A comprehensive assessment of energy efficiency of wastewater treatment plants: An efficiency analysis tree approach

A comp ehensi e assessmen o ene gy e ficiency o was ewa e ea men plan s: An e ficiency analysis ee app oach Alexand os Mazio is a , Ramon Sala-Ga ido b , Manuel Mocholi-A ce b , Ma ia Molinos-Senan e a,c,d, ⁎ a Depa amen o de Ingenie ía Hid áulica y Ambien al, Pon ificia Uni e sidad Ca ólica de Chile, A da. Vicuña Mackenna, 4860 San iago, Chile b Depa men o Ma hema ics o Economics, Uni e si y o Valencia, A d. Ta onge s S/N, Valencia, Spain c Ins i u e o Sus ainable P ocesses, Uni e si y o Valladolid, C/ D . Me gelina, S/N, Valladolid, Spain d Cen o de Desa ollo U bano Sus en able ANID/FONDAP/15110020, A . Vicuña Mackenna, 4860 San iago, Chile HIGHLIGHTS GRAPHICAL ABSTRACT •E ficiency analysis ee was used o e alu- a e he ene gy e ficiency o was ewa e ea men . •The a e age ene gy e ficiency o e alu- a ed was ewa e ea men plan s is 0.287. •Ene gy e ficiency is influenced by he age and echnology o he acili y. ABSTRACTARTICLE INFO Edi o : Damia Ba celo Keywo ds: Ene gy e ficiency Ene gy sa ings Reg ession ees Linea p og amming Boo s ap eg ession Was ewa e ea men plan s Was ewa e ea men plan s (WWTPs) a e ene gy in ensi e acili ies. Con olling ene gy use in WWTPs could b ing subs an ial benefi s o people and en i onmen . Unde s anding how ene gy e ficien he was ewa e ea men p ocess is and wha d i es e ficiency would allow ea ing was ewa e in a mo e sus ainable way. In his s udy, we employed he e ficiency analysis ees app oach, ha combines machine lea ning and linea p og amming echniques, o es i- ma e ene gy e ficiency o was ewa e ea men p ocess. The findings indica ed ha conside able ene gy ine ficiency among WWTPsin Chile exis ed. The mean ene gy e ficiency was 0.287 sugges ing ha ene gy use should cu educe by 71.3 % o ea he same olume o was ewa e . This was equi alen o a educ ion in ene gy use by 0.40 kWh/m 3 on a e age. Mo eo e , only 4 ou o 203 assessed WWTPs (1.97 %) we e iden ified as ene gy e ficien . I was also ound ha he age o ea men plan and ype o seconda y echnology played an impo an ole in explaining ene gy e fi- ciency a ia ions among WWTPs. 1. In oduc ion E alua ing he sus ainabili y o u ban wa e se ices has become a ele- an issue du ing he las wen y yea s (Molinos-Senan e e al., 2016). Sus- ainabili y is usually associa ed wi h he iple bo om line amewo k, i.e., social, economic and en i onmen al dimensions (Ma ques e al., 2015). Howe e , a e he Pa is Ag eemen adop ed a 21s Con e ence o he Pa ies o he Uni ed Na ions F amewo k Con en ion on Clima e Change (COP21), ene gy and g eenhouse gas emissions ela ed issues ha e acqui ed special a en ion. In his con ex , educing he ca bon oo - p in o u ban wa e u ili ies would con ibu e o mee he objec i es o he Pa is Ag eemen in he medium- e m. Was ewa e needs o be ea ed a high s anda ds be o e i is discha ged back o he en i onmen o i s euse (Feng e al., 2022). On a e age, high- Science o he To al En i onmen 885 (2023) 163539 ⁎Co esponding au ho a : Ins i u e o Sus ainable P ocesses, Uni e si y o Valladolid, C/ D . Me gelina, S/N, Valladolid, Spain. E-mail add ess: ma ia.molinos@u a.es (M. Molinos-Senan e). h p://dx.doi.o g/10.1016/j.sci o en .2023.163539 Recei ed 28 Feb ua y 2023; Recei ed in e ised o m 12 Ap il 2023; Accep ed 12 Ap il 2023 A ailable online 4 May 2023 0048-9697/© 2023 The Au ho s. Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). Con en s lis s a ailable a ScienceDi ec Science o he To al En i onmen jou nal homepage: www.else ie .com/loca e/sci o en incomecoun ies ea abou 70 % o he was ewa e hey gene a e whe eas in uppe -middle and lowe -middle income coun ies his a io d ops o 38 % and 28 %, espec i ely (UNESCO, 2017). Un ea ed was ewa e will lead o ade e io a ion on he ecological s a us o wa e bodiessuch as i e s and lakes (Mille e al., 2013;Ganguly and Dewan, 2020). Hence, in he coming yea s, he numbe o was ewa e ea men plan s (WWTPs) will in- c ease o achie e he a ge s defined by he Sus ainable De elopmen Goals (Goal 6) (Uni ed Na ions, 2015). An impo ance esou ce, om an economic and en i onmen al pe spec- i e, o he ope a ion o WWTPs is ene gy. P e ious s udies es ima ed ha he ea men o was ewa e equi es up o 4 % o elec ic ene gy in Uni ed S a es and 0.70 % o elec ici y consump ion in China (Longo e al., 2016, 2020; Niu e al., 2019). In Eu ope, ene gy use o ea ing was ewa e could explain o >1 % o consump ion (Walke e al., 2021). The ene gy in- ensi y o was ewa e ea men p ocess pushes up companies' p oduc ion cos s. P e ious s udies concluded ha ene gy cos s accoun o >60 % o wa e companies' ope a ing expendi u e (Gu e al., 2017). Mo eo e , en- e gy use leads o he gene a ion o g eenhouse gas emissions (GHG) which could ha e a nega i e impac on people and en i onmen (Wang e al., 2018;An e al., 2018;Ca doso e al., 2021). Thus, he ansi ion o- wa ds a sus ainable and ca bon e ficien was ewa e ea men p ocess is o g ea in e es o policy make s. The abo e challenges and objec i es gi e ise o he measu emen o he ene gy pe o mance o WWTPs and he need o ge a be e unde s anding on wha d i es ene gy e ficiency when ea ing was ewa e (Venka esh e al., 2014;To eg ossa e al., 2018;Molinos-Senan e and Mazio is, 2022). Based on he adi ional defini ion o e ficiency, which (in an inpu - o ien ed case) measu es he abili y o a decision-making uni (DMU) o p o- duce he same le el o ou pu s using less inpu s (Coelli e al., 2005), in he amewo k o WWTPs, ene gy e ficiency is defined as a syn he ic index