Review on Energy Resilience

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Energy hubs optimization for smart energy network system to minimize economic and environmental impact at Canadian community

Mohamed Ghorab Natural Resources Canada, CanmetENERGY, 1 Haanel Drive, ON K1A 1M1, Ottawa, Canada

H I G H L I G H T S

• Annual model is conducted to estimate thermal and electrical loads for five hubs. • Load sharing between 20 buildings reduces system capacity sizes and capital costs. • Twenty mixing buildings presenting five different hubs are modeled and investigated. • Optimization model is developed for smart energy system at Canadian community. • PV technology reduces cost saving and CO2 emissions by 80% and 43% respectively.

A R T I C L E I N F O

Keywords: Energy optimization Smart energy network system Conventional Photovoltaic General algebraic modeling system Greenhouse gas emissions

A B S T R A C T

Recently, significant attention has been given to distributed power and thermal generation instead of centralized power plants. This study investigates the energy production of five energy hubs in a smart energy network set up for minimizing the energy cost and Greenhouse Gas (GHG) emissions under different technology scenarios at a Canadian community. The energy hubs present twenty Canadian buildings’ archetypes: sixteen single residential buildings, primary school, quick service restaurant, office, and supermarket. The hubs are interconnected by thermal and electricity grid and are sharing the locally generated energy to meet the demand loads. Three scenarios, each employing different energy technologies at hubs level were developed and investigated. The performance results were compared to a reference case.

A complex optimization model was developed to minimize the energy consumption and environmental im- pact. The results showed that integrating different residential and commercial buildings with associated gen- eration technologies in smart energy network allows energy sharing among them, enhancing overall efficiency and reducing energy and emissions costs. The study also showed that with this approach an electrical storage capacity of 25% of average electrical load had the lowest energy cost and environmental impact. In addition, it was found that integration of photovoltaic and electrical energy storage technologies would provide GHG savings up to 43% when compared to reference technologies.

1. Introduction

1.1. Background

Centralized energy system is typically used today to provide elec- trical and thermal powers to the buildings. However, Distributed Energy Resources (DER) system has the potential to replace the cen- tralized energy system in the near future. The localized network of the DER system will be designed in a small or medium size community and sited near consumers to meet the local energy demand loads. The DER system can include electrical and thermal storage technologies. It also includes waste energy recovery system to generate additional usable

energy and increases the overall system efficiency. The DER system also reduces the transmission and distribution losses as the generated energy is close to the consumer. In addition, the DER system increases the reliability of the energy supply to the customers. It allows using local renewable energy resources by adding more flexibility and provides energy control to the consumers. The local systems need to be properly designed to meet any variations in the thermal (heating and cooling) and electrical demand loads. The rise in energy demands has sig- nificantly contributed to the greenhouse gas emissions. Consequently, more attention has been given to renewable energy resources (e.g. solar, wind, geothermal, hydro, biomass) and locally generated energy (e.g. CHP and conventional systems). During the last two decades, the

https://doi.org/10.1016/j.applthermaleng.2019.01.107 Received 6 June 2018; Received in revised form 15 December 2018; Accepted 30 January 2019

E-mail address: [email protected].

Applied Thermal Engineering 151 (2019) 214–230

Available online 04 February 2019 1359-4311/ © 2019 Published by Elsevier Ltd.

T

DER system has been developed to include solar (PVT and PV), wind and cogeneration (CHP) systems. In order to minimize economic and environmental impacts for commercial, residential and industrial building applications, integrated renewable energy sources are required with energy system optimization to provide heating, cooling, and electrical energies. The research works on Smart Energy System (SEN) and energy system optimization will be described in details through the following two sections.

Smart energy system

Smart Energy System (SEN) is an integrated system of different energy sectors such as electricity, heating, cooling, transportation, ap- pliances, and buildings to achieve and afford solutions by implementing future sustainable and renewable energy systems. There are challenges of integrating electric power supply from the renewable energy re- sources to the electric power grid. The main idea of the integration is managing electrical energy generation and its distribution without causing feedback to the system [1]. Some progress has been made re- cently in the electric vehicle integration to the grid (V2G), vehicle (V2V), to anything V2X [2] at lower level or micro-grids and the in- tegration impact on the vehicle battery life [3]. The electrical energy in the vehicles can be integrated with the micro-grids by scheduling the charging/discharging time depending on available renewable energy and demand. Smart energy management control system could reduce the building energy consumptions significantly. The buildings can be

integrated with thermal and electrical networks by sharing the gener- ated on-site renewable energy from such technologies as photovoltaic thermal (PVT) [4], photovoltaic (PV) [5] and fuel cell (FC) [6]. In most such cases, buildings can achieve net-zero energy status.

Furthermore, many other technologies could be used in building applications such as microturbine [7], internal combustion engine [8], organic Rankine cycle (ORC) integrated with geothermal and solar energy [9], Stirling engine [10], fuel cell [11] and photovoltaic thermal (PVT) [12].

Energy system optimization

Smart heating and cooling energy systems performances have been improved through advanced control and smart management strategies. Multi-agent heating technology market is more stable and better suited for sustainable energy generation than centralized system [13]. In [14] authors focused on thermal energy storage (TES) optimization controls including system components activation, schedule time for charging and discharging. In [15] shifting the use of household’s smart appli- ances was examined either to renewable energy or to auto control by choosing the time of use.

The energy system optimization studies include variables, con- straints and complex problems arising out of the energy system com- ponents and technologies. Indicating and finding a suitable optimiza- tion technique to solve these problems depends on many parameters and energy system complexity. Ashouri et al. [16] applied mixed

Nomenclature

AC alternative current Ahub roof surface area of each hub APanels total solar panels surface area APV photovoltaic surface area APVT photovoltaic thermal surface area AR area ratio AWHP air-water heat pump C cost CHP combined heat and power DC direct current DER distributed energy resources E energy EP electrical purchase ES energy storage ESell electrical selling F Factor Fprice fuel price Fr Friction G energy generation GAMS general algebraic modeling system GSHP ground source heat pump GT energy transfer/exchange I radiation intensity L outlet power from technology MILP mixed integer linear programming MIP mixed integer programming MP mass production MR mass reduction ND number of days NGE natural gas emission factor NOP network optimization program OM operation and maintenance ORC organic Rankine cycle P inlet power to the technology PLR part load ratio

PV Photovoltaic PVT thermal photovoltaic R investment interest rate Sbat storage battery sf subsidy fraction ZEB zero-energy buildings

Greek symbols

η efficiency

Subscript and superscript

cap capacity ch Charging dem demand disch Discharging elec electricity f fuel h Hour hp heat pump hub Hub i Inlet max maximum N number of hubs n hub number 1, 2, 3, 4 or 5 n′ technology lifetime Ntech number of technologies o Outlet price Price s Supply sell selling subs subsidy Tech Technology w Week Wc outside weather conditions yr yearly

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

215

integer linear programming by developing a framework for selecting, sizing and controlling the building energy systems for an energy hub model case study. Maréchal and Kalitventzeff [17] applied a mixed integer linear programming (MILP) technique to select the utility system by combining modeling and practical systems. Moreover, MILP has been used by [18–24] to investigate DER system optimization. They concluded that high-energy technology efficiency had a significant ef- fect on the economic efficiency and CO2 reduction compared to low one. Omu et al. [23] used IBM ILOG CPLEX model and [24] developed a General Algebraic Modeling System (GAMS) model to solve the MILP optimization problem. Omu et al. [23] concluded that optimally de- signed DER system could reduce annual CO2 emissions with more ef- ficient use of primary energy. Parisio et al. [26] presented a robust optimization approach to energy hub management using MILP. Brahman et al. [27] studied optimal electrical and thermal energy management of a single residential energy hub by integrating energy production, conversion, and storage technologies. Ren et al. [28] in- vestigated economic optimization and sensitivity analysis of the pho- tovoltaic system in residential buildings.

Fazlollahi and Maréchal [18] proposed an alternative approach to apply the multi-objective optimization by combining mixed integer nonlinear programming with evolutionary approaches. Another ap- proach to solve mixed integer nonlinear programming was divided the problem into nonlinear and a mixed integer by applying a structured process [19]. Furthermore, Scala et al. [25] modeled optimal energy flow management in multicarrier energy networks including inter- connected and distributed energy hubs by using a multi-objective op- timization study.

