A model for energy master planning and resilience assessment of net-zero emissions community

ABSTRACT

New community-scale developments should address both greenhouse gas emissions mitigation and climate adaptation goals. This paper presents a systematic approach to energy master planning (EMP) of net-zero emissions communities via probabilistic analysis of the resilience and cost effectiveness of various energy provision portfolios (supply, conversion and storage) in early design stage. Applied in the EMP of a new university satellite campus, comprising of five buildings with mixed energy uses, both the 2050 net-zero emissions and the energy resilience objectives are met by an energy provision portfolio that consists of air source heat pumps for heating and cooling, and a combination of PV panels, purchased green power, standard (non-green) grid power, battery and thermal heat and cold storage tanks – with only a modest 6% increase in costs compared to a reference solution. The case project demonstrates the financial feasibility of a resilient energy system that also meets a net-zero emissions objective.        

KEYWORDS:

• Net-zero emissions community
• energy master planning
• resilience
• performance-based decision-making
• Monte carlo simulation
• sustainable communities

1. Introduction

The scale of the challenge to keep global warming due to greenhouse gas (GHG) emissions within 1.5 to 2°C requires emissions reduction of urban and built environment projects at scale. Net-zero emissions community-level energy systems provide significant contributions compared to an individual building or land parcel project (International Energy Agency, 2014; Jank, 2017; Sharifi & Yamagata, 2016). At the same time, these systems must also be resilient and adaptive to increasing climate risks. Both net-zero emissions and climate resilience objectives need to be addressed in the project’s early planning and design stages to have a higher impact. Herein, a community-level development is defined as one that is owned, organised or managed by a sole decision-maker body and consists of mixed-use and co-located buildings with on-site generation units connected to both internal energy networks and national energy grids to meet both thermal and electrical energy demands. Achieving a resilient and net-zero emissions community (NZEC) requires a comprehensive and systematic process of energy master planning (EMP) for efficient use of resources and technologies.

Several models and tools suitable for local energy planning using distributed energy resources have been developed (Ma, Wu, Hao, et al., 2018; Mehleri, Sarimveis, Markatos, et al., 2013; Morvaj, Evins, & Carmeliet, 2016; Sameti & Haghighat, 2018; Wouters, Fraga, & James, 2015) and explored (Allegrini, Orehounig, Mavromatidis, et al., 2015; Charani Shandiz, Rismanchi, & Foliente, 2021; Connolly, Lund, Mathiesen, et al., 2010; Groissböck, 2019; Mendes, Ioakimidis, & Ferrão, 2011) models and tools suitable for local energy planning using distributed energy resources, but an integrated EMP approach that addresses both GHG mitigation and resilience goals considering technical system configurations in the early design stage is lacking. Other works have introduced high-level approaches for precinct-level EMP (Santillan, Syn, Charani Shandiz, et al., 2022). However, the current approach to EMP is not systematic (or non-integrated) and typically ad hoc influenced by the designers and/or energy system planners’ own experience and/or expertise (building footprint focused vs. energy network focused) (Charani Shandiz, Rismanchi, & Foliente, 2021; Huang, Yu, Peng, et al., 2015; Jank, 2017; Vogt, 2003). In most energy system planning studies, the typical focus has been on optimising a predefined problem within a given set of financial constraints. Conventional approaches hinder the exploration of new and innovative and more sustainable solutions. A performance-based approach – that focuses on the required performance (the ‘outcomes’, which, in this paper, is a climate-resilient NZEC), rather than the solution (the ‘means’) (G. C. Foliente, Huovila, Spekkink, et al., 2005; G. Foliente, 2000) – framed against a set of technical performance, environmental and financial/economic objectives (i.e., multi-dimensional) provides a systematic method of exploring the performance of traditional and new and innovative solutions on a project by project or context-based manner. Finally, the availability of modelling tools that consider both building and energy grid perspectives and enable their impact assessments across the set of performance objectives would complete the required components of an integrated EMP for NZECs.

Unfortunately, the current tools, models and approaches for EMP at the community scale are few and case-specific (Charani Shandiz, Rismanchi, & Foliente, 2021). Approaches, models and tools are often narrow in scope (i.e., model scale, buildings typology, energy demand, etc.) and assessment criteria. In particular, net-zero emissions and energy resilience objectives are not explicitly common joint considerations for NZECs. The list or database of technology options is often narrow and limited to a few business-as-usual options. Given the current impacts of climate-related extreme events, ignoring energy system resilience would lead to elevated risks to community users, owners and investors. Energy resilience assessment and quantification in the early stages of the design is not commonly undertaken (Charani Shandiz, Foliente, Rismanchi, et al., 2020), or limited only to high-level climate risk identification and adaptation planning.

The authors’ previous works present systematic EMP and energy resilience frameworks (Charani Shandiz, Foliente, Rismanchi, et al., 2020; Charani Shandiz, Rismanchi, & Foliente, 2021). The key factors and steps include: (a) explicit identification of client objectives and considering the relevant regulatory requirements, project context data, technologies database, and the adopted NZEC definition or scope; (b) planning approach, engineering formulation/tools for planning, operation and design; (c) system design and assessment against NZEC definition; (d) multi-dimensional assessment of selected metrics; and (e) selecting the acceptable technology options using the appropriate decision-making process (Charani Shandiz, Rismanchi, & Foliente, 2021). The energy resilience framework comprises three layers of resilience (engineering-designed, operational and community-societal) (Charani Shandiz, Rismanchi, & Foliente, 2021). An initial set of energy resilience metrics has previously been identified for planning and assessment, especially for EMP of community systems.

This article presents the development and application of a multi-dimensional (considering environmental, financial and engineering/technical aspects) EMP model and an expansive technology options database (containing both technical and financial data) that can be employed in the early stages of EMP to support decision-making to meet both net-zero emissions and energy resilience objectives at the community level. The model enables probabilistic analysis of system resilience performance in the early stages of the design. The key components and capabilities of the performance-based approach to EMP of NZEC are demonstrated in the development of a new University of Melbourne campus in Australia, paving the way for further developments and applications that address both climate mitigation and adaptation goals in community-level and/or precinct-wide projects. This paper is part of a PhD Thesis where further details regarding the model, data sets, assumptions and more are elaborated (Charani Shandiz, 2022).

1.1. Research scope, key definitions and limitations

The EMP at the community level is defined herein as a process of planning, design and performance evaluation of the energy system(s) and its operation to meet the community’s financial, environmental and social aspects. Therefore, the energy master plan sits in the context of the physical site boundary and its surroundings, considers the relevant range and mix of demand and supply scenarios, and is herein proposed to be a key component to the more common (community) project master plan” (Charani Shandiz, Rismanchi, & Foliente, 2021).

A ‘community-level energy system’ is a group or cluster of newly built (and/or existing) buildings with mixed uses located near each other, connected to national energy grids (and an internal grid). The entire system is considered as a whole with one owner (or organiser) in charge of decision-making. It comprises physical, administrative and GHG emissions balance boundaries. Although the term ‘community’ in a social context has a different meaning, hereby ‘community’ means the focus is on the medium-scale energy system, precinct or campus, both greenfield or brownfield. Some examples include a university campus, school, hospital, mall, etc. In NZEC, the energy uses are met by on-site and/or off-site renewable energy sources.

At the community level, there are a variety of energy supply and energy storage options as well as energy networks available, see Figure 1. The goal is to understand how the ‘planning, design and performance of the energy system’ can lead to meeting the required performance of the community, including technical, financial, and environmental aspects.

Figure 1. Schematic concept of community-level energy system and examples of energy provision technologies.

Community EMP has three stages: conceptual design, preliminary design and detailed design (Charani Shandiz, Rismanchi, & Foliente, 2021). The focus here is on the conceptual and preliminary design phases. Although data is often limited at the early stages of design, there is greater scope and flexibility in exploring NZEC technology and design solutions, and thus, can be a very valuable exercise.

Most of the cities and communities are located in moderate climates. Extreme climates usually have a low population density and require specific design and technologies for energy-related projects. Therefore, the focus here is on temperate climate regions which are categorised as C based on the Koppen-Geiger climate classification (Peel, Finlayson, & McMahon, 2007). The EMP model is not applicable to energy-intensive industrial processes (e.g., refinery, smelting, etc.). Also, the model considers the energy demands on a lumped node and it does not consider the nodal energy flows. These and additional features should be considered in further development of the model. One of the limitations of this work is the scarcity of measured data in early-stage development. Although the EMP model can be applied to both brownfield and greenfield, for the selected demonstration study a greenfield is considered where the authors have been engaged in the planning and design process.

