Industrial Project with Verdia Pty Ltd School of ......iv Acknowledgements I wish to acknowledge my supervisors at Murdoch university, Dr Martina Calais; Senior Lecturer Engineering,

Economic Analysis of Reducing End-User Peak Demand with Renewable Energy


Industrial Project with Verdia Pty Ltd

School of Engineering and Information Technology

Emily Salter

Bachelor of Engineering

(Renewable Energy and Electrical Power)

June 2017


ii

Declaration

I hereby affirm that this thesis is my own work and that to the best of my knowledge, contains no material

previously published or produced by any party in fulfilment, partial or otherwise, of any other degree or

diploma of another University or institute of higher learning except where due acknowledgement is made

in the text.

Emily C. Salter

3rd June 2017


iii

Abstract

Large electricity end users are currently not economically inspired to invest in renewable energy due to

low energy charges and high peak demand charges. To investigate if peak demand reduction could

contribute to economic benefits, an excel model was developed to determine the economic benefit of

reducing end user peak demand with renewable energy. Moving loads to low demand periods and

improving load efficiency is the most successful method to reduce peak demand. It was identified that

load reduction by solar penetration is well suited to Western Australian end users, however solar

generation is vulnerable to losses of up to 80%. Energy storage systems were simulated to provide

contingency in the event of solar generation loss. For the case study investigated, capital investment was

estimated to take between 10 and 11 years to be returned in the form of large scale generation

certificates, consumption savings and demand savings.

iv

Acknowledgements

I wish to acknowledge my supervisors at Murdoch university, Dr Martina Calais; Senior Lecturer

Engineering, for her continual inspiration and guidance and Dr Gareth Lee, Deputy Head for Teaching and

Research of the School of Engineering and Information Technology, for providing excellent advice and

support.

Srini Laxman, the then Technical Director at Verdia has provided essential academic, professional and

moral support. The general manager at Verdia, Doug Tapper has been exceptionally patient during the

past year allowing my completion of university and her thesis as an industrial project

Dr John E. Stanton; Adjunct Professor at the University of Western Australia, godfather and best friend

has been tremendously supportive and helpful in the completion of this thesis. My parents Margot and

Richard Salter, have always been caring and hardworking role models, and Carl Kobelke, loving and

supportive boyfriend.

Finally, I wish to thank my Grandmother, Joan Hammond, who inspired me to finish my Bachelor of

Engineering.

v

Contents

Definitions ................................................................................................................................................... vi

Abbreviations ............................................................................................................................................. vii

List of Symbols ........................................................................................................................................... viii

Figures ......................................................................................................................................................... ix

Tables .......................................................................................................................................................... ix

1 Introduction ....................................................................................................................................... 10

1.1 Verdia ........................................................................................................................................ 11

2 Background ........................................................................................................................................ 12

2.1 Market Operator ....................................................................................................................... 12

2.2 Network Generators ................................................................................................................. 13

2.3 The Network Operator .............................................................................................................. 13

2.4 Energy Retailers ........................................................................................................................ 13

2.5 Peak Demand ............................................................................................................................ 14

3 Available Research and Resources .................................................................................................... 16

4 Scope ................................................................................................................................................. 19

5 Approach to modelling peak demand reduction ............................................................................... 21

5.1 Market Operator Demand Charge Pricing Mechanisms ........................................................... 21

5.2 Network Operator Demand Charges ........................................................................................ 21

5.3 Peak Demand ............................................................................................................................ 23

5.4 Cost of Supply ........................................................................................................................... 23

5.5 Load Rescheduling .................................................................................................................... 25

5.6 Lighting ...................................................................................................................................... 28

5.7 Solar .......................................................................................................................................... 29

5.8 Energy Storage .......................................................................................................................... 33

5.9 Power Factor Correction ........................................................................................................... 34

5.10 New Cost of Supply ................................................................................................................... 35

5.11 Estimating project costs ............................................................................................................ 36

6 Development ..................................................................................................................................... 37

6.1 Model Navigation ...................................................................................................................... 38

6.2 Model Operation....................................................................................................................... 38

7 Model Validation ............................................................................................................................... 41

7.1 Simulated and Real Data Comparison....................................................................................... 41

7.2 Case Study ................................................................................................................................. 45

8 Conclusion ......................................................................................................................................... 53

8.1 Future Work .............................................................................................................................. 54

9 References ......................................................................................................................................... 55

vi

Definitions

Billing Data Electricity accounts, provides National Meter Identifier (NMI), energy

charges, network charges, market charges and consumption.

Cooling Load A Reiterative, air-conditioning or ventilation load

End User Electricity consumer at the end of the electricity supply chain

Energy Charges Charge for electricity per unit of electricity (kWh)

Electricity Retailer Retail bodies who buys capacity from the market to sell on to end

users

Electricity Network South West Interconnected System.

Load Data Interval data measurements of the end user’s load

Market Charges Charge for supply passed through from Australian Energy Market

Operator

Network Charges Charge for supply passed through from Network Operator

Off Peak Hours Defined by network operator times outside 8am to 10pm Monday to

Friday, energy costs are usually lower during these times as demand

is usually less.

Peak Hours Defined by network operator as 8am to 10pm Monday to Friday,

energy costs are usually higher during these times as demand is

usually greater.

Western Power State Government (currently) owned statutory corporation

responsible for building and maintaining the South West

Interconnected System (SWIS).

Supply Electricity supply

Troffer Linear fluorescent Light fixture,

vii

Abbreviations

AEMO Australian Energy Market Operator

NEM National Electricity Market

NMI National Meter Identifier

WEM Wholesale Electricity Market

SWIS South West Interconnected System

viii

List of Symbols

P Real Power

Q Reactive Power

S Apparent Power

kW kilo-Watt

kVAr kilo-Volt Amperes Reactive

kVA kilo-Volt Amperes

PFtarget Target Power Factor

Qtarget Target Reactive Power

Starget Target Apparent Power

W Watt, 1 Joule per Second

ix

Figures

Figure 1: The South West Interconnected System (Western Power, 2017) ............................................... 10 Figure 2: The Wholesale Electricity Market (AEMO,2016) ......................................................................... 12 Figure 3: Bundled and Unbundled Electricity Tariffs .................................................................................. 14 Figure 4: Commercial Load Profile Example ............................................................................................... 14 Figure 5: Homer Pro's Advanced Grid Rate Properties Input ..................................................................... 18 Figure 6: Rolling Peak Illustration ............................................................................................................... 22 Figure 7: SWIS Load Data vs Temperature ................................................................................................. 23 Figure 8: 12 peak trading intervals ............................................................................................................. 25 Figure 9: IRCR Ratios ................................................................................................................................... 25 Figure 10: Original and Proposed NTDL Profiles ......................................................................................... 26 Figure 11: Net Load Change, NTDL ............................................................................................................. 27 Figure 12: Original and Proposed Rescheduled Loads ............................................................................... 27 Figure 13: Original Lighting and Proposed LED Loads ................................................................................ 28 Figure 14: Original, Rescheduled and LED Reduced Loads ......................................................................... 29 Figure 15: Modelled PV Output .................................................................................................................. 30 Figure 16: Original, Rescheduled, LED and PV Reduced Loads................................................................... 30 Figure 17: AEMO SWIS Load Data vs Murdoch Meteorology Temperature Data ...................................... 32 Figure 18: AEMO SWIS Load Data vs Murdoch Meteorology Solar Radiation Data ................................... 32 Figure 19: National Renewable Energy Laboratory PV Watts Average and Minimum simulated 2000kW

Solar Output ............................................................................................................................................... 33 Figure 20: Target Maximum Demand ......................................................................................................... 34 Figure 21: Power factor correction algorithm ............................................................................................ 35 Figure 22: Demand Reduction implementation and economic benefit ..................................................... 36 Figure 23: Model Colour Coding Key .......................................................................................................... 38 Figure 24: Navigation Panel ........................................................................................................................ 38 Figure 25: Model Overview ........................................................................................................................ 39 Figure 26: Assumed LEDs installed with respect to time............................................................................ 42 Figure 27: Simulated and real LED load reduction ..................................................................................... 43 Figure 28: Average Simulated and Real LED Reduced Load, May .............................................................. 43 Figure 29: Murdoch University’s Annual Load Profile ................................................................................ 45 Figure 30: Murdoch University’s Annual Load Profile ................................................................................ 46 Figure 31: Solar Output .............................................................................................................................. 48 Figure 32: Power Factor Distribution Histogram ........................................................................................ 48 Figure 33: Simulated Power Factor Distribution Histogram ....................................................................... 49 Figure 34: Corrected Power Factor Distribution Histogram ....................................................................... 49 Figure 35: Simulated Annual Load .............................................................................................................. 50 Figure 36: Simulated Annual Load Profile .................................................................................................. 50 Figure 37: Simulated Progressive Demand Reduction ............................................................................... 52

Tables

Table 1: Hypothetical Demand Reduction Project ..................................................................................... 45 Table 2: Corresponding Real Power Measurements, Peak Trading Intervals............................................. 46 Table 3: Capacity Year and Individual Reserve Capacity Requirements ..................................................... 47 Table 4: Peak Trading Season Real Power Measurements ......................................................................... 51 Table 5: Simulated Capacity Year and Individual Reserve Capacity Requirements .................................... 51 Table 6: Maximum Demand Reduction ...................................................................................................... 52 Table 7: Simulated Total Savings ................................................................................................................ 53

10

1 Introduction

Australia’s electricity systems are undergoing major industrial, technological and economic changes

(Finkel et al. 2017). With significant drops in renewable energy costs and increasing technology

improvement, distributed energy generation is very rapidly becoming more viable than conventional

centralised generation sources. Australia has two major electricity systems, the national electricity market

and the wholesale electricity market. The wholesale electricity market supply’s electricity to the South

West of Western Australia. The network responsible for transmitting and distributing this electricity is

referred to as the South West Interconnected System (SWIS). It connects electricity consumers to

electricity generators South West of the line from Kalbarri to Ravensthorpe, including Kalgoorlie, Figure

1.