Moreover, conventional thermal and electrical systems optimization has been investigated, for operating cost and emissions [29] and for operational cost and unit size [30,31]. Likewise, district thermal (heating and cooling) and electrical systems optimization and man- agement have been studied for economic operational strategies [32–37]. Whereas, Maroufmashat et. al. [38] investigated the economic and emission optimization of network energy for three hubs and two energy technologies.

1.2. Problem description and study objective

Energy hubs have a great potential for integrating energy system models to achieve sustainable multi-energy systems. Optimum energy operation with less environmental impact requires an integrated energy control system at both hub and community levels.

Incorporating multiple energy sources both conventional and re- newable need to increase substantially to achieve smart energy system status at community level. Integrated conventional and renewable en- ergy resources presented at the energy hubs are strongly dependent on the individual system performance and the demand load profile at each hub. Many factors influence the optimum energy planning design at community level such as local emission regulation, capital cost, avail- able renewable energy resources, fuel, and electrical costs, incentive programs for renewable energy, local air quality, building applications, and weather conditions.

There is a gap in the literature related to integration of conventional and renewable energy resources, using smart energy optimal strategy for minimizing energy cost and carbon footprint. Canadian buildings’ archetypes including residential, educational and commercial energy hubs are developed and investigated by employing different energy technologies and smart energy optimization for minimum energy cost and environmental impact. Energy hubs optimization model is devel- oped using General Algebraic Modeling System (GAMS) programming platform.

In this study, five hubs located in Toronto, Canada presenting twenty building archetypes are investigated. The building archetypes include sixteen single residential buildings, a small office, a quick ser- vice restaurant, a supermarket, and a multi-floor primary school.

Table 1 presents the flat surface area of the twenty buildings for the five hubs.

1.3. System description

Fig. 1a presents five hubs network with mixed building archetypes. The performances for the five hubs were annually modeled, simulated and optimized under Toronto, Canada weather conditions. Electrical grid, conventional and renewable technologies were used at each hub to provide electrical and thermal energies to meet the demand loads. Excess electrical and thermal energies can be shared among the hubs. In addition, a provision of exporting/importing electricity to and from the grid is provided service instances when there is a mismatch between the generation and demand.

Energy generation and distribution among the hubs were in- vestigated to identify the optimum technologies and energy production at each hub to meet thermal and electrical demand loads with minimum energy cost and CO2 emissions. The multi-objective optimization ap- proach is applied by developing a mixed integer programming (MIP) code using GAMS platform to solve the complex system optimization problem.

The input parameters to the mathematical model include thermal and electrical demand loads for each hub, weather conditions (e.g. outside temperature, solar radiation, wind speed, etc.), gas and elec- tricity tariffs, gas emission tax and financial data (capital cost, interest rate, incentive, lifetime, etc.).

Case studies shown in Table 3 are modeled, simulated and in- vestigated for optimum energy cost and environmental impact. These case studies are categorized into three scenarios and reference system (base case). The reference case includes conventional technology system where single boiler and chiller are installed at each hub to provide thermal energy (heating and cooling) while electrical energy is imported from the utility grid to meet the electrical demand loads. The first scenario of the case studies has similar technologies for the re- ference case in addition to electrical energy storage system (batteries) to store electrical energy during off-peak hours and use it back during on-peak hours. The second scenario includes the same technologies for the reference case with integrated PVT/PV systems to provide thermal and electrical energies to the hubs. The third scenario is the same as the second scenario but with replacing the boiler and chiller with a heat pump to provide thermal energies. A small boiler is used in the third scenario to provide domestic hot water energy.

Fig. 1b presents a schematic of energy flows between different technologies, grid and building loads at each hub for scenarios 2 and 3. A heat pump in scenario 3 replaces the chiller in scenario 2. The elec- trical energies from the PVT and PV systems in scenarios 2 and 3 are using to meet part or full electrical loads, recharge battery storage or export to the grid. The generated thermal energies from boiler, PVT and heat pump are shared between the five hubs to meet the space heating and DHW loads.

1.4. Thermal and electrical demand loads

Fig. 2 shows the hourly profiles of irradiance in kW/m2 and outside temperature in °C under Toronto weather conditions. These profiles are

Table 1 Building type and surface area.

Hub Building type Number Footprint surface area (m2)

1 Single RB 16 3200 2 Quick service restaurant 1 232 3 Supermarket 1 4180 4 1-story small office 1 511 5 Primary School 1 860 Total 20

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

216

used to simulate the annually thermal loads for the building’s arche- types, and to estimate thermal and electrical energy generations from renewable technologies (PVT, PV systems).

The commercial and educational building’s envelopes used in this study are in accordance with the National Energy Code of Canada for Buildings (NECB). The residential buildings are built according to the R- 2000 building code. Annual dynamically simulations were conducted for each hub to estimate hourly thermal and electrical load profiles using TRNSYS platform under Toronto weather conditions. Moreover, to estimate the thermal and electrical load profiles of the sixteen re- sidential buildings, a single residential building is simulated and then thermal and electric diversity factors are applied. The diversity factor is defined as a ratio of the measured power over the maximum possible power value [39]. The diversity factor values of 70% and 85% [40] are applied in the current study.

Fig. 3 presents the weekly average thermal and electrical load

profiles for the five hubs. A single week is selected for each season to present buildings average load profiles. The simulation results show that the supermarket (hub 3) has the highest electrical demand load among other hubs with peak values during the daytime following by the primary school. The 1-story small office has the lowest electrical load between 8:00 and 24:00 hrs compared to the other hubs. The electrical loads are almost the same from 00:00 to 8:00 hrs for 1-story office, 16 single residential buildings, and quick service restaurant, as shown in Fig. 3a. On the other hand, the thermal load for the primary school has the highest values in comparison with the other hubs as presented in Fig. 3b. The heating load for residential buildings drops during the daytime from 9:00 to 16:00 hrs due to lower set point temperatures and solar heat gains. The internal heat gain during the daytime in the winter season decreases the heating load for other building hubs including supermarket, office and quick service restaurant. In the summer season, the DHW is the only source for heating load.

a) Energy transfer between the hubs/ grids

b) Energy flow at each hub

Boiler

Chiller / HP

To Heating Load

Thermal energy from HP in winter

Import Electricity from Grids

To Cooling Load

To Electrical Load

Thermal cooling Flow

Electrical Energy Thermal heating Flow

Gas Flow

PV Elec. Energy

Battery

Thermal Energy to other hub (s)

Inverter

Inverter Inverter

PV system

Electrical export to grid

Electrical Transfer to other hub (s)

Thermal Energy from other hubs

Cooling Energy from other hub (s)

Cooling Energy to other hub (s)

Elect. Energy from other hub (s)

Fig. 1. Schematic of Energy system interface between the hubs and technologies.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

217

2. Optimization methodology

The optimization problem is formulated as an advanced optimiza- tion model since each hub has a unique option to install different technologies with different capacity sizes to meet the thermal and electrical demand loads of the buildings. Each hub is connected with other hubs through pipelines and electric wires to share the excess thermal and electrical energies as shown in Fig. 1a. A network opti- mization program (NOP) is developed using the General Algebraic Modeling System (GAMS) [41]. The NOP can be used for optimization problem with a large number of hubs and energy technologies such as

conventional, GSHP, AWHP, CHP, ORC, PVT, PV, fuel cell, wind, and electric storage systems.

