FP7-SMARTCITIES-2013(ICT) Objective ICT-2013.6.2 Data Centres in an energy-efficient and environmentally friendly Internet

DC4Cities An environmentally sustainable data centre for Smart Cities

Project Nº 609304

Final Market Analysis D2.4

Responsible: Sonja Klingert [UNIMA] Contributors: Sonja Klingert, Thomas Schulze, Torben Möller, Nicole Wesemeyer [UNIMA]; Wolfgang Duschl, Florian Niedermeier, Hermann de Meer [UNIPASSAU]; Federico Lombart Badal, Francesc Casaus, Eduard Martín [IMI]; Silvia Sanjoaquin Vives, Elena Alonso Pantoja, Gonzalo Jose Diaz Velez [GN]; Mehdi Sheikhalishahi [CREATE-NET]; Frederic Wauters, Maria Perez Ortega, [FM]; Marta Chinnici [ENEA] Document Reference: D2.4 – Final Market Analysis Dissemination Level: PU Version: 3.0 Date: 30 Nov 2015

Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

EXECUTIVE SUMMARY The final market analysis, D2.4, builds on the findings of the first market analysis, D2.2, where regions in Europe offering good starting conditions for marketing a DC4Cities based tool were identified. Now, for the second stage of the analysis, the plan was to evaluate these findings for specific cities, which at the same time play a major role in the data centre industry and are among the most advanced smart cities in Europe. The analysis therefore included Barcelona, as DC4Cities partner, and the data centre “Big Five” cities in Europe: London, Frankfurt, Amsterdam, , and Madrid. The results partly reinforced the conclusions of the first market analysis, but some findings also came as a surprise once urban agglomerations were analysed in more detail: The most thorough analysis could be executed for Barcelona, due to a good data basis as Barcelona partners belong to the consortium of DC4Cities. The result is promising, less for today’s market, because the share of intermittent renewable energy at the local electricity mix is still comparably low. But the prospects for a favourable development are good, both regarding the energy supply side and the data centre market that seems to be offering adequate flexibility. Also, Amsterdam is among the preferred starting regions to market DC4Cities, due to a high level of nearby wind energy and the political will to increase the share of renewable energy. A limitation comes from the side of the data centre industry, which in great parts relies on the colocation business model with limitations to power flexibility. The great surprise is Frankfurt, which although being less favoured with solar irradiation than many southern cities has a strong political commitment to fully rely on renewable energy in 2050. A feasibility study projected a great increase of solar and wind energy – which together with a booming data centre industry offers perfect conditions to market DC4Cities. The limiting factor is again the focus on a colocation business model. Due to a lack of readily available intermittent renewable energy resources and a clear roadmap to achieve a high share of renewable energy in the local electricity mix, London, the hottest data centre market in Europe, does not suggest a positive market outlook. The same applies to Madrid – a surprise, because of an abundance of solar irradiation; but the economic downturn destroyed the local data centre market to a great degree, so that the economic basis is small. Also Paris, even though comparable to Frankfurt regarding solar irradiation, does not offer favourable conditions for DC4Cities, because of an extremely high density of power demand and a political focus of geothermal and solar thermal energy. The most important conclusions from these findings are that DC4Cities should aim at developing a business model targeted at colocation data centres and that the political atmosphere has a great clout on the development of a market for DC4Cities.

ii Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

CONTRIBUTORS TABLE

Document Section Author’s Name(s) Reviewers

SONJA KLINGERT (UNIMA), MEHDI SHEIKHALISHAHI Introduction SILVIA SANJOAQUÍN VIVES (CREATE-NET) (GN)

The European DC MEHDI SHEIKHALISHAHI SONJA KLINGERT (UNIMA) Market (CREATE-NET)

Data Management in TORBEN MÖLLER, NICOLE SILVIA SANJOAQUÍN VIVES European Smart Cities WESEMEYER (UNIMA) (GN)

WOLFGANG DUSCHL, FLORIAN Selected Smart Cities: NIEDERMEIER, HERMANN DE MARTA CHINICCI (ENEA) London MEER (UNIPA)

Selected Smart Cities: SONJA KLINGERT (UNIMA), GONZALO JOSE DIAZ VELEZ Paris FREDERIC WAUTERS (FM) (GN)

Selected Smart Cities: WOLFGANG DUSCHL FREDERIC WAUTERS (FM) Amsterdam (UNIPA)

Selected Smart Cities: SILVIA SANJOAQUÍN VIVES THOMAS SCHULZE (UNIMA) Frankfurt (GN)

SILVIA SANJOAQUÍN VIVES, ELENA ALONSO PANTOJA, Selected Smart Cities: GONZALO JOSE DIAZ VELEZ FREDERIC WAUTERS (FM) Madrid (GN), MARTA CHINNICI (ENEA)

SILVIA SANJOAQUÍN VIVES, ELENA ALONSO PANTOJA, GONZALO JOSE DIAZ VELEZ, Selected Smart Cities: THOMAS SCHULZE ALEJANDRA SAYANS JIMÉNEZ Barcelona (UNIMA) (GN); EDUARD MARTÍN, FRANCESC CASAUS, FEDE LOMBART (IMI)

FLORIAN NIEDERMEYER Summary and Outlook SONJA KLINGERT (UNIMA) (UNIPA)

iii Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

TABLE OF CONTENTS

I.1. Introduction...... 1 I.1. Purpose of the document ...... 1 I.2. Scope ...... 1 I.3. Methodology ...... 2 I.3.1. Basic Methodology for all Focus Cities ...... 2 I.3.2. Specific Approach for Barcelona ...... 4 I.2. Definitions and acronyms ...... 5 II. The European Data Centre Market and DC4Cities – an Overview ...... 6 II.1. Collection of Data Centre Information ...... 6 II.1.1. Surveyed DCs ...... 7 II.1.2. Large Scale EU DCs ...... 8 I. 1.2c EU DC Code of Conduct ...... 12 II. 1.3 Energy Landscape within EU Countries with high share of European DC market ...14 III. 1.3 Conclusion ...... 15 III. Data Management in European Smart Cities ...... 17 III.1. Questionnaire ...... 17 III.1.1. Focus on Data Centres ...... 18 III.1.2. Focus on Renewable Energy ...... 20 III.2. Conclusion ...... 22 IV. Barcelona ...... 23 IV.1.1. Commercial and Enterprise Data Centres ...... 23 IV.1.2. Potential for Renewable Power ...... 25 IV.1.3. DC4Cities potential in Barcelona ...... 36 IV.1.4. Summary...... 43 V. Options for DC4Cities in Selected Smart Cities ...... 44 V.1. Amsterdam ...... 44 V.1.1. Public Strategy and Options ...... 44 V.1.2. Commercial and Enterprise Data Centres ...... 45 V.1.3. Potential for Renewable Power ...... 46 V.1.4. Summary and Analysis ...... 48 V.2. Paris ...... 49 V.2.1. Public Strategy and Options ...... 49 V.2.2. Commercial and Enterprise Data Centres ...... 51 V.2.3. Potential for Renewable Power ...... 51 V.2.4. Summary and Analysis ...... 54 V.3. London ...... 56

iv Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

V.3.1. Data Centres in London ...... 56 V.3.2. Business Models and SLAs ...... 58 V.3.3. Energy production and Smart city policies in London ...... 59 V.3.4. Summary...... 61 V.4. Frankfurt ...... 62 V.4.1. Data Centres in Frankfurt ...... 62 V.4.2. Energy Production, Usage and Smart City Policies in Frankfurt ...... 65 V.4.3. Summary...... 68 V.5. Madrid ...... 69 V.5.1. Data Centres in Madrid ...... 69 V.5.2. Energy context in Madrid ...... 72 V.5.3. Smart City Policies in Madrid ...... 74 V.5.4. Summary...... 75 VI. Conclusion and Outlook ...... 77 VII. References...... ix VIII. Appendix ...... xi VIII.1. Overview Data Centres in Europe ...... xi VIII.2. Data Centres in Barcelona ...... xi VIII.2.1. Public DC table ...... xi VIII.2.2. Additional information Public DCs ...... xi VIII.2.3. Private DCs ...... xi VIII.3. Data Centres in Amsterdam ...... xi VIII.4. Data Centres in Paris ...... xi VIII.5. Data Centres London ...... xi VIII.6. Data Centres Frankfurt ...... xi VIII.7. Data Centres Madrid ...... xi

TABLE OF FIGURES Figure 1 – Business models of CoC Participants ...... 13 Figure 2 – DCIE of CoC participants ...... 14 Figure 3 – United Nations statistical divisions for Europe ...... 18 Figure 4 – Smart Cities using different kind of DCs ...... 18 Figure 5 – The ratio of own and commissioned DCs in Northern Europe ...... 19 Figure 6 – The ratio of own and commissioned DCs in Southern and Western Europe ...... 19 Figure 7 – The ratio of electricity and other costs of Helsingborg’s DCs ...... 20 Figure 8 – The planned percentage of renewable in the electricity mix today and in future ...21 Figure 9 – The sunshine hours per year of Smart Cities based on their answer ...... 22 Figure 10 – Distribution DCs Barcelona according Tiers ...... 23

v Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 11 – Geographical distribution DCs Barcelona ...... 24 Figure 12 – DCs distribution according their business model...... 24 Figure 13 – Final energy consumption per capita in 2012 [BCN12] ...... 26 Figure 14 – Evolution of Energy consumption, population and GDP [BCN12] ...... 26 Figure 15 – Distribution of final energy consumption form [BCN12] ...... 27 Figure 16 – Electricity generation mix in Barcelona, 2008 [PECQ08] ...... 27 Figure 17 – Electricity consumption mix in Barcelona, 2008 [PECQ08] ...... 28 Figure 18 – Spanish electricity mix, 2008 [PECQ08] ...... 28 Figure 19 – Wind generation potential in Barcelona ...... 29 Figure 20 – Photovoltaic generation potential in Barcelona ...... 30 Figure 21 – Location of different PV projects carried out by the City Administration ...... 30 Figure 22 – PV panels in Les Corts cemetery, Barcelona ...... 31 Figure 23 – PV panels in Joan Miró library, Barcelona ...... 31 Figure 24 – Evolution of installed PV power in Barcelona ...... 34 Figure 25 – Expected PV typical profiles in 2020 ...... 35 Figure 26 – Expected evolution of the regional energy mix [PEC08] ...... 36 Figure 27 – Solar irradiation potential ...... 37 Figure 28 – Wind energy type profiles ...... 38 Figure 29 – Baseline scenario DC consumption vs %RES ...... 40 Figure 30 – DC4Cities scenario DC consumption vs %RES ...... 41 Figure 31 – Locations of DCs in Amsterdam [CEO13], p.8 ...... 45 Figure 32 – PV plants and wind turbines in Amsterdam ...... 47 Figure 33 – Monthly energy output from fixed-angle PV system ...... 52 Figure 34 – Daily irradiance on a fixed plane on a typical winter and summer day ...... 53 Figure 35 – Development of power versus energy demand in MGP [DRI12] ...... 54 Figure 36 – The distribution between the two business models B2B and B2C ...... 57 Figure 37 – Floor size of collected DCs in London. The line marks the average floor size ....57 Figure 38 – Power connected load for the considered data centres in London...... 58 Figure 39 – Business models of collected DCs ...... 59 Figure 40 – Electricity generation by source in UK ...... 59 Figure 41 – Installed solar power capacity in London in comparison of other regions in the England in kW [AEC13] ...... 60 Figure 42 – Data Centre Characteristics ...... 62 Figure 43 – Data centre size...... 63 Figure 44 – Data Centre Business Models in Frankfurt ...... 63 Figure 45 – Power Connected Load of DCs in Frankfurt ...... 64 Figure 46 – Exemplary suitability map for PV in the area of the Equinix FR4 DC in Frankfurt ...... 65 Figure 47 – Energy Situation Today [FRA15] ...... 66

vi Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 48 – Energy Situation in 2050 [FRA15] ...... 67 Figure 49 – Prediction of a typical spring week [FRA15] ...... 67 Figure 50 – Prediction of a typical autumn week [FRA15] ...... 68 Figure 51 – Map of the most important DCs in Madrid ...... 70 Figure 52 – Business models of DCs in Madrid ...... 71 Figure 53 – Final energy use in Madrid by form, 2013 [MAD15] ...... 72 Figure 54 – Evolution of PV production 2006-2013 [MAD15] ...... 74

LIST OF TABLES Table 1 – DC 1 Card Information ...... 7 Table 2 – DC 2 Card Information ...... 7 Table 3 – Telecity DC Card Information ...... 8 Table 4 – Equinix DC Card Information ...... 9 Table 5 – Digital Realty DC Card Information ...... 9 Table 6 – Interxion DC Card Information ...... 10 Table 7 – eShelter DC Card Information ...... 11 Table 8 – Virtus DC Card Information ...... 11 Table 9 – EU DC CoC Card Information ...... 12 Table 10 – EU DC CoC Card Information ...... 14 Table 11 – Summary of the current electricity situation in Barcelona ...... 28 Table 12 – Barcelona current situation [BCN12] ...... 33 Table 13 – Distribution of installed PV power by plant size24 ...... 33 Table 14 – Expected PV output in 2020 [PEC08] ...... 34 Table 15 – Expected Barcelona power consumption in 2020...... 35 Table 16 – Expected distribution of PV installed power by plant size in Barcelona, 2020 .....35 Table 17 – Barcelona’s representative DC main characteristics ...... 36 Table 18 – DC yearly energy consumption and potential savings ...... 37 Table 19 – PV main characteristics ...... 37 Table 20 – PV type profiles ...... 38 Table 21 – %RES type days for Barcelona mix grid ...... 38 Table 22 – Type profile weight ...... 39 Table 23 – Energy prices distribution ...... 39 Table 24 – Energy prices ...... 39 Table 25 – Yearly energy prices simulation ...... 39 Table 26 – Baseline scenario energetic, environmental and economic balance...... 40 Table 27 – DC4Cities scenario energetic, environmental and economic balance ...... 41 Table 28 – 2020 scenario energetic, environmental and economic balance ...... 41 Table 29 – Summary of the results obtained for the representative DC ...... 41

vii Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Table 30 – Summary of the results obtained at city level ...... 42 Table 31 – Data centers floor space [CEO 13, p.10] ...... 46 Table 32 – Wind production in Amsterdam ...... 47 Table 33 – Summery Data Analysis Amsterdam ...... 48 Table 34 – Summary Data Analysis Paris ...... 54 Table 35 – The Mayor's CO2 emissions reduction targets in London compared to 1990 [LON14] ...... 61 Table 36 – DC Market in Madrid ...... 69 Table 37 – Summary of the current electricity situation in Madrid [MAD15] ...... 72 Table 38 – Madrid current situation ...... 73 Table 39 – Installed PV power distribution by size ...... 74 Table 40 – Expected distribution of PV installed power in Madrid, 2020 ...... 74

viii Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

I.1. INTRODUCTION I.1. Purpose of the document

DC4Cities aims at running data centres (DCs) in a highly adaptive way with regards to the power supply offered by intermittent renewable energy sources like photovoltaic (PV) and wind power sites. The technical feasibility of the suggested approach has been developed, implemented and tested in the DC4Cities trials throughout the last two years. The results until now (mid second trial phase) are very positive (see D6.2): It is feasible to adapt the workload of a DC to a given and regularly updated power consumption plan, the so-called Consolidated Power Plan (CPP, see D6.2) without incurring substantial cost or in exchange of incurred cost with a reward. This is only one side of the medal; in order for the DC4Cities approach to have a real impact on the European DC landscape and on the integration of DCs into European smart cities and national energy grids DC4Cities needs to penetrate the European DC market. To explore this potential is the task of the market analysis. I.2. Scope

D2.4 builds on the first iteration of the market analysis D2.2, delivered in summer 2014. It unveiled smart cities' strategies in Europe integrated into ambitious smart city plans on European level including objectives to increase renewable energy shares. It looked at the geographical dispersion of smart cities, striving DC markets and availabilities of intermittent renewable energy sources and identified a set of promising starting areas for marketing DC4Cities. It also pointed out a major problem of pinning down specific market sizes for DC4Cities: the data base. It proved to be impossible on European level to get precise information on the DC market; also the identification of smartness levels in cities was ambiguous. Therefore, for this market analysis the DC4Cities consortium decided to focus on a set of smart cities, which are at the same time among the most vivid DC industry hot spots (the so-called European “Big Five”):  Barcelona: in the pole position both from the (geographical) potential of renewable energy and smart city category; however not a big player in the DC markets. Not among the Big Five, but a project partner where a high level of detail in the analysis was possible.  Amsterdam: A DC hot spot with aspiring plans both for the smart city development, regarding the use of renewable energy sources and the location policy for DCs.  Paris: Also a DC hot spot with ambitious smartness plans that at first sight seems to be well equipped with renewable energy.  London: THE top European DC market. Rather unprivileged though from the point of view of sun and wind sources. It was put on the list of focus cities as a reference from the DC market point of view.  Frankfurt: Together with London the hottest DC market in Europe; however, from the level of renewable energy it was deemed lagging behind cities like Amsterdam or Madrid. It was also added to the list as a reference as part of the Big Five rather than as a candidate for high shares of renewable energy.

1 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

 Madrid: At the beginning of the year, Madrid was still categorized among the European big five DC hotspots; however since then it has lost this attribute. On the other hand, it has a high potential for PV based renewable energy. The analysis of those focus cities will identify both the present DC market potential as well as the market potential for 2020 or 2030 (depending on the availability of planning data). I.3. Methodology

It is a great challenge to provide market estimations for the focus cites of the DC4Cities final market analysis. The reason is that the level of accessible and attainable data is completely different in the listed cities, depending on a plethora of factors like personal ties with the city, statistics and documents to be found via internet research. Only for the city of Barcelona the detail of data is sufficient to allow a quantitative analysis. DC4Cities is an approach envisioned for contexts in which public administrations promote transformations in cities in favour of an increase of energy self-sufficiency through the usage of renewable energies. As explained in D2.2, with the aim of accomplishing the European energy goals in the horizon 2020 – 2030, cities are evolving their energy systems in order to increase energy efficiency and the share of renewables in the energy mix, and to decrease CO2 emissions. The market size for DC4Cities then needs to be evaluated considering this future scenario. Independently of the lack of definition of external factors and data basis size for each city, the methodology of analysing the collected data should follow a unique approach based on a minimum set of data to be researched in all 6 focus cities. For Barcelona, however, as an active partner in the DC4Cities consortium the data situation is privileged compared to the other focus cities – therefore a more intricate approach is feasible.

I.3.1. Basic Methodology for all Focus Cities

The goal of the basic methodology is to coarsely estimate the magnitude of a market for DC4Cities in the focus cities or qualitatively evaluate the opportunities for DC4Cities in the respective city. This analysis defines a scope for applying DC4Cities based on the data collection carried through in the context of the market analysis task. It will rely on the following data and assumptions: 1. Information about the number of DCs in the respective cities The number and sizing of DCs in any of the focus cities will be collected as much as possible. In most cases due to the absence of an official “DC registry” it is not possible to have an objectively complete number of DCs in a city; we will always know only a subset of them. From this analysis we exclude small server closets or server rooms; and we don’t have sufficient information about enterprise DCs. 2. Collected information about the three basic business models of DCs: colocation vs. application based or cloud computing DCs. HPC is excluded from the analysis. The analysis of D2.1 and D2.2 resulted in the perception that the level of flexibility of DCs is to some degree dependent on the business model. Therefore, this is valuable data to be collected. For colocation DCs, at the moment DC4Cities does not offer a simple solution. The reason is that colocation can cover various degrees of outsourcing application and server management to the DC (from the point of view of the outsourcing company), but the prevalent situation is that a colocation DC is just housing servers and is not involved in their management. There are business models, where colocation centres offer various degrees of backup and virus management – these could apply DC4Cities in a

2 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

limited way. As there are no reliable figures to be found in statistics, from experience, we assume that only 10% of colocation DCs can apply DC4Cities. In addition, we assume all other DCs business models have enough flexibility to be able to apply DC4Cities in a more or less limited way. All DCs where no information about the business model is available, are treated as if the same ratio as prevalent in the respective city were applicable to them. 3. Information about the power and energy consumption of these data centres. Data about the power and energy consumption of the DCs will be used where available. In case there is no information available, the analysis will be rather qualitative than quantitative. 4. Information about the renewable local energy infrastructure and/or renewable energy yield, as detailed as possible as well as information about the national power grid energy mix. These data will be collected for the situation today and extrapolated based on the political guidelines as given in information source 5. Where data are not available and need to be estimated, this will be transparently stated. 5. Information about the smart city policies with regards to both data centres and/or goals for renewable energy generation/infrastructure/public DCs as candidates. The information collected here will be used to evaluate how the market potential will develop in future: the higher the gap between today’s infrastructure and the goals for 2020, the higher is the potential in relation to today’s market. 6. General information about weather and climate data for solar and wind energy harvests where no information about yields from solar and wind sites is available. 7. Information about the flexibility of the DCs The ideal data would consist of shares of flexible load differentiated according to durations. The more fine-grained the data we have, the more meaningful the analysis would be. In most cases, there is no data available; then we use the following set of assumptions:  For a prevalently cloud computing oriented business model we will assume that between 5-15% of the load can be shifted. These are not minimum/maximum values, but rather assumptions for more or less shifting opportunities. We assume that the load can also be shifted to the next day.  For a prevalently colocation oriented business model we will assume that between 5-20% of the load can be shifted. These are not minimum/maximum values, but rather assumptions for more or less shifting opportunities. We assume that the load can also be shifted to the next day.  For a prevalently application oriented business model we will assume that between 10-30% of the load can be shifted. These are not minimum/maximum values, but rather assumptions for more or less shifting opportunities. We assume that the load can also be shifted to the next day. Apart from coarse estimations in our trial DCs (IMI about 20%, APSS about 5%, CSUC about 20%) assumptions are based on our general experience. 8. For all computing styles we will assume more or less flat power profile This is a simplifying assumption due to missing data, however based on experience, for most DCs only the upper 10% are fluctuating.

