Techno-Economic Analysis of Non-Wire Alternative (NWA) Portfolios Integrating Energy Storage Systems (ESS) with Photovoltaics (PV) or Demand Response (DR) Resources Across Various Load Profiles

Abstract

The Non-Wire Alternative (NWA) approach has gained attention as a strategy to replace or defer traditional grid infrastructure upgrades by leveraging integrated solutions combining Energy Storage Systems (ESSs) with Distributed Energy Resources (DERs). The overall feasibility and economics of distributed flexibility solutions can be enhanced by leveraging the synergies among various DERs for NWA deployment. This study presents the results of a techno-economic analysis of an NWA portfolio that integrates Photovoltaic (PV) generation and Demand Response (DR) resources with ESSs. Three representative load profiles are analyzed under different load growth scenarios: a balanced mix of industrial, commercial, and residential loads; residential-dominant loads; and commercial/industrial-dominant loads. The analysis shows that the combined deployment of PVs and DRs significantly reduces the required ESS capacity. Furthermore, economic analysis based on Benefit–Cost Analysis (BCA) demonstrated that combining ESSs with either PVs or DRs enhances economic efficiency compared with an NWA portfolio that relies on ESSs alone, particularly under low-capacity factor conditions. However, the effectiveness of a DR or PV varies depending on the load profile. DR is less effective when the peak load durations are prolonged, whereas PV offers limited economic benefits under residential loads with the evening peak demand. These techno-economic results highlight the importance of tailoring NWA portfolios to specific load conditions to maximize both technical performance and economic value.

1. Introduction

In response to the climate crisis, a roadmap has been established to achieve carbon neutrality by 2050, with an interim target set for 2030, to emphasize the urgency of mitigating greenhouse gas emissions [1]. As the global share of renewable and Distributed Energy Resources (DERs) in electricity generation continues to expand, various efforts are underway to develop the supporting technologies, regulations, and policies to support distributed energy systems [2,3]. Several countries have enhanced their power system operations and market structures to facilitate the energy transition [4].

According to recent trends, renewable energy will account for approximately 43% of total global power sector investments by 2020 [5]. Investments in renewable energy capacity, as well as transmission and distribution infrastructure, are expected to increase [6]. In line with this trend, energy transition strategies are being developed to promote renewable energy deployment and reduce GHG emissions [7]. Consequently, power systems are shifting from a centralized structure based on large-scale power plants to a decentralized structure based on DERs [8].

However, the large-scale integration of variable renewable energy sources, such as solar and wind, presents significant challenges to the stability and flexibility of power systems [9,10]. This issue is particularly pronounced in South Korea, which operates an isolated power system with no interconnections to neighboring countries [11]. With renewable energy generation being concentrated in specific regions, issues such as line congestion, voltage instability, and reliability risks have become significantly severe [12]. Addressing these challenges requires cost-effective and environmentally sustainable alternatives, given the limitations in reinforcing the existing power grids [13,14].

Among the proposed solutions, Energy Storage Systems (ESS) have emerged as key resources for enhancing grid flexibility by shifting the energy supply and improving system reliability [15,16]. However, the economic feasibility of deploying an ESS as a standalone asset remains limited owing to high capital costs and relatively modest revenues [17,18]. To overcome these challenges, the Non-Wire Alternative (NWA) approach has gained increasing attention [19]. NWAs involve deferring or replacing traditional grid infrastructure upgrades with integrated solutions—typically involving ESS and DERs—based on long-term (10–15 years) load forecasts [20].

In recent years, several countries have introduced NWA to defer conventional grid infrastructure investments by leveraging DERs, ESS, and DR. For instance, Orange & Rockland (O&R) in New York deployed a 12 MW/57 MWh battery energy storage system to alleviate peak load demand without expanding physical infrastructure [21]. Similarly, NV Energy in Nevada and the Independent Electricity System Operator (IESO) in Ontario, Canada, have initiated NWA projects incorporating solar-plus-storage and competitive DER procurement mechanisms [22,23]. In Europe, although a standardized NWA regulatory framework is yet to be established, recent studies have explored the techno-economic feasibility of front-of-the-meter solar-plus-storage systems and market-based flexibility services [24]. In Japan, while the term NWA is not formally adopted, national smart grid pilot programs have demonstrated similar functions through the integration of DERs, DR, and storage to enhance operational flexibility and defer grid upgrades [25]. These international examples illustrate the practical potential of NWAs under diverse regulatory and market conditions.

In South Korea, recent efforts to implement Non-Wire Alternatives have been initiated through a pilot project launched on Jeju Island in 2024 [26]. Additionally, a renewable energy bidding system was introduced in the Korean electricity market to alleviate grid constraints by encouraging the participation of DERs such as ESS, solar, and wind. These initiatives aim to integrate real-time bidding mechanisms and promote more flexible grid operations without expanding existing grid infrastructure. By optimizing the complementary functions of various DERs within a portfolio, NWA strategies can enhance both cost efficiency and technical performance [27].

This study aims to evaluate the economic feasibility of ESS by analyzing its performance within a hybrid resource integration framework as part of an NWA portfolio. Specifically, the synergistic integration of the ESS with Photovoltaic (PV) systems and Demand Response (DR) resources is evaluated to determine its cost efficiency and contributions to NWA outcomes. Furthermore, this study analyzes the impact of regional load characteristics on the economic feasibility of ESS-integrated NWA portfolios.

Most previous studies have assessed either the technical performance or the economic viability of NWA solutions under fixed, simplified assumptions [28,29,30]. In contrast, our systematic method incorporates detailed load compositions—such as industrial, commercial, and residential—along with hourly load profiles and projected load-growth trajectories into a scenario-based ESS sizing framework. By doing so, the analysis reflects more realistic system conditions and improves the accuracy of minimum ESS capacity estimates under uncertainty.

The remainder of this paper is organized as follows: Section 2 presents an integrated method for analyzing the load characteristics. In Section 3, load profiles are classified, and a Benefit–Cost Analysis (BCA) framework is introduced to evaluate the NWA configurations of the distribution system, including ESS, PVs, and DRs. System constraints and economic formulations are also discussed. In Section 4, a case study is conducted using forecasted load data from Jeju Island. Finally, Section 5 presents the conclusions of this study, along with suggestions for future research.