2.1. Hub and network energy modeling

The energy hub concept is presented in Fig. 4. It includes different input energy resources and their conversion to various outputs. The energy outputs can be thermal from boiler and heat pump or electrical from PV technology or both thermal and electrical from PVT, fuel cell (FC) and CHP technologies. At each hub, there are unique input and output parameters which are dependent on available energy resources

0

1

2

3

4

5

6

7

8

9

10

Ir ra

d ia

n ce

( kW

/m 2 )

Time (Month)

January February March April May June July August September October November December

-30

-20

-10

0

10

20

30

40

O u

ts id

e T

e m

p e

ra tu

re (

°C )

Time (Month)

January February March April May June July August September October November December

Fig. 2. Hourly irradiance (kW/m2) and outside temperature (°C) profiles under Toronto weather conditions.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

218

and used technologies. Since there is multi-power flow at each hub, a converter or multiplication factor is applied to the input energy re- sources to achieve the outlet energy flow as presented in Eq. (1). The efficiency matrix (ηoi) for the input power flow and converter unit factor (Foi) are defined based on the type and capacity of used tech- nologies at each hub. The efficiency matrix elements have zero values

when there is no conversion from the inputs to outputs. However, if there is energy transfer from Pi to Lo, a coupling factor is applied to represent the technology efficiency. In case, there is more than one energy conversion technology at the hub, the coupling factor of tech- nology efficiencies is applied. Moreover, in order to make the matrix multiplication consistent for the optimization problem, an input energy

a) Electrical

b) Thermal

0

100

200

300

400

500

600

700

800

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

El e

ct ri

ca l L

o ad

(k W

)

Time (hr)

Week 216 Single RB Quick service restaurant Supermarket 1-story small office Primary School

0

100

200

300

400

500

600

700

800

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

El e

ct ri

ca lL

o ad

(k W

)

Time (hr)

Week 1516 Single RB Quick service restaurant Supermarket 1-story small office Primary School

0

100

200

300

400

500

600

700

800

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

El e

ct ri

ca l L

o ad

(k W

)

Time (hr)

Week 2816 Single RB Quick service restaurant Supermarket 1-story small office Primary School

0

100

200

300

400

500

600

700

800

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

El e

ct ri

ca lL

o ad

(k W

)

Time (hr)

Week 3716 Single RB Quick service restaurant Supermarket 1-story small office Primary School

0

20

40

60

80

100

120

140

160

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Th e

rm al

L o

ad (

kW )

Time (hr)

Week 216 Single RB Quick service restaurant Supermarket 1-story small office Primary School

0

20

40

60

80

100

120

140

160

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Th e

rm al

L o

ad (

kW )

Time (hr)

Week 1516 Single RB Quick service restaurant Supermarket 1-story small office Primary School

0

20

40

60

80

100

120

140

160

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Th e

rm al

L o

ad (

kW )

Time (hr)

Week 2816 Single RB Quick service restaurant Supermarket 1-story small office Primary School

0

20

40

60

80

100

120

140

160

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Th e

rm al

L o

ad (

kW )

Time (hr)

Week 3716 Single RB Quick service restaurant Supermarket 1-story small office Primary School

Fig. 3. Hourly electrical and thermal load profiles for each hub at different weeks (a) electrical, and (b) thermal.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

219

variable is defined for each converter unit factor.

⎡

⎣

⎢ ⎢ ⎢ ⎢

⎤

⎦

⎥ ⎥ ⎥ ⎥

= ∗

⎡

⎣

⎢ ⎢ ⎢ ⎢

⎤

⎦

⎥ ⎥ ⎥ ⎥

= ⎡

⎣ ⎢ ⎢

⎤

⎦ ⎥ ⎥

L

L

L

η F

P

P

P

η η η

η η η

. .

.

.

. , Where

· · ·

· o

O

oi oi i

I

oi

I

oi

O OI

1 1 11 1

1

(1)

Energy network model includes groups of different energy hubs, connected to the network energy distribution systems. Electricity and natural gas are the main input energy resources to all case studies. The thermal and electrical energy outputs from each hub meet the demand loads for that hub first then the excess energies are transferred to other hubs through the network energy distribution systems. The energy ex- change among the hubs is presented in Eq. (2) where (Eo) is the total energy transferred from one hub (sender) to other hubs and ET is the energy transfers from the sender hub to another hub (receiver). The net energy received at each hub (Ei) from other hubs (excluding thermal and electrical energy transfer losses) is presented in Eq. (3). Where EL is the energy losses between the hub (“n”) and other hubs (N).

∑ ∑= → = − ∀

∉

E hub E hub n where N n hub hub

N

( ) ( ) 1 & &o N

T

(2)

∑= → − → ∀ ∉E hub E n hub EL n hub hub hub N( ) ( ( ) ( ) ) &i N

T

(3)

2.2. Objective function

A multi-objective optimization function is used to minimize the annual total energy cost and environmental impact for the twenty buildings combined in five hubs. The environmental impact is pre- sented by emission cost as a function of electricity and natural gas consumptions. Then the sum approach (SA) [42] is applied to convert the objective function to a single one. The objective function (CTotal) in Eq. (4) includes cost of capital (Ccapital), fuel (Cfuel), electrical (Celec), operational and maintenance (COM) and emissions (Cemission). The rev- enue costs such as subsidy for using renewable technology (Csubs) and selling electric power to the grid (Cselling) are subtracted from the an- nual cost.

∑ ∑ ∑ ∑ ∑= ⎧ ⎨⎩

+ + + +

− − ⎫ ⎬⎭

C C C C C C

C C

(

)

Total Min

capital hub

Tech w h fuel elec OM emission

subs selling (4)

The GAMS optimization model includes many variables such as capital cost for each technology, fuel and emissions costs, maximum electric storage capacity, OM cost, thermal and electrical energy gen- eration technologies, part load energy profile, imported and exported electric power from or to the grid. The optimum decision variables include generated energies from different technologies at each time step for each hub. The transferred energy decision variables from the hub (s) to other hubs are presented as binary variables.

2.3. Model constraints and costs

The technology system constraints define the generated energy limitation from each technology at each hub for every time step. The input energy (Pi) for each technology in Eq. (1) is constrained by the technology minimum and maximum capacity limitations, as shown in Eq. (5).

≤ ≤P P Pmin i max (5)

On the other hand, when a technology is active, the generated en- ergy should be between the maximum capacity and minimum allowing output at part load operation as shown in Eq. (6) for each hub, tech- nology and time step.

∗ ≤ ≤ ∀PLR Tech G Tech hub Tech h wMax ( ) Max ( ) , , ,Cap hub Tech h w Cap. , , , .

(6)

The energy generated from the PVT and PV systems is dependent on solar radiation, outside weather conditions, solar panel surface areas, and solar panel performance. The number of solar panels is limited by available surface area which is equal to 75% of the building roof areas for each hub. The PVT and PV models [43] are dynamically simulated for a year under Toronto weather conditions to obtain respective thermal and electrical energy outputs. Area ratio (AR) is ratio of PVT solar panels’ surface area (APVT) to the total PVT and PV panels’ surface areas (APVT + APV). The AR is varying from 0.0 to 0.75. For AR = 0.0, means that the system does not include PVT technology and all used solar panels are PVs. The PV generation is constrained by surface area, radiation, electrical panel performance and inverter efficiency as pre- sented in Eq. (7).

≤ ∗ ∗ ∗ ∀G η I A η hub h w, ,hub pv h w inverter h w pv elecpv, , , , (7)

At each hub, the generated thermal energy (G) plus received energy (ET) from other hubs should meet the thermal load at every time step (h, w) as presented in Eq. (8). In addition, the generated thermal energy from all hubs should balance with the total thermal demand loads for the five hubs at every time step as shown in Eq. (9).

@ each hub;

∑ + ≥ ∀G thermal ET Dem thermal h w( ) ( ) , Tech h w

h w h w ,

, , (8)

For all hubs;

∑ ∑ ∑≥ ∀G thermal Dem thermal h w( ( ) ) ( ) , hub

Tech h w hub h w

, ,

(9)

On the other hand, the generated electric power at each hub, plus stored electrical energy in the batteries plus purchased electricity from the grid should balance with the electrical energy demand at each hub as shown in Eq. (10) for every time step (h, w). Furthermore, for all hubs, electrical energies which are generated, stored, imported, trans- ferred among the hubs and exported to the grid should balance the electrical energy demand loads for all hubs at every time step (h, w) as presented in Eq. (11). The storage of electrical energy from the previous time step (h-1, w) will be available to use at the current time step (h, w).

