P edic i e model o sus ainable exploi a ion o geo he mal esou ces in
A ica: The case o Olka ia geo he mal ield
C. Zu i
a,*
, D. Fiaschi
a
, X.S. Musonye
b
, H.S. Mukhongo
b
, M. Na ula
c
, I.P. Da Sil a
c
a
Depa men o Indus ial Enginee ing, Uni e si y o Flo ence, I aly
b
Kenya Elec ici y Gene a ing Company, Pension Plaza-Nga a, Nai obi, P.O. Box 47936, Kenya
c
S a hmo e Uni e si y, Nai obi, Kenya
ARTICLE INFO
Keywo ds:
Geo he mal ene gy
A ican ene gy de elopmen
Renewable ene gy
Sus ainable decision making
ABSTRACT
Geo he mal ene gy is a c ucial enewable esou ce o a sus ainable u u e, especially in A ican na ions cu by
he Ri Valley, which holds as un apped po en ial. Howe e , high up on cos s and de elopmen isks emain
key challenges. This s udy in oduces a simpli ied model calib a ed wi h eal da a om Kenya's Olka ia
geo he mal ield. The model enables apid p elimina y assessmen s o bo h echnical and economic pe o mance,
equi ing minimal inpu da a. Addi ionally, i inco po a es a Li e Cycle Assessmen o e alua e en i onmen al
impac s, an aspec a ely explo ed in A ican geo he mal s udies. The esea ch analyses a ious echnological
con igu a ions, including Single Flash, Double Flash, and O ganic Rankine Cycle (ORC) sys ems, aiming o
imp o e e iciency wi hou addi ional d illing. Findings show ha in eg a ing an ORC wi h exis ing lash sys ems
can boos ene gy ou pu by up o 20.1 %, wi h only a modes ise in he Le elized Cos o Elec ici y. Compa ed o
he cu en Olka ia IV se up, hyb id sys ems demons a ed lowe ca bon emissions and educed ma e ial esou ce
use pe ene gy ou pu . Resul s con i m ha ORC in eg a ion o e s he mos sus ainable pa hway o de eloping
high- empe a u e geo he mal esou ces in he Eas A ican Ri . This app oach balances ene gy e iciency, eco-
nomic easibili y, and en i onmen al impac , p o iding aluable guidance o u u e powe plan de elopmen in
egula o y-cons ained se ings. This wo k is ully consis en wi h he objec i es o Sus ainable De elopmen
Goals (SDG) 7 and 13.
In oduc ion
Geo he mal ene gy is ecognized as a iable op ion o pa ially
eplacing ossil uel-based baseload elec ici y gene a ion, he eby
con ibu ing o he mi iga ion o ene gy- ela ed g eenhouse gas emis-
sions (Sha min e al., 2023). The A ican con inen , whe e elec ici y
demand is p ojec ed o inc ease subs an ially o e he nex h ee de-
cades, possesses signi ican geo he mal po en ial, p edominan ly
loca ed wi hin he Eas A ican Ri Sys em (EARS) (Omenda & Simiyu,
2015).
Cu en es ima es indica e an exploi able capaci y o app oxima ely
15 GW wi hin he EARS (Elba ba y e al., 2022); howe e , only 1.02 GW
has been de eloped o da e, wi h nea ly 95 % o his capaci y concen-
a ed in Kenya (Omenda e al., 2025). This conside able unde u iliza-
ion is p ima ily a ibu able o he high echnical and inancial isks
inhe en o ups eam de elopmen phases, speci ically su ace explo a-
ion and explo a ion d illing, which en ail subs an ial capi al
expendi u e and o e unce ain e u ns (Sho all e al., 2015).
Nume ous s udies ha e explo ed he wide ange o echnical, eco-
nomic, and en i onmen al challenges acing geo he mal ene gy de el-
opmen (Kang e al., 2022; Kuma e al., 2022; Tomac & Sau e , 2018).
In he A ican con ex , as in o he geo he mal a eas loca ed in de el-
oping coun ies, addi ional cons ain s hinde p og ess. E hiopia is a
pe inen example, whe e he geo he mal sec o aces a sho age o
quali ied expe ise. This sho age could be mi iga ed h ough a ge ed
capaci y building, such as collabo a i e aining pa ne ships wi h
o eign ins i u ions, engaging expe guidance and es ablishing dedi-
ca ed ins i u ions o s eng hen local capabili ies (Ben i e al., 2023).
Fu he mo e, he absence o speci ic laws and he go e nmen 's limi ed
ocus on geo he mal de elopmen esul in licensing delays, unde -
sco ing he u gency o de ining a clea na ional s a egy o geo he mal
in eg a ion in o he ene gy mix (Ben i e al., 2023).
Simila ly, he Uganda case highligh s challenges ela ed o limi ed
awa eness and in o ma ion, policy and ins i u ional challenges,
* Co esponding au ho .
E-mail add ess: [email p o ec ed] (C. Zu i).
Con en s lis s a ailable a ScienceDi ec
Ene gy o Sus ainable De elopmen
jou nal homepage: www.jou nals.else ie .com/ene gy- o -sus ainable-de elopmen
h ps://doi.o g/10.1016/j.esd.2025.101886
Recei ed 28 Augus 2025; Recei ed in e ised o m 2 No embe 2025; Accep ed 9 No embe 2025
Ene gy o Sus ainable De elopmen 90 (2026) 101886
A ailable online 15 No embe 2025
0973-0826/© 2025 The Au ho s. Published by Else ie Inc. on behal o In e na ional Ene gy Ini ia i e. This is an open access a icle unde he CC BY license
(
h p://c ea i ecommons.o g/licenses/by/4.0/ ).
inadequa e esea ch and de elopmen , and insu icien human capaci y
expe ise (Mu umba e al., 2021).
Add essing hese in e connec ed challenges in a coo dina ed manne
is essen ial o unlocking he ull po en ial o geo he mal esou ces in
A ica and ensu ing hei sus ainable con ibu ion o he con inen 's
ene gy u u e. In his ega d, ongoing in e na ional esea ch e o s
should p io i ize he de elopmen o ad anced, simpli ied, and epli-
cable emo e sensing and geoscien i ic modeling ools. Such ools would
suppo ea ly esou ce assessmen , guide policy, enhance communi y
accep ance, and e eal a egion's ue geo he mal po en ial in he
con ex o A ica and o he de eloping economies.
