Time of use pricing strategy for Indian microgrids subordinate to grids with unreliable coverage

(1)

IN

DEGREE PROJECT ENERGY AND ENVIRONMENT,

SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2018,

Time of use pricing strategy for Indian microgrids subordinate to grids with unreliable coverage

NIKLAS BRÄNNLUND

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF INDUSTRIAL ENGINEERING AND MANAGEMENT

(2)

TRITA -ITM-EX 2018:659

www.kth.se

Supervisor & examiner:

Hatef Madani Larijani

(3)

-1-

Acknowledgement

The author of this thesis would like to thank everyone at Fortum Sverige AB for their support.

Especially Catarina Naucler, Tobias Goodden for all your help and guidance. I would also like to give a special thanks to supervisor and examiner Hatef Madani Larijani.

(4)

-2-

Abstract

In the last decade microgrids has become a usual way of providing electricity resilience in rural India.

With the electrification of rural villages happening rapidly the national grid fails to provide reliable electricity distribution, and microgrids has proven a good way of improving upon the lacking electricity quality. Although many cases exist where a microgrid and the national grid serves the same village, in very few of these cases the microgrid and the national grid interconnects exporting and importing electricity from each other, but instead operates in parallel, rendering the microgrid in to a backup system. This lack of interconnection is partly due to the lack of experience and increased complexity of operation. For many microgrid systems in India the preferred electricity pricing strategy has been to utilize a fixed tariff with one price per kwh regardless of the current distributed electricity mix or the time of day. But recent introduction of smarter electricity meters creates the possibility to implement a more complex pricing strategy such as time of use tariffs. This thesis looks at how microgrids subordinated to an unreliable grid can use electricity meters to create a dynamic time of use tariff increasing operational freedom. To assess a time of use tariff, microgrid setups (generic to microgrids in rural India) are generated for different cases of national grid unreliability, then using optimization software a time of use tariff is generated tailored to that specific operational environment, these tariffs are discussed and compared to the traditional use of fixed tariffs. The simulations show how that a time of use tariff of four price levels can be used to better assure high return for consumption during hours of low production while stimulating consumption during hours of high power production. The tariff suggested in this thesis shows tendencies to increase the risk for the stakeholders while also creating a market structure better suited when expecting growth of electricity consumption.

(5)

-3-

Tidsvarierande prissättningsstrategi

för Indiska mikronät i områden med

opålitlig elförsörjning

(6)

-4-

Sammanfattning

De senaste tio åren har mikronät kommit att bli ett vanligt sätt att betrygga elförsörjning på den indiska landsbygden. På grund av det nationella elnätets mycket snabba expansion har elförsörjning för ett stort antal avlägsna byar inte kunnat prioriteras, mikronät har istället blivit en lösning för att lokalt kunnat tillgodose elbehovet. Även om ett mikronät och det nationella elnätet tillsammans tillgodoser elbehovet för en by finns det få fall där dessa är ihopkopplade och drar nytta av varandras produktion. För dessa mikronät saknas möjligheten av att exportera och importera elektricitet och istället blir mikronätet indirekt ett backuppsystem som kan användas då det nationella nätets elförsörjning ligger nere. Denna avsaknad av koppling mellan mikronät och nationellt elnät beror i största utsträckning på en låg erfarenhet och den ökade driftkomplexitet detta skulle innebära. För många indiska mikronät prissätts elen baserat på ett fast pris per kwh oavsett driftstatus, men med hjälp av smarta elmätare skulle mer komplexa prissättningsalgoritmer kunna implementeras. Denna rapport undersöker hur ett indiskt mikronät kopplat till ett överliggande nät med otillförlitlig drift med hjälp av mätdata kan skapa en dynamisk prissättningsstrategi baserat på rådande driftförhållanden. För att uppskatta denna prissättningsstrategi genereras fallstudier om typiska indiska mikronätsinstallationer som sedan optimeras med hjälp av simulering. Mikronätsdrift testas under olika nivåer av otillförlitlighet i det nationella elnätet och en möjlig prissättningsstrategi tas fram för varje alternativ och diskuteras mot varandra samt jämförs med den traditionella användningen av ett fast elpris. Simuleringarna visar på att en prissättningsstrategi där fyra prisnivåer fått associeras till driftkostnadsscenarion kan användas för att säkerhetsställa en högre inkomst under situationer med högdriftkostnad och samtidigt stimulera ökad elanvändning under timmar med hög produktion. Den undersökta prissättningsalgoritmen visar dock på att inverteringsrisken kan öka för mikronätsägaren men samtidigt kunna stimulera en marknadsstruktur mer lämpat för tillväxt av den lokala elkonsumtionen.

(7)

-5-

1 Introduction

India’s energy demand is rising, and with a large part of the population living in remote rural areas, electrification has been a challenge. Many rural villages are still without connection to the national grid, and for those connected, the power supply can be very unreliable and intermittent, with daily interruptions and long periods of down time. The government of India is aiming to deliver electrical power to over 18,000 un-electrified villages during 2018. Some can be accessed by the national grid using extensions, but for about 20% of the them, off grid power will be necessary (Suryad et al. 2017).

In recent years India has seen a large development and deployment of microgrids, independent and isolated power systems that uses small scale power generation to provide electricity quality and resilience to areas where the national grid has been insufficient (Suryad et al. 2017). In village where both a microgrid and the national grid serves electricity demand, there is the possibility of interconnecting the two, providing more stable electricity delivery. However, historically microgrids and the national grid has been operating in parallel, with the microgrid functioning as an independent backup system (OKAPI, 2017).

In many cases an interconnection between the two parties would benefit both (OKAPI, 2017). Allowing electricity export from microgrids, the national grid would be able to increase its electricity recourses.

And by allowing import, the microgrids would be able to lower their need for electricity generation (or storage) capacity. In a plausible future scenario, when the national grid extends to a village where a microgrid is operating, the national grid could simply connect to the existing microgrid, taking advantage of its existing infrastructure. The microgrid would act as the distributor to the village, importing and exporting electricity from and to the national grid.

The reason microgrids in rural India can exist is however due to the unreliability of the national grid.

Microgrids offer reliable electricity with high quality and can thus compete with the heavily subsidised price tag of the national grid (OKAPI, 2017). Villagers choose to pay more if they can access electricity whenever they need it. Because of this, the main goal in microgrid operation is to ensure reliability in distribution. Currently the main pricing strategy used by microgrids in rural India is fixed tariffs with one price rate regardless of electricity mix, time of day or amount of usage (Suryad et al. 2017). Smart electricity meters and digitalisation of the microgrid operation opens for more complex pricing mechanisms that may favour growing electricity dependence and (with increasing energy resources) ultimately may favour social development (IEA, 2015).

