Ad

La

Cb

a

ARRAA

KDEECEG

1

aaeriEpqc

rsetel1

h2

Sustainable Cities and Society 23 (2016) 23–36

Contents lists available at ScienceDirect

Sustainable Cities and Society

jou rna l h om epage: www.elsev ier .com/ locate /scs

nalysis of the electricity consumptions: A first step to develop aistrict cooling system

uca Pampuria, Nerio Cereghetti a, Davide Strepparavaa, Paola Caputob,∗

Istituto di Sostenibilità Applicata All’ambiente Costruito, Scuola Universitaria Professionale della Svizzera Italiana, Campus Trevano, Via Trevano,H-6952 Canobbio, SwitzerlandArchitecture, Built Environment and Construction Engineering Department, Politecnico di Milano, Via Bonardi 9, 20133 Milano, Italy

r t i c l e i n f o

rticle history:eceived 23 December 2015eceived in revised form 17 February 2016ccepted 18 February 2016vailable online 26 February 2016

eywords:istrict cooling

a b s t r a c t

Space cooling represents an increasingly and often understated important energy demand even in moder-ate climates. Usually space cooling is provided at building level by electric driven appliances. This impliesseveral problems in summer due to the peak of electricity consumption.

Analytical methods for investigating energy consumption due to space cooling demand at urban orregional level are needed. The research here reported is focused on the territory of southern CantonTicino, in Switzerland, and is based on real data provided by the local electricity company. The researchinvestigates the electricity consumptions of big users in order to verify if there is a significant cooling

lectricity consumptionnergy consumerslusteringnergy signatureIS

demand, how this demand affects electricity consumptions and if this demand can be satisfied by districtcooling (DC). The possible DC connections were selected by a defined procedure and mapped by GIS, aswell as the density of the electricity consumption and the peak power. Three main areas suitable for DCwere identified. The analysis demonstrates that DC could represent an alternative to electric driven airconditioning systems, with benefits for the consumers, the utilities and the environment.

© 2016 Elsevier Ltd. All rights reserved.

. Introduction

The path to carbon neutral communities represent long lastingnd urgent challenges for all governments. For example, the EU haspproved the “Roadmap for moving to a competitive low carbonconomy in 2050” (EU COM112/2011, 2011) with the objective toeduce greenhouse gas emissions by 80–95% by 2050 in compar-son to those of 1990. Analogously Switzerland has approved thenergy Strategy 2050 that includes the main aims for the energyolicy towards 2050 (BFE, 2015). Among these aims, as conse-uences of abandoning nuclear power, a reduction of the electricityonsumption is required.

Energy statistics show that European and Swiss buildings areesponsible of more than the 40% of the total final energy con-umption (Odyssee, 2012; BFE, 2015), representing an importantnergy saving opportunity. More in details, energy statistics showhat buildings space heating accounts for 68% and 71% of end-use

nergy consumption respectively in Europe and Switzerland, whileighting and electrical appliances account for 15%, water heating for2% and cooking for 4% (Odyssee, 2012; BFE, 2015).

∗ Corresponding author. Tel.: +39 0223999488; fax: +39 0223999484.E-mail address: paola.caputo@polimi.it (P. Caputo).

ttp://dx.doi.org/10.1016/j.scs.2016.02.015210-6707/© 2016 Elsevier Ltd. All rights reserved.

Despite the energy efficiency measures developed in Europe andin Switzerland, the global energy consumption in buildings con-tinues to rise. In particular, considering the focus of the presentpaper, BFE (2015) stated that energy consumption related to airconditioning had increased of 39% from 2000 to 2013. Many author-itative researches have investigated space cooling in terms ofpredicting cooling loads in buildings, but a significant level ofcomplexity is recognized in this framework as reported also inDamnu, Wanga, Zhai, and Li (2013). The present research sug-gests another approach, that starts from the available electricityconsumption, since space cooling demand is usually provided bycompression chillers. The increasing of the electricity demand dueto space cooling is in contradiction with the defined energy targets,as empathized also by the media during the hot summer of 2015(IEA, 2015).

In general, it is difficult to obtain reliable data about electric-ity consumption for a large number of users (Caputo, Costa, &Zanotto, 2013). Furthermore, with few exceptions, these data areavailable as a whole, without the details of electricity consump-tion for cooling demand. The evaluation of the summer surplus due

to space cooling is difficult because often data do not cover a fullyear and significant sample of users. For these reasons, a methodol-ogy for analyzing this hidden energy demand and for studying howit can be satisfied without affecting electricity consumptions was

24 L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36

F AIL (ifi

dd(rcvemecbtttc

2d

adeetbc

tvctsc

dc

ig. 1. Territorial context of the study (left) and map of municipalities supplied bygure legend, the reader is referred to the web version of this article.)

eveloped. The present paper reports the results of a research con-ucted on the territory of southern Canton Ticino, in SwitzerlandFig. 1). Even though this territory is small, it can be considered rep-esentative of a large arear with similar climatic and morphologicalonditions. The research starts from the analysis of real data pro-ided by the local electricity company (AIL SA, 2015) that supplieslectricity to most of southern Canton Ticino (Fig. 1). Following theethodology reported in Section 3, the research investigates the

lectricity consumptions of big users in order to verify how spaceooling affects electricity consumptions and if it can be satisfiedy district cooling (DC). The analysis demonstrates that, despitehe temperate climatic conditions, there is a significant consump-ion of electricity due to the space cooling demand, especially inhe tertiary sector and that DC can be an efficient alternative toonventional cooling devices.

. District cooling as an alternative to conventional coolingevices

El Barky and Dyrelund (2013) report an interesting summarybout the advantages of moving from conventional space coolingevices, chillers operating at building level, to DC. The operation oflectric chillers implies large investments in power peak plants andven for blackouts in many cities around the world. A light reduc-ion of the peak demand can be achieved in high thermal capacityuildings, by other measures and users behavior, not always appli-able.