ha in eg a es he olume o was ewa e ea ed, he amoun o pollu an s e- mo ed and he ene gy equi ed o ea was ewa e (He nández-Sancho e al., 2011). Hence, ene gy e ficiency is a me ic o benchma king ene gy pe o mances o WWTPs (Longo e al., 2016). This app oach di e s om he ene gy in ensi y concep defined as he ene gy consumed pe uni ol- ume o was ewa e ea ed (kWh/m 3 ) which igno es he main unc ion o a WWTP, e.g., emo ing pollu an s om was ewa e (Cas elle -Viciano e al., 2018). F om a me hodological poin o iew, he e a e wo main echniques o measu e e ficiency o DMUs, i.e., pa ame ic (econome ic) and non- pa ame ic (linea p og amming). Bo h app oaches compa e inpu s and ou pu s o DMUs and de i e ela i e e ficiency measu es. Da a en elop- men analysis (DEA), a non-pa ame ic app oach, can be used o e alua e he ene gy e ficiency o WWTPs (Ca doso e al., 2021) because o i s abili y o in eg a e mul iple inpu s and ou pu s in a single composi e indica o (Gue ini e al., 2016). DEA elies on he cons uc ion o he e ficien p o- duc ion on ie using obse ed da a on inpu s and ou pu s o he uni s e al- ua ed, i.e., e ficiency is no es ima ed econome ically (Yada e al., 2022). DEA builds a piecewise and linea on ie and assumes ha de ia ions o DMUs om he on ie a e due o ine ficiency only. Pas esea ch e alua ing he ene gy e ficiency o WWTPs is limi ed and ocused on he use o DEA me hod. He nández-Sancho e al. (2011) and He nández-Cho e e al. (2018) used a non- adial DEA model, o es ima e ene gy e ficiency o a sample o Spanish WWTPs. Gue ini e al. (2017) ap- plied a double boo s ap DEA model o assess he ene gy e ficiency o I al- ian WWTPs. The objec i e o he pape by Molinos-Senan e (2018) was o compa e he ene gy e ficiency among was ewa e ea men echnologies using a me a on ie DEA model. Longo e al. (2016) used he DEA-CCR (Cha nes,Coope and Rhodes) and DEA-BCC (Banke , Cha nes and Coope ) models o assess he ene gy e ficiency o WWTPs om di e en coun ies. The same DEA models we e used by Yang and Chen (2021) o es ima e he ene gy e ficiency o Chinese WWTPs. Longo e al. (2018) p oposed a o- bus ene gy e ficiency DEA o es ima e bias-co ec ed EE sco es o WWTPs. Despi e he posi i e ea u es o DEA o es ima e ene gy e ficiency o WWTPs, i is a de e minis ic app oach which means ha i is sensi i e o ou lie s. Mo eo e , i su e s om o e fi ing sugges ing ha e ficiency es ima es may no be accu a e (Es e e e al., 2020). This is because echni- cal ine ficiency sco e o each uni is es ima ed as he de ia ion o each ac- i i y o p oduc ion plan om he on ie o he p oduc ion possibili y se . The WWTPs´ ene gy e ficiency es ima ions by Molinos-Senan e and Mazio is (2022) migh su e om o e fi ing p oblem as well because hey used s ochas ic pa ame ic en elopmen o da a (S oNED) me hod which is a combina ion o DEA and s ochas ic on ie analysis app oaches. To deal wi h o e fi ing issues in e ficiency es ima ion and imp o ing he obus ness o he esul s, Es e e e al. (2020) de eloped a newly echnique, called e ficiency analysis ees (EAT) which b ings oge he machine lea n- ing and linea p og amming echniques. EAT o e comes he o e fi ing p oblem by applying a p uning p ocedu e based upon c oss- alida ion. I allows de e mining e ficiency e alua ion ou -o -sample o he assessed uni s (WWTPs) and he e o e, es ima ing he op imal le els o ene gy use o WWTPs. Es e e e al. (2020) demons a ed ha he EATme hod ou pe - o ms agains o he non-pa ame ic echniques. The e o e, i p o ides eli- able e ficiency sco es being app op ia e o benchma king analysis and policy decision making. Agains his backg ound, he main objec i e o his s udy is o p o ide a comp ehensi e assessmen o he ene ge ic pe o mance o a sample o WWTPs based on he EAT me hod. This app oach allows quan i ying he op imal le el o ene gy ha could be used o ea was ewa e based on di - e en pollu an s quali y-adjus ed olume h esholds. Because ene gy e fi- ciency is es ima ed a WWTP le el, po en ial ene gy sa ings i WWTPs we e e ficien a e also quan ified. Finally, we in es iga e he influence o he age and seconda y ea men echnology on he ene gy e ficiency o WWTPs. Ou s udy ex ends he cu en s and o li e a u e as ollows. To he bes o ou knowledge, his is he fi s ime ha an app oach ha combines bo h machine lea ning and linea p og amming echniques is used o measu e ene gy pe o mance o was ewa e ea men p ocess. The use o EAT me hod o es ima e ene gy e ficiency sco es o WWTPs o e comes he lim- i a ions o DEA app oach p e iously used by he li e a u e. Mo eo e , o he fi s ime, he op imal le el o ene gy use o WWTPs is es ima ed. This in o ma ion is e y ele an o he wa e egula o and wa e companies o define a ge s ha could p og essi ely be me . This no el piece o wo k was applied o a sample o Chilean WWTPs. 2. Me hodology In his sec ion we p esen he me hodology used o assess he ene ge ic pe o mance o se e al WWTPs. I is based on h ee s ages. The fi s and second s ages a e ela ed o he applica ion o he EAT me hod whe eas he hi d s age ocuses on iden i ying ac o s influencing he p e iously es- ima ed ene gy e ficiency sco es. In he fi s s age, he EAT app oach uses eg ession (decision) ees o de i e he p edic ed alue o he esponse a - iable (i.e., ene gy use in his case s udy). This alue is de i ed a e sepa a - ing he whole sample in o se e al non-o e lapping egions based on a se o ules ( h esholds) o he p edic o a iables (i.e., olume o was ewa e ea ed o emo e se e al pollu an s in his case s udy) (Rebai e al., 2019)(seeFig. 