E^2DC 2014 Cambridge, UK 30/06/2014 1 Enabling Green Data Centres in Smart Cities Massimo Bertoncini, Engineering Ingegneria Informatica, Project Coordinator Vasiliki Georgiadou, Green IT Amsterdam Green networked data centres as energy prosumers in smart city environments Project FP7 SMARTCITIES 2013 Grant Agreement n : 609211 3rd International Workshop on Energy Efficient Data Centres Cambridge, UK June 10, 2014
Project Identity Card E^2DC 2014 Cambridge, UK 30/06/2014 2 Acronym: GEYSER Full Title: Green networked data centres as energy prosumers in smart city encironments Starting Date: 1 November 2013 Duration: 36 months EU Maximum financial grant: 2.979.000,00 Partners: 10 Country coverage: Italy, Greece, Ireland, Romania, Germany Switzerland, The Netherlands
The Consortium E^2DC 2014 Cambridge, UK 30/06/2014 3
GEYSER Consortium E^2DC 2014 Cambridge, UK 30/06/2014 4 End users ASM Terni (Power DSO) SARA (DC Operator), GITA (Green IT/Smart City ) ENG (DC Owner/Cloud Service Provider) Industrial Players: ICT Players (ENG, SiLO) Smart Grids/Data Centres Technology/Solution Providers (ABB) Smart Energy Intelligent SaaS Providers (Wattics) Academy/R&D TUC, RWTH, ZHAW Builds on the legacy of the FP7 GAMES European R&D project
Background: Data centres ARE a local affair! Data centre energy efficiency is clearly becoming more and more an urgent issue to address for DC Operation and Strategic Managers Need to focus on intelligent solutions addressing DC performance AND energy management and optimization One solution to mitigate energy costs has recently been data/application migration trend towards free-cooled data centres located in Norway, Iceland or in the middle of some cold nowhere ( follow-the-least-energy cost strategy) Major drawbacks Customers security concerns Data networking costs are not negligible and affect the overall performance cost Only a financial-driven exercise No impact on the overall environmental sustainability and resource efficiency On the other hand a significant share of data centres are situated within urban contexts (Bank. Hospitals, public/private offices) E^2DC 2014 Cambridge, UK 30/06/2014 5
Smart City and Local Self-sustainable systems E^2DC 2014 Cambridge, UK 30/06/2014 6 The trend is for managing and solving locally (at district/city levels) the major city challenges/requirements, reducing in this way value chain length and costs (es. Sustainable agriculture, zero waste, local energy supply systems reducing transmission losses) The long-term goal is to move forward towards sustainable locally optimized urban contexts
Local Sustainable Energy Supply Systems Rising shares of fluctuating Renewable Energy Sources (PV farms, Wind farms, biomass) are distributed along Low or Medium Voltage branches of the electricity networks However electricity networks traditionally are not specifically designed to accommodate such surplus power and, above all, the bi-directional power flow occurring when a surplus of power locally generated could not be consumed close to the generation point and should be transported somewhere The integration of the power generated by distributed RESs has posing serious stability and security problems to the power network operators Smart Grid has emerged in the last years as an effective organizational and technological paradigm to allow power distributors to proactively manage their actual power network, and accordingly defer significant investments for grid expansion E^2DC 2014 Cambridge, UK 30/06/2014 7
Smart grids: Flexible Management of Active Energy Resources (1/2) E^2DC 2014 Cambridge, UK 30/06/2014 8
Smart grids: Flexible Management of Active Energy Resources (2/2) E^2DC 2014 Cambridge, UK 30/06/2014 9 Peak Shaving Optimized consumption Power back to grid Load Levelling Power back to grid Optimized consumption
E^2DC 2014 Cambridge, UK 30/06/2014 10 The state of the art No active links among DCs and Smart Cities (no energy exchange) No or very few links with other DCs (no or very few IT load exchange)
Data Centre versus System-level Energy Efficiency Urban Data Centres are operated in uncoordinated way Their Energy efficiency has been so far addressed in an optimal, yet isolated way Urban Data Centres have a large yet largely unexploited potential to contribute to smart city local energy consumption optimization and smart grid optimized operation.....provided that DCs would decide to come out from their ivory tower Energy efficiency should be addresses in a holistic way, by considering the interaction of Data Centres with the relevant stakeholders within urban contexts (electricity distributors, district heating operators, energy suppliers, smart city/district energy managers) E^2DC 2014 Cambridge, UK 30/06/2014 11