@ each hub;

∑ + ≥ − ∀−G elec ES elec Dem elec E P T h w( .) ( .) ( .) ( , ) , Tech h w

h w h w h w ,

1, , ,

(10)

For all hubs;

∑ ∑

∑

⎛

⎝ ⎜ + +

⎞

⎠ ⎟

= ∀

−G elec ES elec E P TorSell

Dem elec h w

( .) ( .) ( , )

( .) ,

hub Tech h w

h w h w

hub

, 1, ,

(11)

Energy Hub In

pu t P

or ts

O ut

pu t P

or ts

P1 P2 Pi

PI

L1 L2 Lo

LO

Fig. 4. Input and output of energy hub.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

220

At each hub, excess generated electricity is normally storing in a battery storage system and used it back when the electrical demand load (s) is higher than the generated electricity. The electrical energy from the grid can be stored in the batteries during off-peak hours and used it back during on-peak hours. Charging the batteries is calculated using Eq. (12). The charging occurs during off-peak hours (from 00:00 to 7:00 and from 19:00 to 24:00) and when there is an excess of electric power generation from all hubs. The storage of electrical energy does not happen if the total electrical demand load is higher or equal to the average electrical load (LAV) for all hubs and during on-peak hours. The discharge electrical energy from the stored energy system to the hub/s (G SBat) in Eq. (13) is mainly dependant on the electrical demand load at time step (w,h) and energy storage at previous time step (w,h-1). The rate of electrical energy storage at each time step is assumed to be first order of forward discrete function as shown in Eq. (14). The electrical energy storage at each time step presented by Eq. (15) is a function of the storage level and discharge electrical energy at the previous time step and the amount of charging electricity at the current time step. The electrical storage technologies cannot charge and discharge at the same time. The electrical energy storage level for charging and discharging in Eqs. (16a) and (16b) respectively are limited by the maximum battery storage capacity at each hub.

@ each hub,

> → = ∗

− ∀

L elec Dem elec ES elec η L

elec Dem elec h w

If ( .) ( .) ( .) (

( .) ( .) ) ,

AVh w h w chh w inverter charging av

h w

, , , ( )

, (12)

= = ∀−G SBat elec ES elec ES elec h w( , . ) ( .) ( .) ,h w dischh w h w, , 1, (13)

= − + ∀−ES elec ES elec ES elec ES elec h ẇ ( .) ( .) ( .) ( .) ,h w h w h w lossesh w, , 1, ,

(14)

= + −

− ∀

− −ES elec ES elec ES elec ES elec

ES elec h w

( .) ( .) ( .) ( .)

( .) ,

h w h w chh w outh w

lossesh w

, 1, , 1,

, (15)

≤ ∗ES elec η ES( .)dischh w inverter discharging max, ( ) (16a)

and,

∗ ≤η ES elec ES( .)inverter charging chh w max( ) , (16b)

There are various constraints restricting energy exchange among the hubs. At each hub, the energy cannot transfer and receive at the same time. Therefore, there is energy flow direction at each hub. Moreover, each hub cannot distribute energy to another hub unless it meets its own energy demand loads first as presented in Eq. (17). Which means that at every time step, the hub stops to exchange energy with other hubs if the energy generated is less than or equal to the energy demand loads at that hub. Moreover, at each hub, transferred thermal energies between that hub and other hubs is equal to generated thermal energy from all technologies minus the thermal demand load at each time step, as shown in Eq. (18). Likeness for each time interval, the generated electricity at each hub is equal to electricity demand load at that hub plus transferred electricity to other hubs as shown in Eq. (19).

@ each hub; If

< →

= ∀

G heat elec Dem heat elec GT heat elec

h w

( , .) ( , .) Δ ( , .)

0 , h w h w h w, , ,

(17)

∑= − ∀GT heat G heat Dem heat h wΔ ( ) ( ) ( ) ,h w Tech h w h w

, , , (18)

∑= − ∀GT elec G elec Dem elec h wΔ ( .) ( .) ( .) ,h w Tech h w h w

, , , (19)

The selling and purchasing to or from the grid happen only for electrical energy at community level (for all hubs). Excess or shortage of

electric power cannot happen at the same time. The relative magnitude of the energy generation (thermal and electrical) is mainly dependent on energy demand loads and used technologies. The net excess electric power from the entire hubs at each time step is selling back to the grid. Each individual hub cannot sell the excess electric power to the grid until it satisfies first electrical energy demand for its own and for other hubs. At each time step, if the sum of generated electrical energy from different technologies at each hub is less than the sum of the electrical demand loads for all hubs, the selling electrical energy to the grid is equal to zero as presented in Eq. (20). The exported electricity to the grid is equal to the difference between the sum of generated electricity and the sum of electricity demands minus the sum of electrical energy storage at each hub for each time step, as shown in Eq. (21). For all hubs, imported electrical energy from the grid (purchasing electricity) is needed to meet the electrical energy difference (ΔGTh,w (elec.)). This happens when generated electricity by technologies plus electrical en- ergy storage is less than the sum of electrical demand loads, as shown in Eq. (22).

If;

∑ ∑ ∑≤ → =

= ∀

G elec Dem elec GT elec ESell

h w

( .) ( .) Δ ( .)

0 , hub

h w hub

h w hub

h w h w, , , ,

(20)

And if;

∑ ∑

∑ ∑

∑

> + →

= = ⎛

⎝ ⎜

− + ⎞

⎠ ⎟ ∀

G elec Dem elec E SBat ESell

GT elec G elec

Dem elec E SBat h w

( .) ( ( .) ( ) )

Δ ( .) ( .)

( ( .) ( ) ) ,

hub h w

hub h w h w h w

hub h w

hub h w

hub h w h w

, , , ,

, ,

, , (21)

And if;

∑ ∑

∑ ∑

∑

+ < →

= = ⎛

⎝ ⎜

−

+ ⎞

⎠ ⎟ ∀

−

−

G elec E SBat Dem elec EP

GT elec Dem elec

G elec

E SBat h w

( .) ( ) ( .)

Δ ( .) ( .)

( ( .)

( ) ) ,

hub h w h w

hub h w h w

hub h w

hub h w

hub h w

h w

, 1, , ,

, ,

,

1, (22)

The objective optimization function in Eq. (4) is calculated using the following cost Eqs. (23)–(29). The capital cost is calculated based on the maximum capacity of each technology at each hub times the unit cost of each technology multiply by the payback cost over the technology entire lifetime (n′), as shown in Eq. (23). The investment interest rate (R) is equal to 3.5%.

∑ ∑= ∗ ∗ −

∀

+

C Cap C R

hub TechMax ( . ) 1

,Cap hub Tech

Tech Cap Tech

R

. .( ) 1

(1 )n '

(23)

The fuel cost is calculated based on natural gas fuel consumption by the technology (boiler) multiply by generated energy times the fuel price [44] divided by technology efficiency as presented in Eq. (24).

∑ ∑ ∑ ∑= ∗ ∗ ∀C F G heat ND η

hub Tech h

w

( ) , ,

,

fuel hub Tech h w

price hub Tech h w

Tech

, , ,

(24)

The electrical cost in Eq. (25) is calculated based on the imported electric power from the grid multiply by electricity price. Time of use (TOU) rate is applied to calculate the electricity price [45]. Three dif- ferent electricity rates (high, medium and low) are used in Eq. (25). The

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

221

TOU electricity rate is changing in the summer and winter seasons.

∑ ∑ ∑= ∗ ∗ ∀C EP ND EP hub h w, ,Elec hub h w

hub h w price

. , , (25)

The operation and maintenance costs for each technology are cal- culated based on annual fixed cost, which is equal to the maximum capacity of each technology at each hub multiply by the unit cost of the technology multiply by a fraction value (fr), as shown in Eq. (26).

∑ ∑= ∗ ∗C Max Cap C fr( . )OM hub Tech

Tech Cap Tech.( ) (26)

The CO2 emissions cost is calculated based on the tax charge per ton of CO2 emissions ($25 per ton of CO2). The natural gas and electricity Ontario utility emission factors of 0.235 and 0.59 kg/kWh, respectively are used to calculate the CO2 emissions from technologies and utility grid. The total CO2 emissions from each hub (MPCO2) is calculated based on the natural gas consumption and imported electric power from the grid. On the other hand, generated energy by renewable technology system (s) reduces the CO2 emissions with a value of MRCO2 (mass re- duction of CO2). Equation (27) is used to calculate the annual CO2 emissions cost. Where ND is presenting the number of days per week.

∑ ∑ ∑ ∑= ∗ ∗ − ∀C C ND MP MR

hub Tech h w

( ( ) )

, , ,

emissions hub Tech h w

emissionsperton CO CO2 2

(27)

Many renewable technologies are eligible for an appropriate subsidy to incentivize investment in low-carbon energy technologies to calcu- late the revenue stream. The subsidy cost is calculated using Eq. (28) based on subsidy fraction (sf) of the technology capital cost.

