Deliverable 2 1 Report - City Characterisation

GRANT AGREEMENT NO. : PROJECT ACRONYM: PROJECT TITLE: FUNDING SCHEME: THEMATIC PRIORITY: PROJECT START DATE: DURATION:

608775 INDICATE Indicator-based Interactive Decision Support Information Exchange Platform for Smart Cities STREP EeB.ICT.2013.6.4 1st October 2013 36 Months

and

DELIVERABLE 2.1 Characterisation of the City

Date 11th April 2014

Submitted By Will Turner (TCD)

30th April

Will Turner (TCD)

PU PP RE CO

Review History Reviewed By Version Ruth Kerrigan (IES), Oliver Draft Kinnane (TCD) Valeria Ferrando (IES), Richard Final Quincey (IES)

Dissemination Level Public Restricted to other programme participants (including the Commission Services) Restricted to a group specified by the consortium (including the Commission Services) Confidential, only for members of the consortium (including the Commission Services)

X

This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no. 608775

Table of Contents EXECUTIVE SUMMARY ..................................................................................................................................................4 1. Introduction ..........................................................................................................................................................6 2. Sources .................................................................................................................................................................9 2.1. Power plant performance parameters .........................................................................................................9 2.1.1.

Ramp rate, ramp time and minimum run time ..................................................................................10 2.1.2.

Heat rate .............................................................................................................................................11 2.1.3.

Dispatchability ....................................................................................................................................11 2.1.4.

Load factor ..........................................................................................................................................12 2.2. Centralised energy producers.....................................................................................................................12 2.3. Distributed energy producers and micro-generation.................................................................................13 2.4. Renewables.................................................................................................................................................13 2.4.1.

Wind....................................................................................................................................................13 2.4.2.

Hydroelectric ......................................................................................................................................16 2.4.3.

Solar ....................................................................................................................................................17 Photovoltaic (PV) ............................................................................................................................................17 Grid-connected PV ..........................................................................................................................................17 Concentrated Solar Power (CSP) ....................................................................................................................18 2.4.4.

Bioenergy ............................................................................................................................................19 3. Sinks ....................................................................................................................................................................22 3.1. Buildings......................................................................................................................................................22 3.1.1.

Sensitivity analyses .............................................................................................................................23 3.1.2.

Building characteristics .......................................................................................................................24 Internal loads ..................................................................................................................................................25 Insulation ........................................................................................................................................................25 Glazing ............................................................................................................................................................25 Air leakage ......................................................................................................................................................25 HVAC Systems .................................................................................................................................................25 Occupant Behaviour .......................................................................................................................................25 Building size ....................................................................................................................................................26 Sustainable Systems .......................................................................................................................................26 3.1.3.

Top-down modelling approach...........................................................................................................26 3.1.4.

Building typological categories ...........................................................................................................29 Region .............................................................................................................................................................29 Construction date ...........................................................................................................................................30 Form................................................................................................................................................................32 Use ..................................................................................................................................................................34 Retrofit history................................................................................................................................................36 3.2. Electric vehicles (EVs) .................................................................................................................................37 3.2.1.

Hybrids ................................................................................................................................................38 3.2.2.

Automobile efficiencies ......................................................................................................................38 3.2.3.

EV power sources ...............................................................................................................................39 30/04/2014

Grant No. 608775

1

Li-Ion ...............................................................................................................................................................39 3.2.4.

EV charging and infrastructure ...........................................................................................................40 3.2.5.

EVs and the smart grid........................................................................................................................41 3.3. Electric public services ................................................................................................................................41 3.3.1.

Street lighting .....................................................................................................................................41 3.3.2.

Telecommunications ..........................................................................................................................42 4. City networks ......................................................................................................................................................45 4.1. Electricity networks ....................................................................................................................................45 4.1.1.

The smart grid .....................................................................................................................................45 New grid paradigms in smart energy subsystems ..........................................................................................46 Power generation ...........................................................................................................................................47 Smart transmission and energy management ...............................................................................................48 4.1.2.

Electricity distribution grids ................................................................................................................49 Transmission lines...........................................................................................................................................49 Substations .....................................................................................................................................................51 4.2. Natural gas ..................................................................................................................................................51 5. Energy storage ....................................................................................................................................................54 5.1 Energy storage technologies ......................................................................................................................54 6. Socioeconomics – dynamic pricing .....................................................................................................................56 6.1. Bulk usage meters ......................................................................................................................................56 6.2. Peak load pricing.........................................................................................................................................56 6.3. Quantifying the application of dynamic pricing .........................................................................................60 6.4. Review of dynamic pricing trials .................................................................................................................60 6.4.1.

Residential trials .................................................................................................................................60 Demand-Side Findings ....................................................................................................................................62 Supply-Side Findings .......................................................................................................................................62 6.4.2.

Non-residential trials ..........................................................................................................................62 6.5. Dynamic pricing and utility customer behaviour .......................................................................................63 7. Network theory and cities ..................................................................................................................................65 7.1. Complex network theory ............................................................................................................................65 7.1.1.

Graphs, nodes, edges, and degrees ....................................................................................................65 7.1.2.

Connectivity, adjacency and direction................................................................................................65 7.1.3.

Weighted graphs.................................................................................................................................66 7.1.4.

Characteristic path length ..................................................................................................................66 7.1.5.

Global efficiency .................................................................................................................................67 7.2. Types of network ........................................................................................................................................68 7.2.1.

Primal and dual networks ...................................................................................................................68 7.2.2.

Small-world networks .........................................................................................................................68 7.2.3.

Scale-free networks ............................................................................................................................68 7.3. Centrality ....................................................................................................................................................69 7.3.1.

Degree centrality ................................................................................................................................69 7.3.2.

Closeness centrality ............................................................................................................................69 7.3.3.

Betweenness centrality ......................................................................................................................70 30/04/2014

Grant No. 608775

2

7.3.4.

Efficiency centrality ............................................................................................................................70 7.3.5.

Straightness centrality ........................................................................................................................70 7.3.6.

Information centrality.........................................................................................................................71 7.4. City network characterisation ....................................................................................................................71 7.4.1.

Characterisation of small, urban areas ...............................................................................................71 7.4.2.

Characterisation of entire cities .........................................................................................................73 8. Summary .............................................................................................................................................................78 References ..................................................................................................................................................................79

30/04/2014

Grant No. 608775

3

EXECUTIVE SUMMARY The aim of INDICATE Task 2.1 is to identify and characterise the components of the city that are relevant to its energy performance. This characterisation is necessary to aid the development of the complex system model in Task 2.2 and the 3D dynamic simulation model in Task 4.3. To this extent the city can be considered as a greater system of component subsystems, with nodes representing the system components and links representing the physical or abstract relationships between them. Identified as being relevant to the energy performance of the city are:     

Energy sources: Energy sinks: Energy networks: Energy storage: Socioeconomics:

Sites where energy is generated (supply-side) Sites where energy is consumed (demand-side) The infrastructure that distributes energy between the sources and the sinks Sites where generated energy may be retained for later use The impact of city inhabitants on energy supply and demand.

The characterised energy sources have been split into centralised, distributed and renewables. Centralised energy sources are large power stations such as coal-, oil- and gas-fired power plants, and nuclear power plants. The distributed energy sources can be smaller, more localised versions of centralised power plants such as diesel generators and combined heat and power (CHP) plants. Renewable energy sources include hydroelectric dams, wind turbines, solar power and biomass power plants. The nature of some renewable technologies such as wind and solar means that they can fall under either category of centralised (e.g., wind/solar farms) or distributed energy sources (e.g., individual wind turbines/solar panels).

Buildings are the dominant energy sinks in the city, with residential buildings accounting for the largest proportion of energy consumption. The energy performance of buildings was characterised by first reviewing existing building simulation sensitivity analysis studies from the literature to identify which building characteristics have the greatest impact on energy performance. These building characteristics were then classified into ‘top-down’ property subsets which could be described by typical categories used in studies on building typology. This approach allows the energy performance of buildings to be described by just a few inputs, simplifying city-scale energy simulations by reducing processing and labour overheads. Other energy sinks considered are electric vehicles, street lighting and telecommunication networks.

Energy distribution networks important to the smart city are the electricity and gas networks. These networks, along with the concept of the smart grid, were discussed and characterised.

Energy storage systems are necessary to help level the disparity between energy supply and energy demand. This is of particular importance for renewably-generated energy which is intermittent and variable in nature due to climate and weather. Both large-scale bulk grid storage and smaller-scale technologies such as batteries have been characterised for inclusion in the complex system model.

City inhabitants directly impact the flow of energy around a city. The magnitude of demand load is determined by user activities depending on time of day and weather. Energy supply is then adjusted to meet the demand load.

User activities, and hence demand, can be influenced by dynamic pricing offered by utility companies with the 30/04/2014

Grant No. 608775

4

aim of smoothing out the daily demand profile and reducing peak demand. Different dynamic pricing tariffs and their energy impact have been characterised for the INDICATE tool.

Finally, a discussion on complex network theory has been included and its application to cities, as network theory will form the basis of the network optimisation routines for INDICATE.

30/04/2014

Grant No. 608775

5

1.

Introduction

Task 2.1 focuses on the characterisation of the smart city and its energy networks for the purpose of the INDICATE model development. Specifically it focuses on describing the city as an integrated system of component subsystems of buildings (sinks), energy supply systems (sources) and networks (connecting links). With respect to the built environment; building typology, age, operational performance etc. are characterised as inputs.

Parameters related to centralised, distributed, renewable energy sources and network characteristics represent inputs from the supply side. This information will be used to create the 3D Dynamic Simulation Model (DSM) of the City in Task 4.3.

There are three main objectives for Task 2.1:   

Characterise the city as an integrated system of component subsystems Form the basis of the complex system model in Task 2.2 Aid the development of the 3D DSM of the city in Task 4.3.

The key aim of INDICATE Task 2.1 is to identify and characterise the components of the city that will enable its energy performance to be modelled. To this extent, one interpretation of the form of a city is to consider it as a system of subsystems. Many individual subsystems such as the energy supply, storage and distribution systems combine to form a much larger city-scale system (Figure 1).

Figure 1: A city can be represented as a system of component subsystems

In terms of the energy performance of a smart city, the component systems of interest are the supply and demand subsystems, the energy networks, energy storage sites, electrified transport networks and the socioeconomics that account for the energy influence of city inhabitants. From an energy-modelling perspective, it is convenient to characterise the smart city as a collection of nodes and links. The nodes represent the city 30/04/2014

Grant No. 608775

6

components where energy is generated, energy is stored for later use, or where energy is consumed. The links are the networks that allow energy to flow throughout the city between the sites of generation, storage and consumption. A brief description of the city subsystems follows.

SUPPLY SIDE: Sites where energy is generated for use by the city. This generation can occur at conventional sites (e.g., centralised fossil fuel or nuclear power stations); distributed micro-generation sites (e.g., localised photovoltaic (PV) and wind turbines); or more contemporary renewable sites (e.g., wind farms, solar farms, hydroelectric dams).

DEMAND SIDE: The energy sinks are the components of the city which use the energy provided by the supply side. The sinks can be buildings, electrified transport systems, and public services such as street lighting and telecoms.

ENERGY NETWORKS: The connecting links between the sources and the sinks are the energy distribution networks, or grids. Transmission lines are used to physically link the power stations with the buildings.

Traditionally, the flow of energy has been in one direction, from the source to the sink via the grid. Of increasing interest within the smart city is the possibility of micro-generation sites which allow buildings to act as sources, and the energy they produce to feed back into the grid and then be used by other sinks. Electricity and gas grids should be modelled in the smart city.

ENERGY STORAGE: Energy storage refers to facilities where energy can be retained for use at a later date. In terms of the smart city, this can include batteries of electric vehicles and photovoltaic (PV) systems, and also large-scale grid energy storage such as pumped storage hydroelectricity, compressed air and flywheels.

SOCIOECONOMICS: Energy demand is driven by the needs of the inhabitants of a city. Understanding the behaviour of energy users such as building occupants is important if energy use is to be modelled correctly.

Certain policies such as Time of Use (TOU) energy tariffs are designed to impact user behaviour to address the balance energy supply and demand. Such factors should be accounted for in a model of the smart city.

Table 1 summarises the city subsystems and their components.

Table 1: City component subsystems

City subsystem

Supply

Demand

Energy networks 30/04/2014

Component  Centralised power stations (e.g., coal-fired, gas-fired and nuclear power stations)  Distributed power (e.g., localised photovoltaic, combined heat and power (CHP) plants)  Renewables (e.g., hydroelectric dams, wind farms)   

Buildings (residential, commercial, industrial) Transport networks (electric rail, electric vehicles) Public services (street lights, telecoms)

 

Electricity (transmission, substations) Gas (transmission lines, reservoirs) Grant No. 608775

7

Energy storage

 

Bulk storage (e.g., pumped storage hydroelectricity, compressed air, fly wheels) Micro-storage (e.g., PV batteries, electric vehicle batteries)

Socioeconomics

 

Dynamic pricing (Time of Use tariffs, electricity peak demand) User behaviour (occupants, energy consumers)

30/04/2014

Grant No. 608775

8

2.

Sources

EU electricity generation comes from a variety of sources. Figure 2 shows the breakdown of electricity generation in the EU by source (tonnes of oil equivalent) (European Commission 2014). The predominant source is the burning of fossil fuels (51%) such as coal (solid fuels), natural gas and crude oil. Nuclear energy makes up 28% of the total, while renewable energy accounts for 20%. The breakdown of renewable electricity production is also shown in Figure 2. Nearly 70% of EU renewable electricity is generated from biomass and waste, with hydropower (19%) and wind (8%) making up the majority of the remainder.

100%

1%

90%

20% 28%

Nuclear energy Solid fuels Natural gas

Solar energy, 2%

80% 70% 60%

Geothermal energy, 3% Wind, 8%

50% Crude oil

12%

Renewable energy Other

20%

40%

30% 20%

Hydroelectric, 19% Bioenergy & waste, 68%

10%

19%

0%

Figure 2: EU electricity generation by source (left) and breakdown of renewable generation (right) for 2010 (European Commission 2014)

For the INDICATE tool the dominant electricity production technologies will be considered. Fossil-fuel burning power plants (coal, oil and natural gas) generally come under the umbrella of centralised energy producers, as does nuclear power. Distributed energy producers are usually small-scale versions of the centralised energy producers, or small-scale renewable technologies. Renewable energy will be considered separately and will include biomass, hydroelectric, wind and solar power. Geothermal power will not be included for now because it makes up a very small portion of EU electricity production and is restricted to very specific geographical locations.

While solar power represents an even smaller proportion of EU electricity production compared to geothermal, its characteristics make it suitable for inclusion in the model of the smart city.

2.1.

Power plant performance parameters

Electrical demand or load is time dependent and typically follows a 24-h cycle (Figure 3). Electrical load is at a minimum in the middle of the night while the majority of people are asleep. Load then begins to rise in the early morning as people wake, reaches a peak in the early afternoon and then falls again throughout the evening. The shape of the load curve can be affected by climate - for example, hot weather can provoke high electrical demand from the use of air conditioning.

30/04/2014

Grant No. 608775

9

The minimum requirement for electricity is called the baseload demand. The baseload is met by baseload generators which have low operating flexibility and low variable operating costs (costs that change with output, such as the cost of fuel). They can increase their electrical output (ramping) to meet some of the demand above the baseload, and also reduce their output at times of lower demand. The intermediate demand occurs during the day. This load is met by intermediate generators (also called load-following or cycling generators). These generators are more flexible than baseload generators and have a quicker response time to fluctuating demand.

Some of the time the intermediate generators will be running but not putting electricity onto the grid. This is known as spinning reserve and is used to meet unexpected load or act as a backup to other generators that might fail.

The peak demand is the maximum load on the grid and is met with peaking generators. These power plants are usually the most expensive to run and only operate for several hundred hours per year. They have the quickest ramp rate and can put electricity onto the grid at very short notice with the ability to go from zero output to maximum output in just a few minutes (Kaplan 2008).

Figure 3: Typical 24-h electricity load pattern (Kaplan 2008)

Typically, power plant performance can be described using several parameters:    

Ramp rate, ramp time and minimum run time Heat rate Dispatchability Load factor.

2.1.1. Ramp rate, ramp time and minimum run time The ramp rate describes how quickly a power plant can increase its power output. Ramp rate is usually defined as the increase in power output as a percentage, per some unit of time. The ramp time is the time that it takes for a 30/04/2014

Grant No. 608775

10

power plant to reach minimum stable electricity output from a cold start. The minimum run time is the shortest amount of time that a plant can operate after it has been switched on (Blumsack 2014). Table 2 shows typical ramp rates, ramp times and minimum run times for some power generating technologies.

Table 2: Typical ramp rates, ramp times and minimum run times for some power plants (Blumsack 2014; Vuorinen 2009)

Technology Simple-cycle combustion turbine Combined-cycle combustion turbine Nuclear Wind turbine Hydroelectric

Ramp rate [%/min]

Ramp time

Min. run time

20 to 40

Minutes to hours

Minutes

5 to 10

Hours

Hours to days

1 to 5 Variable Variable

Days Minutes Minutes

Weeks to months None None

2.1.2. Heat rate The heat rate describes the efficiency of a power plant to convert fuel into electrical energy. The heat rate, is defined as the fuel input divided by the energy output. Mathematically the heat rate is simply the inverse of the efficiency, so a lower heat rate means a more efficient generator. Heat rates are only applicable to generators that burn a combustible fuel such as coal, gas or biofuel. Table 3 shows average heat rates and efficiencies for electricity generation technologies in Ireland in 2007 (Connolly et al. 2010).

Table 3: National power generation parameters for Ireland (2007) including heat rate, efficiency and load factor (Connolly et al. 2010; Chiodi et al. 2011)

Plant type Natural gas Coal Peat Oil Wind Natural gas CHP Hydroelectric Pumped hydroelectric Biomass

Capacity [MW] 3525 852 345 1014 724

Fuel used [TWh] 28.72 12.95 5.05 4.18 -

Electricity generated [TWh] 13.38 5.35 2.11 1.94 1.88

Heat rate [-] 2.1 2.4 2.4 2.2 -

Efficiency [%] 47 41 42 46 -

Load factor [%] 55 87 32

273

6.08

1.83

3.3

30

57 or 85

216

-

0.66

-

-

26

292

-

0.35

-

-

14

Cocombusted with fossil fuel

0.41

0.14

2.9

34

57

2.1.3. Dispatchability A plant with a lower heat rate will be dispatched (i.e., put online) before a plant with a higher heat rate because it is economically more favourable and has a lower variable cost. Nuclear, coal and geothermal plants are expensive to build but have low fuel cost, hence low variable cost. These plants are typically dispatched first to meet 30/04/2014

Grant No. 608775

11

baseload and will run for the majority of the year. Combined-cycle gas plants burn expensive natural gas so will be dispatched after the baseload plants in order to meet intermediate load. Simple-cycle combustion turbines also burn expensive gas but are less efficient than the combined-cycle plants, so are dispatched last to meet peak loads (Kaplan 2008). For a power plant to be dispatchable it must be controllable by grid operators.