1). The EAT app oach inco po a es he concep o ee dis- posabili y (Es e e e al., 2021) and he e o e, he p edic ed alue o he e- sponse a iable is no he a e age alue bu he op imal (o maximum)one, which in ou s udy allows us o es ima e he op imal use o ene gy by WWTPs. In he second s age, p oduc ion on ie s and ene gy e ficiency sco es a e es ima ed using linea p og amming echniques. The es ima ed p oduc- ion on ie akes he shape o a s ep unc ion (Es e e e al., 2020)( Fig. 2). In he hi d s age, boo s ap unca ed eg ession echniques a e applied o s a is ically iden i y cha ac e is ics o he WWTPs influencing hei ene gy e ficiency. Le 's assume ha he e is a ec o o p edic o a iables defined as x1,...,xmwi h xi∈Rm. This se o a iables is employed o p edic a ec- o o esponse a iables defined as y,...,ynwi h yi∈Rn.TheEATme hod spli s he obse a ions in o wo nodes, Rand Lby selec ing a p edic o a iable jand a h eshold sj∈Sjwhe e Sjcap u es he se o likely A. Mazio is e al. Science o he To al En i onmen 885 (2023) 163539 2 h esholds o he a iable j o spli he da a in o (Es e e e al., 2022). The sepa a ion o he obse a ions in o se e al egions based on h esholds om he esponse a iables is done by minimizing he sum o he mean squa ed o e o . I s ma hema ical o m is as ollows: R L ðÞþR R ðÞ¼ 1 n∑xi;yi ðÞ∈ Lyi−y L ðÞðÞ 2þ1 n∑xi;yi ðÞ∈ Ryi−y R ðÞðÞ 2ð1Þ In Eq. (1) is he node o he eg ession ee; R L ðÞand R R ðÞa e he mean squa ed e o o le node and igh o he ee, i.e., Land R, espec i ely;nis he size o he sample and y L ðÞand y R ðÞa e he p e- dic ed alues o he esponse a iable ha a e es ima ed on he le and igh node o he ee, espec i ely. The eg ession ee can be isualized in Fig. 1. The p edic ed alues o he esponse a iable in he le and igh node o he eg ession ee a e de i ed om he ollowing equa ions: y L ðÞ¼max max yi:xi,yi ðÞ∈ L g ,yI Tkj ∗! L, R ðÞ L ðÞ  ð2aÞ y R ðÞ¼max max yi:xi,yi ðÞ∈ R g ,yI Tkj ∗! L, R ðÞ R ðÞ  ð2bÞ whe e Tdeno es he sub- ee ha is o med u ilizing he EAT echnique and he numbe o spli s is shown by k. Mo eo e , yI Tkj ∗! L, R ðÞ L ðÞ  and yI Tkj ∗! L, R ðÞ R ðÞ  show he se o lea nodeso he ee c ea eda e achie - ing he k- h spli ha Pa e o domina es node Land R(Es e e e al., 2020, 2021, 2022). The concep o Pa e o dominance is illus a ed in he Fig. 2 whe e a case o wo inpu s, x1and x2is conside ed. In Fig. 2 node ′ Pa e o-domina es node because a ′¼2,2ðÞ<b¼9,9ðÞwhe e aand b p esen poin s o nodes ′and , espec i ely. Node ′“is p e e able o” node because i employs less inpu s han node (Es e e e al., 2020). As pa o he second s age o he me hodology applied, he p oduc ion on ie ha he EAT app oach es ima es is p esen ed by he ollowing equa ion: d PTTk¼x,yðÞ∈Rmþ1 þ:y≤dTkxðÞ  (3) whe e dTkx ðÞ is he p edic o es ima o ega ding he sub- ee Tk: The ene gy e ficiency sco e o each uni assessed (i.e., WWTP) is mea- su ed by sol ing he ollowing linea p og amming: φxk,yk ðÞ¼min φ s: : ∑ ∈~ T∗λ a j≤φxjk,j¼1,...,m ∑ ∈~ T∗λ d T∗a ðÞ≥yjk, ¼1,...,p ∑ ∈~ T∗λ ¼1 λ ∈0,1 g ,i¼1,...,n (4) In Eq. (4) φdeno es he ene gy e ficiency sco e, a ,dT∗a ðÞðÞa e inpu - ou pu s poin s o all ∈T∗whe e * is he final sub- ee, and λa e in ensi y a iables ha a e used o es ima e he p oduc ion on ie (Es e e e al., 2020). I is no ed ha he ene gy e ficiency sco e can ake any alue be- ween ze o and one. A WWTP ha has an ene gy e ficiency sco e equal o one (φ¼1:0Þmeans ha i is 100 % ene gy e ficien . A WWTP ha ob ains an ene gy e ficiency sco e less han one (φ<1:0Þ, means ha i is ene gy ine ficien and can educe ene gy use o become mo e e ficien . We quan- i y he po en ial ene gy sa ings using he ollowing equa ion: Ene gys¼Ene gyc∗1φðÞ (5) whe e Ene gysdeno es he po en ial sa ings in ene gy use i he WWTP was ene gy e ficien and; Ene gycis he ac ual le el o ene gy use o each WWTP e alua ed. The hi d s ep o ou analysis is o ge a be e insigh o wha could d i e ene gy e ficiency o WWTPs. In doing so, we eg ess he ene gy e fi- ciency sco e o each WWTP assessed using he EAT app oach (φ) agains a se o s uc u al cha ac e is ics o he acili ies. Since he ene gy e ficiency sco e akes a alue be ween ze o and one, we employ unca ed eg ession. In pa icula , we u ilize boo s ap unca ed eg ession echniques de el- oped by Sima and Wilson (2007). The eg ession model akes he ollow- ing o m: φi¼δ0þδiμ′ iþεi(6) whe e φiis he ene gy e ficiency sco e; δ0is he cons an e m; μ′ iis he se o s uc u al cha ac e is ics o each WWTP iassessed, and δia e pa ame e s ha he eg ession model es ima es. Finally, εiis he e o (noise) e m which ollows he s anda d no mal dis ibu ion (Sima and Wilson, 2007). 3. Case s udy desc ip ion The iden ifica ion o ou lie s and/o a ypical obse a ions becomes un- damen al in non-pa ame ic me hods (De Wi e and Ma ques, 2010a). A pee index app oach 1 (De Wi e and Ma ques, 2010b) was applied o he o iginal da abase, which emb aces 238 WWTPs, o iden i y a ypical Fig. 1. Example o a eg ession ee. Fig. 2. Pa e o-dominance nodes. 