3 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

In case information about on-site generation of renewable energy is available this will have a positive effect on the marketing potential of DC4Cities as the respective DC will be counted as potential DC4Cities candidate. Putting together the information on power and energy of data centres and combining them with the information/assumptions on flexible load based on the characteristics of DCs will yield the aggregated power/energy load of DC in the focus city as well as the aggregated flexible power/energy. Then we can compare these with the peak power of the renewable site infrastructure and/or with the renewable energy harvest in the city and with the total energy/power consumption in the city. The basic methodology approach will result in a foremost qualitative analysis of the market potential of DC4Cities considering a certain percentage of collected data centres might be inclined to use DC4Cities under the current and future conditions. One basic assumption is that DC4Cities is given out for free, e.g. from the smart city administration. I.3.2. Specific Approach for Barcelona

The main guidelines of the basic methodology previously explained have been followed in order to evaluate the potential of DC4Cities in Barcelona. However, because of the possibility of acquiring more accurate information than in other cities, a deeper analysis has been performed, obtaining a preliminary value of the potential of DC4Cities. The steps of this more detailed methodology are as the following: 1. Selection of a representative DC In order to acquire some precise information about the number and size of DCs that will be potentially suitable for DC4Cities in Barcelona, two different sources have been assessed. This process will be described in section IV.1.1. . On the one hand, public reference documents have been consulted and, on the other hand, questionnaires have been sent to private DCs. From this information, a particular DC has been selected as representative to evaluate DC4Cities impact. 2. Information about the flexibility and energy efficiency improvement of the DC Regarding the flexibility, the same assumptions for the general methodology has been considered: cloud computing business model (5% to 15%), collocation business (5% to 20%) and application business model (10% to 30%). For the workload that can be flexible an energy savings equal to 70% have been considered. This value represents the main results achieved for Barcelona trials in Phase I. 3. Information about the energy context: renewable energy availability and energy prices In accordance with the Barcelona Self-sufficiency Energy Plan, which boosts the installation of small/medium photovoltaic power plants, it is assumed that DC4Cities installation will be accompanied by the installation of a local photovoltaic power plant on the roof of the DC or next to its facility (e.g. parking). The usage of local energy will have priority. As a result of the high energy intensity of a DC, most of the electricity demand will be provided by the grid. This approach considers that the flexibility achieved with the use of DC4Cities can help to integrate wind energy at grid level; that is shifting consumption to valley periods, in which electricity demand is lower and a lower energy price is available for consumers. This approach is in line with the current trends to integrate new electricity consumers, as electric vehicles and is also necessary for optimizing the grid functioning, since the demand at global level, i.e. considering not only DCs but also other consumers, could be flattened. Furthermore, this is also consistent with the ideal context described in D2.2, since dynamic energy prices are the factor used for shifting

4 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

energy consumption. As stated in D2.2 dynamic prices are expected to be strongly correlated to the relation between production technologies availability and electricity demands. 4. Evaluation of the energetic, economic and environmental impact due to DC4Cities In order to understand the potential penetration of DC4Cities, it is needed to assess the energetic, environmental and economic improvements that a DC will gain thanks to the usage of our system. For the selected DC a yearly evaluation has been performed, using a similar methodology as the one used for evaluating DC4Cities impact in trails; therefore, we choose a set of representative profiles that can serve to extrapolate results from a set of daily simulations to the whole year. The KPIs used to evaluate the impact are: RenPercent, CO2 savings, energy savings and energy expenses savings (see D7.2 for more details). 5. Extrapolation at city level Taking into account the results obtained from a single DC and knowing a set of DCs that can be potentially suitable for implementing DC4Cities, the impact of DC4Cities will be extrapolated to the whole city. I.2. Definitions and acronyms

Acronym Description

DC DATA CENTRE

PHOTOVOLTAIC (POWER, PV ENERGY)

GoO GUARANTEES OF ORIGIN

MGP METROPOLE GRAND PARIS

RENEWABLE ENERGY RES SOURCES

TIME OF USE (D BASED TOU PRICING OF ENERGY)

ENERGY MANAGEMENT EMA-SC AUTHORITY OF A SMART CITY

5 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

II. THE EUROPEAN DATA CENTRE MARKET AND DC4CITIES – AN OVERVIEW

In this section, we provide a very high level analysis of the potential market for DC4Cities in European DCs. This market analysis is relying on two major sources of information:  A questionnaire distributed among 50 DCs in Europe  Internet research  Bi-lateral (partly interview-like) communication via email. In spite of this multitude of data sources, it was still not possible to get information in a level of detail that allowed for a thorough analysis and modeling of the DC landscape in Europe in order to assess the chances of marketing DC4Cities. According to Greenpeace, globally, DC operators and Internet companies have started to make ambitious plans for increasing renewable energy supply in order to address high energy costs coupled with client concerns about the environmental impact of application services [CPR15]. On the other hand, more and more DC customers are looking for DCs that are powered by renewable energies. Even colocation customers are making greenhouse gas and renewable energy commitments [CPR15]. Recently, 84% of North American DC operators identified the need to consider renewable energy for meeting future energy needs due to environmental challenges, energy costs, and transition to renewables [CPR15]. Additionally, the transition to a green energy economy is emerging and is being integrated within DCs. Electricity is the biggest cost for DCs long-term plans 1. Renewable energy costs are declining and we are observing competitive prices for renewable generated electricity [CPR15]. For example, since 2008 solar energy costs have fallen 80% globally [CPR15]. Alongside with this trend, DC4Cities provides an economic approach that takes into account energy costs as well as renewable energies for DC operators and IT services operators in order to define their own energy goals or pursue goals provided by a smart city. DC4Cities has a potential market within EU regions with both low and high renewable energy availability because the local energy mix also depends on the share of green energy within national grids. DC4Cities has a high potential market if 100% of the electricity is produced by renewables. Until that time DC4Cities can contribute to close the gap to approach 100% renewable energy goal. In cases of 100% green energy availability DC4Cities can be used to save energy costs by adapting energy consumption to cheap times if/within regions with dynamic pricing scheme for electricity consumption. In addition, having 100% renewables, and surplus of the supply, DC4Cities federation can be an asset to execute workload of other DCs with poor share of renewables. II.1. Collection of Data Centre Information

We have collected some information about a few big DC chains within Europe through a questionnaire, interview, Internet research, and email communication. The questionnaire2 captures the most important key data about the DC infrastructure, in particular its energy parameters, business models, and flexibility aspects of the DC with its 21 questions. We have sent out the questionnaire to 50 DCs with twice reminders in order to recall them to respond. In the next sections, we provide a short report on each DC, and provide our analysis about the potential application of DC4Cities within each DC.

1http://www.itweb.co.za/index.php?option=com_content&view=article&id=70866 2http://goo.gl/forms/1SdLjsou9o

6 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

We base our analysis on a few assumptions, plans, and trends, and our first phase trial results. Being a visionary project as we approach the future, DC4Cities market would evolve further; considering both energy landscape, and smart cities development, etc. On the other hand, how would be the pricing model of DC4Cities is very important; whether it is at the DC level, application level, or both make a difference in DC4Cities revenue. II.1.1. Surveyed DCs

Through the questionnaire we received three responses. At the end, one respondent did not agree to publish information about their DC. All in all, the questionnaire's responses indicate that DCs seem to have great interest in DC4Cities’ approach to increase renewable usage. In the following, we provide our analysis about the two DCs.

II.1.1.a. DC 1

Table 1 – DC 1 Card Information PUE # Power Energy #sites Business Data source servers model

1.5 10000 4000 Mwh Cloud, Questionnaire Colocation

DC 1 is a data center provider chain in Europe with 41 sites in 12 locations. The following information is about one of their DCs located in France. This DC benefits from renewable energies by having a contract with energy providers that guarantee a certain level of renewables. They buy from their power provider (EDF) green power (this is called “Equilibre” certificate). They are already using energy efficiency solution within DC. They are continuously looking for energy efficiency solutions. They are interested in energy efficiency solutions due to the following reasons: Cost saving, Sustainability, Legal constraints, and Responding to Demand Response requests. In addition, they are interested in solutions that beyond saving energy optimizes the usage of green energy sources. The DC's SLAs are 100% on power distribution (Power availability) and 99,97% on cooling. Analysis: This DC is very big in terms of the number of servers, and power consumption, with a quite low PUE (1.5). Their business model is quite diverse, making it also an interesting option for DC4Cities. However, they are already supplied by a certain level of renewables. Nonetheless, DC4Cities can be used to increase the renewable usage beyond the level provided by the energy provider and enabling the integration of intermittent renewables. In sum, there is good potential for green energy improvements considering the diverse range of business models, the fact that the DC is part of a huge chain to which the innovation could be spilled-over and considering their general interest in optimizing the usage of green energy sources.

II.1.1.b. DC 2

Table 2 – DC 2 Card Information PUE # Power Energy #sites Business Data source servers model

1.65 Avg:700kW 5000 Mwh 1 Colocation Questionnaire Max:800kW

7 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

DC 2 is a data center provider in Switzerland. The Tier level of DC is IV. They are already using energy efficiency solution within DC, including free cooling, variable speed drives, aisle containment, etc. In addition, they are expressed interest in solutions that beyond saving energy optimize the usage of green energy sources due to independence of the power grid and aspiring eco-reputation. Analysis: There is a 100kW power fluctuation. This means power profile within DC is not flat, making it an interesting option for DC4Cities adaptive approach. Depending on applications characteristics within DC, some portion of this power variation might be shiftable to non-peak times by DC4Cities in order to improve green energy usage. In order to apply DC4Cities approach within this DC, it should provide a contract with its clients about the DC4Cities clauses, and terms in order to provide interesting incentives for the clients.

II.1.2. Large Scale EU DCs

In this section, we provide our insights and analysis about some of the large scale European DC players. EU DC market is very diverse throughout Europe including small to medium in- house DCs, and world-wide DC providers, like Equinix, and Digital Realty.

II.1.2.a. Telecity

Table 3 – Telecity DC Card Information

PUE # Customer Energy #sites Business Data servers Power model source

Manchester: 2015: 12 Colocation Internet 1.35 117.1 MW locations 5 (cities) Helsinki: <1.3 3 H1 2014: 106.3MW London: FY 2014: 1.35 4 111.7MW

“Telecity is one of the largest retail colocation providers in Europe, operating in 12 major markets across the continent and providing a European home for major online service such as streaming music platform Spotify and professional social network Xing, Akamai, Facebook, Amazon Web Services, and Microsoft.“, from Greenpeace report [CPR15]. Telecity DC locations are Amsterdam, Dublin, Frankfurt, Helsinki, Istanbul, London, Manchester, Milan, Paris, Sofia, Stockholm, Warsaw 6. They are also part of EU DC CoC 7. According to Greenpeace report, “Telecity does not set any long-term ambition to increase its use of renewable energy, or even recognize access to renewable electricity as a priority. This deficiency is a significant disadvantage to Telecity’s ability to compete for customers who

3http://www.telecitygroup.com/our-company/news/2013/telecitygroup-launches-new-state-of-the-art- data-centre-in-helsinki-finland.htm 4http://www.telecitygroup.com/data-centres/london-data-centre-powergate-tour.htm 5http://telecitygroup.com/presentations/Telecity-Interim-report-2015.pdf 6http://telecitygroup.com/presentations/Telecity-Interim-report-2015.pdf 7http://www.telecitygroup.com/environment.htm

8 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities want to be assured that their hosting provider is working to power its online operations with greater amounts of renewable electricity.” [CPR15] Analysis: Telecity has multiple sites within 12 European cities. Each DC is very big in terms of the number of servers, and power consumption. Their business models are colocation, and cloud. There are potential for green energy improvements due to being a big DC, and offering a diverse range of online services (like HP trial in DC4Cities scenario). More importantly, they could be a great adopter of DC4Cities due to Greenpeace perspective, that is significant disadvantage of not committing to long-term plans for renewables. By using DC4Cities they can mitigate this deficiency, and attract more clients. In addition, Telecity benefits from diverse range of green energies due to its DCs locations. Thus, some DCs can establish bi-lateral contracts to exploit DC4Cities federation in order to increase the share of renewable usage.

II.1.2.b. Equinix

Table 4 – Equinix DC Card Information PUE # Power Energy #sites Business Data source servers model

EMEA: Global: 2200 EU: 29 DCs 8 Colocation Internet 1.29- GWh (2014) EMEA: 130,000 1.42 sqm

Global: 100+

Equinix is a big DC chain for the Internet, with 29 DCs in Europe and over 100 DCs spread around the world globally. EU DCs are located in France, Germany, the Netherlands, Switzerland, the UK.Table provides some key data about Equinix DCs globally. Equinix made a long-term commitment to be 100% renewably powered. At least for the short term, Equinix has established a clear competitive advantage.” [CPR15]. In addition, Equinix claims a 30% renewable energy mix at the global level [CPR15]. According to Greenpeace report, Clean Energy Supply percent to DCs for Netherlands, London, Paris, Switzerland, Germany are 12%, 15%, 13%, 57%, and 22%, respectively [CPR15]. Analysis: Equinix is very big colocation provider, located in European DC market hubs, with huge power consumption. In short-term, there are good potential for DC4Cities in order to increase green energy usage beyond 30% due to being a big DC chain, until the 100% green target. In the order of penetration degree, we envision further renewable usage improvement for Netherlands, Germany, London, Paris, Switzerland by adopting DC4Cities.

II.1.2.c. Digital Realty

Table 5 – Digital Realty DC Card Information PUE Power Energy #sites Business Data source model

Amsterdam: 1.15 11.5 MW EU: 22 Colocation Internet DCs Dublin: 1.15 16 MW Cloud Global: London1: 1.28 9 MW

8http://www.equinix.com/locations/europe-colocation/europe-data-centers/

9 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

London2: 1.3 to 1.6 30 MW 131 DCs

Paris: 1.4 5 MW

“Digital Realty is the largest digital landlord in the world, with 131 data centers worldwide, roughly three-fourths of which are located in the US, 18% in Europe and 4% in Asia, totaling over 24 million square feet of rental space. Digital Realty operates on the wholesale end of the colocation spectrum, providing entire data center properties to large online brands, many of which now have adopted long-term commitments to be 100% renewably powered, including Facebook, Rackspace, Salesforce, Google, and very recently Equinix and Amazon Web Services.“ [CPR15] Table provides some key data about Digital Realty DCs. They operate DCs in the following European cities: Amsterdam, Dublin, London, Paris, Geneva, and Manchester. Digital Realty has recently made some plans to adopt a sustainability policy. They offer a new program (Clean Start program) to facilitate the purchase of renewable energy credits (RECs) for clients that sign a new lease with Digital Realty. This kind of contract could be a blueprint for setting up a colocation-oriented business model for DC4Cities. According to Greenpeace report, given the size of Digital Realty’s data center portfolio, and the number of major customers that have already made 100% renewable commitments, and their European locations hubs with promising share of renewables, Digital Realty should be able to offer a much more impactful program. [CPR15] In 2014, 41% of their energy consumption came from non-CO2-emitting sources approximately; from this 41%, 22% is categorized as renewable.9 Analysis: Digital Realty is very big colocation provider, with huge power consumption. Digital Realty provides not much more than the house of the servers; so this is an example for a service provider that does not have any kind of influence on the workload running in its DC. According to Greenpeace Report, “Digital Realty’s first offering falls far short of what most companies are seeking in terms of additionality; Digital Realty should look to the policy recently adopted by Equinix for a stronger model among colocation operators. More specific performance targets, including renewable energy or carbon goals such as those adopted by Telecity and Equinix, are needed to better define where Digital Realty wants to go and how it plans to get there.“ [CPR15] Therefore, we envision higher impact of DC4Cities within Digital Realty compared to the Telecity, and Equinix in order to address their customers’ needs, and to increase share of renewables. In order to adopt DC4Cities, we need to establish some contracts for setting up a colocation-oriented business model.

II.1.2.d. Interxion

Table 6 – Interxion DC Card Information PUE # Power Energy #sites Business Data source servers model

Stockholm: 40 DCs Colocation Internet 1.09 10 11 EU Cloud countries

9https://www.digitalrealty.com/solutions-services/sustainable-innovation/ 10http://www.datacenterknowledge.com/archives/2013/03/06/interxion-uses-seawater-to-cool- stockholm-data-centers/

10 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Interxion is another leading provider of cloud, and colocation DCs in Europe with headquarter in Amsterdam. They have 40 DCs across 11 European countries. Full list of cities are Amsterdam Brussels, Copenhagen, Dublin, Frankfurt, London, Madrid, Paris, Stockholm, Vienna, and Zurich. They locate DCs close to city centers to provide excellent power availability and connectivity. Table provides some key data about Interxion DCs globally.It claims to supply 90% of power from sustainable sources, including water, solar and wind.11 In addition, they state that have 100% sustainable energy within all the aforementioned cities. Analysis: Interxion DCs are located in major cities with good share of renewables. They already supply their DCs with 90% sustainable energy. We don't know how much of these “real” and even intermittent renewable energies or just GoOs are. Probability is high, that Interxion uses GoOs, and therefore we see good opportunities to exploit DC4Cities in order to switch partly to local, intermittent resources (e.g in Amsterdam). Also, DC4Cities federation capabilities can be of a great asset for DC federation of Interxion.

II.1.2.e. eShelter

Table 7 – eShelter DC Card Information PUE # Power Energy #sites Business Data servers model source

Frankfurt: 8 sites Colocation Internet 213MW Germany EU: 316MW Switzerland

Austria eShelter is a regional DC provider within Germany Switzerland, Austria, and Switzerland. They claim to supply their energy demand by 100% renewable energy sources12. The table provides some key data about eShelter DCs. Analysis: Our analysis for eShelter is like Interxion. With 100% renewable energy supply we see still opportunities to save energy costs by exploiting cheap energy prices within a dynamic pricing program by switching to local, intermittent renewable energy sources.

II.1.2.f. Virtus

Table 8 – Virtus DC Card Information PUE # Power Energy #sites Business Data servers model source

London1<1.5 15.6 2 Colocation Internet MW London2<1.2 28 MVA

11http://www.interxion.com/why-interxion/sustainability/ 12https://www.e-shelter.de/sites/default/files/field/file/fra1_en_2014.pdf

11 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

VIRTUS Data Centres13 is a colocation DC provider within London region with 2 sites. They are located close to the city but outside the congestion zone. They claim 100% energy from renewable sources.Table provides some key data about Virtus DCs. Analysis: Our analysis for Virtus is like eShelter, and Interxion. With 100% renewable energy supply we see still opportunities to save energy costs by exploiting cheap energy prices within a dynamic pricing program.

II.1.2.g. CenturyLink, Colt, Interroute

CenturyLink, Colt, and Interroute are additional European DC chains that focus on colocation, but also offer cloud business models to some degree. They combine a power demand of several hundreds of MWs, spread over together 68 sites. They do not publish information about their energy mix and strategy. Analysis: With hardly any information about the energy strategies of these chains, even a rough analysis is not possible. However, their sheer size together with the geographical spread and a trend towards committing to renewable energy sources should open up a huge market potential for DC4Cities approaches, provided that the cost for license and implementation are low compared to the expected benefit. I. 1.2c EU DC Code of Conduct

EU DC CoC is a voluntary initiative aimed to bring interested stakeholders together, including the coordination of other similar activities by manufacturers, vendors, consultants and utilities. EU DC Code of Conduct (CoC) provides a portfolio of European DCs that take serious actions in order to implement energy efficiency measures [BER13]. According to “Presentation of the Awards” in 2015, this portfolio offers a wide range of DCs including enterprises who value energy efficiency, and the use of renewables, with 111 companies registered as participants, 255 data centres approved as participant, covering 6.0 TWh of energy consumption, and 210 endorsers [COC15]. The 57% of DCs are stand-alone. As it is illustrated in Figure 1 most of the DCs are the traditional enterprise, followed by hosting. DCs are improving efficiency over time. The average PUE was under 1.7 in 2013. Table 9 – EU DC CoC Card Information PUE # Power Energy #sites Business Data servers model source

Min: 1.25 6 TWh 111 Colocation, Internet companies Cloud, Avg: 1.7 EU DC CoC Application 255 DCs Max: 2.86

PUE and DCIE distribution of participating DCs are illustrated in Figure 2

13http://virtusdatacentres.com/locations/

12 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 1 – Business models of CoC Participants The best practices, with no capital expenditures or major changes to business practice, are among the top implemented one within DCs [BER13]. The “Power Management” best practice was among the worst adopted due to dealing at the hardware level [BER13]. In addition, the “Selection of efficient software” practice was difficult to implement due to lack of markets to offer energy efficiency software [BER13]. Analysis: DCs within this dataset could more easily adopt DC4Cities services as they have energy efficiency measures in place, and care about energy challenges. In addition, DC4Cities could have a high impact on DCs given that there are various DC types with a low PUE on average (1.7) within different countries in Europe. On the other hand, DC4Cities can be proposed as part the Code in order to be implemented as a best practice by the interested DCs due to being a management software, and its transparent and easy integration within a DC infrastructure. In sum, we see a good potential market within EU DC CoC ecosystem. In addition, we have established a relationship with EU DC CoC in order to promote DC4Cities within this Code. We see a great opportunity for DC4Cities on this front. DC4Cities can penetrate within DCs, and then selling more advanced DC4Cities services can be the outcome of this initiative.

13 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 2 – DCIE of CoC participants II. 1.3 Energy Landscape within EU Countries with high share of European DC market

DC Federation capabilities of DC4Cities could further help DCs to improve renewables energy usage due to diversity of renewable energy generation. Renewable energy production within Europe is heterogeneous sourcing mainly solar photovoltaics, wind power, biomass, hydro power, and geothermal power. Even a country, with different cities and regions, provides green energy from various sources. For instance, southern Italy provides substantial wind power, while within northern part of Italy in Trentino-SudTyrol there are also huge amount of power generated from hydro [TER15]. Therefore, DCs could negotiate for federation contracts in order to increase the renewable energy usage of their services by using the DCs that provide higher renewables. Furthermore, EU member states are moving fast to strengthen renewable energy goals. For example, Germany has ambitious renewable energy goals to power its total energy need with 80% renewable energy by 2050 under the Energiewende.14 Denmark is planning to produce half of its electricity by 2020 with wind; thus, Denmark will have at least 50% renewable energy by 202015. Germany and Denmark are neighbors with a high share of renewables, even today. Table provides some key data about EU DC market share, and DC total energy consumption for big players in this landscape [HIN14]. Table 10 – EU DC CoC Card Information Country Year DC market Energy Business model Data share source

Germany 2014 25% 10 Twh Colocation, Cloud, [HIN14] Application

14http://www.germany.info/Vertretung/usa/en/06__Foreign__Policy__State/02__Foreign__Policy/05__ KeyPoints/ClimateEnergy__Key.html 15http://www.greenpeace.org/international/Global/international/briefings/climate/2014/BRIEFING- Denmarks-commitment-to-100pct-renewable-energy.pdf

14 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Germany 2020 12 billion Kwh Colocation, Cloud, [HIN14] Application

UK 2014 21% 9 Twh Colocation, Cloud, [HIN14] Application

France 2014 15% 7 Twh Colocation, Cloud, [HIN14] Application

Netherlands 2014 6% 3 Twh Colocation, Cloud, [HIN14] Application

In 2014, German DCs consumed up to 10 billion Kwh [HIN14]. It is predicted that the consumption will reach to 12 billion Kwh by 2020 [HIN14]. Currently, the largest DC market within Europe belongs to Germany with a share of 25% [HIN14]. This huge share of DC market with a high share of renewables in Germany provide a high level of confidence for this country to be the first potential customer of DC4Cities. After Germany, UK with 21%, and France with 15% share of the European DC market [HIN14] could be the second, and the third customers of the DC4Cities, respectively. In recent year, DC capacity has been increased by more than 17% in the Netherlands [HIN14]. This big increase led the Netherlands to take the fourth place in European DC market by 6% market share [HIN14]. In addition, Dutch electricity prices are moderate respect to Germany. These trends make the Dutch DC market very interesting for DC4Cities. The total energy consumption by servers, and DCs in Germany, the UK, France, and the Netherlands are approximately 10 Twh, 9 Twh, 7 Twh, and 3 Twh, respectively [HIN14]. These countries account for 55% of the total energy consumption of all servers and DCs in Europe (53 Twh) [HIN14]. Therefore, the four countries are definitely the four pillars of the DC4Cities market uptake. III. 1.3 Conclusion

The preliminary analysis, driven by analyzing the data we have collected through various sources, indicate that the majority of DCs are taking into account renewables; and they are taking actions, plans in order to achieve their energy, and renewable targets. Surveyed DCs through questionnaire have great interest in DC4Cities approach. We envision that DC4Cities penetration within large scale DCs like Telecity, Equinix, and Digital Realty are probably higher than Interxion, eShelter, and Virtus. This is due to the claim of the latter DCs about their current renewable energy supply. On the other hand, we cannot really give any opinion about Colt, CenturyLink, and Interoute due to lack of information about their renewable energy goals, and plans. DCs within EU DC CoC can more easily use DC4Cities services as they have energy efficiency measures in place. They have definitely interest in DC4Cities approach. In addition, we would like to propose DC4Cities as a Code within this Code of Conduct. Therefore, we see a great opportunity for DC4Cities marketing within this set of DCs.

15 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Capitalizing on DC4Cities federation within central, western and northern European DCs, coupled with fairly high share of renewables, and diversity of green energy within these regions, we see a great opportunity on this front also for DC4Cities marketing. For instance, DCs in Germany and Denmark, with good share of green energy and diverse energy landscape, can exploit DC4Cities federation for further and diverse approach to improve green energy, capitalizing on renewable energy availability diversity. We conclude that majority of the DCs are willing to invest in renewable energy solutions, like DC4Cities even today. All in all, we identify the four aforementioned countries as the four pillars of the DC4Cities market uptake. Nonetheless, there are DC4Cities marketing potentials within countries like Ireland, Italy, Switzerland, Spain, Sweden, Finland; however with less share of the market.

16 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

III. DATA MANAGEMENT IN EUROPEAN SMART CITIES

The marketing strategy of DC4Cities will consist of two different target groups: commercial DCs (see section II) and DCs that are operated or commissioned by smart city adminis-trations. This section aims at assessing the chances to market DC4Cities in Smart Cities throughout Europe. As already stated, DC4Cities is optimally used in an environment in which it is required to integrate renewable energies. It is therefore very important to get current data about the DCs and the energy mix of the most successful Smart Cities for the recognition of future possibilities of the project. With the help of [MSC14] the 68 most advanced Smart Cities were identified and a questionnaire was sent targeting at data management in smart cities. III.1. Questionnaire

The decision to send out a questionnaire was taken because hardly any of the needed information could be gathered from an internet research. Also, via this tool smart cities receive a first piece of information about DC4Cities and a contact person is identified which can be used for later exploitation. Below the complete set of questioned key data can be found: 1. The current number of data centres, which are either operated or which are commissioned by the smart city administration 2. The maximum power requirement in the year 2013 in kW 3. The energy consumption in the year 2013 in kWh 4. The overall data centre costs in the year 2013 (may include a differentiation between owned and commissioned data centres) 5. The energy costs of the data centre(s) operated by the smart city in the year 2013 6. The energy mix of the year 2013 (or the percentage of renewable energy) 7. The rough estimate of total data throughput through the grid per year in Terabyte or higher (may include a differentiation between owned and commissioned data centres) 8. The expected growth rate of the total data throughput through the grid per year in the coming years 9. The memory capacity in Terabyte and the expected growth rate in the coming years 10. The expected growth rate of renewable energy for the coming years 11. The hours of sunshine per year 12. Reasons why the administration thinks that their smart city is on the list of the most successful smart cities? This questionnaire was sent to 68 cities, of which 13 replied. The raw data can be found in the appendix. Some of the cities asked to be anonymized, they are referred to as “City1”, “City2” and so on. For each city it is noted, in which cultural division of Europe it can be found. This allows an analysis of cultural factors and conclusions in which parts of Europe DC4Cities can have a stronger impact. The cultural division is based on data from the United Nations statistical divisions for Europe and is shown in Figure 3. Blue is Northern Europe, red is Eastern Europe, green is southern Europe and light blue is West- and Central Europe.