The dispatchability of renewable power is more complicated due to its variable and intermittent power output due to weather. While wind and solar plants produce energy they can be used to replace baseload, intermediate and peaking generators. Energy storage technology will make variable renewable power more reliable in the future, but is currently expensive. Hydroelectric power is inexpensive, reliable and dispatchable provided that there is sufficient water in the reservoir that is not being competed for by drinking water supply, irrigation and recreation. Hydroelectricity is typically used to displace peaking plants because it has a very high ramp rate and short ramp time (Kaplan 2008).

2.1.4. Load factor Most power plants, independent of the means by which they generate power, have a maximum possible power output known as the ‘rated power’. A 1 MW power plant operating continuously for a year (24 hours x 365 days = 8760 hours) would produce 8760 MWh of energy. The load factor of a power plant is the actual annual output divided by the maximum annual output, expressed as a fraction or percentage. For example, if a 12 MW hydroelectric plant produces 30 GWh of electricity in a year, the load factor will be 30,000 / (12 x 8760) = 0.285 or 28.5% (Boyle 2004). Typical load factors are approximately 70% for baseload plants, 50% for intermediate plants and 25% for peaking plants (Kaplan 2008). See Table 3 for some load factors used for modelling Irish electricity generation (Chiodi et al. 2011).

2.2.

Centralised energy producers

The majority of all centralised power plants convert a fuel into heat energy via combustion (or fission in the case of nuclear power plants). The heat energy is converted to rotational mechanical energy by a heat engine or turbine. A generator is then used to convert the mechanical energy into electrical energy by rotating a conductor inside a magnetic field. Most fossil-fuel power plants use either a steam turbine or a combustion turbine to convert thermal energy into mechanical energy. Steam turbines are used to produce approximately 90% of the world’s electricity (Wiser 2000). Biomass can be used in co-generation plants where some amount of biomass is used as a substitute for fossil fuel, or it may be burned in dedicated biomass power plants. Combined-cycle power plants (or combined heat and power (CHP)) have both a combustion turbine fired by natural gas, and a steam turbine that uses the wasted heat from the combustion turbine to produce additional electricity and increase the efficiency of the power plant.

Suggested centralised power plants that should be included in the INDICATE model:  Coal-fired  Oil-fired  Natural gas-fired (simple-cycle and combined-cycle)  Nuclear Characterisation of centralised energy producers uses the parameters described in Section 3.1: Power plant performance parameters, above.

30/04/2014

Grant No. 608775

12

2.3.

Distributed energy producers and micro-generation

Distributed power describes technologies which allow the production of electricity on a small, local scale close to where it is used (micro-generation). Due to economies of scale, distributed generation is currently more expensive per kWh than centralised generation. However, there are benefits such as energy security, reduced transmission infrastructure, the ability to utilise a range of energy sources, and reduced environmental impact.

A range of distributed energy producers are available such as backup diesel generators, wind turbines, PV and CHP. Small-scale (~MW) natural gas generators can produce electricity with low hazardous emissions, meaning they can be situated close to populations and therefore suitable for CHP. PV and wind will become more widespread once battery technology matures and the cost of storing electricity decreases.

Suggested distributed power plants that should be included in the INDICATE model:  CHP plants  Diesel generators  Wind turbines  Solar PV arrays  Bioenergy Characterisation of fossil-fuel-burning distributed energy producers will use the parameters described in Section 2.1, above. Renewable distributed energy producers will use the parameters from Section 2.4, below.

2.4.

Renewables

Renewable energy now encompasses a wide range of electricity generation technologies. The most prevalent technologies are the ones that should be accounted for in the INDICATE tool: wind, hydroelectric, solar and bioenergy (see Figure 2). Some renewable energy producers can be used as individual, localised energy sources, such as solitary wind turbines at manufacturing plants or roof-mounted residential PV panels. With some renewable technologies it also possible to increase the scale of energy generation by combining many individual units into large collective arrays such as wind farms and solar farms.

2.4.1. Wind Kinetic energy from the wind is turned into electrical energy using a wind turbine and generator. Wind turbine output power is a function of wind speed at the height of the turbine hub. Wind speed data is recorded typically by weather stations at a height of 10 m. To obtain the wind speed, , at the height of the wind turbine hub, we can use the power law relationship from Awbi (2003): (

)

(

)

(1)

Where is the wind speed at the height of the weather station, . The coefficients and ; and the exponents and define the terrain roughness and atmospheric boundary conditions around the wind turbine and weather station respectively. A similar approach to calculating local wind speeds can be found in the ASHRAE Handbook of

30/04/2014

Grant No. 608775

13

Fundamentals (2004). A more accurate, but more complex wind speed calculation method that uses a logarithmic function to more closely model the real physics can be found in Deaves and Harris (1978).

The total power available to a wind turbine from the wind is given by: (2)

Where is the density of air [kg/m3], is the swept area of the turbine blades [m2], is the wind speed at the turbine hub [m/s], and is the power coefficient [-]. Wind turbine efficiencies are limited to 59.3% by Betz’s law (Albring 1967). Typical wind turbine efficiencies that account for system losses in the bearings, gear box, and generator are of the order of 10 to 30% of the total available power from the wind. As wind turbines are built to achieve a maximum efficiency at a certain wind speed (due to cost and engineering considerations), the efficiency will also be a function of wind speed.

The actual wind turbine power output is specific to the individual generator and usually represented by a wind turbine power curve (Figure 4). The theoretical power is that available to a wind turbine with an efficiency of 100%. As this is impossible, power generation curves for 20% and 40% efficient wind turbines are shown. The ‘cut-in wind speed’ is the minimum wind speed at which the turbine will start generating useful power. The ‘rated wind speed’ is the speed at which the turbine generates its maximum power output. Above the ‘cut-out wind speed’ the turbine will cease to generate power. This is a safety control imposed by the engineers to prevent damage to the wind turbine (Manwell et al. 2009).

30/04/2014

Grant No. 608775

14

Figure 4: Wind turbine power curve showing the theoretical power available from the wind, and the actual power generated by turbines with 20% and 40% efficiencies (source: engineeringtoolbox.com)

The energy generated by a wind turbine is dependent on the wind speed which is highly intermittent and variable in nature. The wind speed frequency distribution at a site gives information on the number of hours each wind speed is met (Figure 5). Wind energy curves for turbines with 20% and 40% efficiencies are shown, along with the theoretical maximum. The total energy generated for a year is the area under the wind energy curve.

30/04/2014

Grant No. 608775

15

Figure 5: The wind speed frequency distribution, and the energy generated by wind turbines with 20% and 40% efficiencies (source: engineeringtoolbox.com)

At the time of writing, micro wind turbines are available for small-scale commercial or domestic use rated between 1 kW and 5 kW. Industrial wind turbines currently range between approximately 25 kW and 8000 kW (8 MW) (Manwell et al. 2009).

In order to model wind power generation in the INDICATE tool it will be necessary to know wind speeds at the turbine sites, and the following properties of the wind turbine:       

Height of the turbine hub Blade swept area Cut-in and cut-out wind speeds Rated wind speed Rated power Turbine efficiency Wind turbine power curve shape.

Individual wind turbines can be used as distributed energy sources, while collections of wind turbines can form centralised energy generating wind farms.

2.4.2. Hydroelectric Hydroelectric power accounts for 6.7% of the world’s annual energy output, and 77% of all renewably-produced energy (BP 2013). Energy is generated by a hydroelectric plant by converting the gravitational potential energy of water into electricity via a turbine. The power output, P, is:

30/04/2014

Grant No. 608775

16

̇

(3)

Where ̇ is the mass flow rate of water [kg/s], is acceleration due to gravity [9.81 m/s], is the height of the water above the turbine [m], and is the efficiency of the power plant. Extracting energy from moving water with a turbine and generator is very efficient as the kinetic energy of the water is directly converted into electricity with no intermediate chemical or thermal process. Typical hydroelectric power plant efficiencies range from around 80% for small-scale plants, up to 90% for large-scale plants (Boyle 2004).

A dam is used to build up a reservoir of water behind the hydroelectric turbine. The dam can be used to store water (and its potential energy) at times of low demand, and then release it at times of high demand. The potential energy of the water can be stored for years and then released, with the turbine reaching maximum power output within minutes. This characteristic makes hydroelectricity ideally suited to smart grids as it can be used to cover baseline loads in the event of baseline power plant failures, meet unexpected high demand, or compensate for times of low renewable energy production. However, there exists year-to-year variation in the total potential energy available due to variations in rainfall (Boyle 2004).

2.4.3. Solar Solar power generally falls under two categories – photovoltaics (PV) where the photovoltaic effect is exploited to generate DC (direct current) electricity from light, and concentrated solar power (CSP) or solar thermal where the heat energy of light is harnessed to heat water. PV systems can be connected to the grid so that excess electricity generated may be sold to utility companies and not wasted.

Photovoltaic (PV) For the CitySim model, Robinson et al. (Robinson et al. 2009) use a simplified solar PV model from Duffie and Beckman (2013) to calculate the power generated by solar PV panels. The power output of the array, , is a function of the collector area, [m2], the total incident shortwave irradiance, [W/m2], the efficiency of the solar PV panel operating at maximum power point, , and the efficiency of the power conditioning equipment, : (4)

The efficiency of the panel at maximum power point accounts for the temperature-dependence of the cell’s conversion efficiency, relative to standard test conditions (STC). See Duffie and Beckman (2013) for a complete description of the model.

Grid-connected PV Solar PV systems can be connected to the grid using an inverter. The inverter converts the DC power generated by the PV panels into AC (alternating current) power suitable for the grid. Alternatively, arrays of solar panels can be used to form a larger-scale centralised solar farm. Metrics used to describe the quality of the PV grid connection are: (Eltawil & Zhao 2010) 30/04/2014

Grant No. 608775

17

Array yield,

(5)

Reference yield,

(6)

Final yield,

(7)

Capture losses,

(8)

System losses,

(9)

All with units of [kWh/kWpd]. [kWh] is the energy output of the PV array, [kW] is the peak power output, 2 2 [kWh/m d] is the mean daily irradiance in the array plane, [kWh/m d] is the reference irradiance at standard test conditions, and [kWh] is the energy going from the array to the grid.

Concentrated Solar Power (CSP) The power output of CSP collectors can be modelled using the approach from the ScenoCalc tool (SP Technical Research Institute of Sweden 2014). The tool is used to calculate the power output of flat plate, evacuated tube, concentrating, and unglazed solar thermal collectors for use with certification standards. ScenoCalc does not simulate solar thermal collection systems; rather it allows the performance of different systems to be compared by converting collector model parameters into energy performance values.

Main inputs for ScecnoCalc are the mean fluid temperature of the collector, the aperture area, solar radiation incidence angle modifier coefficients, the tilt and azimuth angles of the collector, and the geographical location of the system. The output of the tool is the collector yield [kWh] for a range of mean collector liquid temperatures.

Kämpf (2009) adopted the ScenoCalc approach to simulate solar thermal collectors using an empirical quadratic relationship: (

[

]

Where is the optical efficiency of the collector glazing [-], coefficients, is the mean collector liquid temperature [K], and

[

] )

(10)

[W/m2K] and [W/(m2K)2] are heat loss is the mean ambient temperature [K].

To account for solar power in the INDICATE model it is necessary to know the position information of the Sun relative to the ground/solar collector, weather conditions such as cloud coverage, plus the following properties of the solar power system:   

Solar collector surface area [PV and CSP] Incident shortwave irradiance on the collector surface [PV and CSP] Total system efficiency [PV]

30/04/2014

Grant No. 608775

18

   

Optical efficiency [CSP] Heat loss coefficients [CSP] Collector and ambient temperatures [CSP] Array, reference and final yields [grid-connected solar].

Both solar PV and CSP units can also be combined into large-scale centralised arrays forming solar farms.

2.4.4. Bioenergy Bioenergy is the broad term used to describe the energy derived from biomass – living matter or recently living organisms. Biofuels are derivatives of biomass. The carbon released from burning biofuels is approximately equal to the net carbon that the biomass absorbs from the atmosphere during its lifetime, hence biofuels are considered carbon neutral. However, the whole lifecycle of the biofuel including processing and transport will generally have a net carbon output, albeit a fraction of that from fossil fuels (Boyle 2004). There are five main categories of biomass materials used to produce biofuels: (Jansen 2013; Biomass Energy Centre 2014)     

Virgin wood from forestry activities or waste from wood processing e.g., palm kernels shells, wood pellets, woodchips, sawdust Energy crops: high-yield crops grown specifically for energy applications e.g., hybrid eucalyptus, Jatropha, Pongamia, and perennial grass Agricultural residues from harvesting or processing e.g., bagasse from sugarcane, corn husks, coconut shells and straw Food waste from food and drink manufacture, preparation and processing, and post-consumer waste e.g., used cooking oil, tallow and greases Industrial waste and co-products from manufacturing and industrial processes e.g., paper pulp, sewage sludge.

Biomass may be burned and converted into heat or electricity, or may be processed into a biofuel. The main conversion routes between biomass and energy or biofuel are shown in Figure 6.

30/04/2014

Grant No. 608775

19

Figure 6: Main conversion routes for bioenergy (UNDP 2000)

Bioenergy can be derived from many different sources, depending on the type of biomass used and the manner in which it is converted into energy or fuel. In the context of city-scale energy modelling, the processes by which the bioenergy is produced are unimportant. Of greater interest is the way in which the electrical or heat energy from bioenergy is made available to the city. Table 4 shows power plant output ranges and net efficiencies for the main thermochemical biomass conversion technologies.

Table 4: Main thermochemical biomass energy conversion routes to heat and electricity (UNDP 2000)

Conversion system Combustion Combined heat and power (CHP) Standalone (biomass) Co-combustion (fossil and biomass) Gasification CHP Diesel engine Gas turbine Biomass-Integrated Gasification/Combined Cycle Digestion Wet biomass materials 30/04/2014

Power range

Net efficiency (%, Lower Heating Value)

100 kWe to 1 MWe 1 to 10 MWe 5 to 20 MWe 20 to 100 MWe

60 to 90 (overall) 80 to 99 (overall) 30 to 40 (electrical) 20 to 40 (electrical)

100 kWe to 1 MWe 1 to 10 MWe 30 to 100 MWe

15 to 25 (electrical) 25 to 30 (electrical) 40 to 55 (electrical)

Up to several MWe

10 to 15 (electrical)

Grant No. 608775

20

There is a large range of biomass power plant technologies, but fundamentally they produce energy by burning biomass or biofuel, then using a steam turbine to convert the heat produced from combustion into electricity.

Table 5 shows the energy content (mass and volumetric) of typical biomass energy sources and some fossil fuels for comparison. Because of the similarities between fossil fuel power plants and bioenergy power plants, the same characteristics may be used for both.

Table 5: Average heat energy content of biomass and fossil fuels for the UK (Boyle 2004)

Fuel Wood (green, 60% moisture) Wood (air-dried, 20% moisture) Wood (oven-dried, 0% moisture) Charcoal Paper (stacked newspapers) Dung (dried) Grass (fresh-cut) Straw (harvested, baled) Sugar cane residues Domestic refuse (as collected) Commercial wastes (UK average) Oil (petroleum) Coal (UK average) Natural gas (at supply pressure)

Energy content [GJ/tonne] 6 15 18 30 17 16 4 15 17 9 16 42 28 55

Energy content [GJ/m3] 7 9 9 Material specific 9 4 3 1.5 10 1.5 Material specific 34 50 0.04

For the purpose of the IDICATE tool, biomass power plants can be treated as both large-scale, centralised energy producers and smaller-scale distributed energy producers. Biomass can also be co-combusted with some fossil fuels to reduce carbon emissions. This can be accounted for with a fossil/biomass input fuel ratio.

30/04/2014

Grant No. 608775

21

3.

Sinks

Within the context of the city, sinks are the sites where energy is consumed. For energy modelling purposes, the sinks that the INDICATE tool will account for are buildings, electric public transport networks and electric public services.

3.1.

Buildings

Fundamental to energy use of buildings are four property subsets – construction, function, geometry, and systems (Figure 7):

Figure 7: Property subsets of a building that can influence its energy performance









Construction refers to the materials that make up the fabric of the building, as well as the properties and quality of the envelope and the façade. A heavyweight masonry building will have a different energy performance compared with the same building made from timber (mainly due to thermal mass effects). A building façade consisting of a high proportion of glass will be much more susceptible to solar gains and conduction losses compared with a building that has a more opaque façade. Likewise, a building with a leaky envelope will perform differently from a building with a more airtight envelope.

Function refers to the designated purpose of the building, or how the building is used. Whether the building is a public swimming pool or retail outlet will drastically affect the amount of energy it consumes independently of its geometry and construction.

Geometry is the physical description of the building and the space it occupies. The energy consumption of a building will be affected by its size; design; orientation; and location relative to other buildings, geographical topology and features.

Systems provide services to the building occupants. Examples include the heating, ventilation and airconditioning (HVAC) equipment inside a building that maintain an indoor environment suitable for the intended function of the building. Other systems include the lighting and hot water systems.

Conventional building simulation tools such as Virtual Environment (IES 2013), Energy+ (US DOE 2013b) and DEST (DeST Group 2011) use a ‘bottom-up’ approach to create a virtual representation of a single building. Accurate 3D geometry of the building’s overall form, rooms and zones is created within the building simulation software, or by 30/04/2014

Grant No. 608775

22

using computer aided design (CAD) packages. The geometry is then translated into the simulation environment where the walls, ceilings, roofs and floor are assigned construction properties such as the type of building fabric material. Typically the HVAC systems inside the building and the internal loads are then specified before a simulation is performed using external weather data to describe climatic conditions. This approach can require hundreds of individual inputs to create an accurate model of a single building. Clearly the scale change involved with modelling cities and collections of buildings entails that preserving a similar level of complexity may lead to unnecessarily high labour and computing overheads. It is therefore important to model the building characteristics that will allow the right balance between simulation runtime and simulation accuracy. A ‘topdown’ approach is more suitable, where groups of buildings are assigned typical properties based on typological classifications such as construction date, region, retrofit history and so on. Average values from building typology studies and energy performance data on existing buildings can be used to define the building characteristics.

This report makes recommendations as to what building characteristics should be included in the INDICATE tool so that the energy performance of buildings can be simulated while minimising model complexity.

3.1.1. Sensitivity analyses In order for a building energy model to produce useful results, certain physical characteristics of the building need to be represented in the model. Some of these characteristics will have a stronger influence than others on the model results for the energy performance of the building (Cordon 1992). Therefore, it is necessary to identify which characteristics yield the greatest energy impact via sensitivity analysis, and then ensure that those characteristics are suitably represented in the model (Buchberg 1969; Buchberg 1971; Athienitis 1989; Lomas & Eppel 1992; Cammarata et al. 1993).

