Akji kth device in class cj of M Gi pkji Power demand for the device Akji Kji Number of devices in class cj of M Gi pji Total power demand of class cj in M Gi uji Total utility value for pji. L Upper limit of the number of priority levels rest Power resource available at time t in a day.

## Background

A microgrid is usually connected to the main grid through a circuit breaker (point of common coupling (PCC), as shown in Figure 1.1), which allows the entire microgrid to be disconnected from the main grid when necessary (for example, during a fault). in the main network). Otherwise, the microgrid remains synchronized with the main grid under normal circumstances, allowing energy to be imported/exported from/to the main grid.

Challenges

## Related Works: An Overview

On the other hand, for the grid-connected mode, the usual objective will be to maximize profit. In this scheme, consumers are often motivated by offering incentives (through monetary benefits or compensations) such as discounts on their electricity bills, etc., in response to their active participation [107].

## Objectives

The purpose of the design is to minimize the total cost by exactly matching a given load curve for that area. This framework is built on the following modules, i) accurate system modeling given, inputs such as consumer loads and their behavior, available power capacities, intermittency in renewable energy, charge-discharge specifications, etc., ii) mathematically accurate problem formulation, ii) synthesis of efficient solution strategies (both optimal and heuristic ), iii) design of data generation frameworks for generating real microgrid scenarios and iv) performance evaluation of the proposed strategies through extensive experiments conducted on the generated microgrid scenarios.

## Summary of Contributions

*Equitable Electricity Distribution and Pricing for a Network of**Brownout based Blackout Avoidance Strategies in Smart Grids**An Efficient Framework for Brownout–based Appliance Scheduling**A Generic Framework for Designing a Cost–Optimal Microgrid*

Based on the alternative demand choices announced by the LAs for the considered microgrids, EDCo must decide on a policy for distributing its available power among the microgrids such that the total utility of the system is maximized. Hourly data corresponding to a specific week in the month of June (1 June to 7 June in a typical meteorological year (TMY3)) is taken as input.

## Organization of the Thesis

From the definitions above, the main characteristics of a smart grid and their benefits can be summarized as follows. i) Two-way flow of power and information: Using advanced information and communication technologies (ICT), an SG enables two-way flow of power and information. With this feature, consumers can install renewable energy resources such as wind turbines, solar panels, etc. in their own premises and the production from these sources can be fed back to the grid.

## Microgrids

### Distributed Energy Resources

Fuel cells also produce both electricity and heat, but are based on converting chemical energy from hydrogen or other fuels, such as batteries. In comparison, fuel cells are more efficient than conventional combustion systems such as diesel generators, in terms of toxic emissions.

## Demand Side Management and Demand Response Programs

### Demand Response

According to the US Department of Energy (DOE), DR is defined as, "a rate or program established to motivate changes in electricity use by end-use customers, in response to changes in the price of electricity over time, or to provide incentive payments designed to cause lower electricity consumption during times of high market prices or when grid reliability is compromised". Federal Energy Regulatory Commission defines DR as, "changes in electric use by end users from their normal consumption patterns in response to changes in the price of electricity over the course of time, or, to incentive payments designed to induce lower electricity consumption at times of high wholesale market prices or when system reliability is compromised”.

## DR-aided Power Scheduling

### DR Programs–Typical Objectives

In [98], the authors also try to minimize the consumer's bill, where in addition to household reactive loads, the set of appliances to be planned also includes storage devices and plug-in hybrid electric vehicles. The authors in [42] presented the design of a home energy management system (HEMS) that takes a day in advance as input.

## Selection and Sizing of DERs

The primary objective of the microgrid sizing problem is to minimize the total generation cost so that the prescribed loads within the region are reliably met. In general, microgrid sizing problems can be considered as generalizations of unit commitment (UC) problems and economic dispatch (ED) problems.

## Summary

### Example

Given the demand choices of the LAs and the available power of EDCo P as 15 units, the maximum system utility can be calculated by the optimal solution of Equation 3.1 (see Section 3.1, Chapter 3). In this scenario, 3 and 12 power units will be distributed to LA1 and LA2 respectively.

### Problem Definition

The EDCo's equitable power distribution objective can therefore be achieved even in the presence of rationally acting LAs with selfish motivations. In the next section, we present the design of such a truthful mechanism that appropriately addresses the EDCo's power distribution problem.

## Truthful Mechanism Design

### DP based Equitable Power Allocator (DPA)

*Complexity Analysis for DPA Algorithm*

Given, N LA, M demand choices for each load collector and PL power levels, the computational complexity of the proposed DPA algorithm is analyzed as follows. The outermost loop (Step 4) runs for each of the N LAs, the inner loop (Step 5) runs for each PL power distribution level, and finally the innermost loop (Step 7) repeats for each of the choices of request M.