1.1 Study

This thesis investigates a pricing strategy for a time of use (TOU) tariff for microgrid operation typical to rural India (solar power for electricity generation and batteries for electricity storage). The thesis aims to create a simulation tool capable of answering the following research questions:

How can price rates for a time of use tariff be generated?

How does a time of use tariff affect investment value?

How does the national grid intermittency effect microgrid operations?

When is a time of use tariff applicable and favourable over a fixed tariff?

To investigate this, a simulator for microgrid operation is created to take national grid intermittency as an input. This as the reliability of the national grid should heavily affect microgrid operation. The simulator is created using the calculation software Matlab.

Four different reliability scenarios are generated ranging from coverage with relatively low amount of down time to microgrids operating completely off grid. The average national grid behaviour for rural villages in India is used as a benchmark (at 24% down time). Using this benchmark scenario cases of

(11)

-9-

higher unreliability can be generated, this is used to test microgrid operation in down times of 50%, 80%

and 100%. For every case, the simulation takes a specific installed capacity of solar photovoltaic cells (PV) and generated its corresponding storage capacity (seen to least amount needed to assure reliable electricity). Using an optimization algorithm, a proposed time of use tariff is generated for the setup and the investment value of the microgrid seen to a 10-year operational period is calculated, comparing the new time of use tariff to the fixed tariff.

1.2 Research scope

When discussing possible pricing mechanisms for an interconnection between the national grid and a microgrid operating in a village, the existence of a marketplace operating between microgrid, village and national grid is implied. Mengelkamp et al. (2018) describes in their report on microgrid marketplaces a way of categorizing the fundamental areas of interests that affect how the microgrid marketplace should function in to seven categories. These areas range from analysing the physical and political environment of the microgrid, to information gathering and how the market will function in terms of business logic.

These categories are used as a framework to explicitly define on what assumptions and delimitations this thesis will be conducted.

1.2.1 Policies and regulations

National and global policies and regulation ultimately determine how the grid will be setup and how the market may function. There might be incentives, subsides or taxes that would favour the use of specific generation sources or limit the use of others (Mengelkamp, et al., 2018).c This will be covered in the background section and lay as a foundation for any assumptions made regarding prices and willingness to pay when designing the case study.

1.2.2 Superordinate grid connections

This includes national grids and external microgrids. These can provide balancing services with alternative sources of energy as well as arbitrage opportunities. This category evaluates these connections and how they affect the microgrid in terms of energy security, balancing and market participation. A microgrid running in island mode for long periods of time might have an increased need for energy storage and demand side management (Mengelkamp, et al., 2018). This part is one of the main focuses of the thesis, how the relation between the national grid and the microgrid governs microgrid operation is investigated through computer simulation.

1.2.3 Microgrid setup and objective

Factors related to the pursued objectives of the microgrid is evaluated. This includes defining the stakeholders involved in the microgrid operation. These stakeholders might participate in the energy market, and if that is the case, how will they participate. Also different stakeholders have different and sometimes conflicting goals (e.g. energy security, increased renewable energy in the mix or maximize profit) that would affect the microgrid operation (Mengelkamp, et al., 2018).

Microgrids studied in this thesis are assumed to provide the service of electricity resilience. In the type of village investigated, electricity is provided from a national grid but with frequent outages. To be able to sell electricity from the microgrid, they must be able to assure continued electricity delivery to a village without interruptions and large amount of down time. For the microgrid owners the objective is firstly to provide resilient electricity, and secondly to have a positive return on the microgrid investment. The return has to be calculated on a relatively short time span as large investments in electrification of India is made and might threaten microgrid operations in the long term.

(12)

-10- 1.2.4 Information system

The information systems used to gather data, providing market platform and connecting market participants is evaluated. How energy meters are used, what information that is needed and with what time resolution. How the information system is designed might affect operation, algorithms for pricing and market dynamics. All market participants should have equal opportunities for trading without overly complicating the platform. Analysing the energy system as a separate category gives the opportunity to discuss the trade off between data resolution and increased market complexity (Mengelkamp et al., 2018).

For the simulated microgrids, energy meters are assumed to exists such that they are able to communicate information with hourly precision and that information of current generation and demand is obtainable.

It assumes that a management system exists that can control the electricity flow following simple operational logic without the need of higher computational power than adding or subtracting power flows. Hourly time steps in the simulations is deemed to be high enough resolution in order to generate a good approximation of yearly behaviour. On the household level the market mechanism cannot be too complicated or assume ownership of any technology such as computers or cell phones. Information about price rates and current price has to be easily communicated, preferably on the electricity meter itself. The investigated time of use tariff is designed to use a price step function of relative few levels, thus the current price rate could be communicated using light diodes on the electricity meter or with a small display.

1.2.5 Market mechanism

Market structures determines how the market allocates pricing mechanisms, how the bidding process works and how buyers and sellers interact. This includes possible markets allocation and payment rules, design of a bidding language and how buy and sell orders should be coordinated. Also the use of multiple market time horizons e.g. intraday and day ahead markets can be used to increase efficiency but with the drawback of increasing complexity (Mengelkamp et al., 2018). This is outside the scope of this study as the interactions between buyers and sellers are mainly the microgrid importing and exporting electricity and villagers consuming electricity. For the national grid interactions, the microgrid is assumed to be able to import and export freely without any future planning, paying a fixed tariff depending on import and export rate. Villagers are assuming to be able to freely consume electricity, receiving a monthly bill based on their consumption patterns.

1.2.6 Pricing mechanism

The pricing mechanism is implemented via the market mechanism. Depending on the market and grid, schemes such as tariffs, subsides or time of use prices will affect how the market functions. The price should be based on the fees and taxes present as well as the marginal cost of generation, but mechanisms such as price signals to indicate scarcity/surplus could also be used (Mengelkamp et al., 2018). The main goal of this thesis is to evaluate the use of fixed tariffs compared to a time of use tariff. The time of use tariff investigated uses desired consumption patterns to generate price levels depending on electricity mixes and the current cost of generation.