Lam and Hui (1996) performed a sensitivity analysis on inputs for a modelled office building in Hong Kong. They found that the annual building energy consumption was most sensitive to internal loads, window systems, building envelope, indoor temperature set points and HVAC system efficiencies. Lam and Hui also found that both building peak demand loads and annual building energy consumption were sensitive to the same characteristics.

Macdonald (2002) states that the three minimum property groups necessary to perform a thermal building simulation are: the thermo-physical properties of the building materials (specific heat capacity, thermal conductance etc.), causal gains (internal loads from equipment and occupants), and air infiltration. Macdonald goes on to perform an extensive sensitivity analysis on the different variables of a building simulation. He concludes that the variables which have the greatest impact on the final energy solution are the internal loads within the building, the air infiltration rate, the conductivity of the insulation, and the weather file used to specify the outdoor environment.

Heller et al. (2011) conducted a sensitivity analysis on simulation variables for commercial buildings. In agreement with Macdonald, they concluded that the factors which had the largest impact on simulated energy performance were the internal loads (mainly due to lighting), the building envelope (glazing, air infiltration, and thermal conductance), and the HVAC system (type, size, efficiency, control). Heller et al. outlined building component packages of best and worst practice for the aforementioned factors, and found that the final energy use of a building could be made to vary by up to 260%, depending on climate. Heller et al. also considered the energy

30/04/2014

Grant No. 608775

23

impact of operator and occupant behaviour. They found that the combined effect of the two could alter the building energy use by up to 170%, again, climate dependent.

Struck (2009) and Struck et al. (2011) carried out a sensitivity analysis on ten architectural parameters for commercial buildings (mechanical parameters such as HVAC systems were not considered). They then ranked the parameters in order of impact. The top 5 parameters were glazing g-value (or Solar Heat Gain Coefficient), officeto-gross floor area ratio, glazing u-value, air infiltration, and façade u-value. In a later study, Struck (2012) concluded that the top five parameters that affect building energy use are internal loads, glazing g-value, glass-towall ratio, wall u-value and the ventilation air volume.

A more general study on the energy performance of European buildings by the BPIE (BPIE 2011) states that factors that influence residential energy use are the characteristics of the heating system, building envelope, climate, occupant or user behaviour, and social conditions (e.g., not using heating or cooling systems due to fuel poverty, or austerity measures).

Simulations carried out as part of the EU-funded IMPRO-Buildings (Environmental Improvement Potential of Residential Buildings) study (Nemry et al. 2008) showed that across Europe, residential energy use is dominated by heat loss through the building envelope and heat loss due to ventilation and infiltration. Solar gains also impact energy use in buildings with a high window-to-wall ratio such as high-rise apartment blocks.

Considering the findings from the literature above, in order to capture or simulate the energy performance of a building the following characteristics need to be considered:        

Internal loads Glazing Insulation Envelope leakage HVAC systems (Heating, cooling and ventilation) Occupant behaviour [Building size] [Sustainable systems].

When all other parameters are kept constant (e.g., thermal comfort criteria and the ability for the HVAC equipment to meet heating/cooling loads), the impact that the above characteristics have on the building energy performance will scale with the size of the building. For that reason, building size has been added to the list of characteristics. Also added to the list is ‘sustainable systems’, which represents passive and sustainable energy systems, such as solar thermal water heating and PV systems. Sustainable systems are important to consider in the framework of smart cities and low-carbon building technology, and so are relevant to the INDICATE project.

3.1.2. Building characteristics A brief description of the building characteristics identified as being important for building energy modelling follows.

30/04/2014

Grant No. 608775

24

Internal loads The internal loads of a building are the combined effect of the sensible (thermal) and latent (moisture) gains from indoor sources. Occupants, lighting and appliances contribute to the internal loads of a building. It is important here to note one of the differences between residential and non-residential buildings. Heat gains in non-residential buildings are dominated, typically, by internal loads, whereas heat gains in residential buildings are dominated by external climatic conditions such as solar gains and ambient temperature (US DOE 2001). Therefore, internal loads are more significant when modelling non-residential buildings.

Insulation Insulation describes the conduction heat loss rate through the building envelope due to its construction material. In Europe the effectiveness of an insulating material is described by its ‘U-value’, measured in W/m2K. Walls, ceiling, and floors all have an associated U-value depending on their fabric material. The North American equivalent of the U-value is the R-value.

Glazing Glazing or windows have three properties of interest for energy modelling: the level of thermal insulation offered by the glazing, denoted by its U-value, the ability of the window to diminish solar gains, denoted by its G-value (G-value is expressed as a fraction of solar energy transmitted by the glass), and the combined area of the windows. This is usually expressed as a fraction of the wall area, or the window-to-wall ratio.

Air leakage Air leakage describes the air permeability or air tightness of the building envelope. The air change rate of the building is a combination of the infiltration rate (air exchanged inadvertently with outside due to indoor/outdoor pressure differences and the envelope leakage of the building) and the ventilation rate (intended air exchanged with outside due to ventilation systems).

HVAC Systems Heating, cooling and ventilation systems are services used to provide a thermally comfortable (or operationally necessary) and healthy indoor environment with good indoor air quality (IAQ) for building occupants.

Occupant Behaviour The energy-related behaviour of building occupants has been shown to affect building energy consumption by up to 300% (Andersen et al. 2007). While simulating occupant or user behaviour is still in its infancy, it is important that any simulation tool should at least acknowledge, if not attempt to account for, the impact of the occupant on energy use. In its simplest form, examples of occupant behaviour with a building energy impact are switching lights on and off, interacting with the heating thermostat, or the opening and closing of windows.

30/04/2014

Grant No. 608775

25

Building size The volume of a building or more specifically, the conditioned volume is the most important size characteristics for energy modelling, as it dictates the amount of air inside the building that needs to be heated, cooled, or replaced via ventilation. However, volume is rarely reported in building databases, with floor area being a more common metric (BPIE 2011). It may be necessary to calculate or estimate the volume from information on floor area, ceiling height and number of storeys.

Sustainable Systems Sustainable systems generate electricity or provide heating or cooling (air and/or water) for the building at zero or low cost of energy compared to their sustainable counterparts. Sustainable systems are integral to decreasing the energy consumption of buildings.

Reverting back to our model of a building for energy modelling purposes (Figure 7), we can classify the above characteristics into their relevant building property subsets of: construction, function, geometry and systems (Table 6).

Table 6: Building characteristics important for energy modelling classified into their building property subsets

Construction  Glazing  Insulation  Air leakage

Function

Geometry

 Internal loads  Occupant behaviour

 Building size

Systems  HVAC systems  Sustainable systems

3.1.3. Top-down modelling approach Studies on building typology such as BPIE (2011), TABULA (IEE 2014), ENTRANZE (Kranzl 2014), RECS (US EIA 2009) and CBECS (US EIA 2003) have identified several categories which are useful for describing buildings with similar characteristics. Of particular interest when characterising buildings for energy modelling is:     

Construction date Form Region Retrofit history Use.

The following building characteristics need to be represented in our top-down city model: internal loads, glazing, building insulation, ventilation rate, HVAC systems, building size, and occupant behaviour. With the exception of occupant behaviour, these can all be considered intrinsic properties of a building, and so are more suitable to a bottom-up approach to energy modelling. For our top-down modelling approach we need to generalise these characteristics into the categories defined by studies on building typology: construction date, form, region, retrofit history and use.

30/04/2014

Grant No. 608775

26

It can be argued that the construction properties of a building are dependent on the date of construction, the region where the building was constructed, and whether or not the building has undergone an energy performance retrofit. Likewise, the function of a building is dependent on its intended use; the geometry of a building is dependent on its form; while the systems inside a building are dependent on its construction date, region, retrofit history and use. Expressed in equation form we obtain: (

) ( (

(11) (12)

)

(13)

)

(

)

(14)

Knowledge of a building’s construction typological categories will enable some of its energy properties to be established. For example, if we know the construction date, region and retrofit history of a building, its air leakage, glazing and insulation properties may be estimated. Likewise, knowledge of a building’s use will enable the occupant behaviour and internal loads to be estimated. If we know the physical form of a building then we can estimate its volume and wall/floor area ratio, and if we know a building’s construction date, region, retrofit history and use, we can estimate the properties of its HVAC equipment and its sustainable systems. Estimates for the building properties can be based on typical or average values obtained from building typology studies.

Finally, if we assign the building properties identified to be important for energy modelling to their typology categories via the building property subsets we obtain our final top-down building characterisation (Figure 8).

Figure 8: Final building energy characterisation

Table 7 summarises the building property subsets and the relevant building/system characteristics.

30/04/2014

Grant No. 608775

27

Table 7: Building properties by property subset

Construction Glazing

 Insulation (U-Value)  Solar transmittance (G-Value)  Window-to-wall ratio

Insulation

 U-Value (floors, walls, ceilings)

Air leakage

 Air changes per hour at 50 Pa of (de)pressurisation

Function Internal loads

 Heat gain from lighting systems (sensible)  Heat gain from plug loads (sensible)  Heat gain from occupants and occupant activities (sensible & latent)

Occupant Behaviour

 Occupancy pattern (e.g., office hours, residential , 24/7)  Energy attitude of occupants (wasteful, average, profligate)

Geometry Volume



Conditioned volume of the building

Wall-to-floor area



The ratio of the total wall area (including windows) to the total floor area of the building

ratio Systems Heating systems

 Fuel type  Input power  System efficiency

Cooling systems

 Fuel type  Input power  System efficiency

Ventilation systems

 Input power  System efficiency  Airflow rate

Sustainable systems

   

30/04/2014

Application (energy production/water heating) Fuel type Input power System efficiency

Grant No. 608775

28

3.1.4. Building typological categories The typological building property categories will be discussed below, based on the out from previous studies in the literature.

Region INDICATE is a European project and so the building characterisation focuses on EU buildings. However, for the INDACTE tool to reach its widest possible market base it should be possible to model smart cities in other regions.

Seven countries currently account for 65% of global construction growth: China, India, the United States, Indonesia, Canada, Russia and Australia (Global Construction Perspectives 2013).

The BPIE (2011) split the EU-27 countries into three regions based on similarities in climate, building typology and the housing market: 

North and West Europe (Pop. 281m, 50% of building floor space) Austria, Belgium, Switzerland, Germany, Denmark, Finland, France, Ireland, Lithuania, the Netherlands, Norway, Sweden, UK



South Europe (Pop. 129m, 36% of building floor space) Cypress, Greece, Spain, Italy, Malta, Portugal



Central and East Europe (Pop. 102 m. 14% of building floor space) Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovenia, Slovakia.

The IMPRO-Buildings (2008) study also used three regions defined by their associated number of heating degree days (HDD): 

North Europe (4001 to 5823 HDD) Lithuania, Latvia, Estonia, Sweden, Finland



South Europe (564 to 2500 HDD) Malta, Cyprus, Portugal, Greece, Spain, Italy, France



Central Europe (2501 to 4000 HDD) Belgium, the Netherlands, Ireland, Hungary, Slovenia, Luxembourg, Germany, United Kingdom, Slovakia, Denmark, Czech Republic, Austria, Poland.

While it is useful to consider building region by HDD, unlike the BPIE study the IMPRO-Buildings study does not include non-residential buildings.

The RECS (2009) database in the United States uses several levels of classification to divide the country: 

Geography Northeast, Midwest, South, West (with subdivisions for each region)

30/04/2014

Grant No. 608775

29



Climate Very cold/cold, Mixed-humid, Mixed-dry/Hot-dry, Hot-humid, Marine.

The buildings are further split into ‘rural’ and ‘urban’ classifications, and ‘metropolitan’, ‘micropolitan’, and ‘not metropolitan or micropolitan’.

Due to the very large range of region-specific building types throughout the world it is very difficult to make recommendations as to which region classifications should be used for the INDICATE project. Perhaps the best method to adopt would be to build up a dataset over time, starting with broad European regions such as those used by the BPIE and the IMPRO-Buildings project.

Suggested region sub-classifications:    

North and West Europe South Europe Central and East Europe Then allow for the provision of other regions throughout the world and regions with finer geographical resolution (e.g. country specific).

Construction date The energy performance of a building is strongly correlated with its age, unless the building has undergone an energy retrofit. The BPIE (2011) analysed the age and energy performance of residential buildings across Europe.

They were able to consolidate the floor area data into three distinct age bands:   

Old: pre-1960 Modern: 1961 to 1990 (most EU countries experienced a construction boom where the residential building stock nearly doubled) Recent: 1991 to 2010 (the BPIE study was conducted in 2011)

While the BPIE acknowledges that there is significant country-to-country variation in building energy performance within their three age bands, distinctions need to made in order to simplify analysis and facilitate comparison.

Within the regions of South, North and West, and Central and East Europe, the age distribution of European buildings is shown in Figure 9.

30/04/2014

Grant No. 608775

30

Figure 9: Construction date of the European building stock by region (BPIE, 2011)

For finer time resolution we can refer to the ENTRANZE study to display building construction date according to decade (back to the 70s, then 1945 to 1969, and pre-1945) (Figure 10).

2000-2008 1990-1999 1980-1989 NW 1970-1979

S

1945-1969

CE

< 1945 0

10

20

30

40

50

60 Millions

Figure 10: Construction date of EU dwellings (millions) by region. Data from ENTRANZE (Kranzl 2014)

Outside of Europe, the RECS database also splits the building age classifications into decades, starting at pre-1940, then 1940 to 1949 etc. Using decades is the obvious way to classify buildings by construction date, especially considering how quickly building construction standards have developed since the 1940s.

Suggested age sub-classifications: ď&#x201A;ˇ ď&#x201A;ˇ

Pre-1940 Then decades from the 1940s onwards.

30/04/2014

Grant No. 608775

31

Form Broadly, buildings can be categorised into residential (single-family houses or multi-family apartment blocks) or non-residential (commercial buildings such as offices and retail outlets, and industrial buildings such as factories).

Approximately 76% of European building floor space is residential buildings, with the remaining 24% classified as non-residential. Residences account for approximately 25% of EU-27 final energy consumption, and 68% of the final building energy consumption (European Commission 2014). Therefore, residential buildings will form a significant proportion of energy consumption in the model of the smart city. European residences are split into two-thirds single family, and one-third multi-family (European Commission 2013; BPIE 2011) (Figure 11).

Nonresidential, 24%

Multi-family, 27%

Single family, 49%

Figure 11: European building floor space mix (European Commission 2013; BPIE 2011).

Residential form Customarily, residential buildings are split into two categories â&#x20AC;&#x201C; single-family and multi-family. Single-family dwellings are typically occupied by a solitary family unit. They can be detached, semi-detached or part of a terrace of houses. Multi-family dwellings are generally larger buildings split into multiple residences. They can include apartment blocks, flats, or high-rise towers.

The TABULA project (IEE 2014) is a joint venture between 17 European partners to classify residential buildings from 13 EU countries. Within the TABULA project residential buildings are split into four categories of singlefamily house, terraced house (single-family), multi-family house, and apartment block. In the EU-based study IMPRO-Building (Nemry et al. 2008), residential buildings are split into the three categories of single-family, multifamily, and high-rise blocks. The three types of building account for 53%, 37% and 10% of the EU building stock, respectively. Multi-family buildings were defined to have eight storeys or less, high-rise buildings had 9 storeys or more. The BPIE (2011) survey splits residential buildings into just two categories of single-family dwellings and apartment blocks.

30/04/2014

Grant No. 608775

32

The RECS (US IEA 2009) database in the US classifies residential buildings as single-family, apartment blocks and mobile homes. The single-family buildings are split into detached and attached, while the apartment blocks are split into buildings with two to four units, and then five units or more. Table 8 shows the number of units and the energy use intensity (EUI) of the different residential homes in the United States.

Table 8: Energy use intensity for residential US homes from RECS (US EIA 2009)

Building Type Single-Family: Single-Family Detached Single-Family Attached Multi-Family: Apartments in 2-4 Unit Buildings (multi-family) Apartments in 5 or More Unit Buildings (high rise) Mobile Homes

Number of units [millions] 78.6 71.8 6.7 28.1 9.0 19.1 6.9

EUI [kWh/m2a] 135 134 145 189 218 172 197

For the INDICATE project we must decide which sub-classifications of residential buildings to include within the model. Ideally this decision should be made based on European data. However, from the existing data sources analysed for this report, there is a lack of disaggregation of building energy use between the different residential sub-classifications. Therefore, using data from the US-based RECS, the difference in energy performance between single-family detached and attached houses is small (less than 10%), so we do not need to consider these subclassifications separately. The difference between 2 to 4-unit (multi-family) and 5 or more unit (high rise) buildings is larger (over 20%), so they should both be considered in the INDICATE model.

Non-residential form Non-residential buildings belong to an almost infinite range of physical form. Many different shapes exist, ranging from standard cuboid shapes, to more elaborate shapes depending on spatial constraints or the imagination of the architect. For city-scale energy modelling, it would be impossible to account for all commercial building forms and geometries. It may be suitable to limit the form of commercial buildings to simple cuboids, or combinations of cuboids, and then determine the building energy performance from other properties such as the building properties identified above. However, it is uncertain whether this is sufficient to capture accurately building energy performance. Further investigation will be required in order to make an informed decision on how the form of commercial buildings should be represented in the model.

Suggested building sub-classifications: Residential  Single-family  Multi-family (2 to 4 units) and  High rise (5 units or more).

Non-residential  Cuboids or combinations of cuboids.

30/04/2014

Grant No. 608775

33

Use The use of a building, determined by the functions performed within it, directly impacts the energy performance.

For example, a residential building could be used as a dwelling or it could be used as a small office, resulting in two different energy profiles for the same physical building. Historically, classifying non-residential buildings is very difficult due to the broad range of building uses, and even multiple uses within the same building. Based on their own survey results, the BPIE classified building use into residential, wholesale and retail, offices, educational, hotels and restaurants, hospitals, sports facilities and ‘other’. They acknowledge the vast disparity found in commercial buildings, and also the prevalence of survey respondents who classified their building as ‘other’, suggesting that further investigation is required in order to ascertain what types of buildings fall into this category.

Figure 12: European building floor space mix (BPIE, 2011)

Similar building classifications are found in other typology studies. The ENTRANZE study uses classifications of wholesale and retail, offices (private and public), hotels and restaurants, health, education and ‘other’. DESTATIS (the German national statistics office) use classifications of institutional buildings, office and administration, agricultural and farming, operational, factory and workshops, retails and warehouse, hotels and restaurants and ‘other’ (DESTATIS 2012). DESTATIS incorporate industrial-type buildings such as factories and workshops. While it is difficult to provide a general description of this type of building, they should be accounted for in an energy model of the smart city as they will be large users of energy and have a high potential for utilisation of renewable energy. Table 9 summarises building use classifications from the BPIE, ENTRANZE and DESTATIS studies.