17 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 3 – United Nations statistical divisions for Europe Most of the cities that answered to the questionnaire are from Northern Europe. City 1 and 7, Brussels, Gent and Aachen are Western and Central Europe, and only city 10 is from Southern Europe. This great discrepancy may be due to a discrepancy in interest (which does not relate to BCN of course that was not part of the general overview but will be subject to a detailed analysis).

III.1.1. Focus on Data Centres

There are three ways of how DCs can be administered by a city government. 1. The City uses only own DCs and absolutely no commissioned ones 2. The City has completely outsourced the DCs and uses absolutely no own ones 3. The City uses a mix of own and commissioned DCs The way DCs are administered is important to the DC4Cities project, because it shows, who will be mainly the adopter of DC4Cities. If the city administers the DCs on their own, then city governments will be the main adopter otherwise it will be IT service enterprises.

Use only outsourced DCs 33% 39% Use only own DCs

Use a mix of own and commissioned DCs 28%

Figure 4 – Smart Cities using different kind of DCs

18 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 4 illustrates this issue: one third of Smart Cities used only outsourced DCs, 28% had only own DCs, and 39% used a mix of own and commissioned DCs. It can therefore be concluded that DC4Cities has to account for for every kind of DC administration, none of them has shown to be dominant, taken a deviation of 5% into account. In the case of commissioned DCs it is hard to get information about the DCs because the smart city is not authorized to pass on data from their service provider or is not able to get at this data at all. The Smart Cities which only use own DCs can provide data regarding their DCs rather easily. However, the other smart cities However, with the shown trend of outsourcing in Figure 5 and Figure 6 it can be concluded that regional IT-services provider, who manage the DCs of Smart Cities, are also important and potential consumer. IT Service provider and Smart cities own the same amount of DCs and it is therefore important to include both as important users of a DC4Cities System. An interesting aspect of the ratio of own and commissioned DCs is shown in Figure 5 and Figure 6. These two figures clearly show a huge difference between the ratios in Northern Europe and southern coupled with Western Europe. The countries in Northern Europe have a trend to outsource DCs to IT-Services, while countries in Southern and Western Europe have the trend to have almost only own DCs.

21% number of own DCs

number of commissioned DCs 79%

Figure 5 – The ratio of own and commissioned DCs in Northern Europe

14% number of own DCs

number of commissioned DCs 86%

Figure 6 – The ratio of own and commissioned DCs in Southern and Western Europe The average estimation of the increase of data throughput for the future is 60%, but the cities expectation differ widely. According to the data, City10, an anonymous city in Southern Europe expects a growth rate about 30%, with their total data throughput of 300Tb they will have a forecasted total data throughput of 400Tb. These are low values in comparison to Gent (Belgium) which has more than the double total data throughput compared to e.g. City10. Gent

19 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities has a high use of DCs as the energy consumption of nearly one million kWh. But it seems like the project DC4Cities is not helpful for Gent with a 100% renewable power grid. The city governments state that until now DCs have not been a major factor of success of the Smart Cities. This means that smart cities do not have a focus on their DCs. However, if they start putting their focus on the ICT sector, smart cities have to improve their own DCs or they could oblige the service provider via contracts to help the city to increase its “smartness”. This depends on the situation of the smart city as already discussed previously in this chapter. The result is that DC4Cities needs to attract some attention of smart cities to the ICT sector. This could be done for example by explaining how DC4Cities can improve energy efficiency and reduce greenhouse gas emissions or how it reduces energy costs that are a major cost factor of DC cost.

44% Energy costs of DC 56% Other DC costs

Figure 7 – The ratio of electricity and other costs of Helsingborg’s DCs As an example, Figure 7 shows the electricity costs of the DC in Helsingborg compared to the other DC costs. The energy costs make up 44% of the total costs. With a current renewable energy percentage of 93% (as stated for Helsingbourg) it should be easily possible to use DC4Cities and further increase their share of renewable energy sources in the DC energy mix provided that a non-negligible part of the energy mix is intermittent. III.1.2. Focus on Renewable Energy

The questionnaire asked for the current and the targeted percentage of renewable energy in the DC’s energy mix, which is determined by the local energy mix. As Figure 8 shows, there is a clear trend by the smart cities to increase the percentage of energy from renewables tremendously until 2050. This depends also on the grid in the regional area and further on a countrywide level. Cities depend on the incoming energy of the region, which itself depends on the countrywide energy production. This can only be solved with “vertical agreements”, e.g. between city and region or region and country. This trend shows the high interest in renewable energy but also that it needs time to achieve this target. This information supports the project DC4Cities because it helps to put the target of using more renewables into practice. The fact according to the basic rule of the market economy is that a higher demand for renewable energy has the consequence that the supply will increase as well. With this theory the possibility exists that the demand and supply for fossil energy could decrease. So the project could possibly increase the amount of available renewable energy and consequently support the Smart City to reach their planned percentage of renewable energy. E.g. City2 wants to get 100% free of using fossil energy or City6 wants to use only renewable energy by 2050 at the latest, in both cases DC4Cities can be helpful to reach this goal.

20 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 8 – The planned percentage of renewable in the electricity mix today and in future16 Figure 8 shows the use of renewable energies and the plans of the cities for increasing the amount of renewable percentage in the grid. It is important to notice that cities are not fully in control of their grid. Energy is steadily “flowing”, e.g. it cannot be chosen that only green energy in the grid is used. It is therefore necessary to include higher level administration to reach the goal of the smart cities. Only with higher renewable energy goals by a higher level administration the goals of the cities can be reached. It unfortunately unclear if Brussels and City2 produce that much renewable energy on their own, or if they buy certificates to gain a percentage that high. However, northern cities have a higher portion of renewable energy than western cities and also plan to have a higher rate in the future. This surely is the result of the strong wind and hydroelectricity and their already existing infrastructure in Northern Europe. For the other parts of Europe the amount of renewable energy is currently lower, which is not supportive for the implementation of DC4Cities at the moment, as there is a high probability that no ideal power plan can be created by the software or the DCs are not or hardly able to implement these power plans. So the escalation mechanism of the software will be triggered often. But like Figure 8 shows, there is a high request to increase this percentage especially in northern Europe. E.g. City 6, City 12 and Helsingborg (all located in Northern Europe) target to reach the 100% until 2050 – they have a lot of wind power and thus ideal conditions for DC4Cities. The current percentage and the request to increase this percentage is lower in Central and Western Europe than in Northern Europe. Almost all Smart Cities have the target to reach 20% renewables in their energy mix until 2020 which is a fair foundation to use DC4Cities. This difference in targets between Northern and Central Europe might result from the high availability of renewable energy sources with a lot of hydro and geothermal in the North, additionally to the intermittent wind energy source. It is worth remarking that the high differences observed between cities would need a more detailed analysis in order to evaluate correctly the energy context in these European cities and, in general in smart cities located in each of the four divisions (north, west, east and south). Therefore, this section has to be considered only as an overview. For the focused cities a broader evaluation of the current electricity mix and its trends has been developed.

16 These data are retrieved from the questionnaire filled in by smart city representatives. Unfortunately not all cities answered this question so that we don’t know the plans of e.g. the Western Europe Cities and the Southern Europe City.

21 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 9 – The sunshine hours per year of Smart Cities based on their answer Figure 9 shows the long-term average of the sunshine hours per year of different Smart Cities. With the DC4Cities focus on solar and wind power Figure 9 can say which city fits because of its potential solar energy production to DC4Cities. City3, City1 and City2 have a high amount of sunshine hours which builds a great foundation for the solar energy production and consequently for DC4Cities. The other Smart Cities have a smaller amount of sunshine hours but that does not mean that the Smart City is unsuitable, because the wind power is also important. However, there is no possibility to get yearly constant data. III.2. Conclusion

Based on the previous analysis several results can be concluded. Firstly, DC4Cities might aim at smart cities in Western and Central Europe as they certainly displayed a stronger interest in DC4Cities as the addressed southern smart cities (apart from Barcelona), even though these may have good geographical conditions. This is further corroborated by a remark we received from the single Southern City that answered to our questionnaire and did not even mention the objective of increasing the share of renewable energy sources: its smart city development plans go solely into the direction of ICT based new services and an internet-based life-style. The renewable energy in Northern Europe is mainly based on hydro and geothermal energy, which can be regulated well and makes DC4Cities unnecessary. However, this needs to be analysed individually for each target city, as e.g. Denmark uses a lot of – intermittent – wind energy. Western and Central Europe use more solar and wind energy and have therefore a more fluctuating renewable percentage, which is more suitable for the use of DC4Cities. Secondly, smart cities in Northern Europe have more often their DC outsourced than smart cities in Western and Central Europe – so the procedure of overlooking the optimization targets is a little more difficult. The target audience therefore are mainly IT services in Northern Europe and city governments in Western and Central Europe. This has to be conjured, when the target audience is approached. Thirdly, the current percentage of renewable energy in Central and Western Europe is expected to rise strongly in the future. DC4Cities can be used to support the rise with its positive effects on grid stability and can therefore is the rise of renewable energies. It may also have a big impact on DC costs, as you have seen in the example of Helsingborg’s DCs. And finally, it is necessary to make vertical connections between local, regional and countrywide administration, because the energy mix is not separable and the cities need depend on the grid to reach their goal in increasing renewable energies.

22 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

IV. BARCELONA

IV.1.1. Commercial and Enterprise Data Centres

Throughout this section, we will show information gathered concerning DCs located in Barcelona and its metropolitan area. Because of not having an official registry of the DCs of Barcelona, we have had to rely on sources of information on the internet and specialized magazines. Therefore, the results that we are going to expose are approximate since we do not have a complete list, nor have all the data of the DCs on the list. In addition to drawing up the list, we have sent some questionnaires to DCs managers. The percentage of answers has been very low - about 25%.

IV.1.1.a. Number of Data Centres

In total, we have found 15 companies offering DC's services in any of its forms located in Barcelona or locations nearby. Regarding the number of companies operating their own DC in Barcelona, we do not have a census. However, in this case we calculate that at least 5 of them could fit with the objectives set in this project. According to the available information on the levels of security, most of the DCs are Tier III, both public and private ones. In the case of private DC, we could find some examples of Tier II. This kind of DC is older and its criticality in the continuity of the service is not always very demanding. Public DCs are more modern and have higher levels of security so that they begin to offer Tier IV services.

Tiers

2-Tier 3-Tier 4-Tier

Figure 10 – Distribution DCs Barcelona according Tiers Based on the received questionnaires, public DCs have experienced a 5% increase and expect a slightly higher increase for the next year. This increase is mainly due to customers who had his own DC and have decided to move their services to a public DC. In other cases, this increase corresponds to customers without DC or with their services hosted on other public DCs. The size of the DCs in Barcelona tends to be between 800 m2 and 2100 m2 so we could say that the medium size DC is the most common in this city. We have found a geographical distribution among companies providing services of DCs in its various forms (Figure 11). Nine DCs are located in the city of Barcelona, two are located in an adjacent town to Barcelona (Hospitalet de Llobregat) and four are located 12 kilometres from the centre of Barcelona in an industrial complex between Sant Cugat del Vallès and Cerdanyola de el Vallés.

23 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 11 – Geographical distribution DCs Barcelona Taking into account the collected data about security and size, DC4Cities fits well if we take as a reference the stated classification in the paragraph 3.1.3 of the deliverable D2.3.

IV.1.1.b. Business Models and SLAs

The majority of DC providers in Barcelona choose Cloud services as main model business although they rarely only offer one solution (see Figure 12). Typically, besides Cloud, they also offer managed hosting, managed services or app hosting solutions. We have found two companies that only provide Colocation services and two more that offer Colocation along with other Cloud services.

Business Model 8

6

4

2

0 Colo + Cloud Colocation Cloud Cloud + Others

Business Model

Figure 12 – DCs distribution according their business model Most providers offer neutral carrier services (70%), giving the possibility to the customer to choose their provider of IP services. Regarding SLA guarantees, we have found common aspects that were present in nearly all SLAs: the power availability, room temperature and humidity. We more occasionally have also found aspects related with the electricity supply system. However, never or almost never the guarantee terms refer to efficiency or flexibility, so DC4Cities has in this sense an opportunity to help these companies with these kind of services. Taking into account what we argued in D2.3 on the adequacy of DC4Cities depending on the business model and the data available from the analysed DCs, Barcelona is an ideal target for the implementation of the DC4Cities products.

24 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

IV.1.1.c. Energy Situation of DCs in Barcelona

As already mentioned above, most DCs in Barcelona are medium sized. This is also true if you look at the power consumption data, ranging from 800kW to 2,500kW. All of them have UPS facilities and diesel generators that allow them certain autonomy to possible supply cuts. They have also installed redundant power lines in some cases. For those DCs that supplied us with details about their energy supply contracts, we have observed that they have special prices although the discounts vary according to the customer and the supplier. Of the surveyed, 30% said that its power supply comes 100% from renewable energy sources and that the "certificate of origin" should be delivered by the supplier each year. Due to the low number of completed surveys received, we cannot ensure that this 30% is a reliable value. IV.1.2. Potential for Renewable Power

IV.1.2.a. Energy Context in Barcelona

In order to evaluate the energy context in Barcelona, the following sources were assessed:  Barcelona’s Energy Balance from 2012 [BCN12] This document is published by Barcelona’s Energy Observatory (under the City Hall’s control) using data from Barcelona’s Energy Agency. It provides an overview of the city’s energy consumption by form and use, as well as the situation regarding local generation and self-consumption.  The Energy, Climate Change and Air Quality Plan in Barcelona from 2008 [PEC08] In 2008, the city produced an Energy, Climate Change and Air Quality Plan that analysed the recent evolution, situation and projections, and defined a strategy to reach a set of energy and environmental goals for the period 2011-2020.  Registro de Instalaciones en Régimen Especial17 The Spanish Ministry for Industry, Energy and Tourism holds a publicly accessible registry of special regime (distributed generation: CHP, waste and renewable generation), which provides type of installation, installed power and location for most generation assets in the specified category. Barcelona’s energy consumption per capita (see Figure 13) is amongst the lowest in Europe, partly due to its mild climate (in Barcelona less than 45% of the existing residential buildings have a heating system or cover their heating demand). Nevertheless, the energy consumption ratio is low comparing to other Mediterranean capitals such as Rome and Athens, thus suggesting an interesting level of energy efficiency related to comparable cities. Figure 14 shows the evolution of energy consumption, GDP and population in the city. It can be extracted that energy consumption has grown less than GDP in the last years. Therefore, and also taking into account the increase in population, the chart suggests continuous improvements in energy efficiency.

17 https://oficinavirtual.mityc.es/ripre/informes/informeinstalaciones.aspx

25 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 13 – Final energy consumption per capita in 2012 [BCN12]

Figure 14 – Evolution of Energy consumption, population and GDP [BCN12] Out of the total energy consumption, the most relevant form is electricity (43%), followed by natural gas (36%). The use of electricity has grown strongly in the past two decades due to the introduction of new technologies and equipment, such as the deployment of air conditioning and the increase of economic activity in tertiary buildings. Barcelona’s Energy Plan [PEC08] provides detailed information regarding generation and consumption by sources and technologies for the year 2008. Therefore, data from 2008 will be used whenever recent data is not available. In 2008, the electricity generation plants of Barcelona and the Besòs area generated the equivalent to 68% of the total energy consumption of Barcelona and Sant Adrià del Besòs. Nevertheless, an important fraction of electricity produced is exported to other municipalities and therefore Barcelona relies strongly on electricity imports. From the total electricity consumption, the electricity generation plants of Barcelona and Sant Adrià provided a share of 22% of consumption whereas 78% of electricity consumed was imported. As can be extracted from the figures in the charts (electricity generation (Figure 16) and consumption (Figure 17) mix), most of the electricity produced in the Barcelona bulk generation plants is exported to the industrial site in the surroundings.

26 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Final Energy consumption by form, 2012 (GWh) Petroleum LPG products 187.4 3,472 1% 21%

Electricity 7,163 43% Natural gas 5,802 35%

Electricity Natural gas Petroleum products LPG

Figure 15 – Distribution of final energy consumption form [BCN12] Combined cycles produced in 2008 4,876 GWh and only 1,273 GWh were consumed within the city. This is due to the distribution grid architecture: the power line to which some of the combined cycle plants are connected does not feed all of Barcelona, and therefore not all the energy can be consumed within the city.

Electricity generation mix in Barcelona Combined Cycle Fuel/Gas

PV 3.3% 3.6% CHP 89.9% 6.8% 0.0% 4,876.2 Mini-Hydro GWh 3.1% 0.1% Biogas & Solid waste

Figure 16 – Electricity generation mix in Barcelona, 2008 [PECQ08] Regarding RES consumption18, only a share of 4.8% of the total electricity consumption in Barcelona came from renewable and waste sources in 2008, which corresponds to 361.8 GWh. This was a low value in comparison to the average electricity mix at national level, which was approximately 16% renewable, and is caused by the predominance electricity from nuclear plants, which although are located next to Barcelona offer generation services at national level, in comparison to other sources, as hydro and wind generation. This share increases up to 11% considering also other DER technologies, i.e. adding cogeneration.

18 Residual heat from waste incineration and biogas are also considered.

27 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Electricity consumption mix in Barcelona Nuclear Combined Cycle 16.9% (natural gas) 1,273.7 GWh Hydro

CHP 2.7% 1.2% 6.1% 0.5% Fuel/gas 10.9% 0.1% Coal 68.2% 4.0% 0.1% Wind 0.1% Solar

Urban waste

Figure 17 – Electricity consumption mix in Barcelona, 2008 [PECQ08]

Spanish electricity mix Combined Cycle Nuclear Coal 17.0% 0.5% 8.8% Wind 8.1% Hydro 18.4% CHP 22.1% 1.0% Solar 11.4% 0.9% 0.7% Biomass 33.2% Waste Fuel/gas

Figure 18 – Spanish electricity mix, 2008 [PECQ08] The following table summarizes Barcelona’s electricity situation in terms of self-generation and renewable energy usage. Table 11 – Summary of the current electricity situation in Barcelona

Barcelona’s electricity consumption (GWh) 7,529.13

Imported electricity (GWh) 5,892.27

Self-produced electricity consumed within the city (GWh) 1,636.86 (21.7%)

Renewable electricity consumed (GWh) 361.8 (4.8%)

Renewable electricity generated within the city (GWh) 202.9 (2.7%)

DC4Cities aims at integrating intermittent renewable sources such as PV and wind energy. From the total renewable electricity consumption in 2008, 7.6 GWh came from PV installed within the city and 37.7 GWh came from wind energy located out of the city.

28 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

IV.1.2.b. Regulatory framework and action plans by Public Administrations

IV.1.2.b.1. Local goals and action plans As a complement to the [PEC08] and updating the previous energy plan (PMEB 2002-2010) Barcelona’s Energy Agency (which belongs to the City Council) has established an energetic self-sufficiency plan with the aim of increasing the renewable production within the city in 2024 up to a 10%19 of the total energy consumption [EBA15]. This should be achieved through energy efficiency and the installation of small-scale renewable generation plants all over the city (including PV, solar thermal, and others). This is a significant increase from the current value, which was slightly over 2% in 2012. As shown in D2.3 section 4.2.5, the local Energy Agency has assessed the energetic resources for small wind and photovoltaic energy, and it provides public information of the most suitable locations for PV and wind power plants (see following figures).

Figure 19 – Wind generation potential in Barcelona20 The wind generation potential is highest in the outer parts of the city, since wind speed is reduced by obstacles (buildings) as air currents advance through the city. Solar PV generation potential, on the other hand, depends on the height and distribution of the buildings. The highest potential corresponds to rooftops that do not have any shadows casted by surrounding buildings. Potential for this renewable source is high, as can be observed in the following figure. In order to accomplish the City’s self-sufficiency goal, on the one hand, the Local Energy Agency is promoting the energy efficiency and the usage of renewable energies in public buildings, as shown in the examples below. The City Council has carried out renewable generation projects regarding distributed PV generation. City projects are aiming at improving self-sufficiency in public buildings and equipment. The following figure shows the location of some of the different PV plants installed by the local administration. Consuming locally generated electricity reduces grid usage, thus eliminating transportation and distribution losses and reducing stress on the grid, and therefore postponing the need for investments. Furthermore, it contributes to simpler grid management by reducing congestion.

19 This figure considers local energy production only energy produced from local sources, such as wind, solar, residual heat from waste incineration, and biogas. The two latter can be considered local, since this waste is generated within the City (or metropolitan area). Local generation does not include generation using imported fossil fuels such as natural gas. 20 https://ajuntament.barcelona.cat/autosuficiencia/es/webapp.php

29 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 20 – Photovoltaic generation potential in Barcelona21

Figure 21 – Location of different PV projects carried out by the City Administration22 The following figures illustrate examples of self-sufficiency projects carried out in the city. The figure below shows a view of the PV panels in the cemetery of Les Corts, in Barcelona (Figure 22). In this case, the PV panels in this location are part of a larger project including efficiency improvements in cooling, solar thermal energy usage, as well as electricity and thermal (heat and cold) storage. Electricity storage in batteries aims to maximize self-consumption.

21 https://ajuntament.barcelona.cat/autosuficiencia/es/webapp.php 22 http://www.tersa.cat/es/instalaciones-fotovoltaicas-de-barcelona-ciudad_2930. October 2015

30 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 22 – PV panels in Les Corts cemetery, Barcelona23 The project was carried out by the local Energy Agency. This is the first self-sufficient public building in Barcelona. The City Hall’s Energy Agency has promoted the installation of PV panels in other public buildings, such as public schools and libraries. The figure below shows PV panels integrated into a public library’s roof.

Figure 23 – PV panels in Joan Miró library, Barcelona The Metropolitan Area of Barcelona has recently approved the launch of preliminary studies aiming to create a metropolitan energy operator, which would be in charge of managing all metropolitan energy assets, including waste and water treatment plants. Further responsibilities would include the planning and management of investments in generation assets and efficiency improvements. Further actions by Barcelona’s Energy Agency include the installation of solar thermal and geothermal assets to reduce the consumption of imported energy (mostly natural gas or fossil- fuel generated electricity) and increase the use of renewable energy. On the other hand, local environmental legislation is being adapted to support the usage of renewable energies in buildings and public spaces. For example, since 2011 the local environmental legislation establishes an obligation regarding the installation of PV panels in new buildings (and in some cases, existing buildings whenever a relevant refurbishment takes place) when technically feasible and if enough solar resource is available. This regulation is

23 http://premsa.bcn.cat/wp-content/uploads/2014/09/140926_DOSSIER_CementiriAutosuficient.pdf

31 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities mainly aimed at tertiary buildings such as offices, shopping malls, sports centres, and other spaces that are potentially open to the public; and does not affect residential buildings. Legislation requires the installation of 7 Watts-peak per square meter of construction dedicated to specific uses, and larger than a certain size depending on the use. As previously mentioned, the energy efficiency increase is key for the accomplishment of the city goals. The Coty Hall’s Energy Agency has also launched initiatives in this field. A representative example is the development of a Home Energy Management System (HEMS) for private dwellings. This tool has been recently offered to a group of citizen without charge and has the aim of reducing the energy consumption through the provision of information and advices to tenants.

IV.1.2.b.2. National and regional regulation Over the past decade, the different governments have performed deep and relevant changes in the electricity sector legislation, with a strong impact in renewable generation deployment and self-consumption.

RD RD RD RD RD 661/2007 1/2012 9/2013 413/2014 900/2015

- RD 661/2007 established a favourable legislation at national level, with effective retribution schemes to favour special regime (most renewables, waste and CHP) generation. This Royal Decree has enabled investments and, consequently, the share of these technologies in the electricity mix has grown significantly. This has led to an increase between 2008 and 2012 of the special regime electricity production by more than 30,000 GWh. - RD 1/2012 suspended retribution schemes for new plants, discouraging new investments and creating uncertainty in the electricity sector. This law was published in a context of economic crisis and was the consequence of a strong imbalance between the electricity system costs and incomes. - RD 9/2013 is the current electricity sector law, which establishes, besides other changes, new retribution schemes for special regime plants. This change in retribution had retroactive effects for existing plants, thus increasing legal uncertainty. - RD 413/2014 provides the retribution values for the new schemes, which are significantly lower than in the previous situation (RD 661/2007). - RD 900/2015 establishes a new technical and economic framework for special regime plants, but only for those directly connected to the end-consumers for self- consumption. This new regulation reduces uncertainty introduced since 2012 for this kind of plants, which will be normally in the small-medium power range. Nevertheless, it may discourage investment for economic reasons: most of the self-consumers that install a RES power plant must make variable payments depending on the energy consumed from their own plants, therefore, energy expenses savings are reduced. Moreover, for some specific cases, they receive no payment for electricity exported to the grid. However, this framework is more favourable compared to the laws introduced in the last three years. In conclusion, current national legislation introduces economic barriers to RES deployment, and may obstruct some of the established goals at city level, or increase the level of local public support, regulation or investment necessary to reach the established targets. However, the elimination of the legal uncertainty introduced since 2012 may have a positive impact on investment. There is no relevant legislation at regional level for Barcelona.