1.2.7 Energy management trading system

An energy management trading system would be acting on the market carrying out a specific bidding strategy in order to secure energy for a market participant. The bidding strategy is dependent on all of the above stated areas and ultimately can be seen as the driving force of the system (Mengelkamp et al., 2018). This is outside of the scope of this thesis.

(13)

-11-

2 Background

2.1 Microgrid concept

In the recent years microgrids has become increasingly abundant. This is in large due to the electrification of remote communities as well as the need for increased resilience and backup in some areas connected to the main grid (IEA, 2017). A microgrid is defined by a local electric grid that integrates local generation with the local demand side, this means that electricity production from small scale generators deliver electricity to consumers in their near vicinity. Generators includes both prosumers and home systems as well as centralized small scale production facilities operated to cover the electricity demand of the microgrid. Usually the term distributed generation or DG is used to describe this type of electricity generation.

The proximity between supplier and user are very beneficial as it decreases the need of large infrastructure systems and thus lowers electricity losses. Thus microgrids usually requires less transmission and distribution reducing the need of these kind of facilities as well as reducing related energy losses and costs (Schwaegerl and Tao, 2013). The possibility to disconnect from a main grid also limits propagation of fault, and simplifies management and service (Parhizi et al., 2015). It also can work to increase system resilience, reliability and energy quality (Schwaegerl and Tao, 2013). This is partly due to their ability to operate in a so called island mode, shutting the connection to a superordinate grid to independently rely on its local generation, with the possibility to connect back up and disconnect at will (IEA, 2015). This trade also makes microgrids an essential building block for smart grids with large opportunities for optimization and implementations of algorithms for microgrid operation (Schwaegerl and Tao, 2013).

However, to be able to function as an island for longer periods of time (or operate completely independent without a superordinate grid connection), the microgrid has to have a large margin of production or high energy storage capacity.

Microgrids enables a more effective hosting of distributed generation compared to larger grid (IEA, 2015) and has shown to stimulate growth of micro-turbines, photovoltaics, fuel cells and small scale wind turbines (Schwaegerl and Tao, 2013). This have led to microgrids generally having a high percentage of generators using renewable energy sources, electricity from which can help to lower energy prices as well as generating less emissions (Schwaegerl and Tao, 2013). However, the intermittent nature of renewable energy sources does contribute to increased forecast errors and requires balancing. This can be done using energy storing units such as batteries, fuel cells or flywheels. In case studies the combined use of batteries and photovoltaics has showed to be both beneficial for reducing electricity cost and for reducing emissions (Parhizi et al., 2015). Electricity storage units can also be used for arbitrage, taking advantage of volatility in the electricity prices. With smart meters, consumers can install their own generators or storing devices, contribute to the capacity of the microgrid and becoming prosumers (Parhizi et al., 2015).

The initial investment of a microgrid has shown to be relatively high, but with great return if operated properly (Parhizi et al., 2015). The increased reliability of electricity that microgrids provide, has also shown to benefit economic and social development (IEA, 2015). Due to the nature of microgrids, renewable energy can be produced far from the main grid and still return high energy quality (IEA, 2015).

(14)

-12-

2.2 The microgrid sector in India

India aims to provide electricity to all villages by May 2018 and to all people by Mars 2019. Microgrids are an essential building block in order to reach this. As they often are operated by private energy service companies where grid connection is unreliable or missing, they can provide rapid electrification of remote areas regardless of the main grid (OKAPI, 2017).

The political landscape revolving microgrids has evolved a lot during the last two decades. In 2003 electricity generation and distribution was de-licenced in rural areas which made it possible for privately owned microgrid companies to form. Later in 2005 the government decided to prioritize main grid expansion and development only where economically feasible, and encouraged microgrid development as an alternative route to electrification in rural and remote areas. Due to incentives in the form of government funding, microgrids became a great business opportunity. The remote and off grid locations that qualifies for government funding was later redefined to areas with less than 6 hours of daily electricity supply (OKAPI, 2017).

In 2016 the National Tariff Policy was formulated discussing the possible problems that can arise when reliable grid coverage reaches areas with an operational microgrid in place. The policy suggests that power purchase from the microgrid to the main grid should be compulsory providing a mutually beneficial relationship. Regardless of this policy, there are still many microgrids that has not engaged in any cooperation with the main grid, and many places where the grid and microgrid are running in parallel.

Customers of these villages has showed a willingness to pay higher tariffs for higher reliability of electricity from the microgrid than to be dependent on the main grid (OKAPI, 2017).

A study on electricity availability (presented in the 2017 Electricity Support Monitoring Initiative) showed that many remote and rural locations suffer from intermittent supply of electricity from the main grid.

During a monitoring period of 6 hours few of the monitored locations showed constant electricity supply, and during a monitoring period of two weeks many locations even showed days without electricity supply.

For the periods when electricity was delivered, the voltage levels were often below the intended. (OKAPI, 2017).

This unreliability of electricity supply has stimulated the increase of microgrids run by private energy service companies providing services as distribution, generation, metering and billing. This business can be costly when operated in remote locations and high customer tariffs usually has to be deployed, while the main grid tariff is highly subsidized and thus very hard for microgrid owners to compete with from a price perspective. An interconnection between micro grid and main grid would have the potential to benefit both parties, lowering electricity costs while increasing resilience and service (OKAPI, 2017).

In their report regarding integration of microgrids in India, Okapi (2017) lists cooperation between privately operated microgrids and the government as one of the most important steps to reach India’s electrification goals. Despite all this, there still has not been many operational implementations of interconnections between microgrid and main grid. After interviewing stakeholders Okapi (2017) concludes that this probably is due to limited understanding and experience.

Microgrids with superordinate grid connection can as an alternative power generation provide a competitor to the otherwise commonly vertically integrated business of electricity generation and distribution. By participating on electricity markets, the microgrid can induce a more dynamic market with electricity prices being set by open competition. With approximate and on site production the microgrid can sell electricity at low distribution and transmission costs (Chowdhury and Crossley, 2009).

Also microgrids tend to have a higher amount of renewable sources in the energy mix which works to keep marginal cost low and thus highly competitive to superordinate grids. However, this also constitutes the need for external grid connections and/or electricity storage to balance the intermittent production (Schwaegerl, C. and Tao, L., 2013).

(15)

-13-

2.3 Role and implications of electrical energy storage

Electrical energy storage is essential in order for a microgrid to effectively utilize its high percentile of intermittent generation while operating islanded from a superordinate grid (Schwaegerl, and Tao, 2013).