2. Load Profile Characteristics

To ensure the accuracy of load forecasting, it is essential to accurately identify and classify the characteristics of electrical loads [31]. In particular, classifying the consumer’s electricity usage as residential, commercial, or industrial is critical in various applications, including ESS scheduling, DR program design, and tariff structuring [32,33].

However, in practical power system operations, load data often lack detailed customer information. In such cases, evaluating the similarity between an unknown load profile and a set of predefined reference load profiles (RLPs) offers an effective approach for inferring the underlying load type [34,35]. RLPs are representative load curves that capture the characteristic temporal patterns of each consumer category such as peak demand hours, nighttime consumption levels, and weekday–weekend variations.

By assessing the similarity between an unlabeled load and these RLPs, the loads can be classified based on their temporal alignment with known usage patterns. This approach is particularly effective in data-scarce environments compared with traditional supervised learning–based classification methods.

This study focuses on the temporal characteristics of three representative load types—residential, commercial, and industrial—based on their corresponding RLPs. Residential loads typically exhibit dominant peaks corresponding to evening household activities. These loads significantly decrease overnight, with notable differences between weekdays and weekends. Commercial loads, including those of offices, retail establishments, and educational facilities, are primarily concentrated during standard business hours, with substantial declines during nights and weekends. Industrial loads typically exhibit two dominant peaks in the morning and afternoon hours. In facilities operating continuous or multi-shift production systems, load levels often remain stable at night; however, sudden changes can occur owing to equipment startups or shutdowns. Given their large-scale and constant load patterns, industrial loads significantly influence power system operations, making them key targets for demand response and energy efficiency initiatives.

Load similarity can be evaluated using various analytical techniques, ranging from simple statistical metrics to more advanced methods such as Euclidean distance, Dynamic Time Warping (DTW), and cosine similarity [36]. Among these, Euclidean distance is widely used owing to its computational simplicity, fast processing time, and ease of interpretation, especially when compared to more complex methods. It is also effective in capturing overall shape differences in time series data, particularly when the data are normalized and time-aligned, as in this study. These characteristics make it a suitable and practical baseline for classification tasks based on reference patterns.

In this study, RLPs were adopted from previous studies [37]. To maintain consistency and enable meaningful comparisons across different load types, all load curves were normalized using minimum–maximum (min–max) scaling to a range between 0 and 1 [38].

$${{\mathrm{X}}_{\mathrm{i}}}^{\prime }={\displaystyle \frac{{\mathrm{X}}_{\mathrm{i}}-{\mathrm{X}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}{{\mathrm{X}}_{\mathrm{m}\mathrm{a}\mathrm{x}}-{\mathrm{X}}_{\mathrm{m}\mathrm{i}\mathrm{n}}}}$$

(1)

where i denotes a specific hour of the day, and ${\mathrm{X}}_{\mathrm{i}}$ represents the load at that hour. The variables ${\mathrm{X}}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ and ${\mathrm{X}}_{\mathrm{m}\mathrm{i}\mathrm{n}}$ denote the maximum and minimum hourly load values within the observed ${\mathrm{X}}_{\mathrm{i}}$ range, respectively. The normalized load value, denoted by ${{\mathrm{X}}_{\mathrm{i}}}^{\prime }$, represents the load normalized via min–max scaling.

Figure 1 illustrates the scaled RLPs used in this study.

3. NWA-BCA Prediction Model

The load forecasting method is proposed in Section 3.1 and Section 3.2. Subsequently, the calculation of ESS capacity and the NWA benefit–cost ratio (BCR) are presented in Section 3.3 and Section 3.4, respectively. A flowchart of the NWA-BCA prediction model is shown in Figure 2.

Figure 2 illustrates the framework of the proposed method in this study, comprising a multi-stage process for minimum ESS capacity estimation and NWA economic assessment. In the first step, 24 h sample load data are compared against representative RLPs using Euclidean distance to extract a load characteristics coefficient, which quantifies the similarity to residential, commercial, or industrial load types. This coefficient is then used to forecast future load profiles based on 8760 h data.

The forecasted hourly load is subsequently adjusted by subtracting PV generation and DR participation to derive the net load profile. Based on the net load, the minimum required ESS capacity is determined to mitigate overload conditions while satisfying system constraints. Finally, a BCR is calculated to evaluate the economic feasibility of the NWA approach.

3.1. Load Characteristic Coefficient

To analyze the characteristics of unlabeled loads, it is essential to define appropriate load characteristic coefficients [39,40]. In this study, a systematic method is introduced to classify unlabeled loads by comparing them with predefined residential, commercial, and industrial RLPs. For each load category, the RLP is generated based on the average daily load curve constructed using load data in a pervious study [41]. The similarity between the unlabeled load and each RLP is measured using the Euclidean distance metric, which is widely adopted to measure dissimilarities in time-series data.

The Euclidean distance $D$ between an unlabeled load vector $X^{\prime }=$[$x_1$,$x_2$,$x_3$, …, $x_{24}$] and a reference load vector $Y^{\prime }=$ [$y_1$,$y_2$,$y_3$,  , $y_{24}$] is defined as where $i$ denotes the hourly time index from 1 to 24, and $x_i$ and $y_i$ represent the load values at the $i$th hour for the target and RLP, respectively.

$$D=\sqrt{\sum _{i=1}^{24}{\left(x_i-y_i\right)}^2}$$

(2)

This distance metric reflects the absolute difference between the two load profiles. However, because raw distance values are not directly interpretable as similarity measures, they are transformed into normalized similarity scores using a monotonic inverse transformation. To quantify the similarity between an unlabeled load profile and each RLP, the following similarity score $S$ is used.

$$S={\displaystyle \frac{1}{1+D}}$$

(3)

where $D$ represents the Euclidean distance between an unlabeled load profile and an RLP. It quantifies the absolute difference in hourly load values between the two profiles over a 24 h period. This transformation normalizes the similarity score to range between 0 and 1, where higher values indicate a greater similarity. In this study, the similarity score $S$ is used as the load characteristic coefficient for further classification and scenario analysis.