2. Model development

2.1. Resource availability potential calculation methods

Available on-site resources such as variable renewable energy should be estimated separately to be used as an input to the EMP model. This allows for lower data and computation power requirements and, therefore, lower time and cost requirements. Local resource availability databases can be developed in the future to facilitate this even further.

The model takes these parameters, which are mainly dependent on external parameters, as time series (associated with the time step of the simulation). The resolution of the inputs depends on the design/client/project requirements. The following sections describe the calculation methods for resource availability potential.

2.1.1. Solar potential

The potential energy output of solar PV can be estimated using solar radiation on different potential planes and technology specifications in Equation 1(1) $\overline{E_h}=S\eta \overline{H_h}$(1) .

(1) $\overline{E_h}=S\eta \overline{H_h}$(1)

where $S_{\hspace{0.17em}}$is the area of the array (assume one m²), ${\eta }_{\hspace{0.17em}}$ average efficiency of technology, $\overline{H_h}$ average hourly solar radiation on the plane of the module, and $\overline{E_h}$ average hourly solar PV/thermal collector output. The area [m²] required for converting one kW of solar energy to one kW output electricity or thermal energy can be calculated accordingly considering the space required for additional components for these technologies, see Appendix B.

2.1.2. Wind potential

Output power of the wind turbine is estimated based on the wind velocity and typical community-scale wind turbine technical details. The wind speed at the hub height is estimated using Equation 2(2) $V_2=V_1.(Z_2/Z_1{)}^a$(2) .

(2) $V_2=V_1.(Z_2/Z_1{)}^a$(2)

where $Z_1$ is Height 1 (lower height), $Z_2$ Height 2 (upper height), $V_1$ velocity at height $Z_1$, $V_2$ is velocity at height $Z_2,$and $a$ is wind shear factor (0.26 suggested in (Thermal Energy System Specialists) for obstructed surfaces). Based on the meteorological stations’ standards, the wind speed is usually measured at 10 m in height.

2.1.3. Geothermal potential

Geothermal energy potential for the use of heat pumps is considered in the higher coefficient of performance (COP) of ground source heat pump (GSHP) compared to non-ground source alternatives. However, the ground heat capacity should be estimated by assuming the energy output capacity per pile (which depends on the length of piles, ground characteristics, and so on) and the number of piles fitting in the area per square meters.

2.2. Technology database

A technology database is introduced herein in order to support EMP at the community level. The technical and financial features of a wider range of community-level technologies (for heating, cooling and electricity provisions) including energy supply, conversion and storage technologies are collected, which are necessary for the design and assessment of the technology portfolio. Data are collected from the literature, technical catalogues and different national and international inventories and cross-checked. A brief description of the database items is provided in Appendix A.

The technology database is introduced in Appendix B. The cost and technical specification of technologies vary depending on the market supply, installation, type/magnitude of the project, location and site specifications (urban/remote, sloped/flat area, etc.), profit margins, taxes, technology advancements, and more. The technology database provides the basis of several technologies and should be updated for each specific project to include context-specific information.

The cost of energy distribution (e.g., pipes and wires) and control technologies in the community are not in the scope of this work. These costs are often negligible considering the capital and operational costs in energy provision. In the future, the EMP model can be enhanced to include the cost of spatial modelling features and energy distribution technologies.

3. Energy master planning model formulation

The EMP model for the planning, design and assessment of communities is introduced in this section by incorporating the EMP and energy resilience frameworks introduced by (Charani Shandiz, Foliente, Rismanchi, et al., 2020; Charani Shandiz, Rismanchi, & Foliente, 2021). Figure 2 shows the EMP model algorithm in a flowchart.

Figure 2. Flowchart of the EMP model algorithm.

The EMP model condition in step a of Figure 2 for net-zero emissions case is a zero emissions constraint. The net-zero balance calculation method depends on the local/national regulations or voluntary certifications specifications; the balance calculation methods are discussed in (Charani Shandiz, Rismanchi, & Foliente, 2021). The community conditions and inputs are added in Step (a). The problem formulations are in Step (b) for identifying optimal sizing, configuration and system operation strategy. In Step (c), the feasible solution found should comply with the conditions set in the previous steps. A multi-dimensional assessment of selected metrics of the performance of the selected community-level energy system option is performed in Step (d). The outputs and results are in Step (e).

Various components of the EMP model are described in the following sections.

3.1. Model structure

A schematic diagram of the model structure is shown in Figure 3. The energy demand is satisfied by energy supply from a combination of energy grid, on-site renewable energy sources (RES), fuels that sometimes need to be converted through a process to the specific commodity, or otherwise directly feed the demand. Energy can also be stored or exported to the energy grid.

Figure 3. EMP model structure.

The EMP model is a revision and reconstructed version of Ficus, which is itself a derivative of URBS (Atabay, 2017b; Dorfner, Schönleber, Dorfner, et al., 2019). The source code is obtained from an online open-source platform (GitHub repository) under GNU General Public License (Atabay) and improved following the frameworks proposed in (Charani Shandiz, Foliente, Rismanchi, et al., 2020; Charani Shandiz, Rismanchi, & Foliente, 2021).

Ficus’s objective is to find the optimal design capacity of the alternative technologies (generation and storage) as well as the optimal operation of all energy supply and storage technologies to satisfy the energy demand in each time step throughout the year with minimal cost or emission while respecting all constraints (e.g., technology, area and so on). Ficus is primarily used for industrial energy systems but has the flexibility to be applied to other community-level energy systems. It is mixed-integer linear programming (MILP) based with the main objective of minimising the investment and operational costs, and is a deterministic optimisation model.

The main additions to and revisions of Ficus’s original code include GHG emissions objective function, area constraints, carbon price, and resilience assessment module. GHG emissions objective function is added, which is to find the optimal design and operation of the system by minimising emissions (see Section 2.3.2). Besides the demand and supply constraints and other constraints described in (Atabay, 2017b), emissions and cost can be used as a constraint for emissions (emissions budget) and cost budget. Also, specific area breakdown availability for the energy supply and storage is another constraint consideration (see Section 2.3.3). A carbon price mechanism is added to the source code. The energy resilience assessment module is elaborated in Section 2.4.4.

The upgraded EMP model is written in the open-source environment Python (v 3.7)/Pyomo. In this research, the Academic version of Gurobi (v 9.1.2) is used as the solver, but also CPLEX and GNU Linear Programming Kit(GLPK) solvers can be used. Inputs and outputs are in spreadsheets (Excel). The inputs and outputs of the EMP model are described in Figure 4. The model’s planning horizon is assumed to be one year with hourly time steps. Different planning horizons (e.g., short-term or long-term) and time steps (e.g., multiple hours or minutes) can be used in the model. This depends on the period of balance, level of details/sensitivity and computing power available.

Figure 4. Inputs and outputs of the EMP model.

A summary description of model sets is provided in Table 1. To avoid repetition, only the key formulations and assumptions as well as the additions to the source code, are elaborated in the following sections. More detailed information about Ficus are presented in (Atabay, 2017b).

Table 1. Description of model sets.

Set	Description
tXT	Time step and period
cXC	Commodities
sXS	Storage units
pXP	Processes
dXD	Energy demand
E	Emissions

3.2. Objective functions

Either of the two objective functions available in the model can be selected and are described as in Equation 3(3) $Min{C}^{tot}=\text{min}\left(\sum C^{inv}+C^{O\&M}{C}^{imp}+C^{\exp }+C^{netw}+C^{cp}+C^{subs}\right)$(3) .

Cost objective:

(3) $Min{C}^{tot}=\text{min}\left(\sum C^{inv}+C^{O\&M}{C}^{imp}+C^{\exp }+C^{netw}+C^{cp}+C^{subs}\right)$(3)

where C^tot is total annualise cost, C^inv is the annualised investment cost, C^O&M is annual operation and maintenance cost, C^imp is annual energy import cost, C^exp is energy export revenue, C^netw is annual network charges, C^cp is the annual carbon price cost (or other emissions related fees) and C^subs is annual subsidies (or any other emissions related incentives).

The objective function is the total annualise cost. The energy export cost is assumed to be negative. All costs are annual value (e.g., O&M) or are annualised by using the annuity factor. The annuity factor or the present value of a cash flow is to calculate the equivalent annual cost to convert the investment cost over a period into the present value. This is calculated for different components separately, considering their economic life. One of the limitations here is that the weighted average cost of capital (WACC) is considered constant for the entire length of the project:

(4) $ANF=\hspace{0.17em}\frac{{\left(1+r\right)}^n.r}{{\left(1+r\right)}^n-1}$(4)

where n is the depreciation period (economic life span, here is assumed to be equal to the life span of the technology) [years], and r is WACC [%].