Table 9: Non-residential building classifications by source

BPIE (2011) 30/04/2014

ENTRANZE (2014) Grant No. 608775

DESTATIS (2012) 34

Wholesale and retail Offices Educational Hotels and Restaurants Hospitals Sports facilities Other

Wholesale and retail Offices (private) Offices (public) Education Hotels and restaurants Health Other

Retail and warehouse Office and administration Operational Hotels and restaurants Factories and workshops Institutional Agricultural and farming Other

CIBSE benchmarking (Bruhns et al. 2011) provides a much more comprehensive list of building uses (Table 10).

Obviously more classifications of building available to the INDICATE tool will mean more flexibility for the user.

However, the accuracy of the tool will depend on the accuracy of the data for the different building uses. Care must be taken to ensure that building uses adopted in the INDICATE tool are representative of real buildings and their energy use.

Table 10: Building uses from CIBSE benchmarking

CIBSE (2011) General office High street agency General retail Large non-food shop Small food store Large food store Restaurant Bar, pub or licensed club Hotel Cultural activities

Entertainment halls Swimming pool centre Fitness and health centre Dry sports and leisure facility Covered car park Public buildings with light usage Schools and seasonal public buildings University campus Clinic Hospital - clinical and research

Long term residential General accommodation Emergency services Laboratory or operating theatre Public waiting or circulation Terminal Workshop Storage facility Cold storage

Perez-Lombard et al. (2008) extracted data from the CBECS (US EIA 2003) database and presented energy use intensities for non-residential buildings (Table 11). While the absolute values for the EUIs are for US buildings, the ratios may be useful to show how energy use scales between different types of non-residential buildings. Of particular interest is the difference in EUI between hotels (316 kWh/m2a) and restaurants (814 kWh/m2a). These two types of building are typically considered together in the same classification (as by the BPIE and ENTRANZE) yet according to data from CBECS they have an energy performance which is different by almost a factor of three.

However, it should be noted that many larger hotels actually contain restaurants, so the figures may be distorted â&#x20AC;&#x201C; another example of difficulties faced when trying to generalise building energy use.

Table 11: Energy use intensity for US buildings. Table from PĂŠrez-Lombard et al. (2008) based on data from US EIA (2001; 2003)

Building Type Residential (RECS) Retail Schools Offices Hotels Supermarkets 30/04/2014

EUI [kWh/m2a] 147 233 262 293 316 631 Grant No. 608775

Ratio (relative to residential) 1.0 1.6 1.8 2.0 2.1 4.3 35

Hospitals Restaurants

786 814

5.3 5.5

Producing a final list of building uses for the INDICATE tool may not be so useful. Instead, a dynamic list that grows with the availability of accurate data could be a more sensible approach. The main building uses should be included, plus the inclusion of larger buildings with high energy uses that may have the potential to exploit renewable energy or demand response such as factories and swimming pools.

Users of the INDICATE tool should be able to input their own data on building properties and energy performance in order to fine-tune the tool to their own purposes and promote flexibility of the INDICATE model. In the absence of user-specific data the tool should have predetermined or ‘default’ entries for the different building uses and characteristics.

Suggested building uses:      

Educational Factories and workshops Hospitals Hotels Offices Residential

     

Restaurants and bars Sports facilities Stadia Supermarkets Swimming pools Wholesale and retail

Plus the capability for the tool to utilise a dynamic list of building uses Retrofit history The degree to which the energy performance of a building can be improved by energy retrofit is dependent on the specific retrofit measures. In terms of residential buildings, the most simple and cost effective way to improve energy performance is to insulate the attic (COM 2006). Further improvements can be achieved from ‘simple’ measures such as insulating walls and floors, and air sealing the building envelope, all the way up to installing new building technologies such as solar panels and heat pumps. In commercial buildings the most cost effective measure to achieving better energy performance is to improve the energy management system (COM 2006).

There is no widely-accepted consensus on the definition or classification of retrofit packages. The INDICATE tool should allow for a range of energy retrofit levels so that future energy efficiency measures can be accounted for.

A straightforward approach would be to allow retrofit to current building codes or standards, and also ‘deep energy retrofit’. Different building standards could be included within the model so that the user may specify which standard is applicable to their city model. Deep energy retrofits have different definitions (e.g., 50% energy savings (RMI 2012)) but the term is used to describe retrofit measures which take the energy performance of the building beyond that required by codes and standards. It must then be decided whether the deep energy retrofit option is relative to the original energy performance of the building or relative to the relevant building standard or code.

Suggested retrofit options:

30/04/2014

Grant No. 608775

36

 

3.2.

Retrofit to building standard/code (allowing for the provision of country-specific and user-definable standards/codes) Deep Energy Retrofit (relative to building standards or existing energy performance).

Electric vehicles (EVs)

Electric vehicles (EVs) are seen as a measure to reduce carbon emissions, exhaust emissions, fossil fuel consumption and traffic noise. Advantages of the electric motor (EM) over the internal combustion engine (ICE) include: (Kühne 2010)        

Locally emission free Quiet during operation No requirement for a gearbox Increased lifetime due to reduced wear Less problematic recuperation of braking energy Reduced engine brake abrasion Zero idling energy consumption More able to utilise renewably-produced electricity.

The main disadvantages are range and cost, although prices are falling and range is increasing with the development of battery technology.

EV classifications are: (Serra 2012) 

Battery EV (BEV)

An EV that is powered solely by an electrochemical battery.



Fuel Cell EV (FCEV)

An EV powered by a fuel cell. FCEVs are likely to be hybridized with a battery or ultra-capacitor power source.



Ultra-capacitor EV (UCEV)

En EV powered by an ultra-capacitor. UCEVs are likely to be hybridized with a battery of a fuel cell power source.



Hybrid EV (HEV)

Any vehicle which combines electric and combustion power sources.



Grid-enabled EV (GEV)

Any EV which can be connected to the electricity distribution grid for power exchange. This can include BEVs, PHEVs and UCEVs.



Plug-in Hybrid EV (PHEV)

Any HEV which can be connected to the grid to charge its batteries. PHEVs can be driven by either their combustion engine or electric motor.



Extended Range EV (EREV)

A PHEV which can only be driven using its electric motor.



Full (or Strong) HEV

An HEV which can be driven for short distances using only the electric motor, but cannot be charged from the grid.



Power-assist Hybrid EV

An HEV with a reduced electric motor and battery, whereby their propulsion role is restricted to assisting the combustion engine via torque boosts during heavy acceleration.

30/04/2014

Grant No. 608775

37



A combustion engine-driven vehicle that uses a small electric motor and battery for regenerative braking and engine shutdown during idling.

Micro Hybrid EV

3.2.1. Hybrids Hybrid EVs (HEV) represent a combination of internal combustion engine and electric motor technologies. There are four distinct hybrid vehicle architectures: (Mi et al. 2011) 

Series hybrid

The ICE and electric motor are connected in series to the drive train. The internal combustion engine is the primary fuel converter changing petrol into mechanical energy. The mechanical output of the ICE powers a generator which drives the electric motor. The electric motor then provides mechanical power to the wheels. The decoupling of the ICE from the drive train simplifies the control of the engine, allows the engine to be operated at optimum speed, and removes the need for a transmission system. A battery may also be used to drive the electric motor.



Parallel hybrid

The ICE and electric motor are connected to the drive train in parallel, so that both propulsion systems can be used to deliver power to the wheels together or separately. The electric motor can be used as a generator to recover kinetic energy from braking.



Series/parallel

This is a combination of series and parallel hybrid architecture that allows the fuel efficiency and drivability to be optimised based on the operating conditions of the vehicle. The vehicle can be operated as either a series hybrid or a parallel hybrid. A mechanical link is introduced between the ICE and drive train so that the ICE can deliver power to the wheels. A second electric motor is used as a generator.



Complex hybrids

Complex architecture that offers higher speeds and load efficiencies via upgraded planetary gear sets that provide greater flexibility in managing the proportion of mechanical and electric power delivered to the wheels (Serra 2012).

3.2.2. Automobile efficiencies There are two primary systems within the modern automobile – the vehicle propulsion (VP) system and the energy storage (ES) system. The VP system encompasses all of the components that participate in the conversion of energy stored by the ES system into kinetic energy at the wheels (Serra 2012). This includes the engine or motor, plus the transmission, crankshafts, differentials etc. The ES system in a traditional motor vehicle consists of the combustion fuel and the storage tank. In an EV the ES system is the battery. Both the VP and ES systems have their own efficiency.

Regarding the efficiency of automobiles there are several types of efficiency: 

Tank-to-wheel (T2W)

30/04/2014

This is the product of the efficiency of the VP and ES systems. It accounts for all of the efficiencies within the vehicle. T2W efficiency for ICE vehicles is approximately 15%. For BEVs the T2W efficiency is approximately 75% (Serra Grant No. 608775

38

2012). 

Well-to-Tank (W2T)

This is the efficiency of the system which provides energy to the vehicle i.e., in the case of EVs it is the efficiency of the electricity grid, while in the case of ICE vehicles it is the efficiency of the extraction, refining and distribution of fossil fuels.



Well-to-Wheel (W2W)

This is the combination of the T2W and W2T efficiencies which allows the calculation of the final efficiency of the automobile. For ICEs the W2W efficiency is approximately 13%. Assuming night-time charging using electricity produced by baseload power plants, the W2W efficiency of a BEV is approximately 29% i.e., more than twice that of an ICE-driven vehicle (Serra 2012).

3.2.3. EV power sources Most EVs use electrochemical batteries as energy storage systems. The most common batteries currently in use or at an experimental stage are Lead-acid, Ni-MH, Li-ion, Na-NiCl and Zn-O2. Properties of the different battery technologies are compared in Table 12. Other proposed energy storage technologies currently under consideration include flywheels, electrostatic storage with ultra-capacitors and hydrogen storage using fuel cell systems (Dixon 2010).

In terms of the INDICATE tool the actual battery technology is irrelevant; the most important consideration from a modelling perspective is the battery characteristics of energy density, power density and cycle life.

Table 12: EV energy storage systems (Dixon 2010)

Energy Source USABC* Lead-acid Ni-MH ZEBRA Zn-O2 Flywheels Ultra-capacitors

Energy density [Wh/kg] 200 35 70 110 200 40 5

Power density [W/kg] 400 150 220 150 100 3,000 2,000

Cycle life [charges] 1,000 700 1,500 1,500 1 (Electric Fuel) 5,000 500,000

Cost [US$/kWh] 100 150 1,500 700 5,000 20,000 25,000

*USABC (United States Advanced Battery Consortium targets)

Li-Ion Lithium-ion refers to a range of battery chemistries that include: (Serra 2012)    

NCA (lithium nickel cobalt aluminium) LMO (lithium manganese spinel or lithium manganese polymer) LMO/LTO (lithium titanate) LFP (lithium iron phosphate).

Properties of Li-ion batteries are shown in Table 13.

30/04/2014

Grant No. 608775

39

Table 13: Lithium-ion battery characteristics (Serra 2012). As the technologies are immature exact figures cannot be presented for all battery chemistries.

Li-ion chemistry NCA

Energy density [Wh/kg] 170

Power density [W/kg] Highest

150 150 140

Good Good OK

LMO LMO/LTO LFP

Cycle life

Cost

Good: 350,000 cycles OK Very good Very good

High High Highest Low

3.2.4. EV charging and infrastructure There are currently three different options available to recharge EV batteries: 

Slow-charge

The EV is plugged into a low-voltage (e.g., 120, 220 or 240V) power supply and left to charge. This typically occurs when the vehicle is parked overnight at home, or during the day at the office. Full charge time can be in the range of four to eight hours for current technology.



Fast-charge

Increasing the supply voltage allows charge time to be decreased. Table 14 shows three fast-charge standards that have recently emerged. Fast-charging can reduced full charge time down to 30 minutes.



Battery swapping

Separating the battery of an EV from the EV itself is seen as a way to reduce the initial financial investment in an EV. EV owners would not own the battery outright; rather hire the battery from a battery service provider (BSP). A depleted battery can then be swapped for a fully charge one in a matter of seconds.

Table 14: Specification of three fast-charging standards (Khaligh 2010)

Standard SAE

Region North American

IEC 62196

Europe

TEPCO

Japan

30/04/2014

Charging parameters Voltage: 300-600 V DC or 3-phase AC Current: 80 to 400 A Power output: 90 to 240 kW (Not finalized) AC Standard: Maximum AC Power output: 172.5 kW Voltage: 690 V, 50 to 60 Hz Maximum AC current: 250 A DC Standard: Maximum DC Power output: 240 kW Voltage: 600 V Maximum DC current: 400 A Input: 3-phase 200 V Maximum DC output power: 50 kW Maximum DC output voltage: 500 V Maximum DC output current: 100 A

Grant No. 608775

40

Infrastructure will be required to allow the recharging of EVs (charge spots) or battery swapping stations.

Charging demand will vary from one station to another depending on its location. The best locations for the charge spots and battery swapping stations can be determined by network optimisation.

3.2.5. EVs and the smart grid Grid-connected EVs can be used as ancillary storage capacity for the smart grid, known as vehicle-to-grid (V2G).

The energy stored in EV batteries can be used to meet excess demand rather than dispatching expensive and carbon intensive peaking power plants. Then, when electricity supply outstrips demand the excess energy from the grid goes back into charging the EV battery. These times of high demand are short-lived and infrequent meaning that the owner or user of the grid-connected EV is unlikely to be affected. Control measures at the point of the EV also allow the owner to limit the amount of charge in the EV battery made available to the grid to prevent excessive discharging. It is estimated that vehicles only spend 4% of their existence being driven; even during rush hour 92% of vehicles are parked, and therefore theoretically available to the smart grid (Kempton & Tomić 2005). The use of BEVS could potentially help load balancing of the future smart grid, improve overall grid efficiency and reduce the need to increase grid capacity. The total V2G storage capacity should be accounted for in the INDICATE tool.

3.3.

Electric public services

The energy demand of electric street lighting and telecommunications networks will be considered in the following sections.

3.3.1. Street lighting Mills (2002) estimated that street lighting accounts for 8% of global lighting energy consumption, using 1507 PJ of energy annually. In Ireland, it is estimated that there are 420,000 street lights in the country representing an electrical demand of 50 MW, using 205 GWh of electricity and producing 110,000 tonnes of CO2 per annum (SEAI 2011). The SEAI report estimates that street lighting accounts for 15 to 35% of the total energy usage of Irish local authorities. In the same report the authors describe the difficulty in developing a robust energy performance indicator (EPI) that will allow effective comparison of street lighting use across the country. However, they go on to suggest that ‘energy use per light’ could be used. Generally, street light performance is described by the luminous efficacy, given in units of lumens/Watt (i.e., the human-visible light generated by the lamp divided by the power required to operate it). Table 15 and Figure 13 show some typical street lighting characteristics for different lighting technologies.

Table 15: Typical lighting characteristics and adoption for street lighting in Ireland (SEAI 2011; US DOE 2013a)

Lighting technology MBFI (High pressure mercury) SOX (Low pressure sodium) SON (High pressure sodium)

30/04/2014

Approx. lifespan (years) 3 3 5

Luminous efficacy (lumens/watt) 50 160 90

Grant No. 608775

Use in Ireland ≤ 3% 55% 42%

41

Metal halide Mercury vapour LED (Light emitting diodes) T5 â&#x20AC;&#x201C; CFL Fluorescent

5 8 to 10 9 to 12 5

65 to 120 40 to 601 100 to 140 45 to 80

Unknown/negligible Unknown/negligible Unknown/negligible Unknown/negligible

For urban and suburban areas a street lighting efficacy of 75 lumens/Watt is recommended (NYSERDA 2002). In addition to the lamp or bulb, the majority of street lighting requires a ballast â&#x20AC;&#x201C; the parts of the light fixture that incorporates control equipment such as current limiters, dimmers and starters. The lighting ballast also has an associated electrical energy consumption. Using the above values from SEAI (2013) for Irish street lighting we can obtain an average power usage of 120 W per street light (50 MW divided by 420,000 lights). This number will decrease as advanced lighting technologies are adopted. Figure 13 shows luminous efficacies for a range of lighting technologies (US DOE 2013a).

Figure 13: Current and projected luminous efficacy ranges for common light sources as of Jan 2013. Luminare efficacy considers the entire lighting system as opposed to just the lamp/blub (US DOE 2013a).

Streetlight spacing is dependent on factors such as the height of the light fixture, the light output of the bulb, the angle of dispersion of the light and the lighting configuration (e.g., lights installed on one side of the road or two).

A general rule is to use spacing of 12 to 15 metres for roads with pavements for pedestrians, and 40 to 45 metres for roads with no pedestrians2.

3.3.2. Telecommunications Information and communication technologies (ICT) are estimated to account for between 2 and 4% of global carbon emissions, with one-sixth of this attributable to telecommunications networks (Vereecken et al. 2011).

1

From (NYSERDA 2002)

2

From http://www.pps.org/reference/streetlights/ [accessed March 2014]

30/04/2014

Grant No. 608775

42

While this is a small fraction of worldwide energy use when compared to that used by buildings, the energy consumption of telecommunications is growing rapidly, and expected to double by 2020 (Pickavet et al. 2008).

Average mobile (cellular) phone subscription rates for European countries are approximately 125 per 100 inhabitants, indicating that there is more than one mobile subscription per European inhabitant (European Commission 2014). European telecommunications coverage by technology (Figure 14) and technology power consumption (Figure 15) will allow estimations for telecoms energy usage to be made (European Commission 2012; Vereecken et al. 2011). In the absence of data on data transfer speeds for Europe, the midpoint of each range in Figure 15 could be used as an approximation.

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% DSL

VDSL

FTTP

WiMAX

Standard cable

Docsis 3 cable

HSPA

LTE

Satellite

Figure 14: Telecommunications coverage (households) by technology for the EU in 2012 (European Commission 2012)

30/04/2014

Grant No. 608775

43

Figure 15: Power consumption per subscriber for various telecommunications technologies (Vereecken et al. 2011)

30/04/2014

Grant No. 608775

44

4.

City networks

In this section, two major city networks: electricity and natural gas are presented. Discussion on the electricity network includes the â&#x20AC;&#x2DC;smart gridâ&#x20AC;&#x2122; and its components, grid paradigms, power generation, transmission and distribution grids, and grid management. The natural gas distribution system is outlined and modelling issues such as the calculation of nodal flow and pressure are presented.

4.1.

Electricity networks

Discussion of the smart grid and the electricity distribution network follows.