The relative similarity between different load types is assessed using the proposed method, which enabled a more effective similarity analysis than simple distance-based methods. The most representative load patterns and their underlying characteristics within the predefined residential, commercial, and industrial categories are identified based on the similarity scores.

3.2. Load Forecasting Method

After determining the load characteristic coefficient for each distributed feeder load using a similarity analysis, time-dependent coefficients are integrated into the load forecasting model to reflect the varying growth rates based on load characteristics, as shown in Equations (4) and (5).

$${d^{\prime }}_{L,t}=d_L·\left({\displaystyle \frac{{\beta }_C{\gamma }_{C,t}+{\beta }_R{\gamma }_{R,t}+{\beta }_I{\gamma }_{I,t}}{{\beta }_C+{\beta }_R+{\beta }_I}}\right),\forall t\in T$$

(4)

$${a^{\prime }}_t=a_t·{\left(1+{d^{\prime }}_{L,t}\right)}^N,\forall t\in T$$

(5)

where $a_t$ denotes the current hourly load at hour t; $d_L$ represents the overall annual load growth rate; the coefficients ${\beta }_C$, ${\beta }_R$, and ${\beta }_I$ indicate the characteristic contributions of the residential, commercial, and industrial load types, respectively, and are empirically determined based on the electricity consumption statistics of the target region. The variable ${\gamma }_{x,t}$ denotes the normalized load ratio of type $x$ (residential, commercial, or industrial) at hour t, obtained through the min–max scaling of RLPs for each load type. Unlike a simple arithmetic mean, the weighted mean used in Equation (4) accounts for the relative contributions of each load type by normalizing with the sum of ${\beta }_C$, ${\beta }_R$, and ${\beta }_I$, thereby providing a more accurate representation of growth dynamics across the overall load profile. It should be noted that inter-sectoral correlations are not explicitly modeled in this approach, as the objective is to approximate aggregated demand characteristics rather than to capture interaction effects among load types.

When ${\gamma }_{x,t}$ approaches 1, it indicates that the load of type $x$ reaches its peak at hour t, thereby allocating a greater portion of the projected load growth to that specific hour. Conversely, when the normalized value approaches 0, the amount of load growth allocated to that hour is significantly reduced.

Once the time-dependent growth coefficient ${d^{\prime }}_{L,t}$ is derived from Equation (4), the forecasted load ${a^{\prime }}_t$ over a period of N years is calculated by using a geometric growth model, as shown in Equation (5). In this formulation, at ${a^{\prime }}_t$ represents the projected hourly load, and the coefficient ${d^{\prime }}_{L,t}$ captures the variation in growth rate across time based on the prevailing load characteristics. As a result, the forecasted load profile does not increase uniformly but reflects sector-specific and time-sensitive dynamics.

3.3. ESS Capacity Estimation

To enhance the economics of the NWA, it is essential to minimize the required capacity of the ESS, which is typically the most cost-intensive resource [42,43].

3.3.1. Objective Function

The objective function for determining the minimum required capacity of the ESS is expressed in Equation (6) as follows [44,45]:

$$\min C_{ESS}$$

(6)

Equation (6) represents the objective function for determining the minimum ESS energy capacity required to mitigate the overload of the distribution system over a period of 1 year. In this study, the energy capacity of the ESS is denoted as $C_{ESS}$, which corresponds to the maximum state of charge (SOC) observed during the entire planning horizon. Therefore, the objective is to minimize $C_{ESS}$ = $\underset{t\in T}{\max }SOC\left(t\right)$, which effectively represents the minimum energy storage capacity required for the ESS to operate in compliance with all operational constraints while mitigating network overload conditions.

The power conversion system (PCS), expressed in megawatts (MW), controls the charging and discharging power of the ESS. It determines the maximum rate at which the ESS can charge or discharge energy to or from the grid. A higher PCS capacity enables faster system response, particularly during grid events such as peak shaving and frequency regulation. Although the objective in Equation (6) aims to minimize ESS size, this sizing result is directly linked to economic implications through the cost analysis conducted in the subsequent sections. The net present cost is evaluated based on the determined ESS capacity across various scenarios, enabling a comprehensive techno-economic assessment.

Minimizing the peak SOC across all time steps ensures that the ESS is strictly sized based on operational requirements. The optimization problem can be formulated subject to constraints, including charge and discharge power limits, non-simultaneous charging and discharging, SOC dynamics, and the requirement to mitigate network congestion. This approach enables the cost-effective sizing of the ESS by identifying the minimum capacity required to support essential grid functions.

This formulation ensures that the ESS is appropriately sized to satisfy the performance requirements of the system, avoiding unnecessary oversizing and enhancing the overall economic efficiency. The ESS is subject to both physical and operational constraints, including limitations on the charging and discharging power and energy capacity. These constraints must be incorporated into the optimization problem to ensure a feasible and operationally realistic solution. The detailed ESS constraints are defined below.

3.3.2. Charge and Discharge Power Limit Constraints

The charging power $P_{ch}\left(t\right)$ and discharging power $P_{dis}\left(t\right)$ of the ESS at time t must not exceed their respective maximum allowable values as defined in Equation (7):

$$0\le P_{ch}\left(t\right)\le P_{ch}^{max},0\le P_{dis}\left(t\right)\le P_{dis}^{max},\forall t\in T$$

(7)

where $P_{ch}^{max}$ and $P_{dis}^{max}$ represent the maximum charging and discharging powers, respectively, and T is the set of time intervals.

3.3.3. Prohibition of Simultaneous Charging and Discharging Constraints

The ESS is not permitted to charge or discharge simultaneously. This operational constraint is enforced using a binary control variable u(t) ∈ {0, 1} as defined in Equation (8):

$$P_{ch}\left(t\right)\le u\left(t\right)·P_{ch}^{max},P_{dis}\left(t\right)\le (1-u(t\left)\right)·P_{dis}^{max},\forall t\in T$$

(8)

3.3.4. SOC Dynamics and Limit Constraints

The SOC of the ESS evolves over time based on the net energy flow and must remain within its operational limits, as defined in Equations (9) and (10):

$$SOC\left(t\right)=SOC(t-1)+{\eta }_{ch}·P_{ch}\left(t\right)·∆t-{\displaystyle \frac{P_{dis}\left(t\right)·∆t}{{\eta }_{dis}}},\forall t\in T$$

(9)

$${SOC}^{min}\le SOC\left(t\right)\le {SOC}^{max},\forall t\in T$$

(10)

where ${\eta }_{ch}$ and ${\eta }_{dis}$ indicate the charging and discharging efficiencies, respectively; $∆t$ is the time step; and ${SOC}^{max}$ and ${SOC}^{min}$ are the minimum and maximum SOC levels, respectively.