In case of proper morphologic, climatic and density features ofhe built environment, DC can be adopted as an alternative to con-entional cooling devices. As well explained by El Barky (2013), DCan bring advantages similar to those of district heating (DH) dueo centralized production (i.e. risk reduction for the individual con-umers and cost effectiveness for the energy supply companies). DC

an contribute to use local sources that otherwise would be wasted.

Euroheat and Power (2006) reports that DC is normally pro-uced by means of one or more of the following technologies thatan be also combined:

n red) in Canton Ticino (right). (For interpretation of the references to color in this

• conventional compressor-based cooling;• absorption cooling, transforming waste heat into cooling;• free cooling, using cool ambient air or cool water from the ocean,

lake or river.

Euroheat and Power (2006) affirms that a DC system can reachefficiencies typically 5 or even 10 times higher than the efficienciesof electric chillers, as it is shown also in Table 1, where primaryresource factors (PRF) are adopted. Primary energy refers to energythat has not been subjected to any conversion or transformationprocess (e.g. oil in the oil fields). Primary energy may be resourceenergy or renewable energy or a combination of both. Resourcerefers to a source depleted by extraction (e.g. fossil fuels) andrenewable energy to a source that is not depleted by extraction(e.g. biomass, solar). The use of the primary resource factor (PRF)enables to measure the savings and losses occurring from energygeneration to the delivery to the building. The primary resourcefactor expresses the ratio of the non-regenerative resource energyrequired for the building to the final energy supplied to the build-ing. The primary resource factor represents the energy delivery butexcludes the renewable energy component of primary energy.

Since southern Canton Ticino has several natural sources suit-able for DC such as the Lake of Lugano and several local rivers, freecooling (i.e. the adoption of cold water for cooling purposes) couldbe the most appropriate technology in this case.

2.1. District cooling potential in Europe and Switzerland

DC still represents an uncommon technology for providing cool-ing. For example, the current DC market share represents only 1–2%of the cooling market in Europe and DC is almost not adopted inSwitzerland. Despite of this, there are interesting examples in Asianand European cities (IEA, 2015; Xiang-li, Lin, & Hai-wen, 2010), i.e.

Stockholm, Helsinki etc.

Euroheat and Power (2006) estimated that the European cool-ing market would be around 660 TWh/y in the period 2012 to2020. This corresponds to an electricity consumption of 260 TWh,

L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36 25

Table 1Performance of different cooling solutions (Euroheat and Power, 2006).

Solutions EER PRF

Conventional room and central air conditioning 1.5–3.5 1.7–0.7

District cooling solutionsIndustrial chillers with efficient condenser cooling and/or recovered heat to district heating 5–8 0.5–0.3Free cooling/industrial chillers 8–25 0.3–0.1Free cooling and cooling spills 25–40 0.1–0.06Absorption chiller driven from heat from waste or renewable sources 20–35 0.13–0.07

E loitabP

cfe8sSacweAie(ssgstM

da

2

st

nrZ

os

D

•

•

•

ER: (Seasonal System) Energy Efficiency Ratio. This states the output of yearly expRF: Primary Resource Factor.

onsidering a seasonal system energy efficiency ratio of 2.5. There-ore, according to Eurostat (2015), it is possible to summarize thatlectricity consumed due to cooling needs represents about the% of the yearly electricity generation in EU. These data demon-trate the large potential of penetration for DC in Europe and inwitzerland, a small country in the European Alpine region. Sinceverage temperatures in Switzerland are likely to increase due tolimate change, it is expected that energy demand for space heatingill decrease while space cooling, which is currently almost non-

xistent in private households in Switzerland, is likely to increase.s recommended by Christenson, Manz, and Gyalistras (2006) and

n CH2014-Impacts (2014), this phenomenon should not be under-stimated. This last study reports that Swiss heating degree daysHDD) are going to decrease from 5% to 20% in 2050 in compari-on to the reference period (1980–2009). Conversely, in the sametudy it is projected that Swiss cooling degree days (CDD) areoing to increase from 100% to 400% in 2050 in comparison to theame reference period. Analogous considerations are reported inhe framework of other interesting researches in Europe (Taseska,

arkovska, & Callaway, 2012; Berger et al., 2014).For these reasons studies about how the increasing cooling

emand can be satisfied reducing its impact on the energy systemsre needed.

.2. Main benefits of district cooling

The phenomenon of electric summer peak due to cooling istressed also by Jing, Jiang, Wu, Tang, and Hua (2014), underlininghe high operating costs.

Interesting evaluations about benefits related to DC as an alter-ative to building size electric driven air conditioning system areeported in Bjerregaard (2013), De Carli, Galgaro, Pasqualetto, andarrella (2014) and Gang, Wang, Gao, and Xiao (2015).

As main benefit, DC can contribute to smooth the demand curvef electricity, especially if free cooling, such as rivers and lakes, orurplus heat through absorption chillers were adopted.

El Barky (2013) reports an interesting summary of benefits ofC on users, utility companies and communities as follows:

benefits for users: cost reduction (in particular in case of freecooling or use of waste heat), no noise from chillers, reduced riskreparations, better certification of the energy performance of thebuilding and available space on rooftop and basement for otherpurposes;benefits for utility companies: a new profitable market, a newproduct and service in portfolio, synergies between power, heat

and cooling production, especially in case of distributed genera-tion;benefits for communities: environmental and energy improve-ments related to a more efficient use of the resources, possibilityof creating synergies in the energy and infrastructural planningof their territory.

le cooling energy per unit of yearly electrical energy input in the system.