32 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

IV.1.2.c. Current situation on the renewable energy production and projection to 2020

IV.1.2.c.1. Current situation Barcelona’s Energy Balance of 2012 has been examined, and the following data has been obtained: Table 12 – Barcelona current situation [BCN12]

PV energy production 2012 (GWh) 15.48

PV installed power 2012 (kWp) 12,388

At the time the document was published, no wind power plants existed within the city, and as of 2014 according to the RIPRE24 there is a single wind generation installation within the City, with a total power of 10 kW. The following table shows the installed PV plants by power range, according to the RIPRE. Most of the installed power (93%) corresponds to plants with power greater than 20 kW25. Table 13 – Distribution of installed PV power by plant size24

Power Range Percentage Percentage over over number of installed power plants 100 – 2,000 8% 71% kW 50 - 100 kW 12% 13% 20 - 50 kW 23% 9% 1.7 - 20 kW 57% 7% The following chart (Figure 24) shows the evolution of installed PV power in Barcelona, distinguishing between municipal (blue) and non-municipal (red) up to 2014. The chart shows a slow and steady growth in installed power in the period 1999-2007, with a boom in the period 2008-2011 due to great regulatory support. The fraction of municipal PV installed power from 2009 to 2014 is estimated using data from previous years and performing estimations based on the level support from public administrations. The chart in Figure 24 shows a stagnation in installed power from 2011 due to the change in regulation. Nevertheless, although recently established regulation has reduced rewards, legal uncertainty has decreased. Moreover, the energy self-sufficiency plan establishes a favourable scenario, since the local authorities will support new PV power plants. In conclusion, this could have a positive impact and it is expected a PV growth in the future in accordance with the city goals.

24 https://oficinavirtual.mityc.es/ripre/informes/informeinstalaciones.aspx 25 There can exist not significant variations, since not all the small power plants (<10 kW) are registered since 2011, since the regulation has been modified.

33 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

14,000 12,000 10,000 8,000

kWp 6,000 4,000 2,000 0 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014

Municipal PV [kWp] Non-municipal PV [kWp]

Figure 24 – Evolution of installed PV power in Barcelona

IV.1.2.c.2. Growth in the period 2014-2020 According to the Energy, Climate Change and Air Quality [PEC08] goals, the energy generation from PV and mini-wind power plants will increase strongly to fulfill the 2020 goals (340% increase from 2008 situation). Supposing a constant number of equivalent generation hours (energy output per kW), the PV installed power can be escalated proportionately to the energy generation. No wind generation took place in 2008 within the city, and the target for 2020 is 0,04GWh/year [PEC08]. This value can be considered insignificant compared to the generation expected for PVs. Table 14 – Expected PV output in 2020 [PEC08]

2008 2012 2020

Electricity consumed 7,531.8 7,162.8 9,837.1 (GWh)

Electricity generated 7.62 15.48 33.12 (GWh)

Installed power (kWp) 6,116.5 12,388 26,585

These projections expect a 10% yearly growth in installed power for the period 2012-2020. Stagnation in the past years due to regulation will have to be compensated with higher growth rates in the following years to reach the target (14% in the period 2014-2020). A set of representative daily profiles has been developed for the second phase of the DC4Cities trials in Barcelona. Specifically, three types of days have been represented: sunny, variable and cloudy. Figure 25 below shows the trial profiles escalated to match the expected installed power in 2020 in Barcelona. In a sunny day, the peak power output of all the PVs combined would be approximately 17 MW.

34 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

18

16

14

12

10 Sunny

8 Variable

Power(MW) 6 Cloudy

4

2

0 1 3 5 7 9 11 13 15 17 19 21 23

Figure 25 – Expected PV typical profiles in 2020 This peak PV output (17 MW) is small compared to the city’s consumption (roughly 1%), and therefore will probably not be a problem to integrate using tools such as DC4Cities, other demand response mechanisms, and energy storage. Table 15 – Expected Barcelona power consumption in 2020

Barcelona yearly electricity consumption 2020 9,837.1 (GWh) [BCNEB12]

Barcelona’s average power consumption (MW) 1,123

Barcelona’s peak power consumption estimation 1,684 (150% of average power) (MW)

Taking into account the evolution of PV installed power in previous years, the current distribution of installed power regarding plant size and the set targets for 2020 (Table 15), an estimation regarding the number and size of PV installations in 2020 has been performed (Table 16): Table 16 – Expected distribution of PV installed power by plant size in Barcelona, 2020 Power Range Number of Total plants Power (kW) 100 – 2,000 kW 30 18,875 50 – 100 kW 44 3,456 50 – 20 kW 85 2,393 20 – 1,7 kW 211 1,861 Total 369 26,585

Besides the increase of PV generation within the city, a large increase in wind power generation is expected until the same horizon (2020), as shown in the chart below. The weight of wind power in the Catalan electricity mix is expected to grow to 18% in 2020. This means that an important fraction of electricity will come from intermittent sources, which may be especially active at times of low demand and price (i.e. at night). In the case of Spain and many liberalized electricity markets, electricity price internalizes the system situation regarding demand and supply at every moment. In a situation of relevant unmanageable

35 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities energy amounts in the grid at times of low demand, price would decrease. Therefore, DC4Cities can contribute to integrate local PV through self-consumption mechanisms, and wind power from the grid using a price-based optimization.

100% 1% Other RES 80% 18% PV 60% Hydro 40% Wind 20% Other non-RES 0% CHP 2008 2020

Figure 26 – Expected evolution of the regional energy mix [PEC08]

IV.1.3. DC4Cities potential in Barcelona

Following the methodology described in I.3. opportunities of installing DC4Cities in Barcelona have been explored. As the main goal of this market analysis is to size the market for each city, in order to evaluate the feasibility of introducing our product in Barcelona and its competitiveness in front of competitors, it is needed to evaluate not only the energetic impact of DC4Cities, but also the economic impact. At this stage of the project, there is no definition about the price of our solution. Therefore, a preliminary evaluation will be performed estimating the maximum investment that a DC could assume considering the yearly energy expenses savings achieved thanks to DC4Cities. It is worth noting that the introduction of DC4Cities into the market could be due to other factors, as e.g. the need of accomplishing or being aligned to the city energy action plans for both, private or public DCs. This accomplishment could in some cases be an obligation, through new local/national environmental legislation, and in some others the result of participating in a public procurement process, e.g. if the DC is interested in providing services to the city council. Furthermore, for private DCs the driver could be its own environmental strategy. As has been described in IV.1.2.b.1. Barcelona aims at increasing its self-sufficiency through the promotion of local renewable power plants and the increase of the energy efficiency and, thus, DC4Cities is a product totally aligned with these objectives. Hereafter the main steps and results to size the market for DC4Cities in Barcelona as well as the assessment of the impact are presented: Step 1: Selection of a representative DC In order to select a representative DC for testing the consequences of installing DC4Cities, the questionnaires collected by IMI have been examined. The DC selected has the following main characteristics: Table 17 – Barcelona’s representative DC main characteristics Business Property Size Yearly consumption Indicators Model kW consumed kWh consumed Cloud/ Private IT provider, leader Around 1,049 9,183,405 Colocation in the Spanish market and 1,000 offering worldwide services racks and & projects. 1,700 m2. Based on the information collected in the questionnaire and the experience of Barcelona trials, it is considered a flat IT consumption profile, which means an hourly consumption of 1,048.33 kWh.

36 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Step 2: Information about the flexibility and energy efficiency improvement The selected DC has a cloud and colocation oriented business model. As described in the basic methodology for all focus cities, a flexibility in the range of 5-15% is assumed for cloud oriented business models. The range increases up to a 20% for colocation oriented business models. Considering this, a shiftable load equal to 10% has been chosen, since this is an average value. Regarding energy efficiency improvements, Barcelona trial results are considered as representative. Thus, energy savings achieved with DC4Cities will be equal to 70%. Table 18 – DC yearly energy consumption and potential savings Yearly consumption Indicators Maximum shiftable consumption DC4Cities final energy savings 918,340 kWh/year 642,838 kWh (≈ 100 kWh/h) Step 3: Information about the renewable sources availability and energy prices According to the current energy context in the city and the trends for the following years, local authorities in Barcelona will support the installation of photovoltaic power plants for self- consumption. For this reason, it has been assumed that the DC in the DC4Cities scenario will be fed by two different energy sources: a local photovoltaic renewable plant installed on the roof and electricity coming from the grid. For the baseline scenario, the DC only consumes electricity coming from the grid. Local renewable energy availability In order to size the photovoltaic power plant, data provided by the Local Energy Agency (see Figure 27) has been used. The figure shows the location of the DC26. The areas marked in dark red represent the areas with a high solar irradiation, thus, this DC has a suitable location for installing this kind of installations.

Figure 27 – Solar irradiation potential27 The following table summarizes the main characteristics of the simulated PV power plants: Table 19 – PV main characteristics

Roof Auxiliary Module Modules Module Module Total usable space (55% surface surface units peak peak usable) (m2) (m27unit) (m2/unit)

26 Due to the need of maintaining the confidentiality of the DC, the exact location is not mentioned here 27 http://ajuntament.barcelona.cat/autosuficiencia/es/webapp

37 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

space power power (m2) (Wp) (kWp)

2,384.64 1,311.55 1.7 1,073.09 631 250 157.75

PV hourly production for this example has been determined using a simulation tool28. There has been selected the most frequent type profiles, one representing a sunny day and another representing a cloudy day29 (see Table 20).

Table 20 – PV type profiles

Sunny 700.94 0 0 0 0 0 0 0.64 7.63 29.46 57.39 79.78 93.45 100.27 98.49 88.84 72.02 47.85 20.63 4.18 0.30 0 0 0 0 day kWh Cloudy 189.21 0 0 0 0 0 0 0 0 2.27 13.37 22.36 29.76 34.33 32.90 27.15 18.99 8.03 0.06 0 0 0 0 0 0 day kWh

Renewable energy availability from the grid As has been previously explained, the weight of renewable energies in the electricity mix in Barcelona is lower than the average in Spain and this is because of the high concentration of bulk generation within the city and in the surroundings. In accordance with [PECQ08] wind energy is the most used renewable energy in Catalonia and its weight in the Catalan electric mix grid in 2015 was around 14%. For this reason, RES type profiles defined for the grid will be based on typical wind energy profiles. It is not possible to select wind profiles only for the Barcelona or Catalonian grid, since they are not available. Hence, RES type profiles will be calculated for the national grid, i.e. using the same approach as for Barcelona trial (see D6.2). The source is the Spanish System Operator, Red Eléctrica Española. Wind energy availability normally increases at valley hours, i.e., when electricity demand decreases. There are two different common patterns. Wind energy production weight is higher during valley hours, either due to a higher production in energy terms or to a higher production in relative terms, since demand is much lower.

Figure 28 – Wind energy type profiles30 The following Table 21 shows the profiles that have been selected as representative: Table 21 – %RES type days for Barcelona mix grid

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

%RES BCN 30% 30% 30% 29% 29% 29% 28% 26% 24% 23% 23% 23% 25% 25% 27% 27% 27% 27% 26% 24% 23% 24% 26% 28% grid 1 %RES BCN 34% 35% 34% 34% 34% 34% 33% 31% 28% 27% 26% 26% 27% 27% 28% 29% 29% 29% 27% 26% 26% 28% 30% 32% grid 2

28 Software PVsyst 6.3.0 29 Since these profiles have been developed for the the secong phase of Barcelona trial, more details will be provided in D6.3. 30 https://demanda.ree.es/eolicaEntreFechas.html

38 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

The following Table 22 shows the distribution of the profiles considered for the yearly simulation: Table 22 – Type profile weight

Grid type profiles PV type profiles Weight

Day 1 BCN grid 1 Sunny 50.0%

Day 2 BCN grid 2 Cloudy 50.0 %

Energy prices Taking into account this consumption (big electricity consumer), it can be stated that this DC will have a dynamic pricing tariff, either hourly energy prices indexed to the wholesale or a 6- period tariff. It is assumed that the DC has a 6-period tariff. The following tables show the different energy prices patterns by day and month: Table 23 – Energy prices distribution

Table 24 – Energy prices TV Period (€/kWh) P1 0.121253 P2 0.100387 P3 0.094028 P4 0.079584 P5 0.076415 P6 0.064674

In order to calculate an average price representative for the whole year, the different daily profiles for each period have been weighted. Table 25 – Yearly energy prices simulation

P1 P2 P3 P4 P5 P6 €/kWh 0 0% 0% 0% 0% 0% 100% 1 0.06467 1 0% 0% 0% 0% 0% 100% 1 0.06467 2 0% 0% 0% 0% 0% 100% 1 0.06467 3 0% 0% 0% 0% 0% 100% 1 0.06467 4 0% 0% 0% 0% 0% 100% 1 0.06467 5 0% 0% 0% 0% 0% 100% 1 0.06467 6 0% 0% 0% 0% 0% 100% 1 0.06467 7 0% 0% 0% 0% 0% 100% 1 0.06467 8 0% 20% 0% 16% 12% 52% 1 0.07561 9 0% 20% 8% 8% 12% 52% 1 0.07677 10 12% 8% 8% 8% 12% 52% 1 0.07927

39 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

11 20% 0% 8% 8% 12% 52% 1 0.08094 12 20% 0% 8% 8% 12% 52% 1 0.08094 13 8% 12% 8% 8% 12% 52% 1 0.07844 14 8% 12% 8% 8% 12% 52% 1 0.07844 15 8% 12% 0% 16% 12% 52% 1 0.07728 16 8% 12% 8% 8% 12% 52% 1 0.07844 17 8% 12% 8% 8% 12% 52% 1 0.07844 18 20% 0% 8% 8% 12% 52% 1 0.08094 19 12% 8% 8% 8% 12% 52% 1 0.07927 20 12% 8% 8% 8% 12% 52% 1 0.07927 21 0% 20% 8% 8% 12% 52% 1 0.07677 22 0% 20% 0% 16% 12% 52% 1 0.07561 23 0% 20% 0% 16% 12% 52% 1 0.07561

Step 4: DC4Cities impact assessment In the baseline scenario without DC4Cities, applying the assumptions described previously in Table 26, DCs RenPercent is equal to 28.07%, which implies approximately the emission of 2,035.22 tCO2. Moreover, energy expenses have been estimated to be 677,051 €. Table 26 – Baseline scenario energetic, environmental and economic balance

Case Consumption (kWh) Renewable (kWh) RenPercent (%) tCO2 Cost (€)

Baseline 9,183,405.00 2,577,770.79 28.07% 2,035.22 677,051.86

The following figure compares the DC total electricity consumption with the renewable electricity consumption, which follows the wind energy availability pattern.

BCN grid 1 BCN grid 2

100% 100% 1,000 1,000 90% 90% 80% 80% 800 800 70% 70% 600 60% 600 60% 50% 50% 400 40% 400 40% 30% 30% 200 20% 200 20% 10% 10% 0 0% 0 0% 1 3 5 7 9 11 13 15 17 19 21 23 1 3 5 7 9 11 13 15 17 19 21 23

%RES IT Consumption (kWh) %RES IT Consumption (kWh)

Figure 29 – Baseline scenario DC consumption vs %RES The implementation of DC4Cities allows to increase the energy efficiency and to adapt the consumption pattern to both, local and grid renewable availability. The following figure shows a graphical representation of the main changes that may appear. It can be observed that due to the high energy intensity of a DC, PV energy production is very low in comparison to the consumption, and therefore, adaptation will be mainly done in accordance to the grid profiles. In order to evaluate the electricity consumed from the grid before and after DC4Cities, the light and dark blue lines have to be compared. As can be observed, after implementing DC4Cities, consumption is going to be lower during peak hours, since electricity from the PV is consumed and flexible workload is shifted to night periods, when more wind energy is available. It is worth noting that there is no need to increase the power contracted to the DC, even though workload is more concentrated in a shorter timeframe.

40 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

BCN grid 1 BCN grid 2

1,200 100% 1,200 100%

1,000 1,000 80% 80%

800 800 60% 60% 600 600 40% 40% 400 400

20% 20% 200 200

0 0% 0 0% 1 3 5 7 9 11 13 15 17 19 21 23 1 3 5 7 9 11 13 15 17 19 21 23 Photovoltaic (kWh) Photovoltaic (kWh) %RES BCN grid 1 %RES BCN grid 2 Grid consumption (kWh) Grid consumption (kWh) Non-flexible consumption (kWh) Non-flexible consumption (kWh) IT Consumption (kWh) IT Consumption (kWh) Figure 30 – DC4Cities scenario DC consumption vs %RES RenPercent increases up to 29.6%. Although this could be understood as a low improvement, the energetic and environmental impact of DC4Cities is high. The electricity consumption decreases approx. 7% and also 181.94 tCO2 emissions are avoided. Moreover, energy expenses savings are estimated to be 62,249.13 €. Table 27 – DC4Cities scenario energetic, environmental and economic balance

Consumption Renewable RenPercent Case tCO Cost (€) (kWh) (kWh) (%) 2

PV&DC4Cities 8,540,566.65 2,527,855.80 29.60% 1,853.28 614,802.73

According to the Barcelona energy context, a high growth in the electricity generation from RES is expected until 2020, therefore, DC4Cities impact has been also assessed considering this future scenario. The following table summarizes the main results. As can be observed, in a scenario with a higher penetration of wind energy, improvements increases significantly. Table 28 – 2020 scenario energetic, environmental and economic balance

Consumption Case Renewable (kWh) RenPercent (%) tCO Cost (€) (kWh) 2

Baseline 2020 9,183,405.00 3,437,027.72 37.43% 2,028.00 677,051.86

PV&DC4Cities 8,540,566.65 4,333,064.88 50.74% 949.53 614,802.73 2020

The following Table 29 shows the summary of the main results. Energy savings as well as CO2 emission reductions are relevant, it can be stated that the impact of DC4Cities is high. Moreover, the energetic DC invoice is also significantly reduced. It is noteworthy that the variable payments due to the new regulation that affects PV have not been considered, since the effect is low in comparison to other variables. Table 29 – Summary of the results obtained for the representative DC

Energy Economic Economic RenPercent tCO2 savings savings savings (10 increase decrease (kWh) (€) years) (€)

41 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Baseline VS DC4Cities+PV 642,838.35 1.53% 181.94 62,249.13 622,491.31

Baseline VS DC4Cities+PV 642,838.35 13.31 % 1,078.5 62,249.13 622,491.3131 2020

As mentioned before, at this stage of the project the investment needed for applying DC4Cities is not known. However, the energy expenses savings order of magnitude in 10 years is considered high enough as to amortize the investments for both, the PV power plant and DC4Cities. In order to make a sensitivity analysis, the case with the minimum flexibility in the range assumed for each business model is supposed (5%). Moreover, with the aim of evaluating the less favourable case, it is supposed that DC4Cities energy savings could decrease up to a 50% in comparison to trial results. With these assumptions, energy savings, in comparison with baseline results, are 223,585.13 kWh with a decrease in CO2 emissions of 89.98 tonnes and a RenPercent increase of 1.44%. In this unfavourable case, economic savings in ten years are estimated in 314,391.27 € Step 5: Extrapolation at city level The extrapolation at city level is not direct, since the data obtained does not allow to estimate the global impact at city level. However, a preliminary order of magnitude can be given. As explained above, there are around 15 companies that offer public DC services in the Barcelona Metropolitan Area and most of them offer cloud and/or colocation services, although together with other business models. DCs owned by private companies have been discarded for this analysis, since an evaluation case by case would be needed. Although the characteristics and size are different for all the public DCs, some similarities have been found:  Size: from 800 m2 to 2,000 m2. Therefore, the size is variable in a limited range. The selected DC is in the upper part of this range.  Business models: most of them offer cloud services, but together with other business models. There are also a couple of colocation oriented DCs. Hence, most of the DCs are able to partially adapt its energy consumption, since some of the services they provide are cloud services. Since all the DCs offer cloud or collocation services, the flexibility selected for the representative case can be also representative for the other DCs in Barcelona. However, the size is more variable and it is not known if most of the DCs are in the lower or in the upper part of the range. Taking into account this values, it is selected a maximum extrapolation factor equal to 5, i.e., considering an average impact equal to one third of the impact achieved for the representative example. Table 30 – Summary of the results obtained at city level Energy savings RenPercent tCO2 Economic Economic savings (MWh) increase decrease savings (k€) 10 years (k€) Baseline VS 1.53% 909,7 DC4Cities+PV Baseline VS 3,214.19 311.24 3,112.46 DC4Cities+PV 22.67% 5,428.45 2020

31 No Energy prices projections have been performed. Therefore, energy expenses savings have no variations.

42 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

IV.1.4. Summary

Barcelona can be considered as a city with a high potential for applying DC4Cities. In general the impact of changes in the energy consumption of DCs thanks to DC4Cities can have a huge impact. The total impact at city level have been estimated in 3,214.2 MWh (≈0.04% from the total Barcelona’s electricity consumption). From the smart city point of view Barcelona is also an ideal place to apply DC4Cities, even though considering the existent barriers to RES at national level. Barcelona’s Local Energy Agency has launched an energy self-sufficiency concept with the aim of increasing both the usage of local renewable energies and the energy efficiency, and therefore, specific actions plans are being carried out. Some of the actions are in line with the definition of the EMA-SC describes in previous deliverables, as e.g. the recently approved study aiming to evaluate the creation of a metropolitan energy operator. Furthermore, projections states that wind energy coming from the grid will also increase significantly up to 2020. This is a favourable scenario for applying DC4Cities. Global energy efficiency improvements at DC level has been estimated to be 7% and renewable usage shares also increases significantly in contexts with a high availability of renewable energy. For example, for the 2020 energy context projection, RenPercent could be improved from ≈37.5% to 50.7% applying DC4Cities and installing a new PV power plant. Most of this impact comes from DC4Cities. For example, energy savings from the PV represents only the 2% of the total energy savings achieved, since PV power is low in comparison to the DC power consumption. Moreover, DC4Cities can have a valuable impact to reach energy expenses savings, since consumption can be shifted to valley consumption hours, in which the availability of wind energy is higher and energy prices are lower. This also has a positive effect to the distribution network, since grid congestions could be reduced. Energy expenses savings at city level has been estimated in 311 k€/year. In ten years, the amount is presumed to be 3.1 M€. Concerning DC flexibility and energy efficiency achievable improvements, due to an unavailability of public information, generic data have been used. The sensitivity analysis done evaluates the impact of modifying significantly the hyphotesis considered. In a more unfavourable scenario, with only a 5% of flexible workload and an improvement of the energy efficiency equal to a 35%, although the amount of energy saved in comparison with total consumption could seem small (223 MWh/year for a single DC, ≈2.5%), the results are still relevant. The savings obtained at city level will decrease only to 1,117.93 MWh/year (≈ 0.02% from the total Barcelona’s electricity consumption). Furthermore, similar conclusions could be extracted for economic savings obtained in ten years thanks to DC4Cities, in this worse case, 300 k€ savings could be reached just for the DC evaluated and the amount at city level will be 1.57 M€. Regarding the impact achieved in the renewable energies usage, although in relative terms RenPercent improvement in the DC would be quite similar with a lower flexibility, in absolute terms the impact would decrease similarly to energy savings. It is worth remarking that these values have to be considered only as a preliminary order of magnitude. The analysis aims at identifying the potential market in the city, taking into account the impact that could be achieved, and not at evaluating detailed energy and emissions savings. It would be required to have a broader study in order to confirm if this impact could be achieved, and this is out of scope for this deliverable. This is also applicable to energy expenses savings.

43 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

V. OPTIONS FOR DC4CITIES IN SELECTED SMART CITIES V.1. Amsterdam

The market analysis of the city of Amsterdam consisted of an analysis of the city level objectives, the DCs profiles and business models.