3.3.5. Distribution Line Overload Prevention Constraints

To ensure the secure and cost-effective operation of distribution networks with integrated ESS, it is essential to prevent the total power flow from exceeding the thermal limits of distribution lines [46]. Accordingly, the following constraint is introduced to mitigate potential line overloading and ensure safe operation of the distribution network, as defined in Equations (11) and (12):

$$P_{net\_load}\left(t\right)={a\prime }_t-P_{PV}\left(t\right)-P_{DR}\left(t\right),\forall t\in T$$

(11)

$$P_{net\_load}\left(t\right)-P_{dis}\left(t\right)+P_{ch}\left(t\right)\le P_{line}^{max},\forall t\in T$$

(12)

where $P_{net\_load}\left(t\right)$ denotes the net load at time t. The net load is defined as the original load demand ${a^{\prime }}_t$ minus the PV generation $P_{PV}\left(t\right)$ and the load reduction from the DR resources $P_{DR}\left(t\right)$. This value reflects the actual power requirements that must be supplied by the distribution system. The term $P_{line}^{max}$ represents the thermal capacity limit of the distribution line. This constraint represents the net power flow through the distribution network, considering both the native load and operational impacts of the ESS, PV, and DR. By maintaining that the power flow remains within the physical limits of the line, the system avoids thermal violations that could compromise network reliability or result in additional operational costs.

From the economic perspective of power systems, this condition enables the flexible scheduling of ESS operations—particularly during peak demand periods—where ESS discharge can mitigate line congestion and defer expensive investments in network reinforcement.

3.4. BCA

In this study, the economics of the NWA portfolio are assessed using the BCA framework [47]. BCA is a metric used in cost–benefit analysis, representing the ratio of the present value of total program benefits to that of total program costs [48]. A BCA value greater than 1 indicates that the expected benefits exceed the associated costs, suggesting that the project is economically viable. In this study, the BCA is calculated using a modified version of the net present value (NPV), as defined in Equation (13):

$$\mathrm{B}\mathrm{C}\mathrm{A}=\sum _{y=1}^N{\displaystyle \frac{(B_y-C_y)}{{(1+R)}^y}}\times {\displaystyle \frac{1}{(C_0-B_0)}}$$

(13)

where N denotes the total program duration in years, and R represents the discount rate. The difference between $B_y$ and $C_y$ corresponds to the net benefit in year y. The term $C_0$ refers to the initial investment cost, including expenditures for PV and ESS installations, and $B_0$ represents the avoided cost of reinforcing the distribution system due to NWA deployment. If a continuous increase in the load requires immediate grid reinforcement, the deferral benefit $B_0$ becomes 0.

Figure 3 illustrates the financial benefits obtained by deferring traditional grid investments through demand reduction enabled by the NWA framework [49]. As shown in the figure, the forecasted peak load without NWA implementation (gray dashed line) is expected to exceed the distribution line limit in year $y_a$, resulting in system overload. In traditional planning, grid reinforcement is typically scheduled for year $y_a$. However, by applying the NWA, the increase in the peak load can be mitigated (blue solid line), deferring the grid upgrade to year $y_b$. Accordingly, the deferral benefit—comprising capital investment and operation and maintenance (O&M) cost savings—is evaluated based on the difference between years $y_b$ and $y_a$, with year $y_a$ as the baseline reference.

The cost and benefit components of the NWA portfolio vary depending on the planning objectives and the quantities of each NWA resource required for the configuration.

Table 1. Parameters of BCA.

Type		Components	Values
Benefit	$B_0$	Distribution system reinforcement	3,000,268 (USD)
	$B_y$	Distribution system O&M	5004 (USD)
	$B_y$	ESS charge/discharge revenue	0.06 (USD/kWh)
	$B_y$	PV generation revenue	0.1 (USD/kWh)
	$B_y$	PV renewable energy certificate (REC)	514.1 (USD/REC)
Cost	$C_0$	ESS installation (PCS)	145.3 (USD/kW)
	$C_0$	ESS installation (SOC)	273.2 (USD/kW)
	$C_0$	PV installation	877 (USD/kW)
	$C_y$	ESS O&M	3.5 (USD/kW)
	$C_y$	PV O&M	13.7 (USD/kW)
	$C_y$	DR incentive	0.07 (USD/kWh)

lists the specific cost–benefit components of the NWA portfolio. All cost and benefit components are evaluated as net present values (NPVs), using a 5% discount rate and 2024 as the base year. According to Equation (13), the $B_0$ represents the benefit resulting from avoided grid reinforcement. $B_y$ includes annual benefit components such as O&M savings and revenue from DER operations. $C_0$ denotes the initial capital costs associated with the installation of ESS and PV systems, while $C_y$ refers to annual operating costs such as O&M expenses and DR incentives.

The cost and benefit parameters were determined using the average market values observed in the South Korean electricity sector. The unit cost for distribution system reinforcement was referenced from the standard construction cost data provided by Korea Electric Power Corporation (KEPCO) [50]. Specifically, the baseline cost for distribution system reinforcement was set at USD 3,000,268 with an annual O&M cost of USD 5004. The arbitrage revenue of ESS was estimated using the daily range of the system marginal price (SMP) in 2024, assuming that the ESS operates by charging during low-SMP hours and discharging during high-SMP hours. The revenue from ESS operations was calculated based on a charge/discharge revenue of USD 0.06 per kWh derived from the daily SMP range in South Korea during 2024 [51]. The annual revenue from PV systems was estimated based on two components: electricity sales and Renewable Energy Certificate income. For PV systems, the generation revenue was assumed to be USD 0.10 per kWh, and the REC value was set to 514.1 REC/kW, reflecting the national market average during 2024 [52,53].