On the other hand, the impact of DC on local infrastructures andfinance is very important. DC requires significant investments, astable policy framework, and collaboration between various stake-holders. This means that DC should be promoted in the frameworkof appropriate conditions. In this case, DC can become a very impor-tant tool toward improving energy efficiency, reducing greenhousegases emissions and renewable energies integration (Bjerregaard,2013).

3. Methodology

In the previous sections, it is stated that the cooling demand isconstantly increasing and that it strongly affects the consumptionof electricity. Moreover, it is stated that electricity consumptiondue to cooling deserves to be deeply analyzed. This paper reportsa method for exploring cooling demand of big electricity users atregional scale, southern Canton Ticino, and focuses on DC potentialin the same region.

The case study is very interesting because it originates froma research supported by the local electricity company (AIL SA,2015) that supplies electricity to most of southern Canton Ticino,as reported in Section 1 and Fig. 1. This case attests the inter-est of energy utilities in DC underlining the innovation and thereplicability of the methodology. Furthermore, it represents a pre-cious occasion for managing real data and for foreseeing potentialsynergies among DC, DH and distribute generation. To develop ahypothetical DC project, it is fundamental to map the potentialcustomers in a certain area. Therefore, since the beginning of theanalysis, the chosen methodology includes an important use of GIS.GIS supported tools are very attractive for energy utilities becauseallow a quick overview of the potential operations in a definedterritory.

3.1. Datasets and selection of the users

AIL SA provided a geo-referred dataset of the electricity con-sumptions among all the users of the served territory. As bigelectricity users, only consumers with more than 100 MWh/y weretaken into account (AIL, 2014a). In general, they are tertiary orindustrial users. Additionally, other two available databases wereconsidered: the federal register of buildings (RBD, 2014) and thelist of the users connected to the water grid for industrial uses,provided in AIL (2014b).

AIL (2014a) reports the list of users with a yearly electricityconsumption bigger than 100 MWh, with the address of the users.The electric consumptions are recorded every 15 min. The datacollected demonstrate that there are 957 users with an electricconsumption over 100 MWh/y. This corresponds to a total con-sumption of 487,500 MWh/y.

In the RBD (2014) are listed the most important data for describ-ing buildings and dwellings in addition to their address.

AIL (2014b) reports data about water taken from the lakeof Lugano and supplied for industrial uses. Approximately

2 Cities

6au

ti

u

cut

2ice

ubce

eeTtiscpww

tffe

3

aItdpopirt2ms

t

actssc

lowing step of the research regards the definition of the “coolingprofile”1 (CP) of each user, including the calculation of the elec-tricity consumed and of the maximum power due to space cooling

6 L. Pampuri et al. / Sustainable

million m3 per year of water are extracted at a temperature ofbout 7–9 ◦C. The data include technical information about thesers (addresses, type of uses, etc.).

The first step of the research regarded the identification of usershat surely present a significant cooling demand and their location,n order to verify the feasibility of a DC system.

Data were elaborated in order to carry out a classification of thesers based on their activities (mainly of the tertiary sector).

First, some big users were excluded from the list of hypotheticalooling consumers because of their activity (if the activity of theser implies a negligible cooling demand) or because of inconsis-encies or lacks encountered during the data elaboration.

In particular, the users related to the following activities (AIL,014a), were excluded: aqueducts, temporary connections, breed-

ng and stables, gas cabinets, construction sites, churches andonvents, docks, private garages, plants for water treatment, incin-rators, landfills and public lighting.

Second, the users connected to the water grid for industrialses were excluded from the list of hypothetical cooling consumersecause it is conceivable to assume that they already satisfy theirooling demand using industrial water, since this service is costffective.

Electricity data of the users that pass the selection were thenlaborated. In particular, in order to investigate cooling demands,lectricity data were analyzed referring to outside temperature.he period from April to September was considered because inhese months the heating demand in Ticino is negligible, so, evenf there were electric driven heating systems, the electricity con-umption would not be affected by heating. Hypothetical spaceooling users should have electric consumptions that increase pro-ortionally when the outdoor temperature rises. Therefore, usersith electric consumptions not depending on outside temperatureere excluded from the analysis, as explained in Section 3.2.

Finally, through the support of GIS, the farthest users (far morehan 500 m from other four potential connections) were excludedrom the analysis because their connection to a DC system is noteasible due to the long distance from the hypothetical grid, asxplained in Section 3.3.

The process of the selection is shown in Fig. 2.

.2. Energy accountability and external temperature

A precious help in defining how electricity consumptions areffected by cooling needs is represented by energy accountability.n Switzerland, energy accountability is considered as an impor-ant tool for analyzing, monitoring and controlling the energyemands in buildings and the related economic impacts. The mainurpose of energy accountability is the continuous monitoringf energy fluxes and related costs. This approach also permits tolan energy retrofit interventions and other energy management

mprovements. Energy accountability includes also interesting rep-esentations of data by schemes and graphs. Among these, there ishe useful representation of the energy signature (Zhao & Magoulès,012), a graphs that reports on the abscissas the outdoor weeklyean temperatures and on the ordinates the weekly energy con-

umptions (heat or power).The energy signature of a building can be compared with defined

ypical curves, as reported in Fig. 3.In particular, type 1 means that energy consumptions are not

ffected by the trend of the outdoor temperature; type 2 indi-ates that energy consumptions linearly increase with the outdoor

emperature during all the year; type 3 indicates that energy con-umptions linearly increase with the outdoor temperature only inummer season (electric driven cooling system) and type 4 indi-ates that energy consumptions linearly increase when the outdoor

and Society 23 (2016) 23–36

temperature drastically decrease in winter season (electric drivenheating system).