Emissions objective:

(5) $Min\hspace{0.17em}{E}^{tot}=\mathrm{m}\mathrm{i}\mathrm{n}(\sum \left(E_{\hspace{0.17em}}^{imp}+E_{\hspace{0.17em}}^{exp})\right)$(5)

(6) $E_t^{imp}=\hspace{0.17em}\sum _t^{\hspace{0.17em}}\sum _c^{\hspace{0.17em}}{p}_{c,t}^{imp}\times \hspace{0.17em}e{f}_{c,t}$(6)

(7) $E_t^{exp}=\hspace{0.17em}\sum _t^{\hspace{0.17em}}\sum _c^{\hspace{0.17em}}-p_{c,t}^{exp}\times \hspace{0.17em}e{f}_{c,t}$(7)

Where E^tot is the total annual GHG (CO₂ equivalent) emissions, E^imp is the annual emissions related to energy import and E^exp is the emissions offsets for importing clean energy to the grid (i.e., emissions avoided). The imported (Equation 6(6) $E_t^{imp}=\hspace{0.17em}\sum _t^{\hspace{0.17em}}\sum _c^{\hspace{0.17em}}{p}_{c,t}^{imp}\times \hspace{0.17em}e{f}_{c,t}$(6) ) and exported (Equation 7(7) $E_t^{exp}=\hspace{0.17em}\sum _t^{\hspace{0.17em}}\sum _c^{\hspace{0.17em}}-p_{c,t}^{exp}\times \hspace{0.17em}e{f}_{c,t}$(7) ) related emissions are calculated at each time step t and for each commodity c for the amount of energy p and the relevant emissions factor ef_c,t at the time step t and for commodity c. These are summed over the entire period of analysis.

Another objective added to Ficus is: minimising the operational GHG emissions associated with energy consumption in the community. As discussed earlier in the NZEC definition, the formulation of the emissions objective directly impacts the EMP of the community. When zero emission is the objective (emissions offset does not count) the objective function is as given in Equation 8(8) $Min\hspace{0.17em}{E}^{tot}=\mathrm{m}\mathrm{i}\mathrm{n}(\sum (E_{\hspace{0.17em}}^{imp}))$(8) (with no energy export component):

(8) $Min\hspace{0.17em}{E}^{tot}=\mathrm{m}\mathrm{i}\mathrm{n}(\sum (E_{\hspace{0.17em}}^{imp}))$(8)

In this case, no emissions should be generated to achieve the zero emissions objective.

3.3. Constraints

Conservation of energy (power/energy balance) is a key constraint. The sum of energy and power imported, exported, generated, consumed, stored in or stored out, and losses has to be zero for every commodity in every time step over the analysis period, see Equation 9(9) $\sum _c^{\hspace{0.17em}}{\rho }_{p,c,t}^{out}+\sum _{c,\hspace{0.17em}s}^{\hspace{0.17em}}{\rho }_{s,c,t}^{out}+\hspace{0.17em}{\rho }_{c,t}^{imp}=\sum _c^{\hspace{0.17em}}{\rho }_{p,c,t}^{in}+\hspace{0.17em}\sum _c^{\hspace{0.17em}}{\rho }_{s,c,t}^{in}+{\rho }_{c,t}^{exp}+d_{c,t}$(9) .

(9) $\sum _c^{\hspace{0.17em}}{\rho }_{p,c,t}^{out}+\sum _{c,\hspace{0.17em}s}^{\hspace{0.17em}}{\rho }_{s,c,t}^{out}+\hspace{0.17em}{\rho }_{c,t}^{imp}=\sum _c^{\hspace{0.17em}}{\rho }_{p,c,t}^{in}+\hspace{0.17em}\sum _c^{\hspace{0.17em}}{\rho }_{s,c,t}^{in}+{\rho }_{c,t}^{exp}+d_{c,t}$(9)

where the sum of power entering or produced by processes, storage units and energy import is equal to the sum of power exiting or consumed by processes, storage units, energy export and energy demand. ${\rho }_{p,c,t}^{out}$ is the power flow of the commodity c at time step t out of the process, ${\rho }_{p,c,t}^{out}$ is the power flow of the commodity c at time step t out of the energy storage and ${\rho }_{c,t}^{imp}$ is the power flow of commodity c imported at time step t. The other side of Equation follows the same naming convention where in stands for entering, exp for energy export and d for energy demand.

Area constraint means that the sum needed area on-site for installation of the RES and mechanical plant for each development stage of Fishermans Bend (FB), including unoccupied areas, rooftops, car parks and so on, must not be greater than the available area:

(10) $\sum \left(A_{expos}+A_{ug}+A_{site}+A_{in}\right)\le A_{tot}^{ava}$(10)

(11) $\sum \left(A_{expos}\right)\le A_{expos}^{ava}$(11)

(12) $\sum \left(A_{ug}\right)\le A_{ug}^{ava}$(12)

where $A_{tot}^{ava}$ is total available area, $A_{expos}$ occupied exposed area (e.g., rooftop), $A_{ug}$occupied area underground, $A_{site}$occupied area on site, and $A_{in}$ occupied area indoor. The superscript $av{a}_{\hspace{0.17em}}$refers to the available area in each category. Cost or emissions budget limits can be added or released depending on the case. Commodities and technologies are constrained by technical operational (e.g., efficiency, lifespan, and so on) and design capacity limits (e.g., the minimum and maximum capacity).

Various energy careers availability, energy network (e.g., electricity, gas and so on) connected or island (i.e., not connected to energy networks) scenarios can be modelled by adjusting the constraints in the model.

3.4. Multi-dimensional performance-based metrics and assessment

The two key objectives that the EMP model needs to address are: (a) how to reduce operational GHG emissions related to energy consumption and (b) how to evaluate the community-level energy system’s resilience. However, to assess the community energy system’s performance comprehensively, three high-level factors for comparisons of different design options and final decision-making (or, what will be referred from hereon as three bottom lines) are selected. For each bottom-line, the main metrics (or indicators) are defined. Indicators can usually be selected based on the client’s requirements, best practice guidelines, sustainability scheme recommendations, or certification or rating scheme requirements. Central to this is the net-zero emissions and resilience targets set for 2050. However, additional metrics can be added to the model when other aspects of the system need to be evaluated. Three (Level 1) dimensions or bottom lines and the relevant or supporting metrics selected for this paper are summarised in Table 2.

Table 2. Selected indicators under the three (Level 1) dimensions or bottom lines considered in the EMP model.

Level 1	Level 2	Level 3
Environmental	Carbon footprint	Operational GHG emissions [kgCO2e/year]
	Resource depletion	Land occupied [m^2]
Financial	Equivalent annualised cost [AUD/year]	Capex [AUD]
		Opex [AUD/year]
		Revenues [AUD/year]
Engineering/technical	Share of RES	Energy used [kWh]
		Energy delivered [kWh]
	Engineering-designed resilience	Critical load not served [kWh]
		Duration of critical load not served [hr]

Although the social criterion is a particularly important aspect of sustainability and stakeholder decision-making, it is outside the scope of the present work. In addition, it should be mentioned that identifying the weighting and aggregation methods (Gan, Fernandez, Guo, et al., 2017) for multi-criteria decision-making methods and finding the optimal decision is also out of the scope of this work. This should be dealt with using appropriate methods such as Analytical Hierarchy Process, Preference Ranking Organization Method for Enrichment of Evaluations (PROMETHEE), Elimination And Choice Translating Reality (ELECTRE) method, and so on (Pohekar & Ramachandran, 2004). The metrics definition and assessment methods are described in the following sections.

3.4.1. Environmental performance assessment

Four primary metrics are selected that can represent the decision-maker’s objectives and to provide an appropriate evaluation of the environmental impacts of design solutions. A special focus is on the operational GHG emissions.

3.4.1.1. Operational GHG emissions [kg CO₂e/year]

The operational GHG emissions are related to the amount of energy related to operation of the community level energy system used from each energy carrier over a year. It can be evaluated as follows:

(13) $E^{tot}=\sum (E_{\hspace{0.17em}}^{imp}+E_{\hspace{0.17em}}^{exp})$(13)

(14) $E_t^{imp}=\hspace{0.17em}\sum _t^{\hspace{0.17em}}\sum _c^{\hspace{0.17em}}{p}_{c,t}^{imp}\times \hspace{0.17em}e{f}_{c,t}$(14)

(15) $E_t^{exp}=\hspace{0.17em}\sum _t^{\hspace{0.17em}}\sum _c^{\hspace{0.17em}}-p_{c,t}^{exp}\times \hspace{0.17em}e{f}_{c,t}$(15)

The impact of fuel switching is automatically considered in the emissions balance calculation. That is, when one source of energy (e.g., gas) is substituted with another (e.g., grid or renewable source electricity), their respective emissions are carried over in the calculation (with those associated with renewable source electricity accounted for as avoided emissions).