The capital costs for ESS installation were divided into PCS and SOC components, amounting to USD 145.3/kW and USD 273.2/kW, respectively, and the annual O&M cost for the ESS was set to USD 3.5/kW [54]. For PV systems, the installation cost was assumed to be USD 877/kW, with an annual O&M cost of USD 13.7/kW [55]. Additionally, DR incentives were included in the benefit evaluation, with a compensation rate of USD 0.07/kW [56].

4. Case Study

The NWA-BCA prediction model was applied to hourly load data over a 10-year forecasting period from 2024 to 2033. The load data used in this study were obtained from distribution feeder datasets in South Korea. During the data preprocessing stage, a similarity analysis was conducted on three representative feeders to extract the characteristic load patterns, which are then applied to the load forecasting process. The required SOC capacity of the ESS was estimated over a 10-year period using the forecasted load profiles. Subsequently, the performance of the NWA portfolio was evaluated using the BCA formulation.

In this study, three distinct load profile distributions were considered. The representative load profiles for Cases 1, 2, and 3 are shown in Figure 4, Figure 5, and Figure 6, respectively.

4.1. Load Similarity Analysis

Table 2. Load characteristics coefficient.

Type		Components	Values
Case 1	${\beta }_C$	Commercial characteristic coefficient	0.398
	${\beta }_R$	Residential characteristic coefficient	0.345
	${\beta }_I$	Industrial characteristic coefficient	0.330
Case 2	${\beta }_C$	Commercial characteristic coefficient	0.369
	${\beta }_R$	Residential characteristic coefficient	0.486
	${\beta }_I$	Industrial characteristic coefficient	0.287
Case 3	${\beta }_C$	Commercial characteristic coefficient	0.565
	${\beta }_R$	Residential characteristic coefficient	0.374
	${\beta }_I$	Industrial characteristic coefficient	0.502

summarizes the load characteristic coefficients β, which are derived from a similarity analysis between the normalized load profiles of the commercial, residential, and industrial sectors (Figure 1) and the load profiles of Cases 1, 2, and 3 (Figure 4, Figure 5 and Figure 6). The similarity is quantified using Equation (2), and the resulting coefficients represent the degree to which each case resembles the standard load types.

Analysis of the load characteristic coefficients in Case 1 shows comparable values across the residential, commercial, and industrial components, indicating a balanced load pattern that includes all three sectors. Although the overall load remains low and stable throughout most hours of the day, a pronounced peak at 11:00 a.m., followed by a sharp decline at 2:00 p.m., indicates a significant degree of temporal variability within the load profile.

In Case 2, the residential load coefficient is significantly higher than those of the others. The corresponding load profile presents distinct temporal characteristics, most notably a sharp peak in the late evening hours. The load begins to rise rapidly at approximately 4:00 p.m., peaks between 7:00 and 8:00 p.m., and then sharply declines. This peak indicates the dominance of residential consumption, as evening load increases are typically associated with household electricity usage behaviors, such as lighting, cooking, and appliance operation, following standard working hours.

In Case 3, the commercial and industrial load coefficients are comparatively high. The load profile demonstrates a broad and sustained peak period during daytime hours, gradually increasing from approximately 5:00 a.m. and reaching an initial peak near 11:00 a.m. A temporary dip is observed around noon, followed by a secondary peak at approximately 3:00 p.m., which is characteristic of typical industrial consumption patterns. The load remains high throughout the late afternoon and gradually declines toward midnight. This dual-peak structure and the extended high demand period suggest a mixed-load composition. This profile is predominantly driven by commercial and industrial consumption and is characterized by sustained electricity usage during standard business hours.

4.2. NWA Portfolio

To perform load forecasting based on the load characteristic coefficients defined in Equation (4), it is necessary to calculate the net load that incorporates the effects of both PV generation and DR reduction. The PV generation data are externally sourced at an hourly resolution, covering a full annual horizon of 8760 h. For DR, it is assumed that a 4 h event is dispatched during the period of highest peak demand. Furthermore, to investigate how variations in the availability of PV and DR resources influence the required ESS capacity, an NWA portfolio is constructed, as summarized in

Table 3. NWA portfolio with PV and DR.

		NWA Portfolio
Overall Load Growth $({\mathit{d}}_{\mathit{L}}$)		1%		3%		5%
NWA Resources (MW)		PV	DR	PV	DR	PV	DR
Case 1	(a)	0	0	0	0	0	0
	(b)	0	1	0	1	0	1
	(c)	0	2	0	2	0	2
	(d)	1	0	1	0	1	0
	(e)	1	1	1	1	1	1
	(f)	1	2	1	2	1	2
	(g)	2	0	2	0	2	0
	(h)	2	1	2	1	2	1
	(i)	2	2	2	2	2	2
Case 2	(a)	0	0	0	0	0	0
	(b)	0	1	0	1	0	1
	(c)	0	2	0	2	0	2
	(d)	1	0	1	0	1	0
	(e)	1	1	1	1	1	1
	(f)	1	2	1	2	1	2
	(g)	2	0	2	0	2	0
	(h)	2	1	2	1	2	1
	(i)	2	2	2	2	2	2
Case 3	(a)	0	0	0	0	0	0
	(b)	0	1	0	1	0	1
	(c)	0	2	0	2	0	2
	(d)	1	0	1	0	1	0
	(e)	1	1	1	1	1	1
	(f)	1	2	1	2	1	2
	(g)	2	0	2	0	2	0
	(h)	2	1	2	1	2	1
	(i)	2	2	2	2	2	2

Scenarios (a)–(i): (a) Base, (b) PV 0 MW + DR 1 MW, (c) PV 0 MW + DR 2 MW, (d) PV 1 MW + DR 0 MW, (e) PV 1 MW + DR 1 MW, (f) PV 1 MW + DR 2 MW, (g) PV 2 MW + DR 0 MW, (h) PV 2 MW + DR 1 MW, (i) PV 2 MW + DR 2 MW.

.

Table 3 defines the NWA portfolios used in this study by specifying the installed capacities of PV and DR under different annual load growth rates (1%, 3%, and 5%). The numerical values (0, 1, and 2) represent arbitrary capacity levels in MW, where 0 indicates no deployment, 1 indicates a moderate level (1 MW), and 2 indicates the maximum level (2 MW). Scenarios (a) through (i) are designed to reflect increasing levels of DER integration, allowing for comparative analysis of their impact on the minimum ESS capacity and BCA results.