In the analyzed case study, the energy signatures were providedas a graph representing the daily mean power in function of thedaily mean temperature. At the end of this phase of analysis, thefollowing typologies of users were founded:

• users without heating and cooling demands provided by elec-tric driven systems. In this case, electricity consumptions aremeanly uniform throughout the year. These users were obviouslyexcluded from the further phases of the analysis;

• users with cooling demand provided by electric driven systems.In this case, electricity consumptions increase during the summerseason;

• users with heating and cooling demand provided by electricdriven systems. In this case, electricity consumptions increaseduring the winter and the summer season.

The last two typologies, exemplified in Fig. 4, were consideredin the further steps of the research.

3.3. Map of the electricity consumers with cooling demands

At the end of the data elaboration process described in Sections3.1 and 3.2, it was possible to define the final list of users withspace cooling demand and to provide a map of their location byGIS. This is reported in Fig. 5, where the green dots represent userswith cooling demand, the red dots represent users without coolingdemand and the yellow dots represent users that have to be furtherinvestigated. This methodological step is fundament in planning DCsystems as confirmed also by the technical literature. For example,Euroheat and Power (2006) stated that in order to identify the mostoptimal location of a new DC plant within a certain area, it can bevaluable to use an automated mapping tool. The estimated coolingdemand and geographical location of each building in a defined areapermit to calculate also the economic difference between a con-ventional cooling system and a DC system. This operation includesa classification of building size, cooling demand and building type.Moreover, a careful consideration of the local natural resources canreally increase both profitability and energy efficiency.

The GIS map of the locations of the potential cooling user inthe analyzed territory (Fig. 5) permits also to define a geographiccriterion for the selection of users suitable for DC systems. This cri-terion serves to detect a minimum space cooling density in orderto define where the DC grid is technically and economically feasi-ble. Considering the characteristics of the climate and of the builtenvironment, the following criterion was adopted: each user can beconsidered as a hypothetical connection of the DC system if thereare at least other four users within a distance of 500 m as reportedin Fig. 6, with grey buffers. This preliminary criterion can be mod-ified or improved in the following phases of developments of theresearch, as reported in Section 4.

3.4. Definition of cooling profiles

At the end of the data elaboration described in Sections 3.1, 3.2and 3.3, it was possible to edit the final map of users for which spacecooling is supposedly provided by electric driven systems. The fol-

demand.

1 The profile of the electricity adopted for space cooling.

L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36 27

Selec�on of users wit h remote me tering Data clea ning

Selec�on of suitable users depending on

their t ypology

Elimina� on of users without energy data

Elimina� on of users conn ected to ind ustrial

water net

Eli mina� on of users whose energy

consu mp�on do es no t depend on ou tdo or

tempera ture

GIS ma p of suitable users

Calcula�on an d GIS map of cooling energy and

power

Calcula�on an d GIS map of possible DC ar eas

Fig. 2. Scheme of the selection of the electricity consumers potentially interested in DC.

s of en

wSidtw

u

•

•

•

3

(

Fig. 3. Typologie

First, for each user the electricity data provided by AIL (2014a)ere analyzed taking into account the period from April to

eptember 2013. As explained in Section 3.1, during this periodn Ticino heating demand is negligible, so the clustering process,escribed in the following, is not affected by consumptions relatedo heating. Conversely in the same period it is possible to have daysith or without cooling demand.

The methodology for cooling profiles (CPs) calculation of a singleser is constituted by the sequence of the following steps:

Data clustering: electricity consumption data are analyzed toobtain meaningful sets of days;Clusters grouping: clusters obtained in the step 1 are groupedconsidering their distribution over the days of the week and theperiod;CPs calculation: the cooling profiles are calculated using the clus-ters groups defined in the previous step.

.4.1. Data clusteringFor each user the electricity data in the period April–September

182 days) are analyzed with a resolution of 15 min (96 values per

ergy signatures.

day). The basic idea of this step is to separate days in sets relatedto cooling demand and others without this contribution.

Data clustering is performed using K-means algorithm (Arthur& Vassilvitskii, 2007). This algorithm converges to a good solution(not necessarily the optimum) given a number of clusters (k). Thealgorithm output is constituted by k clusters, each of them definedby the belonging days and a centroid as mean reference. In this case,a centroid is a 24-h profile with 96 values. The generic centroid’selement i is the mean of the corresponding i values related to thecluster’s days.

Therefore to the fact that k is an input, the followed strategywas to develop an algorithm that performs K-means more timeswith k higher and higher and then to choose the best solution. Toimplement this iterative approach a MATLAB® script (scout) wasdeveloped. The following table reports the input parameters of thescript.

Scout permits to perform K-means for (maxNumClusters-

minNumClusters) times, with k starting from minNumClusters untilmaxNumClusters, discard solutions with clusters related to popula-tions smaller than minPopCluster and choose the best result in theremaining subset. The last choice is actuated considering as best

28 L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36

ure: el

st

5iit

u

3

sCt

Fig. 4. Examples of energy signatures that depend on outdoor temperat

olution the one minimizing the standard deviations’ sum relatedo clustered data.

Each K-means running is performed with a replication equal to0 to avoid the risk of local minima. As said before, the cluster-

ng algorithm can converge to a local minimum, depending on thenitial conditions. The algorithm is launched more times by replica-ions and the optimum solution can then be obtained reasonably.

Fig. 7 shows the centroids found by scout algorithm for a genericser. In this case the solution is represented by six clusters.

.4.2. Clusters grouping

Once performed the clustering and calculated the centroids, the

econd step to obtain CPs is to find associations between clusters.onsidering Fig. 8 is reasonable to divide them into the followinghree groups:

ectricity for cooling and heating (above); electricity for cooling (below).

• G1, composed by centroids 2, 4 and 6. It shows a schedule ofoperation from 8 a.m. to 7 p.m. step up approximately;

• G2, constituted by centroids 3 and 5. It shows a schedule of oper-ation from 8 a.m. to 10 p.m. approximately;

• G3, with centroid 1. It shows a schedule of operation is stableapproximately constant all over the day.