Batteries provides a great balancing tool as they booth can serve as a load or a generator, charging when load lower than production and discharging when higher. Due to power losses in every delivery step, batteries should typically be operated under these charging conditions in order to maximize power delivery (Wenzhong, 2015).

Batteries also provides the capability of peak shifting in conditions where forecasting is possible. If the usual production pattern fails to follow load peaks, electricity storage can be used to shift these load peaks to occur when electricity production is higher, e.g. increased load during the afternoon could be met with less capacity increase, allowing demand to better follow production. Peak shifting is also possible on longer time spans when whether conditions might imply lower future production, while load levelling typically provides a smoother short term production curve (Wenzhong, 2015).

2.4 Pricing mechanism

2.4.1 Storage pricing mechanisms

Capturing the value of electric energy storage in a microgrid system is hard due to the amount of services they provide. Comparative studies looking at similar systems with and without storage has shown that electricity prices even might be lower in systems with high amount of storage capacity, primarily due to load levelling decreasing price volatility and price spikes. This implies that the revenue that can be obtained for large systems with energy storage investments is relative low compared to the benefits to the system. Because of this the value of energy storage should not be entirely dependent of the capacity value, but rather the sum of the capacity value and the value of the provided services such as spinning contingency reserves and load levelling (NREL, 2013).

2.4.2 Levelized cost of electricity

Levelized cost of electricity (LCOE) is a commonly used metric for comparing and valuing electricity generation. It combines initial investment costs with running costs like operation, maintenance and fuel.

This provides a good way of comparing generation sources that utilizes different energy sources as their running and initial costs may vary from each other. Calculation of LCOE is done by dividing a generators present value of total expected costs with the present value of total expected production. LCOE can be viewed as the minimum cost that the generated electricity can be sold for in order to break even, and are therefore a good tool to use when researching pricing mechanisms for an energy system. When calculating LCOE for PV cells, a degradation term can also be added to account for less production as the panels age. LCOE for PV cells has shown to decrease as efficiency increase. This is true for PV panels of higher rated efficiency, as well as efficiency altering measurements such as automated axis tracking. This shows that the initial cost of PV panels is offset by the value of production (Lai and McCulloch, 2017).

When evaluating electrical energy storage (EES) it is necessary to look at the levelizied cost of storage (LCOS) instead of the LCOE, this as EES does not produce electricity. The LCOS for EES is however hard to determine properly as services provided by the storing possibility often are of more value that the actual revenue it creates (as previously mentioned). Because of this, the approach of just dividing present value of total costs with present value of total stored power does not paint the entire picture.

Thus methods for calculating LCOS quickly becomes complex. (Lai and McCulloch, 2017).

(16)

-14-

The direct use of produced power is always beneficial due to transmission- and storage related losses, this means that charging will generally only be done using surplus electricity. By isolating this term from the energy that is directly delivered to the load a LCOE term can be derived to provide a more accurate measurement of how costs are affected due to energy storage. The LCOE for EES can then be described as the cost of delivering electricity from the EES to the load as a sum between the LCOE for the surplus electricity (with losses related to the EES contributed for) and the LCOS for the EES as seen in Equation 1, where the LCOS simply is the net present costs divided by the net present power stored (Lai and McCulloch, 2017).

Equation 1: Levelized cost of electricity for electricity storage 𝐿𝐶𝑂𝐸_𝐸𝐸𝑆 = 𝐿𝐶𝑂𝐸_{𝑠𝑢𝑟𝑝𝑙𝑢𝑠} + 𝐿𝐶𝑂𝑆

Minimizing LCOE for a microgrid system implies maximizing its revenue, however for microgrid systems that often operates in islanded mode, minimizing LCOE would also imply less reliability with more power outage (Hittinger et.al, 2015). Because of this trade off, there are incentives to optimize the microgrid system configuration. But as the load seldom follow distinct patters and in many microgrid cases evolves and changes (as more capacity gets installed, or availability and price affects usage), many of these models have a hard time to account for the uncertainties connected to microgrid operation. Using linear programing approaches runs the risk of being too deterministic and becomes hard to extrapolate upon (Hittinger et.al, 2015).

2.5 Simulating PV production and electricity storage

2.5.1 Modelling solar production

Modelling solar PV production is not trivial as it depends on factors of high uncertainty and unpredictability such as weather conditions, cleanness of cells, operational temperature and operational conditions. Thus when modelling PV cells for decision making regarding economy or investments, analysing model uncertainty becomes extra important (Goss et al. 2017).

The electrical yield from a PV module is at the top level determined by data on local solar irradiation.

The main sources for model uncertainty is how this irradiance gets translated to electrical energy output.

Two main factors contributing to this uncertainty is the operational environment and the technical specification of the module. Operational environment describes how the module is installed, slope- and azimuth angle of the module as well as the ambient operating temperature. The technical specification is of how the module operates at standard test conditions and specifies how much electricity that is generated at certain irradiance and ambient temperature. When making assumptions regarding any parameters within these two factors, results should be viewed with caution (Goss et al. 2017).

The effect temperature has on the model performance can usually be assumed to be in the area of around 0.5 % loss per degree Kelvin (for crystalline silicon devices). When calculating the temperature rise inside the module, there are environmental parameters as well as installation parameters that should be taken in to account. Environmental parameters are mainly the solar irradiance, wind speed and ambient temperature. Installation parameters are the inclination of the module as well as if the module is mounted on a plane surface or standing free. To calculate the operating temperature many empirical tables and expressions exist, ranging from energy balances to simpler models using coefficients for typical cases.

When modelling large time steps of one hour or above, simpler models are deemed to be accurate, but for increased resolution more complex models might be needed (Goss et al. 2017).

(17)

-15-

Azimuth and slope angles affects energy yield as well. For climates with large variance in seasonal solar position these affects can be used to tailor energy yield to demand, with steeper slopes generating more energy during winter and vice versa. The change in production (due to slope and azimuth angle) ultimately depends on the location and should be taken in to account when modelling (Goss et al. 2017).

2.5.2 Modelling battery storage

Batteries are one of the few energy storage technologies that are available to respond within seconds, making them a very attractive choice for energy storage within electricity systems of unpredictable production or load. Lithium based batteries has shown high potential as they can provide fast charge rates sustained over large lifetime throughputs. The specific energy per kg is higher in lithium compared to lead acid batteries and the commercial availability of lithium-ion batteries are increasing fast (Julien et al., 2018).