Figure 1: The South West Interconnected System (Western Power 2017)

One of the Wholesale Electricity Market’s greatest challenges is in meeting peak demand (IMO, 2014).

Peak demand is defined as above average sustained power use. Compounded from millions of electricity

consumers or end users, peak demand usually occurs daily, throughout business hours, and annually

throughout the summer months. Woven throughout complex energy pricing mechanisms, peak demand

normally accounts for half an end user’s electricity expenses. Most end users are charged for their

electricity supply by fixed consumption charges. Large end users are charged according to all components

of electricity supply. Their cost of supply varies according to their peak demand, not just their

consumption. Traditionally these charges have been accepted as a fixed expense that must be paid. Over

the recent years, peak demand management has become a growing area of interest (Benetti et al. 2015).

Demand side management and demand response systems have been utilised to reduce end user grid

burden and to take advantage of energy savings (Benetti et al. 2015). In most cases this has been achieved

either by onsite generators or load shedding (Benetti et al. 2015).


11

In the last few decades there has been a shift towards renewable energy and energy efficiency. As the

affordability of renewable energy has improved, many end users have adopted renewable energy to

reduce costs and improve their sustainability. In Australia, solar power has proved to be a particularly

viable approach to reduce the electricity expense of small end uses (Piccinini 2015). High electricity

consumption costs and low investments required are providing fast returns on investment.

For large end users with large capacity charges and low consumption charges, there is little benefit of

simply reducing their consumption. Instead, it is realised there may be significant economic reward to be

gained by reducing consumption in such way that reduces peak demand. It is acknowledged that

renewable energy resources are not reliable enough to reduce peak demand independently. An end user

must be able to dependably reduce their peak demand burden on the electricity systems; this would allow

a reduction to the end user’s allocated power capacity, and hence reduce their expenses.

Reliably reducing an end user’s peak demand can be achieved in various ways. Removing or reducing loads

is the most reliable way to manage peak demand. If that is not suitable for an end user, onsite renewable

generation and energy storage can be used. This report focuses on utilising a combination of

techniques/technologies with the goal of flattening an end user’s peak demand.

A model was developed during an industrial project with Verdia, to determine the potential of flattening

an end user’s peak demand. The model is used to identify demand reduction savings and analyse the

viability of renewable energy for large end users. It utilises a combination of energy efficiency solutions,

renewable energy onsite generation, power correction and energy storage. The possibility of reducing the

peak demand (viably) of an end user depends on various factors, characteristics and variability of their

peak demand and the correlation between their load and solar resource.

1.1 Verdia

Verdia Pty Ltd is a new national energy services company (Verdia 2016). It provides project advisory,

development, management, finance and asset management services. Verdia operates independently

from any suppliers or contractors. This allows Verdia to drive competitive tension between suppliers and

hold installers accountable for their quality of service. Verdia is constantly developing its quality partner

network through due diligence. Verdia’s market niche sits with larger end users requiring end-to-end

project management.


12

As the company builds momentum, larger end users have been engaging Verdia to assess the energy

saving implications of renewable energy projects; hence the emergence of a requirement for such an

evaluation model to determine the economic benefit of reducing peak demand with renewable energy.

The model discussed in this paper has been developed to provide insight to the viability of renewable

energy reducing peak demand.

2 Background

Electricity supply systems are very complex. Electricity supply systems comprise of energy generation,

interconnecting transmission and distribution networks and a vast quantity of fluctuating loads. It is

important to ensure energy generation remains in balance with loads and transmission/distribution loss.

The network operator manages this on a daily basis. As more loads are added to an electricity system,

more generation power must become available. This is overseen and coordinated by the market operator

(IMO 2014).

2.1 Market Operator

There is one market operator in Australia, The Australian Energy Market Operator (AEMO). The market

operator manages the operation and development of both the national electricity market and the

wholesale electricity market (AEMO 2016a). It conducts long term forecasting to ensure sufficient

generation capacity for the future. The AEMO must coordinate market generators i.e power stations to

ensure there is enough electricity to supply all market customers (AEMO, 2016). The Network Operator

(currently Western Power for the SWIS) manages the transmission and distribution network to deliver

electricity from Market Generators to Market Customers. Market Customers purchase electricity to sell

in the retail market to end users. These relationships are shown in Figure 2, green representing

coordination and management, grey as energy flow and black as payment.

Figure 2: The Wholesale Electricity Market (AEMO,2016)

AEMO

Market Generator

Network Operator

Energy Retailers

End User


13

There are various trading mechanisms: Reserve capacity, Bilateral Contracts, short term energy markets

(STEM) and dispatch/balancing process (AEMO 2016b). The reserve capacity mechanism impacts the cost

of demand. It ensures there is sufficient capacity and demand side management. Energy Retailers contract

capacity credits to the AEMO, who assigns capacity credits to market generators (AEMO 2016b). If

capacity is insufficient, AEMO auctions for more, so long as surplus cost is shared across market

customers. This is done on an annual basis to ensure there is enough capacity available in the market.

2.2 Network Generators

Network generators produce electricity: usually via a power station, wind farm or solar farm. Network

generators charge end users based on per unit consumption via their electricity retailers. Generators

charge a higher rate for electricity during peak hours; 8am to 10pm week days, when electricity is in higher

demand. This simple pricing mechanism aims to encourage off peak consumption – with the goal of

reducing peak demand.

2.3 The Network Operator

In most electricity networks, there is more than one network operator. The SWIS is unique to Australia;

Western Power is the sole network operator (AEMO 2016b). Western Power is responsible for operation,

maintenance and development of the SWIS. They must ensure all customers receive satisfactory

electricity supply (Western Power 2011). End users are charged according to various network tariffs

(Western Power 2016a). This report is focussed on end users serviced under Metered Demand Exit

Service, or Contracted Maximum Demand Exit Service tariffs. End users on these tariffs account for less

than 1% of those connected to the SWIS; however their annual consumption accounts for more than 40%

of total electricity consumed annually (Western Power 2016b).

2.4 Energy Retailers

Energy retailers buy, or bid for capacity, for their customers: the end users (Australian Energy Market

Operator 2015). To simplify energy costs for end users, energy retailers usually bundle all electricity supply

components into a fixed per unit energy price (Australian Energy Market Operator 2015). This allows

most end users to be charged by the amount of energy consumed in kilowatt hours (kWh), Figure 3.


14

Figure 3: Bundled and Unbundled Electricity Tariffs

Energy retailers prefer not to bundle energy tariffs for large end users. Large end users are a considerable

burden on electricity markets. Unbundling energy tariffs ensure all components of electricity supply are

passed through to the end user. This is enticing for end users as it usually results in lower costs; however,

end users become exposed to risk of high capacity charges if their demand increases.

2.5 Peak Demand

Peak demand in the South West Interconnected System occurs on both a daily and annual basis. Daily

peak demand is driven by business operation usually running from 8am to 7pm. Between December and

March, in 2014 maximum demand occurred between 5.30pm and 6.00pm (IMO 2014, 18). Power usage

throughout businesses usually typically follows a profile shown in Figure 4.

Figure 4: Commercial Load Profile Example

Annually, peak demand occurs through the summer months: December to March. The increased summer

temperatures cause cooling loads, such as ventilation, air-conditioning or refrigeration to operate more

12:00:00 AM 4:00:00 AM 8:00:00 AM 12:00:00 PM 4:00:00 PM 8:00:00 PM

kW

Power Usage Characteristic of Commercial Operations

Energy Retailer

Bundled Tariff

kWh

Electricity Supply Components

Generators Network

Operators Market

Operator Environmental

Schemes

kWh kVA kW kWh

Energy Retailer

Unbundled Tariff

End User

kW kWh kVA kWh

End User


15

frequently and at greater power levels. This period is referred to as the peak trading season. An end user

in Western Australia, is most likely to use their maximum demand during this time. Large end users such

as manufacturers, hospitals, and institutions are major contributors to peak demand. To ensure the

electricity network has sufficient capacity to meet such demand, maximum capacity requirements are

assigned to large end users. Large end users are assigned the maximum capacity they require annually.

Although maximum capacity may only be required a few times a year, large end-users must pay year-

round for the access of this capacity. The high cost of supplying peak demand has driven peak demand

reduction and management. Peak Demand Management is the management of maximum power use.

16

3 Available Research and Resources

Peak Reduction is a rapidly growing subject. Benetti et al. (2015) provide a systematic literature review

and classification of more than two hundred papers relating to electric load management (ELM). The main

subjects of ELM are peak demand response/management/reduction and smart grids. They find that

research or survey papers released every year are almost growing exponentially; the majority of these are

published in the United States of America. Much research is conducted from the perspective of electricity

network optimisation; managing combined peak demand is one of their greatest challenges.

Demand response is usually requested by large consumers, usually in the form of onsite generation to

release capacity into the networks. Demand response motivation is driven by providing the end user credit

for freeing up capacity. In most contexts, this is far more viable than pursuing strategies of network

augmentation. In contrast, demand reduction/management motivation is achieved by increasing end

user’s cost of demand. The meaning of demand reduction and management tend to cross over in many

papers. Usually reduction/management is achieved by energy efficiency or renewables.