To estimate the associated operational GHG emissions (from energy consumption in buildings, explained in the previous section), emissions factor for each energy carrier should be considered. Emissions factors depend on the emissions intensity of energy production. Therefore, it varies in different locations and at different times. Producing energy from clean renewable sources results in having an operational emissions factor equal to zero.

3.4.1.2. Land use [m²]

The land requirement for each technology at the community level energy system. This can be calculated by summing up the area that each technology occupies:

(16) $A_{tot}^{ava}=\sum (A_{expos}+A_{ug}+A_{site}+A_{in})$(16)

(17) $A_{\hspace{0.17em}}^{\hspace{0.17em}}=A{f}_c^{\hspace{0.17em}}.P_c^{\hspace{0.17em}}$(17)

where $A{f}_c^{\hspace{0.17em}}$ [m²/kW] is the area factor for each technology and $P_c^{\hspace{0.17em}}$ [kW or kWh] is the technology designed capacity/size.

3.4.2. Financial performance assessment

The annualised cost is considered as the primary metric for financial performance. The secondary metrics for calculating the annualised cost are capital expenditures per year, operational expenditures per year, and revenue per year. The third level metrics for operational expenditures are the fix and variable costs and demand charges, taxes, and so on. The selected metrics are based on typical client/project requirements.

3.4.2.1. Equivalent annualised cost [AUD/year]

This is to calculate the total cost of all technologies and systems operation. To standardise and convert all costs into an equivalent cost the capex is multiplied by an annuity factor, while all other annual costs – such as fixed and variable operation costs, energy import, cost of peak demand, fees/taxes – are added and the revenues/energy export and incentives/subsidies are subtracted.

(18) $Equivalent\hspace{0.17em}annulised\hspace{0.17em}cost=\sum _i^{\hspace{0.17em}}{C}_{investment}\times ANF+\hspace{0.17em}{C}_{fixed\hspace{0.17em}opex}+C_{variable\hspace{0.17em}opex}+C_{import}+C_{network}+C_{fee\hspace{0.17em}or\hspace{0.17em}tax}-C_{revenue,export}-C_{incentives,subsidies}$(18)

3.4.2.2. Capital expenditures (Capex) [AUD]

Includes all investment costs related to energy supply and distribution, such as the cost of technologies, pipes and wires (sum of the capital costs).

3.4.2.3. Operational expenditures (Opex) [AUD/year]

Includes all costs related to the Operation and Maintenance (O&M) of energy supply and distribution systems, such as energy costs, water consumption costs, cost of maintenance, and labour for operation.

3.4.2.4. Revenues [AUD/year]

Revenues related to exporting the excess electricity to the energy grid.

3.4.3. Engineering performance assessment

Three indicators are selected to evaluate the engineering/technical performance of the community level energy system.

3.4.3.1. Energy use [kWh]

Energy used by various services of the community at each time step, divided into electricity, heating and cooling. Energy use is more an indicator of energy efficiency, while energy source shows RES/fossil. All energy uses related to the buildings, including heating, cooling, electricity (appliance energy, plugs, etc.) uses, should be considered. Specific energy uses such as transportation uses, electric vehicle or industrial processes are excluded. Energy uses of each building should be considered as follows:

(19) $Q_{tot,\hspace{0.17em}c}=Q_{b1,\hspace{0.17em}c}+Q_{b2,\hspace{0.17em}c}+\dots +Q_{bn,\hspace{0.17em}c}$(19)

$Q_{bn,c}=Q_{in,c}\times efficiencyofeachsystemortechnology$

where Q_b stands for the energy consumption of individual buildings and Q_tot energy demand of the entire community. Q_c refers to energy uses including heating, cooling and electricity, and Q_in,c refers to the fuel/resource used in each process.

3.4.3.2. Energy produced/source [kWh]

Energy produced by each technology at each time step for each energy source.

3.4.3.3. Share of RES [%]

The ratio of energy demand answered by RES is one of the key criteria in sustainability rating systems and in some countries their legislation even considers a minimum mandatory share of RES for each building/community.

(20) $Share\hspace{0.17em}of\hspace{0.17em}RES=RE\hspace{0.17em}delivered/total\hspace{0.17em}energy\hspace{0.17em}delivered$(20)

3.4.4. Energy resilience assessment

The multi-layered energy system resilience framework and metrics proposed in (Charani Shandiz, Foliente, Rismanchi, et al., 2020) is incorporated to enable quantification of the energy resilience. The resilience evaluation is added to the EMP model as a post-optimisation module, enabling a probabilistic quantification of energy resilience. The required parameters and inputs for the resilience assessment are described in Table 3.

Table 3. Energy resilience input parameters description.

Parameter	Description
Critical energy demand [kW]	Energy required for the critical part of demand (electricity, heating, cooling)
Type of hazard/outage	Type of outage and disrupted technologies can be grouped by hazard type (e.g., storm, flood and so on)
Disrupted service	The commodity/service (electricity, heating, cooling) disrupted due to the outage
Starting hour of disruption	The time that disruption starts
Duration of disruption [hour]	The length of the disruption
Percentage of generation available	The amplitude/intensity of disruption (impact of disruption)

To quantify energy resilience, the outage caused due to a hazard/disruption to the energy supply network(s) and/or technology(ies) are reduced to the percentage of generation available of the community system operation from the starting hour of the disruption, for the duration of the disruption. The other available technology supply and storage options that are not disrupted will be functioning to satisfy the critical energy demand. Accordingly, ‘critical energy not served (kWh)’ and ‘duration of critical energy not served (hr)’ as the key selected metrics are quantified using the resilience trapezoid shown in Figure 5. However, the flexibility of the model allows for the addition of other metrics (see (Charani Shandiz, Foliente, Rismanchi, et al., 2020; Charani Shandiz, Rismanchi, & Foliente, 2021)) when needed and data is available. The two key selected metrics used and their formulations are described as follows using the resilience trapezoid shown in Figure 5.

Figure 5. Energy resilience module workflow based on (Charani Shandiz, Foliente, Rismanchi, et al., 2020).

The critical load not served (CLNS) is defined as the amount of critical demand lost in the system due to the climatic disruption, i.e., the area under the resilience trapezoid. In case of the preparation measures and energy storage options, the backup energy stored in the storage (A_s), sum of S_c,t over the period of analysis, should be subtracted from the area under the trapezoid:

(21) $CLN{S}_{c,t}^{\hspace{0.17em}}=\left(\sum _{\hspace{0.17em}}^{\hspace{0.17em}}{p}_{c,t}^{\hspace{0.17em}}\hspace{0.17em}\right)-\hspace{0.17em}{d}_{c,t}^{\hspace{0.17em}}\hspace{0.17em}+\left(s_{c,t}^{\hspace{0.17em}}\right)\hspace{0.17em}\left[kW\right]$(21)

This turns to the following formula for the entire duration of disruption:

(22) $CLN{S}_c^{\hspace{0.17em}}=\sum _T^{\hspace{0.17em}}(CLN{S}_{c,t}^{\hspace{0.17em}})=\underset{t_0}{\overset{t_f}{\int }}{R}_{t}dt=A_1+A_2+A_3+\left(-A_s\right)\hspace{0.17em}\left[kWh\right]\hspace{0.17em}$(22)

The duration of critical load not served (DCLNS) is defined as the duration of critical demand lost in the system due to the climatic disruption, i.e., the length of the long base of resilience trapezoid. In case of a full back-up storage system, the energy supplied by the storage (L_s) must be included in DCLNS calculation:

(23) $DCLN{S}_{c,t}^{\hspace{0.17em}}=\left\{\begin{array}{c}=1,if{p}_{c,t}^{\hspace{0.17em}}-d_{c,t}^{\hspace{0.17em}}+(s_{c,t}^{\hspace{0.17em}})<0\\ =0,if{p}_{c,t}^{\hspace{0.17em}}-d_{c,t}^{\hspace{0.17em}}+(s_{c,t}^{\hspace{0.17em}})\ge 0\end{array}\hspace{0.17em}\left[hr\right]\right.$(23)

This turns to the following formula for the entire duration of disruption:

(24) $DCLN{S}_c^{\hspace{0.17em}}=\sum _T^{\hspace{0.17em}}DCLN{S}_{c,t}^{\hspace{0.17em}}=\hspace{0.17em}{t}_f-t_1+\left(-L_s\right)=\hspace{0.17em}{L}_1+L_2+L_3+\left(-L_s\right)\hspace{0.17em}\left[hr\right]$(24)

If a storage system is available in the technology portfolio, it is assumed the energy storage is full at the commencement of each disruption as most climatic related disruptions can be predicted shortly before and therefore the storage can be charged fully as a preparation measure, see Figure 5.