Then the distribution of clusters elements is analyzed takinginto account the days of the week (Fig. 8) and the daily average ofLugano’s temperature during the period April–September (Fig. 9).

Fig. 8 shows how G2 is strongly related to Thursdays, G3 to

Sundays and G1 to the other cases. Therefore, it is reasonable toestablish two types of CP, one relating to the clusters G1 and anotherconcerning the clusters G2. No CP is considered for G3, being con-stituted only by a centroid.

L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36 29

Fig. 5. Location of the potential DC users (green dots: users with cooling demand; red dots: users without cooling demand; yellow dots: users which have to be furtherinvestigated). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 6. Analysis of the density of the cooling demand in relation to a buffers of 500 (grey circles around each user).

30 L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36

cluste

iid

3

bte

C

C

C

ort

tdtc

(f

E

P

wp

b

ts

Fig. 7. Centroids related to a

On the other hand, Fig. 9 shows that G1 centroids are distributedn cool (cluster 6), warm (cluster 2) and hot (cluster 4) days. A sim-lar result regards G2 centroids, with cluster 5 related to warm/hotays and cluster 3 to the cool ones.

.4.3. Cooling profiles calculationThe third step is to calculate the cooling profiles as the difference

etween centroids related to hot/warm and cool days, consideringhe weekdays grouping described in Section 3.4.2. The followingquations report the CPs definitions for the solution shown in Fig. 7.

P1 (t) = Centroid4 (t) − Centroid6 (t) (1)

P2 (t) = Centroid2 (t) − Centroid6 (t) (2)

P3 (t) = Centroid5 (t) − Centroid3 (t) (3)

CP1 is obtained by clusters 4 and 6 and is related to hot daysf the week except Thursdays and Saturdays. CP2 is similar, butelated to warm days. Instead, the third cooling profile is associatedo hot Thursdays.

It is important to consider that clusters 6 and 3 can be seen ashe minimum bands of power consumption, related to all types ofemand except cooling (i.e. lighting and appliances). Indeed thesewo clusters have been obtained considering days with negligibleooling demand (Fig. 9).

For each CP the daily electric consumption for cooling purposesEcool) and the related peak power (Pmax) are calculated with theollowing equations:

cool =n∑

i=1

⎡⎣Pi ×

96∫

t=1

CPi (t) dt

⎤⎦ (4)

max = max (CPi (t)) (5)

here Pi is the number of hot/warm days related to coolingrofile i.

For example, considering CP1 described by Eq. (1), P1 is the num-er of days belonging to cluster 4.

It is important to remark that this paper shows a first approacho the cooling profiles estimation. Especially the next two stepshould follow this study to have an exhaustive analysis:

ring solution for a single use.

• The complete automation of CPs calculation has to be imple-mented, reasonably focusing the methodology on a smaller setof users and without considering their geographical distribution.Actually the methodologies explained in Section 3.4.1 and 3.4.3are automatic, but not clusters grouping (Section 3.4.2);

• Once the previous automation is complete, an uncertainty anal-ysis has to be actuated for errors estimation.

4. Results

Section 3 describes the procedure aimed at defining the electric-ity consumption and peak power due to space cooling requirementof big users in the territory of Canton Ticino served by AIL SA.This important result has to be integrated by other elaborationsnecessary to study the feasibility of a DC grid. Powers and ener-gies described in Section 3 were represented by a GIS map; thentheir density was calculated as energy and power per surface unit,considering a grid of one hectare. This elaboration permits to havean idea of the area in which DC could actually be a promising andeffective technology.

The authors have verified the technical literature in order to findbenchmarks useful in this phase of the analysis, but they becameaware that the data founded were not directly usable. The consultedreferences are mainly related to analogous criteria for planningDH grids (Aste, Buzzetti, & Caputo, 2015; Brum, Erickson, Jenkins,& Kornbluth, 2015) that are more common than DC grids, espe-cially in Europe. Based on an authoritative study provided by theSwiss Federal Office for Energy (Oppermann, Gutzwiller, & Müller,2010), a space heating density of 30,000 MWh/y per square kilo-meter is considered as standard value for developing DH grids inthe context of Canton Ticino. On the contrary, it is difficult to defineanalogous standards for DC. The technical literature related to DCis mainly focused on networks design (Xiang-li et al., 2010), ratherthan on standards related to space cooling density and size of theglobal cooling needs. As an example, Euroheat and Power (2006)reports that statistically district cooling has been introduced downto 0.5 kW per meter of grid of distribution, i.e. in larger and middle

size cities, within expansion areas such as retail and business parks,hospitals, airports, etc.

The authors have stated from the beginning that this paperfocuses on the first steps required to develop a DC system, starting

L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36 31

ped by days of the week.

ffiotltoat

oSfr

Table 2Input parameters of scout MATLAB® script.

Parameter Description Meaningful values

minNumClusters Minimum number of clusters 2maxNumClusters Maximum number of clusters 6–8minPopCluster Minimum population of a

cluster10–15 days

Replications Number of K-means 50

Fig. 8. Clusters grou

rom the analysis of the electricity consumptions. The transitionrom the electricity consumptions for space cooling to the cool-ng demands implies the knowledge of the cooling systems andf the equipment installed into the buildings. Furthermore, theransition from the cooling demands to the cooling power for gridength unit implies, at least, a preliminary design of the layout ofhe DC grid. Since these activities will regard the following devel-pments of this study, it was not possible to take into accountt this stage of the research the benchmarks previously men-ioned.