Some research looks into AEMO pricing mechanisms for encouraging end users to reduce their peak

demand (Nelson and Orton 2016). Increasing cost of demand, or introducing penalty rates, inspire end

users to seek load reduction by demand response, energy efficiency, renewable energy and/or load

shifting.

Reducing residential peak demand is also an area of great interest; household are responsible for most

after hour demand increase. Sadineni & Boehm (2011) look at peak demand reduction/management of

residential dwellings at the substation level (Sadineni and Boehm 2011). Their simulations achieved a 65%

reduction in peak demand by load control and PV systems in a residential village in Las Vegas. They model

the peak reduction using Energy-10 software for residential dwellings and small buildings. Sadineni &

Boehm found load control (air-conditioning thermostats) was particularly effective in reducing peak

demand. Their report was primarily focussed on peak reduction on an average daily basis; there were

limited comments on peak reduction annually.

Modelling Peak Demand

Many models have been developed for estimating peak demand at the utility level for the purpose of

future infrastructure planning. Suganthi and Samuel (2012) review the growing number of models to


17

predict electricity demand. The move to complicated algorithms has allowed estimation accuracy to

improve (Mirlatifi, Egelioglu, and Atikol 2015), and (Dirks et al. 2015). These algorithms incorporate local

weather and population fluctuation predictions to estimate overall peak demand on a network.

Hong, Chang, and Lin (2013) examined how weather has impacted peak demand over the last 30 years,

finding demand, over consumption, to be more greatly influenced, particularly from heating, ventilation

and air-conditioning (HVAC).

Sehar, Pipattanasomporn, and Rahman (2016) present an energy management model to study peak

demand savings from photovoltaics (PV) and storage in demand responsive (DR) buildings. They look at

an established DR building and how its DR ability coupled with PV and ice storage can reduce and/or shift

its load out of a peak demand period. Due to time of peak demand and PV generation, they found the

combination of DR, PV and ice storage was required to maximise peak demand reduction; however ice

storage drastically increased energy consumption.

The National Renewable Energy Laboratory and the International Journal of Smart Grid and Clean Energy

have investigated demand charge mitigation with renewable energy. The consensus is that project

viability is highly dependent on demand charge structure and the shape of the load profile of end users

(Huat Chua, 2013) and (Neubauer, 2013). Additionally, in general, the less battery storage incorporated,

the better payback periods are. This is because battery storage has been traditionally very expensive, and

must be charged by either excess solar or the network in off peak times.

Current Software

Homer was one of the software packages identified to have the ability to simulate demand charge

reduction. Homer’s Advanced Grid Module allows network characteristics, including demand charges, to

be modelled. Homer Pro software can simulate a network with great detail. Energy charges can be defined

as simple rates, real time rates or scheduled rates. The demand charges, or rates can only be specified in

terms of $/kW/month, shown in Figure 5.


18

Figure 5: Homer Pro's Advanced Grid Rate Properties Input

The South West Interconnected System’s network and market demand charges cannot be entered

simply into Homer’s Rate Charges input. The fixed and variable maximum or contracted maximum

demand (CMD) network charges are constrained by thresholds; there is not a linear relationship. Many

other software suites deployed this method for calculating demand charge reduction, however in the

case of the WEM, market demand charges are dependent on the median measured peak trading

interval and temperature ratios. Therefore a more complex pricing mechanism is required.

19

4 Scope

To simulate the complexity of demand charge reduction in the wholesale electricity market, a model has

been developed to investigate the economics of demand reduction using renewable energy technologies.

To improve the reliability of demand reduction, the model concludes that a holistic approach should be

used, including a combination of load rescheduling, efficient lighting, solar generation and energy storage.

The model calculates an end user’s current cost of supply from their original load profile. The model

simulates load reduction from LED upgrades, solar penetration, energy storage and power factor

correction. From the simulated load, the model determines the new cost of supply, and thus calculates

the savings in supply including the reduced demand charges. Due to the extensive use of Microsoft Excel

in Verdia, it followed that the model should be built within such a platform.

To calculated the original cost of supply, network and market pricing mechanisms had to be investigated.

These pricing mechanisms were to be put into an Excel spreadsheet, capable of calculating the cost of

supply from the annual load profile of an end user.

The load rescheduling component was also developed. The model simulates the shifting of existing loads

to off peak times, differentiating weekday loads from weekend loads.

The LED Upgrade component determines the load reduction by comparing the power draw from the

current fixtures to the power draw of the proposed fixtures. This was to be done using Verdia’s LED

replacement tables.

The variability of the Perth solar resource was investigated to determine the potential loss of generation

through the peak trading intervals and its consequences on demand.

The reduction in demand must ensure that the assigned capacity requirements of end users can be

reduced confidently. For weather events that cause significant drop in renewable generation, there must

be contingency to ensure load can be serviced without drawing more capacity. It is envisaged that the

energy storage should supplement the load under times of generation loss by cloud events. It is suggested

that a cloud camera could be coupled with a battery management system to ensure Energy Storage

Systems, ESS, could provide such contingency; these have been increasing in efficiency and reducing in

price over the past few years. The viability of utilising ESS to reduce demand is investigated with the use

of this model.


20

Once the model was completed, it was tested against real recorded data to determine whether the

simulations were representative of a real load. Verdia looks for business cases where the project break

even time would be reached prior to warranty expiry. The breakeven time is the number of years of

project savings required to equal project investment, often referred to as the payback period.

21

5 Approach to modelling peak demand reduction

To model the economic implications of peak demand reduction by a renewable energy project, we need

to determine the difference between the cost of electricity supply before and after the project is. The

current cost of supply comprises the factors of energy generation, network usage, environmental schemes

and market charges. There are 16 network operators in Australia, each with their own pricing structures.

To simplify the model, it has been tailored to the South West Interconnected System within the tariffs of

Western Power. It is intended that, in the future, the model will be adapted to be utilised for end users

on other networks. This is likely to be challenging due to the vast number of network operators and energy

retailers.

5.1 Market Operator Demand Charge Pricing Mechanisms

The market operator must ensure there is sufficient capacity for all end users throughout the year. To

do this, the market operator assigns end users an Individual Reserve Capacity Requirement (IRCR). This

is established by taking the median measurement during the peak trading season. Twelve

measurements of real power requirements, in kilowatt hours (kWh) are taken at random during 2pm

and 6pm throughout February and March. The median value measured determines the IRCR for the next

capacity year, commencing the following October.

5.2 Network Operator Demand Charges

An end user is also charged for peak demand by the network operator. The network operator takes energy

measurements every 30 minutes to identify an end user’s electricity consumption and capacity

requirements. The network operator will then assign a pricing structure, or tariff, depending on the end

user’s capacity requirements.

Western Power is the network operator of the South West Interconnected System. Western Power

provides the following demand tariffs (Western Power 2016a): -

• High Voltage Metered Demand Exit Service (RT5)

• Low Voltage Metered Demand Exit Service (RT6)

• High Voltage Contract Maximum Demand Exit Service (RT7)

• Low Voltage Contract Maximum Demand Exit Service (RT8)


22

These tariffs are assigned to large end users, and are intended to provide an incentive to reduce their

respective peak demand. Western Power measures peak demand in terms of complex power in kilo volt-

amperes (kVA).

5.2.1.1 High and Low Voltage Metered Demand Exit Service

Western Power refers to the high and low voltage metered demand tariffs as Reference Tariff 5 (RT5) and

Reference Tariff 6 (RT6) respectively (Western Power 2016a). End users are supplied under these tariffs if

their seasonal usage is highly variable. The maximum demand is measured using a rolling 12 months peak

function. If the maximum demand is not exceeded within a 12-month period, it is reset to the next highest

peak measured (Western Power 2016a). If, however, the peak is exceeded within 12 months, it will adjust

to that new peak and the cycle starts again, Figure 6 .

Figure 6: Rolling Peak Illustration

To further encourage a reduction in peak demand, a discount proportional to the off peak to total

consumption ratio also applies to end users within this tariff. Demand length charges also apply to end

users. These charges are proportional the distance an end user is from their substation, to account for

energy losses and transformer adjustment required to deliver voltage within the acceptable range. Fixed

metering charges also apply.

5.2.1.2 High and Low Voltage Contract Maximum Demand Exit Service

The market operator refers to these as RT7 and RT8. These tariffs are applicable to end users with

instantaneous power requirements greater than 1000kVA. A contracted maximum demand (CMD) is

calculated according to the end users total connected load, their load factor and power factor (Western


23

Power 2016b)The end user is charged according to their CMD; if the end user can reduce their load factor,

maximum demand/average demand, their CMD will reduce. For any demand greater than the CMD,

excessive network usage charges apply. RT7 and RT8 end users also receive demand length charges and

fixed metering and administration charges (Western Power 2016b).

5.3 Peak Demand

Western Australia’s Mediterranean climate is characterised by warm to hot summers and mild winters.

Most West Australians retreat from the heat to air-conditioned offices, cars and homes. These cooling

loads along with refrigeration, are the primary cause of peak demand. There is a strong relationship

between extremes of temperature and high electrical loads. Figure 7 shows the relationship between

total load on the South West Interconnected System and the temperature during the peak trading

season of 2017.

Figure 7: SWIS Load Data vs Temperature

To examine the load of an end user, measured 30-minute interval load data can be obtained from the

network operator’s online portal (Western Power 2016a).

5.4 Cost of Supply

Interval data is usually provided over several years through 30-minute interval energy measurements.

Interval data is usually received for several years in a z-format csv file. Real energy (kWh) and reactive

energy (kVArh) measurements are provided every half hour; each day is represented by a new row of

data. This data had to be converted to linear form by a z-converter algorithm. To avoid storing excessive

quantities of data, it is assumed that load operation will be consistent year to year. This allows load

analysis to be conducted over a period of one year.