P e iously, esea che s ha e de eloped so wa e ools o e alua ing
he echno-economic easibili y o geo he mal p ojec s, conside ing
ac o s such as esou ce assessmen , plan design, inancial modeling,
and ma ke dynamics. Geo he mal Elec ici y Technology E alua ion
Model (GETEM) is an Excel-based ool used o es ima e he Le elized
Cos o Elec ici y (LCOE) o geo he mal p ojec s, including hyd o-
he mal and Enhanced Geo he mal Sys em (EGS). I allows use s o
inpu echnical and economic pa ame e s and uses a discoun ed cash
low me hod o calcula e cos s and assess di e en scena ios (Mines,
2016). GEO he mal echno-economic PHysics-based sImula o o
Resou ce E alua ion and Simula ion (GEOPHIRES) is a lexible and use -
iendly ool o e alua ing geo he mal sys ems' echnical and economic
pe o mance, including elec ici y gene a ion, di ec hea use, and
combined hea and powe (CHP). I simula es all majo sys em compo-
nen s, such as ese oi s, wells, and su ace plan s, and calcula es key
me ics like LCOE and Le elized Cos o Hea (LCOH) using use inpu s
o de aul co ela ions (Becke s & McCabe, 2019). The Flexible
Geo he mal Economics Modeling (FGEM) ool is designed o assess he
echnical and economic easibili y o lexible geo he mal ope a ions by
modeling he in e ac ion be ween he subsu ace, su ace plan , and
ene gy ma ke s. I suppo s he e alua ion o di e en dispa ch s a e-
gies, including mass low a ia ion, wellhead p essu e modula ion, and
in eg a ion wi h he mal ene gy s o age (TES) and li hium-ion ba e ies
(LiBs) (Aljub an & Ho ne, 2024).
Al hough comp ehensi e, hese ools a e bes sui ed o la e p ojec
s ages, as hey equi e ex ensi e inpu da a on subsu ace condi ions,
wells, plan design, and ela ed economic o en i onmen al ac o s.
Collec ing his da a can be di icul , especially in a eas wi h limi ed da a
a ailabili y, as seen in mos geo he mal egions in A ica. Thus, he
ques ion emains whe he i is possible o p edic he echno-economic
pe o mance o a plan wi h limi ed a ailable da a. Sol ing his ques-
ion could be aluable o assessing geo he mal po en ial on a la ge
scale. Mo eo e , a signi ican gap in he li e a u e is he inabili y o hese
me hods o e alua e he po en ial cascade exploi a ion o he esou ce,
an issue ha emains insu icien ly add essed.
The p ima y objec i e o his wo k is o de elop a echno-economic
model capable o assessing he geo he mal po en ial o elec ici y
gene a ion based on a limi ed se o inpu a iables. The model also
inco po a es he e alua ion o cascade uses o geo he mal esou ces,
aiming o op imize bo h exis ing and u u e ins alla ions. Applying his
ool in de eloping egions, such as A ica, would o e a aluable
quan i a i e con ibu ion o suppo policymake s and guide s ake-
holde s, p o iding a solid ounda ion o ene gy planning (Mohamed
and Kassim (2024); Ami & Khan, 2022; Kuma e al., 2022).
Ano he impo an aspec is including en i onmen al ac o s o
p o ide a mo e comple e and holis ic analysis. In li e a u e, he li e cycle
assessmen (LCA) me hodology is commonly employed o e alua e he
en i onmen al impac s o geo he mal plan s (Pa isi e al., 2020; GECO
p ojec , 2020; Zu i e al., 2022). In many s udies, LCA is combined wi h
exe gy-based e alua ions, leading o an exe go-en i onmen al analysis
(Man ida e al., 2023). This app oach o e s a mo e comp ehensi e
unde s anding o he en i onmen al impac o ene gy sys ems by in e-
g a ing he modynamic e iciency wi h sus ainabili y me ics.
Howe e , despi e i s widesp ead applica ion o LCA globally, such
analyses emain no ably absen in he A ican con ex (Muko o e al.,
2021), highligh ing a signi ican gap in he li e a u e ha his wo k aims
o begin add essing.
An addi ional inno a i e aspec o his wo k is he applica ion o an
LCA o an A ican case s udy, a con ex ha has been la gely unde ex-
plo ed in he exis ing li e a u e. This allows a compa ison be ween
s anda d ene gy sys ems and o he coun ies' geo he mal sys ems.
Fu he mo e, a no el me hodology is p oposed ha in eg a es echno-
economic and en i onmen al analyses, aiming o de ine a decision-
making app oach ha p io i izes he mos sus ainable solu ions,
conside ing ene gy, economic and en i onmen al aspec s.
The e o e, as pa o he collabo a ions es ablished wi hin he Eu-
opean p ojec LEAP-RE (LEAP – RE p ojec , 2020), a simpli ied model
has been de eloped o assess he easibili y o geo he mal esou ce u i-
liza ion o elec ici y gene a ion using a minimal se o inpu pa ame-
e s. The model is s uc u ed in a modula amewo k, allowing o he
e alua ion o di e en echnologies indi idually, as well as hei po-
en ial in eg a ion. This enables a cascading u iliza ion app oach, whe e
highe - empe a u e esou ces a e used i s , ollowed by lowe -
empe a u e ones, o he enhancemen o exis ing geo he mal plan s.
The model is simula ed using da a om Kenya's Olka ia geo he mal
ield, which is in he axial o he cen al Kenyan Ri . Cu en ly, he
Olka ia geo he mal ield is he second mos p oduc i e wo ldwide,
su passed only by he geyse s in he USA (Renkens, 2019). To suppo
he analysis, a da a collec ion e o was conduc ed o pe o m he i s
LCA o a geo he mal plan in A ica, add essing a gap in he exis ing
li e a u e (Muko o e al., 2021). Finally, he esul s o he echno-
economic and en i onmen al assessmen s a e in eg a ed o p o ide a
decision-making amewo k o e alua ing he sus ainabili y o he
p oposed solu ions.
Me hodology
Gene al app oach
To cha ac e ize a geo he mal esou ce, se e al pa ame e s a e
equi ed (Ba bie , 2002; Reed & Ande son, 2011; Di Pippo, 2008).