∑ ∑= ∗ ∀C sf Cost hub Tech hub Tech( ( , ) ) ,Subs hub Tech

Tech cap. . (28)

The annual selling electrical energy cost to the grid is calculated based on the excess electrical energy generated by the technologies and electricity-selling price multiply by the number of days (ND), as shown in Eq. (29). Time of use (TOU) rate [45] is applied for selling the electrical energy back to the grid but the electricity-selling rate is equal to 75% of the electricity purchase rate.

∑ ∑ ∑= ∗ ∗ ∀C ESell ND ESell hub h w, ,Selling hub h w

hub h w price

, , (29)

2.4. Optimization model

General Algebraic Modeling System (GAMS) [41] is a powerful modeling system for mathematical programming and optimization that can be developed and used for complex and large-scale modeling ap- plications. In this study, a GAMS program code was developed to analyze the impact of renewable and integrated energy technologies on energy cost and GHG emissions. It also provides an optimal manage- ment and efficient energy network distributions between the hubs. Solving optimization problem becomes a complex with nonlinear con- straints. A set of assumptions were applied to the optimization model in order to reduce the model complexity. Firstly, a weekly average hourly data (w, h) was used instead of daily hourly data (d, h) to reduce the computational time. Secondly, the excess thermal energy is not selling back to the thermal grid. The network thermal energy losses are ne- glected. Lastly, the excess electrical energy is selling back to the grid with 75% of the electricity purchase price considering the time of use (TOU). Table 2 presents unit capital and maintenance costs, lifetime, operation fraction, for each technology used in the present study.

The current optimum model includes many variables such as initial values, lower, and upper bounds, positive and binary variables. The average iteration number of 10,360 is used in the model. GAMS pro- gramming platform for mixed integer programming (MIP) problem is developed using integrated CPLEX solver to optimize DER for the

different case studies in Table3. In order to have a confidence of obtained optimum solution, dif-

ferent solvers for the programming model are tested with different in- itial conditions. The results show that initial conditions do not have any effect on the optimum results by using MIP with CPLEX solver. The initial conditions affect only the modeling time. Furthermore, narrow range between lower and upper bounds for each variable is applied in the present model to reduce the computation time. Additionally, energy balance at every time step for each individual hub and all system is less than 10−14.

For each case study, the optimum output results include energy conversion technologies, optimum generated energy at each hub, eco- nomic cost and CO2 emissions.

3. Case studies

Five different hubs with conventional and renewable technology systems are modeled and simulated to analyze the economic energy and environmental impact. As presented in Table 1, a combination of five different buildings (residential, commercial and educational) presents Canadian community level. The building archetypes include sixteen single residential houses, Quick service restaurant, supermarket, one story small office, and multi-floor primary school.

Energy consumption reduces by combining the load profiles for mixing building applications, which improve the technology part load conditions. Furthermore, there is a reduction in total technology system capacity due to load shifting between the buildings. The estimated re- duction in technology capacities due to load sharing is ranged from 18% to 24% for heating and cooling systems.

Table 3 illustrates the test matrix for the case studies. As mentioned earlier in Section 1, the conventional system with grid connection is considering a reference case study (base case). Three different scenarios each employing different technology are investigated in this study. First scenario is integrated electrical energy storage battery technology to the base case. Four electrical storage battery capacities are investigated for scenario 1 as shown cases 1–4 in Table 3. The optimum case study of scenario 1 (cases 1–4) is used for scenario 2 with integrated PVT and PV technology systems. Four different area ratios (AR) of PVT and PV technology systems are investigated for the second scenario as pre- sented cases 5–8 in Table 3. Finally, the heat pump replaces boiler and chiller of the optimum case of scenario 1 with integrating PVT and PV renewable technologies to present scenario 3. Three additional case studies of area ratios (AR) are investigated for scenario 3 as shown cases 9 to 11 in Table 3. In total eleven cases with different technology sys- tems and capacities were investigated and analyzed annually for minimum energy cost and environmental impact.

4. Results and discussions

4.1. Annual electrical and thermal demand loads

Fig. 5a presents the annual electrical demand load at each hub. The supermarket has the highest electrical load among the other hubs with the lowest being the 1-story small office. The annual electrical load intensities are 43.2, 1515, 554.5, 98 and 823.9 kW/m2 yr for 16 re- sidential, quick service restaurant, supermarket, 1-story small office,

Table 2 Technology economic parameters.

Technology Capital-cost ($/kW) Life-time (year) OM fraction

Boiler 85 30 0.012 PVT 2240 25 0.01 PV 2000 25 0.005 Battery 125 10 0.01 HP 1200 15 0.01

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

222

Table 3 Test matrix for case studies.

Case #/System Grid Boiler heating Boiler DHW Heat pump AR (PVT) 1-AR (PV) Elect. storage %

Reference system Base case X X X

Scenario 1 Elec. Storage Size % of Av. Load Case 1 X X X 100%

Case 2 X X X 60% Case 3 X X X 40% Case 4 X X X 25%

Scenario 2 PVT area ratio to total Area used with conventional system Case 5 X X X 0.75 0.25 25%

Case 6 X X X 0.5 0.5 25% Case 7 X X X 0.25 0.75 25% Case 8 X X X 0.0 1.0 25%

Scenario 3 PVT area ratio to total Area used with heat pump Case 9 X X 0.1 0.9 25%

Case 10 X X 0.05 0.95 25% Case 11 X X 0 1 25%

a) Electrical demand load

b) Thermal demand load

0

500

1000

1500

2000

2500

16-Single RB Quick service restaurant Supermarket 1-story small office Primary School

El e

ct ri

ca l E

Lo ad

(M W

)

Elec. Demand Load

82,955

2,317 23,382

9,713

394,802

0.0E+0

5.0E+4

1.0E+5

1.5E+5

2.0E+5

2.5E+5

3.0E+5

3.5E+5

4.0E+5

4.5E+5

16 Signale RB Quick service restaurant

supermarket 1-story small office

Primary School

T he

rm al

l oa

d de

m an

d M

W h/

yr

Thermal Energy Demand

Fig. 5. Annual electrical and thermal demand loads for each hub.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

223

and primary school buildings, respectively. Fig. 5b illustrates the annual thermal demand load for each hub. The

results show that the thermal demand load for the primary school is significantly high as the school consist of multiple floors with high fe- nestration to wall ratio which increases the heat transfer through the walls and windows. In addition, the primary school has a high domestic hot water consumption during the school operating time. The thermal demand load for the 16 single residential buildings is lower than that for the primary school. The quick-service restaurant has the lowest thermal energy load among other buildings due to high internal heat gains followed by 1-story small office then the supermarket. The annual thermal demand load intensities per footprint surface area for all ar- chetypes are 206.9, 93.3, 68.4, 36, and 20.1 GJ/m2 for primary school, 16-SRB, office, quick service restaurant, and supermarket, respectively.

4.2. Effect of electrical storage capacity

Battery energy storage system is used to store electrical energy from the grid during off-peak hours and also when the generated electricity exceeds the demand loads. Electrical energy discharges from battery during on-peak hours to reduce the peak load and to avoid higher purchase price. Four battery sizes (100%, 60%, 40%, and 25%) as a percentage of the average electrical load at each hub are investigated to estimate battery optimum size. Fig. 6 illustrates the hourly electrical load profile for the primary school on week 15. The Figure also shows charging and discharging electrical energy during off-peak and on-peak hours, respectively with battery sized 60% of the average electrical load. Larger battery storage capacity is required for charging at level higher than 60% of daily average electrical load and vice versa.

Fig. 7 shows the effect of electrical energy storage levels (100%, 60%, 40% and 25% of daily average electrical load) on the annual cost savings for scenario 1 compared to the reference case (conventional system without electrical energy storage). The results show that at high electrical energy storage level, the annual total cost for scenario 1 is about 7.4% more than for the base case. The annual cost decreases at low electrical energy storage levels. At 25% electrical energy storage level, the annual total cost for scenario 1 is higher than the base case with a value of 0.2%. The results show that there is no significant effect on annual cost saving for scenario 1 at high electrical energy storage level. As a result, the case with 25% of average electrical load (case 4 in Table 3) has the optimum annual cost saving compared to other case studies in scenario 1.