(ºC)

5 (ºC)

10 (ºC)

15 (ºC)

20 (ºC)

25 (ºC)

30 (ºC)

35 (ºC)

40 (ºC)

45 (ºC)

50 (ºC)

MW

500 MW

1,000 MW

1,500 MW

2,000 MW

2,500 MW

3,000 MW

3,500 MW

4,000 MW

4,500 MW

5,000 MW

1 Jan 5 Jan 9 Jan 13 Jan 18 Jan 22 Jan 26 Jan 30 Jan 3 Feb 7 Feb 12 Feb 16 Feb 20 Feb 24 Feb 28 Feb 4 Mar 9 Mar 13 Mar 17 Mar 21 Mar

Tem

per

atu

re

Load

AEMO SWIS Load DATA vs Murdoch MET Temperature DATA

Operational Load (MW) Temperature (ºC)

Half hourly data measurements


24

Energy is a measure of power over time. To convert half hourly energy consumption to half hour power

measurements, the kWh readings are multiplied by 2, shown by equation 1. The assumption is made that

power is an average value across the half hour period.

[1]

This process is also applied to real and reactive energy measurements to convert to power. To calculate

apparent power “S”, the Pythagorean is found from the real power “P” and reactive power “Q”, equation

2.

[2]

From real, reactive and apparent power values the current cost of supply can therefore be computed.

Energy charges are calculated by multiplying total peak hour and off-peak hour consumption by

corresponding per unit tariffs. Network charges are calculated corresponding to High and Low Voltage

Metered Demand Exit Service (RT5 and RT6) or High and Low Voltage Contract Maximum Demand Exit

Service (RT7 and RT8). For RT5 and RT6 the model identifies maximum demand, corresponding demand

range and discount factors. Equation 3 shows how off-peak consumption is proportional to the discount,

aiming to discourage peak consumption.

Discount = (Off Peak Consumption

Total Consumption) ∗ Discount_Factor

[3]

For RT7 and RT8 excess network usage is identified by equation 4, demand exceeding contracted

maximum demand (CMD) is charged with excess usage fees. The maximum excess usage value of each

billing period is used to calculate these charges.

Excess Network Usage = max(demand − CMD)

[4]

where ‘time’ is in hours = 0.5

𝐸𝑛𝑒𝑟𝑔𝑦 = 𝑃𝑜𝑤𝑒𝑟 × 𝑡𝑖𝑚𝑒 = 𝑃 × 0.5

𝑃𝑜𝑤𝑒𝑟 =𝐸𝑛𝑒𝑟𝑔𝑦

0.5= 𝐸𝑛𝑒𝑟𝑔𝑦 × 2

𝑆 = √𝑃2 + 𝑄2


25

All other demand charges are calculated from the CMD. Lower CMD results in lower demand charges, but

higher excess network usage charges. Higher CMD means higher demand charges yet lower excess

network usage charges. This pricing mechanism favours flatter loads, reducing demand peaks means a

lower CMD can be chosen without excess network usage charges.

Market charges are calculated by finding the Individual Reserve Capacity Requirement (IRCR) from the

median real power value throughout the 12 peak interval trading periods shown in Figure 8, (AEMO

2016b).

Figure 8: 12 peak trading intervals

From the median real power demand, the IRCR is calculated using the corresponding IRCR ratios, Figure 9

(AEMO 2016b). IRCR is calculated by multiplying mean real power demand by Temperature Dependent

Load (TDL) and the Total Ratio.

Figure 9: IRCR Ratios

The 12 peak trading intervals and IRCR ratios are published in compliance with the Wholesale Electricity

Market (WEM) Rules. These pricing mechanisms have been input to the Excel model to ensure the current

cost of demand can be calculated from a years’ worth of end user interval data. To calculate the potential

demand reduction, various modules with in the model would simulate effects of rescheduling loads and

replacing lighting fixtures with more efficient LEDs. Solar penetration and energy storage should be sized

to the rearranged, reduced load.

5.5 Load Rescheduling

Running loads outside of peak demand periods is the most effective measures to flatten peak demand. It

is appreciated most loads are time dependent, and rescheduling loads to other times may result in


26

disruption of business productivity and profitability. It is envisaged that part of the assessment process

should be to determine whether any loads can be run at other times, and the power that they require.

Load rescheduling aims to reduce peak demand by changing the operation time of non-time dependent

loads (NTDL) to low demand periods. To model load rescheduling over the existing interval data, the

model calculates a load variation. The model requires an input of time and power rating for NTDLs to

create the original NTDL profile, and the proposed time to reschedule the NTDLs. The model has been set

up to allow load start and finish times to be specified hourly and for weekday inputs to differ from

weekend inputs.

Figure 10 shows the isolated operation time change of a constant NTDL to a period of lower demand.

Figure 10: Original and Proposed NTDL Profiles

The model compares the existing NTDL profile to the proposed NTDL profile to find a net load change,

Figure 11.

kW

100 kW

200 kW

300 kW

400 kW

500 kW

600 kW

12:00 AM 3:00 AM 6:00 AM 9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM

Original and Proposed Load Profiles

Original NTDL Proposed NTDL

Non-Temperature Dependent Loads


27

Figure 11: Net Load Change, NTDL

The model allows both real and reactive power loads to be rescheduled. The corresponding week or

weekend day load reduction for real and reactive power is added to each data stream, to appropriate

week and weekend days. By adding the net load change to the original real power load profile, the

proposed load profile is obtained; shown in Figure 12.

Figure 12: Original and Proposed Rescheduled Loads

Rescheduling load as shown in Figure 12 demonstrates that maximum load is reduced. It is assumed that

loads are modelled at normal steady state operation. Although start up or inrush load (currents) may be

significantly higher than that of normal steady state operation, they only last momentarily, in a half hourly

measurement they are negligible. It is also assumed that the load rescheduling on one week day is

representative of the load rescheduling achievable on all weekdays, the model allows weekend operation

to be rescheduled, or to remain unchanged.

-600 kW

-400 kW

-200 kW

kW

200 kW

400 kW

600 kW


Net Load ChangeNon-Temperature Dependent Loads

kW

500 kW

1,000 kW

1,500 kW

2,000 kW

2,500 kW

3,000 kW

3,500 kW

4,000 kW

4,500 kW

5,000 kW


Original and Proposed Rescheduled Loads

Original Load Proposed Load

Rescheduled


28

5.6 Lighting

Light emitting diodes, LEDs, are significantly more efficient than conventional incandescent, halogen and

fluorescent lights. This can be measured by luminous efficacy, ηL lumens per watt (lm W¯¹). Incandescent

and fluorescent lights have a luminous efficacy of approximately 13 lm W¯¹ and 90 lm W¯¹, whereas

standard LEDs can produce approximately 150 lm W¯¹ (Narukawa et al. 2010). This allows LEDs to

produce the same amount of light using a lot less power. Verdia uses a general rule of thumb that LEDs

should have approximately 50% of the power rating of the fixture they are replacing. This assumption

has been used to help model load reduction by LED luminaire upgrades. Usually a lighting assessment of

an end user’s site would be conducted where fixture types, quantities and power ratings would be

recorded.

Conventional lighting loads usually account for around 30% of total power consumption (Yun, Kim, and

Kim 2012). Using Verdia’s general rule of thumb, 15% of total load can be reduced by replacing all

luminaires with LED fixtures. To model the reduction in load by LED luminaire replacement more closely,

a similar process to load rescheduling is adopted. The user provides existing luminaire type, quantity

and operation time (weekday and weekend). The model identifies luminaire rating from its lighting data

sheet and calculates existing load, Figure 13. The model chooses replacement LED luminaires to satisfy

equivalent lumens from the lighting data sheet and calculates the proposed lighting load, also shown in

Figure 13.

Figure 13: Original Lighting and Proposed LED Loads

kW

50 kW

100 kW

150 kW

200 kW

250 kW

300 kW

350 kW

400 kW

450 kW


Original and Proposed LED Loads

Original Lighting Load Proposed LED Load

Lighting Loads


29

The model calculates the net load reduction by finding the difference of current and proposed LED loads.

The corresponding week or weekend day load reduction is applied to the proposed rescheduled real

power data stream. Example reduced LED load is shown Figure 14.

Figure 14: Original, Rescheduled and LED Reduced Loads

5.7 Solar

As the majority of Verdia Western Australia’s clients are situated in the Perth metropolitan area, it was

decided that the issue of solar generation was to be represented by Perth’s solar generation data. A years’

worth of solar generation data was obtained from the National Renewable Energy Laboratory’s solar

energy production calculator, PVWatts. Optimal solar generation data was obtained for a 100kW system

facing North and at a tilt of 30 degrees. Depending on the capacity required, the solar capacity can be

scaled as desired.

The performance can also be de-rated according to the orientation (azimuth) and inclination. This is done

by identifying the appropriate loss factor from the solar panel orientation and inclination losses data table

from The Australian Solar Radiation Data Handbook, (Australian and New Zealand Solar Energy Society

(ANZSES) 2006). The model also has the capability to reflect the impact of inverter sizing. If generation

exceeds inverter rating, a simple excel “if” function clips generation to match inverter size, demonstrated

in Figure 15.

kW

500 kW

1,000 kW

1,500 kW

2,000 kW

2,500 kW

3,000 kW

3,500 kW

4,000 kW

4,500 kW

5,000 kW


Original, Rescheduled and LED Reduced Loads

Original Load Proposed Rescheduled Load Proposed LED Reduced Load


30

Figure 15: Modelled PV Output

The model has inputs for three different arrays to allow sizing of various sizes, azimuths and tilts. The total

generation is combined into a PV generation data stream. The PV generation data stream is subtracted

from the proposed LED reduced load data stream to provide the PV penetrated load, Figure 16.