V.1.1. Public Strategy and Options

By 2025, the city wants to have realized a CO2 emission that is 40% less than in 1990. They will obtain this ambition through saving energy, by locally generating, storing and using renewable energy, and by efficiently dealing with fossil fuels. The general rules of the Activities Decree Environmental Management apply to the vast majority of DCs. This decree contains environmental rules for business activities. Examples are the demands that are placed on energy consumption, and the prevention or limiting of noise pollution and air emissions. Amsterdam and its surrounding region have Climate and Energy Objectives [CEO 13, p.29]. The Amsterdam Structural Vision indicates that - in order to achieve these objectives - "increasing the energy efficiency of the industry and the greening of the ICT sector is one of the transitional paths on which the municipality is cooperating with citizens, businesses and institutions." This transitional path consists of at least the following components:

1. Saving energy (think: energy-efficient software and servers; modular building of server rooms and related facilities; separating hot and cold corridors; raising the temperature in the server room; applying evaporative cooling and cooling wheels) 2. Switching to renewable energy (think: wind, solar and biomass energy and hydropower) 3. Efficiently using residual (fossil) energy (think: use of thermal storage and cooling storage; maximum use of systems that run on fossil fuel; reusing residual heat from installations and electricity induction)

A DC provides electricity-intensive services, in which the central use of electricity increases, whilst the decentral electricity consumption of customers decreases since the applications, the supporting resources and their consumptions are moved to the DC. Reducing electricity consumption is a matter that concerns the entire chain - from software, hardware to the building. The Consortium Green IT Amsterdam Region is the platform that takes this greening to hand in and around Amsterdam. Using green power is a unique selling point for some DCs, whilst unimportant for others. Researches from the University of Amsterdam and CE Delft indicate that the joint purchasing of green current for DCs can be attractive and leads to less CO2 when it concerns the actual addition of (local) production. Purchasing electricity on the basis of Guarantees of Origin (GoOs) is a less good option since, as a result of imperfections in the market, the purchase of GoOs does not in all cases lead to more sustainable production capacity. The parties Green IT Amsterdam, University of Amsterdam, SARA, TNO, SURFnet, SURF and NL Agency (2013) [CEO 13] have done research showing that, with nuance, the principle of “produce and consume locally as much as possible” also goes for the choice of transporting data and energy from DCs with regard to cloud computing. The consortium states that: “An important conclusion from the study is that the use of locally produced green energy has the largest positive sustainability impact. When using a DC abroad

44 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities that uses green energy, it turns out that the energy needed for the transport of data has a significant CO2 footprint. In many cases, it is more effective to do computing and storage in a Dutch data centre, using local or imported green energy. Moving data through the cloud to green data centres abroad is therefore not necessarily more sustainable than local storage.") [CEO 13, p.31]. “The Amsterdam Energy Strategy [AES 10] foresees an important role for solar and wind energy to make the Amsterdam energy consumption more sustainable and to make the city less dependent upon the increasingly expensive imports of fossil energy. Residents of the city are encouraged to invest in solar panels on their roofs, and co-invest on short term in windmills on the outskirts of the city.” [CEO 13, p.32]. In 2015 the subsidies have been stopped but the payback period has shortened which makes the investment still interesting. Amsterdam businesses can give the energy strategy a strong boost if they invest in solar systems on their roofs and in windmills (in the Netherlands). When setting up, DCs can discuss the possibilities with the experts of the municipality and Green IT Amsterdam.

V.1.2. Commercial and Enterprise Data Centres

V.1.2.a. Number of Data Centres

Amsterdam is the third strategic DC hub in Europe [CBR 01]. In 2013, there were about 63 DCs in the Amsterdam region, of which approximately 40 are located within the city boundaries of Amsterdam. They are located near universities and research laboratories (scientific computation), near internet exchanges such as AMS-IX or close to large businesses as enterprise DCs (Police Computer Centre, ABN AMRO, Nuon and the Computer Centre of the Municipality) [CEO 13] (see Figure 31). Amsterdam is particularly known as internet exchange hub.

Figure 31 – Locations of DCs in Amsterdam [CEO13], p.8

In the Netherlands, there is approximately 200,000 m² GFA (Gross Floor Area), of which more than 140,000 m² GFA is located in Amsterdam. Based on figures from BroadGroup and CBRE [CBR 15], Vattenfall issued the following forecast for growth in the Dutch market: the number of square metres will increase to 385,000 in 2016 (versus 200,000 m² in 2013). Table 31 shows the number of data centers divided in bins based on the floor area.

45 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Table 31 – Data centers floor space [CEO 13, p.10]

V.1.2.b. Business Model and SLAs

Based on the information provided by Data Center Map32 there are 38 colocation DCs and 14 cloud providers. Some of these DCs are owned by the same company and therefore the sum in the DC information table in the annex is different. The Amsterdam market is determined by six large, world-leading parties, in DC sizing; Equinix, Global Switch, Terremark, Interxion, TelecityGroup, and Digital Realty. In addition to this, Dutch parties, such as EvoSwitch, KPN and Switch, are active as well [CEO 01]. The table in the annex (cf. Section VIII.3. ) gives an overview of the data centers located in Amsterdam. V.1.3. Potential for Renewable Power

For the municipality of Amsterdam, the environmental department North Sea Canal Area has calculated that the 40 housing and hosting DC establishments situated within the town limits of Amsterdam have a combined annual energy consumption of 460 million kWh, or 0.46 TWh (2013 figures) [CEO 13]. These users are responsible for approximately 11% of the total annual electricity consumption of all 22,000 companies in Amsterdam. Based on figures from BroadGroup and CBRE, Vattenfall issued the following forecast for growth in the Dutch market: the capacity will increase from 225 MW (2012) to 500 MW (2016). (Note: it is assumed that these figures are higher than the capacity figures of CBRE, because it also includes the consumption of main process-supporting DCs within companies.) Besides a fast data connection, a 'big' electricity supply is of great importance for a DC. When talking about 'big' electricity connections, we are talking about 5 MVA and 10 MVA, but even connections of about 30 or 50 MVA are no exception [CEO 13]. In the Amsterdam region, Liander is the net operator, managing the electricity (and gas) net. A connection obligation is operative here. DCs are obliged to turn to Liander for the necessary connections. The renewable energy production in Amsterdam is composed of wind and PV. The following figure shows the location of the PV plants and the wind turbines33.

32 www.datacentermap.com 33 http://maps.amsterdam.nl/energie_zonwind/?LANG=es

46 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 32 – PV plants and wind turbines in Amsterdam Table 32 gives an overview of the wind production in Amsterdam. Only wind production is taken into account as this represents the largest RE production of the city. Table 32 – Wind production in Amsterdam Wind production MWh Number Total (MWh) Afrikahaven 7500 9 67.500

Westpoort 6500 5 32.500

BP 6800 3 20.400

Nauema 5100 2 10.200 Noordzeeweg 1200 8 9.600

Hemweg 1200 8 9.600

Sloterdijk 1200 4 4.800 WCI 4600 1 4.600

WCI 4000 1 4.000 WCI 1700 1 1.700

164.900

47 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

V.1.4. Summary and Analysis

As a summary the data collection for Amsterdam for the DC consumption and the available renewable energy. Table 33 – Summery Data Analysis Amsterdam

Amsterdam 2013 DC consumption (MWh/y) yearly consumption of data centers in 460.000 Amsterdam Total consumption of all data centers represent 11% of city companies of Amsterdam city 4.181.818 consumption (MWh/y) % DC for which DC4C applicable 34% this % includes all cloud/IaaS data centers and 10% of the colocation data centers % flexibility of workload 10% assumption based on trials Flexible energy (MWh/y) % flexibility * % DC for which DC4C 15.746 applicable * DC consumption Flexible energy as % of Amsterdam 3,4% Flexible energy/ DC consumption DC consumption Renewable energy (MWh/y) only wind generation is assumed 164.900 Renewable energy as % of total 4% Renewable energy/ Total consumption consumption of all companies of of all companies of Amsterdam city Amsterdam city

DC consumption as % of Available 279% DC consumption / Renewable energy RE DC flexible energy as % of 10% Flexible energy / Renewable energy Available RE

The total yearly DC consumption is 460.000 MWh. All cloud/IaaS data centers are considered as offering flexibility as well as 10% of the colocation data centers. These data centers together represent 35% of all data centers of Amsterdam. Of these data centers and based on the results of the DC4Cities trials 10% of their overall consumption is estimated to be flexible in time. Based on these assumptions the amount of flexible energy in DCs in the city of Amsterdam is estimated to be 15.746 MWh/year (3,4% of overall DC consumption). The amount of renewable energy generated by nearby wind turbines is estimated to be 164.900 MWh/year. From the DC4Cities analysis the situation in Amsterdam looks promising because the flexible energy which can be shifted or tuned over time represents a small portion (10%) of the available local renewable production at city level. It is clear from the trials that data centers have a limited amount of flexibility. When being encouraged in Amsterdam to use more renewable energy, this flexible energy will represent 10% of the unpredictable, renewable energy locally produced in the Amsterdam city region. Amsterdam is therefore considered as a city in which a match is possible between the DC flexibility and the renewable energy at hand. This will require the city to support this but as stated by the smart city objectives switching to renewable energy is an important objective of the ICT sector. The consumption of the data centers will also increase significantly, double by 2016 as compared to 2013, but it is expected to remain

48 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities in an acceptable range of the available renewable energy which will also increase as this is another city-level objective. The number of DC4Cities packages that might be sold taking into account the number of data centers present in Amsterdam today is 18 i.e. 14 (cloud/IaaS data centers) and 4 (10% of colocation data centers). This would allow these data centers to use in an efficient way the available renewable energy generated in the Amsterdam region. V.2. Paris

There are three different geographical areas that are relevant when looking into opportunities for renewable driven DCs in the urban agglomeration of Paris.  The city of Paris. This is the city within the official city limits, which in France equals the “département” Paris. However, data sources are not very abundant and the building density is so high that rooftop PV is the only available intermittent renewable energy source (apart from a new 8MW geothermal site).  The “Metropole Grand Paris” (MGP). MGP is a new institutional and organizational body that will come into effect on 01.01.2016 in order to deal with issues regarding area management, economic, social and cultural development, local politics and environmental protection. The rationale behind this idea is to integrate the territorial entities which are highly influenced by the catchment area Paris anyway, e.g. as residential area for its workers. MGP comprises Paris and the departments Hauts-de-, Seine-Saint-Denis, and Val-de-Marne. A common statistic is not yet available, but figures can be aggregated from the department level.  The region “Ile de France”. Ile the France is the region with Paris as a centre; quite a lot of statistics are available for this area, and prior to the creation of MGP it was an adequate estimate to the urban agglomeration of Paris as Paris contains 24% of the area but 89% of the population of Ile de France34. In this study, Ile de France information will only be referred to if no information on city or MGP level is available. V.2.1. Public Strategy and Options

Paris is very active with regards to climate issues; also in connection with the UN climate change conference at the beginning of December 2015. Not only is there a plethora of public and private institutions engaged in climate change activities like APC (“agence parisienne du climat”) or Paris Green35, but also a number of energy related plans depending on the regional level. The same applies to the smart cities concept: Paris is among the 10 smartest cities in Europe [MSC14] and invests heavily into its smart future, for instance by promoting the smart grid36 or by appointing a smart city team to explore options for a smart Paris37.

V.2.1.a. Paris’ Climate Plan: “Le Plan Climat Energie de Paris”

On the level of the city, the setting in Paris is not bad considering the general targets for renewable energy utilization. In 2007 Paris officially adopted the Paris Climate and Energy Plan (“Le Plan Climat Energie de Paris” [COP14a]) that adheres to the factor 4 concept and aims at reducing CO2 emissions by 75% in 2050 compared to 2004. For the intermediary goals of 2020 that means that Paris intends to reduce the GHG emissions as well as energy

34 https://fr.wikipedia.org/wiki/%C3%8Ele-de-France 35 http://www.apc-paris.com/ ; http://www.paris-green.com 36 http://www.paris-green.com/en/realisations-emblematiques-amenagement-et-smart-cities/ 37 http://datasmart.ash.harvard.edu/news/article/planning-the-smart-city-in-paris-736

49 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities consumptions on its territory by 25% compared to 2004 and increase the use of renewable energy or recuperation to 25% of its energy consumption – please note: energy, not electricity. The Paris administration intends to set a good example by increasing these figures to 30%. In the reference year of 2004, the Paris area used 32 GWh of energy with a share of 10% of renewable and recuperated energy. The objectives of the climate plan are constantly being monitored. A first evaluation was presented in 2012; since then there have been yearly updates in the so-called “plans bleus”. One target group in the climate plan which is implemented with interest groups is the tertiary sector. Agreements have been signed with: la RATP (public transportation organization), la CPCU (public heat network organiztion), Point-P, Monoprix, Airfrance, Bouygues Immobilier, la Caisse des Dépots, Eau de Paris, Edf, GrDF, Greenflex, and la Poste. Until now there is no stand-alone DC among those, however, most of these companies will have their inhouse DC. DCs are mentioned only from the buildings point of view; and the issue that is currently being tackled is not regarding adaptive electricity consumption, but regarding heat reuse. Nevertheless, the energy plan sets a priority to locally produced renewable energy. The (2013) figures mentioned for the city administration in the 2014 plan bleu are an energy consumption of 634.9 GWh worth 62M€ of which 227.7GWh are the consumption of electric power, 17.9% of which is produced with renewable sources and recuperation [COP14b], mainly using the power grid. In 2013, all of Paris had an electricity consumption of 14,614 GWh with a share of renewables and recuperation of 18.6% [COP14b].

V.2.1.b. MGP: Plan Locale Energy (PLE)

The PLE [APU14] has to be viewed in connection with the SRCAE (see below). It also adheres to the factor 4 vision38 and was adopted in 2013, i.e. even before the official installation of MGP as an institution. It focuses mainly on builidings as they are consuming 60% of the total energy (not electricity). From buildings perspective they differentiate areas with single housing versus blocks of buildings (less surface for PV) and extrapolate, based on the nature of the buildings, RES coverage for 2050: on average 69%, less in the Paris due to the prevalence of blocks of buildings, more in the outskirts. Also this plan – as in Barcelona – establishes a priority of local consumption. Also DC are viewed from the buildings perspective, not the IT perspective. As such they are viewed as producers of heat rather than producers of IT services, and therefore the re-use of heat plays a major role. PV and wind as energy sources for DCs are not mentioned.

V.2.1.c. Ile de France: Schéma Régional du Climat, de l’air et de l’energie (SRCAE39)

The regional plan SRCAE was developed based on the legal packages Grenelle I and Grenelle II with regionally differentiated target corridors for wind energy. It also requires regions in France to monitor a variety of environmentally and energy related developments like regional climate change factors, energy efficiency, air quality. It is aligned with an overall factor 4 goal that aims at reducing greenhouse gases until 2050 compared to 199040.

38 „factor 4“ is the name of a strategy that aims at increasing sustainability by both increasing production and decreasing resource input by half. http://www.sustainabilitydictionary.com/factor-4/ 39 http://www.driee.ile-de-france.developpement-durable.gouv.fr/le-srce-d-ile-de-france-adopte- a1685.html 40 https://fr.wikipedia.org/wiki/Sch%C3%A9ma_r%C3%A9gional_climat_air_%C3%A9nergie

50 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

This law and the implementation connected to it forms the link between local and national legislation. V.2.2. Commercial and Enterprise Data Centres

V.2.2.a. Number of Data Centres

37 DCs located in Paris have been analysed (see Appendix VIII.4. ). It was not possible for all DCs to collect the same level of information. It was for example not possible to determine the surface and the power consumption of all DCs. Most of the information has been collected starting from the data center map41 and via the web sites of the data center providers. DCs can be divided into two categories: small & medium size ones with a surface smaller than 1000m2 such as Acropolis, Clarinet and Cogent and large ones with a surface higher than 1000 m2 such as Colt, Interoute, TelecityGroup. Even though it is not possible on the level of identified DCs to give an idea about the aggregated sum of MW required by all DCs together, from [DRI12] it is known that Paris nowadays needs to supply around 270MW to DCs and the MGP 715MW. For 2025, around 1000MW power is expected to be additionally required by DCs (for additionally 500.000 m2). From the grid point of view, both DC and electric vehicles are seen as a major challenge which the region aims to deal with for instance by reusing the heat of DCs in order to relieve the grid [DRI12].

V.2.2.b. Business Model and SLAs

The number of cloud/IaaS providers in Paris is still quite limited. Some cloud providers such as agarik and mesh became part of larger groups such as Atos and Plusserver Gbmh. Some large companies such as navlink and SFR offer such cloud solutions. There are smaller, dynamic actors such as Kheops, LINKBYNET and Tas France. The consolidation is a clear indication of a tendency in Paris of an increase in volume of cloud services. This will allow more and more flexible services to be offered in the short and mid- term in Paris which will be controllable by DC4Cities to increase the data centers’ renewable energy usage. Based on the information provided by Datacenter map there are 50 colocation data centers from Paris in France and 19 cloud providers. This gives a fraction of 72% of colocation DCs and 28% of cloud/IaaS DCs. In this fraction the size and power usage of the DCs is not taking into account.

V.2.3. Potential for Renewable Power

All over France, hydro has been THE renewable resource until recently. In 2004 nearly all of the 25GW installed renewable power was hydro, since then first wind and then solar has been developing. In 2013 Hydropower was 25 GW, Wind 8.3 GW and Solar 4.4 GW42.

V.2.3.a. The City of Paris

In the aftermath of signing the “plan climat energie” Paris has invested a lot into PV and thermal solar energy sites so that 2013 31.800 m² PV panels [COP14b] which is equivalent to an installed peak capacity of 1.18 MW distributed in 67 solar installations43. From the consumption

41 http://www.datacentermap.com/datacenters.html 42 http://resourceirena.irena.org/gateway/countrySearch/?countryCode=FRA 43 http://www.statistiques.developpement-durable.gouv.fr/energie-climat/r/energies- renouvelables.html?tx_ttnews[tt_news]=23865&cHash=103c4b14d08e3a8728eea9b75d4fd049

51 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities side, renewable energy and recuperation delivers 17,9% of electricity consumed [COP14b]from the production side, unfortunately, the French statistical office stated that there are no figures available. Nor are there to our knowledge publically available data as to the amount of energy produced with these installations or daily profiles. In order to get an estimation of the production, a simulation tool based on a photovoltaic geographical information system (PVGIS) which was developed in connection with JRC IET was used44. When entering Paris’ geo-data, this tool accesses radiation databases with local weather information. Based on assumptions regarding PV characteristics45 and the KW peak information (1.18MW) average monthly energy yields are calculated. In order to get a feeling as to the difference between winter and summer yields, radiation charts for Paris with average daily harvesting profiles for each month (e.g. January, July) are also presented. The results are that a yearly production of 1,16 GWh is estimated, distributed to a great degree during the summer months: as can be seen in Figure 33 the production in July (141 MWh) is forecasted nearly four times as high as in December (37 MWh). Relating this to the electricity consumption of 2013 of 14.614 GWh [COP14b] the share of solar and wind (no installation) electricity at the local electricity production is minimal and does not rectify the adaptation of a DC workload to solar and wind profiles. For all of France, the situation is not much better: in 2012, all renewable sources accounted for 16,7% of electricity consumption – of which solar and wind, however were a mere 5% and 18% respectively. For 2020 the share of solar electricity at the renewable electricity consumption is projected to remain about the same, whereas the share of wind energy is supposed to grow to 39% at the renewable electricity consumption.[STA15] so that the volatility of electricity supply will increase also in France.

Figure 33 – Monthly energy output from fixed-angle PV system This is also reflected in Figure 34 that depicts the irradiation for fixed solar cells based on the assumed weather data for January and for July. Note that the clear-sky simulation does not yield as different results as the real-sky simulation. Altogether, Paris has a program to install 200.000 m2 of solar panels [COP14a], which is more than 5 times as much as installed in 2013. Considering that in 2013 31800 m2 of solar panels are equivalent to 1.18 MW power, in 2020 7,4 MW could be installed. Using the same

44 http://re.jrc.ec.europa.eu/pvgis/apps4/pvest.php 45 standard assumptions as there is no information regarding the Paris solar installations: Crystalline Silicon technology, building integrated, 14% system losses, 35° slope, orientation towards south, a fixed plane

52 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities estimation tool as before, this would mean a solar production of 7,65 GWh. The objective for 2020 is to consume 24.200 GWh in final energy (not electricity). So from the grid point of view of the local energy grid, a tool like DC4Cities does not bring a great benefit – it could even de-stabilize the grid in times of high power demand as the presence of either sun or wind in the grid is minimal. With regards to the national grid, this might change in future. And looking at DC4Cities from the point of view of a priority of self- consumption as stated in the Paris energy plan, DC4Cities can be applied to DCs with a considerable share of on-site PV or wind energy generation.

Figure 34 – Daily irradiance on a fixed plane on a typical winter and summer day As in Barcelona, Frankfurt and Amsterdam, also for Paris a solar cataster has been created. This shows that the plans to increase the PV infrastructure are being implemented with great force. What also helps the DC4Cities issues it that Paris has the first smart grid neighbourhood in France: Issy-Grid, installed in 201246.

V.2.3.b. MGP

For the area of MGP the picture looks more or less the same: In 2010 the consumption of electricity was 75 TWh, for 2020 it is expected to be between 80-87TWh, for 2030 between 83-95 TWh. The energy (not electricity) consumption of the tertiary factor has a share of 44% of the total, which is deemed to remain more or less constant [DRI12]. In 2010 about about 8 GWh were produced by PV in the area of MGP, only 0,34 by wind which was then less of 1 permill at the total. In connection with the PLE, the theoretical potential of PV in the area of MGP has been estimated at 1,2 TWh based on the rooftops in the area that remain after the production of thermic solar energy [APU14]. Thus, the figure can be viewed as a very conservative estimation. Still, the share of PV at the electricity consumption will be only in the magnitude of one-digit percent points due to the high density of electricity consumption in the area. A big challenge that is foreseen in the area is that power is increasing at a much faster pace than energy consumption (see Figure 35). And a big consumer identified for MGP are DCs. According to [note problematique] the power that needs to be additionally installed in the area due to DCs only is 1 GW. If DCs were scheduling according to the availablity of sun and wind, this might even increase the pressure. However, what would again help would be the self- consumption of DCs whereever and whenever possible.

46 http://www.paris-green.com/wp-content/uploads/2014/04/Fiches_5_IssyGrid.pdf

53 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

In 2013 the installed PV power was 12,5 MW peak47, which using the above mentioned PV yield estimation tool might result in the production of 12,9 GWh of energy.

Figure 35 – Development of power versus energy demand in MGP [DRI12]

V.2.3.c. Ile de France

The power consumption of Ile de France amounts to 16% of the whole French power consumption. It produces less than 10% of the electric energy it consumes – and of these 10% only 15% are renewable and re-use energy based [DRI12]. V.2.4. Summary and Analysis

As a summary the data collected for the Parisian areas, both for the DC and the energy world are merged in Table 3448; in case the figures are estimated or calculated, this is mentioned in the table. Table 34 – Summary Data Analysis Paris

Paris MGP

2013 2020 2013 2020

DC today (connected) - (connected) 715 (connected) 270 MW1 MW1 1.715 MW1(2025)

Electricity 14.614 - 75 TWh1 (2010) 80-871 TWh consumption GWh2(2013)

Totoal Power Average. 1700 - Average.8.500 Average.9.100 (estimated) MW (calc from MW (calc from MW (calc. from 14.614 GWh) 75 TWh) 80 TWh)

47 http://www.statistiques.developpement-durable.gouv.fr/energie-climat/r/energies- renouvelables.html?tx_ttnews[tt_news]=23865&cHash=103c4b14d08e3a8728eea9b75d4fd049 48 Sources: 1 [DRI12], 2 [COP14b], 3[http://www.statistiques.developpement-durable.gouv.fr/energie- climat/r/energies- renouvelables.html?tx_ttnews[tt_news]=23865&cHash=103c4b14d08e3a8728eea9b75d4fd049], 4 [estimated using the JRC tool], 5 [APU14]

54 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

PV power 1,18 MW (peak, 7,4 MW (peak, 12,5 MW3 Average 136 installed)3 estimated, (peak, installed) MW (calc. from based on potential) planned m2 installed)

Wind energy 0 0,341 (2010) 2013 GWh

PV energy 1,16 GWh4 7,65 GWh4 12,9 GWh4 (estimated) (estimated) (estimated)

Miscellaneous 17,9%2 1,2 TWh5 Share Rooftop PV Renewable technical Energy sources potential for at energy MGP area (after consumption thermal solar power); this would go along with average MW of

Looking at Paris and the share of mere 17.9% of renewable energy sources at the electricity consumption of which solar makes up only a small fraction and wind is non-existent, the situation does not look very promising. From the point of view of DCs, we know that the connected power of DCs is about 270 times higher than installed PV peak power (and about a sixth of total power in Paris). This must of course not imply that all PV power in Paris could go to DCs – it is only the difference of magnitudes that should be illustrated here. Even if on average DCs used only half of the connected power49, the ratio would still be so high as to not allow a considerable change in the fraction of renewable energy used by a DC (RenPercent). In spite of this rather bad outlook, DC4Cities can be interesting for the DCs of the Paris city administration that aims at being a good example of renewable integration. Also, the city administration has more space than just the DC housing to install roof-top PVs by using more public buildings. DC4Cities can also be interesting for DCs with a high amount of self-production via roof-top solar cells that aim as using the electricity they produce themselves to the greatest degree possible (this goes along with the objective of the Paris Climate plan to give a priority to self- consumption). Usually the ratio of density of roof-top solar electricity production alone compared to the power consumption of a DC does not offer a lot of scope for increasing the DCs RenPercent. As an example, in D2.3 the installation at CSUC was referenced with an installed peak capacity of 154 W/m2 (D2.3, p.21). The power density of demand in the CSUC trial was 661 W/m2 in the baseline scenario50 (D6.2 p. 26) so that a roof-top installation could cover around 23% of DC power in times where realized power is near to peak power. The reward for a DC in this case is that the marginal cost of the energy consumed is zero. However, for Paris the yield of solar peak power installations is lower than in BCN so that the ratio of supply and demand density is worse. The DC4Cities strategy is thus less rewarding than in BCN even for the consumption of on-site production and therefore also this use case does not promise a huge market in Paris.