In probabilistic energy resilience analysis via Monte Carlo simulation (MCS), the two selected resilience indicators are calculated for N number of potential samples/iterations. The steps and elements proposed by (Haldar & Mahadevan, 2000) are followed in setting up the MCS, including:

1. Problem definition in terms of random variables;

2. Probabilistic quantification of the random variables and their characteristics;

3. Random variables values generation;

4. Experimentation/evaluation of the problem for all random variables;

5. Extraction of probabilistic information for n realisations; and

6. Accuracy and efficiency of the simulation determination.

The outage duration and the percentage of generation available are random variables. Their probabilistic distribution and characteristics are estimated depending on the type of outage and for each specific context/case study. The two selected metrics are evaluated for n iterations. The number of iterations can be estimated depending on factors such as computation power, coefficient of variation, confidence level, relative uncertainty and population of samples available. Figure 5 shows how the MCS is employed to calculate the probabilistic energy resilience indicators.

In the first step of the energy resilience module (Step (a) in Figure 5), the impact and disruption/outage due to climatic hazards on the system are estimated. Then, the system technology design portfolio and its operation performance against the disruption is evaluated (Step (b) in Figure 5). Step (c) assesses system resilience for the relevant resilience layer (i.e., engineering-designed, operational or community-societal). Next, the selected metrics associated with the resilience layer are quantitatively assessed (Step (d) in Figure 5. Finally, the graphical representation and risk of failure graphs are demonstrated (Step (e) in Figure 5). The module’s numerical spreadsheet and graphical output can be used to compare energy resilience and risk of failure for various communities and scenarios.

The focus here is on the overall system resilience related to the engineering-designed resilience layer. The operational resilience and the community resilience layers are crucial components of a resilient system and need to be considered further in future research.

Estimating the resilience metrics using the MCS method can provide a comprehensive view of the probability and potential consequences of a disruption. The resilience quantification module developed in the present work is suitable for early stages of the planning where input data is scarce and a quick evaluation is needed, relatively higher uncertainties are acceptable compared to the detailed design stage. The approach used herein is similar to the approaches widely employed in seismic engineering (i.e., push-over analysis) and economics (i.e., stress test) (Bruneau, Chang, Eguchi, et al., 2003; Obama, 2020; Vugrin, Warren, Ehlen, et al., 2010). In detailed design, the fragility curve and the probability of extreme events should be considered in the modelling. The EMP model can be applied to various cases and for different events to estimate the energy system resilience.

4. Case study description

The number and type of case studies implementing the EMP are limited worldwide. None has been undertaken in Australia. The selected case study herein provides an opportunity to demonstrate the EMP model’s practical application.

The new campus of the University of Melbourne in FB, located about 5 km from the Melbourne city centre (see Figure 6) is selected to demonstrate the application and value of the EMP model. The campus is located in an old industrial site that has been demolished and is being rehabilitated and redeveloped in three stages. The physical scale, typology of building and mixed energy uses (i.e., heating, cooling and electricity) in the selected campus are in line with the framework (Charani Shandiz, Rismanchi, & Foliente, 2021) and allow for demonstrating the capabilities of the EMP model introduced herein. A limitation here is the scarcity of measured data in the early-stage development. Although the EMP model can be applied to both brownfield and greenfield, herein a new development campus is selected to demonstrate the capabilities of the model due to the higher flexibility and options available in a new development. Another motivation for selecting the FB campus was that the campus design is ongoing. The authors have been engaged in the planning and design process. Several meetings, consultations and workshops with the consultants, the University facilities and design teams helped to frame a real-world EMP study that contributes to the literature.

Figure 6. An artist’s impression of the new University of Melbourne campus in Fishermans Bend, Melbourne (the University of Melbourne, 2021).

The focus here is on the first stage of the new campus development, comprising five buildings with a total gross floor area (excluding parking) of 54,987 m² and a total useful floor area of 36,243 m². Energy uses are mainly standard energy uses for offices, small retail, laboratories and lecture theatres. The design objectives of the campus include net-zero emissions and resilient aspects. The energy demand of FB campus is summarised in Table 4.

Table 4. Summary of Fishermans Bend campus energy demand.

Item	Value
Peak electricity demand [kW]	439.3
Peak cooling demand [kW]	1930.2
Peak heating demand [kW]	999.9
Average electricity demand [kW]	204.7
Average cooling demand [kW]	105.5
Average heating demand [kW]	32.5
Total electricity demand [MWh]	1793.4
Total cooling demand [MWh]	924.7
Total heating demand [MWh]	284.9

Melbourne is categorised in a mild temperate zone (Australian Building Codes Board, 2019) or Cfb according to Koppen-Geiger climate classification (Peel, Finlayson, & McMahon, 2007).

Melbourne hourly weather data is taken from TRNSYS (2023Thermal Energy System Specialists, 2023) as a typical meteorological year. Hourly solar radiation in Melbourne on horizontal 30° and 45° planes at zero azimuth angles are estimated using TRNSYS (Thermal Energy System Specialists, 2023). Assuming an average efficiency of 17.5% solar PV and 70% for solar thermal collector for typical technology available on the market, the standardised hourly energy output (zero to one) per module area of these technologies is estimated based on Equation 1(1) $\overline{E_h}=S\eta \overline{H_h}$(1) .

The normalised hourly wind power output (zero to one) is estimated using Equation 2(2) $V_2=V_1.(Z_2/Z_1{)}^a$(2) and assumed specifications of a typical community level wind turbine summarised in Table 5. Other techno-economic specifications of the wind turbine are reported in Appendix B.

Table 5. Wind turbine specifications.

	Cut-in wind speed [m/s]	Rated power at [m/s]	Cut-out wind speed [m/s]	Height [m]
Wind turbine	2.5	12	70	30

For GSHP potential, the thermal energy cap per pile is assumed equal to 1.5 kW, and the area required per pile is assumed 6 m². This can be either considered in the area constraint or in the max capacity of GSHP. The model input variables are summarised in Table 6.

Table 6. Input variable values for the selected case study.

Selected parameters	Reference case
Electricity price	Energy use: 0.051 [A$/kWh] Green electricity: 0.103 [A$/kWh] Connection charge: 147 [A$/kW/year]
Gas price (and biomass)	Gas: 0.071 [A$/kWh] Biomass: 0.032 [A$/kWh]
Feed-in tariff	0.013 [A$/kWh]
Grid emissions factor	0.73 [kgCO2e/kWh]
Carbon Price	A$50/tCO2e

The probability of outage duration of the electricity grid in Melbourne (including the FB campus in the city of Port Philip) due to climate hazards is shown in Figure 7. This is estimated according to the historical outage data available in Melbourne.

Figure 7. Failure duration distribution.

Net-zero emissions balance calculation accounts for the avoided emissions through the renewable energy export to the grid in addition to the on-site or off-site renewable energy generation (e.g., PPA). Avoided emissions is calculated by using the grid emissions factor, which allows for fuel switching. Other emissions balance calculation methods can be considered by adding the conditions in step a of Figure 3.

5. EMP model verification

In this section, the EMP model’s application is demonstrated for two main scenarios, namely: reference case (cost minimisation with no emissions constraints) and net-zero emissions case (cost minimisation while imposing a zero-emissions constraint). Other settings (campus inputs, technology database, area constraints, carbon price and so on) are identical for both cases.

A one-week time slice in Autumn (mid-season when both heating and cooling demands are present) is selected for verification of the EMP model output, in particular to investigate the energy performance of the technologies and validate that the energy demand is met at all times. Validation and calibration of EMP are challenging by nature. Ficus as the core of the EMP model has already been tested and published (Atabay, 2017a; Atabay 2017b). Cross-checking and referencing input data is another strategy taken together with consultation with the experts (e.g., facility manager and consultants). The mathematical equations are obtained from the existing literature and prototyping is employed in the model development. The output results are checked against the thermodynamic laws (e.g., conservation of energy), flow directions, and the balance of supply and demand, as well as with other available data and case studies.

In Figure 8, the hourly electricity supply and demand are shown for one selected week in March (hours 1640 to 1808). It can be seen that the electricity demand is fully met at each time step. The electricity grid the main source of electricity supply covered by on-site PV generation. The excess PV generation is exported to the grid, see hour 1718 and 1742 in Figure 8.

Figure 8. Hourly electricity supply and demand for one selected week in March (hours 1640 to 1808).

Figure 9 to Figure 12 show the hourly cooling and heating supply and demand for the selected week in March (hours 1640 to 1808).

Figure 9. Hourly cooling supply and demand for one selected week in March (hours 1640 to 1808).

Figure 10. Hourly performance of TES c. tank for one selected week in March (hours 1640 to 1808).