Howe e , ou key pa ame e s a e selec ed in he con ex o de eloping a
simpli ied p edic i e echno-economic model. The ou chosen pa am-
e e s include ese oi empe a u e, mass low a e, well dep h and
ambien ai empe a u e. The ou pa ame e s we e selec ed because
hey encapsula e he dominan physical d i e s ha de e mine a
geo he mal sys em's pe o mance, as al eady demons a ed in Zu i and
Fiaschi (2025). P e ious esea ch has also shown ha empe a u e and
mass low a e p ima ily de e mine he usable ene gy and powe ca-
paci y o a geo he mal sys em, di ec ly in luencing he ene gy con e -
sion p ocess and sys em design (Becke s e al., 2017; Becke s & McCabe,
2019). Con e sely, well dep h and ambien empe a u e in luence cos s,
e iciency and ope a ional cons ain s. Toge he , hese a iables ep e-
sen he mos c i ical de e minan s o bo h echnical and economic
pe o mance o a esou ce. Each pa ame e is de ined as ollows:
•Rese oi Tempe a u e (T
geo,in
). I de ines he possible applica ions
(elec ici y p oduc ion/hea /cold) and also he po en ial o he
selec ed echnology;
•Mass low a e (m
geo
). I de ines he capaci y o he ene gy sys ems
and hei componen s, such as hea exchange s, u bines, and o he
mechanical componen s;
•Well Dep h (L
well
). I de ines he dep h o he well and hus esul s in a
possible change in empe a u e a he wellhead and hus a he inpu
o an ene gy sys em;
•Ambien ai empe a u e (T
ai
). I de ines he a e age empe a u e o
he en i onmen and he ela ed in e ac ions a he condense and
cooling owe o powe plan s.
These pa ame e s a e no a bi a y. They o m a minimal ye su i-
cien se o pa ame e s capable o ep oducing he i s -o de
C. Zu i e al.
Ene gy o Sus ainable De elopmen 90 (2026) 101886
2
dependencies ha con ol plan pe o mance in eal sys ems. By
cen e ing he model on he ou inpu s, he amewo k achie es
compu a ional simplici y wi hou comp omising p edic i e eliabili y.
P edic i e model amewo k
The p edic i e model, p esen ed in his wo k, is di ided in o wo
main sec ions: he p oduc ion sec ion and he su ace sec ion. The p o-
duc ion sec ion ep esen s a one-dimensional model o a geo he mal
well ha es ima es he he modynamic s a e o he luid a he wellhead.
The su ace sec ion comp ises he he modynamic blocks ha model
geo he mal ene gy con e sion echnologies, including he Flash powe
plan and he Bina y ORC powe plan .
Each sec ion ep esen s a simpli ied and independen model ha
ecei es inpu s, p ocesses hem, and p oduces ou pu pa ame e s
ep esen a i e o he geo he mal luid cha ac e is ics. In he case o he
he modynamic blocks, he ou pu s also include he powe gene a ed by
he u bine. These ea u es allow each block o be connec ed o ano he
in se ies o in pa allel, adding blocks o he plan diag am like pieces o a
puzzle. Fo each he modynamic block, i s applicabili y depends on he
empe a u e classi ica ion o he inpu esou ce. Fig. 1 shows he con-
cep ual amewo k o his p edic i e model, showing model s ages and
inpu s and ou pu s.
The model is designed so ha he selec ion o each he modynamic
block—ei he Flash o ORC—depends on he empe a u e classi ica ion
o he geo he mal esou ce. This classi ica ion ollows he Lindal dia-
g am, which de ines empe a u e anges sui able o di e en
geo he mal applica ions (Gup a & Roy, 2007; Soelaiman, 2016). By
using his app oach, he model ensu es ha he chosen ene gy con e -
sion sys em is well-ma ched o he he mal cha ac e is ics o he
geo he mal esou ce, as illus a ed in Fig. 2.
Me a-model de elopmen
To de elop an accu a e and as p edic i e model, me a-models we e
designed. Me a-models a e ma hema ical ools ha simpli y he e alu-
a ion o inpu -ou pu ela ionships in complex sys ems (Pala e al.,
2019). Fo ins ance, hey e alua e how changes in inpu s like empe -
a u e, p essu e o low a e a ec ou pu s like powe p oduced o e i-
ciency. To cons uc his me a-model, inpu pa ame e anges we e
selec ed o each he modynamic block, as shown in Supplemen a y
ma e ial A (SM.A). A homogeneous dis ibu ion was c ea ed o each
pa ame e , o ganized in o h ee-dimensional cubic ma ices. Thus, he
esul ing ma ix had h ee-dimensional a ays o size N
p
x N
p
x N
p
, whe e
N
p
deno es he numbe o elemen s along each axis (in ege g ea e han
1). The model was un o e e y combina ion o hese inpu alues o
calcula e he op imal ins alled powe . The ob ained esul s we e o ga-
nized in o ou pu ma ices and subsequen ly in e pola ed using linea
mul idimensional in e pola ion (Chan e al., 1997) o c ea e he inal
me a-model. The alue o dimensions o inpu ma ix N
p
was ca e ully
chosen o c ea e a dense inpu g id and keep he p edic ion e o below
0.5 %, ensu ing a good balance be ween accu acy and compu a ional
speed. The ou -s ep schema ic ep esen a ion o he me a-model
de elopmen p ocess is illus a ed in Fig. 3.
This app oach allows he model o p edic ou comes ac oss a wide
ange o condi ions wi hou unning a ull- ime, ime-consuming and
cos ly simula ion each ime, hence sa ing ime and cos while main-
aining eliabili y. All he modynamic blocks we e implemen ed in Py-
hon using he CoolP op lib a y and a e desc ibed in he
The modynamic blocks sec ion. The app oach uses a ew inpu pa am-
e e s, and he me a-model gene a ion enables in e pola ion ac oss
ealis ic ope a ional egimes wi hou needing a ull high- ideli y simu-
la ion o e e y case, hus p ese ing p edic i e powe while keeping
compu a ional cos low.
Fig. 1. Concep ual amewo k o he model showing he model p ocessing s ages and hei inpu s and ou pu s.