4.3. Effect of photovoltaic thermal and photovoltaic panels area ratio

Total solar panel area (APanels) at each hub presents 75% of the roof surface area. The Area ratio (AR) is tested at different values (AR = 0.75, 0.5, 0.25 and 0.0) to study the effect of using a different combination of PVT and PV solar panels on annual optimum total cost and CO2 emissions for the community buildings. At AR = zero, all used solar panels are photovoltaics (PVs). Fig. 8 presents the effect of the area ratio (AR) on the optimum annual total cost saving for scenario 2 (boiler, chiller, battery, PVT and PV system) compared to the reference case. The results show that there is a significant increase in both annual total cost and purchasing electricity cost savings with further increase of PVs instead of PVTs panels. Fig. 8 illustrates selling electricity cost ratio at different area ratios compared to the case with AR = 0.75 in scenario 2. From the results, the selling electricity cost ratio increases with further increase of PVs solar panels and the case with AR = 0.0 has the highest value of 9.4 among other case studies. Therefore, the PVT technology is not the optimum system for minimum energy cost and GHG emissions because the capital cost of PVTs is higher than the PVs for the same panel surface areas. Moreover, the electric power output from the PVT system is lower than the PV system due to high fluid temperature. As a result, output thermal and electrical energies from the PVT system are not economic compared to the PV system.

Furthermore, Fig. 9 demonstrates the annual cost saving at different AR (10%, 5%, and 0.0%) for scenario 3 compared to the reference case. The results show that the annual cost saving of purchasing electricity

0

20

40

60

80

100

120

140

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

El ec

. E n

er gy

k W

h

Time (hr)

Demand Load

Charge / Discharge from Elec. Storage

Import Elec.

100% Av. Elec. Load

60% Av_Elec. Load

40% Av_Elec. Load

Charging

Discharging

Charging

Fig. 6. Primary school electrical profile including charging and discharging electrical energy on week 15 for 60% of the average electrical load.

-10%

-8%

-6%

-4%

-2%

0%

2%

4%

100% 60% 40% 25%

Sa vi

n g

(% )

Storage Energy with different Storage value % of Av. Elect. Load

Conventional (Boiler + Grid) Storage Elec. Energy

Total Cost

Purchase Electrical cost

Fig. 7. Total annual cost saving for Scenario 1 at different electrical energy storage capacities compared to the base case (conventional system without electrical energy storage).

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

224

increases with a value of 7% due to decrease AR from 10% to 0.0%. Additionally, Fig. 9 shows selling electricity cost ratio of scenario 3 case studies (9, 10 and 11 in Table 3) compared to case 7 in scenario 2 for AR = 75%. From the result, the electrical energy consumption of sce- nario 3 is higher than scenario 2 due to increase electric power con- sumption by the heat pump. Consequently, the selling electricity cost ratio is lower for scenario 3. Furthermore, installing PVs only (AR = 0.0) provides the highest selling electricity cost ratio of 9.1 (case 11) and the case with AR = 10% has the lowest selling electricity cost ratio with a value of 7.8.

As a result, 25% battery storage size is the optimum case for the first scenario. Moreover, the system with integrated photovoltaic solar pa- nels (AR = 0.0) and 25% electrical energy storage system is the op- timum case study for scenarios 2 and 3. The three optimum cases from scenarios 1, 2 and 3 (Cases 4, 8 and 11 in Table 3) are investigated further for optimum scenario at the community level.

0

1

2

3

4

5

6

7

8

9

10

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

AR = 75% AR = 50% AR = 25% AR = 0%

Se lli

n g

co st

R ati

o

Sa vi

n g

(% )

PVT Area/ Total Area Used

(Boiler + PVT & PV +Grid) Storage 25% AV_elec.

Total Cost

Purchase Electrical cost

Selling Elec. Ratio

Fig. 8. Annual total cost saving (compared to base case) and selling cost ratio for Scenario 2 at different solar panel area ratios.

0

1

2

3

4

5

6

7

8

9

10

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

AR = 10% AR = 5% AR = 0%

Se lli

n g

co st

R ati

o

Sa vi

n g

(% )

PVT Area/ Total Area Used

(HP+ Boiler + PVT & PV +Grid) Storage 25% AV_elec.

Total Cost Purchase Electrical cost Selling Elec. Ratio

Fig. 9. Annual total cost saving and selling cost ratio for Scenario 3 at different solar panel area ratios.

-20%

0%

20%

40%

60%

80%

100%

Scenario 1 Scenario 2 Scenario 3

Sa vi

n g

(% )

Effect of Inverter efficiency on total cost saving

Inverter efficiency 80%

Inverter efficiency 90%

Fig. 10. Effect of inverter efficiency on annual total cost saving for scenarios 1, 2 and 3.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

225

4.4. Effect of inverter efficiency

An inverter is used to convert electricity from AC to DC and vice versa for charging and discharging the electrical storage batteries. It is also used to convert the electric power outlet from the solar panels to AC. In this study, 80% inverter efficiency is assumed to be the same for both charging and discharging the batteries. Fig. 10 shows the effect of

the inverter efficiency on annual total cost savings compared to the reference case. From the results, the inverter efficiency has a significant effect on annual total cost saving for scenarios with electrical energy productions. There is increasing in annual cost saving with a value of ≈ 10% due to increase the inverter efficiency from 80% to 90%. However, there is no significant effect of inverter efficiency on the annual total cost saving for scenario 1. Inverter efficiency with a value of 80% is

a) Electrical energy from Grid

b) Electrical energy from PV panel

c) Electrical energy from battery Storage

0

500

1000

1500

2000

2500

16-Single RB Quick service restaurant Supermarket 1-story small office Primary School

El e

ct ri

ca l E

n e

rg y

(M W

h )

Elec. From the GridsBase Case

Scenario 1

Scenario 2

Scenario 3

0

500

1000

1500

2000

2500

3000

3500

4000

16-Single RB Quick service restaurant Supermarket 1-story small office Primary School

El e

ct ri

ca l E

n e

rg y

(M W

h )

Scenario 2 & 3Elec. Energy from PV

0

5

10

15

20

25

30

35

40

16-Single RB Quick service restaurant Supermarket 1-story small office Primary School

El e

ct ri

ca l E

n e

rg y

(M W

h )

Elect. Energy used from Storage Base Case

Scenario 1

Scenario 2

Scenario 3

Fig. 11. Annual optimum electrical energy distribution at each hub for different scenarios.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

226

used in the flowing presented results.

4.5. Optimum energy at each hub and community

Fig. 11a shows the imported electrical energy from the grid to meet the electrical demand loads for the three optimum cases (4, 8 and 11) of scenarios 1, 2 and 3 and the reference case. The entire electrical energy is imported from the grid to meet the electrical demand loads for re- ference case and case 4 in scenario 1 as these two cases do not have any technology to generate electricity. Part of this imported electricity for case 4 (scenario 1) is used for charging the electrical storage battery during the off-peak hours and discharging it back to the system during the on-peak hours. Fig. 11b illustrates the generated electrical energy from PV solar panels (AR = 0.0) at each hub for cases 8 and 11of scenarios 2 and 3 respectively. Since the five hubs are located under the same weather conditions, the generated electricity is mainly dependent on the hub roof surface area. Consequently, the supermarket and 16 single residential buildings have higher electric power generation from PVs technology compared to other hubs. Fig. 11c shows the electrical energy discharged from the batteries at each hub for the cases 4, 8 and 11. The reference case does not include a storage battery system. The results show that the electrical storage batteries can be used only at the quick service restaurant and the primary school hubs (hubs 2 and 5). Since the heat pumps for scenario 3 consume more electrical energy, the electrical energy storage increases for scenario 3 compared to that for scenarios 1 and 2. However, the electrical storage energies used for scenario 1 and 2 are almost the same where they have the same heating and cooling technology systems.

Fig. 12 shows the annual electrical demand loads and energy dis- tribution at the community level for case studies 4, 8 and 11 of sce- narios 1, 2 and 3, respectively and reference case study. The results show that the annual electrical demand load for scenario 3 is higher than that for scenarios 1 and 2 due to high electric power consumption by the heat pump. The electrical energy is imported from the grid to meet the five hubs demand loads for the reference case and scenario 1. Moreover, the generated electricity from PV systems is the same for both scenarios 2 and 3. The annual selling electrical energy (with a negative value) back to the grid is slightly high for scenario 2 compared to scenario 3. Furthermore, the discharged electrical energy from the batteries is very small and scenario 3 has the highest value compared to other scenarios.