49 and considering the IT workload of DCs is often rather flat, this is not a very realistic assumption) the real ratio might be more at 2/3 or 3/4 50 225 W/0,34m2 (D6.2 p. 26) equals 661/m2

55 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Looking into the situation of MGP does not change the picture substantially. The fraction of PV peak power vs. installed DC capacity is around 1,7% and thus only slightly better than for Paris city. In the case of DCs located in the outskirts of Paris the self-consumption might be a more valuable due to more space to install solar power additionally to roof-top solar cells. The outlook to 2020 due to a lack of data is possible only for MGP: The ratio of PV peak power vs. installed DC capacity is better by a factor of four (it will be around 8%), but PV power in the area will still make up only around 1.5% of total power capacity needed for MGP. Therefore, using solar power during the day when the grid is under strain will most probably increase that strain on the power grid and so also in 2020 will probably not be rewarded. This reasoning of course excludes again the priority to consume self-produced power not only for stand-alone DCs but even more so for the smart city administration of Paris that aims at covering 30% of their total energy (not electricity) consumption by renewables and re-use. V.3. London

In the following chapter, the results of our market analysis regarding the city of London (UK) will be presented. The basis for this contribution was research on DCs in London as a first step to elaborate relevant information on their features like size, business models as well as tier levels and categorize them. In parallel, a survey was created with approximately 20 questions to provide more information regarding the questions for DC operators. The survey was distributed to the contact e-mail addresses stated on their individual webpages. The goal of the survey was to receive more specific information regarding their individual operations. This includes for example information about the concrete power demand over several periods, PUE or the usage of integrated renewable energy supplies. All information gathered are presented, categorized and analysed in the following sections. V.3.1. Data Centres in London

In this study, 32 DCs in London are considered. However, this is only a subset. It was not possible to have an objectively complete number of DCs due to the lack of a complete central list. It was noticed that most of the DCs offer the business model “Co-Location” (B2B). This business model refers to a concept in which operators of DCs are housing servers of their costumers and provide connectivity and an internet service provider. The provider of the colocation DC takes care of the infrastructural service management. This includes amongst others the provision of infrastructural service management, UPS systems and automated as well as physical security systems like automatic surveillance systems and guards. In addition grid connection is provided. The customer can as a rule choose a provider of their choice. Beside that there are also DCs which offers directly services to customers (B2C) (e.g. managed hosting, could, webhosting or VoIP). The distribution of the two different models can be seen in Figure 36. By choosing the B2B model DC operators can provide a higher flexibility to their customers as services are not limited to the pre-installed hardware. On the other hand, no visibility of the given processes and procedures inside the servers is given, because each customer administrates his own services which are not visible by the DC operator. Therefore, this B2B is not well suited for DC4Cities, because this information is necessary for an optimal usage of DC4Cities.

56 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

B2B vs B2C

B2B B2C

Figure 36 – The distribution between the two business models B2B and B2C For all DCs that were considered and which stated the tier level on their website the level was three. It is the second highest tier level of DCs. Features of these data centers are for example: the availability of redundant components, multiple independent distribution paths serving the IT equipment but also the possibility of fault-tolerant maintenance51. The average floor size for technical equipment of the DCs found in the market study is approximately 9,500m². The largest DC has a floor size of 65,543m² and the smallest 557m². Therefore, all DCs belong to the categories of medium or large size DCs (cf. D2.3 Table 3 for a taxonomy). The different sizes of the considered DCs are shown in Figure 37. The red line marks the average DC floor size.

Floor size of Data Centres 70,000

60,000

50,000

40,000

30,000

20,000

10,000

0 0 5 10 15 20 25 30 35

Figure 37 – Floor size of collected DCs in London. The line marks the average floor size The more space is provided for technical equipment the higher the potential for more energy consumption. Hence, an indicator of a large energy consumption could be the size of the DCs. An approach could be to take into account DCs with large energy consumption, because positive effects from DC4Cities are much more noticeable than in comparison with DCs with a very low energy consumption. Because of the fact that from the survey no information could

51 https://journal.uptimeinstitute.com/explaining-uptime-institutes-tier-classification-system/

57 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities be extracted about the real energy consumption, information about the connected load were collected which are presented in Figure 38.

Power Connected Load for Data Centres in MW 90.00 80.00 70.00 60.00 50.00 40.00 30.00 20.00 10.00 0.00 0 5 10 15 20 25 30 35

Figure 38 – Power connected load for the considered data centres in London. Considering both graphics (size of DCs and power connected load) it can be seen that both values are not strictly correlated even if the power connected load usually grows with a greater floor size. In comparison with London, Frankfurt provides DCs with a much higher connected load (120 MW) but London has more DCs with a load over 20 MW. Therefore, it can be argued, that enough potential for the usage of DC4Cities in London is given.

V.3.2. Business Models and SLAs

More than 90% of the service providers in London that explicitly mention the existence of their own DCs are co-location DCs with the explained infrastructure above. Nearly half of them offer carrier neutral services. The other half offers at least several possibilities regarding carrier services, so the customer can choose more or less between network and/or energy providers. Besides colocation, some DCs additionally offer services like cloud services, managed hosting, VoIP and webhosting, which are shown in Figure 39. On the websites of the different DCs, different SLAs are mentioned in relation to the availability, the provided security or cooling systems. These statements are very generic and don’t show concrete implementations. Also no flexibilities in the SLAs are mentioned except in [LON14]. Although co-location - due to the already above mentioned problem of the lack of influence on work processes - is not ideal for DC4Cities, there is still potential for improvements. Beside the aspect of alignment of workload to renewable energy sources, mechanisms for energy savings by intelligent usage of different working modes are also available.

58 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Business Models 35

30 30 25

20

15

10

5 4 5 2 1 0 Cloud Managed Hosting Colocation VoIP Webhosting

Figure 39 – Business models of collected DCs A concept in which data center operators act as middleman to offer their customers DC4Cities would therefore be an interesting alternative. For these reasons it can be concluded that London definitely provides potential for the use of DC4Cities, although this potential cannot be fully exploited.

V.3.3. Energy production and Smart city policies in London

In addition to consider only DCs in London, it is also necessary to examine the energy situation in more detail. The improved utilization of renewable energy sources play the key role in DC4Cities therefore a suitable energy situation in the current energy mix is desirable. The current energy mix in the UK is illustrated in the Figure 40.

Figure 40 – Electricity generation by source in UK52 It is visible that the amount of renewables in UK takes only a little part of the overall energy mix in UK. Especially, the solar situation is very poor. Further, looking at London, in particular the ratio of solar energy in comparison to other regions of England is far behind in last place. With a fraction of only 2.6%53 of the total solar electricity of the regions in the UK, London is far behind in last place in comparison. And with an installed capacity of only 57,809 kW,

52 Values: http://www.greentechmedia.com/articles/read/uk-wind-power-surges-but-so-does-coal-use 53 The values of all nine regions of the UK were summarized. 2,6% is the ratio of generated electricity of London in comparison to the other regions.

59 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

London meets only approximately 1% of the estimated possible solar providence electricity needs in London of 20% [AEC13]. Including the sunlight in London over a year, with an installed solar power of approximately 58MW, approximately 88MWh can be achieved over a year54, what is approximately a factor of 1,5. Especially the summer from April to September have the greatest impact here.

Installed solar power capacity in England (kW) 600 503.808 500

400 350.376 283.376 300 257.882 218.205 190.367 187.465 200 92.009 100 57.809

0 South South East of East Yorkshire West North North London West East England Midlands and The Midlands West East Humber

Figure 41 – Installed solar power capacity in London in comparison of other regions in the England in kW [AEC13] Some possible reasons for the poor situation in London could be the following points: lack of potential leadership; city space of small thin buildings with little roof space and lower level of home ownership means fewer households’ installation above [AEC13]Error! Reference source not found.. There are some projects to provide more solar energy like the Blackfriars rail station55 (the world’s largest solar-powered bridge) or the NCP car park at the Olympic park56. Apart from these exceptions the overall situation in London regarding renewables is nevertheless rather poor. Despite the fact of London’s city structure with many small and high buildings, there is an ambition to cover up to 25% of London’s energy needs from local, low carbon sources by 2025. In the past, the installation of solar systems in London has been driven by several projects. In RE:NEW for example, 4,300 housholds have been equipped with solar systems and beside energy efficiency measures, the Enfield’s Civic Centre, the Frie Brigade and 20 fire stations were provided with solar power systems [AEC13]. Additionally, Mayor’s planning policy has secured additionally 7MW of solar capacity in 2013. This was driven among others by several solar array installations in the Greater London Authorithy, Londons Fire Brigade or the metropolitan police services with capacities from 250kw up to 619kw per installation [AECL13]. In the future, the scope focus on London’s homes, because there is the biggest remaining capacity of solar installations.

54 London was entered on the website tool as the city and all basic parameters were led at it is. For the installed peak PV power 57,809 was used and all twelve months of the calculation were summarized (result ca. 88MWh). http://re.jrc.ec.europa.eu/pvgis/apps4/pvest.php 55 http://www.theguardian.com/environment/2014/jan/22/worlds-largest-solar-powered-bridge-opens- in-london

56 http://www.sundog-energy.co.uk/news/recent-pv-projects/320-london-olympics-car-park-roof

60 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Additionally, London will reach several goals regarding CO2 emissions by focusing on the mobility sector and energy efficiency techniques. Different plans are existing, like57: RE:CONNECT, RE:NEW, RE:FIT and the London’s action plan address different strategies to reduce CO2 emissions. Retrofit existing buildings with energy efficient measures, reduce emissions through transport including a significant roll-out of electric vehicles or maximize CO2 reductions from new developments by national building regulations are some further approaches [GLA11]. Additionally, in the London Plan and the Smart London Plan there are approaches for a better decentralized energy structure and demand management.

Currently, London focuses on reducing CO2 emissions compared to 1990 which is shown in the following table: Table 35 – The Mayor's CO2 emissions reduction targets in London compared to 1990 [LON14]

Target year Target CO2 emissions reduction on 1990 levels

2015 (interim target) 20%

2020 (interim target) 40%

2025 60%

2050 At least 80%

Techniques like usage of biogas to generate heat and power, replacement of Transport for London’s buses with hybrid buses and development of energy efficiency in public buildings are employed. But the key word CO2 emission reduction says nothing about the solar or wind situation for London in the near future. The potential is still there, especially in the household section [GLA11] because only fewer than 0.5% of London’s over 3 million homes and only 7% of London’s 3,000 schools have solar panels installed [AECL13]. Anyway, no concrete numbers could be found regarding the energy mix in London, the future energy situation in London, energy consumption of DCs or which specific goals are pursued to improve the situation regarding local photovoltaic plants (e.g. roof installations). V.3.4. Summary

The renewable energy situation in London regarding solar power is very poor currently in comparison with other regions in the England. This is probably due to the building structure and lack of free spaces. The ambition to cover up to 25% of the energy consumption by renewables, is a first important step but the technical possibility could be much higher. No answers regarding our survey for London as well as missing information about the concrete energy consumption of DCs in London, the energy mix situation today, in the near or far future, makes it more difficult to rank the potential of London with regard to DC4Cities. Nevertheless, the targeted objective to reduce the CO2 emissions up to 80% by 2050 has to be highlighted. This can only be reached if in addition to the planned energy efficiency and energy saving strategies also the amount of renewables increases, which allows for a positive outlook. Several plans like RE:CONNECT, RE:NEW, RE:FIT and the London’s action plan addresses this ambition, but no concrete rollout plans are available regarding numbers of concrete installation of solar and wind plants.

57 https://www.london.gov.uk/priorities/environment/publications/delivering-londons-energy-future-the- mayors-climate-change

61 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

London is an important city in Europe where a potential for DC4Cities is given. Different climate change plans are followed to reduce CO2 emissions and increase renewable energy supply. On the other hand, a lot of DCs with high power connected load allows for inferences to a high potential. Even if the potential of DC4Cities cannot be fully leveraged there is still potential for improvements. Although the business models of the DCs which are mainly co-location are not ideal DC4Cities, the DC operators can play the role of a middleman to offer their customers the usage of DC4Cities in both aspects, increase the usage of renewables and also saving more energy. For other service providers like manged hosting or cloud energy savings aspects are probably more relevant from the customer points of view and the usage of renewable energy more from the provider’s point of view. V.4. Frankfurt

In the following the results of our market analysis concerning the city of Frankfurt (more precisely Frankfurt am Main) will be presented. It includes information about DCs, smart city policies and the energy situation.

V.4.1. Data Centres in Frankfurt

Frankfurt is one of the most important cities in the DC market for Europe. In this section we will provide the following information: 1. Information about the number of DCs. 2. Collected information about the three basic business models of DCs 3. Information about the power and energy consumption of these DCs

V.4.1.a. Number of Data Centres

Within our analysis we have found 34 distinct DCs in Frankfurt. However, this is only a subset. It was not possible to have an objectively complete number of DCs due to the lack of a complete central list. Some DC (service) providers are only resellers or make use of co-location services themselves. However, in most cases these circumstances are not visible to a third party like us and therefore difficult to identify. All DCs are either 3 or 4-Tier DCs. The majority are private DCs with a business to business (B2B) model.

Tiers B2B vs B2C

3-Tier 4-Tier B2B B2C

Figure 42 – Data Centre Characteristics

62 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

In general DC4Cities can be used in any tier-level or customer model. However, the higher the tier level, the higher the requirements on redundant infrastructure. Business customers might have higher requirements than normal consumers. However, as DCs with a B2B business model usually have less direct customers than B2C it has still high potential. Taking this into account the DCs in Frankfurt are averagely suited to use DC4Citties with regard to the tier- level and customer model. The average floor size of the DCs found in the market study is approximately 10,000m2. The largest DC has a floor size of 60,000m2 and the smallest 800m2. Therefore, all DCs belong to the categories of medium to large size DCs (cf. D2.3 Table 3 for a taxonomy).

Figure 43 – Data centre size In general, as described in D2.3 larger DCs are, even though all DCs are supported by the DC4Cities approach, more suitable for using our approach. This is because the impact of such DCs to the overall energy management of the city is larger and by that the relation between impact and overhead of using DC4Cities is good. Taking this into account the majority of the DCs found within the market study for Frankfurt are well suited concerning size to be used with DC4Cities.

V.4.1.b. Business Model and SLAs

Almost all (95%) service providers in Frankfurt that explicitly mention the existence of their own DCs are co-location DCs. About 90% of them offer carrier neutral services, i.e. the customer can choose more or less freely between network and/or energy providers.

35 30 25 20 15 10 5 0 Managed IaaS Colocation Cloud SaaS Hosting

Business Model

Figure 44 – Data Centre Business Models in Frankfurt

63 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Besides colocation, some DCs additionally offer services like managed hosting, infrastructure as a service (IaaS), cloud services or software as a service (SaaS). SLAs which were publicly available were all of a very generic nature. They offered a guaranteed availability of the service and for colocation services high security aspects. However, no flexibilities were yet to be found within these SLAs. On the one hand this facilitates flexibilities in areas were no specific guarantee term was provided like for example a guaranteed network latency. On the other hand it provides strict restrictions for the parts which were defined in the SLAs like the availability. In general however, the SLAs did often only provide suggestive guarantees like the availability with not much guarantees with regard to performance. Therefore, DC4Cities has several chances to exploit these gaps to gain higher flexibilities. In general, colocation services are not the ideal service to be offered in conjunction with DC4Cities, as the DC owner has very little impact on the flexibility of their customers. However, they can act as a mediator/reseller and offer contracts to its customers which facilitate the better use of DC4Cities services with a rewards system. In addition, with the help of DC4Cities a better planning can be performed which transforms into a better flexibility as well. Taking these findings into account the DCs found within the market study for Frankfurt are averagely suitable to be used with DC4Cities concerning their business model.

V.4.1.c. Energy Situation of DCs in Frankfurt

As indicated earlier the size of DCs found in the market study are above average. The same goes for the energy supply capacities. The connected load ranges between 1MW and 120MW (see Figure 45) . All DCs have redundant energy lines. In addition they all have UPS facilities and diesel generators. The UPS facilities can maintain the state of the DC online for about 10- 15 minutes which is enough time for switching on the diesel generators. It was not possible to find out exact details about the energy contracts of the DCs. However, taking into account the size of the connected load of the DCs they are all large customers and therefore it can be assumed that they have special tariffs.

140

120

100

80

60

40

20

0 0 5 10 15 20

Figure 45 – Power Connected Load of DCs in Frankfurt Many (30%) offer contracts to their customers enabling to use 100% renewable energy sources. However, we were not able to identify if any of these are at least partially cover by locally installed PV or other locally installed renewable energy sources. However, in general the circumstances for installing local PVs within the city of Frankfurt are good. The figure below

64 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities shows exemplarily a map containing the qualification of the Equinix FR4 DC to install local PVs. The orange colour indicated that it is well suited.

Equinix FR4

Figure 46 – Exemplary suitability map for PV in the area of the Equinix FR4 DC in Frankfurt58 Taking the “Solarkataster” as a basis the situation is similar for almost all parts of the city of Frankfurt. On average all areas are well suited for installing PVs on their rooftop. Therefore, a high potential does exist for using DC4Cities in DCs in Frankfurt for optimizing the usage of local PVs. V.4.2. Energy Production, Usage and Smart City Policies in Frankfurt

Frankfurt has an ambitious plan: 100% renewable energy usage by 2050 [NAU13, FRA15]. In general the city of Frankfurt and its administration is traditionally green and eco-aware. In 2012 Frankfurt was shortlisted for the “2014 European Green Capital awards”59 for example. Besides they are part of many initiatives and actively try to lower the carbon emissions and commit to concrete targets. They for instance joined the Covenant of Mayors target to reduce the CO2 emissions by 40% for the year 2030 [COV12]. Already in 1990 the city founded a municipal energy agency – the “Energiereferat”. They were also the co-founder of the Klima-Bündnis e.V. – Climate alliance. As part of its activities the Department for Energy Management of Frankfurt, who are responsible for 1,800 facilities60, were already able to save 400,000€ per year by setting up an energy control system. This saving is partially forwarded to the facilities by a reward scheme where they can receive up to a 50% bonus on the usage savings. Furthermore, the local parliament decided in 2007 that all publicly owned new buildings needs to fulfil passive house standards. All in all it was able to reduce emissions of city facilities by up to 24% since 1990 by these measures. In addition to fostering eco-efficiency of public buildings the city of Frankfurt is promoting the integration of Combined Heat and Power generation and on-site electricity production. Today, they can already be found in many public offices, schools, fire stations, hospitals and other public buildings. By this, the consumption of

58 http://www.gpm-kom8.de/geoapp/solarkataster/frankfurt/ 59 http://ec.europa.eu/environment/europeangreencapital/shortlist/ 60 http://www.klimabuendnis.org/frankfurtammain.0.html

65 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

primary energy was reduced by about 30% (75,000 tones CO2 per year). Since 2008, Frankfurt is also involved in the Covenant of Mayors [COV12]. Besides the city of Frankfurt launched various campaigns to reduce energy consumption and GHG emissions. In 2008 the campaign “Frankfurt Saves Energy” was launched to reward private households saving energy. This already led to a reduction of around 350 tonnes of CO2. In addition the campaign EKOPROFIT helped to save around 4,2 tonnes of CO2 per year since 2008 by giving private companies access to energy management systems. Just recently Frankfurt has launched the “Masterplan 100% Climate Protection” which has the goal to rely completely on 100% green Energy by the year 2050 [SCH15]. The Frauenhofer institute has made several simulations to analyse and predict the energy production and consumption of Frankfurt am Main today and in 2050. They have simulated if it is possible to rely on 100% renewable energy sources produced only within the close range of the city and how the situation (mainly with a monetary background) changes if 10% would be imported from other parts of Germany like from wind farms in the north sea. The result was, that it will most likely be possible to purely rely on energy produced within the region but that it would be cheaper if 10% of it would be imported. For the simulation they have been taken some assumptions which are not yet clear if they are realistic like for example that the inhabitants will become more energy aware. To this concern DC4Cities can have a big impact as well. Overall the assumptions of the study seem reasonable and the results are positive.

V.4.2.a. Current Energy Situation in Frankfurt

The city of Frankfurt consumes approximately 6,580GWh of electric energy per year [REG15]. About 50% is consumed by commercial institutions and 30% by the industry. According to dena, BMWI there is a saving potential of about 20% in the latter and about 50% in the first. Currently, the energy mix in Frankfurt mainly rely on gas, coal and waste incineration (cf. figure below). Only a very small part (~11%) of the energy production is coming from renewable energy sources.

Energy Mix Today Energy Usage Today

Gas Coal Waste Biomass Solar Wind Nuclear Airport Private Households DCs other

Figure 47 – Energy Situation Today [FRA15] About 30% of the energy used in Frankfurt is consumed by the Frankfurt Airport. About 20% 61 of it is used by private households. About the same amount of energy is used by DCs in Frankfurt today. This means they consume about 1,316GWh/a. On average this leads to a power consumption of 150MW. However, the power connected load of the DCs found in the market analysis are much higher (>500MW). In addition, this percentage will even grow further as the trend continues in the next couple of years.

61 http://www.faz.net/aktuell/rhein-main/stromverbrauch-in-frankfurt-waechst-weiter-13023794.html

66 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

V.4.2.b. Energy Situation in Frankfurt in 2050

The simulation of the Frauenhofer ISE institute has shown that it is feasible to have Frankfurt as a self-sufficient city in 2050 [NAU13]. However, as mentioned before it is cheaper and therefore more feasible to import 10% renewable energy. The energy mix will mainly rely on wind and PV (see Figure 48).

Energy Mix 2050 Energy Usage 2050

Biogas Coal Waste Biomass Solar Wind import Airport Private Households DCs other

Figure 48 – Energy Situation in 2050 [FRA15] Furthermore, the simulations have shown the following characteristics for a typical spring and a typical autumn week (see figures below). As one can see the usage and production curve do not fit 100% of the time. Therefore storage facilities are needed. In addition energy is predicted to be exported as well which however, will likely lead to losses. In general DC4Cities which was not taken into account in the simulations of the Frauenhofer ISE can help to partially reduce the need for storage facilities and therefore can be a fundamental and valuable part to achieving an optimized city based on 100% renewable energies.

Figure 49 – Prediction of a typical spring week [FRA15]

67 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Figure 50 – Prediction of a typical autumn week [FRA15] As the DC industry is predicted to grow further in the next decades and Frankfurt as the main DC hub in Germany plays a major role the energy usage of DCs is predicted to grow further and therefore its influence as well. However, even with the share of DC power consumption today DC4Cities is able to help. Taking the assumption from above that DCs on avg. consume 150MW power and almost all DCs (>90%) are colocation DCs, it is possible to actively lower the need for using storage facilities by adjusting the DC power curve to use more renewable energy. If we assume that about 10% of all colocation DCs would use DC4Cities and with its help it would be able to have an influence on a maximum of 15% of its power consumption, then however, the effect is relatively small (2,25MW ~ 5% less storage need). In case new business models and contracts would be developed that are better suited for colocation DCs the need for storage facilities could be reduced by 50% with the help of DC4Cities. V.4.3. Summary

Frankfurt as the central hub of German data centres is one of the cities in Europe where the potential of using DC4Cities is the highest. In general the impact of changes in the energy consumption of data centres in Frankfurt can have a huge impact as 20% of the energy in Frankfurt is already today consumed by DCs. This 20% is equivalent to about 1,316GWh per year. The pure size of the data centres located in Frankfurt (both in square meters and in energy consumption) form an ideal basis for using DC4Cities reaching a high impact. However, the business models of the data centres which are mainly co-location based prevent this potential from fully being leveraged. New business models including DC4Cities concept based on rewards and penalties should therefore be introduced for co-location data centres in the future. When we consider 10% of all co-location data centres offer flexibilities between 5% and 20% today, DC4Cities could have an impact between 6580MWh/year and 26320MWh/year which is relatively small compared to the overall energy consumption of the city of Frankfurt (~0,1%-0,4%). However, this would drastically change with new business models. This situation is also reflected by the fact that from the data centres found in the market study only about 4 DC4Cities packages (10% of colocation + 1 cloud/Iaas data centre) might be sold in Frankfurt at the current state. This is again due to the high number of co-location data centres and the lack of a corresponding DC4Cities business model. The amount of energy generated by renewable energy sources are about 725.000MWh per year which is about 11%. Compared to other cities this is relatively high. From the smart city point of view Frankfurt is also an ideal place to apply DC4Cities. The (smart) city administration is highly interested in ecological concerns. They have already today successfully applied many energy and CO2 reducing measurements in the public sector and foster eco-efficiency in all sectors, i.e. industry, private households and commercial

68 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

enterprises, too. Their plans for the future are even more ambitious, trying to reduce CO2 emissions by 40% until the year 2030 and by 100% by the year 2050. As simulations have shown this goal is realistic. Here, DC4Cities can have a valuable impact to reach this goal and for example partially reduce the need of storage facilities. However, for this purpose more advanced business models for co-location data centres should be developed in the future to be able to leverage the full potential of DC4Cities for the city of Frankfurt. V.5. Madrid

In this section the results of our DC’ market analysis concerning the city of Madrid will be presented. The sources used are above all the DC's web sites, which provide DC's information profile and DC's business models. Unfortunately, no relevant information concerning the energy consumption is available; we have tried to send a lot of e-mails of DC's energy manager (or other e-mail addresses stated) but without success. However, thanks to the few information that we have had, we will analyse 15 of the bigger (in terms of size) DC's in Madrid. This number is proceeds through the DataCenter962 web-site, DC’s expositions, Plataforma enerTIC63. V.5.1. Data Centres in Madrid

Although the European DC’s market is grown in the recent years, the city of Madrid - as shown in the last report (2015) of CBRE [CBR15] - currently, does not take part to this expanding market, though this market has been important by 2014. However, beside Madrid has not kept pace with the growth of the largest four markets in Europe, is still evident the interest of Madrid’ operators (such ad the telephone operators) to invest in DC market.