Figure 16: Original, Rescheduled, LED and PV Reduced Loads

Loss of generation was very important to consider. Although Perth usually has clear skies and ideal solar

generation conditions through the peak demand trading period, Perth has seen its highest daily rainfall

(hence heaviest clouds) in January and February 2017 (Bureau of Meterology 2017). AEMO identifies

maximum loads occurring within the hot season, so the relationship between temperature and load was

investigated.

Figure 17 shows the SWIS’s total load compared with temperature throughout the hot season peak

trading period in 2017. There appears to be a very strong correlation between high temperature and peak

kW

50 kW

100 kW

150 kW

200 kW

250 kW

300 kW

350 kW

12:00:00 AM 4:00:00 AM 8:00:00 AM 12:00:00 PM 4:00:00 PM 8:00:00 PM

Modelled PV Power Output

PVSyst Output

PVSyst Output withInverter Clipping

from PVSyst Data

kW

500 kW

1,000 kW

1,500 kW

2,000 kW

2,500 kW

3,000 kW

3,500 kW

4,000 kW

4,500 kW

5,000 kW

12:00:00 AM3:00:00 AM 6:00:00 AM 9:00:00 AM 12:00:00 PM 3:00:00 PM 6:00:00 PM 9:00:00 PM

Original, Rescheduled, LED and PV Reduced Loads

Original Load Proposed Rescheduled Load Proposed LED Reduced Load Proposed PV Reduced Load


31

loads; lower temperatures tend to result in lower loads. This is explained by the dominance of cooling

loads on the SWIS (IMO 2014, 18). 10th February 2017, Perth received one of the heaviest days of rainfall

on record (Bureau of Meteorology, 2017). Figure 17 shows solar radiation was greatly impacted. Solar

radiation dropped approximately 60%. The low solar radiation resulted in lower temperatures, shown in

Figure 17 that resulted in less cooling requirements, hence lower total load.

Observation of recent data (Bureau of Meteorology, 2016 and 2017), leads to the same conclusions

through the hot season peak trading period. Heavy cloud or rain events reduces solar radiation; reduced

solar radiation results in milder temperatures and less cooling loads.

32

Figure 17: AEMO SWIS Load Data vs Murdoch Meteorology Temperature Data

Figure 18: AEMO SWIS Load Data vs Murdoch Meteorology Solar Radiation Data

(ºC)

10 (ºC)

20 (ºC)

30 (ºC)

40 (ºC)

50 (ºC)

MW

1,000 MW

2,000 MW

3,000 MW

4,000 MW

5,000 MW

1 Jan 2017 5 Jan 2017 9 Jan 2017 13 Jan 2017 18 Jan 2017 22 Jan 2017 26 Jan 2017 30 Jan 2017 3 Feb 2017 7 Feb 2017 12 Feb 2017 16 Feb 2017 20 Feb 2017 24 Feb 2017 28 Feb 2017 4 Mar 2017 9 Mar 2017 13 Mar 2017 17 Mar 2017 21 Mar 2017

Tem

per

atu

re

Load

AEMO SWIS Load DATA vs Murdoch MET Temperature DATA

Operational Load (MW) Temperature (ºC)

Half hourly data measurements

W/m²

200 W/m²

400 W/m²

600 W/m²

800 W/m²

1000 W/m²

1200 W/m²

1400 W/m²

1600 W/m²

1800 W/m²

MW

500 MW

1,000 MW

1,500 MW

2,000 MW

2,500 MW

3,000 MW

3,500 MW

4,000 MW

4,500 MW

5,000 MW

1 Jan 2017 5 Jan 2017 9 Jan 2017 13 Jan 2017 18 Jan 2017 22 Jan 2017 26 Jan 2017 30 Jan 2017 3 Feb 2017 7 Feb 2017 12 Feb 2017 16 Feb 2017 20 Feb 2017 24 Feb 2017 28 Feb 2017 4 Mar 2017 9 Mar 2017 13 Mar 2017 17 Mar 2017 21 Mar 2017

Sola

r R

adia

tio

n

Load

AEMO SWIS Load DATA vs Murdoch MET Solar Radiation DATA Operational Load (MW)Half hourly data measurements


33

5.8 Energy Storage

The temperature – load – solar radiation correlation does not apply to cloud events. To be able to

quantify the reduction in peak demand by solar penetration, contingency for cloud events must be

available. Minimum hourly solar generation was compared to average hourly solar generation

throughout the peak trading season. Based on examination of National Renewable Energy Laboratory

PV Watts data for Perth, it is estimated that during the peak trading season, solar generation may drop

to as low as 20% of average generation, Figure 19.

Figure 19: National Renewable Energy Laboratory PV Watts Average and Minimum simulated 2000kW Solar Output

Because of this, it is proposed that energy storage should be sized at 80% of average solar output

throughout the peak trading season. The model has been set up to provide excess solar or off-peak grid

charging. There is also input to adjust round trip charging efficiency and depth of battery discharge

values. The energy storage capacity recommended is scaled accordingly.

Energy storage can also be selected to assist in flattening the load. From a target maximum demand,

energy storage required is determined. The area between the power curve and target maximum

demand is the additional energy storage capacity.

kW

200 kW

400 kW

600 kW

800 kW

1000 kW

1200 kW

1400 kW

12:00 AM 3:00 AM 6:00 AM 9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM 12:00 AM

Solar Output

Minimum (kW) Average (kW)


34

Figure 20: Target Maximum Demand

5.9 Power Factor Correction

Power factor is the ratio of real power to complex power. Electricity systems prefer the loads of end

users to have power factor greater than 0.8 (Western Power 2011). The majority of loads draw real

power; however loads containing rotating magnetic fields require inductive power. Reducing the real

component of power requirements with LEDs and PV penetration cause the power factor to reduce

even further. To obtain a satisfactory power factor, capacitor banks can be utilised to inject reactive

power (negative inductive power) to lower the net reactive power.

The model calculates the power factor for every set of measurements in the data stream, the ratio of real

power, P to apparent power, S requirements. If power factor is less than the user’s specified target power

factor PFtarget, the power factor correction required, QPFC is calculated, as in equation 5 and 6.

𝐏

𝐏𝐅𝐭𝐚𝐫𝐠𝐞𝐭

= 𝐒𝐭𝐚𝐫𝐠𝐞𝐭 = √𝐏𝟐 + 𝐐𝟐𝐭𝐚𝐫𝐠𝐞𝐭

𝐐𝐭𝐚𝐫𝐠𝐞𝐭 = √𝐒𝐭𝐚𝐫𝐠𝐞𝐭𝟐 − 𝐏𝟐 = √

𝐏


𝟐

− 𝐏𝟐

[5]

kW

1000 kW

2000 kW

3000 kW

4000 kW

5000 kW

6000 kW

12:00:00 AM 3:00:00 AM 6:00:00 AM 9:00:00 AM 12:00:00 PM 3:00:00 PM 6:00:00 PM 9:00:00 PM 12:00:00 AM

Identification of energy storage required to achieve Target Maximum Demand

Target Maximum Demand Demand after Solar Penetration


35

𝐐𝐏𝐅𝐂 = 𝐐𝐭𝐚𝐫𝐠𝐞𝐭 − 𝐐𝐚𝐜𝐭𝐮𝐚𝐥

𝐐𝐏𝐅𝐂 = √𝐏


𝟐

− 𝐏𝟐 − 𝐐𝐚𝐜𝐭𝐮𝐚𝐥

[6]

QPFC is negative, indicating the requirement of capacitive reactive power to lower the inductive load.

QPFC is calculated in units of reactive power, kVAr. The model utilises a simple algorithm, Figure 21 to

determine the power factor correction required to maintain power factor throughout the year.

Figure 21: Power factor correction algorithm

The total reduction of charges is defined as the peak demand reduction savings. The savings as a

proportion of project cost can be used to determine whether a project is viable and should be

implemented.

5.10 New Cost of Supply

From the resulting load, the new cost of supply can then be estimated. There are limitations on the

accuracy of calculating the new cost of supply. Loads are highly variable; they are subject to change by

individuals, operation or weather behaviour (Hong, Chang, and Lin 2013) and (Neubauer, 2013). Using

an end user’s latest interval ensures that results are as accurate as possible; however, the results are

vulnerable to changes in loads.

In reality, the new cost of supply will not be realised immediately. The AEMO determines an end user’s

individual reserve capacity requirements (IRCR) during the peak trading season AEMO (2014), February

and March. The IRCR determines the capacity to be allocated for end users for the following capacity year,

𝑃𝐹𝑡𝑎𝑟𝑔𝑒𝑡

𝑃𝐹 < 𝑃𝐹𝑡𝑎𝑟𝑔𝑒𝑡

𝑸𝑷𝑭𝑪 = 0 𝑸𝑷𝑭𝑪 = √

𝑷

𝑷𝑭𝒕𝒂𝒓𝒈𝒆𝒕

𝟐

− 𝑷𝟐 − 𝑸𝒂𝒄𝒕𝒖𝒂𝒍

𝑴𝒂𝒙(−𝑸𝑷𝑭𝑪)

𝐹𝑎𝑙𝑠𝑒 𝑇𝑟𝑢𝑒


36

commencing in October. Because of this a reduction in demand occurring just after the Hot Season Trading

Period may not reduce and end user’s demand allocation, hence cost for up to 20 months,

Figure 22.