The behaviour of a battery depends a lot of the technology used and the physical properties of the battery itself. When modelling EES in a time step simulation, the general approach is to determine the current that will be either charging or discharging from the batteries. By using the power required from the system (for either a load or a generator) and dividing this with the voltage of the EES the required current can be calculated (Manwell and McGowan, 1993).

Following this equation, there are two major constraints that determine battery dynamics, the maximum current that can be generated by the system for a specific time step and the voltage corresponding to that time step. The actual voltage usually differs from the rated value, as it depends on factors such as charge or temperature, but for normal operation, the actual voltage and nominal voltage can be assumed as equal (Manwell and McGowan, 1993).

2.5.3 Electricity storage, choice of technology

When designing an electricity storage system for a microgrid, the services required from the electricity storage is the determining factor for what type of battery technology that would be most fitting. For microgrids serving entire villages with unreliable and intermittent power generation, the technology likely to be most suitable is either lead acid or lithium ion batteries (IRENA, 2017a).

The main storage technology that are used in microgrid systems in rural India are lead acid batteries (Suryad et al. 2017). While lead acid batteries might be applicable for small scale storage, when the microgrid capacity and load increase the need for faster battery response will most likely lead to increased maintenance and low life cycles (Corcuera, Estornés and Menictas, 2015).

Lithium ion batteries has proven to have a much more advantageous cycle life and lower maintenance cost (Bresser, Paillard and Passerini, 2015). The main drawback of using lithium ion batteries for village size microgrid storage is their relatively high average cost ($500 per kWh of storage) compared to the much lower of lead acid batteries ($150 per kWh of storage) (IRENA, 2017a). Lithium ion has however seen a recent increase in popularity during the last decade, making the battery technology widely commercial available. In the near future this might push the price to levels where lead acid would no longer be able to compete with the technology (Julien et al., 2018).

(18)

-16-

3 Methodology

3.1 Key assumptions

3.1.1 Electricity resilience

The main part of the microgrids operational goal is to provide electricity resilience, this is also assumed to be the only service that is requested from the microgrid. This is because of the very low and heavily subsided electricity prices tided to the Indian national grid, usually the block tariffs that can be obtained for users of the national grid are far lower than the microgrid can afford to compete with. (Suryad et al.

2017). It is assumed that electricity resilience is achieved at 99.5% coverage, meaning that maximum of 43 hours of outage is allowed annually in order to achieve operational goal.

3.1.2 Investment period

As India invests a lot in order to meet 100% of population electricity access, both expansion and resilience of the national grid can be expected to increase fast. Due to this the demand for electricity resilience might drop and thus a microgrid could be seen as a rather risky investment when viewed as a long-term solution. Using a relatively short time frame of 10 years provides a more realistic picture of the investment value and should reduce the risk of the investment.

3.1.3 Electricity production technology

The microgrid model is using PV cells for electricity production and batteries for electricity storage, this follows the most common type of microgrid setup in rural India today (Suryad et al. 2017).

3.1.4 Degradation and losses

System losses during distribution will be assumed to be incorporated in to the demand as they are proportional to each other. Using this reasoning, system losses during export and import are assumed to be included in the corresponding tariff as well. Capacity losses due to degradation of the batteries are assumed to be accounted for by allowing larger depth of discharge, and capacity losses due to aging of the PV cells are assumed to be accounted for by maintenance costs.

3.1.5 Time step resolution

Hourly time steps are used in simulating the electricity flow in the microgrid system and are assumed to be able to capture enough data on the annual behavior of the microgrid to be able to show characteristics and tendencies. It also enables the use of weather data and statistical information on grid reliability without the use of interpolation. Also as most input data is available in hourly time steps, little difference between hourly and lower simulation resolutions is assumed.

3.1.6 Customer information constraint

The customers are assumed to not be interested or able to take part of complicated information regarding electricity price. Consequently, a tariff cannot be entirely dynamic and ever changing. Also information on current price would have to be easily communicated to the customer without relying on any (customer side) third party information technology such as computers or smartphones. The time of use tariffs suggested in this thesis is based on the assumption that price levels can be communicated using electricity meters installed in the household. For example, this could be done with led lights indicating current price level, or a small display.

(19)

-17-

3.2 The Model

A model is created to simulate the electricity flow in a simplified microgrid system connected to a superordinate grid with unreliable and unpredictable coverage. The microgrid setup is illustrated in Figure 1 and is simplified to exclude components such as inverters between battery banks and PV modules (which might operate on direct current) and the national grid and load (which might be operating with alternating current). Consequently, any losses due AC to DC conversion or vice versa is assumed to be accounted for by electricity demand, as these losses affect the whole system. The cost for electrical utilities such as switches inverters and battery management systems are all assumed to be included in the system initial cost, and their maintenance and operational costs are assumed to be included in the systems annual cost.

As illustrated in Figure 1 there are four major components to the system:

Electricity production from PV cells Electricity storage in battery banks Superordinate grid connection Village electricity demand

Figure 1: Microgrid structure and setup, arrows shows power flow directions

3.2.1 Modelling electricity production from PV cells

The PV arrays DC output for one year is generated using the software “System Advisory Model” (SAM) developed by the National Research Laboratory (NREL) of the U.S Department of Energy. The software allows for choice of weather data by location and takes ambient temperature and model temperature in to account when calculating production efficiency. The generated time step series are saved to file and used as input in the electricity flow model. This assumes that the PV production is always desired at max possible effect and no indented production shedding would be needed. Generation losses during PV generation, such as temperature effects, cell reflection and soiling are assumed to follow default values as provided by SAM, details of what inputs that are used can be viewed in appendix A.

(20)

-18-

3.2.2 Modelling electricity storage in batteries

The battery bank is sized using two constraints. Total outage after simulation cannot exceed 0.5% and the batteries has to be able to deliver power enough to alone be able to cover the maximum demand peak. Using the latter constraint and the software SAM, input values for nominal voltage and maximum current can be generated. SAM provides a library of battery types and uses the batteries C-rate to acquire the parallel connection to number of batteries in string ratio needed to achieve desired battery bank power. These are saved to model parameters, as later discussed in the case study.

Figure 2: Battery charging/discharging algorithm

The algorithm for charging and discharging is fairly simple and works by taking the desired power output of the battery (negative if charging is required and positive if discharge is required). And comparing this desired output to the maximum possible current from the battery bank and the current state of charge (SOC) of the battery. The function then returns the actual output (the desired if possible, otherwise as close as possible) and a new SOC. Roundtrip efficiency is affecting the SOC with its square root while charging and discharging respectively. The algorithm is illustrated in Figure 2 with red boxes showing inputs.