1 O he me hods o iden i ying ou lie s a e le e age, supe -e ficiency and o de -m (De Wi e and Ma ques, 2010b). A. Mazio is e al. Science o he To al En i onmen 885 (2023) 163539 3 obse a ions.As a esul , 35 WWTPs we e emo ed om he da abase lead- ing o a o al o 203 WWTPs whose ene gy e ficiency was e alua ed. The 203 WWTPs ope a e in Chile and use fi e di e en seconda y ea men echnologies: i) con en ional ac i a ed sludge (CAS) (n = 79); ii) ex ended ae a ion (EA) (n = 43); iii) ae a ed lagoon (AL) (n = 16); i ) ickling fil e (TF) (n = 20); and ) o a ing biological con ac o o biodisk (BD) (n = 45). All WWTPs a e ope a ed by p i a e wa e companies because he Chilean wa e indus y was almos ull p i a ized be ween 1998 and 2004 (Molinos-Senan e, 2018). Ne e heless, he Chilean u ban wa e eg- ula o , i.e., he “Supe in endencia de Se icios Sani a ios”(SISS), is in cha ge o moni o ing he quali y o he was ewa e (e fluen ) be o e i is sa ely cha ged o he en i onmen . The da a used in his s udy come om he egula o and is o 2017. Choosing he inpu and ou pu a iables is he mos impo an s age in e ficiency assessmen as he esul s a e highly influenced by his choice (De Wi e and Ma ques, 2010b). Because we a e in e es ed in assessing he ene ge ics o WWTPs, he inpu (o esponse a iable in his s udy) a - iable was he elec ici y consumed by each WWTP exp essed in kWh/yea (Rod íguez-Ga cía e al., 2011;Bodik and Kubaska, 2013;Longo e al., 2016;Molinos-Senan e, 2018). I in ol es all elec ici y used o ea was e- wa e ega dless i i is om enewable o non- enewable sou ces. Un o u- na ely, his in o ma ion is no publicly a ailable and he e o e, any en i onmen al impac analysis abou elec ici y used by WWTPs could no be conduc ed. Acco ding o Wakeel e al. (2016) and Longo e al. (2019), he la ges amoun o ene gy consumed in WWTPs is in o m o elec ici y. The main objec i e o he WWTPs is o emo e pollu an s om was e- wa e o mee e fluen discha ge h esholds defined by he egula ion (Dong e al., 2017). Based on p e ious esea ch (He nández-Sancho e al., 2011;Gómez e al., 2017;He nández-Cho e e al., 2018;Longo e al., 2018;Huang e al., 2021), pollu an quali y-adjus ed ou pu s, es ima ed ac- co ding o Eq. (7), we e used as ou pu s (o p edic o a iablesin his s udy) in he ene gy assessmen o WWTPs. Quali y adjus ed ou pu p¼Volumeww Cpin Cpe Cpin  (7) whe e Volumeww p esen s he olume o was ewa e ea ed measu ed in cubic me e s pe yea . Cpin is he concen a ion o pollu an pin he influen and Cpe is he concen a ion o pollu an pin he e fluen . Hence, he ol- ume o was ewa e ea ed by each WWTP was modified o conside he e - ficiency in emo al o each pollu an p. Ou case s udy akes in o accoun h ee pollu an s: i) biochemical oxygen demand (BOD); ii) suspended solids (SS) and iii) phospho us (P). The e o e, we used h ee quali y-adjus ed ou - pu s o assess he ene ge ic pe o mance o WWTPs. Finally, o explo e he impac o s uc u al cha ac e is ics on he ene - ge ic pe o mance o WWTPs, we conside ed he ollowing a iables in ou analysis: i) he age o ea men acili y measu ed in o al numbe o yea s; ii) he ype o ea men echnology, a ca ego ical a iable ha con- side s he a ailable seconda y ea men echnologies. Table 1 epo s he desc ip i e s a is ics o he a iables used in he case s udy. 4. Resul s and discussion 4.1. Op imal le el o ene gy use in was ewa e ea men plan s To es ima e he le el o ene gy use in WWTPs based on pollu an quali y-adjus ed olume o was ewa e ea ed, he EAT algo i hm was sol ed (Fig. 3). Wi hin he h ee quali y-adjus ed ou pu s based on he pol- lu an s emo ed om was ewa e (BOD, SS and P), i is e idenced ha P quali y-adjus ed ou pu de e mines he maximum le el o ene gy use in WWTPs. This is because, as i is shown in Table 1,Pis hepollu an whose emo al e ficiency a ies mo e among he WWTPs e alua ed. By con as , he s anda d de ia ion o BOD and SS quali y adjus ed ou pu s is mo e bounded which means ha he pe o mance o he WWTPs e alua ed in he emo al o hese pollu an s is mo e homogenous han o P pollu an . Fig. 3 shows ha o hose WWTPs whose P quali y-adjus ed ou pu is mo e 2,267,961 m 3 /yea o was ewa e , he maximum ene gy consump- ion is 6,711,383 kWh/yea which in ol es ha he maximum ene gy con- sump ion is 2.95 kWh pe cubic me e o was ewa e ea ed adjus ed by P e ficiency emo al. When P quali y-adjus ed ou pu is be ween 35,624 and 2,267,961 m 3 /yea , hen he maximum le el o ene gy use could be a 915,708 kWh/yea . I he WWTP annually ea s <35,624 m 3 o P quali y-adjus ed olume o was ewa e , hen ene gy consump ion could each he le el o 213,620 kWh/yea . O e all, he esul s demons a ed ha was ewa e ea men is ene gy in ensi e. As a esul , we need o u - he unde s and how e ficien he ene gy pe o mance o was ewa e ea - men p ocess is. 4.2. Ene gy e ficiency assessmen o was ewa e ea men plan s The s a is ics o he ene gy e ficiency sco es es ima ed o he 203 WWTPs e alua ed a e epo ed in Fig. 4. I is shown ha , on a e age he en- e gy e ficiency o WWTPs was 0.287 which means ha he e alua ed acil- i ies could cu down ene gy use by 71.3 % o ea he same le el o was ewa e o emo e pollu an s. I e eals ha he ene ge ic pe o mance o was ewa e ea men p ocess is poo . This figu e is lowe han he a e - age ene gy e ficiency sco es es ima ed by pas esea ch. Fo Spanish WWTPs, He nández-Sancho e al. (2011) and He nández-Cho e e al. (2018) ound ana e age ene gy e ficiency o 0.310 and 0.460, espec i ely. Fo Chilean WWTPs, Molinos-Senan e (2018) es ima ed an a e age ene gy e ficiency sco e o 0.511. A simila a e age ene gy e ficiency (0.458) was epo ed by Gue ini e al. (2017) o a sample o I alian WWTPs. Finally, Longo e al. (2018) ound a e age ene gy e ficiency sco es be ween 0.12 and 0.40 o a la ge sample o WWTPs om di e en Eu opean coun ies. I should be no ed ha hese p e ious s udies used DEA me hods o es i- ma e ene gy e ficiency sco es which has some limi a ions, whe eas ou s udy es ima ed ene gy e ficiency sco es using he EAT app oach. Hence, di e ences in esul s among s udies migh be due o me hodological app oached used o ene gy e ficiency es ima ions. Acco ding o ou es ima ions, only 4 ou o 203 WWTPs, i.e., 1.97 % o he sample, a e ene gy e ficien (Table 2). These ou acili ies a e iden ified as he bes pe o me s in e ms o ene gy use. Two o