Figure 22: Demand Reduction implementation and economic benefit

If a demand reduction project was to be completed just before the peak trading season, demand charge

savings would be realised in October that year.

5.11 Estimating project costs

The cost of such a demand reduction project is comprised of many aspects. The model allows capital

cost of each project to be varied as required. For example; cost per watt solar, cost per kilowatt hour

energy storage or cost per kilo volt-ampere power factor correction. Verdia conducts regular market

assessments to obtain current pricing indexes.

Hot Season

Peak Trading

Capacity

Year Start

Hot Season

Peak Trading

Capacity

Year Start

October

2017

January -

March 2017

October

2018

January -

March 2018

Demand

Reduction

Cost

Reduction

Demand

Reduction


37

6 Development

As the model is simulating the load over a year using 17,520 real and reactive energy measurements (2

readings per hour x 24 hours per day x 365 days per year), various limitations had to be made to reduce

the file size of the model.

The model is currently limited to the South West Interconnected System, in order to avoid the need to

store additional network operator pricing structures.

The model was developed initially to contain data photovoltaic (PV) data for varied azimuths and tilts.

Storing large amounts of data caused Excels to increase its calculation time. To speed up the performance

of Excel’s, the Solar Panel Orientation & Inclination Losses data was utilised from The Australian Solar

Radiation Data Handbook. This allows a single stream of annual solar data to be simulated at different

azimuths and tilts. The model is currently limited to data for Perth. The viability of peak demand reduction

is dependent on many variables, though the shape of an end user’s load profile is predominant.


38

6.1 Model Navigation

To simplify operation of the model, a colour coding key and navigation panel has been positioned at the

top of every input sheet as shown in Figure 23 and Figure 24. The colour coding provides information

regarding the cell’s role. These colours are consistent with the model flow chart that provides an overview

of model operation as depicted in Figure 25.

• Required input (blue): Worksheets or cells where user is required to input information specific to end

user/project

• Default input (purple): Worksheets or cells where default data inputs are held; it is unlikely that these

would need to be changed more than once a year

• Calculations (white): Worksheets or cells where calculations occur, cells should not be modified

• External Information (yellow): External information required i.e. load data, number of luminaires

• Outputs (green): Worksheets or cells where outputs are provided

Figure 23: Model Colour Coding Key

The navigation panel contains buttons that move the user throughout the model. The buttons are

intended to be intuitive by using universal user interface media control convention symbols commonly

found on media players and remote controls, as in Figure 24. Each button is hyperlinked to corresponding

worksheets, to allow user to move as desired. The Load Assessment worksheet is central to user feedback,

so has been made accessible from any page. It provides an insight to load manipulation and ‘live’ feedback

of peak demand statistics. Navigation buttons are hyperlinked and follow the colour coding convention of

light blue to prompt user input/use.

Figure 24: Navigation Panel

6.2 Model Operation

The model is intended to be utilised after obtaining interval data and conducting a site assessment of the

end user’s premises. Figure 25 shows an outline of the model’s process. Yellow items are representative

of either load data or assessments from site inspections. Blue items signify worksheets within the model

that require user input.

Load Assessment Previous Input Next Input Output Instructions

Required Input Default input, can be

changed Calculations External Information Outputs


39

Figure 25: Model Overview

Power Factor Correction

Energy Storage

Solar

Lighting

Load Rescheduling

Standing and Interval Data

Real Energy Input (kWh) Reactive Energy Input

(kVArh)

New Annual Expense

Non Critical Load Rescheduling

Non Critical Loads Load Assessment

LED Retrofitting Current Lighting Lighting Assessment

Replacement Lighting Data

Solar Generation Array Details Viable Space

NREL PV Watts Data

Sizing Energy Specifications Desired Maximum

Demand

Power Factor Sizing Required Power

Factor

Savings

Calculations

Obtain Standing and Interval Data


40

Purple items indicate worksheets with default data values that may be adjusted; for example, network

operator price lists or solar radiation data. White items specify where calculations are occurring. The

model uses a year’s worth of interval data for the entirety of the calculations. This ensures that annual

variation is maintained throughout the various stages of reducing the load within the model.


41

7 Model Validation

The results of key interest are the modelled new load, and the new cost of supply. To ensure the results

from the model are accurate, it is necessary to compare these results with real data. Ideally, one year’s

worth of 30-minute load data before implementation of a peak demand reduction project would be used

in the model to simulate demand reduction. The simulated results could then be compared with the real

load data, post project implementation. Unfortunately, no data was available specific to a large end user

in Western Australia having conducted such a project.

Instead, to test the model, Murdoch University’s LED upgrade project was utilised. From July 2015,

approximately 4000 conventional fluorescent light fixtures were replaced with efficient LED luminaries,

with further installation still to be completed. An approximate installation schedule was used to simulate

number of fixtures installed throughout 2016. Data from January to June in 2016 was compared to

corresponding data for 2017. It was expected the simulated data would contain variations from the real

data. Student and staff numbers, addition of further units, or more efficient loads and weather variation

could cause the power drawn in 2017 to divert from the level predicted. It was noted that there was a

great variation throughout the summer months due to the abnormal summer of 2017. To filter out the

inconsistent weather impact, data was only compared using the months of April and May.

7.1 Simulated and Real Data Comparison

Murdoch University’s load data and LED upgrade schedule was provided by the university’s Energy

Manager. 30-minute interval load data for both the university’s NMIs (National Meter Identifiers) from

July 2014 to June 2017 had been accessed from the network operator. After converting the z-formatted

data to linear form, the kWh and kVArh streams for each NMI were combined to form total energy data

streams. The data from July 2015 to June 2016 was input to the model.

The LED upgrade project schedule installation was staggered across various zones, to ensure minimal

interruption of staff and students. The approximate lighting numbers and project start and completion

dates for each zone were provided. The lights replaced so far are approximately 2300; 1200mm x 300mm

recessed fluorescent double tube troffers and 1700; 600mm x 600mm fluorescent four tube troffers. Both

the double and four tube troffer lights are estimated to draw 80W. It is estimated that another 20W is

lost through the fixtures. It was assumed that installation rates were constant within each zone’s project


42

timeline; from this an overall rate of installation was obtained. From the installation rates, total installed

fixtures could be determined with respect to time, Figure 26.

Figure 26: Assumed LEDs installed with respect to time

From the total fixtures installed, the total load reduction could be calculated. The load reduction per

fixture is equivalent to the power draw from the original fixture minus the power drawn from the new

fixture. As the original fixtures drew approximately 100W and the new fixtures will draw 40W, the load

reduction per light fixture is 60W. The model utilised this to find the load reduction with respect to each

day throughout the duration of the installation, and could provide the resultant load anticipated for July

2016 – June 2017 when most of the installation had been completed.

There was significant variation when comparing the simulated 2015 – 2016 data to the real 2016 – 2017

data. The data is shown on the same axis in Figure 27

0

500

1000

1500

2000

2500

3000

3500

4000

4500

Jul 2015 Oct 2015 Jan 2016 Apr 2016 Jul 2016 Oct 2016 Jan 2017 Apr 2017 Jul 2017

Tota

l LED

Fix

ture

s In

stal

led

Total LED fixtures installed

-1,000 kW

kW

1,000 kW

2,000 kW

3,000 kW

4,000 kW

5,000 kW

6,000 kW

7,000 kW

Jun 2016 Aug 2016 Oct 2016 Dec 2016 Feb 2017 Apr 2017 Jun 2017

Full year simulated data

Before LED Installation Simulated LED Installation After LED Installation


43

Figure 27.

To ensure week days correlate, the 2017 data needed to be shifted forwards by 2 days to account for the

extra day—2016 being a leap year.

Figure 27: Simulated and real LED load reduction

It is concluded that the significantly cooler spring and summer experienced from October 2016 to April

2017 has caused the deviation from that predicted. Due to the cooler temperatures, cooling loads would

have been lessened, and at lower power levels to those of 2016. To filter most of the error caused by

weather variation, only the data from April and May was assessed. Figure 28 shows the average daily load

profile throughout these months.

Figure 28: Average Simulated and Real LED Reduced Load, May

-1,000 kW

kW

1,000 kW

2,000 kW

3,000 kW

4,000 kW

5,000 kW

6,000 kW

7,000 kW


Full year simulated data

Before LED Installation Simulated LED Installation After LED Installation

kW

500 kW

1,000 kW

1,500 kW

2,000 kW

2,500 kW

3,000 kW

3,500 kW

12:0

0 A

M

12:3

0 A

M

1:00

AM

1:30

AM

2:00

AM

2:30

AM

3:00

AM

3:30

AM

4:00

AM

4:30

AM

5:00

AM

5:30

AM

6:00

AM

6:30

AM

7:00

AM

7:30

AM

8:00

AM

8:30

AM

9:00

AM

9:30

AM

10:0

0 A

M

10:3

0 A

M

11:0

0 A

M

11:3

0 A

M

12:0

0 P

M

12:3

0 P

M

1:00

PM

1:30

PM

2:00

PM

2:30

PM

3:00

PM

3:30

PM

4:00

PM

4:30

PM

5:00

PM

5:30

PM

6:00

PM

6:30

PM

7:00

PM

7:30

PM

8:00

PM

8:30

PM

9:00

PM

9:30

PM

10:0

0 P

M

10:3

0 P

M

11:0

0 P

M

11:3

0 P

M

May

Average Simulated and Real LED Reduced Load, May


44

The average load profiles for May shows there is a close correlation between the simulated LED load and

the measured data after the LED installation. The average simulated load reduction through May was

240kW, the average measured load reduction was 245.38kW. The average error of 5.38kW represents

2.24% under estimation of load reduction. This could be caused by install rate assumption errors, as

estimated load reduction of 5.38kW is equivalent to about 90 lights. The assumption of the power draw

of the original fixture may also vary slightly depending on the age of the fixture. These errors, however,

are also likely to be caused by the variation of weather, impacting resulting loads and changes in student

and staff numbers or behaviours. Given the vast opportunities for load variation between May 2016 and

May 2017, the error of 2.24% was smaller than expected.