3.2.3 Generating electricity demand curve

In order to create a curve following a realistic pattern the total time of daily usage has to be distributed throughout the day following a probability curve of when most people would want to use their electric equipment. In order to create a flexible model that can be applied to different case studies electricity usage is divided in to four major categories: household lighting, household others, commercial lighting and commercial others. Due to the scarceness of statistics regarding average electricity use in rural India, general assumptions regarding the behaviour of these categories is made and presented below.

(21)

-19-

Figure 3: Curve showing probability of use for household electrical utilities

Figure 3 shows the probability of usage profile for households. Lights are assumed to be used sparsely during late night and midday with usage peaking during early morning and after sundown. The probability of other appliances (such as cooking devices, cooling/heating loads, and charging of electronic appliances) are assumed to follow a smoother curve that the lighting due to less probability but with most of the activity during afternoon and a small peak during morning hours. The magnitude of these peaks are set to arbitrary values where electrical use during hours of low probability is less than a third as likely as usage during hours of high probability.

Figure 4: Curve showing Probability of use for electrical utilities in the commercial segments

(22)

-20-

Figure 4 shows the probability of usage for commercial segments. Lighting follows the house hold lighting, but stays on during day, this is assumed due to stores and similar being open. Other appliances such as cooling/heating devices and electronic devices are assumed to be heavily used during daytime with less usage during mornings, afternoon and night time.

The demand curve is generated for an entire year with a 1-hour time step using these curves together with case study data. The case study provides max and min values for installed load (Lmax and Lmin) and max min values for total hours of daily use (hmax and hmin) for the corresponding user type. A random value (L and h) is then generated between these max and min values and divided with the probability of use (p) for the current time step.

Equation 2: Demand calculation for specific user 𝐷𝑒𝑚𝑎𝑛𝑑(𝑡) = 𝐿(𝐿_𝑚𝑎𝑥, 𝐿_𝑚𝑖𝑛) ∙ℎ(ℎ_𝑚𝑎𝑥, ℎ_𝑚𝑖𝑛)

𝑃(𝑡)

Figure 5: Four levels of electricity demand, duration and magnitude (compared to minimal value). Generated curve is aggregated using the case study presented in section 3.6. The All India curve is generated with data from POSOCO

(2016).

(23)

-21-

To test these assumed demand profiles, the case study of this thesis is used in order to generate and aggregate a total electricity demand profile which can then be compared to a electricity demand profile for all of India. The data used for India is presented by POSOCO (2016) as a typical India load curve.

The load curve characteristics can be divided in to four levels of energy use. The morning peak, evening peak, midday and night. In Figure 5 the duration and average magnitude of these four levels are presented as a function of time of day and percentage of minimal demand.

The shape of the two curves are rather similar but differs greatly in magnitude. The All India curve only increases electricity demand by 20% during its highest peak compared to its lowest, while the generated curve increases demand by 400% compared to its lowest. However, due to the vast difference in electricity customers between the two curves, this difference in magnitude is assumed as plausible. The probability curves described in Figure 3 and Figure 4 should however be limited to small villages and omitted in studies where data is available.

3.2.4 Generating national grid coverage

The village load and the battery bank is assumed to be small enough so that the national grid could always deliver more than enough power, when up and running. With this assumption, the national grid could be modeled as a time series of ones and zeros representing on and off. If on - export and import of electricity is possible, and if off - the microgrid runs in island mode. The national grid is also assumed to be unaffected by the operation of the microgrid and the village load, such that no amount of import or export of electricity would trigger an outage.

Two inputs are used when generating this on/off time series. Number of interruptions every year and total percent of yearly downtime. By using two inputs, the duration of an outages can be prolonged instead of downtime hours being evenly distributed throughout the year. This would thus provide a more realistic picture and be able to follow existing statistics and data for grid unreliability.

The equations used to generate outage time series is showed in Equation 3 and the generated transition matrix is showed in Table 1.

The number of interruption yearly (poff) is assumed to be the same as the number of startup (pon), e.g. a year of 10 interruptions also should have 10 startups.

Equation 3: Generating transition matrix for national grid coverage 𝑝_𝑜𝑓𝑓 = 𝑝_𝑜𝑛 = 𝑝 , 𝑐₁₂ = ^𝑝

𝑡−𝑑 , 𝑐₂₁ = ^𝑝

𝑑

During hours of national grid coverage, the chance of the next hour initiating an outage (c12) is the amount of interruptions divided by the total uptime: the difference between the amount of total time (t) and the amount of downtime (d). During hours of downtime the chance of a startup is the amount of startups divided by the total down time.

Table 1: Transition matrix for national grid coverage, where state 2 represents outage Chance to switch state: To: 1 To: 2

From: 1 1 - C12 C12

From: 2 C21 1 – C21

(24)

-22-

This model works for all cases where the numerator is lower than the denominator, for cases of more unreliability or more interruptions the case should be calculated as inverted finishing by modifying the generated time series, changing ones to zeroes and zeroes to ones.

3.2.5 Simulating microgrid operations

The entire simulation model can be divided in to inputs (and input functions), the microgrid operation algorithm, and the outputs and output functions. These three elements can be seen in Figure 6, on the right side the input and output elements are represented with red for input and green for output. On the left side the operation algorithm is illustrated using a flowchart.

Dotted lines are used to group content for readability purposes, the two dotted areas represent data in the “case study” input as well as the content of the “electricity flow” output.

The input elements have been mostly covered in the section above and is the cornerstone for any case studies simulated. Any case study simulated takes time series of PV production, village demand and grid outage as well as the maximum allowed system outage for the simulation time. Input rhombs on the far right shows what case study data is used for generating the corresponding time series.

Figure 6: flowchart representation of the microgrid model, input showed in red with generated time series in yellow, outputs showed in green. Dotted lines represents content of corresponding input/output.

(25)

-23-

The algorithm works by covering the village load in the following merit order:

1. PV production only 2. PV and national grid 3. National grid only 4. Battery bank discharge

When demand is lower than the PV production, the battery bank gets charged, and when there is surplus electricity (and the national grid is up and running) it gets exported.

When demand is higher than PV production, electricity is imported from the national grid and the battery banks gets charged (if the case study allows for national grid charging). When the national grid is out the battery bank is discharged.

