Importance of Aggregated DER Installed Capacity in Distribution Networks

1. Problem Statement

The rapid penetration of Distributed Energy Resources (DERs)—specifically Electric Vehicles (EVs), Heat Pumps (HPs), and Photovoltaic (PV) systems—is fundamentally altering power flows in Low-Voltage (LV) distribution networks. However, Distribution System Operators (DSOs) face a critical observability gap:

• Data Scarcity: DSOs often lack reliable, up-to-date knowledge of the specific DER capacity connected downstream of LV substations.
• Limitations of Current Data: Existing metadata (from connection applications or registries) degrades over time due to device replacements, ownership changes, and undocumented upgrades. Furthermore, privacy regulations and incomplete topology models prevent access to granular, customer-level data.
• Consequences: This lack of visibility forces DSOs to rely on simplified assumptions (e.g., "greenfield" scenarios) or conservative estimates in planning and operations. This leads to suboptimal decision-making in forecasting, congestion management, flexibility quantification, and hosting capacity assessment, potentially resulting in over-investment or underestimated grid risks.

2. Methodology

The paper proposes a framework for estimating Aggregated DER Installed Capacity at LV aggregation points (e.g., secondary substations or feeder heads) without requiring customer-level monitoring.

• Core Concept: Instead of reconstructing individual installations, the objective is to estimate the total installed capacity and technology mix connected downstream of a measurement point.
• Mathematical Formulation:
  The problem is modeled using the net load ($P_{Net}$) measured at an aggregation point:
  $$P_{Net} = P_{Load} - P_{Gen}$$
  Where:
  • $P_{Load}$ is the total consumption, modeled as a function of traditional load ($P_{NonDER}$) and DER loads (EVs, HPs, BESS) dependent on installed capacities ($C_{V1G}, C_{HP}, C_{BESS}$) and exogenous parameters ($\alpha$, e.g., weather).
  • $P_{Gen}$ is the total generation, modeled as a function of DER generation (PV, V2G, BESS) dependent on installed capacities ($C_{PV}, C_{V2G}, C_{BESS}$) and external parameters ($\beta$, e.g., irradiance).
• Estimation Approaches:
  • Model-Based: Explicitly defining functions $f(\cdot)$ and $g(\cdot)$ based on physical principles.
  • Data-Driven: Using machine learning (ML) and optimization to implicitly learn these relationships from historical data.
• Data Sources: The method utilizes data already available to DSOs, such as substation power flows, feeder measurements, and aggregated smart-meter data. It leverages the characteristic temporal signatures of different DER technologies (e.g., PV generation correlates with irradiance; HP demand correlates with temperature) to disaggregate the total load.
• Relation to NILM: The approach is analogous to Non-Intrusive Load Monitoring (NILM) but applied at the feeder/substation level rather than the individual appliance level.

3. Key Contributions

The paper makes three primary contributions to the field of distribution network management:

1. Definition of a Scalable Solution: It establishes aggregated DER installed capacity as a practical, scalable metric that bridges the gap between limited observability and the need for detailed system knowledge.
2. Operational and Planning Use Cases: It systematically maps how aggregated capacity data improves specific DSO tasks:
   • Forecasting: Improving net load prediction accuracy by explicitly modeling DER behaviors rather than treating them as noise.
   • Congestion Management (CM): Enabling proactive identification of congestion risks based on the specific mix of DERs (e.g., synchronized HP demand during cold snaps).
   • Flexibility Quantification (FQ): Providing accurate baselines for estimating available flexibility envelopes (e.g., how much EV charging can be shifted).
   • Hosting Capacity Assessment (HCA): Shifting HCA from hypothetical "greenfield" scenarios to condition-based evaluations that account for existing DERs.
   • Growth Monitoring: Tracking the spatial evolution of electrification to identify clusters of rapid adoption or disparities in uptake.
3. Strategic Framework: It argues that periodic estimation of aggregated capacity is sufficient for planning and operational support, as DER capacity evolves slowly (months/years) compared to real-time power measurements.

4. Results and Evidence

The paper synthesizes findings from existing literature and theoretical analysis to demonstrate the efficacy of this approach:

• Forecasting Accuracy:
  • Studies cited show that incorporating PV capacity estimates into forecasting models reduced errors by 11% compared to direct net load estimation.
  • Disaggregating BTM PV and load for independent forecasting reduced RMSE by 13–16% compared to direct LSTM forecasting.
  • A DER-aware CNN-LSTM model achieved 93.70% accuracy versus 89.32% for a DER-agnostic model.
• Congestion Management: Accurate nodal load forecasts (enabled by DER awareness) have been shown to reduce congestion management costs by up to 97%.
• Flexibility Sensitivity: Research indicates that installed DER capacity is the most sensitive parameter in flexibility quantification, particularly for EVs.
• Planning Impact: By moving from "greenfield" assumptions to metadata-aware models, DSOs can avoid systematic over- or underestimation of hosting capacity, leading to more efficient asset sizing and reinforcement decisions.

5. Significance

The significance of this work lies in its practical applicability for the energy transition:

• Cost-Effective Observability: It offers a path to high-quality grid visibility without the prohibitive cost and privacy hurdles of installing sensors on every customer asset.
• Informed Decision-Making: It transforms DSO operations from reactive to proactive. With accurate aggregated data, DSOs can better anticipate peak loads, manage voltage constraints, and optimize the dispatch of flexibility resources.
• Equitable Energy Transition: By enabling geographically resolved monitoring of DER adoption, the method helps identify disparities in electrification (e.g., low uptake in specific demographics despite grid capacity). This supports policymakers and regulators in designing targeted incentives and ensuring a just energy transition.
• Future-Proofing: As the grid becomes more decentralized, the ability to estimate and monitor the composition of the load (rather than just the total magnitude) becomes a fundamental requirement for grid stability and market efficiency.

In conclusion, the paper argues that Aggregated DER Installed Capacity is a critical, yet often overlooked, metadata layer. Its estimation and regular updating are essential for modernizing distribution network planning and operations in the face of increasing DER penetration.