4.6. CO2 emissions for optimum case studies

Fig. 13 displays hourly CO2 emissions for cases 4, 8 and 11 com- pared to CO2 emissions for the reference case. The results show that the

CO2 emissions ratio has a reduction (with a negative value) for the cases 8 and 11 during the daytime where there is electrical energy generation from the PV solar systems. Since the electrical energy is imported from the grid for case 4 and reference case, the CO2 emissions ratio is equal to unity. Moreover, the CO2 emissions ratio has a minimum value of −7 at the middle of the day during week 15 (spring season) for cases 8 and 11, where the thermal and electrical demand loads are low during the shoulder season.

The annual CO2 emissions ratio at the community level (five hubs) for the three scenarios is illustrated in Fig. 14. From the results, the scenario 1 produces the same annual CO2 emissions as the reference case. On the other hand, CO2 emissions ratios for cases 8 and 11 have a value of −1.43 and −1.41, respectively. Therefore, there is a sig- nificant reduction in CO2 emissions from 41% to 43% by employing PV technology in scenarios 2 and 3.

4.7. Economic analysis for optimum case studies

Fig. 15 presents annual total cost saving for scenarios 1, 2, and 3 (cases 4, 8 and 11) compared to the reference case. The results show that scenarios 2 and 3 (cases 8 and 11) achieve annual total cost saving of 81% and 71%, respectively. These values of the annual total cost saving will decrease if the total cost includes installation and labour costs for the technologies and pipelines network. The installation and labour costs have a significant impact on the annual cost and they are changing from one country to another and from one province to an- other. The results show also that the annual total cost for case 4 (sce- nario 1) is higher than that for the reference case due to high capital cost for battery energy storage systems. The cost saving for case 8 (scenario 2) is higher than case 11 (scenario 3) for two main reasons. First one, natural gas is used as an energy source in case 8, which is cheaper than electricity price. The second reason, case 8 has higher selling electric power back to the grid compared to case 11.

The annual breakdown cost for each optimum case studies (Cases 4, 8 and 11) and reference case is presented in Fig. 16. For the base case, purchasing electricity from the grid presents the highest value of 87.8% following by CO2 emissions with a value of 9% and about 2.4% for fuel cost following by chiller, boiler and OM costs with values of 0.7%, 0.1%, and 0.01% respectively. For case 4 (scenario 1), the purchasing electricity and CO2 emissions costs have a value of 87.6% and 8.9%, respectively. These values are slightly different from the reference case because of using electrical storage batteries (Sbat) in scenario 1. The electrical storage batteries cost presents about 0.2% of the total com- munity system cost. However, the fuel, chiller, boiler and OM costs are remaining almost the same as the reference case. On the other hand, integrated PV solar panels to the system have a significant reduction in

-8000

-6000

-4000

-2000

0

2000

4000

6000

8000

10000

12000

Elec. Demand Load for base case,

scenario 1 and 2

Elec. Demand Load for Scenario

3

Base Case Scenario 1 Scenario 2 Scenario 3

El e

ct ri

ca l E

n e

rg y

(M W

h )

Community (20 buildings/ 5 hubs)Selling

Elec. Energy from PV

Elect. Energy used from Storage

Elec. From the Grids

Fig. 12. Annual electrical energy distribution for the community (20 buildings/5 hubs) at different scenarios.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

227

the purchasing electricity cost up to 15.4% and 17.5% for cases 4 (scenario 2) and 11 (scenario 3), respectively. The selling electricity cost to the grid presents the highest percentage of the total cost with values of 40.5% and 38.6% for scenarios 2 and 3, respectively. More- over, PV solar panels costs are 37.1% and 36.6% for scenarios 2 and 3, respectively. CO2 emission cost presents 5.4% for scenario 2 and 5.2% for scenario 3. The heat pump cost is about 1.5% of the total community

energy systems cost for scenario 3 and the chiller in scenario 2 presents 0.3%. However, the boiler and the fuel costs are high for scenario 2 compared to scenario 3 because a small boiler size is using in scenario 3 to meet the DHW heating load. Furthermore, the electrical storage batteries and OM costs for scenarios 2 and 3 have the same percentages.

Additionally, scenario 3 is modeling to investigate the effect of the subsidy on the system cost. The result shows that the subsidy for PV technology can achieve about 3% saving of the annual system cost compared to the system without subsidy.

5. Conclusions

Annual thermal and electrical demand loads for twenty building archetypes grouped in five hubs were modeled and simulated for eco- nomic and environmental optimization analyses. Three scenarios, each employing different energy technologies at hubs level were developed and investigated. General Algebraic Modeling System (GAMS) pro- gramming platform is used to develop a complex optimization model and to investigate distributed energy resource system optimization for minimizing energy economics and environmental impacts at a Canadian community. The annual cost and environmental optimization studies for the three scenarios are compared with the optimum case study for the conventional system (base case).

The results show that mixing building archetypes accomplishes a reduction in equipment capacity sizes at each hub due to load shifting (sharing) between the buildings. The electrical storage capacity with 25% of average electrical load has the lowest annual energy and en- vironmental costs. Installing photovoltaic solar technology achieve up to 80% and 70% of the annual energy cost saving compare to the re- ference case and CO2 emissions saving of 43% and 41% for scenarios 2 and 3, respectively.

Acknowledgments

The author would like to acknowledge Mike Lubun and Raymond Boulter from CanmetENERGY for providing buildings archetype load

-8

-6

-4

-2

0

2

24

C O

2 R

ati o

Time (hr)

Scenario 1

Scenario2

Scenario 3

Week 2

-8

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

24

C O

2 R

ati o

Time (hr)

Scenario 1

Scenario2

Scenario 3

Week 15

-8

-6

-4

-2

0

2

4

24

C O

2 R

ati o

Time (hr)

Scenario 1

Scenario2

Scenario 3

Week 28

-8

-6

-4

-2

0

2

4

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

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

C O

2 R

ati o

Time (hr)

Scenario 1

Scenario2

Scenario 3

Week 37

Fig. 13. Hourly CO2 emissions ratio for different optimum cases of the three scenarios at different seasons.

-2

-1.5

-1

-0.5

0

0.5

1

1.5

Scenario 1 Scenario 2 Scenario 3

C O

2 e

m is

si o

n s

ra ti

o

Community buildings (20 hubs) CO2 emissions ratio

Fig. 14. Annual CO2 emissions ratio from community buildings at different scenarios.

-20%

0%

20%

40%

60%

80%

100%

Scenario 1 Scenario 2 Scenario 3

To ta

l C o

st S

av in

g (%

)

Community (20 buildings /5 hubs)

Fig. 15. Annual total energy cost saving for community buildings at different scenarios.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

228

profiles data and for their technical advice. I am very thankful to Dr. Evgueniy Entchev for his support and advice on this work and Department of Chemical Engineering, University of Waterloo, ON, Canada for kind support on optimization study of this work. Funding for this work was provided by Natural Resources Canada, Canada through the Program of Energy Research and Development.

References

[1] H. Lund, P.A. Østergaard, D. Connolly, B.V. Mathiesen, Smart energy and smart energy systems, Energy 137 (2017) 556–565.

[2] K. Darcovich, H. Ribberink, T. Laurent, Will longer range EVs act as enablers for V2X, in: IEA-HEV Task 28 Scientific Workshop on Vehicle-to-Everything Technology, Barcelona, Spain, June 6–7, 2018.

[3] H. Ribberink, K. Darcovich, Impact on battery life from V2G activity, in: Electric Power Research Institute – National Electric Transportation Infrastructure Working Council Meeting, San Francisco, CA, USA, Nov 16–17, 2016.

[4] T.T. Chow, A review on photovoltaic/thermal hybrid solar technology, Appl. Energy 87 (2010) 365–379.

[5] E. Matallanas, M.C. Cagigal, A. Gutiérrez, Neural network controller for Active Demand-Side Management with PV energy in the residential sector, Appl. Energy 91 (2012) 90–97.

[6] F. Calise, G. Ferruzzi, L. Vanoli, Transient simulation of polygeneration systems based on PEM fuel cells and solar heating and cooling technologies, Energy 41 (2012) 18–30.