hem, i.e., WWTP66 Table 1 Desc ip i e s a is ics o he a iables used o es ima e ene gy e ficiency sco es o was ewa e ea men plan s. Va iables Uni o measu emen Mean S d. De . Minimum Maximum Elec ici y consump ion kWh/yea 337,563 855,810 443 6,711,383 Was ewa e olumes BOD emo ed m 3 /yea 667,766 1,657,445 461 13,078,100 Was ewa e olumes SS emo ed m 3 /yea 649,010 1,620,520 445 13,182,505 Was ewa e olumes P emo ed m 3 /yea 472,332 1,300,256 9 11,927,636 Remo al e ficiency BOD % 90.58 9.66 28.86 98.93 Remo al e ficiency SS % 88.33 10.81 23.5 99.80 Remo al e ficiency P % 56.63 19.62 1.16 96.06 Age o acili y Yea s 19 6 5 40 Type o ea men echnology (CAS = 1, EA = 2, AL = 3, TF = 4, BD = 5) Ca ego ical 3 2 1 5 Obse a ions: 203. A. Mazio is e al. Science o he To al En i onmen 885 (2023) 163539 4 and WWTP96, we e o bigge size compa ed o he o he ones as shown by he high le els o pollu an s emo ed and ene gy use. These WWTPs a e el- a i ely old plan s wi h a cons uc ion age o 27 and 24 yea s old and use CAS and EA echnology, espec i ely. The o he wo ully ene gy e ficien acili ies a e ela i ely newly buil wi h a li e ha does no exceed he 15 yea s and use TF and BD echnologies o ea lowe le els o was ewa- e . This finding e idences ha WWTPs wi h di e en cha ac e is ics, i.e., age and echnology could be ene gy e ficien . The es o WWTPs wi hin he op 10 o acili ies showed an ene gy e - ficiency sco e which anged be ween 0.70 and 0.870. The ype o ea men echnology used among his g oup o acili ies a ied bu he e is ep esen- a ion o all echnologies conside ed in his s udy. Hence, he ype o sec- onda y ea men used o ea was ewa e is no a echnical limi a ion o achie e ela i ely good ene gy e ficiency. As a as he wo s pe o me s a e conce ned, i is ound ha he a e age ene gy e ficiency sco e anged be ween 0.020 and 0.070 (Table 2). This in- dica es ha his g oup o acili ies needs o make conside able sa ings in he ene gy use o ca ch-up wi h he mos ene gye ficien ones. The majo i y o wo s pe o me s a e old WWTPs wi h an a e age cons uc ion age o 19.6 yea s. This g oup o WWTPs is cha ac e ized by using CAS, EA and AL echnologies o ea was ewa e . By con as , none o he bo om 10 ene gy e ficiency WWTPs uses TF and BD as seconda y ea men . Mo e de ails abou he impac o echnology and age on he ene ge ic pe - o mance o WWTPs a e epo ed in Sec ion 4.3. In o de o u he analyze he a iabili y in he ene ge ic pe o mance o he 203 WWTPs assessed, Fig. 5 shows he dis ibu ion o he es ima ed ene gy e ficiency sco es among WWTPs. I is shown ha he majo i y o he WWTPs a e ine ficien om an ene gy pe spec i e. In pa icula , 99 ou o Fig. 3. E ficiency analysis ee (EAT) o es ima ing use o ene gy in was ewa e ea men plan s, whe e: P emo al deno es phospho ous (P) quali y adjus ed ou pu (Eq. (7)) in m 3 /yea ; Id is he node; n( ) is he numbe o obse a ions and yis he maximum ene gy use in kWh/yea . Fig. 4. S a is ics o he ene gy e ficiency es ima ions o assessed WWTPs. A. Mazio is e al. Science o he To al En i onmen 885 (2023) 163539 5 203 WWTPs (48.8 %) we e ound o ha e an ene gy e ficiency sco e which a iedbe ween 0 and 0.20. This means ha on a e age hese acili ies need o cu down ene gy use by >80 %. Addi ionally, he e we e 64 acili ies (31.5 %) wi h an ene gy e ficiency sco e which anged be ween 0.21 and 0.40. The e o e, he ene gy sa ing po en ial among hese WWTPs could ange be ween 60 % and 80 %. Only 40 ou o 203 acili ies, i.e., 19.7 % o he sample, p esen an ene gy e ficiency sco e la ge han 0.41. O e all, he findings demons a e ha conside able ene gy ine ficiency exis s in he was ewa e ea men p ocess among he assessed acili ies. The es ima ed ene gy e ficiency in ol es ha he assessed WWTPs could sa e ene gy o p oduce he same quali y-adjus ed ou pu s, i.e., o ea he same olume o was ewa e wi h he same pollu an s' emo al e ficiency. Based on he ene gy e ficiency sco es o he 203 WWTPs and hei cu en ene gy use, po en ial ene gy sa ing was es ima ed o be 42,465,302 kWh/yea which is equi alen o an a e age o 0.40 kWh/m 3 . Fig. 6 e eals ha po en ial ene gy sa ings a e he e ogenous among he e alua ed WWTPs. The median alue is 0.32 kWh/m 3 whe eas he 25 h and 75 h pe cen iles a e 0.19 kWh/m 3 and 0.53 kWh/m 3 , espec i ely. The ex eme alues co espond o hose WWTPs wi h he lowes ene gy e - ficiency sco es. Con olling ene gy use could ha e posi i e benefi s o people and en i- onmen . Fi s , by educing ene gy use in he ope a ion o WWTPs, wa e companies could educe ope a ing cos s which could be u he passed on o cus ome s in e ms o lowe bills. The a e age elec ici y p ice o Chile in 2017 was 65.05 €/MWh (CNE (Nacional Ene gy Commission), 2020). Hence, based on he olume o was ewa e ea ed by he assessed WWTPs, hey could sa e a ound 2,762,368 €/yea i hey we e ene gy e fi- cien . I is equi alen o 0.026 €/m 3 o was ewa e ea ed. Second, ene gy sa ings may lead o a conside able educ ion in GHG emissions. The use o enewable ene gy in ea ing sewage could lead o lowe emissions eleased in he a mosphe e. Based on he elec ical p oduc ion mix o Chile in 2017, he GHG emission ac o was 449.73 KgCO2eq/MWh (ME (Chilean Minis- y o Ene gy), 2022). I he WWTPs e alua ed we e ene gy e ficien , hey could sa e 19,098 ons CO 2 eq/yea . Acco ding o he Wo ld Bank (2022) da abase, he a e age annual ca bon emission pe capi a in Chile in 2017 was 4.7 ons o CO 2 eq. Hence, i he WWTPsassessed we e ene gy e ficien , Table 2 Top 10 and bo om 10 ene gy e ficien WWTPs. WWTP iden ifica ion Ene gy e ficiency sco e Ene gy sa ing po en ial (kWh/yea ) Ac ual ene gy (kWh/yea ) BOD quali y-adjus ed olume SS quali y-adjus ed olume P quali y-adjus ed olume Age Technology Top 10 ene gy e ficien uni s WWTP66 1.000 0 869,880 44,051 45,313 35,624 27 CAS WWTP96 1.000 0 1,340,569 2,567,717 2,535,503 2,267,962 24 EA WWTP158 1.000 0 1097 461 445 393 15 TF WWTP194 1.000 0 1572 5234 5235 9 9 BD WWTP157 0.870 923 7103 5299 5240 3874 18 TF WWTP120 0.827 639,915 3,698,930 5,802,318 5,430,766 2,741,136 15 EA WWTP197 0.770 2011 8744 6014 5841 5052 10 BD WWTP111 0.710 126,692 436,869 677,122 642,038 503,489 23 EA WWTP71 0.700 25,092 83,639 72,041 61,653 63,742 14 CAS WWTP123 0.700 241 803 69,089 90,586 91,665 11 AL Bo om 10 ene gy e ficien uni s WWTP23 0.070 38,957 41,889 65,065 62,060 38,491 17 CAS WWTP133 0.069 417,554 448,500 651,829 494,116 517,640 31 AL WWTP13 0.064 216,439 231,238 675,648 692,818 559,040 12 CAS WWTP119 0.061 420,877 448,218 676,596 641,377 581,390 23 EA WWTP80 0.058 280,471 297,740 675,544 677,261 612,112 12 EA WWTP129 0.058 584,746 620,749 1,490,316 1,538,891 618,420 22 AL WWTP7 0.045 254,330 266,314 1,078,200 1,084,140 785,366 16 CAS WWTP10 0.042 676,293 705,943 1,421,667 1,419,844 858,325 19 CAS WWTP105 0.039 510,772 531,500 953,003 942,167 912,534 22 EA WWTP102 0.020 747,810 763,071 2,486,964 2,370,157 1,822,024 22 EA Fig. 5. Ene gy e ficiency sco es o WWTPs e alua ed. A. Mazio is e al. Science o he To al En i onmen 885 (2023) 163539 6 po en ial sa ings in GHG emissions would be equi alen o he annual GHG emi ed by 4063 Chilean people. 4.3. Fac o s influencing ene gy e ficiency In o de o ge a be e unde s anding on wha d i es he ene gy e fi- ciency o was ewa e ea men p ocess we need o look a he esul s e- po ed in Table 3.The eg essionfindings demons a e ha he age o he WWTPs and he ype o seconda y ea men echnology had an impo an pa in explaining ene gy e ficiency a ia ions ac oss acili ies. The age o ea men plan had a nega i e sign and was s a is ically significan . This means ha he olde he WWTP is, he lowe i s ene gy e ficiency could be. The esul indica es ha an inc ease in he age o ea men plan by one yea could lead o a educ ionin i s ene gy e ficiency by0.101 % on a - e age. This migh be explained by he ac ha olde acili ies p esen less ene gy e ficien pumps and ae a ion sys ems in he biological p ocess. I should be no ed ha ae a ion accoun s o he la ges ac ion o WWTPs´ ene gy cos s, anging om 45 % o 75 % o o al ope a ional cos s (Longo e al., 2016). E idence on he influence o he age o WWTPs on ene gy e - ficiency is inconclusi e. On he one hand, He nández-Sancho e al. (2011); Molinos-Senan e e al. (2014) and Gue ini e al. (2017) ound ha ene gy e ficiency o WWTPs was no a ec ed by he age o he acili ies. By con- as ,Molinos-Senan e and Mazio is (2022) concluded he opposi e because hey ound ha WWTPs younge han 10 yea s old p esen ed he la ges en- e gy e ficiency sco es. Fig. 7 shows he dis ibu ion o ene gy e ficiency sco es based on he age o WWTPs. The esul s indica e ha he olde acili ies a e less ene gy e ficien han new ones. In pa icula , WWTPs ha ha e a li e o 1 o 10 yea s showed an a e age ene gy e ficiency sco e o 0.339. In con as , acili ies ha ha e a li e o >10 yea s showed an a e age ene gy e ficiency sco e o 0.263. O e all, he esul s sugges ha he ecen ly buil WWTPs can be mo e ene gy e ficien han olde ones. Howe e , on a e age he e al- ua ed acili ies a e cha ac e ized by high le els o ene gy ine ficiency. Hence, he e is oom o conside able imp o emen in ene gy pe o mance o newly and olde buil plan s. Looking a he ype o seconda y ea men echnology, Table 3 shows ha in he eg ession analysis, his a iable had a posi i e sign and was s a- is ically significan . This means ha on a e age he WWTPs ha ope a ed based on EA, AL, TF, and BD echnologies we e ound o be mo e ene gy e - ficien han hose acili ies using CAS echnology. Whe eas CAS echnology has been widely adop ed wo ldwide o ea domes ic was ewa e , i is now being ecognized as lacking economic and en i onmen al sus ainabili y, es- pecially wi h espec o he ine ficien use o ene gy (Sheik e al., 2014; Ga ido-Base ba e al., 2018). This conclusion was also e idenced by pas s udies (Molinos-Senan e, 2018;Molinos-Senan e and Mazio is, 2022) who ocused on compa ing he ene gy e ficiency o WWTPs using di e en seconda y ea men echnologies. We nex discuss he ela ionship be ween ene gy e ficiency, ene gy sa - ings po en ial and he ype o echnology used. This is shown in Table 4. The esul s highligh ha plan s ha use CAS echnology a e less ene gy e - ficien han he ones ha use o he ypes o ea men echnologies. The e we e 79 plan s ha use CAS echnology and epo ed an a e age ene gy e - ficiency sco e o 0.228. This means ha hese acili ies could become mo e ene gye ficien by educingene gy use by77.2 % on a e age. The po en ial sa ings in ene gy use could be 0.494 kWh/m 3 on a e age. By con as , i was ound ha acili ies who used EA and TF ea men echnologies e- po ed highe le els o ene gy e ficiency sco es. The a e age ene gy e fi- ciency o plan s ha use EA and TF echnologies we e 0.346 and 0.364, espec i ely. Al hough hei ene gy pe o mance was be e han he plan s ha use CAS echnology, he po en ial sa ings in ene gy use a e subs an ial. I is es ima ed ha ene gy po en ial sa ings could each he le el o 0.434 kWh/m 3 and 0.296 kWh/m 3 when acili ies use EA and TF echnol- ogies, espec i ely. I should be no ed ha WWTPs using BD echnology a e hose wi h he lowes po en ial o sa e ene gy (0.270 kWh/m 3 )al- hough hey a e no he mos ene gy e ficien . This is because, cu en ly, his ype o acili ies is using less