C. Zu i e al.
Ene gy o Sus ainable De elopmen 90 (2026) 101886
3
The modynamic blocks
The he modynamic blocks de eloped in his s udy o m he
analy ical ounda ion o assessing he echnologies and hei pe o -
mance. The models we e alida ed agains published co ela ions and
ope a ional da a (Ka imi & Mansou i, 2018; Shamoushaki e al., 2022),
ensu ing consis ency wi h es ablished geo he mal sys em beha iou .
Quan i a i ely, he models ep oduce he expec ed hie a chy o pe o -
mance, whe e double- lash sys ems exhibi 10–15 % highe con e sion
e iciency han single- lash, while bina y sys ems, hough less e icien ,
enable be e eco e y o low-en halpy hea .
P oduc ion sec ion – geo he mal well
The i s he modynamic block ep esen s he geo he mal well. I
desc ibes he upwa d low o geo he mal luid om he deep ese oi o
he su ace, conside ing p essu e losses and co esponding empe a u e
changes, which in u n de e mines whe he he luid eaches he su ace
in liquid phase o liquid and apo phase. I ep esen s a simpli ied
e sion o he Boiling Poin Dep h (BPD) Model (Di Pippo, 2008; G an &
Bixley, 2011), assuming one-dimensional, e ical, and adiaba ic low
wi hin he well. Ini ial condi ion: The geo he mal luid s a s as a sub-
cooled liquid a he bo om o he well. As he luid ises, p essu e de-
c eases due o bo h g a i y and dynamic low e ec s. When he p essu e
d ops o he sa u a ion p essu e co esponding o he luid's empe a-
u e, lashing occu s, pa o he liquid u ns in o apo , and a wo-phase
low begins. Eq. (E.1) es ima es he boiling dep h L — he dep h in he
well whe e geo he mal luid i s s a s o boil (i.e., lashing begins).
L =P1−Psa (Tgeo)
ρ
g+C2˙
m2(E.1)
whe e: P1: P essu e a he bo om o he well (deep ese oi p essu e);
Psa (Tgeo): Sa u a ion p essu e o he geo he mal luid a ese oi em-
pe a u e Tgeo;
ρ
g: Hyd os a ic p essu e g adien (g a i y +densi y e m);
C2˙
m2: Addi ional dynamic p essu e loss due o low and ic ion e ec s.
Eq. (E.1) ep esen s he inal ou come o a se ies o ma hema ical
s eps ha a e ex ensi ely desc ibed and applied in he li e a u e. Fo his
eason, and due o space limi a ions in he a icle, only he ele an
e e ences a e p o ided he e, while he ull de i a ion and de ailed
explana ion a e a ailable in he Supplemen a y ma e ial B (SM.B).
Eq. (E.1) es ima es he boiling dep h L , which is c i ical in de e -
mining he phase s a e o he geo he mal luid a he wellhead. I L
exceeds o equals he o al well dep h (Lwell), he luid emains in liquid
phase up o he su ace, and an O ganic Rankine Cycle (ORC) echnology
is adop ed. Con e sely, i L <Lwell, lashing occu s wi hin he wellbo e,
p oducing a wo-phase mix u e sui able o he lash echnology. Thus,
Eq. (E.1) p o ides he i s decision c i e ion o selec ing he app o-
p ia e powe con e sion echnology based on ese oi p essu e, em-
pe a u e, and low condi ions. The esul s o his phase eed in o he
su ace sec ion phase o he modeling amewo k.
Fig. 2. Geo he mal p edic i e model concep .
Fig. 3. Fou -s ep schema ic me a-model de elopmen .
C. Zu i e al.
Ene gy o Sus ainable De elopmen 90 (2026) 101886
4
Su ace sec ion – lash and bina y powe plan
The Flash sys em model is di ided in o wo main blocks, ep esen ing
he wo undamen al phases o he he modynamic cycle in a geo he mal
powe plan : s eam p oduc ion and sepa a ion and cooling and e icu-
la ion.
•S eam p oduc ion and sepa a ion: This block co e s he ini ial pa o
he cycle, om he wellhead o he u bine inle . The high-p essu e
geo he mal luid passes h ough an expansion al e, causing a
p essu e d op ha pa ially lashes he liquid in o apo . The
esul ing wo-phase mix u e en e s a sepa a o , whe e he apo and
liquid phase a e sepa a ed. The apo ac ion is di ec ed o d i e he
u bine, while he liquid is di ec ed o seconda y use. The sepa a o
egula es he p essu e o ensu e he p ope ope a ion o he u bine.
The amoun o sepa a ed s eam is managed o maximize ene gy
p oduc ion wi hou d opping below he minimum p essu e equi ed
by he condense .
To ensu e s able ope a ion be ween he al e ou le and condense
p essu e, a con ol pa ame e x
el
is in oduced, de ined by Eq. (E.2).
x el =xsep
xmax
=xsep
x(h1ʹ;Pcond)(E.2)
I ep esen s he a io be ween he ac ual apo ac ion x
sep
and
he maximum allowable apo ac ion x
max
. The x
sep
ep esen s he
eal ope a ing alue o he sys em and co esponds o he unknown
a iable calcula ed du ing he simula ion. Con e sely, he x
max
de-
no es he highes a ainable alue, de ined as he condi ion whe e he
luid has he same en halpy (h
1
)as he geo he mal inle bu a he
condense p essu e (P
cond
). This de ini ion ensu es ha , o any alue
o x
el
wi hin he 0 and 1 ange, he sepa a o p essu e emains
highe han he condense p essu e, he eby allowing he u bine o
ex ac powe om he luid e icien ly. Such an app oach enhances
he nume ical s abili y du ing simula ion and simpli ies he o mu-
la ion o he me amodel by p o iding consis en con ol pa ame e s
o a ious geo he mal condi ions.
Ul ima ely, his me hod iden i ies he op imal ope a ing condi-
ions ha maximize u bine ou pu wi hou comp omising sys em
p essu e balance.
•Cooling and eci cula ion: A e passing h ough he u bine, he s eam
is condensed by a cooling sys em consis ing o a condense , wo
pumps, and a cooling owe . Pa o he cooled wa e is einjec ed
in o he sys em, while he emaining pa is used as cooling luid in
he condense . The sys em ensu es ha ope a ing empe a u es
emain wi hin sa e limi s, adap ing o he ex e nal ambien ai
empe a u e.