At the end of the procedure described in Section 3, only 238

f the 957 big electricity consumers originally provided by AILA were selected as users with a space cooling demand suitableor DC as reported in Table 2. This result confirms that an accu-ate analysis is needed in order to select correctly the potential

Fig. 9. Clusters grouped by different

replications to avoid localminima

electric driven cooling users. The table shows the predominanceof the tertiary sector. The calculated total electricity consump-tion and peak power are 9,070 MWh/y and 10.4 MW, respectively,which confirms a great potentiality for DC. In fact, at least for the

types of external temperature.

32 L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36

F (red dt .)

2iaieodtacc

wst

fri

ig. 10. Representation of the density of electricity consumptions for space coolingo color in this figure legend, the reader is referred to the web version of this article

38 selected consumers, the electric consumption for space cool-ng purposes represents respectively the 7% of the total electricitynd the 9% of the total electricity during the day hours (exclud-ng night hours). Although information about cooling systems andquipment installed into the buildings are not available at this stagef the research, a typical value of seasonal efficiency of the electricriven air conditioning systems (i.e. equal to 3) was adopted2. Byhis mean seasonal coefficient of performance a cooling demand ofbout 27,000 MWh/y and a peak of about 30 MW were estimated,onfirming the significance of space cooling needs in relation to thelimatic features and the extension of the considered territory.

The location of the final 238 users is reported in Fig. 10 together

ith the representation of the density of the electricity needed for

pace cooling in MWh/y per hectare. Analogously Fig. 11 reportsheir location together with the representation of the peak power

2 Based on the expertise of the authors, this can be assumed as a mean valueor the context of Canton Ticino. In Section 2.1 a lower value (i.e. equal to 2.5) iseported. This value was lightly increased (to 3) in order to consider the technologicmprovements of the last years in air conditioning systems.

ots represent the 238 selected cooling users). (For interpretation of the references

needed for space cooling in kW per hectare. The main data of thepotential DC users are reported in Table 3.

4.1. Identification of three areas suitable for DC

Information provided by Figs. 10 and 11 constitutes an impor-tant basis to define areas suitable for the development of plansrelated to DC systems. The areas with the highest density were fur-ther outlined. This information was integrated by other data relatedto the characteristics of the territory, including features of the nat-ural and built environment. This is a delicate stage and implies adeep knowledge of the territory, of its characteristics and of thecontext of operation. At the end, three main areas suitable for DCwere identified (namely Grancia, Lugano Paradiso, Piana Vedeggio),as reported in Fig. 12. These three areas present densities between10 and 20 MWh/y per hectare and precisely: Grancia 18 MWh/y per

hectare; Lugano Paradiso 14 MWh/y per hectare and Piana Vedeg-gio 12 MWh/y per hectare, on average.

Grancia has a high density but space cooling demand is almostall concentered in an important commercial center. In this case,

L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36 33

F ts repi

ff

ntoa

VpeV

TU

ig. 11. Representation of the density of electricity power for space cooling (red don this figure legend, the reader is referred to the web version of this article.)

urther investigations are needed in order to accurately verify theeasibility of a DC grid.

The area of Lugano Paradiso is interesting because it is locatedear the net for the distribution of industrial water. Further inves-igations are needed in order to understand the actual potentialityf DC in this area. As a disadvantage, this area includes many usersnd logistic difficulties related to morphology and mobility.

Despite it has the lowest cooling density of the three, Piana

edeggio represents the most suitable area for DC. In this area aroject of DH has been approved; this could bring important syn-rgies toward the realization of DC. Further, the area is near toedeggio River and Lugano Lake that are both interesting sources

able 3sers selected as suitable for DC: typology, number, yearly electricity consumption and p

Category Number Electr

Engineering offices 1 32Camping 1 29Public buildings 1 9

resent the 238 selected cooling users). (For interpretation of the references to color

for DC. Furthermore, in this area there is an important energydemand related to refrigeration, mainly for the operation of phar-maceutical industries. Despite the present research is focused onspace cooling and does not include the analysis of refrigeration, thispoint represents another important synergy toward the develop-ment of DC. In fact, it is estimated that refrigeration implies a furtherelectricity consumption of about 5,000 MWh/y with an averagedensity of 38 MWh/y per hectare. This potential can be added to

that related to space cooling as previously described and reportedin Figs. 10–12.

In all three areas, however, it deserves to be pointed out that theanalysis here reported did not take into account users that are not

eak power for space cooling.

icity for cooling [MWh/y] Max power for cooling [kW]

.4 23.5

.5 69.5 23.6

34 L. Pampuri et al. / Sustainable Cities and Society 23 (2016) 23–36

Table 3 (Continued)

Category Number Electricity for cooling [MWh/y] Max power for cooling [kW]

Surgeries 1 8.6 7Airports 1 5.5 7.4Customs 1 2.6 8.4Schools 2 43.6 42.6Bakeries 2 29.4 58.1Utility companies 2 21.1 23.9Switchboards 3 276.2 160.4Sport centers 4 84.7 135.6Hotels closed in winter 7 208.1 275.7Hospitals 8 1,353.6 1,200.6Restaurants 11 155.8 177.7Hotels 12 1,026.1 665.7Pharmaceutical industries 12 280.5 552.1Banks 16 667.5 734.1Automotive mechanic workshops 16 131.2 175.8Shops 17 763.8 1,174.9Industries 18 771.8 967.9Alimentary shops 32 890.8 978.1Commercial offices 34 1,139.2 1,527.4Common facilities in residential buildings 36 1,139.1 1,459.0

Total 238 9,070.1 10,448.8

Fig. 12. Representation of the three suitable areas for DC (red dots represent the 238 selected cooling users). (For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

Cities

stociinldd

5

ycbac

sfcMi(teaapco

erwaTac

scacDon(ft

cPo

ctaeiicrr

L. Pampuri et al. / Sustainable

erved by AIL SA and users that have a yearly electricity consump-ion lower than 100 MWh, as reported in Section 3.1. The inclusionf other users could drastically increase the cooling density andould improve the cost effectiveness of DC. This is particularlymportant because even if space cooling in Swiss residential build-ngs is not a standard application up to now and in many cases notecessary, it is still more and more used especially in highly insu-

ated dwellings, where comfort cooling becomes more importantue to higher thermal loads and rising summer thermal comfortemands (Dott, Wemhöner, & Afjei, 2011).