45

7.2 Case Study

Murdoch University has been used as a case study to test the viability of a demand reduction project.

Building upon the existing LED upgrade project, additional solar, battery storage and power factor

correction has been used in the model. In addition to the 4000 LED fixture upgrades, the specifications in

Table 1 were entered in to the model.

Table 1: Hypothetical Demand Reduction Project

Component Specifications

Solar Array 2000kW (North Facing, 30-degree tilt)

Energy Storage 7000kWh (off-peak charging)

Power Factor Correction 1500kVA

Load profile

Murdoch University’s (complex power) load profile, Figure 29, shows high summer load (January – March)

and is relatively constant for the remainder of the year. This characteristic is typical for SWIS end users

(AEMO 2016b). Excess network usage charges apply at the times that load exceeds CMD, shown by the

dashed line. This only occurred a few times during March during the period assessed.

Figure 29: Murdoch University’s Annual Load Profile

The annual apparent load profile is examined daily in Figure 30, the dark blue data points show peak hour

load and light blue show off peak hour load. Load frequency can be gauged by the intensity of data points.

Outliers appear very pale while high frequency load values are more intense. Peak hour includes weekdays

from 8 am to 10pm, off-peak hour is all other times. Usually peak hour/weekdays loads are when peak

demand occurs. Average weekday and weekend loads are highlighted by white. Although weekend load

kW

1,000 kW

2,000 kW

3,000 kW

4,000 kW

5,000 kW

6,000 kW

7,000 kW


Annual Load Profile


46

is approximately 2000kVA and average weekday load ranges from 2000kVA to 4000kVA, the maximum

load drawn was almost 6000kVA.

Figure 30: Murdoch University’s Annual Load Profile

Murdoch University’s annual consumption was 22.6 giga-watt-hours (GWh) per annum, 12GWh peak hour

and 10.6GWh off peak hour. During the interval data period, the model reports the maximum demand

(maximum apparent power) occurred during peak hour: 5976kVA at 1pm, 14th March. Maximum off-peak

hour demand was 3415kVA, at 7.30am on 11th February.

Excel’s index function within the model can easily be used to determine the individual reserve capacity

requirement measured across the previous year’s 12 peak trading intervals. These readings measure real

power demand in kilo-watts, kW Table 2.

Table 2: Corresponding Real Power Measurements, Peak Trading Intervals

Peak Trading Interval kW

8/02/2015 16:30 2526

8/02/2015 17:00 2515

8/02/2015 17:30 2447

9/02/2015 16:30 4310

9/02/2015 17:00 4075

9/02/2015 17:30 3880

10/02/2015 16:30 4465

10/02/2015 17:00 4238

10/02/2015 17:30 3660

14/03/2015 16:30 2744

14/03/2015 17:00 2671

14/03/2015 17:30 2530

kW

1000 kW

2000 kW

3000 kW

4000 kW

5000 kW

6000 kW

7000 kW


Peak and Off Peak Hour Load

Off Peak kVA

Peak kVA

Average Weekend Load

Average Weekday Load


47

The median of these measurements is 3202kW. This median measurement becomes applicable at the

beginning of the capacity year in October. For each month in the capacity year, the median measurement

is multiplied with AEMO’s temperature dependent load (TDL) ratio and total ratio, Table 3. This value is

the individual reserve capacity requirement, IRCR. At the end of the capacity year, the median measured

value resets to the most recent peak trading season, and so the cycle continues.

Table 3: Capacity Year and Individual Reserve Capacity Requirements

Capacity Year Capacity Month TDL_Ratio Total_Ratio IRCR

2015

January 2016 1.5639 0.9932 4974

February 2016 1.5675 0.9914 4976

March 2016 1.5724 0.9888 4978

April 2016 1.5719 0.9891 4978

May 2016 1.5741 0.9881 4980

June 2016 1.5752 0.9875 4981

July 2016 1.5668 0.9915 4974

August 2016 1.5675 0.9908 4973

September 2016 1.5709 0.9893 4976

2016

October 2016 1.2709 0.9969 6047

November 2016 1.2708 0.9968 6046

December 2016 1.2709 0.9967 6046

Demand Reduction project

The hypothetical project is tested for its ability to reduce maximum peak demand.



Energy Storage 7000kWh (sized off-peak charging)


The hypothesised 2000kW solar system is to be installed at Murdoch University in Perth at a latitude of

32º and a tilt of 15º. According to the National Renewable Energy Laboratory’s PV Watts Outputs, the

system would produce 3240MWh/annum. Average daily output is estimated to be 8,900kWh per day. The

output profile shows average generation reaches approximately 1200kW in Figure 31. Looking at the

minimum hourly output, generation drops as low as 20% of average.


48

Figure 31: Solar Output

To provide dependability for the solar penetration, energy storage is sized to be 80% of the average daily

solar generation, sized to 7,000kWh. A battery management system is assumed to manage off peak grid

charging Murdoch University’s original power factor, ratio of real power to complex power, was very

reliable; it rarely drops below 0.95. The power factor histogram in Figure 32 shows there are no

measurements where power factor is unacceptable (below 0.8).

Figure 32: Power Factor Distribution Histogram

The model simulates the power factor with real power injection from solar penetration and energy

storage (and real power load reduction from LED upgrades). As the real power demand as seen by the

grid has decreased but inductive power demand has not, this results in a significant power factor

reduction. Figure 33 shows a high occurrence of power factor recordings below 0.8, not acceptable.

kW

200 kW

400 kW

600 kW

800 kW

1000 kW

1200 kW

1400 kW

12:00 AM 3:00 AM 6:00 AM 9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM 12:00 AM

Solar Output

Minimum (kW) Average (kW)


49

Figure 33: Simulated Power Factor Distribution Histogram

To achieve a power factor of 0.95, the power factor correction algorithm determined a maximum reactive

power of 1500kVAr was required to reduce the inductive load as seen by the grid.

Figure 34: Corrected Power Factor Distribution Histogram

Numerous market research reports indicate battery storage costs have dropped exponentially over the

last few years (US Department of Energy, 2017), (Chung et al, 2016), (Khalilpour et al, 2016) and

(University of Minnesota's Energy Transition Lab, 2017). Depending on the provider and storage

technology, the reports estimate that energy storage costs approximately $300/kWh. If the total project

was to be commissioned within the next year, it is estimated that it would require an investment of

$9.5million.


LED Upgrade 4000 Lights


Energy Storage 5000kWh (sized for half a day reserve and off-peak charging)


50


The following, Figure 35, shows the simulated resultant load. Peak demand is still apparent throughout

the peak trading season, however maximum peak demand reduced by approximately 1000kVA.

Figure 35: Simulated Annual Load

The simulated annual apparent load profile shown daily in Figure 36, the dark blue and light blue data

points again represent peak and off-peak hour load. Average weekend load is simulated to range between

2000kVA and 2800kVA, weekend load has decreased by approximately 500kVA. Maximum demand

reaches about 4600kVA.

Figure 36: Simulated Annual Load Profile

Murdoch University’s resultant simulated annual consumption is 17.2 giga-watt-hours (GWh) per annum;

8.2GWh peak hour and 9GWh off peak hour. During the interval data period, the model reports the

simulated maximum demand (maximum apparent power) occurred during peak hour: 4896kVA at 8am,

kVA

1000 kVA

2000 kVA

3000 kVA

4000 kVA

5000 kVA

6000 kVA

Jun 2015 Jul 2015 Aug 2015 Sep 2015 Oct 2015 Nov 2015 Dec 2015 Jan 2016 Feb 2016 Mar 2016 Apr 2016 May 2016

Simulated Annual Load

kVA

1000 kVA

2000 kVA

3000 kVA

4000 kVA

5000 kVA

6000 kVA


Peak and Off Peak Hour Load

Final Off Peak kVA

Final Peak kVA

Average WeekendLoadAverage WeekdayLoad


51

15h March. Maximum off-peak hour demand was 3415kVA, also 15h March at 7.03am. To appreciate the

future IRCR value, it has been assumed that all components of the demand reduction project installation

has been completed by the peak trading season to ensure full impact of demand reduction is measured

during the new peak trading season intervals.

Table 4: Peak Trading Season Real Power Measurements

Peak Trading Interval kW

8/02/2016 16:30 1450.0

8/02/2016 17:00 1608.1

8/02/2016 17:30 1847.9

9/02/2016 16:30 3723.7

9/02/2016 17:00 3734.1

9/02/2016 17:30 3815.7

10/02/2016 16:30 3706.7

10/02/2016 17:00 3724.0

10/02/2016 17:30 3683.1

14/03/2016 16:30 1798.9

14/03/2016 17:00 1919.9

14/03/2016 17:30 2088.6

The median of these measurements in Table 4 is 2886kW. To avoid complication, it is assumed that the

ratios remain the same. Once the capacity year has reset, new IRCRs will change, as shown in Table 5.