4.1.1. The smart grid The smart grid is regarded as the next generation power grid. It uses two-way flows of electricity and information to create a widely distributed automated energy delivery network. By utilizing modern information technologies, the smart grid is capable of delivering power in more efficient ways and responding to wide ranging conditions and events. The smart grid should be able respond to power generation, transmission, distribution, and consumption events that occur anywhere in the grid, and adopt corresponding strategies. More specifically, the smart grid can be regarded as an electric system that uses information, two-way energy flows, cyber-secure communication technologies, and computational intelligence in an integrated fashion across electricity generation, transmission, substations, distribution and consumption to achieve a system that is clean, safe, secure, reliable, resilient, efficient and sustainable (Fang et al. 2012). The comparison between the existing grid and the smart grid is given in Table 16.

Table 16: Comparison between the existing grid and the smart grid (Farhangi 2010)

Existing Grid Electromechanical One-way communication Centralized generation Few sensors Manual monitoring Manual restoration Failures and blackouts Limited control Few customer choices

Smart Grid Digital Two-way communication Distributed generation Sensors throughout Self-monitoring Self-healing Adaptive and islanding Pervasive control Many customer choices

In order to realise this new grid paradigm, the National Institute of Standards and Technology (NIST), USA provided a conceptual model (Figure 16), which can be used as a reference for the various parts of the electric system where smart grid standardisation efforts are taking place (NIST 2010). This conceptual model divides the smart grid into seven domains. Each domain encompasses one or more smart grid actors, including devices, systems, or programs that make decisions and exchange information necessary for performing applications. The interaction of roles in different domains is carried out using secure communication channels.

30/04/2014

Grant No. 608775

45

Figure 16: The NIST updated conceptual model for a smart grid (NIST 2010)

The smart grid from a technical view point can be divided into three major systems: smart infrastructure, smart management and smart protection systems. The smart infrastructure is then classified into the three following subsystems: (Fang et al. 2012) ď&#x201A;ˇ ď&#x201A;ˇ ď&#x201A;ˇ

The smart energy subsystem for advanced electricity generation, delivery, and consumption The smart information subsystem for advanced information metering, monitoring, and management in the context of the smart grid The smart communication subsystem for communication connectivity and information transmission amongst systems, devices, and applications in the context of the smart grid.

New grid paradigms in smart energy subsystems Two paradigms widely regarded as important components of the future smart grid which also take advantage of other smart grid technologies are the micro-grid, and grid-to-vehicle and vehicle-to-grid (G2V & V2G).

Micro-grid is a localised grouping of electricity generation, energy storage, and loads. Under normal operation, the micro-grid is connected to a traditional power grid (macro-grid). The users in a micro-grid can generate low voltage electricity using distributed generation, such as solar panels, wind turbines, and fuel cells. The single point of common coupling with the macro-grid can be disconnected, allowing the micro-grid to function autonomously (Kaplan & Sissine 2009). An example of a micro-grid is shown in Figure 17 where the lower layer shows the physical structure of the micro-grid, including four buildings, two wind generators, two solar panel generators, and one wireless access point (AP). The buildings and generators exchange power using power lines, and exchange information via an AP-based wireless network. The top layer shows the information flow within the micro-grid, while the middle layer shows the power flow network (Fang et al. 2012).

30/04/2014

Grant No. 608775

46

Figure 17: An example of a micro-grid (Fang et al. 2012)

Grid-to-Vehicle (G2V) and Vehicle-to-Grid (V2G): An electric vehicle (EV) is a vehicle that uses one or more electric motors for propulsion. In G2V, EVs are often powered by stored electricity originally generated by an external power source, and thus need to be charged after the batteries deplete. In V2G, V2G-enabled EVs can communicate with the grid to deliver electricity to the grid when they are parked and connected (Kempton et al. 2008).

Power generation In contrast to power generation in the traditional power grid, smarter power generation becomes possible as two-way flows of electricity and information are supported. A key power generation paradigm enabled by smart grids will be distributed generation (DG) (Fang et al. 2012). DG takes advantage of distributed energy resource systems (e.g., solar panels and small wind turbines), which are often small-scale power generators (typically in the range of 3 kW to 10,000 kW), in order to improve power quality and reliability. The IEA point out that a power system based on a large number of small and reliable DGs can operate with the same reliability and a lower capacity margin than a system of equally reliable large generators (Fraser 2002).

Implementing DG(s) in practice is not an easy due to: i. ii.

DG involves large-scale deployments for generation from renewable resources (Molderink et al. 2010), and the usual operation costs of distributed generators for generating one unit of electricity are high compared with that of traditional large-scale central power plants (Fraser 2002).

DG was therefore predicted to evolve from the present system by (Fraser 2002): i. ii. iii.

Accommodating DGs in the current power system Introducing a decentralized system of DGs cooperating with the centralized generation system, and Supplying most power by DGs and a limited amount by central generation.

30/04/2014

Grant No. 608775

47

Smart transmission and energy management The smart transmission grid can be regarded as an integrated system that functionally consists of three interactive components: smart control centres, smart power transmission networks, and smart substations (Li et al. 2010). Based on existing control centres, the future smart control centres enable many new features such as analytical capabilities, monitoring, and visualization. The smart power transmission networks are conceptually built on the existing electric transmission infrastructure. However, the emergence of new technologies (e.g., new materials, electronics, sensing, communication, computing, and signal processing) can help improve the power utilization, power quality, and system security and reliability, thus drive the development of a new framework architecture for transmission networks.

Two in-home power distribution systems were proposed by Takuno et al. (2010), in which information is added to the electric power itself and electricity is distributed according to this information. The first system is a circuit switching system based on alternating current (AC) power distribution, and the second is a direct current (DC) power dispatching system that uses power packets. The two systems have the potential to act as an intelligent power router. More specifically, supplied electricity from energy sources can be divided into several units of payload. A header and footer are attached to the unit to form an electric energy packet. When the router receives a packet, the packets are sorted according to the addresses in the headers and then sent to the corresponding loads. Using energy packets allows the provision of power to be regulated easily by adding control and monitoring.

Brenna et al. (2012) proposed integrated management of energy flows between different subsystems. A sustainable energy microsystem (SEM) can be used for the integration of these subsystems, allowing the optimisation, integration and management of many services. The SEM acts as a flexible energy hub supplying and storing multiple energy carriers. An example of a smart city utility network that supplies several SEMs is shown in Figure 18 where each SEM is supplied by a unique point of delivery (POD). The centrality of customers is proposed not only in economical and commercial terms, but also with energy and network implications. Customers could vary their cumulative load profile and/or the power produced locally in the SEM in order to reduce the exchange of energy with the utility.

Figure 18: A smart city utility network that supplies several SEMs (Brenna et al. 2012) 30/04/2014

Grant No. 608775

48

4.1.2. Electricity distribution grids The electricity distribution grid primarily consists of transmission lines and substations.

Transmission lines In either traditional or smart grid, the transmission lines are vital components. Figure 19 shows the transmission line network in Dublin, in January 2013 (EirGrid 2013). The network consists of 110, 220, 275, and 400 kV transmission lines, cables and substations. Electricity is transported over long distances at high voltages to reduce transmission losses (primarily to heat from the resistive properties of the transmission line materials). The voltage must then be decreased for safe use at the end point. Electrical transformers are used at substations to step up or step down the voltage. High-voltage transmission lines (120 kV or more) are generally overheard, or more expensively, underground.

Total transmission losses were estimated to be 6.8% for the United States electricity grid in 2008 (Bowles 2008).

Approximately 2.5% of electrical power is lost for every 1000 km of transmission lines (Vassell & Maliszewski 1969).

Figure 19: Transmission line network in Dublin, Ireland, January 2013 (EirGrid 2013)

Typical limiting factors on power transfer along short, medium, and long transmission lines are demonstrated in Figure 20. The thermal limit is related to the material properties of the line and is constant independent of length.

Since stability is a system property rather than a material property, stability limits change depending on the length of a line and other system conditions. Thermal considerations generally limit power transfer on short lines, 30/04/2014

Grant No. 608775

49

while longer lines tend to be stability limited. In particular, power transfer on medium-length lines is usually constrained by voltage stability, while the longest lines are limited by their transient stability (MIT 2011).

Figure 20: Three primary constraints of transmission lines (MIT 2011)

One key consideration is the balance between low transmission costs for power resources local to load centres and favourable capacity factors of resources distant from load centres. This trade-off is illustrated in Figure 21.

The optimal solution is a transmission overlay serving resource zones of different types (MIT 2011).

Figure 21: Trade-off between transmission cost and capacity factors

The interconnectedness of the grid compounds the difficulty in controlling power. Multiple transmission lines often intersect at one substation, making it impossible to change the flow on one line without affecting others. As a consequence, energy flowing from one location to another follows multiple paths and may cross jurisdictional 30/04/2014

Grant No. 608775

50

boundaries. These so-called loop flows can create adverse or beneficial physical and economic effects in several jurisdictions (MIT 2011).

Substations Substations act as nodes on the electricity distribution grid. There are five main types of substation: 

Transmission substations

Connect two or more transmission lines



Distribution substations

Transfer power from the transmission system to the distribution system



Collector substations

Used in distributed generation grids to collect renewably-generated power and put it onto the transmission system



Converter substations

Used to change the frequency of the current, or convert AC to DC or vice versa



Switching substations

A substation without transformers operating at a single voltage, used for switching current to back-up transmission lines.

The primary concern of power distribution system expansion planning is to determine the number, location and size of future distribution substations. A model to automatically select the optimal sizing and location of distribution substations without requiring any candidate substation locations was developed by Hongwei et al.

(1996). Eltawil et al. (2010) recently proposed a new planning optimization model for distribution substation siting, sizing, and timing, which involves using linear functions to express the total cost function. The latter model includes different electrical constraints such as voltage drops, substation and transformer capacities, power flow, and radial flow constraints.

4.2.

Natural gas

Natural gas accounts for 36% of EU-27 building energy consumption (Kranzl 2014); used for heating, hot water and cooking. The gas distribution network in the UK is known as the National Transmission System (NTS). The NTS is a high-pressure (approx. 85 atm) gas network that consists of: (Zaiser 2003)     

Terminals Compressor stations Pipelines Storage facilities Offtakes.

Gas is refined and processed at large coastal terminals where it is made suitable for the NTS. Impurities are removed from the gas and the calorific content is measured to ensure that it meets minimum acceptable values (approx. 39 MJ/m3 in the UK). Compressor stations are used to pressurise the gas to overcome frictional losses from the network pipelines and ensure that the gas can be transmitted safely. The majority of compressor stations use gas-powered jet engines to pressurise the natural gas. When gas supply outstrips demand, storage facilities are used to take gas off the network and hold it in reserve. At the storage facilities the natural gas is 30/04/2014

Grant No. 608775

51

converted into liquefied natural gas (LNG) by cooling it to -161°C. LNG has a much higher density than natural gas (a factor of 600) allowing large volumes of gas to be stored. Offtake installations are used to remove gas from the NTS and supply it to Local Distribution Zones (LDZs) or gas-fired power stations. LDZs are smaller transmission networks which supply natural gas to domestic households at lower pressure (approx. 25 atm) than the NTS. Figure 22 shows the National Transmission System for the UK (Zaiser 2003).

Figure 22: The UK’s National Transmission System (NTS) used for the storage and distribution of natural gas (Zaiser 2003).

Acha and Hernandez-Aramburo (2008) describe a steady-state gas-flow model commonly used in gas network analysis. The model uses a Newton-Raphson (N-R) nodal method to calculate the pressure values at the network nodes and the gas flow rates in the pipelines. Assumptions for the model are:    

Temperature of the gas flow remains constant No speed variations of the gas flow Constant gas density throughout the network Constant friction factor for all pipe lengths.

30/04/2014

Grant No. 608775

52

The nodal pressure is calculated by assuming equivalence to Kirchoff’s current law (KCL) in electricity networks: (15)

( )

Where ( ) is the nodal flow balance vector as a function of pressure , is the loads vector with dimension nodes, is the branch-nodal incidence matrix, and is the flow vector in the node branches with dimension pipes. Equation (15) is a set of non-linear equations that must be solved iteratively to obtain the nodal pressures in the network. Nodes can be one of two types: pressure nodes or load nodes. Pressure nodes act as reference points for the network and have a fixed pressure allowing the gas flow vector to be calculated e.g., at terminals or compressors. Load nodes describe points where the gas pressure needs to be calculated from a known load vector e.g., at offtakes.

Gas flow is calculated based on Weymouth’s equation. The gas flow function [( from node to node is defined as: [(

) ]

(

) ] for any transmission pipe

(16)

)

Where ( ) is the gas flow in pipe , is the flow exponent based on the pressure level of the network, and ⁄ , where is the friction factor for pipe ( is pipe length in metres and is pipe diameter in 3 millimetres). The model assumes a friction factor of 11.7x10 m/mm (Acha & Hernandez-Aramburo 2008).

Modelling of gas transmission networks uses analogies with electricity networks. Table 17 shows the analogue properties and variables.

Table 17: Gas/electricity analogue variables for transmission networks (Acha & Hernandez-Aramburo 2008).

Property Potential Flux Power Power loss Resistance

Gas network Pressure (N/m2) Flow (m3/s) Pressure*Flow (W) ΔPressure*Flow (W) Friction factor (k)

Electricity network Voltage (V) Current (A) Voltage*Current (W) ΔVoltage*Current (W) Impedance (Ω)

Pipe diameters in the UK are: (National Grid (UK) 2009)   

Industry: 600 mm Business: 300 mm Domestic: 20 mm to 180 mm.

30/04/2014

Grant No. 608775

53

5.

Energy storage

Storage of energy or electricity is important for the integration of renewables with the smart grid. The variable and intermittent nature of renewable power generation means that supply and demand can often be out of sync.

To utilise renewably-generated electricity more efficiently it must be storable for utility ‘load levelling’ at later times. The majority of electric vehicles also rely on electricity storage so that the vehicles can transport their own power sources.

Choice of storage option depends on the characteristics in Table 18 (Momoh 2012).

Table 18: Characteristics of energy storage technologies (Momoh 2012)

Storage characteristic Unit size Storage capacity Available capacity Self-discharge time Efficiency Durability or Life-cycle Autonomy

Mass/volume energy densities Response time Cost Reliability Other

5.1

Description Physical size and scale of the storage technology Total energy that can be stored Average power output The time it takes for the storage technology to discharge while not connected to a load Ratio of the energy output (discharge) to the energy input (charge) Number of charge-discharge cycles that the storage technology can undergo while still meeting performance specifications Ratio between energy capacity and the maximum discharge power of the unit.

Used as an indicator to describe the length of time that a storage technology can continuously provide energy The energy storage capacity per unit mass or unit volume of the technology How quickly the stored energy can be made available for use The financial cost to install, maintain and operate the unit The ability for the storage technology to guarantee a level of service Feasibility, required monitoring and control equipment, operational constraints, environmental impact, ease of maintenance, simplicity of design, operational flexibility, response time for energy release

Energy storage technologies

Energy storage technologies applicable to the smart grid are shown in Table 19. A brief description, along with the advantages and disadvantages of each technology is given.

Table 19: Energy storage technologies suitable for the smart grid. Adapted from Momoh (2012), Divya et al. (2009), and EPRI (2012)

Storage technology Flow batteries

30/04/2014

Description Rechargeable battery consisting of two chemical electrolyte components separated by a membrane.

Advantages  Storage capacity limited only by the size of the electrolyte reservoir  Quick response time  No self-discharge

Grant No. 608775

Disadvantages  Complicated in comparison to normal batteries  Poor energy density  High running costs from pumps, controls and 54

Storage technology

Description

Advantages

Advanced batteries

Advanced rechargeable batteries e.g., lithium-ion, polymerion, nickel metal hybrid, sodium sulphur.

Super capacitors (SC)

Double-layer capacitor that uses electrostatic fields to store energy rather than chemicals. Capable of storing much more charge than a standard capacitor.

Cryogenically cooled superconducting coil that stores energy in an electromagnetic field.

Super conducting magnetic energy storage (SMES)

Pumped storage hydroelectric (PSH)

Compressed air energy storage (CAES)

Flywheel energy storage (FES)

30/04/2014

Water is pumped uphill to a reservoir thus storing its potential energy. When the water is released downhill its potential energy is converted into electricity using a turbine and generator. Accounts for 99% of worldwide grid energy storage (EPRI 2012).

Air is compressed and stored in airtight underground vessels or reservoirs. When released, the compressed air expands through a specialised combustion turbine.

A flywheel in a vacuum enclosure is accelerated to high rotational speed.

Energy is removed from the system by slowing down the flywheel.

Disadvantages electrolyte storage

 Good energy density so smaller than lead acid batteries  Slow self-discharge time  Good efficiency  Very high durability  Rapid charging and discharging (low impedance)

 Expensive  Complicated in structure  Deep discharging detrimental to durability

 Very quick response time  High power output  No moving parts  No self-discharge  Very high efficiency  Used in high power applications  Largest capacity grid energy storage technology  Quick response time

 Low energy content  Impractical due to reliance on cryogenics

 Allows storage of offpeak generated electricity  Aid the conservation of natural gas

 Natural gas is burned to reheat the expanding air, causing low efficiency

    

 Self-discharge a problem due to frictional losses

Quick response time Temperature invariant High energy density Durable Low maintenance

Grant No. 608775

 Low energy density  Low voltage  Quick self-discharge time

 Requires water to be stored at elevation, so application is geographically restricted  Potential environmental impact

55

6.

Socioeconomics – dynamic pricing

Meeting peak electricity demand is an increasing problem for utility companies. Times when demand is very high require the operation of ‘peaking’ power plants to meet the increased load. These peaking power plants (which are typically coal or gas-fired) are expensive to run and have a high carbon output. As a result, utility companies are increasingly offering dynamic pricing tariffs to residential, commercial and industrial customers. Dynamic pricing is a mechanism to make the economic cost of electricity for the consumer time dependent. The price of electricity at times of low demand will be lower than at times of high demand. This encourages behavioural changes from the user in order to move electricity demand away from the peak times and reduce peak electricity loads. The simplest exponent of dynamic pricing is to offer lower night-time electricity prices to residential customers to encourage them to run energy-intensive appliances such as dishwashers and washing machines at night when the baseline load is low.

As part of Task 2.1 for INDICATE, it is necessary to characterise and quantify the effects that dynamic pricing has on both the demand and supply sides of the electricity network, so that energy flows can be modelled correctly.

This will include looking at TOU tariffs offered by utility companies, how they affect demand through user behaviour (of residential, commercial and industrial customers), and how this influences the supply side and operation of peaking power plants.

Dynamic pricing is the charging of different energy rates at different times of the day and year to reflect the timevarying cost of supplying energy. Because demand fluctuates based on lifestyle and weather conditions, the energy system typically has to keep spare peaking generation capacity online for times when demand may surge on short notice (see Section 3.1). Often, these peaking power plants are only run for 100–200 hours a year, and the associated cost of maintaining a standby supply adds to the average cost of providing energy.

6.1.