Towards a new active role for Data Centres DCs are expected to turn on into flexible energy players at the crossroads of Smart City and Smart Energy Grids DCs may become collaborative stakeholders within the framework of upper level smart city ecosystems DC Energy Efficiency should be managed in synergistic combined way with system-level energy efficiency DC may gain substantial (financial) benefits from the integration and collaboration with Utilities (Power Distributors, Energy Traders,...) Acting as Energy Prosumers matching utilities reqs may lower DC Energy Efficiency or Operational Performance (and potentially infringe SLAs with customers) However fair incentive sharing will reward DC at the extent to outperform penalties due to lowering of energy efficiency or operational performance E^2DC 2014 Cambridge, UK 30/06/2014 12
The GEYSER vision Think Global Act Local (GLOCAL) E^2DC 2014 Cambridge, UK 30/06/2014 13
Data centres as active connection hubs at the crossroads of Data and Energy Networks E^2DC 2014 Cambridge, UK 30/06/2014 14
The GEYSER approach for integrating Data Center in Smart grids and Smart City Data Centres as adjustable adaptive power consumers Follow-the-sun strategies Use intermittent renewable energy sources when they are available Innovative Value Proposition and Novel service offering to smart grids and utilities operators System services offering tailored to local balancing markets (Voltage regulation, reactive power, congestion management, load leveling, peak shaving Energy (power, heat) retailers/suppliers/traders and Smart Ditrict/City Energy Manager Energy consumption flexibility to optimize the global district/city energy consumption While... Optimizing IT load management matching utilities needs for services aimed at more efficient power network stability and security of supply E^2DC 2014 Cambridge, UK 30/06/2014 15
What we are doing A monitoring and control middleware framework to be deployed through a set of fully modular plug-ins to current data centres which will include: An optimized intelligent pervasive sensing and monitoring infrastructure, which combine techniques from intelligent processing (data mining, learning) and pervasive sensing aimed at: off-line energy load profiling on-line sensing in real time the actual energy consumption of IT infrastructure (servers) and facility infrastructure, and forecast the energy consumption in the coming time snapshots A real time data centre multidimensional optimization control tool, which will produce energy sub-optimal actuation strategies for the cooling subsystems by trading off energy demand of IT infrastructure and cooling, energy flexibility management, renewable energy source supply, network traffic increase over the net against penalties for potential violation of SLAs. A smart grid real time green energy marketplace, which includes a smart districts energy management broker and a smart energy (both power and heat) grid simulator. E^2DC 2014 Cambridge, UK 30/06/2014 16
The GEYSER Framework High Level Architecture Network of DCs Multi-criteria Energy Efficiency Optimization Network of DCs Multi-criteria Energy Optimizer Customers QoS & SLA Negotiation Broker Adaptive IT Load Migration Broker Network of DCs Energy Budget Broker Data Security & Privacy Energy-Budget Situational Awareness Monitoring & On Demand Adaptive Control DC Real-time Multi-criteria Energy Efficiency Optimization Real-time Multi-criteria Energy Optimizer Smart Grid/Smart City Integration Real-time Energy Marketplace Broker Customers QoS & SLA Negotiation Broker Data Centre Adaptive IT Load Migration Broker Data Centre Energy Budget Broker Smart Grid/Smart City Interface Subsystems Closed Loop Control for IT Load Optimization Non-IT Energy Sinks Monitoring & Control Continuous Monitoring, Feedback & Adaptive Control Real-time Energy Monitoring/Sensing & Adaptive Control Subsystem IT Energy Sinks Monitoring & Control Continuous Sensing & Control Renewable Energy Sources Monitoring & Control Non-Renewable Energy Sources Monitoring & Control Closed Loop Control for Energy Optimization HAVC Buildings Non-IT Energy Sinks Storage Processing Networking IT Energy Sinks Geothermal PV Windmills Energy Renewable Energy Sources Non-Renewable Energy Sources E^2DC 2014 Cambridge, UK 30/06/2014 17 UPS Electricity Generator
E^2DC 2014 Cambridge, UK 30/06/2014 18 Enabling The Next Generation of Data Centres DC level local optimization monitor, control, reuse and optimize their overall ENERGY consumption & production, from renewable energy sources in particular, within the framework of a unified representation of energy Smart Grid & Smart City level global optimization actively involved as energy prosumers within a smart city green energy marketplace Virtual Data Centre positioned at the forefront of a distributed computing architecture in a cloud-oriented networked scenario
E^2DC 2014 Cambridge, UK 30/06/2014 19 Our Approach: 4 Scenarios Scenario "10"- Workload Federated Data Centres Scenario "11"- Workload and Energy Federated Data Centres Scenario "00"- Single Data Centre within a Smart City/District IT Workload Relocation Scenario "01"- Data Centres as coordinated energy elements within a Smart City/District Coordinated Energy Management