### VCG-based Equitable Pricing (VCG-EP)

*Complexity Analysis of VCG-EP*

The nth iteration of the inner loop (in lines 5 to 11) fills the field δijn→p with the sum of pji (the power level of the ith subarea at the jth priority level) and β(i−1)n→p (step 6). The input consists of 168 entries (ie, the number of times T = 168), each corresponding to a specific hour in a typical week.

Example (continued

## Experimental Results

### Data Generation Framework

Cat-1 devices have the highest priority in terms of the urgency of uninterrupted power supply. Given the type of household (lower and upper power consumption limits), each household is randomly assigned several devices from the table, so that the total power consumption of the household is within the upper and lower limits.

### Results

The figure shows that the average power level that can be assigned to an LA decreases with an increase in the number of aggregators, for a fixed value of available power (P). It can be seen from this figure that the execution time increases sharply with an increase in the number of aggregators, for any given value of P.

## Summary

In this chapter, we will deal with the design of real-time power distribution mechanisms for dealing with power shortages in dynamic scenarios. The formulation of this work is based on resource allocation strategies commonly used for executing sensitive QoS tasks in real-time systems.

## System Model and Problem Formulation

### Example 1

The remaining PR power is then used to meet the non-essential power needs of the subareas.

## Dynamic Programming Based Priority level Allocator (DP)

P L Total number of power allocation levels available at power resolution pu Ki Number of priority levels available for this sub-area. The power allocation levels are defined as the number of discretely quantified power allocation units taken into account (pu) given the total available power PR (pu defines the smallest unit of power that can be allocated to a sub-area; used to indicate the difference between two consecutive levels of asset allocation that can be considered in the problem).

## Streamlined DP-based Priority Level Allocator (SDPA)

### Data Structure

Partial solutions (nodes) within the linked list ϕ21 are obtained by considering subareas S1 and S2, but restricting subarea S2 to operate only at the first priority level. Finally, at the end of the second iteration, SDPA merges the linked lists within Φi and constructs a single linked list ψi, which consists of optimal partial solutions considering up to subareas {S1,.

### Detailed Algorithm

The complexity of the innermost loop is governed by the number of nodes (or partial solutions) G in ψi−1, which is upper bound by the total number of available power levels P L. Therefore, the complexity of each iteration of the outer loop SDPA becomes O(LG).

## Proportionally Balanced Priority level Allocator (PBPA)

### Example

The subarea S1 with the highest principal value (present at the root of H), is removed from the stack.

### Handling Dynamic Demand-Supply Variations

While a max-heap keys are used to perform level enhancements during demand underloads, a min-heap is used for level degradations during overloads. So in this case, use SDPA or PBPA (exact version; Algorithm 5) to mitigate such power imbalances (depending on the size of the total service area under a utility). 5) The outage-based power allocation problem considered in this work has an important social perspective.

### Fairness

Such adjustments can be made through a slight variation of the PBPA algorithm that progressively upgrades/degrades the priority levels assigned to subareas starting with their current levels, instead of starting with level 0 (as stated in line 4 of algorithm 5 (PBPA)). So in addition to always serving essential loads of all sub-areas, repeated dynamic reassignment of power through the procedure discussed above should ensure that a sub-set of sub-areas are not unfairly selected over and over again for power allocation (because they yield higher rewards) while a few other sub-areas are unnecessarily starved.

## Experiments and Results

### Data Generation Framework

Thus, the lower a device's category, the greater the probability that the device is mapped to a high priority level. Entry at a given priority level (say, j) is also associated with a fixed tariff (tarji) per kW.

### Simulation Results

It can be seen from the figure that the total rewards obtained by all strategies increase with the number of available ones. It can be seen from the figures that as the number of stages increases, the SDPA algorithm achieves better rewards in all scenarios.

## Summary

Power shortage scenarios within such microgrids are typically caused due to a higher percentage of renewable energy penetration in the microgrid system. The microgrid utility defines a fixed number of alternative electricity rates to which its consumers can subscribe, depending on their devices.

## System Model and Formulation

### ILP Formulation

*Unique interval for elastic appliances**Power resource constraints**Objective function*

For any time interval ∈ {1..T}, the total rated power of all devices whose preferred interval contains the time window t must not exceed the total available power resource rest. The objective is to maximize the total revenue achieved by the micro-network over all interval selections of the appliances.