Bulk usage meters

Energy tariffs can be affected by the granularity of usage data that is recorded by the customer’s meter. Mass market customers (residential and non-residential) typically have bulk usage meters with a single data register, which simply accumulates the usage over time. As such, these customers can only be billed for the electricity use according to the following types of pricing:  

Flat rates: all usage charged at a uniform rate during a given period in time (e.g., 30-day billing cycle) Tiered rates: usage charged at a different price based on blocks of usage (e.g., first 500 kWh vs. next 500 kWh) during a given period of time (e.g. 30-day billing cycle).

Such energy rate designs do not convey the variability over time (hour-to-hour, day-to-day, season-to-season) in the cost to produce electricity.

6.2.

Peak load pricing

The concept of moving from time-invariant electricity prices to ‘peak-load‘ pricing, where prices are more closely tied to variations in the marginal cost of generating energy, has been around since the 1970s (Kahn 1970). The 30/04/2014

Grant No. 608775

56

key underlying principle is that more cost effective prices during times of less demand will reduce peak demand and the need to build enough capacity to meet it, and this will lead to an overall increase in economic welfare.

The marginal cost of electricity varies over time because the demand for electricity varies widely, it is uneconomical to store electricity in most applications and the optimal mix of generating capacity to balance supply and demand at all hours includes fluctuating combinations of base load capacity, construction and operational costs. When demand is low, it is cleared with base load capacity with marginal operating costs and as demand rises, generating capacity with higher marginal operating costs are required to balance supply and demand (Joskow & Wolfram 2012). In general, marginal costs are low at night and high during the day, low when temperatures are moderate and potentially very high when temperatures are either extremely high or extremely low, depending on the price of substitute fuel and the attributes of appliance stock in an area.

A number of developments over the last decade have elevated interest and uptake in dynamic pricing. The evolution of spot price changes which can alternate as frequently as every ten minutes reflects changing supply and demand conditions. In addition, technological progress continues to drive down the costs and increase functionality for data acquisition, processing, storage and transmission. Smart meters send real-time consumption data to utilities and make feasible various forms of real-time pricing that tie retail prices to dynamic wholesale prices.

Using smart meters, utilities are now capable of recording energy usage on a much more frequent basis, enabling mass market customers, who previously had bulk usage meters, to be introduced to new types of pricing programs that better reflect differences over time in the cost to produce energy. These time-based rate programs include: 

 





Time-of-use pricing (TOU): typically applies to usage over broad blocks of hours (e.g. on-peak = 6 hours for summer weekday afternoon; off-peak = all other hours in the summer months) where the price for each period is predetermined and constant Real-time pricing (RTP): pricing rates generally apply to usage on an hourly basis, and applicable prices fluctuate accordingly Variable peak pricing (VPP): a hybrid of TOU and RTP where the different periods for pricing are defined in advance (e.g., on-peak = 6 hours for summer weekday afternoon; off-peak = all other hours in the summer months), but the price established for the on-peak period varies by utility and market conditions Critical peak pricing (CPP): when utilities observe or anticipate high wholesale process or power system emergency conditions, they may call critical events during a specified time period (e.g., 3pm to 6pm on a hot summer weekday afternoon). The price for energy during these time periods is substantially raised.

Two variants of this type of rate design exist – one where the time and duration of the price increase are predetermined when events are called, and another where the time and duration of the price increase may vary based on the grid’s need to have loads reduced Critical peak rebates (CPR): When utilities observe or anticipate high wholesale market prices or power system emergency conditions, they may call critical events during pre-specified time periods (e.g., 3pm to 6pm summer weekday afternoons). The price for electricity during these time periods remains the same, but the customer is refunded at a single, predetermined value for any reduction in consumption relative to what the utility deemed the customer was expected to consume.

30/04/2014

Grant No. 608775

57

All of the pricing programs listed above are commonly referred to as dynamic pricing because prices are not known with certainty ahead of time. Each of the dynamic pricing options represents a different combination of risks and rewards for the customer (Figure 23). RTP offers consumers potentially the highest reward compared to traditional flat-rate pricing, but also the highest risk. TOU offers consumers the least potential reward at the lowest risk (Faruqui & Palmer 2011). The relationship between risk and reward is an important one, as it lies at the centre of initiatives to incentivise uptake and use. Collectively, these programs allow customers and utilities to take greater advantage of grid and wholesale market variability and of the capabilities of smart grid customer systems.

Figure 23: Dynamic Pricing Risk and Reward (Faruqui & Palmer 2011)

In practice, the measurable benefits of a reduction in energy use from a supplier point of view will be influenced by load management or shifting options available to them and the selection and application of appropriate price mechanisms to deliver on this. An overview of load shifting options is set out below in Figure 24, alongside Table 20 that describes related mechanisms for delivery.

30/04/2014

Grant No. 608775

58

Figure 24: Energy Supply Load Shifting Options (Breukers & Mourik 2013) Table 20: Forms of Load Management (Breukers & Mourik 2013)

Type of shift

Definition

Related price mechanism

 Peak shaving and clipping both aim at a non-shiftable reduced electricity consumption in critical peak periods (when overall demand is high).  Relevant for households.  Encouraging an increase of energy use during off-peak periods (in order to make the production and supply system more efficient, e.g. to use intermittent renewable generation or to increase cost-effectiveness of certain energy intensive technologies).  Currently more relevant for large energy users but in the future with micro grids and decentralised generation on household level households will also benefit from valley filling.  Regular moving of demand from times of high to times of low demand (resulting in demand increase during off-peak hours and demand decrease during peak hours)  Relevant for households.

CPP, CPR potentially with TOU

Strategic Conservation

 Overall and constant reduction in consumption.

Strategic Load Growth

 Strategic load growth allows for an overall increase of load level because of the installation of automation or additional technologies

In principle, dynamic pricing is not the first option to achieve strategic conservation. Installation of energy efficient appliances and changing of routine behaviour is more suited.

RTP

Peak Clipping

Valley Filling

Load Shifting

30/04/2014

Grant No. 608775

TOU, RTP, and to a lesser extent CPP, CPR

TOU most effective, in combination with CPP, CPR

59

Flexible Load Shape

6.3.

such as EV that will also allow for a more flexible load shape to develop.

The load growth can, just as is the case with valley filling, also take place strategically during specific moments of the day or certain days to match generation by intermittent and renewable sources such as windy or sunny days or moments. For large users and in future with increased integration of EV and renewable also for households. ď&#x201A;ˇ This entails the ability of the demand side to respond to sudden generation changes in real time by providing reserve - e.g. when wind produced electricity is lower due to forecast errors. ď&#x201A;ˇ For households and companies that have reserve capacity (e.g. electric vehicles; decentralized energy generation).

RTP, CPP, CPR

Quantifying the application of dynamic pricing

Within Europe, Directive 2012/27/EU (2012) mandated that an 80% roll-out of smart meters be achieved by 2020 and it is therefore necessary to quantify the potential effects on energy use of this continued roll-out in the context of modelling in the INDCIATE tool.

Where utilised in practice, utility companies routinely offer a mixture of dynamic pricing options to optimise efficiencies in operation. As such, an appreciation of the variability of methodology definition and subsequent application is important. Pricing program development is framed in the context of many different factors including cost neutrality; reflective pricing against different time bands; consideration of system demand peaks; the inclusion of different energy costs; the definition and selection of study tariff groups etc. (Conlon 2008). These and other considerations will be typically shaped by environment-specific attributes. Having regard for this, it is important to note that in Europe, the need for load shifting during a limited set of hours in a year is less significant than in countries with extreme climates or large seasonal swings, and therefore CPP and CPR have not, as yet, been widely applied (Breukers & Mourik 2013). For the purpose of Task 2.1, the principle aim was a standardised approach to quantifying the effects on demand and supply-side aspects.

6.4.

Review of dynamic pricing trials

A review of existing trial studies that report on the impact of price incentives to shift and/or reduce energy consumption across mass market consumers was undertaken. While a significant body of research exists on residential trials, research on non-residential uses is only emerging. 76% of European building floor space is residential buildings (European Commission 2014) and as such, an emphasis was placed on this sector.

6.4.1.

Residential trials

A total of 6 European trials were examined in the first instance, drawing on the findings of Breukers and Mourik (2013). The findings of these European-specific trials were then compared with a more expansive piece of research undertaken by The Brattle Group in the United States, which draws on their international database of dynamic rate experiments.

30/04/2014

Grant No. 608775

60

TOU and CPP mechanisms were deployed in the identified European studies with the key findings set out in Table 21. Table 22 shows the dynamic pricing time periods for each European trial (Breukers & Mourik 2013) Table 21: Summary findings from European residential dynamic pricing trials (Derived from Breukers and Mourik, (2013))

Peak Reduction

Peak to offpeak differential (approx.)3

Trial

Country

Pricing Mechanism

No. of Participants

Overall Consumption Reduction

ESMCBT (2009-2010) EDRP EDF Project (2007-2010)

Northern Ireland

TOU

2,920

2.5%

8.8% - 11.3%

143 – 271%

165%

France

TOU

194

2.3 – 4%

4% (weekdays) to 8% (weekends)

-

4%

126 - 286%

None

1-2%

267%

TEMPO EDF (1989-1992)

France

TOU & CPP

800 in 1989 to 300,000 after 2004

PowerShift (2003-2004)

Northern Ireland

TOU

100

Elforsk Pilot (2003-2005)

Sweden

CPP

93

Intellekon (2008)

Germany

TOU

2000 (part TOU only)

-

9.7%

Load cut to an average of at least 50% during high periods 2%

-

250%

Table 22: Timing of dynamic pricing tariffs for the trials from Breukers and Mourik (2013)

Dynamic pricing trial

ESMCBT (2009-2010)

EDRP EDF Project (2007-2010)

TEMPO EDF (1989-1992)

   

Time of use (TOU) periods Night time (11pm to 8am) Day time (8am to 5pm and 7pm to 11pm) Peak times (5pm to 7pm) A weekend tariff (varying rates for same time periods).

  

Peak period (4.30pm to 7.30pm) Night period (11pm to 6am) Off-peak period (6am to 4.30pm and 7.30pm to 11pm).

 

Peak hours (6am to 10pm) Off-peak hours (midnight to 6pm and 10pm to midnight).

3

This is the approximate percentage price difference in energy between peak and off-peak periods in the trial. A range is provided where different study price periods applied in a trial.

30/04/2014

Grant No. 608775

61

PowerShift (2003-2004)

Elforsk Pilot (2003-2005) Intellekon (2008)

Weekdays:  Midnight to 8am – Green (low)  8am to 4pm – Amber (medium)  4pm to 7pm – Red (high)  7pm to Midnight – Amber (medium) Saturdays and Sundays:  Midnight to 8am – Green (low)  8am to 7pm – Amber (medium)  7pm to Midnight – Green (low) Critical Peak Pricing with annual maximum of 40 hours at the peak price.  Peak (10am-6pm)  Off-peak (6pm-10am).

Demand-Side Findings At a broad level, the findings indicate that consumers understand and respond to critical-peak pricing programs. A review of the 6 identified trials indicates a reduction in overall consumption patterns of between 0 to 10%.

Consideration on the variance in these figures can be attributed to the composition and level of reach of the trials themselves and the predominance of TOU price mechanisms within this, including associated risk/reward relationship factors. Two of the six trials did not measure consumption reduction directly and a fuller picture of demand-side impacts was garnered from a comparison with summary outputs on international research undertaken in 2011. The Brattle Group’s assessment of 109 tests of time varied rates from 24 pilots in North American, Australia and Europe, indicated that customers reduced peak load on dynamic rates relative to flat rates, with a median peak reduction (or demand response) of 12%. Almost 30 tests produced results in the range of 10–15%, and many more exhibited larger responses (Faruqui & Palmer 2011).

Supply-Side Findings TOU is not aimed at reducing overall energy demand, merely at shifting the demand from one period to another and the reviewed European TOU-based trials realised peak reductions between 1 to 11%. Faruqui and Palmer’s research (2011) confirms that the magnitude of customer’s response varies with the price incentive. The peak reduction of 50% exhibited in the Elsforsk CPP pilot, relative to that recorded in the other European pilots, suggests there is validity in this judgement. The recorded peak to off-peak values highlight the elasticity of substitution in the identified trials, with related benefits for required supply. Again, variances in recorded values are influenced by trial-specific factors such as weather, consumer attitudes, the specifics of the rate design (number of pricing periods and their timing and duration), and the manner in which the rates were marketed.

6.4.2.

Non-residential trials

The ability of non-residential energy users to take advantage of dynamic pricing tariffs depends on the nature of their business and the ability to shift electrical loads to a period where the tariffs are more favourable. For example, commercial property users operate between fixed hours (e.g., 8am to 6pm) and have less flexibility to shift loads from a higher tariff structure. In contrast, industrial energy users who operate 24 hours per day may have the flexible means to tailor productivity over different time periods. Findings from a New York 30/04/2014

Grant No. 608775

62

demonstration project confirmed that price and demand response opportunities for commercial and industrial (C&I) businesses were unique to their electric load characteristics, control capabilities and operational constraints (Kim et al. 2013).

The findings of non-residential trials are generally mixed. Based on their research of Californiaâ&#x20AC;&#x2122;s Statewide Pricing Pilot, Faruqui and George (2005) highlighted that small to medium C&I customers conclusively reduced peakperiod energy use in response to time-varying prices. The research also illustrated that the response varied in relation to the level of energy demand, as linked to business type. CPP and TOU tariffs were tested for C&I customers with billing demands under 200 kW, which were segregated into two groups: (1) peak demands less than 20 kW (LT20) and (2) peak demands between 20 and 200 kW (GT20). The findings showed that LT20 customers exhibited an average peak-period4 reduction on critical5 weekdays of 6% (1.5% on normal weekdays), comparable to an average peak-period reduction of 9.1% for GT20 customers on critical weekdays (2.4% on normal weekdays).

An assessment of mandatory TOU pricing in Connecticut (Jessoe & Rapson 2011) found that despite a significant shift in marginal prices, C&I customers did not exhibit reductions in peak load or overall usage. The study concluded that C&I customers were either more price inelastic or that the pricing regime in the trial was not effective at transmitting meaningful economic incentives to customers.

There are limited examples of non-residential trials in Europe and there is currently a new emerging field of research in this area. For example, UK Power Networks, and EDF Energy are currently conducting trials in this area.

As with the residential trial findings, the approach of supply-side utility companies will depend on load management or shifting options available to them. Final tariffs will reflect demand in different time periods, with the aim of incentivising shifting load from peak/costlier periods.

6.5.

Dynamic pricing and utility customer behaviour

Different end-users are likely to have different attitudes, motivations, behaviours, knowledge and other resources available to them, which effect how they respond to and participate in dynamic pricing interventions. There is strong consensus that supporting technology is key to the success of dynamic pricing (Breukers & Mourik 2013).

Such technology typically includes a wide range of support or feedback devices including smart meters, In-HouseDisplays (energy dashboards) and web portals. This is supported by empirical evidence. Faruqui et al. (2010) demonstrated that multiple pricing pilots in the US with a combination of TOU, CPP, CPR and multiple enabling technologies generated the highest peak clipping and load shifting.

4

Defined as between 2pm and 7pm for the purpose of the trial.

5

On a small number of identified â&#x20AC;&#x2DC;critical daysâ&#x20AC;&#x2122;, the peak period price was modelled at roughly five times higher than the standard rate and six times higher than the off-peak rate.

30/04/2014

Grant No. 608775

63

Interestingly, culture plays an important part in consumer reaction to programs. Aggregated results from trials by Stromback et al. (2011) highlight that European participants are among the most responsive to smart meter enabled programs, responding better than residents in the USA and Canada Figure 25.

Figure 25: Recorded responses to smart metering programmes (Stromback et al., 2011)

A crucial precondition for effective use of smart meters is that the end-user understands the technology and its benefits, and educating customers is important to success. Technological improvements, including more cost effective advances in support systems are likely to result in utility companies seeing the net benefits of wide scale deployment of the advanced metering infrastructure.

Different viewpoints exist on whether low income earners are more likely to be negatively impacted by dynamic pricing, however research shows that they are as equally price-responsive as average customers to dynamic rates (Faruqui & Palmer 2011).

30/04/2014

Grant No. 608775

64

7.

Network theory and cities

Multiple networks can be found within a city e.g., the energy distribution network, the water network, telecoms networks, traffic networks etc. This section reviews network theory concepts and how those concepts can be used to characterise city networks. First, complex network theory is described. Second, the concepts used to define urban networks are discussed. Then the theory of the multiple central assessment (MCA) with its three phases: establishing centrality analysis for cities, expanding the scope from centrality to network analysis, and increasing scale of analysis up to entire cities, are summarised. The application examples for the three phases of the MCA method are finally presented.

7.1.

Complex network theory

The following is a summary of general terms used to describe complex networks.

7.1.1. Graphs, nodes, edges, and degrees A network (or graph in complex network theory) in its most basic form comprises of a collection of nodes (or vertices) and edges. The nodes are the individual points in a network, and the edges are the connections between pairs of nodes. For example, a network could be a list of cities (nodes) and the airline flight paths (edges) between the cities. The degree refers to the number of edges connected to a node.

In functional form, a network

can be represented as: (

)

(17)

Where is a non-empty set of nodes, and is a set of unordered edges (pairs of nodes) (Porta et al. 2006). In this report, a node will be referred to by its order in the set , with . If there is an edge between nodes and , the edge being indicated as ( ), the two nodes are said to be adjacent or connected.

There may be more information available than simply the nodes and edges; for example, depending on the network the nodes may have characteristics such as size, height or colour, and the edges may represent length, age or processing power (Estrada et al. 2010).

7.1.2. Connectivity, adjacency and direction Two nodes of a graph are said to be connected if there exists a pathway between them. If every pair of nodes of a graph is connected by some pathway then the graph is said to be connected. If at least one node of the graph is not connected to another node, then the graph is said to be disconnected. The gas distribution grid would be considered a connected network; whereas a small, rural community that relies on deliveries of oil to provide energy could be considered a disconnected network.

The shortest path between two nodes is called the geodesic. If two nodes are linked by one edge then the nodes are adjacent.

30/04/2014

Grant No. 608775

65

A graph can be defined using an adjacency matrix [ ] where if nodes and are connected and if they are not connected (Estrada et al. 2010). Figure 26 shows a graph of seven nodes (a) and its adjacency matrix (b), with rows and columns .

6

3

2 5 4 7

[

(a)

1

] (b)

Figure 26: Example of a graph (a) and its adjacency matrix (b).

A directed graph is one in which the edges have an orientation e.g., the flow of traffic down a one-way street.

Undirected graphs have no edge orientation e.g., two-way traffic.

7.1.3. Weighted graphs Sometimes it is useful to consider a weighted or valued graph: (

)

(18)

Where is a set of elements (or weights) assigned to each edge. Depending on the network the weights can represent characteristics such as financial cost or physical distance.

A weighted graph can be described using two matrices, the adjacency matrix , and a matrix containing the edge weights. In the particular case of a spatial (or geographic) graph (a graph with nodes that have a precise position in a two-dimensional or three-dimensional Euclidean space and with links that are real, physical connections) it is useful to work with lengths instead of weights. The length matrix is an matrix where is the metric length of the link connecting and (a quantity inversely proportional to the weight associated with the edge) (Porta et al. 2006).