GEYSER - compliant DC for Smart City Urban Data Center Power Marketplace Bid Placement Smart City Energy Mix Demand Networked Data Center IT Workload adaptation Heat Power request Heat request Energy (Power, Heat) Trader DSO Network Technical constraints mgmt E^2DC 2014 Cambridge, UK 30/06/2014 20
E^2DC 2014 Cambridge, UK 30/06/2014 21 GEYSER compliant DC for DSO Urban Data Center Load Flexibility Aggregation Local Balancing Market Smart district Node Networked Data Center IT Workload adaptation Building prosumer DSO
E^2DC 2014 Cambridge, UK 30/06/2014 22 The GEYSER Data Model GEYSER does not reinvent the wheel but builds upon EU FP7 GAMES data model integrates with EU FP7 COOPERATE Neighbourhood Information Model DC Components Energy Consumption Energy Production DC Internal Context DC Energy Efficient Optimization Actions DC External Context DC Energy Efficient Indicators
E^2DC 2014 Cambridge, UK 30/06/2014 23 The GEYSER Data Model part 1 Heating, Ventilation, Air Conditioning Facility Heat Recovery Infrastructure Non-IT Components Lighting DC Energy Consumption Components Building Building Energy Management System Simple Components Servers Storage Network IT Components IT Equipment Power Distribution Unit Applications Combined Components Server Cluster
E^2DC 2014 Cambridge, UK 30/06/2014 24 The GEYSER Data Model part 2 PV Green Energy Sources Wind Geothermal DC Energy Production Components Brown Energy Sources Diesel Generator On-site Energy Storage Components UPS
E^2DC 2014 Cambridge, UK 30/06/2014 25 The GEYSER Data Model part 3 Dynamic adjustment of temperature, humidity, and lighting Non-IT Resources Heat Storage (pre-cooling) and release Energy Consumption UPS-based, compressed air storage systems, flywheel, ice storage tanks DC EE Optimization Actions IT Resources Maximise utilization of green energy produced Load Management Dynamic Power Management Energy aware load management Turn on/off DC components Energy Production Shift delay tolerant workload to match the renewable energy production peak
E^2DC 2014 Cambridge, UK 30/06/2014 26 The GEYSER Data Model part 4 IT Cooling Energy / Power consumption (loads) Energy produced locally (electricity or heat) Transformer Lighting Building DC Monitored Parameters Heating recovered Power demand shifting CO2 emissions Performance Economic performance Applications performance
Data Model integrated with EU EEB Model via FP7 COOPERATE model E^2DC 2014 Cambridge, UK 30/06/2014 27
E^2DC 2014 Cambridge, UK 30/06/2014 28 The System Architecture: a possible pilot site instance Weather Forecast Cloud Energy Market Open Source Optimization Engine GEYSER Platform Open Source GEYSER DB Wattics DB Insights Wattics Display Web Page Interface ABB Decathlon ODBC Interface Open Source Communication Layer Wattics API OPC Client OPC Server History DB OPC SNMP Modbus TCP BMS DC IT Equipment (servers, ) DC Infrastructure Equipment (power meters, CRAHs, )
E^2DC 2014 Cambridge, UK 30/06/2014 29 Chiller Chiller The GEYSER Simulation Environment All components related to data centres and smart cities are to be represented as models Hardware In the Loop: different components in the simulation can be replaced with real hardware: Real time testing of the complete data centre and interaction with the Grid Detailed testing of the data centre operation and its components with a limited model of the grid Power Hardware in the Loop of the efficiency of the cooling of one or more computational racks Grid Simulator Smart City Simulator Energy Market Simulated Onsite Energy Generator Electrical Interface Hydraulic Interface DC-Control System Racks Racks Data Centre Simulator Thermal Interface Hot Aisle Sensors
GEYSER Main Innovations New generation of DCs as key players in the smart energy city and/or smart grids/dso Considering DC as a stakeholder exchanging ENERGY, including electricity AND heating/cooling DC-level optimized management of energy, including DCs Renewable Energy Source powered or with green energy contracted at the meter Comprehensive set of Local Balancing Services offered to DSO/electricity operator (e.g. Ancillary services like voltage regulation, demand response/consumption flexibility), going beyond Demand Response DC participating to the Optimized ENERGY management of a Smart District/Smart City by Energy provisioning to the Local Energy Marketplace Cooperative business models (and enabling IT platform) for nurturing DC cooperation among all the DC ecosystem stakeholders (DCs, energy providers, DSOs, DC End users) by fair incentive sharing Optimized ENERGY green energy usage as rational behavioural strategy of DCs participating to Smart City real time green energy-led marketplaces E^2DC 2014 Cambridge, UK 30/06/2014 30
Thank you for your attention! Any questions? The journey has just begun...fancy for more? http://www.geyser-project.eu/, http://www.linkedin.com/groups/networked-data-centres-smart-grids-7485057, or just contact us Massimo.bertoncini@eng.it Geyser-project@eng.it E^2DC 2014 Cambridge, UK 30/06/2014 31