ene gy han he o he s based on CAS, EA, AL and TF echnologies. O e all, he esul s indica e ha ene gy pe o mance o WWTPs is influ- enced by i s age and he seconda y ea men echnology i uses. Olde acili ies a e less ene gy e ficien han newly buil ones. Plan s ha use Fig. 6. S a is ics o he po en ial ene gy sa ings o assessed WWTPs. Table 3 Fac o s influencing he ene gy e ficiency o WWTPs. Es ima es o boo s ap un- ca ed eg ession. Va iables Coe S d. e o z-s a p- alue Cons an 3.310 0.121 27.355 0.000 Age o acili y −0.101 0.041 −2.463 0.013 Type o echnology 0.121 0.031 3.903 0.000 sigma 0.251 0.056 4.482 0.000 X 2 (3) 71.54 p- alue 0.000 Obse a ions: 203. Bold s a is ics a e s a is ically significan a 5 % significance le el. A. Mazio is e al. Science o he To al En i onmen 885 (2023) 163539 7 CAS echnology may equi e high le els o ene gy use and hus could be less ene gy e ficien han hose using o he ypes o seconda y ea men echnologies. Pas esea ch (e.g., Ca alho e al., 2012;He nández-Cho e e al., 2018;Ca doso e al., 2021) has e alua ed he p esence o economies o scale in he pe o mance o wa e and was ewa e u ili ies leading o mixed esul s. To be e unde s and he influence o he olume o was e- wa e ea ed, i.e., economies o scale, on he ene gy e ficiency o WWTPs, Fig. 8 shows he es ima ed ene gy e ficiency sco e o each acili y e alua ed agains i s olume o was ewa e ea ed. No di ec ela ionship among bo h a iables was e idenced. The Pea son co ela ion coe ficien be ween ene gy e ficiency sco e and olume o was ewa e ea ed was 0.156 which means ha bo h a iables a e no ela ed. Mo eo e , Fig. 8 shows ha he la ges acili y ea ing 13,596,615 m 3 o was ewa e pe yea p esen s an ene gy e ficiency sco e o 0.19. Hence, po en ial disecon- omies o scale a e iden ified o his (and o he ) e alua ed WWTPs. The 203 assessed WWTPs we e ca ego ized in h ee g oups: i) WWTPs ea ing <100,000 m 3 /yea ; ii) WWTPs ea ing be ween 100,000 and 500,000 m 3 /yea and; iii) WWTPs ea ing >500,000 m 3 /yea (He nández-Sancho e al., 2011). 2 The a e age ene gy e ficiency o each g oup o acili ies was 0.312, 0.259 and 0.303, espec i ely. A K uskal-Wallis es was conduc ed o e i y whe he he di e ences be- ween ene gy e ficiency sco es a e s a is ically significan (He nández- Cho e e al., 2018). The p- alue was la ge han 0.05 which means ha di e ences in a e age ene gy e ficiency among g oups o WWTPs a e no s a is ically significan . 5. Conclusions The emo al o pollu an s om was ewa e o a oid en i onmen al damage is an ene gy in ensi e p ocess. Con ol o ene gy use could ha e a posi i e influence o people and en i onmen . Unde s anding he op imal use o ene gy du ing he ope a ion o WWTPs and wha d i es ene gy equi emen s is o g ea in e es o policy make s and wa e companies´ manage s. A ele an ool o ene gy use imp o emen s in WWTPs is benchma king i s ene gy e ficiency. In doing so, eliable and obus me h- odological app oaches should be used. Hence, in his s udy, o he fi s ime, he e ficiency analysis ee (EAT) me hod was employed o comp e- hensi ely e alua e he ene gy e ficiency o a sample o WWTPs. EAT b ings oge he machine lea ning and linea p og amming echniques o e coming he limi a ions o DEA me hod, which is he mos common me hod used in heli e a u e oassessene gye ficiency o WWTPs. The main findings o his s udy a e as ollows. Fi s , i is ound ha he e ficiency in he emo al o P significan ly influences he use o ene gy by WWTPs. Second, only 4 ou o 203 assessed WWTPs (1.97 %) we e ene gy e ficien whe eas he o he acili ies p esen oom o sa e ene gy. The a e - age ene gy e ficiency was 0.287 meaning ha on a e age WWTPs should cu down ene gy use by 71.3 % o ea he same olume o was ewa e . This is equi alen o a educ ion in ene gy use by 0.40 kWh/m 3 .Thi d,i has been demons a ed ha he age o he acili y and he seconda y ea - men echnology used o ea was ewa e significan ly influence on he en- e gy e ficiency o he WWTPs. Facili ies using CAS we e iden ified as he less ene gy e ficien e idencing he lack o economic and en i onmen al sus ainabili y o his echnology. This esea ch p o ides scien ific guidance o benchma king he ene gy e ficiency o WWTPs and he e o e, o he e ficien ope a ion o he was e- wa e ea men indus y ha can benefi people and he en i onmen . This s udy e idences he poo ene gy e ficiency o he WWTPs assessed and he e o e, e eals he need o de elop policies and implemen ac ions by wa e egula o s and wa e companies o imp o e he ene gy e ficiency o WWTPs. In pa icula , WWTPs could educe ene gy consump ion and inc ease ene gy e ficiency by adop ing se e al measu es such as: i) op imiza ion o p ocesses by ins alling sma me e s and de eloping con- ol sys ems o he op imal ope a ion o pumps and ae a ion sys ems and; ii) eco e y o he ene gy om was ewa e such as hea o elec ici y om sewage sludge. This ene gy could con ibu e o he educ ion in he o e all ene gy equi emen s o he acili y. Gi en he economic and 2 WWTPs we e ca ego ized based on al e na i e olumeo was ewa e ea ed wi hou find- ing p- alues lowe han 0.05 in any case. Fig. 7. A e age ene gy e ficiency acco ding o g oups o was ewa e ea men plan s by age. Table 4 Ene gy e ficiency sco es and ene gy po en ial sa ing o he assessed WWTPs by ype o echnology. Technology Ene gy e ficiency sco es (indica o ) Po en ial ene gy sa ings (kWh/m 3 ) Mean S d. De . Min Max Mean S d. De . Min Max CAS 0.228 0.147 0.042 1.000 0.494 0.359 0.000 1.841 EA 0.346 0.230 0.020 1.000 0.434 0.404 0.000 2.200 AL 0.278 0.180 0.058 0.700 0.465 0.277 0.002 0.790 TF 0.364 0.240 0.100 1.000 0.296 0.204 