. Conclusions

Despite the energy efficiency measures developed in the lastears, the energy consumption in buildings continues to rise. Spaceooling represents an increasingly important energy demand in theuilt environment, even in moderate climates. Space cooling is usu-lly provided by electric driven appliances and the related energyonsumption is often hidden and understated.

The novelty of this research lies in the adopted approach thattarts from a large amount of electricity consumption data, suitableor statistical processing, rather than from methods for predictingooling loads in buildings (energy simulations, energy audit etc.).any authoritative researches are available in this field but a signif-

cant level of complexity is recognized. For example, Damnu et al.2013) stated that it is not feasible to apply the traditional predic-ion method to evaluate hourly building cooling load at the urbannergy planning stage because of the limited building informationnd complexity of energy prediction. After describing and testing

simplified prediction model, Damnu et al. (2013) analyze theotential causes of prediction errors and the significance of variousooling load influence factors, as demonstration of the complexityf the issue.

Further researches about methods and tools for investigatingnergy consumption due to space cooling demand at urban oregional level are needed. The research presented in this paperould like to give contributions to this need. The research reports

n interesting case of study in the territory of southern Cantonicino (that has climatic features similar to those of many otherreas in Europe) based on real data provided by the local energyompany.

As main purpose, the research investigates the electricity con-umptions of big users in order to verify if there is a significantooling demand, how this demand affects electricity consumptionsnd if this demand can be satisfied by district cooling. To that end aomplex procedure of selection of users was defined. The possibleC connections were estimated and mapped by GIS. The densityf the electricity consumption and the density of the peak powereeded for space cooling were obtained by statistical elaborationsK-means clustering). The areas with the highest densities wereurther outlined and information about the characteristics of theerritory were included.

Three main areas suitable for DC were identified: namely Gran-ia, Lugano Paradiso, and Piana Vedeggio. In particular, the area ofiana Vedeggio presents many features that suggest the feasibilityf a DC grid.

The analysis demonstrates that, despite the temperate climaticonditions, there is a significant space cooling demand, especially inhe tertiary sector. This demand can be efficiently provided by DC asn alternative to electric driven air conditioning systems, with ben-fits for consumers and utilities. Further investigations are neededn order to explore the evolution of the not negligible and increas-

ng space cooling demand in the residential sector. Other analysisould be performed also to estimate the electricity consumptionselated to refrigeration, in particular for industrial users, and howefrigeration could affect DC.

and Society 23 (2016) 23–36 35

The applied methodology could be easily replicated in other con-texts of Switzerland, Europe etc. In Europe, the recent directives anddeclarations (European Parliament, 2009; European Parliament,2010; European Parliament, 2012; European Commission, 2014)request drastic improvements of the energy systems towards moreefficient paradigms and the same can be stated for Switzerland, asreported by BFE (2015), and in other countries with analogous eco-nomic and technical features. In this framework, national or localauthorities are advised to develop integrated heating and coolingstrategies. Combined district heating and cooling makes more effi-cient and cost effective the use of energy resources (El Barky &Dyrelund, 2013; Ascione, Canelli, De Masi, Sasso, & Vanoli, 2014;Jing et al., 2014).

Local authorities should plan district heating and coolingdevices as an integrated part of the urban infrastructure wheneverit is cost effective. DC should be considered as well as DH in energyplanning, toward the coordination of energy supply and structuraldevelopment of the territory.

District strategies could contribute to create the opportunity tomeet low energy and low carbon standards in a more cost effec-tive way using renewable energy and efficient combined heat andpower via the district heating and cooling grids. DC, combinedwith DH, distribute generation and cogeneration systems with ahigh integration of renewable energies, can represent an impor-tant alternative or a complementary element for guarantying theenergy goals at year 2030 and 2050 with a community approachinstead of a building approach. The building approach is indeedfocused on nZEB standards that are difficult to be rapidly appliedto most of the existing built environment since important barriersare still present (Caputo & Pasetti, 2015; Sesana et al., 2015).

Acknowledgements

Many thanks to AIL SA for supporting the research, for provid-ing all the available data and information and for endorsing themethodological approach.

References

AIL SA. (2015), www.ail.ch.AIL. (2014a). Information about users with more than 100 MW h of electric

consumption, internal report, 2014.AIL. (2014b). Information about users of the industrial water network, internal report,

2014.Arthur, D., & Vassilvitskii, S. (2007). K-means++: The advantages of careful seeding.

In Proceedings of the eighteenth annual ACM-SIAM symposium on discretealgorithms (SODA ‘07). pp. 1027–1035. Philadelphia, PA: Society for Industrialand Applied Mathematics. ISBN: 978-0-898716-24-5 (2007).

Ascione, F., Canelli, M., De Masi, R. F., Sasso, M., & Vanoli, G. P. (2014). Combinedcooling, heating and power for small urban districts: An Italian case-study.Applied Thermal Engineering, 71, 705–713.

Aste, N., Buzzetti, M., & Caputo, P. (2015). District heating in Lombardy Region(Italy): Effects of supporting mechanism. Sustainable Cities and Society, 14,43–55.

Berger, T., Formayer, A. H., Korjenic, A., Pospischal, B., Neururer, C., & Smutny, R.(2014). Impacts of climate change upon cooling and heating energy demand ofoffice buildings in Vienna. Energy and Buildings, 80, 517–530.