A key ea u e o he model is i s modula s uc u e, which allows he
ou pu om he i s block o be used as inpu o addi ional he mo-
dynamic cycles, such as in double lash sys ems o in ORC o hea he
seconda y luid.
The ORC bina y he modynamic block includes he main hea
exchange (MHE), u bine, condense , and pump, wi h an addi ional
ic i ious sepa a o used only o me amodel de elopmen . Unlike lash
sys ems, he ORC uses a seconda y wo king luid, allowing lexibili y in
luid choice based on he modynamic and en i onmen al pe o mance.
Selec ed luids (R1233zd(E), R1234ze(Z), isopen ane) we e chosen
because o hei subc i ical beha iou and low Global Wa ming Po en-
ial (GWP), and hei sui abili y o medium en halpy geo he mal sys-
ems consis en wi h geo he mal esou ces wi hin he A ican Ri (Di
Pippo, 2008; DiPippo, 2016; Tammone e al., 2021). Ano he key eason
o selec ing hese luids lies in he compu a ional na u e o he model,
which e alua es a e y la ge numbe o geo he mal esou ce combina-
ions. To ensu e easonable compu a ion imes and, mos impo an ly,
consis en model con e gence ac oss all simula ions, i was necessa y o
main ain he cycle unde subc i ical condi ions. The selec ed luids ully
mee his equi emen , p o iding nume ical s abili y and eliable pe -
o mance h oughou he modeling p ocess. The model de e mines
op imal wo king luid p essu e and mass low h ough an i e a i e
p ocess. Hea exchange is go e ned by he geo he mal empe a u e
di e ence (ΔT
geo
), adjus ed o p e en supe c i ical condi ions. I he
wo king luid becomes supe hea ed, i all en e s he u bine; i no , only
he apo po ion does. The en i e cycle is sol ed espec ing mass and
ene gy balances, ensu ing ealis ic sizing and compa ibili y wi h
ma ke -a ailable hea exchange s.
The ex ended desc ip ion and he equa ions o lash and bina y
blocks a e epo ed in he Supplemen a y ma e ial C (SM.C).
In e ms o echnological bene i s o design, he main ad an age o
he p oposed model lies in i s p e-implemen ed and adap able s uc u e,
which can be easily applied o di e en geo he mal con ex s. The model
au oma ically e alua es he op imal ope a ing pa ame e s o he plan
acco ding o he cha ac e is ics o he geo he mal esou ce. A ep e-
sen a i e example is he x
el
pa ame e used in Flash sys ems, which
allows he model o se app op ia e alues as he geo he mal esou ce
condi ions change o unde o -design ope a ing scena ios.
Economic modeling
The economic e alua ion o he sys em is ca ied ou in wo s ages.
The i s s age in ol es pe o ming a cos analysis o he indi idual plan
componen s, u ilizing he mo-economic co ela ions as hose p oposed
in Tu on e al. (2008). Key pa ame e s necessa y o hese co ela ions,
such as he su ace a ea o he condense and e apo a o , he low a es
managed by he comp esso , and he olume o he sepa a o , a e
de i ed om he he modynamic blocks. Fo he sepa a o , he cos
es ima ion employs he he mo-economic co ela ion p o ided by
Mosa a e al. (2017). The speci ic componen s and he co esponding
equa ions used o each a e de ailed in Supplemen a y ma e ial D (SM.
D). Economic es ima ions o geo he mal wells ypically ely on linea ,
exponen ial, o loga i hmic co ela ions. In his s udy, we adop ed he
co ela ion ailo ed o A ican applica ions as p oposed by Shamoush-
aki e al. (2021). The cos o he well can g ea ly a ec he o al cos o
he plan , bo h in e ms o ini ial in es men and LCOE (Ka imi &
Mansou i, 2018; Shamoushaki e al., 2021), and he e o e special
a en ion is pu on he numbe o well and hei dep h. The numbe o
wells in he analysis depends on he speci ic case. I known, i can be
di ec ly inpu in o he economic model. O he wise, wo op ions a e
a ailable: assume one p oduc ion and one einjec ion well, o es ima e
he well coun using an empi ical co ela ion based on plan capaci y, as
de i ed om li e a u e da a see equa ion and e e ence in he SM.D.
Wi h he economic model desc ibed abo e, he ini ial in es men cos s
and o al ope a ion and main enance (O&M) cos s a e e alua ed o
de e mine he Le elized Cos o Elec ici y (LCOE). The LCOE ep esen s
he ull cycle cos s (bo h ixed and a iable) o a gene a ion echnology
pe uni o ene gy, enabling compa isons ac oss di e en echnologies
i espec i e o hei size, cos s uc u e, o li espan (Uecke d e al.,
2013). In he con ex o enewable ene gy sys ems, he alue o p o-
duced ene gy is in luenced by he esou ce a ailabili y and by he
in e mi en na u e o he ene gy sou ce. Consequen ly, he LCOE is
in insically linked o he a iabili y pa e ns ha de ine he ene gy
gene a ion p o ile (T an & Smi h, 2018). This pa ame e is pa icula ly
impo an in his s udy, as i se es as a signi ican economic indica o
(de Sim´
on-Ma ín e al., 2022). By consolida ing hese cos s in o a single
economic indica o , LCOE e ec i ely quan i ies he cos o elec ici y
uni a y p oduc ion cos ($/kWh), as discussed by de Sim´
on-Ma ín e al.
(2022).
LCOE =
CTCI +CWD +∑n
i=1
Cp
(1+ )i
∑n
i=1Ei
(1+ )i
(E.3)
The equa ion's pa ame e s in Eq. (E.3) a e as ollows: C
TCI
is he o al
capi al in es men cos ; C
p
o powe plan is he To al P oduc ion Cos
C. Zu i e al.
Ene gy o Sus ainable De elopmen 90 (2026) 101886
5
(C
TPC
) (Shamoushaki e al., 2022) o o he applica ion is he annual
ope a ing and main enance cos (C
O&M
) (de Sim´
on-Ma ín e al., 2022);
E
i
is he annual elec ici y p oduced; is he discoun a e; n is he
li e ime o he ene gy sys em; C
WD
deno es he cos o wells d illing. Fo
he calcula ion o C
TCI
, he me hod and pa ame e s used in his wo k
we e aken om Ka imi and Mansou i (2018), Shamoushaki e al.