Figure 11. Hourly heating supply and demand for one selected week in March (hours 1640 to 1808).

Figure 12. Hourly performance of TES h. tank for one selected week in March (hours 1640 to 1808).

It can be seen that the energy needs for both cooling and heating are fully satisfied by ASHP, GSHP rev. and TES tank at all times. The ASHP is the main source of thermal energy supply complemented by TES and GSHP. In Figures 10 and 12, it can be seen how the excess energy is stored in the TES to be used when required. Heat pumps seem to operate mainly during the day when the demand is higher and PV generation is available to increase the self-consumption (i.e., to use the generated on-site energy and minimise the energy exportation to the grid). GSHP rev. covers a small portion of the demand when simultaneous heating and cooling is required.

The graphs presented in this section show that the EMP model can properly optimise the technologies sizing and operation to reduce the total costs while meeting the energy demand at all time steps.

6. Results and discussion

A selected set of EMP model results from the case study is presented and discussed in this section to demonstrate the model capabilities. The annualised cost breakdown of the reference case and the net-zero case are shown in Figure 13.

Figure 13. Annualised cost breakdown of reference community and net-zero emissions community cases.

It can be seen that NZEC is feasible with a total cost of about six percent higher than the reference case for the case study community. This means that the net-zero emissions target for the selected community is practical by 2050. Also, in both cases, the investment, energy importation and network costs are the largest part of the total cost. Although the carbon cost is higher in the reference case, an increase in the investment and import of clean/green energy (or also known as power purchase agreement or PPA) results in a higher total cost for the NZEC case. Revenues from energy export and variable operational costs seem to be negligible compared to the other cost components. The reference community emits over 1000 tonnes of CO₂e per year.

Table 7 summarises the optimal design capacity [kW] of the energy supply technology portfolio including thermal and electrical, as well as grid import, export and PPA. The size is in terms of the energy input capacity (for example, for heat pump the electricity input capacity).

Table 7. Design capacity of technology portfolio for reference case and net-zero emissions case.

Technology name	Reference community [kW]	Net Zero Emissions Community [kW]
CHP	-	-
Boiler (gas, electric, biomass)	-	-
ASHP heating	24	28
ASHP cooling	156	162
Wind turbine	-	-
PV	414	554
Solar thermal	-	-
PPA	0	257
Absorption chiller	-	-
Co-generator	-	-
GSHP reversible	10	-
Peak grid import	433	127
Peak grid export	268	421

It can be seen that the grid import has the highest design capacity for energy provision in the reference scenario and is therefore the cheapest option. PV seems to be an attractive source in both zero and non-zero cases. However, gas and biomass sources of energy do not seem to have a place in the technology portfolios. PPA is necessary for achieving net-zero. Heat pumps are the optimal technology for heating and cooling provisions.

Warm and cold thermal energy storages are present with an energy content capacity of about 7.1 and 12.5 MWh, respectively, in the technology portfolio of both cases – see Table 8. As mentioned in Appendix B, the temperature difference between the inlet and outlet of the cold and warm storage tanks are considered equal to 8°C and 20°C, respectively, with a variable flow rate. The portfolio of net-zero emissions case also includes a 133-kWh battery.

Table 8. Energy storage portfolio for reference and net-zero cases.

	Reference case	Net-zero case
Capacity (kW)
Battery	0	49
TES h. tank	894	899
TES c. tank	1374	1374
Energy (kWh)
Battery	0	133
TES h. tank	7149	7190
TES c. tank	12473	12473

In addition to the gaps discussed in the literature, the complexities in the technology portfolio operation may be another reason for limited use of some technologies, such as thermal energy storage and GSHP. The centralised scheme considered here enables this technology integration, however this might be challenging and financially unattractive for the individual systems.

The linear programming (including the mixed-integer configuration) nature and formulation of Ficus is similar to other tools in existing literature such as (Ma, Wu, Hao, et al., 2018; Mavromatidis, Orehounig, & Carmeliet, 2018; Mehleri, Sarimveis, Markatos, et al., 2013; Morvaj, Evins, & Carmeliet, 2016; Sameti & Haghighat, 2018; Wouters, Fraga, & James, 2015). However, considering the EMP framework, we found that Ficus is missing some constraints and considerations that are important in the decision-making and EMP of communities. Thus, the emissions objective, hourly emissions and carbon price missing in the source code of Ficus (Atabay, 2017b) were incorporated for more accurate modelling. Also, the energy supply and storage area availability constraints were developed. Specific area availability considerations (i.e., rooftop, underground, outdoor and indoor) in the system planning at a larger scale seem to not be a common consideration in most studies, such as (Atabay, 2017b; Ma, Wu, Hao, et al., 2018; Sameti & Haghighat, 2018; Wouters, Fraga, & James, 2015), while at the community level this is a critical consideration due to the limited area available and its monetary value, especially in urban communities. The area constraint is divided into three groups: (a) the area available under the ground, (b) the area available on-site, indoor, and (c) exposed area available on the rooftop. The technologies associated with these three categories are grouped as well. This can be valuable for the architectural design and planning of the campus. Potentially, the technologies can be fitted into the master plan and architectural design of the community for optimal use of space, performance, aesthetics and more. These additions to the model are important for enabling the design of different NZEC definitions and applicability to various communities.

As the key metrics for resilience quantification, the probability distribution of CLNS and DCLNS are shown in Figure 14. The critical energy demand is a part of the total energy demand that is vital to the operation of the community and its relevant services even during a hazard, such as critical research facilities, data centres, security, and so on. The horizontal axis shows the bins for each indicator and the vertical axis demonstrates the probability of occurrence of each bin.

Figure 14. Probability and distribution of critical load not served (CLNS), kWh (left) and duration of critical load not served (DCLNS), hr (right) – reference case, start time 169h.

There is a more than 70% probability that the 500 kWh of critical load in the reference case is not served in the case of a disruption in January, which is considerable. The probability of CLNS for values greater than 2000 kWh reaches approximately zero. The probability of 6 h of critical load not being served is about 60%. This probability gets to near zero for a DCLNS more than 24 h. The reason that the CLNS and DLNS have higher values in the first part of the graph is linked to those high impact lower duration climatic hazard scenarios in the case study when often the critical energy demand cannot be covered by an on-site energy provision portfolio. These one-dimensional graphs can be compared for all scenarios and for different disruption types and times. It can be seen that the distribution curve follows a long tail distribution similar to kernel or log-normal distributions. This graphical representation can help the decision-makers and planners to understand the failure probability associated with each bin.

The bar graphs in Figure 14 can also be plotted in a probability density distribution form where the total area under the resilience probability density distribution in each graph is equal to one. The presented distribution graph shows the probability for each bin, which is easier to gain insights from. The cumulative density function graphs in Figure 15 demonstrate the cumulative probability of failure in terms of the CLNS and DCLNS.

Figure 15. Cumulative probability density of critical load not served (CLNS), kWh (left) and duration of critical load not served (DCLNS), hr (right) – reference case, start time 169h.

It can be seen that the extended disruptions are unlikely and the highest disruption probability is for the first few bins. These graphs can be used by choosing the threshold percentage and then finding the corresponding CLNS or DCLNS. For example, choosing a 95% cumulative probability point means that the disruption will be less than about 800 kWh and 10 h with a 95% probability.

The squared 2D probability of failure graph presented in Figure 16 is created from the contour kernel distribution. The horizontal axis shows the DCLNS in hours with 6-h bins and the vertical axis shows the CLNS in kWh with 500 kWh bins. The colour of the squares shows the combined probability of failure in each bin. The one-dimensional distributions of each of main axes (namely CLNS and DLNS) are plotted on two secondary axes in Figure 16.

Figure 16. Two-dimensional square failure probability density distribution of the reference case (left) and the net-zero emissions case (right) for the Fishermans Bend campus case study.

It can be seen that, due to a higher share of PV and battery, the probability of failure in terms of CLNS and DCLNS in the NZEC case has a reduced extension and intensity compared to the reference case. This counters the common perception that the net-zero communities could be more prone to failure during extreme climate events. Or, in other words, the EMP of a NZEC can achieve a resilient energy system solution.

Although MCS that is used in the probabilistic resilience quantification is also used in reliability engineering for probable working/maintenance hours, the difference is that herein the focus is on the disruption to extremely high impact climate events. In addition, the focus is on the overall system resilience on one node (community as a lumped model). In some extended systems, the nodal energy resilience should be modelled as each node might have a different resilience performance. It is assumed that the energy supply disruption in the energy grid is due to probable extreme climatic event(s). Sometimes, during an extreme event, the grid energy supply is not fully lost, but an extremely high energy price might not be feasible for the community (financial resilience) so that it is as if the supply is disrupted. Similarly, a reduction in energy losses will increase systems energy resilience.