In a weighted graph the shortest path length between and is defined as the smallest sum of the edge lengths throughout all the possible paths in the graph from to . In a non-weighted graph the shortest path length is simply the smallest number of steps required to go from to .

7.1.4. Characteristic path length The characteristic path length is defined as the average length of the shortest paths (with the average being calculated over all the nodes pairs in the network): (Watts & Strogatz 1998)

30/04/2014

Grant No. 608775

66

L

1  d ij N  N  1 i , jN ;i  j

(19)

The characteristic path length is a metric for the connectivity of a network.

7.1.5. Global efficiency As the characteristic path length is not well defined for non-connected graphs, unless an artificial assumption of a finite value for is made also when there is no path connecting nodes and (Porta et al. 2006), a new index, the global efficiency has been defined by Latora and Marchiori (2001). As with the characteristic path length , is a measure of how well the nodes communicate over the network, and it is based on the assumption that the efficiency of the communication between two generic nodes and of the graph is ⁄ . In the case where is inversely proportional to the shortest path length connecting the nodes i.e., unconnected and there is no path linking and , and . The global efficiency of graph is defined as the average of over all pairs of nodes:

E glob G  

1 1 1  ij    N  N  1 i , jN ;i  j N  N  1 i , jN ;i  j d ij

(20)

The global efficiency is correlated to , with a high characteristic path length corresponding to a low global efficiency (Latora & Marchiori 2003). By definition, in the topological (non-weighted graph) case, takes values on the interval [0, 1], and is equal to 1 for the complete graph (a graph with all the possible ( ) edges).

In metric systems (translated into valued graphs), it is possible to normalize such a quantity by dividing ( ) by the efficiency ( ) of an ideal complete system in which the edge connecting the generic couple of nodes , is present and has a length equal to the Euclidean distance between and : (Latora & Marchiori 2001; Latora & Marchiori 2003)

E glob Gideal  

1 1  N  N  1 i  jN d ijEucl

(21)

Eucl where d ij is the Euclidean distance between nodes and along a straight line (the length of a virtual direction

connection

). Hence the normalisation is:

E1glob G  

E glob G l  E glob Gideal 

(22)

A different normalisation proposed by (Vragović et al. 2005) is expressed as: Eucl

glob

E2 30/04/2014

d ij 1   N  N  1 i , jN ;i  j d ij Grant No. 608775

(23) 67

The global efficiency measures the mean flow-rate of information over a graph and has the property of being finite even for unconnected graphs (Porta et al. 2010).

7.2.

Types of network

Many different types of network exist, but the following are relevant to characterise urban networks found within cities.

7.2.1. Primal and dual networks Primal urban networks are represented by a two-dimensional Euclidean space containing zero-dimensional geographic entities such as intersections (nodes), and one-dimensional geographic entities such as streets (edges). This representation is a called a ‘direct approach’ because of the coherence between the dimensions of geographic entities and graph entities (Porta et al. 2006). Dual urban networks represent streets as nodes and intersections as edges. This representation is called an ‘indirect approach’.

Network analysis applied to territorial cases has mostly followed a direct approach. Primal networks are used because their use is more intuitive for systems in which distance has to be measured, not just in spatial terms (e.g., metres) such as in urban street systems, but also in topological terms (e.g., degrees of separation) such as in social systems. Traffic engineers, economic geographers and even geoarcheologists generally follow the primal approach.

7.2.2. Small-world networks A small-world network is a network where the average distance between pairs of nodes is small compared to the overall size of the network (Watts & Strogatz 1998). While most nodes are not adjacent, one node may be reached from another by only a few steps. The typical path length increases logarithmically with the number of nodes (Porta et al. 2006). Most economic, social, natural, and man-made networks exhibit small-world network properties (Porta et al. 2006).

7.2.3. Scale-free networks Scale-free networks are networks that are characterised by the presence of hub, (nodes with a number of connections (degrees) much larger than the average number of connections for all the nodes of the network ). Empirical evidence collected from the analysis of both natural and manmade networks from the real world have shown that for scale-free networks the number of links originating from any one node follows a power law distribution ( ) (Wolfram Research Inc. 2014). The exponent generally varies between 2 and 3. Most real-world networks are scale-free networks (Porta et al. 2006).

Most of the nodes on the network have a small number of links, while the less numerous hubs have an extremely large number of connections. This has extremely important consequences on the resilience of scale-free networks to errors and attacks. The emergence of scaling in a complex network has been recognized as a sign that the system is not static, but rather subject to incremental growth through time and preferential attachment (Strogatz 2001).

30/04/2014

Grant No. 608775

68

7.3.

Centrality

Centrality describes the importance of a node within a network. An application of centrality to an urban network would be how well-used a road is within a city, or how many people visit a particular building. Centrality affects how a city network functions and plays an important role in shaping its growth. The natural movement of people drives the evolution of the topological and functional structure of the city and contributes to the formation of the street network and urban spaces (Hiller 2007).

Central points (i.e., city centres) can be defined as the point of intersection of main routes, where some special configuration of terrain or some particular layout of the river system makes the place compulsory to pass through (Mehaffy et al. 2010). A central point in an urban network is easily accessible from both the immediate surroundings and more distant points, has the potential to sustain higher densities of retail and services, and is a key factor for supporting the formation and vitality of urban nodes (Newman & Kenworthy 1999). The location for urban landmarks (museums, theatres, office headquarters, service and retail) are preferentially close to central points.

There are several different types of centrality, of which the ones applicable to urban networks will be described below.

7.3.1. Degree centrality Degree centrality is based on the idea that important nodes have the largest number of edges with other nodes in a graph. The degree of a node is the number of edges connected to the node (Diestel 2006). The degree of node is defined in terms of the adjacency matrix as k i 

a

ij

. The degree centrality is defined as:

jN

C

D i

 aij ki jN   N 1 N 1

(24)

7.3.2. Closeness centrality Closeness centrality is based on the concept of minimum distance or geodesic as the smallest sum of the edge lengths throughout all the possible paths in the graph from to in a weighted graph, or the minimum number of edges traversed in a topological graph (Bavelas 1950; Sabidussi 1966). Closeness centrality is used because degree centrality is not relevant in primal representations, in which a node's degree (the number of streets incident to an intersection) is substantially limited by spatial constraints (Porta et al. 2006). The closeness centrality of point is defined as:

C iC  Li 1 

N 1  d ij

(25)

jN ; j  i

Where is the farness, or the sum of the distances from node to all other nodes. The closeness is then then the inverse of the farness. Closeness is meaningful for connected graphs only, unless one artificially assumes to 30/04/2014

Grant No. 608775

69

be equal to a finite value when there is no path between two nodes and , and to take on values between 0 and 1 in the case of non-valued graphs.

7.3.3. Betweenness centrality Betweenness centrality is a measure of the number of times a node acts as a bridge along the geodesic between two other nodes (Freeman 1977). The interactions between two nonadjacent nodes might depend on intermediate nodes that can have a strategic control or influence on them. Assuming that communication travels along only geodesics, the betweenness centrality of node is defined as:

C iB  Where

1



N  1N  2  j ,kN ; j  k ; j ,k i

nij (i ) nij

is the number of geodesics linking the two nodes and , and

the two nodes and that contain node . The index maximum when node falls on all geodesics.

(26)

( ) is the number of geodesics linking

takes on values between 0 and 1 and reaches its

The concept of betweenness can also be extended to edges. The edge betweenness centrality, , is based on the same idea that an edge is central if it is included in many of the geodesics connecting pairs of nodes. The betweenness centrality of edge , ..., is defined as:

CB 

1

 N  1 N  2  

i , j 1,, N ;i  j

Where is the number of geodesics between nodes between nodes and that contain edge .

nij ( )

(27)

nij

and , and

( ) is the number of shortest paths

7.3.4. Efficiency centrality Efficiency centrality originates from the idea that the efficiency of the communication between two nodes and is equal to the inverse of the shortest path length (Latora & Marchiori 2001). Thus, the efficiency centrality of node is defined as:

 1 CiE     jN ; j  i d ij 

 1    jN ; j  i d Eucl ij 

   

(28)

7.3.5. Straightness centrality Straightness centrality compares how the path between two nodes on a graph deviates from a straight line (Crucitti et al. 2006). Straightness is a variant of efficiency centrality and originates from the same idea of efficiency centrality with a different normalization (Vragović et al. 2005). The straightness of node is defined as:

30/04/2014

Grant No. 608775

70

d ijEucl 1  C   N  1  jN ; j  i d ij S i

   

(29)

7.3.6. Information centrality Information centrality describes the impact on network efficiency from removing individual nodes (Stephenson & Zelen 1989). Formally, information centrality is defined as the relative drop in the network efficiency caused by the removal from of the edges incident on :

C iI  Where

indicates the network with

on node . The term

E2glob

E 2glob E 2glob G   E 2glob G '  E 2glob E 2glob G 

points and

edges obtained by removing from

(30)

the edges incident

is the global efficiency defined in equation (23), however, a generic performance

parameter can be used. The removal of node edges affects communication between nodes on the graph, thereby

increasing the geodesic lengths. Consequently, the efficiency of the new graph E2glob G ' is lower than that of

E 2glob G  . Information centrality is correlated to the other three centrality metrics degree, closeness and

betweenness centrality.

The edge information centrality, , is a measure relating the importance of the edge to the ability of the network to respond to the deactivation of the edge itself. The network performance, before and after a certain edge is deactivated, is measured by the efficiency of the graph from Equations (21 to 23). The information centrality of edge is defined as the relative drop in the network efficiency caused by the removal from of the edge :

CI 

E E (G )  E (G ' )  E E (G )

(31)

where ( ) is the efficiency of the graph .

7.4.

City network characterisation

Characterising city network using complex network theory can be achieved on different scales. Below will be discussed small, urban areas and the city as a whole, using case studies from the literature as examples.

7.4.1. Characterisation of small, urban areas Crucitti et al. (2006) and Porta et al. (2006) demonstrated the systematic evaluation of different centrality metric distributions over a number of one-square-mile samples of urban patterns in a primal geographic framework.

Figure 27a and Figure 27b show the original maps of two cities: Ahmedabad, India and Richmond, California, USA, respectively. The corresponding road-centreline-between-nodes graphs are shown in Figure 27c and Figure 27d.

The primal graphs in Figure 27c and Figure 27d were constructed using the following rules:

30/04/2014

Grant No. 608775

71

   

Real intersections are turned into graph nodes and real streets are turned into graph edges All graph edges are defined by two nodes (the endpoints of the arc) Intersections between edges are always located at nodes Edges follow the footprint of real streets as they appear on the source map; all distances are calculated metrically.

After the computation of centrality scores for the primal nodes, analogous primal layout maps (red-and-blue maps) were produced.

The two urban patterns are strikingly different. Ahmedabad exhibits a typical, self-organized pattern that spontaneously emerged from historical processes independent of any central coordination. Richmond shows a predominantly planned pattern, following one coordinated layout that has developed in a relatively short period of time. Ahmedabad is a densely interwoven, uninterrupted urban fabric, whereas Richmond shows a traditional gridiron structure.

(a)

(b)

(c)

(d)

Figure 27: One-square mile urban patterns; original maps of (a) Ahmedabad, India and (b) Richmond, USA, and primal road-centrelinebetween-nodes graphs of (c) Ahmedabad and (d) Richmond (Porta et al. 2006).

30/04/2014

Grant No. 608775

72

The closeness centrality of Ahmedabad calculated using a primal approach is shown in Figure 28. Node centrality scores are calculated for the primal weighted graph, where edge lengths are used for the weighting factors (Porta et al. 2006). Figure 28a shows the node closeness , calculated for the whole network. Figure 28b shows the local closeness where is calculated for a sub-network of nodes at distance metres from each node.

Figure 28c shows the local closeness where is calculated for a sub-network of nodes at distance metres from each node.

For primal graphs, the closeness centrality is dominated by the so-called ‘border effect’, i.e., higher closeness scores consistently group around the geometric centre of the image. To some extent less evident in less dense cases, the border effect is overwhelming in denser urban fabrics such as those of Ahmedabad. However, in all cases the border effect affects the spatial flow of enough to prevent the emergence both of central routes and of focal spots in the city fabric (Porta et al. 2006).

C

Figure 28: Closeness centrality (C ) in Ahmedabad, India using a primal approach showing (a) global closeness, (b) local closeness (d < 400 m), and (c) local closeness (d < 200 m) where d is the distance between nodes (Porta et al. 2006). The colour red indicates a high level of closeness, whereas blue indicates a low level of closeness.

7.4.2. Characterisation of entire cities Centrality metrics only describe sub-properties of networks and are strictly relative to individual networks. A wider set of metrics are needed to characterise the networks themselves to allow normalisation and comparison between different networks (Porta et al. 2010). Comparisons between networks can be made using the Minimum Spanning Tree (MST) graph and the Greedy Triangulation (GT) graph. If a real graph and the positions of its nodes in a two-dimensional plane are given, the MST graph is the planar graph (a graph where none of the edges cross over each other) with the minimum number of edges in order to ensure connectedness, while the GT graph is the graph with the maximum number of non-planar edges.

In short, MST and GT graphs represent the planar graphs with the lowest and the highest possible cost, respectively. MST and GT will serve as the two extreme cases to normalise the values of the structural measures to be computed, namely efficiency and cost. The cost is defined as the sum of the length of edges in a network:

W 

a

l

ij ij

(32)

i, j

30/04/2014

Grant No. 608775

73

The graph efficiency is the global efficiency defined in Equations (21 to 23).

The relative cost

and relative efficiency

are defined as:

w  W MST W GT  W MST E  E MST  GT E  E MST

Wrel  E rel

(33) (34)

By definition, the MST has a relative cost and , while GT has and . In general, the counterpart of an increase in efficiency is an increase in the cost of construction, i.e., an increase in the number and length of edges.

The ‘backbone’ of a city can be extracted by deriving spanning trees based on edge betweenness and edge information (Scellato et al. 2006) as defined above. The obtained trees are radically different from those based on edge lengths, and allow an extended comprehension of the most important routes that affect pedestrian/vehicular flows, retail commerce vitality, land-use separation, urban crime and dynamic collective behaviours.

The Maximum Centrality Spanning Trees (MCSTs) are defined as the maximum weight spanning trees where the edge weight is defined as the centrality of the edge. A graph ( ) is a tree if and only if it satisfies the following four conditions:    

has ( ) edges and no cycles has ( ) edges and is connected Exactly one simple path connects each pair of nodes in is connected, but removing any edge disconnects it.

Given a connected, undirected graph ( all the nodes together. Consequently, trees.

), a spanning tree is a subgraph of which is a tree and connects ( ). A single graph can have many different spanning

Using the concept of centrality and the definitions of Minimum Spanning Tree (MST) and Greedy Triangulation (GT), an interesting characterisation of different city patterns can be obtained by plotting relative efficiency ( ) as a function of relative cost ( ) (Cardillo et al. 2006) as shown in Figure 29. Cardillo et al. classified the topological patterns exhibited by several cities:       30/04/2014

Medieval patterns, including both Arabic (Ahmedabad and Cairo) and European (Bologna, London, Venice, and Vienna) Grid-iron patterns (Barcelona, Los Angeles, New York, Richmond, Savannah, and San Francisco) Modernist patterns (Brasilia and Irvine 1) Baroque patterns (New Delhi and Washington) Mixed patterns (Paris and Seoul) Lollipop layouts (Irvine 2 and Walnut Creek).

Grant No. 608775

74

In Figure 29, each point on the plot represents a city. The point of coordinates (0, 0) would correspond to the cost/efficiency of the MST while the point (1, 1) would correspond to the GT network. Irvine 2, having coordinates (0.175, â&#x2C6;&#x2019;0.398), i.e., a negative value of relative efficiency, has been plotted instead as having coordinates (0.175, 0).

Figure 29: Relative efficiency, Erel versus relative cost, Wrel indicating a correlation between structural properties and a priori known classes of cities as medieval, grid-iron, modernist, baroque, mixed, and lollipop fabrics (Cardillo et al. 2006).

Figure 29 shows a certain capacity to characterize the different classes of cities listed above. The plot indicates an overall increasing behaviour of as a function of , with a saturation at ~ 0.8 for values of > 0.30.

Gridiron patterns exhibit a high value of relative efficiency, about 70 to 80% of the efficiency of the GT, with a relative cost from 0.24 to 0.40. The three gridiron cities (New York, Savannah and San Francisco) with the largest value of efficiency, ~ 0.80, have respectively a relative cost equal to 0.34, 0.35, and 0.38. Medieval patterns have in general a lower relative cost and efficiency than gridiron patterns although, in some cases like Ahmedabad and Cairo (the two Arabic medieval cities with the highest relative efficiency), they can also reach a value of ~ 0.80 with a smaller cost equal to 0.29. Modernist and lollipop layouts are those with the lowest relative cost but also the lowest relative efficiency.

Scellato et al. (2006) compared MCSTs for betweenness and information centrality with the minimum length spanning trees (mLSTs) for the cities of Bologna and San Francisco (Figure 30). An mLST is a spanning tree with weight (cost) smaller than or equal to the weight of every other spanning tree of the graph. The weight associated with each edge is set to be equal to the length of the edge and represent the cost of the edge. In the case of Bologna, the MCST based on betweenness and information centrality have 82% and 75% of the edges in common with the mLST respectively. In San Francisco the MCSTs for betweenness and information have 70% and 76% of the edges in common with the mLST. It is worth noting that the two MCSTs have 77% of the edges in common in Bologna, whereas for San Francisco it is 66%.

30/04/2014

Grant No. 608775

75

Figure 30: Spanning trees of Bologna (above) and San Francisco (below). From left to right, mLSTs, betweenness-based MCSTs, and information-based MCSTs (Scellato et al. 2006).

The graphical visualization of the MCST is of interest for urban planners since the trees express the uninterrupted chain of urban spaces that serve the whole system while maximizing centrality over all edges involved. This method identifies the backbone of a city system as the sub-network of spaces that are most likely to offer the highest potential for the standard of life of the urban community in terms of popularity, safety and services locations. This is evident in Figure 30, where the comparison between the trees in the two cities clearly indicates that the spatial sub-system that keeps a city together in terms of the shortest trip length is not the same spatial sub-system that does it in terms of the highest centrality.

It is also worth noting that metric distance is also involved in the algorithms for the calculation of the centrality metrics, so that different kinds of trees considered hereby are rooted in the geographic space. The second thing is that while the shortest length backbone performs effectively when applied to planned urban fabrics like San Francisco, in self-organized evolutionary cases like that of Bologna it does not find continuous routes nor clearly distinguishes a hierarchy of sub-systems in the network, while the highest information and especially the highest betweenness backbones do. The organic patterns are therefore said to be more oriented to put things and people together in public space than to shorten the trips from any origin to any destination in the system, this latter character being more typical of planned cities.