(2022). Fo he e alua ion o LCOE wi h he in eg a ion o geo he mal
wells Hacks ein and Madlene (2021) ha e been adop ed, and i allow o
conside he wells d illing cos . While o he e alua ion o C
TPC
, Sha-
moushaki e al. (2022) was ollowed. Fo C
O&M
, li e a u e e e ence
alues we e aken (Becke s e al., 2021). The discoun a e was se o
each applica ion a 7 % (Becke s e al., 2021) and he li e ime a ies
be ween powe plan and o he applica ion is ixed o 30 yea s (IRENA,
2020; Shamoushaki e al., 2021; Becke s e al., 2021; GEOENVI p ojec ,
2020). A discoun ing a e o 7 % was adop ed o e lec he inancing
condi ions o geo he mal p ojec s in de eloping economies suppo ed
by concessional unding. Empi ical da a om global epo s indica e
ha publicly inanced p ojec s ope a e wi hin a 6–10 % ange,
depending on coun y isk and access o low-in e es capi al. Thus, 7 %
was selec ed as a midpoin . Resea ch shows ha a geo he mal plan can
sus ain ope a ion o 25–40 yea s wi h p ope plan main enance and
se icing. To ensu e compa abili y wi h p e ious echno-economic
analysis and in e na ional s anda ds, he li espan used in his s udy
was se a 30 yea s. All co ela ions used and he gene al exp ession o
he he mo-economic model a e shown in Table 1. The model desc ibed
in his pa ag aph was implemen ed in he Py hon en i onmen .
Case s udies
Olka ia geo he mal ield
The case s udy o which he model de eloped in his wo k is applied
in he Olka ia geo he mal ield, ocusing on da a om wells ha eed
s eam o Olka ia IV. The selec ion o he Olka ia geo he mal ield il-
lus a es he p og essi e unde s anding o geo he mal sys ems h ough
sus ained esea ch-indus y collabo a ion. Pa ne ship be ween KenGen
and lea ning ins i u ions has os e ed con inuous unde s anding and
ad ancemen s in ese oi modeling and op imisa ion, well design and
ela ed s udies. The p e ious s udies ocused on sub-su ace ese oi
cha ac e isa ion (Musonye, 2015; Wanyonyi e al., 2024; Omollo e al.,
2022). This s udy ad ances esea ch on s eam cha ac e is ics, applicable
ene gy con e sion echnologies, and hei economic and en i onmen al
implica ions. The ope a ional da a o he Olka ia IV powe plan ,
co e ing well pe o mance, ese oi condi ions, s eam in ake da a, and
plan con igu a ion, p o ides a eliable basis o model alida ion.
Fu he , i s single- lash echnology has he po en ial o hyb idisa ion,
making i ideal o assessing he echno-economic pe o mance o
ad anced ene gy con e sion schemes o geo he mal esou ces ound in
he Eas A ican Ri . The powe plan is a single lash wi h an ins alled
capaci y o 150 MW
e,
ea u ing wo u bines, each wi h a a ed capaci y
o 75 MW
e
. The ne ou pu capaci y is 140 MW
e
and exploi s a liquid-
domina ed esou ce hos ed in T achy e o ma ion be ween 1.2 and 3
km below he su ace (Musonye, 2015). The plan is se ed by 21 wells:
18 p oduc ion wells and 3 e-injec ions wells. Each u bine is ed by a
s eam line wi h an in ake low a e o 135 kg/s, 180 ◦C empe a u e, and
p essu e be ween 9 and 12 ba . Se e al scena ios we e analyzed, which
a e b ie ly desc ibed below and discussed in mo e de ail in he ollowing
sec ion:
•Scena io Base case: This scena io ep esen s he case s udy o Olka ia
IV, hus using he da a ega ding he wells, a Single Flash sys em was
calcula ed
•Scena io Double Flash (FII): In his scena io, ano he lash is added o
he esul s ob ained o a single lash by connec ing wo he mody-
namic blocks
•Scena io T iple Flash (FIII): In his scena io, an addi ional he mo-
dynamic block is added o he FII case o simula e a hi d le el o
powe gene a ion
•Single Flash and Bina y cycle (FI - ORC) scena io: he e an ORC- ype
he modynamic block is added o he basecase scena io
•Double Flash and Bina y cycle (FII - ORC) scena io: He e, howe e , a
he modynamic block is added o he FII scena io ORC.
Li e Cycle Assessmen
Da a collec ion was ca ied ou o calcula e he en i onmen al
impac o a sus ainabili y analysis. The app oach ollowed was Li e Cycle
Assessmen (LCA), pe he amewo k es ablished by ISO 14040 and ISO
14044. The objec i e is o de e mine he en i onmen al impac o he
Olka ia IV powe plan , iden i y he mos signi ican ecological in-
dica o s and compa e hem o o he sys ems. The app oach used o his
analysis is a c adle- o-ga e ype. The sys em bounda ies encompass all
p ocesses ela ed o he cons uc ion, ope a ion, and main enance pha-
ses o he powe plan . The end-o -li e phase is no conside ed due o he
lack o a disposal p og am. The unc ional uni selec ed is he elec ical
ene gy p oduced by he powe plan in kWh, conside ing a use ul 30-
yea li e ime and a capaci y ac o o 0.94, co esponding o 8234 h/
yea o ope a ion.
En i onmen al model
The p ocesses o he cons uc ion phase which we e modeled in his
se lemen a e he geo he mal wells and wellheads; he building hos ing
he powe plan , and he collec ion pipelines. The a e age geo he mal
wells dep h is 3000 m each. The collec ion pipelines a e di ided in o 28
km o p oduc ion and 5 km o e-injec ion pipe. The machine y and
powe plan ha e no been modeled wi h p ima y da a om KenGen
because o lack o da a. Consequen ly, machine y and plan da a we e
adop ed om he model p oposed by Ka lsd´
o i e al. (2015). Fo he
scena io analysis in ol ing he ORC plan , he e e ence in en o y o
he machine y was aken om Zu i e al. (2024).