[7] E. Barbieri, P. Spina, M. Venturini, Analysis of innovative micro-CHP systems to

meet household energy demands, Appl. Energy 97 (2012) 723–733. [8] J. Kim, W. Cho, K. Lee, Optimum generation capacities of micro combined heat and

power systems in apartment complexes with varying numbers of apartment units, Energy 35 (2012) 5121–5131.

[9] D. Tempesti, G. Manfrida, D. Fiaschi, Thermodynamic analysis of two micro CHP systems operating with geothermal and solar energy, Appl. Energy 97 (2012) 609–617.

[10] H. Ribberink, K. Lombardi, L. Yang, E. Entchev, Investigation of a hybrid renew- able-microgeneration energy system for power and thermal generation with re- duced emissions, Proc. Inst. Mech. Eng. Part A J. Power Energy 227 (1) (2013) 62–72.

[11] Y. Hamada, K. Takeda, R. Goto, H. Kubota, Hybrid utilization of renewable energy and fuel cells for residential energy systems, Energy Build. 43 (2011) 3680–3684.

[12] S. Obara, S. Watanabe, B. Rengarajan, Operation method study based on the energy balance of an independent microgrid using solar-powered water electrolyzer and an electric heat pump, Energy 36 (2011) 5200–5213.

[13] V.O. Pruissen, V.A. Togt, E. Werkman, Energy efficiency comparison of a cen- tralized and a multi-agent market based heating system in a field test, Energy Procedia 62 (2014) 170–178.

[14] R. Ooka, S. Ikeda, A review on optimization techniques for active thermal energy storage control, Energy Build. 106 (2015) 225–233.

[15] Charlotte B.A. Kobus, Elke A.M. Klaassen, Ruth Mugge, Jan P.L. Schoormans, A real-life assessment on the effect of smart appliances for shifting households’ elec- tricity demand, Appl. Energy 147 (2015) 335–343.

[16] A. Ashouri, S. Fux, M. Benz, L. Guzzella, Optimal design and operation of building services using mixed-integer linear programming techniques, Energy 59 (2013) 365–376.

[17] F. Maréchal, B. Kalitventzeff, Process integration: selection of the optimal utility system, Comput. Chem. Eng. 22 (1998) 149–156.

Boiler 0.1%

Chiller 0.7% Fuel cost

2.4%

purchase Elect. 87.8%

OM 0.01%

Emissions 9.0%

Base Case Boiler 0.1%

Chiller 0.7%

Sbat 0.2%

Fuel cost 2.4%

purchase Elect. 87.6%

OM 0.0%

Emissions 8.9%

Scenario 1

Boiler 0.1% Chiller

0.3%

PV 37.1%

Sbat 0.1%

Fuel cost 1.0%

purchase Elect. 15.4%

Selling Elect. 40.5%

OM 0.2%

Emissions -5.4%Scenario 2

Boiler 0.02% Heat Pump

1.5%

PV 36.6%

Sbat 0.1%

Fuel cost 0.3%

purchase Elect. 17.5%

Selling Elect. 38.6%

OM 0.2%

Emissions -5.2%Scenario 3

Fig. 16. Breakdown annual energy cost for optimum and base case studies.

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

229

[18] S. Fazlollahi, F. Maréchal, Multi-objective, multi-period optimization of biomass conversion technologies using evolutionary algorithms and mixed integer linear programming (MILP), Appl. Therm. Eng. 50 (2013) 1504–1513.

[19] C. Weber, F. Maréchal, D. Favrat, Design and optimization of district energy sys- tems, Comput. Aided Chem. Eng. 24 (2007) 1127–1132.

[20] R. Hongbo, G. Weijun, A MILP model for integrated plan and evaluation of dis- tributed energy systems, Appl. Energy 87 (2010) 1001–1014.

[21] M. Eugenia, S. Haralambos, M. Nikolaos, P. Lazaros, A mathematical programming approach for optimal design of distributed energy systems at the neighborhood level, Energy 44 (2012) 96–104.

[22] D. Mehleri, H. Sarimveis, C. Markatos, G. Papageorgiou, Optimal design and op- eration of distributed energy systems: application to Greek residential sector, Renewable Energy 51 (2013) 331–342.

[23] A. Omun, R. Choudhary, A. Boies, Distributed energy resource system optimization using mixed integer linear programming, Energy Policy 61 (2013) 249–266.

[24] Y. Yang, W. Gao, Y. Ruan, J. Xuan, N. Zhou, C. Marnay, Optimal Model of dis- tributed energy system by using GAMS and case study, in: Conference Proceedings of the International Symposium on Sustainable Development of the Asian City Environment (SDACE), 2005.

[25] L. Scala, A. Vaccaro, A. Zobaa, A goal programming methodology for multi objec- tive optimization of distributed energy hubs operation, Appl. Therm. Eng. 71 (2014) 658–666.

[26] A. Parisio, C. Vecchio, A. Vaccaro, A robust optimization approach to energy hub management, Electric Power Energy Systems 42 (2012) 98–104.

[27] F. Brahman, M. Honarmand, S. Jadid, Optimal electrical and thermal energy management of a residential energy hub, integrating demand response and energy storage system, Energy Build. 90 (2015) 65–75.

[28] H. Ren, W. Gao, Y. Ruan, Economic optimization and sensitivity analysis of pho- tovoltaic system in residential buildings, Renewable Energy 34 (2009) 883–889.

[29] H.M. Groscurth, Th. Bruckner, R. Kümmel, Modeling of energy-service supply systems, Energy 9 (1995) 941–958.

[30] D. Henning, MODEST—an energy system optimization model applicable to local utilities and countries, Energy 22 (1997) 1135–1150.

[31] D. Henning, S. Amiri, K. Holmgren, Modelling and optimisation of electricity, steam

and district heating production for a local Swedish utility, Eur. J. Oper. Res. (2006) 1224–1247.

[32] C. Weber, N. Shah, Optimization based design of a district energy system for an eco- town in the United Kingdom, Energy 36 (2011) 1292–1308.

[33] R. Aringhieri, F. Malucelli, Optimal operations management and network planning of a district system with a combined heat and power plant, Ann. Oper. Res. 120 (2003) 173–199.

[34] A. Hepbasli, Thermodynamic analysis of a ground-source heat pump system for district heating, Int. J. Energy Res. 29 (2005) 671–687.

[35] B. Rolfsman, Combined heat and power plants and district heating in a deregulated electricity market, Appl. Energy 78 (2004) 37–52.

[36] A. Rieder, A. Christidis, G. Tsatsaronis, Multi-criteria dynamic design optimization of a small scale distributed energy system, Energy 74 (2014) 230–239.

[37] E. Fabrizio, V. Corrado, M. Filippi, A model to design and optimize multi-energy systems in buildings at the design concept stage, Renew Energy 35 (2010) 644–655.

[38] A. Maroufmashat, A. Elkamel, M. Fowler, S. Sattari, R. Roshandel, A. Hajimiragha, S. Walker, E. Entchev, Modeling and optimization of a network of energy hubs to improve economic and emission considerations, Energy 93 (2015) 2546–2558.

[39] B. Abushakra, A. Sreshthaputra, J.S. Haberl, D.E. Claridge, Compilation of diversity factors and schedules for energy and cooling load calculations, ASHRAE Research Project 1093-RP Final Report, Energy Systems Laboratory Technical Report, ESL- TR-0I/04-01, Department of Mechanical Engineering, Texas A&M University, February, 2002.

[40] E.C. Kang, E.J. Lee, M. Ghorab, L. Yang, E. Entchev, K.S. Lee, N.J. Lu, Investigation of energy and environmental potentials of a renewable trigeneration system in re- sidential applications, Int J Energies (9) (2016) 760.

[41] General Algebraic Modeling System, < http://www.GAMS.com/ > (GAMS web- site).

[42] A. Soroudi, Power System Optimization Modeling in GAMS, book, Springer, 2017. [43] E. Entchev, L. Yang, M. Ghorab, E.J. Lee, Simulation of hybrid renewable micro-

generation systems in load sharing applications, Energy 50 (2013) 252–261. [44] Enbridge Gas, < http://www.enbridgegas.com > (accessed June 2017). [45] Hydro-one, < http://www.hydroone.com > (accessed June 2017).

M. Ghorab Applied Thermal Engineering 151 (2019) 214–230

230