V.5.1.a. Number of Data Centres

The Madrid economy are inspiring much confidence toward new IT investment indeed, the beginning of 2014 has recorded encouraging hints of new market activity in the IT sector. The economic report for 2014 indicates GDP growth of 0.4% in the first quarter with expectation of a positive outlook for the year. Up until now Madrid's considerable corporate community has shown reluctance toward spending of any kind, an understandable stance given the recent recession [CBR14]. In the fist part of this analysis, we using the data referred to the Madrid DC’s market, as shown in the CBRE report (2014). The Table 36, shows statistic regarding the DC information in the city of Madrid in terms of supply sq. m, availability sq. m., vacancy rate %, collocation take up. Table 36 – DC Market in Madrid

LOCATION SUPPLY AVAILABILITY VACANCY RATE % COLLOC ATION TIER 1 SQ. M. SQ. M. TAKE-UP MARKET YEAR TO DATE

MADRID 29,064 1,006 3,46% 193

62 http://www.datacenter9.com/datacenters/spain 63 http://www.enertic.org

69 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

Where:  SUPPLY: . Retail colocation supply comprises of fitted DC space only; un- built shell phases of the DC are excluded. . Wholesale colocation supply includes both fitted and shell DC space. Typically wholesale operators sell shell space, which is built out to suit customers.  AVAILABILITY: . Retail availability of space is based on fully fitted space vacant and available to sell. . Wholesale availability is based on all vacant space.  VACANCY RATE: . The vacancy rate is a product of availability/total supply.  COLOCATION TAKE UP: . This comprises DC space committed to at retail and wholesale colocation facilities in the year (2014). The aforementioned indicators, above all the “size” indicator, are taken into account in the DC market analysis because, more space provided for IT equipment facilitates the increase of the energy consumption; indeed, it is growing the relationship between the size and energy consumption. However, in order to provide a clear picture of the DCs’ market other indices, such as the power density, must be included. Currently, it is difficult to find the information regarding the size (in terms of sq. m.) and energy consumption of the most important DCs in operation in Madrid that we have taking into account, approximately 15 (see the Appendix) for our analysis. In the following we provide a map regarding the collocation of DCs in the city of Madrid.

Figure 51 – Map of the most important DCs in Madrid64

64 Source: http://www.datacenter9.com/datacenters/spain

70 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

V.5.1.b. Business Model andSLAs

In this study, the most important DCs in Madrid are considered. In details, the analysis about the business model is conducted on 15 DCs, which are considered the most important in the city of Madrid. Taking into account the three categories that we mentioned in the methodology section - Colocation, Cloud, Applications hosting - we provide the following results: - The 93% of the DCs are co-location DCs which offer carrier neutral services; - The 67% of the DCs offer Cloud services. The Figure 52, presents the percentage of applications hosting and also shows the additionally offer services like presents financial services, telephone services. DCs’ wide range of business services can be best summarized in 6 categories, as shown in the Figure 2: 1. Cloud; 2. Financial services; 3. Web-Hosting; 4. VOIP-services; 5. Telephone-services; 6. Managed-services. As shown the Figure 52, approximately the 67% of the DCs in Madrid provide a cloud services platform and only the 13% provide a financial service.

Business Models 15

12

9 10 6

3 5 2 4 3 3 0

Figure 52 – Business models of DCs in Madrid65 On the web site of several DCs’ in Madrid, different SLAs are mentioned related to the network availability, the DC availability, the provided security, cooling systems. However, the information regarding the SLAs is very generic and for this it is impossible to conduct an analysis in depth. Moreover, it is also difficult to find the data and details regarding the process, the workloads; for example, the aspect the alignment of workload to renewable energy sources (flexibility mechanism) are not mentioned in the DC’ Madrid profile, but this could be a potential topic to investigate.

65 Source: the websites of the most important (15) DCs in Madrid.

71 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

V.5.2. Energy context in Madrid

The Energy Balance for the city of Madrid in 2013 [MAD15] was published in October 2015 by the City Hall’s Environment and Mobility Governance Area. It provides an overview of the energy context at city level and the recent evolution. Comparing to Barcelona’s consumption distribution, petroleum products represent a larger proportion (see Figure 53). This can be explained by a more widespread use of gasoil boilers for heating, further increased by the greater heating needs in Madrid. Another possible factor is that transportation represents a higher fraction of energy consumption in Madrid (29% in 2013) [MAD15] than in Barcelona (23% in 2012) [BCN12], possibly because of larger transportation distances due to its greater surface. According to the report, energy self-sufficiency (consumption of energy from local sources such as PV, and heat and electricity from waste) in Madrid reached 2.6% in the year 2012. The rest of the city’s energy demand was covered with imported electricity or fuels. In the case of electricity, generation within the city supposed 6.6% of consumption, including CHP, PV and waste (generation from biogas or direct solid waste incineration).

Energy consumption by source (GWh) Renewables Coal (thermal and 245 fuels) 0.6% 549 1.4% Petroleum Natural products gas 15,073 10,170 38.4% 25.9%

Electricity 13,179 33.6%

Figure 53 – Final energy use in Madrid by form, 2013 [MAD15] Table 37 – Summary of the current electricity situation in Madrid [MAD15]

Electricity consumption 2013 13,179 (GWh)

Electricity generation 2013 (GWh) 870.2 (6.6%)

Renewable and waste generation 319.8 (2.4%) 2013 (GWh)

Biogas + Urban waste (GWh) 298.10

Solar PV66 (GWh) 21.73

66 The document provided an aggregated value for solar PV+thermoelectric. This value is taken as PV, since no thermoelectric plants have been found in Madrid’s RIPRE. Furthermore, thermoelectric plants are generally large-scale plants, which are not frequent within cities.

72 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

V.5.2.a. Regulatory framework and action plans by Public Administrations

V.5.2.a.1. Local goals and action plans Madrid’s local administration has developed an Energy and Climate Change plan with 2020 as horizon, aiming to reduce greenhouse gas emissions and improve energy intensity67. However, this plan does not show explicit RES generation goals or previsions.

V.5.2.a.2. National and regional regulation National regulation for Spain is explained in the section corresponding to Barcelona (see IV.1.2. ). Madrid’s regional authority (Comunidad de Madrid) provided effective subsidies for the installment of RES plants awarded by the regional administration during the years 2011 and 201268. In this case, regional legislation helped to overcome the legal uncertainty caused by national regulation.

V.5.2.b. Current situation on the renewable energy production and projection to 2020

V.5.2.b.1. Current situation Madrid’s Energy Balance from 2013 has been examined, and a value of 21.73 GWh has been obtained. Supposing 1,250 equivalent working hours for the existing plants (reference value in Spain for PV plants), this PV production corresponds to an installed power of 17,384 kWp. Table 38 – Madrid current situation

PV energy production 2013 (GWh) 21.73

Installed PV power 2013 (kWp) 17,394

According to the same document, no wind power plants existed at the time, and according to Madrid’s RIPRE69 no plants exist at the moment. PV production in Madrid has grown steadily throughout the last years, even after regulatory changes at national level (see section IV.1.2.b. ), probably due to regional policies that encouraged the installation of renewable power plants, as explained in section IV.1.2.b. . Madrid’s RIPRE has been consulted in order to assess the distribution of RES power by plant size within the city.

67 http://www.madrid.es/portales/munimadrid/es/Inicio/Energia-y-cambio- climatico?vgnextfmt=default&vgnextoid=0ca36936042fc310VgnVCM1000000b205a0aRCRD&vgnextc hannel=1ccd566813946010VgnVCM100000dc0ca8c0RCRD&idCapitulo=6825716 68 http://www.renovablesmadrid.com/ 69 Registry of Special Regime (DER and most renewables) plants held by the Spanish Ministry of Energy, Industry and Commerce https://oficinavirtual.mityc.es/ripre/informes/informeinstalaciones.aspx

73 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

25

20

15

10

5 PV production PV (GWh)

0 2006 2007 2008 2009 2010 2011 2012 2013

Figure 54 – Evolution of PV production 2006-2013 [MAD15] Table 39 – Installed PV power distribution by size70

Power range Installed power (%) Number of plants (%) 100 kW -1250 kW 29% 2% 50 kW – 100 kW 46% 17% 20 kW – 50 kW 12% 11% 1.7 kW – 20 kW 14% 70%

V.5.2.b.2. Growth to 2020 Barcelona and Madrid have applied different policies regarding the promotion of RES. Nevertheless, the results have been similar until national regulation changes introduced in 2012. Taken into account this evolution, and recent regulatory changes, it is foreseen that future evolution is also similar. Based on the RIPRE’s plant distribution by size, and foreseeing the same relative increment for Madrid than the one expected in Barcelona for the period 2012-2020, a distribution of plants by size can be projected for 2020 (see Table 40). Table 40 – Expected distribution of PV installed power in Madrid, 2020

Power range Installed power (kW) Number of plants 100 kW – 1,250 kW 10,896 17 50 kW – 100 kW 17,188 186 20 kW – 50 kW 4,426 122

1.7 kW – 20 kW 5,145 764

Total 37,654 1,089

V.5.3. Smart City Policies in Madrid

Madrid is the capital and largest city of Spain. Madrid contributes 11.7% Spain’s GDP71 regarding the energy situation, the plan was provided by previous paragraph. In terms of strategic action for Smart Cities, the city of Madrid is involved in many initiatives

70 https://oficinavirtual.mityc.es/ripre/informes/informeinstalaciones.aspx 71 http://www.mayorsinaction.eu/fileadmin/user_upload/general_folder/Co- Power_events/Madrid/2_Castano_City_of_Madrid_Energy_and_Climate_Change_Plan.pdf

74 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities such as:  energy and climate change action plan (2014);  air quality plan;  sustainable urban mobility plan;  urban planning;  RIPLE plan;  energy and Climate Change Action Plan Horizon 2020. In the city of Madrid, the city managers are directing their efforts in order to improve the specific priorities of the city. “For instance, in the case of Málaga or Amsterdam there is a strong push from the respective energy players involved. In cities such as Madrid and Stockholm, elements like public safety and traffic congestion management, and water management in the case of Madrid, were the initial starting points. In other cities, such as Santander and Göteborg (the ZigBee City), the first step was the creation of a pervasive communication infrastructure”72. Moreover, it is interesting to show the latest commitments in terms of Smart cities policiies to Madrid; for example on “market Place of the European Innovation Partnership on Smart Cities and Communities” web site73 it is possible to find a list if this commitments such as:  7001 - PRIPARE Smart City Strategy (ICT-SECTOR) - To ensure that ICT solutions integrated in EIP smart cities will be compliant with future privacy regulation (2015);  6817 - OPTICITIES Development of public/private partnerships and the experimentation of innovative ITS services (TRANSPORT AND MOBILITY, ICT- SECTOR) - Development of public/private partnerships and the experimentation of innovative ITS services (2015);  4817 - Smart cities framework for the Mediterranean area (ENERGY, TRANSPORT, MOBILITY, ICT - SECTOR) - Policy & Regulations / Integrated Planning (2014). According to the last analysis concerning the strategy of the city of Madrid in terms of Smart Cities [MEL14], in order to achieve increasingly higher level of GDP pc, the SC plan could be articulated around a further development of ICT, an improvement of multimodal accessibility, greater efficiency of the public transport network and an increased productivity and the size of the companies. The results obtained by [MEL14], based on the Urban Audit database, provide a picture of the Smart cities of Madrid taking into account the variables, which reflect several aspects that characterize a SC. As suggested by Caragliu et al. [CBN], the aspects included for example the urban capital, urban accessibility, public transport and the penetration of the information society. For example, one of the indicators available in the Urban Audit database to verify the integration of the new technologies in the development of cities is the percentage of households with Internet access (for more details see [MEL14])

V.5.4. Summary

“Madrid has traditionally been the Tier 1 (or smaller than Tier 1-2007) European Data Center Markets. This is because although Madrid is the administrative capital of Spain, the commercial and corporate communities are more disparate, with both Madrid and Barcelona as key bases. As such, Madrid has a relatively small corporate occupier rate compared to other Tier 1 cities like Frankfurt, Paris, London, Amsterdam and Barcelona. As we have seen elsewhere, notably

72 http://www.portalidc.com/resources/white_papers/IDC_Smart_City_Analysis_Spain_EN.pdf 73 https://eu-smartcities.eu/place/madrid

75 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities in Paris, as corporate requirements grow in size, they begin to procure space directly, and as such we expect to see direct corporate procurement begin to grow in Madrid”74. Since, Spanish capital’ corporate community increasing, and as such we are more optimistic about Madrid's medium term performance12. The investments in DCs could become an attractive niche segment in the Madrid market. Despite a good market outlook in terms of renewable energy production (RIPRE’75 plan), smart cities initiatives, the investors remain cautious in terms of investments in DCs (as we can see in the report CBRE-Q2_2015). As marketability improves, it may be possible to introduce new investment on DC in terms of renewable source. Madrid is leadership in the development of solar, thermal and wind power, and this aspect is crucial for the flexibility mechanism and in order to adopt the DC4Cities solution. Indeed, in view of Madrid’s renewable energy market the DC4Cities solutions can be improve and accelerate gradually the renewable energy market for DCs. Moreover, the recent successes of Frankfurt and London serve to highlight the current disparity developing between the Madrid and Paris DCs’ market. Indeed, even if the Madrid key strength is the development of energy renewable resources, the disparity between the Madrid and Frankfurt or London DCs’ markets is still significant. Indeed, side from cloud, connectivity remains at the core of the majority of new requirements; indeed, those centres recognised as being connectivity significant receiving heightened interest. A contrasting situation exists in Paris and Madrid where demand, largely driven by localised corporate occupiers, is somewhat sporadic with market conditions remaining more challenging [CBR14]. From the DC’s market perspective, if we take into account that Madrid in 2015 is no longer considered part of the major DCs’ market (Frankfurt, London, Amsterdam and Paris) then, the challenge will be first of all improve this market and after, to adopt an DC ‘energy saving solution as DC4Cities. Regarding DC4Cities impact, even taking into account important improvements expected in Madrid in terms of renewable energy usage (local PV and non-manageable renewables – mostly wind power – outside the city) the disparity between the Madrid and Frankfurt or London DC4Cities markets will still probably be significant due to the large differences in terms of DC market. Hence, from the renewable perspective, the expected distribution of PV installed power for 2020 will be of 37,654 kW for 1,089 PV systems, therefore a significant growth in installed power is expected for the next few years. At this condition, thanks to the PV installed power predictions, Madrid can be considered suitable location in order to apply DC4Cities. From the energy point of view, Madrid and Barcelona show similar evolutions in terms of renewables (PV installed within the city, relevant wind power penetration in the grid). Therefore, the application of DC4Cities will be the same: maximizing self-consumption of locally installed PV, and adaptation to wind power availability in the grid, which will shift grid electricity consumption to valley hours. Thus, the impact and penetration of DC4Cities in DCs will probably be similar in both cities. Moreover, from the smart city point of view Madrid could be also a suitable city to apply DC4Cities, since several of smart cities framework and initiatives (see section V.5.3. ) are put in place.

74 Source: http://archive.datacenterdynamics.com/focus/archive/2008/08/madrid-europes-minor- market 75 https://oficinavirtual.mityc.es/ripre/informes/informeinstalaciones.aspx

76 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

VI. CONCLUSION AND OUTLOOK

This work has provided an overview of development in the European DC market and smart city policies in Europe as framework for marketing a DC4Cities based approach. It has analysed in detail the potential impact of DC4Cities in Barcelona and additionally has delivered thorough analyses of the “Big Five” DC hot spots in Europe which by more than mere coincidence are also among the farthest developed European smart cities. As a result it can be stated that even though the hype of energy saving in DCs is receding, there is a very realistic evaluation from the part of DC management about the future increase of the share of renewable sources in the energy/electricity mix of the DC. With electricity supply being the life-line of DCs all respondents expressed a high interest in the concept of integrating renewable electricity supply into their daily operations. The overview about smart city policies and states showed that there is a great disparity in interest in renewable energy shares and geographical locations – all in all, smart cities in northern, western and central Europe responded more actively to a questionnaire raising the topic of how to deal with sensor-based mushrooming big data from an energy perspective. Southern smart cities have a slightly different focus on how to become smart. This is also corroborated by the analysis of the focus cities.  For Barcelona the prospects to create a veritable DC4Cities market are very good, considering that the smart city is launching studies with the aim of evaluating the creation of a public organism with some of the functionalities envisioned for the EMA-SC, a metropolitan energy operator that will manage the public renewable power plants. On the other hand, the prevailing business model in DCs is cloud and application services which offer a lot more flexibility compared to colocation. The impact of DC4Cities operations on the energy consumption in Barcelona is estimated at around 3 GWh, with RenPercent improvements in 2020 expected to be more than 10 percent points for a sample DC. Also the city’s policy to foster and strengthen local energy self-sufficiency can be a driver to market DC4Cities, even more if the city decided to accompany this endeavour with investment in incentives.  Also for Amsterdam the results give a fairly positive evaluation and an even more positive outlook. With a share of 11% of the total commercial electricity consumption, modifications of DC power profiles can have a considerable effect on the power grid even today. Both an increase of DC consumption and ambitious targets to reduce CO2 emissions by 40% until 2025 and 75% until 2050 (against 1990) relying to a great degree on wind energy creates a market for DC4Cities, even though nowadays renewables are responsible for only 4% of overall city consumption. With an extremely high share of colocation DCs, the number of packages that might be sold today (18) for DCs is comparably low.  Contrary to this, Paris does not offer a huge scope to market DC4Cities. The most important reason for this is that even though also Paris has plans to increase its share of renewable energy sources at the local electricity mix (at the moment 17.9) it focuses on geothermal energy which can be scheduled so that the DC4Cities approach seems unnecessary. However, from the point of solar irradiation Paris is more blessed than e.g. Frankfurt76, and therefore does offer some opportunities for DCs that aim at optimizing their self-generated PV electricity – although again, colocation is the prevailing business model.

76 http://www.espon.eu/main/Menu_Publications/Menu_MapsOfTheMonth/map1101.html#

77 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

 The same applies to London. Unfortunately, even though London is the number one DC hot spot in Europe, both energy and DC related factors clearly limit the potential for a DC4Cities market: nearly all DCs offer colocation services with severe limitations as to the clout on workload management, and even though solar irradiation is not as bad as expected PV plays only a minor role in London’s today’s energy policy. There are plans to cover until 2025 25% of London’s energy needs by local, low carbon energy sources, but there is no public information available regarding how this can be done. Therefore the prospects for a London DC4Cities market might change considerably in the next decade.  The big surprise as to the market potential of DC4Cities in European smart cities is Frankfurt. Compared to other smart cities, Frankfurt’s local energy coverage plans are extremely ambitious: they aim at a rate of 100% until 2050. A feasibility study shows that this goal can be reached relying by more than 2/3 on sun and wind. Together with the information that DC electricity consumption makes up around 20% of electricity demand in Frankfurt, the impact of implementing DC4Cities can be huge and thus the market in Frankfurt should definitely be developed. The only drawback is that nearly all DCs offer colocation services with limited flexibility.  Madrid finally, is an example for the influence of business and politics on the economic viability of DC4Cities: Even though the irradiation of sun is slightly better than in Barcelona and wind is about the same, the prospects for marketing DC4Cities in Madrid are by far not equally good. This is mostly owing to the economic downturn in Madrid, which in the aftermath also reduced the DC market liquidity up to a point where Madrid has ceased to be assessed as DC hot spot in Europe. DCs, though still a local market, seem not to play an important role in energy consumption in Madrid, which can be concluded from the fact that no data on DC electricity consumption on city level are available. Additionally, pursuing smart city plans, the city of Madrid firstly aims at mobility and IT service goals and gives less priority to the investment into driving the utilization of abundant renewable energy sources. As a consequence of these findings we draw the following conclusions: Politics: as already deduced from the first market analysis D2.2 the political and legal framework largely determines the market for DC4Cities. D2.2 related this statement preliminarily to the nature of incentives for DCs to take part in the DC4Cities scheme (resulting in the recommendation to improve e.g. the European ETS scheme). The new findings point into an additional direction: not only the economic viability of DC4Cities for a single DC is greatly influenced by the existence of favourable legal and economic conditions, but also the technical rationale. The more ambitious plans to invest into renewable electricity generation are pursued by a public organization (e.g. city, region, nation, EU), the higher the share of renewables in the electricity generation (e.g. 27% in Germany!), the more volatile the electricity supply in the grid and the more it makes sense to adapt workload to this volatility. Power grid view: Even though power in the grid may be volatile to some extent in the grid; if the share of renewables is low, adapting workload to this volatility might not support but rather stress the grid by shifting workload from times with low overall power consumption to times with high power consumption. Depending on prevailing TOU based energy tariffs, this might even result in higher energy cost if the per unit price depends on day and night shifts instead of the supply of renewables. Smart city policy: From the beginning DC4Cities focused on smart cities as drivers to foster the integration of local and intermittent renewables into the local power grid (distribution grid). It turns out, however, that the level of smartness of a city is less important for the marketability than its ambition to have a high share of intermittent renewable energy sources and its activities to reach this goal. This is also related to the fact that the overall economic situation e.g. being in an upturn or downturn impacts the marketability of DC4Cities.

78 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

EMA-SC versus EMA: DC4Cities is offered for two different scenarios which can be reconciled via smart city policies. The EMA-SC setting envisions an energy management authority that aims at optimizing the consumption of renewable energy on city level with DCs being one consumer group amongst many. EMA on the other hand represents the concept of a DC optimizing autonomously the share of renewable energy in its energy mix. Both approaches can be merged if the smart city adopts a concept of self-sufficient energy supply and thus rewards the priority of consuming on-site generated power. The latter has proven to be the most promising use case for the time being, with the EMA-SC concept as a next step that smart cities could take on a roadmap towards self-sufficient power supply. Urban view: The findings of D2.4 have revealed a high disparity in opportunities for a DC4Cities based approach in cities with even similar sun and wind energy conditions. This fact reinforces the conviction that the level of urban agglomerations is to be preferred to any other level of analysis. Business modelling requirements: Until now, DC4Cities business models targeted DCs that to a large degree operate the workload themselves (this also includes enterprise DCs although they due to a lack of data, they have not been explicitly targeted in this study). Against the background of very high shares of colocation DCs in otherwise suitable smart cities, the development of colocation business models is strongly advised. Digital Realty, the “world's largest wholesale colocation services and data center solutions provider with over 130 secure locations globally”77 has recently started to offer specific contracts that enable their clients to purchase so-called Renewable Energy Credits (REC). A DC4Cities exploitation strategy should plug onto this strategy. This should also extend to attractive offers for European DC chains that could capitalize on geographical federation in order to achieve RenPercent goals.