Table 3 presents the NWA portfolios that integrate PV and DR resources under different load growth scenarios. Three levels of load growth were considered. For each growth level, nine portfolio scenarios (labeled (a)–(i)) were defined across three representative load types (Cases 1, 2, and 3). The values in each cell represent the relative capacities of PV and DR resources, expressed in normalized units to facilitate comparative analysis.

Table 4. Minimum ESS of the PCS and SOC capacity ($d_L=1$).

Type		ESS PCS (MW)	ESS SOC (MWh)	Total Charging (MWh)
Case 1	(a)	1.2	1.6	18.1
	(b)	0.3	0.4	0.6
	(c)	0.3	0.4	0.6
	(d)	0.6	0.8	7.9
	(e)	0.2	0.4	0.3
	(f)	0.2	0.4	0.3
	(g)	0.4	0.6	4
	(h)	0.1	0.2	0.1
	(i)	0.1	0.2	0.1
Case 2	(a)	2.4	7.2	86
	(b)	1.3	3.2	15.3
	(c)	0.2	0.4	0.44
	(d)	2.4	7	66.4
	(e)	1.3	3	14.7
	(f)	0.2	0.4	0.28
	(g)	2.4	6.8	59.2
	(h)	1.3	3	14.2
	(i)	0.2	0.4	0.44
Case 3	(a)	1	6.8	56.8
	(b)	1	2.6	15.1
	(c)	1	2.4	15.1
	(d)	0.4	1.8	6.1
	(e)	0.2	0.4	0.8
	(f)	0.2	0.4	0.8
	(g)	0.1	0.2	0.18
	(h)	0	0	0
	(i)	0	0	0

Scenarios (a)–(i): Various combinations of DR (0–2 MW) and PV (0–2 MW); see Table 3 for details.

,

Table 5. Minimum ESS of the PCS and SOC Ccpacity ($d_L=3$).

Type		ESS PCS (MW)	ESS SOC (MWh)	Total Charging (MWh)
Case 1	(a)	3	11.8	475.6
	(b)	2	6.4	241.6
	(c)	2	6.4	202.3
	(d)	2.4	11.2	353.1
	(e)	1.6	5.8	153.9
	(f)	1.6	5.6	126.2
	(g)	2.3	10.8	289.5
	(h)	1.5	5.4	114.6
	(i)	1.5	5.4	92.8
Case 2	(a)	4.6	16.8	515.3
	(b)	3.5	11.4	219.5
	(c)	2.4	7.8	126.9
	(d)	4.6	16.4	448.1
	(e)	3.5	10.8	175.9
	(f)	2.4	6.6	97
	(g)	4.6	16	404.6
	(h)	3.5	10.6	154.6
	(i)	2.3	6.4	85.3
Case 3	(a)	3.1	37.2	1212
	(b)	2.8	26.2	866
	(c)	2.8	20.6	768.1
	(d)	2.5	25	643.9
	(e)	2	19	424.8
	(f)	2	14	390.6
	(g)	2	18.6	321.8
	(h)	1.7	13.2	197.7
	(i)	1.7	9.6	186.6

Scenarios (a)–(i): Various combinations of DR (0–2 MW) and PV (0–2 MW); see Table 3 for details.

and

Table 6. Minimum ESS of the PCS and SOC capacity ($d_L=5$).

Type		ESS PCS (MW)	ESS SOC (MWh)	Total Charging (MWh)
Case 1	(a)	5	38.2	3294
	(b)	3.9	32.6	2718
	(c)	3.7	30.8	2488
	(d)	4.6	37	2535
	(e)	3.6	31.4	2150
	(f)	3.6	26.2	1940
	(g)	4.5	35.8	2146
	(h)	3.8	30.2	1780
	(i)	3.8	25	1585
Case 2	(a)	7.1	41.6	2079
	(b)	5.9	36	1632
	(c)	4.8	30.8	1357
	(d)	7	35	1854
	(e)	5.9	29.4	1424
	(f)	4.8	24.6	1165
	(g)	7	29.8	1701
	(h)	5.9	24.2	1289
	(i)	4.8	18.6	1045
Case 3	(a)	5.4	190.6	4915
	(b)	4.9	162.8	4429
	(c)	4.9	135.2	4105
	(d)	4.8	162.4	3647
	(e)	4.4	134.6	3238
	(f)	4.4	107	2988
	(g)	4.3	134.2	2705
	(h)	3.8	112	2364
	(i)	3.8	90	2178

Scenarios (a)–(i): Various combinations of DR (0–2 MW) and PV (0–2 MW); see Table 3 for details.

present the minimum required PCS and SOC capacities of the ESS, and the total charging energy for the three representative cases based on various NWA portfolio configurations from scenario (a) to (i). The baseline scenario without NWA implementation and that with full NWA integration were compared. For consistency, the charging and discharging efficiencies (${\eta }_{ch}$ and ${\eta }_{dis}$) were both set to 90%. Scenario (a) represents an ESS-only configuration, in which the NWA portfolio consists solely of an ESS. By contrast, scenario (i) corresponds to a fully integrated NWA portfolio that integrates both PV and DR resources to their maximum extent, thereby minimizing the required ESS capacity. Parameter $d_L$ denotes the annual load growth rate, which is set to 1%, 3%, and 5%. The deferral period for grid reinforcement is set to 10 years across all load profile cases and the discount rate is assumed to be 5%.

According to Table 4, Table 5 and Table 6, even when the same NWA portfolio is applied, the ESS PCS and SOC capacity values differ across scenarios for each case due to load patterns and compositions to each case. In scenario (a), the ESS capacity and total charging values are the highest across all cases. This indicates that the absence of DER leads to a substantial increase in ESS capacity and operational requirements. For instance, in Case 3, scenario (a) requires 5.4 MW of PCS capacity, 190.6 MWh of SOC, and a total charging value of 4915 MWh. In contrast, in scenario (i), where DERs are fully integrated, these values are reduced to 3.8 MW, 90 MWh, and 2178 MWh, respectively. This means that ESS requirements tend to decrease as DR and PV resources are incorporated into the system.