Table 5: Simulated Capacity Year and Individual Reserve Capacity Requirements

Capacity Year Capacity Month TDL_Ratio Total_Ratio IRCR

2016

January 2017 1.5639 0.9932 4482

February 2017 1.5675 0.9914 4485

March 2017 1.5724 0.9888 4487

April 2017 1.5719 0.9891 4487

May 2017 1.5741 0.9881 4489

June 2017 1.5752 0.9875 4489

July 2017 1.5668 0.9915 4483

August 2017 1.5675 0.9908 4482

September 2017 1.5709 0.9893 4485

2017

October 2017 1.2709 0.9969 3656

November 2017 1.2708 0.9968 3656

December 2017 1.2709 0.9967 3656

As the IRCRs are always referring to the previous season’s median measured peak, demand reduction will

not be appreciated by the market operator until the start of the new capacity year in October. Regarding

the network operator, Murdoch University is on the contracted maximum demand tariff; the reduced


52

peak demand will reduce excess network usage charges, but significant demand charge reduction will not

be realised until the contracted maximum demand is adjusted.

A progressive load reduction graph, Figure 37 derived from the model shows each component’s

contribution to reducing the peak. The graph shows the maximum annual values to illustrate the impact

each component has on the maximum demand experienced within a single year.

Figure 37: Simulated Progressive Demand Reduction

The red shows the original load almost reached a maximum of 6000kVA. The simulated LED upgrade

estimated to reduce this to about 5700kVA. With the load reduced by efficient LED fixtures, solar further

reduces maximum demand to 4800kVA. The energy storage, though, could flatten peak demand on most

days; it has been sized only to provide reliability within the worst-case scenario. In this case, it cannot be

assumed that energy storage as sized will significantly reduce peak demand. Energy storage is charged

during off peak times for supplementing peak load. Power factor correction is simulated to further reduce

demand by 180kVA. According to the model, the combined project reduces the maximum demand by

1,430kVA.

Table 6: Maximum Demand Reduction

Component Demand Reduction (kVA)

LED Upgrade 300

Solar Array 850

Energy Storage 100

Power Factor Correction 180

kW

1,000 kW

2,000 kW

3,000 kW

4,000 kW

5,000 kW

6,000 kW

7,000 kW Demand Reduction

Max of kVA Max of New Load W. LED (kVA)

Max of New Load W. Solar (kVA) Max of New Load with Energy Storage (kVA)

Max of Final Load (kVA)


53

Total 1430

Resultant savings are calculated from subtracting the initial cost of supply from the simulated cost of

supply. The resultant savings, are calculated to be approximately $900,000. Demand savings are

estimated to make up 24% of the total savings achieved. Considering the project investment, of $9.5

million, the payback period is 10.5 years.

Table 7: Simulated Total Savings

Demand Saving $220,000

Consumption $410,000

LGC Savings $270,000

Total $900,000

8 Conclusion

The model developed during this industrial project has identified significant potential savings for the

reduction of peak demand. Generally, for a large end user, managing peak demand to achieve

dependable demand charge savings provides a more viable project than solely renewable energy

projects that can only deliver consumption savings. A range of factors have been identified that place

limits on the consistency of realising such demand savings. The impact of particular weather events

during the peak trading interval directly influences dominant cooling loads; these, in turn, impact on

market capacity ratios. Demand savings are also at the mercy of load use behaviour and the nature of

business operation. I suggest that savings should be acknowledged, not from the previous year’s

expense, but from forecasted expenditure.

This model is still in the early stage of development. In terms of validating demand savings, the next step

is to incorporate a mechanism that is able to scale loads according to local temperature events. The

further step would be to adjust the model to operate within other network operator tariffs in other

Australian states.


54

8.1 Future Work

This project has outlined the significance of cooling loads in the South West Interconnected System. It is

apparent that, in warm climates, peak demand is primarily a product of temperature. This feature

underlines the necessity for appropriate insulation and intelligent building design.

With the growing demand for energy storage both for the reliability of renewable energy and the

electric vehicle industry, the economies of scale are driving the reduction in the cost of energy storage.

This will enhance investment in such projects.

Managing peak demand of the load is far simpler, and much more efficient and affordable, than relying

on centralised sources of generation. As more and more distributed generation comes on line,

Australia’s electricity systems will need to adapt. It certainly seems, and the independent review, (Finkel

et al, 2017, 30) predicts that, Australia’s electricity systems will see a rapid uptake of investment into

larger scale renewable energy generators and microgrids.


55

9 References

AEMO. 2016a. AEMO 2016 Annual Report AEMO. AEMO. 2016b. "Wholesale Electricity Market (WEM) market procedures." AEMO.

https://www.aemo.com.au/Electricity/Wholesale-Electricity-Market-WEM/Procedures.

Australian and New Zealand Solar Energy Society (ANZSES). 2006. "Australian Solar Radiation Data Handbook (ASRDH) – Edition 4, Fully Revised." In: Australian and New Zealand Solar Energy Society (ANZSES).

Australian Energy Market Operator, AEMO. 2015. MARKET PROCEDURE: RESERVE CAPACITY SECURITY. Australian Energy Market Operator.

Benetti, Guido, Davide Caprino, Marco L. Della Vedova, and Tullio Facchinetti. 2015. "Electric load management approaches for peak load reduction: A systematic literature review and state of the art." Sustainable cities and society 20:124-141. doi: 10.1016/j.scs.2015.05.002.

Bureau of Meterology. 2017. "Daily Rainfall." Commonwealth of Australia, Last Modified 17/05/2017 Accessed 17/05/2017. http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=136&p_display_type=dailyDataFile&p_startYear=&p_c=&p_stn_num=009225.

Dirks, James A., Willy J. Gorrissen, John H. Hathaway, Daniel C. Skorski, Michael J. Scott, Trenton C. Pulsipher, Maoyi Huang, Ying Liu, and Jennie S. Rice. 2015. "Impacts of climate change on energy consumption and peak demand in buildings: A detailed regional approach." Energy 79:20-32. doi: http://doi.org/10.1016/j.energy.2014.08.081.

Finkel, Alan, Karen Moses, Chloe Munro, Terry Effeney, and Mary O’Kane. 2017. Blueprint for the future: Independent review into the future security of the National Electricity Market. Deparetment of Environment and Energy.

Hong, Tianzhen, Wen-Kuei Chang, and Hung-Wen Lin. 2013. "A fresh look at weather impact on peak electricity demand and energy use of buildings using 30-year actual weather data." Applied Energy 111:333-350. doi: http://doi.org/10.1016/j.apenergy.2013.05.019.

IMO. 2014. "SWIS Electricity Demand Outlook."3. Mirlatifi, A. M., F. Egelioglu, and U. Atikol. 2015. "An econometric model for annual peak

demand for small utilities." Energy 89:35-44. doi: http://doi.org/10.1016/j.energy.2015.06.119.

Narukawa, Yukio , Masatsugu Ichikawa, Daisuke Sanga, Masahiko Sano, and Takashi Mukai. 2010. "White light emitting diodes with super-high luminous efficacy." Journal of Physics D: Applied Physics 43 (35).

Nelson, Tim, and Fiona Orton. 2016. "Australia's National Electricity Market: Optimising Policy to Facilitate Demand‐Side Response." Australian Economic Review 49 (2):146-168. doi: 10.1111/1467-8462.12151.

Piccinini, Antony. 2015. "ROOFTOP SOLAR PV – residential and commercial." Energy News 33 (1).

Sadineni, Suresh B., and Robert F. Boehm. 2011. "Measurements and simulations for peak electrical load reduction in cooling dominated climate." Energy (Oxford) 37 (1):689-697. doi: 10.1016/j.energy.2011.10.026.

Sehar, Fakeha, Manisa Pipattanasomporn, and Saifur Rahman. 2016. "An energy management model to study energy and peak power savings from PV and storage in demand responsive buildings." Applied Energy 173:406-417. doi: http://doi.org/10.1016/j.apenergy.2016.04.039.

Suganthi, L., and Anand A. Samuel. 2012. "Energy models for demand forecasting—A review." Renewable and Sustainable Energy Reviews 16 (2):1223-1240. doi: http://doi.org/10.1016/j.rser.2011.08.014.

https://www.aemo.com.au/Electricity/Wholesale-Electricity-Market-WEM/Procedures

https://www.aemo.com.au/Electricity/Wholesale-Electricity-Market-WEM/Procedures

http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=136&p_display_type=dailyDataFile&p_startYear=&p_c=&p_stn_num=009225

http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=136&p_display_type=dailyDataFile&p_startYear=&p_c=&p_stn_num=009225

http://doi.org/10.1016/j.energy.2014.08.081

http://doi.org/10.1016/j.apenergy.2013.05.019

http://doi.org/10.1016/j.energy.2015.06.119

http://doi.org/10.1016/j.apenergy.2016.04.039

http://doi.org/10.1016/j.rser.2011.08.014


56

Verdia. 2016. "About Verdia." Verdia. http://www.verdia.com.au/. Western Power. 2011. Technical Rules. Economic Regulation Authority. Western Power. 2016a. 2016/17 Price List edited by ELECTRICITY NETWORKS CORPORATION. Western Power. 2016b. 2016/17 Price List Information edited by ELECTRICITY NETWORKS

CORPORATION. Western Power. 2017. "What we do." https://westernpower.com.au/about/what-we-do/. Yun, Geun Young, Hyoin Kim, and Jeong Tai Kim. 2012. "Effects of occupancy and lighting use

patterns on lighting energy consumption." Energy and Buildings 46 (Supplement C):152-158. doi: https://doi.org/10.1016/j.enbuild.2011.10.034.

http://www.verdia.com.au/

https://westernpower.com.au/about/what-we-do/

https://doi.org/10.1016/j.enbuild.2011.10.034

Documents

Industrial Project with Verdia Pty Ltd School of ......iv Acknowledgements I wish to acknowledge my supervisors at Murdoch university, Dr Martina Calais; Senior Lecturer Engineering,