The ope a ion phase consumes 40.4 m
3
o wa e pe day o plan
ope a ion and app oxima ely 126 MWh o elec ici y o hea dis ibu-
ion. O his elec ici y, 16 % is aken om he na ional g id and 84 %
om plan p oduc ion. Di ec emissions o he a mosphe e a e mainly
CO
2
, H
2
S, H
2
and CH
4
. Six make-up wells a e expec ed o be buil in 25
yea s o main ain he geo he mal s eam low a e a he equi ed le el.
In addi ion, lub ican oil is consumed o he machine y, while Sodium
Ca bona e is used o an ico osion and an i-scaling o he pipes. All
in en o y da a a e epo ed in Supplemen a y ma e ials F (SM.F). The
esul s shown in his pape e e o he single sco e. The con ibu ion
Table 1
Economic co ela ions o he mo-economic model.
Cos ype Rela ion
To al cycle cos TCB=∑i− h componen
i=1s componen Ci
Cos o si e p epa a ion Csi e =0.05*TCB
Cos o se ice acili ies Cse =0.05*TCB
Alloca ed cos Calloc =0
To al di ec pe manen in es men CDPI =TCB+Csi e +Cse +Calloc
Cos o con ingencies and con ac o 's ee Ccon =0.18*CDPI
To al dep eciable capi al CTDC =CDPI +Ccon
Cos o plan s a up Cs a up =0.1*CTDC
Pe manen in es men CTPI =CTDC +Cs a up
Wo king capi al CWC =0
To al capi al in es men CTCI =CTPI +CWC
Cos o wages and bene i s CWB =0.035*CTDC
Cos o sala ies and bene i s CSB =0.035*CTDC
Cos o ma e ials and se ices CTPI =CWB
Cos o main enance o e head CMO =0.05*CWB
Di ec manu ac u ing cos s CDMC =CMO +CTPI +CSB +CWB
Fixed manu ac u ing cos CFIX =0.02*CTDC
To al annual cos o manu ac u e CCOM =CDMC +CFIX
Gene al expense CGE =0
To al p oduc ion cos Cp =CCOM +CGE
C. Zu i e al.
Ene gy o Sus ainable De elopmen 90 (2026) 101886
6
analysis o he main indica o analyzed is epo ed in he Supplemen a y
ma e ial F (SM.F.) The en i onmen al model is ealized using OpenLCA
2.4, and he backg ound da a comes om Ecoin en 3.10. I is impe a-
i e o s a e ha he LCA analysis elies on he Eu opean-based da a-
bases and in en o y due o he limi ed a ailabili y o local da a. Fo his
eason, ep esen a i e p o ide s om he Ecoin en da abase we e used,
selec ing p ocesses labeled “ma ke - GLO,” which deno e he global
ma ke , in o de o bes e lec global a e age condi ions, as is widely
done in he li e a u e (Blanc e al., 2020; GEOENVI p ojec , 2020;
Douziech e al., 2021). While his may in oduce some egional bias, he
selec ed da ase ensu es me hodological consis ency and compa ibili y
ac oss he analyzed scena ios. This assump ion may sligh ly a ec ab-
solu e impac alues, bu does no al e he o e all en i onmen al ends
and conclusions de i ed om he analysis.
Resul s
Valida ion echno-economic
Resul s me amodels
The alida ion o he he modynamic model, pe o med by
compa ing se e al powe plan s epo ed in he li e a u e, has been
included in Supplemen a y ma e ials G (SM.G) due o space limi a ions
in he main ex . The gene al esul s o he me amodels a e epo ed in
he ollowing pa ag aph.
The modynamic me amodels – powe p oduc ion
Fig. 4 compa es h ee geo he mal echnologies, by analyzing
ins alled powe as a unc ion o low a e and geo he mal luid em-
pe a u e, wi h an ex e nal empe a u e ixed a 30 ◦C. The me a-models
es ima e powe ou pu , highligh ing signi ican di e ences in ope a-
ional empe a u e anges. The Single Flash sys em achie es high powe
ou pu a geo he mal empe a u es abo e 250 ◦C, whe e he Bina y ORC
is ine ec i e. The Double Flash ou pe o ms he Single Flash, pa icu-
la ly be ween 100 ◦C and 150 ◦C, inc easing powe by 6–7 %, and abo e
200 ◦C, whe e gains exceed 20 %. Howe e , in high ambien empe a-
u es, ising condensa ion empe a u es educe e iciency, limi ing
powe gains o 0.5–1 %, making he sys em less iable.
Compa ing Fig. 4A and B, he Double Flash expands he ins alled
powe a ea beyond 50 MW and shows s eepe powe g ow h wi h
inc easing low a e (m
geo
) and geo he mal empe a u e (T
geo
). The Bi-
na y ORC, wi hin i s ope a ional ange, enables sligh ly la ge plan s
han he lash sys ems a he same geo he mal condi ions. Fo example,
a 200 ◦C and 400 kg/s, he ORC eaches 20.9 MW, Single Flash 20.1
MW, and Double Flash 23.6 MW. Fo lowe low a es (<200 kg/s) a
maximum T
geo
, ins alled capaci ies emain mode a e: Single Flash (~37
MW), Double Flash (~46 MW, +24 %), ORC (~12 MW).
A key dis inc ion lies in ope a ional empe a u e anges: ORC unc-
ions be ween 80 and 200 ◦C, while lash sys ems ex end om 100 ◦C o
>350 ◦C, allowing g ea e ene gy exploi a ion a high empe a u es.
While powe ou pu is simila when ope a ing a he same empe a u e,
he lash sys ems le e age highe empe a u e esou ces o g ea e e -
iciency. P edic ion unce ain y a ies, being lowes o Single Flash,
highe o ORC, and highes o Double Flash. These indings cla i y each
echnology's capaci y and op imal condi ions based on low a e and
empe a u e pa ame e s.
Economic me amodels – powe p oduc ion
This sec ion p esen s he esul s o he economic me amodels. Vali-
da ing he model equi es a simila app oach o he he modynamic
model bu wi h ac ual LCOE alues o speci ic cases, which a e a ely
p o ided alongside necessa y he modynamic and economic pa ame e s
in li e a u e. Thus, he model's p edic ions can only be compa ed o
known e e ence alues.