However, such reductions are more pronounced under low-load conditions, while the integration of DER becomes less effective under high-load growth scenarios. For example, Table 4 shows that ESS capacity in Case 3 can be completely eliminated under scenario (i). In contrast, Table 6 shows that a substantial ESS capacity is still required under the same scenario. These results suggest that the required ESS capacity varies depending on each case’s load pattern and composition even with the same NWA portfolio. Therefore, customized planning that reflects regional demand characteristics is essential when implementing NWAs. In particular, when the load growth characteristics and DR participation are effectively coordinated, ESS investment requirements can be significantly minimized. Based on these results, a BCA is conducted for all scenarios (a)–(i) across the three representative case studies presented in this study.

4.3. Numerical Results of BCA

The BCA results for the NWA portfolios presented in Table 3 are summarized in

Table 7. BCA results of NWA portfolios.

Type	${\mathit{d}}_{\mathit{L}}$	BCA
		(a)	(b)	(c)	(d)	(e)	(f)	(g)	(h)	(i)
Case 1	1	0.03	0.05	0.05	1.15	1.33	1.33	1.42	1.53	1.53
	3	0.06	0.05	0.05	0.35	0.53	0.53	0.58	0.83	0.83
	5	0.13	0.12	0.12	0.21	0.23	0.26	0.31	0.34	0.38
Case 2	1	0.02	0.01	0.05	0.62	0.72	1.33	0.71	1.02	1.48
	3	0.04	0.03	0.03	0.26	0.33	0.46	0.44	0.56	0.74
	5	0.07	0.07	0.07	0.19	0.2	0.22	0.32	0.36	0.43
Case 3	1	0.02	0.02	0.02	0.95	1.33	1.33	1.54	1.61	1.61
	3	0.05	0.05	0.06	0.2	0.24	0.3	0.41	0.51	0.62
	5	0.04	0.04	0.04	0.06	0.07	0.09	0.1	0.11	0.14

Scenarios (a)–(i): Various combinations of DR (0–2 MW) and PV (0–2 MW); see Table 3 for details.

.

According to the results summarized in Table 5, the BCA value is the lowest in the scenario involving only the ESS. The synergistic integration of additional resources, such as DR or PV generation, significantly increases the BCA value, particularly under low load factor conditions. However, under high load factor conditions, even with full utilization of DR or PV generation, the BCA does not improve significantly. In such cases, it may be more economical to either scale up the deployment of distributed energy resources or reinforce the power infrastructure.

Moreover, even when the same NWA portfolio is applied, the resulting BCA values varied depending on the load characteristics of each case. This emphasizes the importance of considering regional demand profiles and load patterns when evaluating the economic viability of NWA implementation.

Table 8. ESS capacity variation with DR participation ($d_L=3$).

Type	Values
	Scenario (i) in Case 1			Scenario (i) in Case 2			Scenario (i) in Case 3
	DR Rate 80%	DR Rate 100%	DR Rate 120%	DR Rate 80%	DR Rate 100%	DR Rate 120%	DR Rate 80%	DR Rate 100%	DR Rate 120%
ESS PCS (MW)	1.5	1.5	1.5	2.8	2.3	1.9	1.7	1.7	1.7
ESS SOC (MWh)	5.4	5.4	5.4	8.2	6.4	5	10.6	9.6	8.4
Total Charging (MWh)	94.4	92.8	92.7	100.3	85.3	77.4	187.1	186.6	186.5
BCA	0.83	0.83	0.83	0.65	0.74	0.83	0.59	0.62	0.67

Scenario (i): Maximum deployment of DER resources (PV 2 MW + DR 2 MW).

presents a comparative analysis of Case 1-(i), Case 2-(i), and Case 3-(i) under a 3% annual load growth, with varying DR participation rates of 80%, 100%, and 120%. In Case 1-(i), the ESS PCS and SOC capacities remain constant at 1.5 MW and 5.4 MWh across all DR participation rates. This indicates that DR has a relatively limited effect on reducing the ESS SOC compared to PV systems, resulting in low sensitivity to DR participation. In Case 2-(i), the required PCS and SOC capacities decrease from 2.8 MW and 8.2 MWh to 1.9 MW and 5.0 MWh as the DR participation rate increases. Case 2-(i) is the most sensitive to DR participation because DR is more effective than PV in reducing evening peak loads. In Case 3-(i), while the PCS capacity remains constant at 1.7 MW across all DR conditions, the SOC capacity decreases from 10.6 MWh to 8.4 MWh with higher DR participation. This means that DR is less effective in mitigating specific peak loads, so the SOC changes more than the PCS when DR participation increase.

Table 9. BCA variation with discount rate ($d_L=3$).

Type	R	BCA
		(a)	(b)	(c)	(d)	(e)	(f)	(g)	(h)	(i)
Case 1	3	0.06	0.06	0.05	0.39	0.58	0.59	0.64	0.91	0.91
	5	0.06	0.05	0.05	0.35	0.53	0.53	0.58	0.83	0.83
	7	0.05	0.05	0.04	0.32	0.48	0.48	0.53	0.75	0.75
Case 2	3	0.05	0.03	0.03	0.28	0.36	0.51	0.48	0.61	0.82
	5	0.04	0.03	0.03	0.26	0.33	0.46	0.44	0.56	0.74
	7	0.04	0.03	0.02	0.23	0.3	0.42	0.4	0.51	0.67
Case 3	3	0.06	0.06	0.06	0.22	0.27	0.34	0.46	0.57	0.69
	5	0.05	0.05	0.06	0.2	0.24	0.3	0.41	0.51	0.62
	7	0.05	0.05	0.05	0.18	0.22	0.28	0.38	0.47	0.57

Scenarios (a)–(i): Various combinations of DR (0–2 MW) and PV (0–2 MW); see Table 3 for details.

presents the BCA values for Case 1, Case 2, and Case 3 under discount rates of 3%, 5%, and 7%. The results from scenarios (a) to (i) show that BCA values tend to decrease as the discount rate increases. Additionally, scenarios with higher BCA values indicate greater sensitivity to discount rate. This implies that the discount rate is an important factor in the economic evaluation of NWA.

Table 10. Effectiveness of NWA resources with ESS and DR ($d_L=3$).