The EMP model and resilience evaluation module allow for probabilistic resilience analyses using the MCS technique, beyond the high-level risk identification, depending on the data available and the client/design requirements. The energy resilience assessment approach herein is inspired by the approaches typically employed in seismic engineering (one of the fields that pioneered resilience estimation and enhancement measures (Bruneau, Chang, Eguchi, et al., 2003; Vugrin, Warren, Ehlen, et al., 2010) and encoded some of them in building codes) and in financial parameters. For example, in the seismic design of buildings, an initial building is designed and then a specified seismic load is imposed on the system to undertake a push-over analysis and estimate its system response and identify the required performance (e.g., deformation, load resistance) to meet the target objectives, or identify the required enhancements to bring the building’s performance to meet the requirements of the selected seismic load. Similarly, stress testing is another way of evaluating the resilience of engineering and social systems (e.g., used in the global economic recession in 2008) (Obama, 2020). This approach is suitable for the early stages of the planning, where input data is scarce and a quick evaluation is needed; relatively higher uncertainties are acceptable in this stage compared to those in detailed design stage. Detailed fragility analyses and system components modelling may also be needed in more advanced design phases.

Since practical and case-study applications of such multi-dimensional approach in energy master planning of NZEC are lacking (Charani Shandiz, Rismanchi, & Foliente, 2021), different case studies of NZEC projects should be analysed and compared to determine the bottlenecks and successes and how stakeholders are influenced.

The EMP model is not focused on improving the optimisation methods; however, it is using the optimisation logic established in the source code (i.e., Ficus) to demonstrate the EMP approach to achieve NZEC and assess its multiple dimensions, in particular energy resilience. Thus, network optimisation and detailed economic analysis and optimisations are not in the scope of this work. In future, the model can be expanded to include spatial and network aspects, economic factors and markets such as variable WACC and time varying investment and so on.

To investigate the uncertainty of model outputs given uncertain inputs and externalities, a variety of scenarios should be developed. These scenarios can be based on a knowledge-based deterministic approach, for example, as in Walker, Labeodan, Boxem, et al. (2018) and account for externalities and the uncertainties that are involved in EMP (Prasad, Bansal, & Raturi, 2014). Further work on the scenario development framework for EMP is needed, along with its practical application to a broad range of case studies.

As discussed in (Charani Shandiz, Foliente, Rismanchi, et al., 2020), targeting also the community/societal resilience layer could provide the highest resilience capacity. However, the focus herein was only on the engineering-design energy resilience layer, as evaluating community resilience and incorporating the feedback on system requirements and performance are very complex and challenging. In this work, the energy/emission reduction hierarchy is not discussed as energy demand is an input to the EMP model. The influence of energy efficiency and demand reduction strategies on cost, emissions and energy resilience are reported in (Charani Shandiz, 2022).

Detailed data (e.g., demand or disruption estimation) are not available or known in the early stages of the design, therefore relevant methods should be used to estimate data for early-stage modelling and analysis (e.g., using technical experience and/or indirect data). Although the data might be community, location and time dependent, the model can still be applied to various cases/contexts by including the required data. For example, efficiency reported here is at full load, the effect of part load efficiency and the effect of external factors (e.g., the temperature variations) should be considered in the detailed design phase. Additional indicators can be considered in the future for environmental and financial aspects to understand the effect of market price.

7. Implications and future directions

The developed approach in this paper lays the groundwork for further research and development in the field. The implications for practice and policy are identified and further research are discussed below.

The EMP model enables exploring the suitability of a wide range of technology options for different community objectives and constraints under various scenarios to quantify key decision-making metrics and uncertainties of the options for solutions under consideration. The risks and consequences of failure related to energy resilience of communities against various extreme climatic events can be analysed and therefore prepared for from the initial design stages. When systematically applied in projects, these can potentially lead to more cost-effective, sustainable and resilient energy system solutions. Also, the approach empowers communities to have more control over their energy system design and performance and access the energy system based on their values as prosumers rather than sole consumers. This is aligned with the global shared economy trend along with the decentralisation of energy systems. The reduced amount of data required and the design consideration of the EMP model enables it to be accessible to a wide range of both technical and non-technical stakeholders.

In future, research should focus on the development of extensive databases and archetypes for various greenfield and brownfield communities based on monitored data of NZEC case studies to further enhance this capability. Other dimensions, applications and implications of the EMP model should be investigated. With further applications of the EMP model to a broad range of case studies and scenarios, the impact of policy, uncertainties and other factors can be analysed. And this may lead, over time, to the development of context-appropriate policy and/or technical guidance for different typologies of community-level energy systems. In the technical aspects of modelling, the impact of technical, market and regulatory settings needs more research to be tested and simplified for early-stage design. Some of these aspects include: energy distribution network, technology placement, temporal lag in the thermal systems, temporal market and regulatory settings, quality of energy, temporal lag, and so on. There are further opportunities to enhance and extend the multi-dimensional performance evaluation and resilience assessment measures, incorporate more sophisticated stochastic modelling and analysis capabilities, integrate feedback data from technical and social systems (real or near-real time) and integrate the model’s capabilities with a range of digital twin technologies and/or related platforms that are supported by spatial data infrastructure, geospatial modelling and visualisation capabilities.

8. Conclusions

A comprehensive and systematic approach to EMP of NZEC with the additional capability for quantitative energy system resilience assessment is developed to explore the performance of multiple technology solution options across a range of criteria in the early-stage planning and design stages. The EMP model enables a systematic treatment of the key drivers, challenges and/or constraints in community level and/or precinct-wide development projects to achieve both net-zero emissions and energy resilience targets using probabilistic analyses. The model draws from an enhanced technology options database (with technical and financial information) that has been developed for the Australian context.

Applied in the EMP of the University of Melbourne’s new campus, which comprises five buildings with mixed energy uses (heating, cooling and electricity), both the 2050 net-zero emissions and operational energy resilience objectives were demonstrated to be technically and financially feasible. Amongst the solution options considered an energy provision portfolio that consists of 28 kW and 162 kW air source heat pumps for heating and cooling, respectively, and 554 kW of PV panels, 257 kW of green power, 127 kW grid power (non-green), 49 kW of battery, 899 kW and 1374 kW of thermal heat and cold storage tanks demonstrated to meet the dual objectives. The case study results counter the common perception that energy systems for NZECs could be less reliable during extreme climate events, and that making the energy system more resilient while also meeting a net-zero emissions objective may not be financially feasible.

As expected, the results are highly sensitive to the energy price as well as the cost of technologies. It was observed that the reduction of the cost of technologies or increase in energy price or implementing incentives and penalties will make the net zero energy targets more achievable with more attractive return on investment.

The ability to model the potential technical, environmental, financial and social impacts of a range of energy system configurations at community-level developments and assess their resilience to a range of potential disruptions early in the EMP process gives key stakeholders and/or decision-makers the opportunity to explore a wide range of potential sustainable and innovative solutions. Systematic and widespread applications to community-level and/or precinct-wide development projects can potentially lead to more cost-effective, sustainable and resilient energy system solutions at scale.

In the future, the EMP model should be further applied to a diverse type and size/scale of case study projects and scenarios, include a more comprehensive set of performance evaluation and resilience assessment measures (especially social impact measures), incorporate more powerful modelling and analytical capabilities, and be integrated with a range of advanced digital engineering tools, technologies and platforms.

Abbreviations

AUD	=	Australian dollar
CLNS	=	critical load not served
CO2e	=	carbon dioxide equivalent
COP	=	coefficient of performance
DCLNS	=	duration of critical load not served
EMP	=	energy master planning
FB	=	Fishermans Bend
GHG	=	greenhouse gas
GSHP	=	ground source heat pump
kW	=	kilowatt
kWh	=	kilowatt hour
MCS	=	Monte Carlo simulation
NZEC	=	net-zero emissions community
PPA	=	power purchase agreement
PV	=	photovoltaic
RES	=	renewable energy sources
SDGs	=	sustainable development goals
UN	=	United Nations

Acknowledgments

The first author thanks The University of Melbourne for providing a Melbourne Research Scholarship Award.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The work was supported by Melbourne Research Scholarship Award.

Notes on contributors

Saeid Charani Shandiz

Saeid Charani Shandiz has extensive experience in sustainability, energy modelling, data science and technologies, especially related to sustainable and climate-resilient development of infrastructure. He has several years of work experience and has been involved in a wide range of sectors and disciplines including built environment and energy systems projects in Australia, Italy, Iran, and Morocco, as well as completing advanced academic degrees.In his PhD project, Saeid developed a systematic approach, that has applications and implications for wider systems and infrastructure, to achieve both net-zero emissions and climate resilience goals in energy planning, design, and assessment of community-scale energy systems. He is the author of several scientific papers on subjects such as net-zero emissions buildings and communities, building science, energy resilience and energy master planning.

