2013). Rainfall-induced differential settlements of shallow foundations onheterogeneous unsaturated soils. Geotechnique 63, No. 15,1346–1355.

Moridis, G., King, M. & Jansen, J. (2007). Iowa stored energy parkcompressed-air energy storage project: Compressed-air energystorage candidate site selection evaluation in Iowa: DallasCenter feasibility analysis. Prepared for the Iowa Stored EnergyPlant Agency by The Hydrodynamics Group LLC, p. 46.

Olivella, S., Gens, A., Carrera, J. & Alonso, E. (1996). Numericalformulation for a simulator (CODE_BRIGHT) for thecoupled analysis of saline media. Engng Comput. 13, No. 7,87–112

Succar, S. & Williams, R. H. (2008). Compressed air energystorage: Theory, resources, and applications for wind power.Princeton, NJ: Princeton Environmental Institute.

THG (The Hydrodynamics Group) (2011). Iowa stored energyplant agency compressed-air energy storage project: Finalproject report – Dallas Center Mt. Simon Structure CAESsystem performance analysis. Report prepared for the IowaStored Energy Plant Agency, Des Moines, IA, p. 52.

WHAT DO YOU THINK?

To discuss this paper, please email up to 500 words tothe editor at [email protected]. Your contribution willbe forwarded to the author(s) for a reply and, ifconsidered appropriate by the editorial panel, will bepublished as a discussion.

24 10

8

6

4

2

00 800 1000600400200 1200

18

12

Cum

ulat

ive

air f

low

: kg

Cum

ulat

ive

liqui

d flo

w: k

g

6

00

Porosity distribution:short correlation lengthPorosity distribution:uniform porosityPorosity distribution:long correlation length

Porosity distribution:short correlation lengthPorosity distribution:uniform porosityPorosity distribution:long correlation length

300250200

Time: d Time: d

(a) (b)

15010050 350

Fig. 10. (a) Cumulative air flow intake during the 1 year period of air injection at constant pressure (i.e. formation of air bubble).(b) Cumulative water production during the subsequent 3 years of compression/decompression cycles


Coupled hydromechanical analysis of an underground compressed air energy storage facility in sandstone

Documents