BFE. (2015), www.bfe.admin.ch.Bjerregaard, J. (2013). Society’s stake in district cooling. International Magazine on

District Heating and Cooling,. 〈www.dbdh.dk〉 (Journal n 0.3/2013).Brum, M., Erickson, P., Jenkins, B., & Kornbluth, B. (2015). A comparative study of

district and individual energy systems providing electrical-based heating,cooling, and domestic hot water to a low-energy use residential community.Energy and Buildings, 92, 306–312.

Caputo, P., Costa, G., & Zanotto, V. (2013). A methodology for defining electricitydemand in energy simulations referred to the Italian context. Energies, 6(12),6274–6292.

Caputo, P., & Pasetti, G. (2015). Overcoming the inertia of building energy retrofit

at municipal level: The Italian challenge. Sustainable Cities and Society, 15,120–134.

CH2014-Impacts. (2014). Toward Quantitative Scenarios of Climate Change Impactsin Switzerland. pp. 136. Bern, Switzerland: OCCR, FOEN, MeteoSwiss, C2SM,Agroscope, and ProClim.

3 Cities

C

D

D

D

E

E

E

E

E

E

E

6 L. Pampuri et al. / Sustainable

hristenson, M., Manz, H., & Gyalistras, D. (2006). Climate warming impact ondegree-days and building energy demand in Switzerland. Energy Conversionand Management, 47, 671–686.

e Carli, M., Galgaro, A., Pasqualetto, M., & Zarrella, A. (2014). Energetic andeconomic aspects of a heating and cooling district in a mild climate based onclosed loop ground source heat pump. Applied Thermal Engineering, 71,895–904.

ott, R., Wemhöner, C., & Afjei, T. (2011). Heating and cooling with heat pumps inSwiss residential buildings. In Proceedings of the 10th IEA Heat Pump Conference2011, 16–19 May 2011, Tokyo, Japan, 2011.

amnu, L., Wanga, Z., Zhai, Z. J., & Li, X. (2013). A simplified method to predicthourly building cooling load for urban energy planning. Energy and Buildings,58, 281–291.

l Barky, T. K. (2013). The district cooling potential. International Magazine onDistrict Heating and Cooling,. 〈www.dbdh.dk〉 (Journal n 0.1/2013)

l Barky, T. K., & Dyrelund, A. (2013). Why district heating and district cooling gohand in hand. International Magazine on District Heating and Cooling,.〈www.dbdh.dk〉 (Journal n 0.3/2013).

U COM 112. (2011). A Roadmap for moving to a competitive low carbon economy in2050. Brussels: Communication from the Commission to the EuropeanParliament, the Council, the European Economic and Social Committee and theCommittee of the Regions no. 112.

uroheat and Power. (2006). Ecoheatcool work package 5. Possibilities with moredistrict cooling in Europe, 2006.

uropean Commission. (2014). 2030 Climate and energy goals for a competitive,secure and low-carbon EU economy. European Commission. 〈www.europa.eu〉(Published on 22nd January 2014)

uropean Parliament. (2009). Directive 2009/28/EC of the European Parliament and

of the council of the 23 April 2009 on the promotion of the use of energy fromrenewable sources and amending and subsequently repealing Directives2001/77/EC and 2003/30/EC.

uropean Parliament. (2010). Directive 2010/31/EU of the European Parliament andof the Council of 19 May 2010 on the energy performance of buildings (recast).

and Society 23 (2016) 23–36

European Parliament. (2012). Directive 2012/27/EU of the European Parliament andof the Council of 25 October 2012 on energy efficiency, amending Directives2009/125/EC and 2010/30/EU and repealing Directives 2004/8/EC and2006/32/EC.

Eurostat. (2015). Energy balance sheets—2013 data-2015 edition. 〈http://ec.europa.eu/eurostat/web/products-statistical-books/-/KS-EN-15-001〉, Release date:02/07/2015.

Gang, W., Wang, S., Gao, D., & Xiao, F. (2015). Performance assessment of districtcooling systems for a new development district at planning stage. AppliedEnergy, 140, 33–43.

IEA (International Energy Agency). (2015). Implementing agreement on districtheating and cooling including combined heat and power. 〈www.iea-dhc.org〉.

Jing, Z. X., Jiang, X. S., Wu, Q. H., Tang, W. H., & Hua, B. (2014). Modelling andoptimal operation of a small-scale integrated energy based district heating andcooling system. Energy, 73, 399–415.

Odyssee. (2012). Energy Efficiency Trends in Buildings in the EU. Lessons from theODYSSEE MURE project.

Oppermann, G., Gutzwiller, S., & Müller, E. A. (2010). Schlussbericht Projektphase 1,Bundesamt für Energie, Forschungsprogramm 153,994/0040, BFE-Projektnummer102,385.

RBD. (2014). RBD: Federal Register of Buildings and Dwellings (status 31.12.2013),www.bfs.admin.ch/bfs/portal/en/index/themen/00/05/blank/01/03.html.

Sesana, M. M., Cuca, B., Iannaccone, G., Brumana, R., Caccavelli, D., & Gay, C. (2015).Geomapping methodology for the GeoCluster Mapping Tool to assessdeployment potential of technologies for energy efficiency in buildings.Sustainable Cities and Society, 17, 22–34 (2015)

Taseska, V., Markovska, N., & Callaway, J. M. (2012). Evaluation of climate changeimpacts on energy demand. Energy, 48, 88–95.

Zhao, H., & Magoulès, F. (2012). A review on the prediction of building energyconsumption. Renewable and Sustainable Energy Reviews, 16, 3586–3592.

Xiang-li, L., Lin, D., & Hai-wen, S. (2010). Optimal design of district heating andcooling pipe network of seawater-source heat pump. Energy and Buildings, 42,100–104.