If (during a time step), the demand is higher than the combined production, import and discharge, this time step is logged as an outage. The algorithm repeats until the maximum allowed hours of outage is higher than the logged amount of outages, increasing the battery bank capacity every loop.

3.2.6 Output and post processing

Outputs from the microgrid operation algorithm (illustrated as green fields in Figure 6) consists of time series for the state of charge and power flow of the battery, time series of exported and imported electricity and the required battery capacity. These outputs are used in the techno economic part of the model to process parameters related to the investment value and comparing the use of a fixed tariff to a time of use tariff. The inputs to these processes are illustrated in the bottom left area of Figure 6.

Battery life is an important metric to take in to consideration due to the large investment in battery capacity is a major contributor to the profit of the investment. The batteries nominal calendar life has to be compared to their cycle life. If the number of battery cycles are above the battery cycle life during the expected calendar life, a new adjusted battery life is returned.

3.3 Techno economic model

The techno-economic model is used to calculate investment value of a microgrid case study. The model uses outputs from the simulation model (as discussed in the previous section) to calculate pricing strategies based on (by the case study) predefined economic parameters. The main purpose of the model is to investigate how a time of use pricing strategy compares to the use of a fixed tariff. This is done by first assessing power delivery scenarios where different customer behavior is desired. For example, when there is a lot of excess production of electricity from the PV cells the desired customer behavior is for them to consume more. When there is a national grid outage there is a desire for the costumers to consume less, lessen the use of the battery bank and reserving electricity.

Using an optimization algorithm and constraints (further discussed below) these prices levels are assigned a suggested value based on the cost of the corresponding power delivery scenario. This creates a time of use tariff that could easily be communicated to a user depending on the number of scenarios used.

The final step of the model is to calculate investment value of the case study. The internal rate of return will be the main metric for assessing investment value and deciding on what microgrid setup that are preferable given a national grid downtime scenario.

(26)

-24- 3.3.1 Defining price levels

Using the simulation model and looking at the dispatch algorithm, three main delivery scenarios can be observed. When electricity is produced and in abundance compared to the village demand, when import of electricity from the national grid is needed to cover for demand, and finally when battery power is needed to prevent shortage. The desired consumption pattern from the microgrids point of view would follow these three scenarios with most of the village electricity demand being satisfied by direct PV power without storage. If this was the case, there would be less need for importing electricity from the national grid and also the battery bank could be made much smaller.

Figure 7: Price level scenarios from time of cheapest electricity to most expensive.

These scenarios are assumed as the points where electricity price would have to differ the most when using a time of use tariff. A fourth scenario does however exist, when the battery bank is charging during PV production hours, there could be an incentive to raise the electricity price slightly to prevent the village and battery bank from competing for the produced power and to ensure that the battery bank charges as fast as possible.

(27)

-25-

These four scenarios are used as the bases for the time of use tariff. Thus generating a tariff model with four possible price levels. Figure 7 illustrates how these price levels relate to the microgrid operation algorithm.

The following definition is used for the price levels:

1. PV production with surplus electricity 2. PV production with no surplus electricity 3. Supplement electricity from the national grid 4. Supplement electricity from battery storage

For every time step in the microgrid simulation the related price level is denoted, following the above definition. For every time step a price level vector can be created (such that for a price level of one the first element is set to one followed by zeroes, for a price level of two, the second element of the vector is set to one with the rest being left as zeroes, and so on). These vectors would contain as many elements as price levels and for every vector all elements would be zeros except for the one element corresponding to the active price level, this element would be a one. These vectors are used to create a price level matrix with rows corresponding to simulation time step and columns corresponding to the price level vector. A small example for a simulation of three time steps and with four price levels are shown in Equation 4.

Equation 4: Price level matrix for 3 simulation steps and tree price levels, 1 denotes active price level. For time step one the price level is one, for step two the level is tree and for step three the level is two.

[1 0 0 0

0 0 1 0

0 1 0 0] = 𝑃_𝐿𝑉𝐿

3.4 Calculating time of use tariff

For each price level there has to be a price tag. These time of use rates are calculated using Matlab and the built in optimization function fmincon, a multivariable function solver with the possibility of assigning constraint to the optimization problem (mathworks.com, 2018).

3.4.1 Demand response related to time of use tariff

This curve can now be used for the optimization setup and revenue calculations. Below follows how the optimization problem is defined and how the constraints are formulated.

3.4.2 Price rate goal

The goal of the price rate is simply to maximize revenue of the electricity sold, increasing the electricity cost for each price level as far as the constraints allow. This formulated simply with the use of Equation 5.

Equation 5: Gross microgrid revenue, for a simulation n time steps long max𝑝⃑ 𝑓(𝑝⃑) = [𝑑₁, 𝑑₂, … , 𝑑_𝑛] [𝑝_𝑙𝑣𝑙₁

⋮

𝑝_𝑙𝑣𝑙_𝑛] 𝑝⃑

(28)

-26-

If di is the demand for the simulation step i, and if 𝑝_𝑙𝑣𝑙_𝑖 is the price level vector corresponding to the active price level for time step i (Equation 4). Equation 5 would return the gross revenue for price rate vector 𝑝⃑. This equation is used in the equation solver to solve for the 𝑝⃑ that would generate the max possible revenue within given constraints as described below.

3.4.3 Constraint 1, price rates has to incline

The first constraint is simply that the price levels has to be inclining, this to achieve desired consumption incentives. If the village is supposed to consume most at price level 1, it would have to be priced lower than 2, 3, and 4. Thus the first constraint is simply: 𝑝₁ ≤ 𝑝₂ ≤ 𝑝₃ ≤ 𝑝₄ for a price vector of 𝑝⃑.

3.4.4 Constraint 2, monthly house hold bill cannot increase

For the second constraint, the monthly average household bill cannot be higher than it would have been with a fixed tariff. With a time of use tariff, households have the possibility to shift their consumption along their willingness to pay.

Figure 8: Example of household consumption adjusted to follow PV production

When incorporating a time of use tariff, there would not be entirely unrealistic to observe a change in consumption pattern as the village moves some of their consumption from hours of high price rates to hours of low price rates. To account for this the model conducts all calculations related to the time of use tariff, with a demand curve that is adjusted a towards the PV production. This shift in consumption pattern cannot be allowed to be too large as much of the village electricity demand would be tied to patterns such as work hours and daylight. The new curve is generated using the assumption that 25% of the electricity demand curve could be moved to follow the PV production curve, an example of which can be viewed in Figure 8. This is also assumed to be low enough that the change in behavior does not affect the willingness to pay.