Recently, Strano et al. (2013) compared the structural properties of the street networks of ten different European cities using their primal representation. The properties of the geometry of the networks and a set of centrality measures highlighting differences and similarities between cases were investigated. The total length of tree-like

30/04/2014

Grant No. 608775

76

(dead ends) links was related as a function of the cycle area for the considered cities as shown in Figure 31 where the distributions show power-law behaviour with an exponent of approximately 0.8.

Figure 31: Total length of tree-like (dead ends) links as a function of the cycle area for the considered cities (Strano et al. 2013).

30/04/2014

Grant No. 608775

77

8.

Summary

Component subsystems within the city that are important to model in order to capture its energy performance were identified as the: i. ii. iii. iv. v.

Energy sources (centralised, distributed and renewable energy generators), Energy sinks (buildings, electric transport networks, electric public services), Energy networks (electricity and gas networks), Energy storage (bulk grid storage and localise), and the impact of Socioeconomics (dynamic pricing).

Each of the subsystems was characterised within the framework of energy modelling, to form the basis of the complex system model in Task 2.2 and aid the development of the 3D dynamic simulation model in Task 4.3.

Network analysis applicable to city networks was described and examples of applications were given.

30/04/2014

Grant No. 608775

78

References Acha, S. & Hernandez-Aramburo, C., 2008. Integrated modelling of gas and electricity distribution networks with a high penetration of embedded generation. In CIRED Seminar 2008: SmartGrids for Distribution. Frankfurt, Gernmany: IEE, pp. 28–28.

Albring, W., 1967. Introduction to the Theory of Flow Machines. ZAMM - Zeitschrift für Angewandte Mathematik und Mechanik, 47(2).

Andersen, R.V., Olesen, B.W. & Toftum, J., 2007. Simulation of the Effects of Occupant Behaviour on Indoor Climate and Energy Consumption. In Proceedings of Clima 2007: 9th REHVA world congress: WellBeing Indoors, Helsinki, Finland. Helsinki, Finland.

ASHRAE, 2004. ASHRAE Handbook- Fundamentals SI., Atlanta GA: American Society of Heating, Refrigeration and Air Conditioning Engineers.

Athienitis, A.K., 1989. A computer method for systematic sensitivity analysis of building thermal networks.

Building and Environment, 24, pp.163–168.

Awbi, H.B., 2003. Ventilation of Buildings 2nd ed., London, UK: Spon Press.

Bavelas, A., 1950. Communication Patterns in Task-Oriented Groups. The Journal of the Acoustical Society of America, 22(6), p.725.

Biomass Energy Centre, 2014.

Biomass http://www.biomassenergycentre.org.uk/.

Energy

Centre.

Website.

Available

at:

Blumsack, S., 2014. Basic economics of power generation, transmission and distribution. Penn State e-Education.

Available at: https://www.e-education.psu.edu/eme801/node/530.

Bowles, M., 2008. State electricity profiles 2008. US Energy Information Administration. Available at: http://www.eia.gov/electricity/state/.

Boyle, G., 2004. Renewable Energy: Power for a sustainable future 2nd Editio., Oxford, UK: Oxford University Press.

BP, 2013. Statistical Review of World Energy (June 2013).

BPIE, 2011. Europe’s buildings under the microscope - A country-by-country review of the energy performance of buildings. , p.132.

Brenna, M. et al., 2012. Challenges in energy systems for the smart-cities of the future. In 2012 IEEE International Energy Conference and Exhibition (ENERGYCON). IEEE, pp. 755–762.

Breukers, S.C. & Mourik, R.M., 2013. The end-users as starting point for designing dynamic pricing approaches to change household energy consumption behaviours. , (March), p.94.

Bruhns, H., Jones, P. & Cohen, R., 2011. CIBSE Review of energy benchmarks for display energy certificates Analysis of DEC results to date. , (May).

30/04/2014

Grant No. 608775

79

Buchberg, H., 1971. Sensitivity of room thermal response to inside radiation exchange and surface conductance.

Building Science, 6(3), pp.133–149.

Buchberg, H., 1969. Sensitivity of the thermal response of buildings to perturbations in the climate. Building Science, 4(1), pp.43–61.

Cammarata, G., Fichera, A. & Marletta, L., 1993. Sensitivity analysis for room thermal response. International Journal of Energy Research, 17(8), pp.709–718.

Cardillo, A. et al., 2006. Structural properties of planar graphs of urban street patterns. Physical Review E, 73(6), p.066107.

Chiodi, A. et al., 2011. Modelling Electricity Generation - Comparing Results: From a Power Systems Model and an Energy Systems Model. In 30th International Energy Workshop. Palo Alto, California: Stanford University.

COM, 2006. Action plan for energy efficiency: realising the potential. , p.25.

Conlon, P., 2008. Development of Domestic and SME Time of Use Structures for a Smart Metering Program in Ireland.

Connolly, D. et al., 2010. Modelling the existing Irish energy-system to identify future energy costs and the maximum wind penetration feasible. Energy, 35(5), pp.2164–2173.

Cordon, G.C., 1992. Input-output sensitivity of building energy simulations. ASHRAE Transactions, 98(1), pp.618– 626.

Crucitti, P., Latora, V. & Porta, S., 2006. Centrality measures in spatial networks of urban streets. Physical Review E, 73(3), p.036125.

Deaves, D.M. & Harris, R.I., 1978. A Mathematical Model of the Structure of Strong Winds, CIRIA. DeST Group, 2011. DeST. DESTATIS, 2012. DESTATIS. Publikationen im Bereich Bautätigkeit, Wohnungsbau. Available at: https://www.destatis.de/DE/Publikationen/Thematisch/Bauen/BautaetigkeitWohnungsbau/Bautaetigkeit.ht ml.

Diestel, R., 2006. Graph Theory (Graduate Texts in Mathematics), Divya, K.C. & Østergaard, J., 2009. Battery energy storage technology for power systems—An overview. Electric Power Systems Research, 79(4), pp.511–520.

Dixon, J., 2010. Energy storage for electric vehicles. In 2010 IEEE International Conference on Industrial Technology. IEEE, pp. 20–26.

Duffie, J.A. & Beckman, W.A., 2013. Solar Engineering of Thermal Processes 4th ed., John Wiley & Sons.

EirGrid, 2013. EirGrid Transmission System. Available at: http://www.eirgrid.com/media/All-Island Transmission Map.pdf.

30/04/2014

Grant No. 608775

80

Eltawil, M. a. & Zhao, Z., 2010. Grid-connected photovoltaic power systems: Technical and potential problems—A review. Renewable and Sustainable Energy Reviews, 14(1), pp.112–129.

EPRI, 2012. Bulk Energy Storage Impact and Value Analysis.

Estrada, E. et al., 2010. Network Science E. Estrada et al., eds., London: Springer London.

European Commission, 2012. Broadband coverage in Europe in 2012. , p.212.

European Commission, 2013. EU Energy in Figures: Statistical Pocketbook, Luxembourg: Publications Office of the European Union.

European Commission, 2014.

Eurostat. http://epp.eurostat.ec.europa.eu/portal/page/portal/eurostat/home/.

Available

at:

European Parliament, 2012. Directive 2013/27/EU, European Union.

Fang, X. et al., 2012. Smart Grid — The New and Improved Power Grid: A Survey. IEEE Communications Surveys & Tutorials, 14(4), pp.944–980.

Farhangi, H., 2010. The path of the smart grid. IEEE Power and Energy Magazine, 8(1), pp.18–28.

Faruqui, A. & George, S.S., 2005. Quantifying Customer Response to Dynamic Pricing. The Electricity Journal, 18(4), pp.53–63.

Faruqui, A., Harris, D. & Hledik, R., 2010. Unlocking the €53 billion savings from smart meters in the EU: How increasing the adoption of dynamic tariffs could make or break the EU’s smart grid investment. Energy Policy, 38(10), pp.6222–6231.

Faruqui, A. & Palmer, J., 2011. Dynamic pricing and its Discontents. Regulation, 34.

Fraser, P., 2002. Distributed Generation in Liberalised Electricity Markets. International symposium on distributed generation power system and market aspects, p.1G–12.

Freeman, L.C., 1977. A Set of Measures of Centrality Based on Betweenness. Sociometry, 40(1), p.35.

Global Construction Perspectives, 2013. Global Construction 2025.

Heller, J., Heater, M. & Frankel, M., 2011. Sensitivity Analysis: Comparing the Impact of Design, Operation, and Tenant Behavior on Building Energy Performance, Vancouver, B.C.: Ecotype and the New Buildings Institute.

Hiller, B., 2007. Space is the Machine: A Configurational Theory of Architecture Online., London, UK: Space Syntax.

Hongwei, D. et al., 1996. Optimal planning of distribution substation locations and sizes — model and algorithm.

International Journal of Electrical Power & Energy Systems, 18(6), pp.353–357.

IEE,

2014. Typology Approach for Building http://episcope.eu/building-typology/.

30/04/2014

Stock

Grant No. 608775

Energy

Assessment

(TABULA).

Available

at:

81

IES, 2013. Virtual Environment (IESVE).

Jansen, R.A., 2013. Second generation biofuels and biomass, Weinheim, Germany: Wiley-VCH. Jessoe, K. & Rapson, D., 2011. Commercial and Industrial Demand Response Under Mandatory Time-of-Use Electricity Pricing, Joskow, P.L. & Wolfram, C.D., 2012. Dynamic Pricing of Electricity. American Economic Review, 102(3), pp.381– 385.

Kahn, A.E., 1970. The economics of regulation: Principles and institutions 1st ed., John Wiley & Sons.

Kämpf, J.H., 2009. On the Modelling and Optimisation of Urban Energy Fluxes. Ecole Polytechnique Federale de Lausanne.

Kaplan, S., 2008. Power Plants: Characteristics and Costs, Washington D.C. Kaplan, S. & Sissine, F., 2009. Smart grid: Modernizing electric power transmission and distribution, TheCapitol.Net, Inc. Kempton, W. et al., 2008. A test of vehicle-to-grid (V2G) for energy storage and frequency regulation in the PJM system, Kempton, W. & Tomić, J., 2005. Vehicle-to-grid power implementation: From stabilizing the grid to supporting large-scale renewable energy. Journal of Power Sources, 144(1), pp.280–294.

Khaligh, A., 2010. Battery, Ultracapacitor, Fuel Cell, and Hybrid Energy Storage Systems for Electric, Hybrid Electric, Fuel Cell, and Plug-In Hybrid Electric Vehicles: State of the Art. IEEE Transactions on Vehicular Technology, 59(6), pp.2806–2814.

Kim, J.J., Yin, R. & Kiliccote, S., 2013. Automated Price and Demand Response Demonstration for Large Customers in New York City using OpenADR. , (October).

Kranzl, L., 2014. ENTRANZE Data Tool. Policies to Enforce the Transition to Nearly Zero Energy Buildings in the EU27. ENERDATA. Available at: http://www.entranze.enerdata.eu.

Kühne, R., 2010. Electric buses – An energy efficient urban transportation means. Energy, 35(12), pp.4510–4513.

Lam, J.C. & Hui, S.C.M., 1996. Sensitivity analysis of energy performance of office buildings. Building and Environment, 31(1), pp.27–39.

Latora, V. & Marchiori, M., 2003. Economic small-world behavior in weighted networks. The European Physical Journal B - Condensed Matter, 32(2), pp.249–263.

Latora, V. & Marchiori, M., 2001. Efficient Behavior of Small-World Networks. Physical Review Letters, 87(19), p.198701.

Li, F. et al., 2010. Smart Transmission Grid: Vision and Framework. IEEE Transactions on Smart Grid, 1(2), pp.168– 177.

30/04/2014

Grant No. 608775

82

Lomas, K.J. & Eppel, H., 1992. Sensitivity analysis techniques for building thermal simulation programs. Energy and Buildings, 19(1), pp.21–44.

Macdonald, I.A., 2002. Quantifying the Effects of Uncertainty in Building Simulation. University of Strathclyde, UK. Manwell, J.F., McGowan, J.G. & Rogers, A.L., 2009. Wind Energy Explained 2nd Editio., Wiley.

Mehaffy, M. et al., 2010. Urban nuclei and the geometry of streets: The “emergent neighborhoods” model. Urban Design International, 15(1), pp.22–46.

Mi, C., Masrur, M.A. & Gao, D.W., 2011. Hybrid Electric Vehicles: Principles and Applications with Practical Perspectives, Mills, E., 2002. Why we’re here: The $230-billion global lighting energy bill. In Right Light 5. Nice, France, pp. 369– 385.

MIT, 2011. The Future of the Electric Grid, Cambridge, MA: Massachusetts Institute of Technology.

Molderink, A. et al., 2010. Management and Control of Domestic Smart Grid Technology. IEEE Transactions on Smart Grid, 1(2), pp.109–119.

Momoh, J., 2012. Smart Grid, Hoboken, NJ, USA: John Wiley & Sons, Inc. National Grid (UK), 2009. How gas is produced, transmitted and distributed. , p.2.

Nemry, F. et al., 2008. Environmental Improvement Potentials of Residential Buildings (IMPRO-Building), Seville, Spain: IPTS JRC. Newman, P. & Kenworthy, J.R., 1999. Sustainability and cities: overcoming automobile dependence, NIST, 2010. NIST Framework and Roadmap for Smart Grid Interoperability Standards, Release 1.0. NIST Special Publication 1108, pp.1–145. Available at: http://www.nist.gov/smartgrid/upload/FinalSGDoc2010019corr010411-2.pdf.

NYSERDA, 2002. How-to guide to effective energy-efficient street lighting. , (October).

Pérez-Lombard, L., Ortiz, J. & Pout, C., 2008. A review on buildings energy consumption information. Energy and Buildings, 40(3), pp.394–398.

Pickavet, M. et al., 2008. Worldwide energy needs for ICT: The rise of power-aware networking. In 2008 2nd International Symposium on Advanced Networks and Telecommunication Systems. IEEE, pp. 1–3.

Porta, S., Crucitti, P. & Latora, V., 2006. The network analysis of urban streets: A dual approach. Physica A: Statistical Mechanics and its Applications, 369(2), pp.853–866.

Porta, S., Latora, V. & Strano, E., 2010. Networks in Urban Design. Six Years of Research in Multiple Centrality Assessment. In E. Estrada et al., eds. Network Science. London: Springer London.

RMI, 2012. Introducing the Retrofit Depot: Deep Energy Retrofit Guides. , (May).

30/04/2014

Grant No. 608775

83

Robinson, D. et al., 2009. CITYSIM: Comprehensive Micro-Simulation Of Resource Flows For Sustainable Urban Planning. In International IBPSA Conference. pp. 1083–1090.

Sabidussi, G., 1966. The centrality index of a graph. Psychometrika, 31(4), pp.581–603.

Scellato, S. et al., 2006. The backbone of a city. The European Physical Journal B, 50(1-2), pp.221–225.

SEAI, 2011. Energy Efficiency & Public Lighting Overview Report.

Serra, J.V.F., 2012. Electric vehicles: Technology, policy and commercial development 1st ed., Abingdon, UK: Earthscan.

SP Technical Research Institute of Sweden, 2014. Description of ScenoCalc (Solar Collector Energy Output Calculator), a program for calculation of annual solar collector energy output. , 1(22).

Stephenson, K. & Zelen, M., 1989. Rethinking centrality: Methods and examples. Social Networks, 11(1), pp.1–37.

Strano, E. et al., 2013. Urban street networks, a comparative analysis of ten European cities. Environment and Planning B: Planning and Design, 40(6), pp.1071–1086.

Strogatz, S.H., 2001. Exploring complex networks. Nature, 410(6825), pp.268–76.

Stromback, J., Dromacque, C. & Yassin, M.H., 2011. Empower Demand: The potential of smart meter enabled programs to increase energy and systems efficiency.

Struck, C. et al., 2011. Differential sensitivity of the energy demand for an office building to selected architectural design parameters. In Cleantech for Sustainable Buildings. Lausanne, Switzerland, pp. 817–822.

Struck, C., 2012. Uncertainty propagation and sensitivity analysis techniques in building performance simulation to support conceptual building and system design. Technische Universiteit Eindhoven.

Struck, C. & Kotek, P., 2009. On the Application of Uncertainty and Sensitivity Analysis with Abstract Building Performance Simulation Tools. Journal of Building Physics, 33(1), pp.5–27.

Takuno, T., Koyama, M. & Hikihara, T., 2010. In-Home Power Distribution Systems by Circuit Switching and Power Packet Dispatching. In 2010 First IEEE International Conference on Smart Grid Communications. IEEE, pp.

427–430.

UNDP, 2000. Energy and the challenge of sustainability, New York, NY: United Nations Development Programme.

US DOE, 2013a. Energy Efficiency of LEDs, PNNL-SA-94206.

US DOE, 2013b. EnergyPlus.

US DOE, 2001. Low-energy building design guidelines: Energy efficient design for new Federal facilities. Federal Energy Management Program Booklet. , p.7.

US EIA, 2003. Commercial Building Energy Consumption Survey (CBECS).

30/04/2014

Grant No. 608775

84

US EIA, 2009. Residential Energy Consumption Survey (RECS).

US EIA, 2001. Residential Energy Consumption Survey (RECS).

Vassell, G. & Maliszewski, R., 1969. AEP 765-kV System: System Planning Considerations. IEEE Transactions on Power Apparatus and Systems, PAS-88(9), pp.1320–1328.

Vereecken, W. et al., 2011. Power consumption in telecommunication networks: overview and reduction strategies. IEEE Communications Magazine, 49(6), pp.62–69.

Vragović, I., Louis, E. & Díaz-Guilera, A., 2005. Efficiency of informational transfer in regular and complex networks. Physical Review E, 71(3), p.036122.

Vuorinen, A., 2009. Planning on Optimal Power Systems, Vammala, Finland: Vammalan Kirjapaino Oy. Watts, D.J. & Strogatz, S.H., 1998. Collective dynamics of “small-world” networks. Nature, 393(6684), pp.440–2.

Wiser, W.H., 2000. Energy Resources: Occurrence, Production, Conversion, Use, Springer.

Wolfram Research Inc., 2014. Wolfram http://mathworld.wolfram.com.

MathWorld.

Wolfram

web

resources.

Available

at:

Yuan-Yih Hsu & Lu, F.-C., 1998. A combined artificial neural network-fuzzy dynamic programming approach to reactive power/voltage control in a distribution substation. IEEE Transactions on Power Systems, 13(4), pp.1265–1271.

Zaiser, M., 2003. Online learning webpage: The National Transmission System. The University of Edinburgh: The Learning Technology Section.

Available at: http://www.see.ed.ac.uk/~mzaiser/4thyear/websites05/MacRonald/Website/3-2.html.

30/04/2014

Grant No. 608775

85