0.000 0.663 BD 0.302 0.203 0.100 1.000 0.270 0.200 0.000 1.060 A. Mazio is e al. Science o he To al En i onmen 885 (2023) 163539 8 en i onmen al ele ance o he use o ene gy issues, enhancing he ene gy e ficiency o WWTPs should be a p io i y o egula o s. In his con ex , wa e egula o s should define obliga o y ac ions o wa e companies o educe hei ca bon oo p in such as pe iodically conduc ing ene ge ic au- di s, ea ing sewage sludge h ough anae obic diges ion o p oduce biogas when he WWTP is la ge han a p edefined size, economically incen i ize he use o enewable ene gy, e c. Mo eo e , wa e companies should also be anspa en ega ding he ene gy used o ea ing was ewa e h ough he use o ca bon oo p in labels which could be sha ed wi h cus ome s ia wa e bills and in he webpage o he wa e egula o . Al hough his s udy p o ided a no el me hodological app oach o benchma k he ene ge ic pe o mance o WWTPs, i is no exemp o limi a- ions wi h espec o exogenous a iables which migh influence es ima ed ene gy e ficiency sco es. Fi s ly, he a iable age conside ed in his s udy co esponds o he yea in which each WWTP s a ed o ope a e. The e o e, i does no conside po en ial upg ades made o hem since his yea . Sec- ond, he esponse a iable “elec ici y used”in eg a es all elec ici y used in he WWTP. The e o e, i emb aces elec ici y used o ea was ewa e and sewage sludge. Because di e en echnologies could be used in WWTPs o ea sewage sludge, achie ing di e en ea ed sewage in e ms o quali y migh also influence ene gy e ficiency o WWTPs. Finally, he e a e o he po en ial exogenous a iables such as load ac o , dilu ion ac o and was ewa e empe a u e ha migh influence ene gy e ficiency o WWTPs. Due o he lack o public a ailable in o ma ion, hese a iables we e no included in he ene gy e ficiency assessmen conduc ed in his s udy. In o he wo ds, he main limi a ion o his s udy is ela ed o he lack o a ailable da a. Addi ional in o ma ion on exogenous a iables influencing ene gy e ficiency o WWTPs mus be conside ed o ge a be e unde s anding on he ene ge ic pe o mance o WWTPs and suppo indi- idual WWTP ene gy e ficiency op imiza ion. CRediT au ho ship con ibu ion s a emen Alexand os Mazio is: Concep ualiza ion; W i ing-O iginal D a ; Supe - ision. Ramón Sala-Ga ido: Fo mal analysis; Me hodology. Manuel Mocholi-A ce: Me hodology; Valida ion. Ma ía Molinos-Senan e: Concep ualiza ion; W i ing-O iginal, Supe ision. Da a a ailabili y Da a will be made a ailable on eques . Decla a ion o compe ing in e es The au ho s decla e ha heyha e no known compe ing financial in e - es s o pe sonal ela ionships ha could ha e appea ed o influence he wo k epo ed in his pape . Acknowledgemen This wo k was suppo ed by he RegionalGo e nmen o Cas illa y León and he EU-FEDER (CLU 2017-09, CL-EI-2021-07). Re e ences An, M., He, W., Dege u, D.M., Liao, Z., Zhang, Z., Yuan, L., 2018. Spa ial pa e ns o u ban was ewa e discha ge and ea men plan s e ficiency in China. In . J. En i on. Res. Pub- lic Heal h 15, 1892. Bodik, I., Kubaska, M., 2013. Ene gy and sus ainabili y o ope a ion o a was ewa e ea men plan . En i on. P o . Eng. 39 (2), 15–24. Ca doso, B.J., Rod igues, E., Gaspa , A.R., Gomes, Á., 2021. Ene gy pe o mance ac o s in was ewa e ea men plan s: a e iew. J. Clean. P od. 322, 129107. Ca alho, P., Ma ques, R.C., Be g, S., 2012. A me a- eg ession analysis o benchma king s ud- ies on wa e u ili ies ma ke s uc u e. U il. Policy 21, 40–49. Cas elle -Viciano, L., He nández-Cho e , V., He nández-Sancho, F., 2018. Modelling he en- e gy cos s o he was ewa e ea men p ocess: he influence o he aging ac o . Sci. To al En i on. 625, 363–372. CNE (Nacional Ene gy Commission), 2020. Annual ene gy s a is ics epo . A ailable a : h ps://www.cne.cl/wp-con en /uploads/2021/12/Anua ioCNE2020.pd . Coelli, T.J., Rao, D.S.P asada, O’Donnell, C.J., Ba ese, G.E., 2005. An In oduc ion o E fi- ciency and P oduc i i y Analysis. Kluwe Academic Publishe s, Bos on. De Wi e, K., Ma ques, R.C., 2010a. Influen ial obse a ions in on ie models, a obus non- o ien ed app oach o he wa e sec o . Ann. Ope . Res. 181 (1), 377–392. De Wi e, K., Ma ques, R.C., 2010b. Designing pe o mance incen i es, an in e na ional benchma k s udy in he wa e sec o . CEJOR 18 (2), 189–220. Dong, X., Zhang, X., Zeng, S., 2017. Measu ing and explaining eco-e ficiencies o was ewa e ea men plan s in China: an unce ain y analysis pe spec i e. Wa e Res. 112, 195–207. Es e e, M., Apa icio, J., Rod iguez-Sala, J.J., Zhu, J., 2022. Random o es s and he measu emen o supe e ficiency in he con ex o ee disposal Hull. Eu . J. Ope . Res. 304 (2), 729–744. Es e e, M., Apa icio, J., Rabasa, A., Rod iguez-Sala, J.J., 2020. E ficiency analysis ees: a new me hodology o es ima ing p oduc ion on ie s h ough decision ees. Expe Sys . Appl. 162, 113783. Es e e, M., Rod iguez-Sala, J.J., Lopez-Espin, J.J., Apa icio, J., 2021. Heu is ic and back acking algo i hms o imp o ing he pe o mance o e ficiency analysis ees. IEEE Access 9, 17421–17428. Feng, Z., Liu, X., Wang, L., Wang, Y., Yang, J., Wang, Y., Huan, Y., Liang, T., Yu, Q.J., 2022. Comp ehensi e e ficiency e alua ion o was ewa e ea men plan s in No heas Qinghai-Tibe pla eau using slack–based da a en elopmen analysis. En i on. Pollu . 311, 120008. Ganguly, R., Dewan, H., 2020. Applica ion o decision making ool o de e mine e fluen qual- i y index o exis ing sewage ea men plan s. J. Ins . Eng. (India): Se . A 101 (1), 207–219. Fig. 8. Ene gy e ficiency sco e o each WWTP and olume o was ewa e ea ed. A. Mazio is e al. Science o he To al En i onmen 885 (2023) 163539 9