Behzad Rismanchi

Dr Behzad Rismanchi is a Senior Lecturer in building energy at the Department of Infrastructure Engineering. He is a professional member of Engineers Australia and is a certified energy manager (CEM) with over 15 years of experience in research, design and optimisation of energy systems.

Greg Foliente

Greg Foliente is Professor at the University of Melbourne, a Senior International Expert for the Global Buildings Performance Network (GBPN) and the President and Board Chair for the non-profit International Initiative for a Sustainable Built Environment (iiSBE). He leads systems-based transdisciplinary research – integrating modelling and engineering knowledge with social and environmental sciences and emerging digital and geospatial technologies – to improve the safety, sustainability and resilience of built and urban environments at various scales. His knowledge, experience & interests span the full life cycle of knowledge development from academic research, to applied R&D and their translation into practice, policy & standards. He is an expert consultant to – and research collaborator with – industry, government and international institutions including UN agencies. Amongst many forms of recognition, he received the prestigious James Croes Medal from the American Society of Civil Engineers (ASCE), a best paper award from the American Society of Mechanical Engineers (ASME) and numerous visiting appointments from leading universities in the US, Europe and Asia. He received his PhD and MSc degrees at Virginia Polytechnic Institute and State University (Virginia Tech) and executive leadership and management training at the Australian Institute of Company Directors and the MIT Sloan Executive Education.

Lu Aye

Lu Aye (F.AIE, F.AIRAH) is a Professor of Energy Engineering at the University of Melbourne. Prof. Aye has established an internationally recognised research cluster, Renewable Energy and Energy Efficiency Group in 2008 at the university. He has been the leader of the group since then. He has over 40 years of engineering experience in university teaching, research, development, demonstration and commercialisation of low-carbon technologies, including solar PV and solar thermal systems. His research areas include heating, ventilation, air-conditioning and refrigeration systems, solar energy engineering, waste to resources, complex systems modelling, and life cycle environmental impact assessment. Prof. Aye been recognised as a leading expert in modelling, simulation, optimisation and forecasting of complex systems behaviours. He utilised computational and participatory approaches for modelling socio-ecological systems under deep uncertainty. These models have been applied to identify the effects of policy interventions and robust decision making. He has authored and co-authored over 300 peer reviewed scientific publications. Prof. Aye has been a chief investigator in many Australian Research Council (ARC) and industry grants. During the past five years, in collaboration with his colleagues, he attracted over $ 30 million in research income.

Appendices Appendix A

Description of the techno-economic database items and definitions.

Capacity range [kW] or [kWh]: the minimum and maximum capacity of the technology depending on the market availability and/or technical constraints. This also plays a role in the cost of technology.

Capital cost [AUD/kW] or [AUD/kWh]: The total capital cost required for each kW of the design capacity or energy content capacity.

Fixed operating cost [AUD/kW/year]: Cost of operation and maintenance that does not depend on the energy output generated, including miscellaneous annual costs such as periodical maintenance, insurances, capacity shortage penalty, tax, and so on.

Variable operating cost [AUD/kWh]: The cost of operation and maintenance that varies depending on the energy generated (hours of operation, number of start-stop cycles, number of full/partial/base load hours, and more) including energy consumption of other related components (auxiliary load), chemical and water consumptions, handling the related by products/waste, and wear and tear maintenance.

Efficiency [%]: Shows how efficient the conversion is from one source/type to another source/type of energy. Used mainly to calculate the fuel consumption and emissions.

Space requirement [m2/kW]: The typical space/area that is required for the mechanical equipment.

Average service life [years]: The number of years that the technology can operate as an average. After the end of the service life the technology should be replaced by a new one.

Appendix B

The techno-economic database of energy generation technologies at the community level is summarised in Table B1. The reported costs are per kW or kWh of output.

The techno-economic database of energy storage technologies is summarised in Table B2.

Technical specification of the storage technologies can be added by the user. The following details in Table B3 are assumed for this study based on the typical technology catalogues available in the market.

The temperature difference between the inlet and outlet of the cold and warm storage tanks are considered equal to 8°C and 20°C, respectively, with a variable flow rate.

All costs are average values (from various sizes and products) and include tax. Components (e.g., inverter, wires, pipes, etc.) and installation costs are included. The equipment factors should be considered in the detailed design for final costs. Some costs are converted from other currencies to their Australian Dollar (AUD) equivalent.

Table B1. Techno-economic database of energy generation technologies at the community level.

Technology Type	Capacity range (min, max) [kW]	Capital cost [A$/kW]	Fixed cost [A$/kW/year]	Variable cost [A$/kWh]	Efficiency [%]	Space required [m2/kW]	Service life [years]	Reference
Commercial PV system	(10, 10000)	1600	17	0	17.5	5	25	(Clean Energy Reviews, 2021; Graham, Havas, Brinsmead, et al., 2019; Graham, Hayward, Foster, et al., 2018; Natural Resources Canada RETScreen, 2023; Graham, Hayward, Foster, et al., 2020; Ramboll, 2019; Solar Choice, 2021)
Wind turbine	(100, 10000)	3210	85.6	0	Power curve	0.8	23	(Graham, Hayward, Foster, et al., 2018; Natural Resources Canada RETScreen, 2023; Graham, Hayward, Foster, et al., 2020; Ramboll, 2019)
Solar thermal (evacuated)	(10, 10000)	1701	11.9	0	70	1.1	30	(Karunathilake, Hewage, Mérida, et al., 2019; Natural Resources Canada RETScreen, 2023; Ramboll, 2019; Rockenbaugh, Dean, Lovullo, et al., 2016)
Heat pump air source (outdoor)	(10, 10000)	2780	3	0	360	0.2	25	(Natural Resources Canada RETScreen, 2023; Ramboll, 2019; Rawlinsons Quantity Surveyors, 2020)
Absorption chiller (outdoor)	(10, 10000)	3017	3	0.01	72	0.2	25	(Department of Energy, 2017; Ramboll, 2019)
Heat pump ground source h/c	(10, 10000)	3224	3	0	430	2	25	(Lu, Narsilio, Aditya, et al., 2017; Natural Resources Canada RETScreen, 2023)
Cogen engine or turbine (gas)	(100, 10000)	3745	214	0.01	45	0.1	25	(Natural Resources Canada RETScreen, 2023; Ramboll, 2019)
Combined Heat and Power (biomass/gas)	(100, 10000)	7072	214	0.01	27 (e) & 83 (h)	0.2	25	(Graham, Hayward, Foster, et al., 2018, 2020; Karunathilake, Hewage, Mérida, et al., 2019; Natural Resources Canada RETScreen, 2023; Ramboll, 2019)
Boiler condensing (gas)	(10, 10000)	304	3.1	0.01	95	0.01	25	(Department of Industry science energy and resources Australian Government; Natural Resources Canada RETScreen, 2023; Ramboll, 2019; Rawlinsons Quantity Surveyors, 2020)
Boiler (biomass)	(10, 10000)	1104	51.5	0.01	99	0.2	25	(Karunathilake, Hewage, Mérida, et al., 2019; Ramboll, 2019)
Boiler (electric)	(10, 10000)	320	1.7	0	99	0.01	20	(Ramboll, 2019)

Table B2. Techno-economic database of energy storage technologies at the community level.

Technology Type	Capacity range [kW]	Capital cost [A$/kW]	Capital cost [A$/kWh]	Fixed cost [A$/kW/year]	Fixed cost [A$/kWh/year]	Variable cost [A$/kWh]	Efficiency [%] in and out	Space required [m2/kW]	Service life [years]	Reference
Battery	10–10000	1535	0	0	8	0	90	0.03	15	(Aboumahboub, Brecha, Shrestha, et al., 2020; Clean Energy Reviews, 2021; GHD, 2018; Graham, Hayward, Foster, et al., 2018, 2020; IRENA, 2017; Solar Choice, 2021)
Thermal Energy Storage tank	10–10000	-	40	0.4	0	0	90	0.2	30	(IEA ENERGY TECHNOLOGY SYSTEM ANALYSIS PROGRAMME, 2013; International Renewable Energy Agency, 2020; Ramboll, 2019; Roth, Zogg, & Brodrick, 2006)

Table B3. Technical specifications of the energy storage technologies.

Technology Type	Max power to energy ratio [1/h]	Efficiency in [%]	Efficiency out [%]	Self discharge [%/hr]	Cycles max [-]	Depth of discharge [%]
Battery	0.5	90	90	0.00004	1000000	100
Thermal Energy Storage tank	0.125	90	90	0.000001	1000000	100