### A Running Example

As shown in Equation 5.5, the energy resource constraint must be applied to all time slots t (∈ {1.4}). In the next section, we present our proposed algorithmic strategy that addresses the above scheduling problem.

## Revenue–aware Appliance Scheduler (RaAS)

### Data Structures

Each such request node (e.g. a) contains a device id (a.id), id of the specific interval requested for (a.cin) and a key (a.key) whose value indicates the scheduling priority for the node. This range is divided into ranges of equal sizes such that e.g. Bin (Binc) points to the linked list of request nodes whose key values are within the range.

### Detailed Description of RaAS

If the condition holds: i) index i of the ith device is appended to the list for the first time slot of the partial schedule S (step 16), ii) the resource demand powi of Ai is subtracted from the total available resource rest at each time slot in the interval P Iij (step17) and, iii) the total income t rev is increased the income earned by the allocation of Ai atP Iij (step 18). The worst-case time complexity of the algorithm is given as O(PNa . i Mi), which indicates the sum of the interval selections over all the devices.

### Illustrative Example

Before discussing the experiments and results in detail, we illustrate the operation of the RaAS algorithm by considering the same hypothetical microgrid used for the running example in Section 5.1.2. the table shows: i) the Bindata structure at the beginning of iteration 1, ii) the amount of power resources currently available at each of the four time slots considered. The Comments column for row 2 lists the node extracted from Bin2, the updated partial schema S with device A2 inserted into the list for slot 1, and the updated revenue obtainable.

## Experiments and Results

### Data Generation Framework

This data generation framework takes into account the following inputs (grey boxes in figure 5.3): i) Number of time slots, ii) Price tariff alternatives, iii) Possible household/enterprise types and their consumption limits, iv) Lower and upper limits on the number of households for each type, v) An empirical categorization of commonly used devices, vi) Penalty factor and moderation constant and, vii) Lower and upper bound on available power at different time slots for the next day. This step takes as input i) the number of time slots in a day, ii) the power demand requests of all devices (step 3, 4 and 5) and, iii) the lower and upper bound of the available power (70% to 90) % ).

### Detailed Results

The optimal ILP strategy is also sensitive to the number of times to be scheduled. Further, the execution time can be observed to increase linearly as the number of houses increases.

## Summary

*Photovoltaic (PV) Power**Wind Turbines**Microturbines and Fuel Cells**Energy Storage Systems (ESS)*

The power generation of wind turbines depends solely on the availability of the wind. The empirical WT model based on the power curve of the turbine is given in comparison.

## Proposed Mixed Integer Linear Program (MILP)- Optimal Sizing Problem

### Cost minimizing objective function

IC represents the total investment cost of the microgrid and is calculated as the sum of three product terms, (i) the sum of unit costs (Ciinv) of the NDG number of non-renewable DGs and its installation status wi, (ii) the sum product of unit costs ( Cjinv) for NN DG number of renewable energy DGs and its installation status wj and (iii) sum of unit costs (Ckinv) of NBAT number of batteries and its installation status wk. The minimization objective is subject to various technical constraints that ensure reliable and secure operation of the microgrid.

### Constraints

Equations 6.19 and 6.20 determine the start-up costs incurred by the generator in the time slot t. Charge or discharge limit: In any time slot t, the battery unit k can either be charged or discharged, but not both.

## Experiments and Results

### Case 1: Sizing Microturbines and Fuel cells

The cost of a single microturbine/fuel cell unit is approximately $650,000, for a rated power of 1000kW system. Given, the approximate CO2 emissions for 1kW of power generated with a microturbine to be 720g, the total emissions can be calculated as kilograms.

Case 2: Sizing fossil fuel generators with renewable generators

Case 3: Sizing Distributed Generators with batteries

## Summary

Zhuang, “Agglomeration of large numbers of residential devices for demand response applications,” IEEE Transactions on Smart Grid, vol. Chattopadhyay, "Game Theory Frameworks for Demand Response in Electricity Markets," IEEE Transactions on Smart Grid, vol.

A typical layout of microgrid power system

Profits obtained by an LA, for varied inflations

Typical Loads during summer

A conceptual smart grid model, NIST Framework, Release 4.0, 2020

Taxonomy of Demand Response Programs

Compilation of the results for different number of aggregators (N ) and

Profits obtained by an LA, for varied inflations

System Model

General structure of lists ψ i or any list in Φ

Iterative steps of SDPA

Average allocated Priority level Vs Power

Reward Vs Power

Comparative results of Reward (in Percentages) with varying number of

Comparative results of average execution time (in ms) with varying num-

An outline of the proposed scheme

A pictorial representation of the data structure

Data Generation Framework- Flow Diagram