Fig. 4. Me amodel su ace ep esen ing he powe ins alled o single (A), double lash (B) and ORC bina y (C).
C. Zu i e al.
Ene gy o Sus ainable De elopmen 90 (2026) 101886
7
A majo con ibu o o o al plan cos and LCOE is he cos o wells,
p ima ily in luenced by hei numbe and dep h. To assess his impac ,
ou scena ios wi h well dep hs o 1000 m, 2000 m, 3000 m, and 4000 m
we e analyzed. Fig. 5 illus a es he LCOE (c$/kWh) as a unc ion o
geo he mal esou ce low a e (1–80 kg/s) and empe a u e, wi h
e e ence alues o pho o ol aic (8.7 c$/kWh, yellow), wind (8.4 c
$/kWh, pu ple), and gas u bine (18 c$/kWh, ed) om T inomics B.V.
(2020). The maximum cos h eshold is 30 c$/kWh, beyond which he
echnology is no iable.
Fo low low a es and empe a u es, he LCOE is high bu dec eases
apidly as condi ions imp o e. Howe e , inc easing well dep h shi s he
low-LCOE a ea signi ican ly o he igh , educing cos -e ec i eness.
Re e ence LCOE anges om IEA (2023) and IRENA (2017) s and a
12–8.8 c$/kWh and 14–4 c$/kWh, espec i ely. Addi ionally, he
analysis conside s a maximum low limi o 140 kg/s, ocusing on
in e media e-sized plan s (20–30 MW). La ge plan s, capable o
handling highe low a es, emain cos -e ec i e e en a lowe em-
pe a u es, pa icula ly when low a e and empe a u e inc ease
signi ican ly.
Case s udies: Olka ia IV
Fig. 6 p esen s he ool's p edic ions o bo h ene gy and economic
analyses in a comp ehensi e manne . In he ene gy analysis (Fig. 6A),
he ed ba ep esen s he o iginal Olka ia IV plan da a, which ea u es
a u bine a ed a 150 MW and a ne ou pu o 140 MW. In compa ison,
he Basecase p edic ion calcula es a ne powe o 137.9 MW, only 1.7 %
lowe han he ac ual alue, based on an e alua ed liquid low o
app oxima ely 721 kg/s a 179 ◦C om he sepa a o . Building on his,
he “F II” scena io in oduces a second lash u bine ha adds 26 MW o
yield a o al ou pu o 163.9 MW, u ilizing a seconda y low o 626 kg/s
a 111 ◦C om a second sepa a o . Fu he , he “F III” scena io con-
ibu es an addi ional 4.5 MW ( o al 168.4 MW), co esponding o
ins alled capaci y inc eases o 16.9 % o F II and 20.1 % o F III ela i e
o he Base case. When in eg a ing an ORC sys em, he “F I – ORC”
scena io—using he same inpu da a as F II—achie es an inc ease o
app oxima ely 29.2 MW (abou 19.2 %), while he “F II – ORC” scena io,
which builds on he F II ou pu , adds an ex a 5.6 MW o a o al inc ease
o 20.89 %. A ele an aspec ha should be highligh ed when
compa ing geo he mal ene gy wi h o he enewable sou ces conce ns
he capaci y ac o . The inc ease in ins alled powe ob ained in his
s udy may appea modes when compa ed o ha o wind o pho o ol-
aic sys ems. Howe e , i is impo an o no e ha he capaci y ac o ,
which is de ined as he a io be ween he ac ual annual ope a ing hou s
and he o al hou s in a yea , i is signi ican ly highe o geo he mal
powe plan s, ypically anging be ween 85 % and 95 % (IRENA, 2025).
In con as , he capaci y ac o o wind and pho o ol aic (PV) sys ems
s ongly depends on he geog aphical loca ion, wi h global a e ages o
app oxima ely 34 % o wind and 15–22 % o PV sys ems (IRENA,
2025). This, o cou se, has a signi ican e ec on he o al amoun o
ene gy gene a ed by such sys ems. Consequen ly, o an equi alen
ins alled capaci y, a geo he mal powe plan p oduces subs an ially
mo e ene gy han wind o PV sys ems. The economic analysis (Fig. 6B)
shows ha he base scena io is es ima ed a oughly $517 million in o al
cos s (CAPEX plus OPEX o e a 30-yea li espan), wi h OPEX accoun ing
o 31.3 % and CAPEX o 68.7 % o his alue. The F II scena io incu s a
cos inc ease o app oxima ely 19.8 % compa ed o he base case, and
he F III scena io sees a 27.3 % inc ease. Scena ios inco po a ing an ORC
sys em gene ally lead o highe cos s o a modes powe gain o abou 3
MW, wi h he “F I – ORC” and “F II – ORC” scena ios showing cos in-
c eases o 24.1 % and 27.3 %, espec i ely— indings ha align wi h
li e a u e epo s (IRENA, 2017; Sig ússon & Uihlein, 2015) a ibu ing
Fig. 5. Me amodel su ace ep esen ing LCOE (c$/kWh) o di e en scena ios well-dep h. Wells dep h (WD): A. 1000 m; B. 2000 m; C. 3000 m; D. 4000 m. (Red
line =Gas u bine; yellow line =Pho o ol aic; pu ple line =Wind.) (Fo in e p e a ion o he e e ences o colo in his igu e legend, he eade is e e ed o he
web e sion o his a icle.)
C. Zu i e al.
Ene gy o Sus ainable De elopmen 90 (2026) 101886
8
highe ins alla ion and main enance expenses o bina y ORC sys ems
compa ed o lash sys ems. No ably, he LCOE, ep esen ed by he o -
ange line and do s, exhibi s minimal a ia ion ac oss he scena ios,
inc easing sligh ly om app oxima ely 4.01 c$/kWh in he Basecase o
4.17 c$/kWh in he mos economically disad an ageous scena io.
Li e Cycle Assessmen
The single sco e o he analyzed scena ios, e alua ed using he
Fig. 6. Ba g aph o ins allable powe (le ) and ba g aph o economic cos ( igh ) o di e en scena ios o Olka ia IV.
Fig. 7. Single sco e compa ison o all geo he mal Scena ios and o he ene gy sou ces.
C. Zu i e al.
Ene gy o Sus ainable De elopmen 90 (2026) 101886
9