Type	Values
	Case 1		Case 2		Case 3
	(a)	(c)	(a)	(c)	(a)	(c)
ESS PCS (MW)	3	2	4.6	2.4	3.1	2.8
ESS SOC (MWh)	11.8	6.4	16.8	7.8	37.2	20.6
Total Charging (MWh)	475.6	202.3	515.3	126.9	1212	768.1
BCA	0.06	0.05	0.04	0.03	0.05	0.06

Scenario (a): Base case with no DER resources; Scenario (c): Maximum DR-only deployment (2 MW), without PV.

presents a comparative analysis of the effectiveness of NWA resources with DR integration across different load profiles. In all cases, the application of DR reduced the required ESS capacity compared with scenarios without DR. For example, in Case 1, the PCS capacity decreased from 3.0 MW to 2.0 MW, whereas the SOC reduced from 11.8 MWh to 6.4 MWh. In addition, the total ESS charging (or discharging) volume decreased with the implementation of DR. Notably, in Case 2, the energy requirement was significantly reduced from 515.3 MWh to 126.9 MWh, indicating that DR can effectively reduce the operational burden on the ESS.

These results suggest that DR contributes to reducing the required ESS capacity by mitigating the peak loads and enhancing the operational flexibility of the system. However, from the perspective of BCA, DR does not lead to a clear improvement in BCA outcomes. This is primarily because, although DR reduces the required ESS capacity, it does not generate revenue independently.

Furthermore, the reduction rates in the PCS, SOC, and total ESS charging volumes vary across the three cases, indicating that the impact of DR is dependent on the underlying load characteristics and system conditions. As the duration of the peak load increases, the effectiveness of the DR tends to decrease. Consequently, in Case 3, which has a higher proportion of commercial and industrial loads, the reduction in the ESS SOC capacity owing to DR implementation is less pronounced than in the other cases. Therefore, when integrating DR into the NWA portfolio, it should be strategically combined with other resources, such as PV and ESS, to maximize the overall economic feasibility.

Table 11. Effectiveness of NWA resources with ESS and PV ($d_L=3$).

Type	Values
	Case 1		Case 2		Case 3
	(a)	(g)	(a)	(g)	(a)	(g)
ESS PCS (MW)	3	2.3	4.6	4.6	3.1	2
ESS SOC (MWh)	11.8	10.8	16.8	16	37.2	18.6
Total Charging (MWh)	475.6	289.5	515.3	404.6	1212	190.6
BCA	0.06	0.58	0.04	0.44	0.05	0.41

Scenario (a): Base case with no DER resources; Scenario (g): Maximum PV-only deployment (2 MW), without DR.

presents a comparative analysis of the effectiveness of the NWA portfolios consisting of ESS and PV resources across three representative case studies. In Cases 1 and 3, the integration of PVs results in reductions in both the PCS and SOC capacities compared to the scenarios without PV. Notably, in Case 3, the PCS capacity decreases from 3.1 MW to 2.0 MW, while the SOC capacity decreases from 37.2 MWh to 18.6 MWh. These findings suggest that PV resources can directly offset daytime loads during periods of high solar irradiance, thereby reducing the required ESS capacity.

In addition, the total ESS charging energy is significantly reduced in all the cases with PV integration. For instance, in Case 3, the total charging energy decreased by approximately 84%, from 1212 MWh to 190.6 MWh, thereby reducing the operational burden on the ESS. However, in Case 2, which reflects a residential load profile with peak demand in the evening hours when solar irradiance is limited, the integration of PVs has a minimal effect on reducing the ESS PCS and SOC requirements.

From a BCA perspective, the impact of PV is greater than that of DR. Unlike DR, which depends on load characteristics, PV operates as an independent generation resource, complementing the ESS performance and enhancing the economic viability of the system. Overall, the PV is a key resource for enhancing the cost-effectiveness of NWA portfolios. However, its impact on reducing the ESS capacity is significantly influenced by the underlying load profile.

The trends in BCA values for the previously discussed scenarios are illustrated in Figure 7, Figure 8 and Figure 9. In all three cases, portfolio (i)—which represents the lowest annual load growth rate and the highest level of NWA resource deployment—consistently produced the highest BCA values. However, the number of scenarios with the BCR exceeding 1 varied across cases: six in Case 1, four in Case 2, and five in Case 3. These results suggest that the economic effectiveness of a given NWA configuration depends on specific load characteristics.

In summary, reducing the ESS SOC and PCS capacities through the integration of PV and DR while simultaneously leveraging the generation benefits of PV can help mitigate low BCR values often associated with ESS-only configurations.

5. Conclusions

This study analyzes the impact of integrating PV and DR resources into the NWA portfolio on the sizing and economics of ESS, considering various load characteristics and expected load growth scenarios. Furthermore, three representative load profiles—one characterized by a balanced load mix, another dominated by residential consumption, and the third dominated by commercial and industrial loads—are examined.

The results indicate that the coordinated integration of PV and DR resources significantly reduces the PCS and SOC capacity requirements across all scenarios, with the greatest effect observed in the scenario dominated by commercial and industrial loads. These findings suggest that PVs and DRs are effective in mitigating peak loads and reducing the operational burden on the ESS, thereby enabling more efficient coordinated integration. Furthermore, an economic evaluation based on the BCA shows that scenarios fully utilizing the potential of PV and DR outperform those relying only on the ESS, particularly for systems with lower load factors. Although DR significantly contributes to capacity reduction, its impact on the overall BCA performance is limited. By contrast, PV serves as a distributed generation resource and a means to offset ESS capacity requirements, resulting in a greater positive effect on economic performance.

However, the effectiveness of DRs and PVs is sensitive to the underlying load profiles. Specifically, the benefits of the DR decrease in scenarios with extended peak durations, where commercial and industrial loads predominate. Conversely, the contribution of PVs is maximized under conditions of substantial daytime load but is limited in cases where peak loads occur during evening hours.

The results of this study suggest that the strategic integration of PV or DR resources tailored to specific load compositions and growth characteristics can substantially reduce the ESS investment and enhance the overall system efficiency. Future research should focus on improving the performance and economics of NWA portfolios through the development of advanced DR control strategies, application of high-resolution load forecasting techniques, and expanded integration of renewable energy resources.