77 https://www.digitalrealty.com/

79 Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

VII. REFERENCES

[ACP14] Stadt Aachen : Aachen – a european city with energyefficiency and climate protection activities”, Stadt Aachen 21.7.2014 [AEC13] Assembly Environment Committee of London. Solar power in London – Appendix 1. 2013. URL: https://www.london.gov.uk/moderngov/documents/s46857/10a%20Appendix%201%20- %20Solar%20Scoping%20Paper.pdf

[AES10] Amsterdam: a different energy 2040 Energy Strategy , page 9 URL: https://www.amsterdam.nl/publish/pages/295474/energiestrategieamsterdam2040defengels. pdf

[APU14] Alba, D., Bigorne, J; Richard, O.;Senegas, G.:Un Plan Local Energie (PLE) pour Paris et la Métropole, Atelier Parisien d’Urbanisme (APUR), Paris, 2014 URL: http://www.apur.org/sites/default/files/documents/12p81_plan_local_energie.pdf [BCN12] Agència d’Energia de Barcelona, Balanç energètic de Barcelona 2012. [BER13] Paolo Bertoldi, A Market Transformation Programme for Improving Energy Efficiency in Data Centres, European Commission Joint Research Centre, 2013 [CBN]http://www.intaaivn.org/images/cc/Urbanism/background%20documents/01_03_Nijkam p.pdf [CBR14] European data centres market view 2014 Q1, CBRE report, 2014 [CBR 15] CBRE Market View, European Data Centres, Q1 2015, CBRE, 2015 URL:http://www.cbre.us/services/office/AssetLibrary/Q1%202015%20European%20MarketVi ew.pdf [CEO 13] Data centres establishment policy, The Amsterdam region as Green Data Port, City of Amsterdam, 2013, page 32 [COC15] Data Centres Energy Efficiency: EU DC Code of Conduct, 2015 URL: http://iet.jrc.ec.europa.eu/energyefficiency/ict-codes-conduct/data-centres-energy- efficiency, [COP14a] City of Paris: Plan Climat Énergie de Paris – Actualisation 2012, City of Paris, 2014; URL: http://api-site-cdn.paris.fr/images/70921 [COP14b] City of Paris: Bleu Climat – Annexé au BP 2015, City of Paris, 2014; URL: http://api-site-cdn.paris.fr/images/152847.pdf [COV12] Feldmann, P.: Every day greener, Covenant of Mayors Committed to local sustainable energy, Frankfurt am Main, 2012, URL: http://www.eumayors.eu/IMG/pdf/Frankfurt_Case_Study_Covenant_Mayors.pdf [CPR15] Gary Cook, David Pomerantz, Kassie Rohrbach, and Brian Johnson, Clicking Clean: A Guide to Building the Green Internet, May 2015, URL: http://www.greenpeace.org/international/en/

[DRI12] DRIEE, Ile-de-France: Approvisionnement électrique du Grand Paris, Séminaire sur l'approvisionnement électrique du Grand Paris – 6 mars 2012, Paris, 2012 URL: http://www.driee.ile-de-france.developpement- durable.gouv.fr/IMG/pdf/Seminaire_du_6_mars_2012_- _Presentations_Approvisionnement_electrique_du_Grand_Paris__cle2dcb2f-1.pdf

ix Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

[DRI13] Direction Régionale et Interdépartementale de l’Environnement et de l’Énergie d’Île- de-France: Île-de-France:La soutenabilité du Grand Paris - L’approvisionnement énergétique du Grand Paris Note de problématique; 2013 URL: http://www.driee.ile-de-france.developpement-durable.gouv.fr [FRA15] Stryi-Hipp, G.; Steingrube, A.; Eggers, J.: „Optimierte Energiesystemszenarien für Frankfurt am Main im Jahr 2050, 2015 https://www.ise.fraunhofer.de/de/downloads/pdf- files/gf-60-systemintegration-und-netze/praesentation-optimierte-energiesystemszenarien- fuer-frankfurt-am-main-im-jahr-2050.pdf [GLA11] Mayer of London. Delivering London’s Energy Future. The mayors climate change mitigation and energy strategy. Greater London Authority, October 2011. Pages 7-10. URL: https://www.london.gov.uk/sites/default/files/Energy-future-oct11-exec-summ.pdf [HIN14] Hintemann,R.: Energy consumption of data centers continues to increase in 2014, Borderstep Institute, 2014 [LON14] Stephen Jones Associates, South East Economics: “The Future of London’t Power Supply”, City of London, 2014; URL: https://www.cityoflondon.gov.uk/business/economic-research-and- information/research-publications/Documents/Research-2014/Londons-power-supply-full- report.pdf [MAD15] Fundación para el Fomento de la Innovación Industrial, Escuela Técnica Superior de Ingenieros Industriales de la Universidad Politécnica de Madrid, Ayuntamiento de Madrid: Balance Energético del Municipio de Madrid 2013, Madrid, 2015 [MEL14] Josè M. Mella-Marquez at al,: European Smart Cities: the case of Madrid (Spain), Autonoma University of Madrid, 2014 [MSC14] Manville, C., Cochrane, G., Cave, J., Millard, J., Pederson, J.K., Thaarup, R.K., Liebe, A., Wissner, M., Massink, R., Kotterink, B.: Directorate-General for Internal Policies – Policy Department A: Economic and Scientific Policy: Mapping Smart Cities in the EU, IP/A/ITRE/ST/2013-02, PE 507.480, Brussels, 2014 [NAU13] Neumann, W.: Frankfurt am Main Masterplan - 100 % Klimaschutz bis 2050“, 2013 URL: http://www.masterplan100.de/fileadmin/template/downloads/2013-06- 13_Auftakt_100__erneuerbar_Vortrag_we.pdf [PEC08] Agència d’Energia de Barcelona, Plan de energía, cambio climático y calidad del aire de Barcelona (PECQ 2011-2020), 2008 [REG15] Regionalverband FrankfurtRheinMain, „Kommunaler Energiesteckbrief 2013“, 09.03.2015, URL: http://www.energiewende- frankfurtrheinmain.de/fileadmin/user_upload/content/pdf/Komm_Energiesteckbriefe/Energies teckbrief_Frankfurt_am_Main__krsfr._Stadt.pdf [SCH15] Schumacher, P.; Stroh, K.; Schurig, M.; Ellerbrok,C.; Ramonat, A.; Link, S.:Generalkonzept im Rahmen des Masterplans „100% Klimaschutz“ der Stadt Frankfurt am Main, 2015 URL: https://www.frankfurt.de/sixcms/media.php/738/Masterplan_Klimaschutz_Generalkonzept_La ngfassung.pdf [STA15] COMMISSARIAT GÉNÉRAL AU DÉVELOPPEMENT DURABLE: Repères: Chiffres clés de l’énergie Édition 2014, Service de l’observation et des statistiques; 2015 URL: http://www.statistiques.developpement-durable.gouv.fr/publications/p/2369/969/chiffres- cles-lenergie-edition-2014.html [TER15] Terna: Produzione di energia elettrica in Italia, 2015, URL: http://www.terna.it/

x Project Nº 609304 D2.4 – Final Market Analysis 29.11.2015 DC4Cities

VIII. APPENDIX VIII.1. Overview Data Centres in Europe

PUE Business Power Energy #sites Data source model Many sites 4000 Cloud, 1.5 Surveyed Questionnaire Mwh #Server: Colocation 10000 Avg:700kW 5000 1.65 1 Colocation Questionnaire Max: 800kW Mwh 12 locations (cities): 41 sites Amsterdam 6 Dublin 4 Manchester: 1.35 2015: 117.1 Frankfurt 2 MW 80 Helsinki: <1.3 Helsinki 5 Telecity 78 H1 2014: Istanbul 1 Colocation Internet 106.3MW FY Manchester London: 1.35 2014: 111.7MW 79 4Milan 3 London: 29MW Milan 3 Paris 3 Sofia 1 Stockholm 3 Warsaw 2 London 7 EU: 29 DCs 81 Global: EMEA: EMEA: 1.29- Equinix 2200 130,000 Colocation 1.42 GWh sqm Internet Global: 100+

Amsterdam: 11.5 MW 1.15 EU: 22 DCs Digital 16 MW Colocation Dublin: 1.15 Global: 131 Realty 9 MW Cloud London1: DCs Internet 1.28 30 MW

78 http://www.telecitygroup.com/our-company/news/2013/telecitygroup-launches-new- state-of-the-art-data-centre-in-helsinki-finland.htm 79 http://www.telecitygroup.com/data-centres/london-data-centre-powergate-tour.htm 80 http://telecitygroup.com/presentations/Telecity-Interim-report-2015.pdf 81 http://www.equinix.com/locations/europe-colocation/europe-data-centers/

xi Project Nº 609304 [Document title] [Date] DC4Cities London2: 1.3 5 MW to 1.6 Paris: 1.4 40 DCs Stockholm: Colocation Interxion 82 11 EU 1.09 Cloud Internet countries 8 sites Frankfurt: Germany eShelter 213MW Colocation Switzerland Internet EU: 316MW Austria London1<1.5 15.6 MW VIRTUS 2 Colocation London2<1.2 28 MVA Internet 6 UK:5 sites Colocation CenturyLink 42 MW Frankfurt: 1 Cloud Internet site London: 33MVA Spian: 12MVA Lisbon: 0.8MVA Italy: 5.5MVA Colt London: 1.22 EU: 22 Colocation Germany: Internet 15.68 MVA France: 24.9 MVA Netherlands: 43.1 MVA Colocation Interroute 40 Cloud Internet Min: 1.25 111 Colocation, Internet Avg: 1.7 6 TWh companies Cloud, EU DC CoC Max: 2.86 255 DCs Application

82 http://www.datacenterknowledge.com/archives/2013/03/06/interxion-uses-seawater- to-cool-stockholm-data-centers/

xii PUE Business Power Energy #sites Data source model Many sites 4000 Cloud, 1.5 Surveyed Questionnaire Mwh #Server: Colocation 10000 Avg:700kW 5000 1.65 1 Colocation Questionnaire Max: 800kW Mwh 12 locations (cities): 41 sites Amsterdam 6 Dublin 4 Manchester: 1.35 2015: 117.1 Frankfurt 2 MW 80 Helsinki: <1.3 Helsinki 5 Telecity 78 H1 2014: Istanbul 1 Colocation Internet 106.3MW FY Manchester London: 1.35 2014: 111.7MW 79 4Milan 3 London: 29MW Milan 3 Paris 3 Sofia 1 Stockholm 3 Warsaw 2 London 7 EU: 29 DCs 81 Global: EMEA: EMEA: 1.29- Equinix 2200 130,000 Colocation 1.42 GWh sqm Internet Global: 100+ Amsterdam: 1.15 11.5 MW Dublin: 1.15 16 MW EU: 22 DCs Digital London1: Colocation 9 MW Global: 131 Realty 1.28 Cloud 30 MW DCs Internet London2: 1.3 to 1.6 5 MW Paris: 1.4 Stockholm: 40 DCs Colocation Interxion

78 http://www.telecitygroup.com/our-company/news/2013/telecitygroup-launches-new- state-of-the-art-data-centre-in-helsinki-finland.htm 79 http://www.telecitygroup.com/data-centres/london-data-centre-powergate-tour.htm 80 http://telecitygroup.com/presentations/Telecity-Interim-report-2015.pdf 81 http://www.equinix.com/locations/europe-colocation/europe-data-centers/

xiii Project Nº 609304 [Document title] [Date] DC4Cities 1.09 82 11 EU Cloud Internet countries 8 sites Frankfurt: Germany eShelter 213MW Colocation Switzerland Internet EU: 316MW Austria London1<1.5 15.6 MW VIRTUS 2 Colocation London2<1.2 28 MVA Internet 6 UK:5 sites Colocation CenturyLink 42 MW Frankfurt: 1 Cloud Internet site London: 33MVA Spian: 12MVA Lisbon: 0.8MVA Italy: 5.5MVA Colt London: 1.22 EU: 22 Colocation Germany: Internet 15.68 MVA France: 24.9 MVA Netherlands: 43.1 MVA Colocation Interroute 40 Cloud Internet Min: 1.25 111 Colocation, Internet Avg: 1.7 6 TWh companies Cloud, EU DC CoC Max: 2.86 255 DCs Application

VIII.2. Data Centres in Barcelona

VIII.2.1. Public DC table

Numbe Data Centre URL Business Mode Location Additional Data r

1 Adam www.adam.es Colocation/Clou Cerdanyola del Vallés TIER III d PUE 1.2 Free Cooling

2 ITconic www.itconic.com Cloud Carrer de l'Acer, 30 TIER III (Barcelona) 1700 m2

82 http://www.datacenterknowledge.com/archives/2013/03/06/interxion-uses-seawater- to-cool-stockholm-data-centers/

xiv Project Nº 609304 [Document title] [Date] DC4Cities

3 Colt www.colt.net Colocation Carrer de l’Acer, 26 TIER III (Barcelona) 2,160 m2

4 Ibermática www.ibermatica.com Cloud Travessera de Les TIER III Corts, 39-55 (Barcelona)

5 NTT www.eu.ntt.com/es Cloud/Managed Carrer Numancia, 164- TIER III services 168 (Barcelona)

6 T-Systems www.t-systems.es Cloud/Managed Carrer Sancho de Ávila, TIER III services 110 (Barcelona) PUE 1.3 Free Cooling 80% hours

7 Acens www.acens.com Cloud Carrer Tarragona, 161 TIER III (Barcelona)

8 Claranet www.claranet.es Cloud/Colocatio Carrer Juan Gris, 10-18 TIER IV n/MS (Barcelona)

9 HP www.hp.com Cloud Sant Cugat del Vallès TIER III

10 IBM www.ibm.com Cloud Cerdanyola del Vallès TIER III

11 Level(3) www.level3.com Colocation Hospitalet de Llobregat TIER III

12 Mast Cloud www.mastcloud.net Cloud/MS Cerdanyola del Vallès TIER IV

13 Nexica www.nexica.com Cloud/MS/App Carrer de l’Acer, 30-32 TIER III Hosting (Barcelona)

14 Media Cloud www.mdcloud.es Cloud Av. Diagonal, 177 TIER IV (Barcelona) 800 m2

15 BT www.bt.es Colocation/Clou Hospitalet de Llobregat TIER III d/MS 1000 m2

VIII.2.2. Additional information Public DCs

VIII.2.2.a. 2-ITconic

Number of racks – 850 kW consumed – 1049 kWh consumed – 9.183.405 % increase from last year - 4% Forecast next year - 6% Number of customers - 89 % increase from last year - 4% Forecast next year - 6% Data volume processed - 117.454 Gbps % Batch/Online - 30/70 Increased vol. from last year - 15% Forecast next year - 20%

xv Project Nº 609304 [Document title] [Date] DC4Cities SLA guarantees: Electricity supply Data room temperature Data room humidity VIII.2.2.b. 3-Colt Number of racks – 874 kW consumed – 2650 kWh consumed – 1.850.00 % increase from last year - 5% Forecast next year - 5% Number of customers - 134 SLA guarantees: Power availability Temperature range Humidity range VIII.2.2.c. 15-BT Number of racks – 400 kW consumed – 739 kWh consumed – 5.805.491 % increase from last year - -22% Forecast next year - same consumption Number of customers - 60 % increase from last year - 5% Forecast next year - 10% SLA guarantees: Power availability Room temperature VIII.2.3. Private DCs

As we said in the section IV.2.1.a, we don`t have a census for the private DCs placed in Barcelona. However, we could identify at least 5 of them that would fit with our objectives. For instance La Caixa, Aigües de Barcelona, Allianz, Port de Barcelona or Henkel would be candidates to study the impact of DC4Cities on their facilities. As far as we know, their characteristics as size, amount of consumption and variety of services offered are aligned with the typical targets for this project.

xvi Project Nº 609304 [Document title] [Date] DC4Cities VIII.3. Data Centres in Amsterdam

data centers web site Description Business Models Size Power small /medium /large Colocation Cloud / IaaS Capgemini https://www.capgemini.com/ Capgemini is located in the area yes Amsterdam, The Netherlands (Holland), and the colocation data center is carrier neutral. The data center was last updated on 31-12-2008. Citadel100 Datacenters B.V. i.o http://www.citadel100.com/ Citadel 100 leading provider of cloud yes computing datacenters in Ireland but demand in Amsterdam mainly for colocation services. Colt Technology Services B.V. http://ww2.colt.net/be/en/product yes Amsterdam Duivendrecht- 742m2 Powered by 2MVA s-services/cloud-services/cloud- Amsterdam Oud-Zuid- 1,100m2 Powered by 1.6MVA infrastructure-service-en.htm Computerline Cloud http://www.computerline.com/ yes AMS1,AMS2,AMS3,AMS7 Databarn http://www.databarn.nl yes datacenter parkovyi ucloud http://datacenter.ua/ Datacenter "Parkovyi" with its UCloud yes service offers wide range of cloud resources in a various models (inc. public, private, hybrid) Digital Realty https://www.digitalrealty.com/ yes enterpise cloud http://www.verizonenterprise.com/ yes solutions/cloud/ Equinix (Formerly known as Virtu http://www.equinix.nl/ Colocation infrastructure services (also yes L- AM2, AM3 Amsterdam) used by cloud providers looking to store 17,800+ m2; 2800 cabinets data) euNetworks Datacenter Amsterdam http://citadel100.com/ With solutions from a single rack to a full yes suite, euNetworks Colocation is the perfect data center solution for Carriers, Cloud operators and Enterprises wanting to host equipment in secure, standardized, and managed premises. EvoSwitch AMS1 http://www.evoswitch.com/ EvoSwitch is a leading provider of secure yes 10,000m2 and hyper-connected carrier-neutral data center services. With sustainable data centers at the internet's major global hubs, EvoSwitch is home to the world's leading content and network providers. EvoSwitch puts a strong focus on market development.

Global Crossing Joop http://www.level3.com/en/ yes Global Switch http://www.globalswitch.com/ yes GYRO Center DC-2 http://www.reasonnet.com/ yes ICTroom Company BV www.ictroom.com yes Interconnect VPC https://www.interconnect.nl/dienst Virtual Private Cloud offers a hosted yes en/cloud/ infrastructure service in which flexibility, certainty and security are seamlessly connected. De flexibility of public cloud computing is combined with the safety and control of a dedicated infrastructure. Interconnect offers Infrastructure as a Service (IaaS) on which you can safely built your company's future. Interoute Amsterdam http://www.interoute.com/ yes Interxion http://www.interxion.com/location Interxion's Amsterdam carrier- and cloud- yes L - AMS1, AMS2, AMS3, AMS4, s/netherlands/amsterdam/ neutral data centres provide the latest AMS5, AMS6, AMS7 (Amsterdam) highly secure, scalable infrastructure for mission-critical IT systems, with a wide range of connectivity solutions. ip2 virtual http://www.ip2.nl/ yes KPN CyberCenter http://www.cybercenter.nl yes LeaseWeb https://www.leaseweb.com yes Level 3 http://www.level3.com/en/ yes Luna cloud https://luna.nl/ yes MESH https://www.plusserver.com/?ref=me yes sh NAP of Amsterdam http://www.verizonenterprise.com/ The NAP of Amsterdam is designed to meet yes 2,700 square metres solutions/cloud/ the highest requirements for power, availability and security. Nikhef http://www.nikhef.nl/ yes Phoenix NAP https://phoenixnap.com/cloud- yes services/public-cloud RDC Groep https://www.rdc.nl/web/show/id=868 yes 71/langid=43 Reasonnet http://www.reasonnet.com/ yes SARA Data Center https://www.surf.nl/en/about- SURFsara creates a bridge between yes surf/subsidiaries/surfsara/ research and advanced ICT. We do so with a passion for scientific research in our DNA and with extensive expertise contained in our high-performance infrastructure. This enables us to facilitate scientific research and develop initiatives for the business community. Switch Datacenter Amsterdam http://www.switchdatacenters.com Switch Datacenters is a supplier of yes 8320 m2 / datacenters with its main product high-end colocation space and services for ISP's, System Integrators and large enterprises. systemec http://www.systemec.nl/ cloud / IaaS yes TelecityGroup Netherlands B.V http://www.telecitygroup.com/ TelecityGroup Amsterdam, Netherlands. yes 8300 m2 TelecityGroup has four facilities in Amsterdam offering approximately 8300 m2 of customer space. Each site offers a secure, resilient environment for hosting customer platforms through a combination of connectivity, security, cooling/fire suppression and systems. Terremark http://www.verizonenterprise.com/sol yes True Dc https://www.true.nl/ True datacenter provides 10.000 ft2 of yes 920 m2 (10.000 ft2) secure and highly-connected environments where bandwidth intensive applications, content and information are hosted.

The Datacenter Group http://www.thedatacentergroup.nl/ Co-location in our Vancis data centres yes mean that your data is in safe and reliable hands. Vancis https://vancis.nl/ yes Versatel https://www.tele2.nl/ yes Virtual Datacenter http://www.previder.com/ Previder brings all the benefits of cloud yes computing within your reach, using Previder Cloud Servers. The Cloud is based in Previder's two privately owned, geographically separated data centers linked together by using redundant fiber optic connections. A disaster in one of the data center thus has no effect on the availability of Previder Cloud. The Previder Cloud is based on hardware and software from industry leaders Cisco, NetApp and VMware. VPS.NET https://www.vps.net yes

xvii Project Nº 609304 [Document title] [Date] DC4Cities VIII.4. Data Centres in Paris

Number Data centers Business Models location Power

Colocation Cloud / IaaS location surface 1 Acropolis http://www.acropolistelecom.net/ yes Paris-Bourse 450m2 Paris-Nation 2 Ad Valem http://www.advalem.fr/ yes Unknown 3 ADC1 yes Unknown 1 MW 4 BT Services http://www.globalservices.bt.com/ yes Unknown 5 CELESTE-Marilyn https://celeste.fr/datacenter yes Unknown 6 Claranet http://www.claranet.com/ yes 500m2 7 Cloudata http://www.cloudata.fr/ yes Plessis Robinson Unknown 8 cloudarchitek http://www.agarik.com/en/cloudarchitek/ yes Unknown 9 cloudmaker http://www.agarik.com/cloudmaker/ yes Unknown 10 Cogent https://www.cogentco.com/ yes Cogent Paris 800 m2 Cogent Paris 2 11 Colt http://datacentres.colt.net/ yes yes Paris, Bessières 9454 m2 3.3MVA Paris, Les Ulis Paris, Wattignies 12 DCC-IDF http://www.completel.fr/ yes Aubervilliers 13 Defense Datacenter http://defensedatacenter.com/ yes 1 MW 14 Easynet Nanterre http://www.easynet.com/ yes Unknown 15 Equinix http://www.equinix.com/ yes Paris (Roissy) 26.000 m2 Paris (Saint-Denis) 16 Evoluto http://www.evoluto.be/ yes Small 17 France Paris 2 Data Center yes 18 Green Data Center (GDC) http://www.greendatacenters.fr/ yes GDC1 3650 m2 GDC2 GDC3 19 Global Switch Paris http://www.globalswitch.com/ yes Clichy-Levallois 54MVA 20 Interoute http://www.interoute.com/ yes yes Interxion PAR1, PAR2, PAR3, PAR4, PAR5, PAR6, PAR7 21 KHEOPS Organisation http://www.kheops.org/ yes yes KHEOPS Organisation 1350m2 1, 2 22 Level 3 Paris http://www.level3.com/ yes 23 LINKBYNET STD http://www.linkbynet.com/ yes yes Unknown 24 MDC VeePee http://mdc.veepee.net/ yes Unknown 25 MESH https://www.plusserver.com/?ref=mesh yes large 26 Mediactive Network http://www.groupe-mediactive.fr/ yes Unknown 27 navlink cloud services http://www.navlink.com/IaaS yes Unknown 28 NeoCenter Paris http://www.fr.zayo.com/ yes Unknown 29 SAVEHOUSE MONTREUIL http://www.saveho.com/ yes Unknown 30 SFR iasiad-cloud https://cloud.sfr.fr/ yes Unknown 31 SFR Netcenter http://pme.sfrbusinessteam.fr/ yes Unknown 32 tas France http://www.tasfrance.com/ yes Unknown 33 TelCo Center http://www.telcocenter.fr/ yes Unknown 34 TelecityGroup Paris http://www.telecitygroup.com/ yes Aubervilliers 3400 m2 14 MW Condorcet Courbevoie Cedex 35 TELEHOUSE Paris http://www.telehouse.net/ yes Jeûneurs 2700 m2 Magny-Les-Hameaux Voltaire 36 TeliaSonera http://www.teliasoneraic.com/ yes Unknown 37 tina https://en.outscale.com/ yes Unknown

xviii Project Nº 609304 [Document title] [Date] DC4Cities VIII.5. Data Centres London

xix Project Nº 609304 [Document title] [Date] DC4Cities VIII.6. Data Centres Frankfurt

xx Project Nº 609304 [Document title] [Date] DC4Cities

xxi Project Nº 609304 [Document title] [Date] DC4Cities

VIII.7. Data Centres Madrid

Data Centers Web-site Business Models Location Power

Co-location Cloud/IaaS Location Surface

Telvent Carrierhouse Madrid#1 http://www2.schneider-electric.com Yes Yes Madrid n/a n/a

Interxion Madrid (MAD1-MAD2) http://www.interxion.com/es/ Yes Yes Madrid 4,800 m² n/a

ONO II - Atocha http://www.ono.es/empresas/internet/servicios-adicionales/ No n/a Madrid n/a n/a

Iberbanda Madrid http://www.iberbanda.es/ Yes Yes Madrid 4500 n/a

Interoute Lezama http://www.interoute.es Yes Yes Madrid >100,000 1 MW

Ibercom Telecom http://www.ibercom.com/srv/srv_index.php Yes Yes Madrid >1000 n/a

Eurociber http://www.eurociber.es/ Yes Yes Madrid n/a n/a

TP29 http://www.prosodie.es Yes Yes Madrid n/a n/a

NAP de las Americas Madrid http://www.verizonenterprise.com/solutions/cloud/ Yes Yes Madrid 1000 n/a

GNET Data Center http://www.gnet.es Yes Yes Madrid n/a n/a

inAsset | NixMad https://inasset.es/corporate/ Yes Yes Madrid n/a n/a

Global Switch Madrid http://www.globalswitch.com Yes n/a Madrid 21922 18MVA

IPCORE http://www.ipcore.com Yes n/a Madrid 1200 n/a

Bitcanal http://www.bitcanal.es/ Yes n/a Madrid n/a n/a

Hispaweb Network http://www.hispaweb.net Yes n/a Madrid n/a n/a

xxii Project Nº 609304 [Document title] [Date] DC4Cities

xxiii