To calculate this new demand (Dshift) Equation 6 is used as follows: the monthly average of PV production (PVavg) is used to normalize the total PV production (PVp), and the monthly average of demand (Davg) is

(29)

-27-

used to normalize the total demand (D0). Then a new normal could be calculated using three parts of the demand normal and one part of the PV normal. This new normal is the shifted normal and can be scaled with the average monthly demand (Davg) to generate the new and adjusted curve.

Equation 6: Calculating adjusted demand with 25% shift towards PV production

𝐷_{𝑠ℎ𝑖𝑓𝑡} = 𝐷_𝑎𝑣𝑔

( 𝑃𝑉

𝑃𝑉_𝑎𝑣𝑔 + 3 𝐷₀ 𝐷_𝑎𝑣𝑔 ) 4

Using this new demand curve, the constraint can be formulated as if single household shifts their consumption with 25% following PV production they should not have a monthly bill higher than if a fixed tariff was in place.

3.4.5 Constraint 3, lower/upper -bound price rates

The lower bound price rates tells the solver the absolute minimum price rate for each price level. With expected changes to consumption patterns the actual gross revenue created would be able to shift more on a monthly basis than that of a fixed tariff. This as people shift their consumption unpredictably and the distribution in price levels might be varying as well. Because of this, it is important to generate price rates that always returns profit regardless of consumption pattern.

Whatever the cost of producing and deliver the current electricity is, the microgrid has to sell it for at least that value. This would also help to assure that expansion of generation or storage capacity always would return profit if done based of consumption patterns. Levelized cost of electricity (LCOE) is used to calculate the absolute minimum rate for each price level as well as one for the entire system. The discount rate is set to zero in this stage of calculations in order to conduct all investment analysis post simulation. For cases where a minimum discount rate would be required, this could be changed.

Equation 7: Lower bound price rates, time of use tariff

[ 𝑙𝑏₁ 𝑙𝑏₂ 𝑙𝑏₃ 𝑙𝑏₄

] = [

𝐿𝐶𝑂𝐸_𝑝𝑣 𝐿𝐶𝑂𝐸_𝑝𝑣 𝑃_{𝑖𝑚𝑝𝑜𝑟𝑡} 𝐿𝐶𝑂𝐸_{𝑠𝑡𝑜𝑟𝑎𝑔𝑒}]

+ [

𝐿𝐶𝑂𝐸_{𝑠𝑦𝑠𝑡𝑒𝑚} 𝐿𝐶𝑂𝐸_{𝑠𝑦𝑠𝑡𝑒𝑚} 𝐿𝐶𝑂𝐸_{𝑠𝑦𝑠𝑡𝑒𝑚} 𝐿𝐶𝑂𝐸_{𝑠𝑦𝑠𝑡𝑒𝑚}]

Equation 8: Lower bound price rates, time of use tariff 𝑃_{𝑒𝑥𝑝𝑜𝑟𝑡} ≤ 𝑙𝑏₁

The three LCOE are calculated with parameters as seen above in Table 2 and used for the lower bound as seen below in Equation 7 where 𝑙𝑏_𝑛 is the lower bound for a price level n. The lower bound for price levels 1 and 2 are assumed to not be lower than the revenue for exporting surplus electricity to the national grid. This would otherwise create a situation where more consumption during these hours would reduce the profit. The desired consumption pattern would then be to lessen consumption during these hours and thus not be in line with the objective of the time of use tariff. The upper bound is assumed to three times the fixed tariff and equal to all price levels.

(30)

-28-

Table 2: Parameters used when calculating LCOE LCOE PV

Costs: Initial and annual costs related to the PV array Generation: Sum of annual PV production

LCOE System

Costs: Initial and annual costs of the system (without PV and battery costs) Generation: Sum of village demand

LCOE Storage

Costs : The costs are calculated as the sum of charging-, discharging- and fixed costs:

Discharging: Running costs related to operation of battery bank Charging:

o Surplus solar: the LCOE for the PV array o National grid: the cost of electricity import Fixed: Initial cost of battery bank

Generation: Demand of hours without sufficient PV and national grid outages

3.5 Investment value

The investment value is investigated using internal rate of return (IRR) for a ten-year period. A calculated profit for the simulated year is assumed as average yearly income and used to simplify calculation to a 10^th degree polynomial, as seen in Equation 9. The polynomial is solved using Matlab.

Equation 9: Calculating IRR with static annual profit 0 = 𝐶₀ + ∑ 𝐴𝑃_𝑛

(1 + 𝐼𝑅𝑅)^𝑛

10

𝑛=1

𝐴𝑃_𝑛 = 𝐴𝑃 = 𝐷_𝑠𝑢𝑚 ∗ 𝑃_𝑓𝑡 + 𝐼_{𝑒𝑥𝑝𝑜𝑟𝑡} − 𝐶_{𝑖𝑚𝑝𝑜𝑟𝑡} (1 + 𝐼𝑅𝑅) = 𝑟

0 = −𝐶_𝑜𝑟¹⁰+ 𝐴𝑃𝑟⁹+. . . + 𝐴𝑃𝑟⁰

The annual profit (AP) is calculated using the fixed tariff (𝑃_𝑓𝑡) multiplied the sum of village demand (𝐷_𝑠𝑢𝑚) with export income (𝐼_{𝑒𝑥𝑝𝑜𝑟𝑡}) and import costs (𝐶_{𝑖𝑚𝑝𝑜𝑟𝑡}) added. The IRR is calculated using the assumption that no investments or replacements that falls outside of the operational costs has to be made

Time of use pricing strategy for Indian microgrids subordinate to grids with unreliable coverage

Time of use pricing strategy for Indian microgrids subordinate to grids with unreliable coverage

NIKLAS BRÄNNLUND

Acknowledgement

Abstract

Tidsvarierande prissättningsstrategi

för Indiska mikronät i områden med

opålitlig elförsörjning

Sammanfattning

Table of Contents

1 Introduction

1.1 Study

1.2 Research scope

2 Background

2.1 Microgrid concept

2.2 The microgrid sector in India

2.3 Role and implications of electrical energy storage

2.4 Pricing mechanism

2.5 Simulating PV production and electricity storage

3 Methodology

3.1 Key assumptions

3.2 The Model

3.3 Techno economic model

3.4 Calculating time of use tariff

3.5 Investment value