PaperlayerView exact revision on arXiv

Source-linked AI summary

Modeling and Validation of Electrical Load Profiling in Residential Buildings in Singapore

Luo Chuan, Abhisek Ukil

arXiv:1508.02353v1eess.SYcs.SY

TL;DR

Household electricity-load profiles can support smart-grid technologies, but Singapore’s available consumption data are aggregated rather than appliance-specific. The paper models daily loads for four public-housing types using a bottom-up approach and reports successful validation against EMA statistics and NTU campus measurements. Its measured-data validation primarily represents three-room public housing.

  • Problem

    Singapore’s available residential consumption data aggregate multiple households without showing appliance composition or individual household utilization patterns, although household-load knowledge matters for smart-grid technologies and system planning.

    Source: §I · p. 1, §III · p. 3

  • Method

    A bottom-up model simulates daily loads for four public-housing types by estimating appliance use and accumulating appliance-level consumption into household profiles.

    Source: B. Mathematical Model for Household… · p. 4, §VII · p. 9

  • Results

    The simulated profiles tally with EMA statistics and match NTU CEP measurements; the monitored NTU average was 220 kWh per unit versus a 216 kWh average for 1–2-room and 3-room public housing.

    Source: B. Validation · p. 7, §VI · p. 8

  • Takeaways & Limitations

    The authors state that detailed household-consumption modeling could help future research on energy-efficient building operations within the smart-grid initiative.

    Source: §VII · p. 9

    Measured-data validation uses NTU campus housing, mostly three-room units, so the authors say it should be considered representative of three-room public housing.

    Source: §VI · p. 8

Abstract

from arXiv · show

The demand of electricity keeps increasing in this modern society and the behavior of customers vary greatly from time to time, city to city, type to type, etc. Generally, buildings are classified into residential, commercial and industrial. This study is aimed to distinguish the types of residential buildings in Singapore and establish a mathematical model to represent and model the load profile of each type. Modeling household energy consumption is the first step in exploring the possible demand response and load reduction opportunities under the smart grid initiative. Residential electricity load profiling includes the details on the electrical appliances, its energy requirement, and consumption pattern. The model is generated with a bottom-up load model. Simulation is performed for daily load profiles of 1 or 2 rooms, 3 rooms, 4 rooms and 5 rooms public housing. The simulated load profile is successfully validated against the measured electricity consumption data, using a web-based Customer Energy Portal (CEP) at the campus housings of Nanyang Technological University, Singapore.

I. INTRODUCTION

Household load profiling is presented as important for Singapore’s smart-grid development and electricity-system planning. The study classifies residential building types and models their loads, then validates the profiles against measured data.

I. INTRODUCTION

  • Household load profiling supports smart-grid technologies and is important for distribution-network planning and optimal production-capacity decisions.The introduction also connects accurate household-load knowledge with sizing small-scale distributed energy technologies and demand-side management.Source: §I · p. 1
  • Buildings consumed 93.3% of Singapore’s total electricity in 2013, spanning residential, commercial, and industrial sectors.Residential buildings include public and private housing.Source: §I · p. 1
  • Households accounted for 47.3% of non-contestable load, equivalent to 15.6% of total load.Commercial buildings used 37.9% of consumed energy, while industrial activities used 39.8%.Source: §I · p. 1
  • The study classifies residential buildings, models each type’s load from elementary components, and validates the modeled profiles using measured data.Its stated objectives include studying average loads by residential type and constructing profiles from appliance-level components.Source: §I · p. 1

II. LITERATURE REVIEW

The review describes Singapore’s residential electricity consumption by dwelling type and its relationship with ambient temperature. Because annual temperatures are relatively stable, the study excludes temperature effects from its load model.

A. Residential Building Electricity Consumption in Singapore

B. Effect of Ambient Temperature

  • Monthly residential consumption broadly tracked ambient temperature in 2011, peaking in June and reaching its lowest levels in February and March.Fig. 3 compares monthly electricity consumption with ambient temperature.Source: B. Effect of Ambient Temperature · p. 2, Figure 3 · p. 3
  • The study excludes temperature effects because Singapore’s tropical-marine climate has a relatively stable, constant temperature profile throughout the year.The reported consumption fluctuation between peak and bottom was around 50 kWh and was not considered significant.Source: B. Effect of Ambient Temperature · p. 2

C. Bottom-Up Approach for Modeling

The review contrasts utility-level forecasting with appliance-based bottom-up modeling. The study follows the latter approach, which builds household load from individual appliance loads, while using representative samples and statistical averages to address data needs.

C. Bottom-Up Approach for Modeling

  • Bottom-up modeling builds total household load from elementary components, allowing analysis of each appliance’s effect on the total load.The study follows this approach, whose procedure is shown in Fig. 4.Source: C. Bottom-Up Approach for Modeling · p. 3, Figure 4 · p. 3
  • Prior work used bottom-up end-user models to simulate household consumption, while other reviewed studies focused on utility-level forecasting methods.The forecasting approaches described include neural networks, fuzzy logic, genetic algorithms, and expert systems.Source: C. Bottom-Up Approach for Modeling · p. 3
  • The procedure estimates hourly consumption for each appliance and accumulates those appliance curves into the household load curve.The appliance-load loop runs for a particular appliance, while the household loop repeats until all appliances are counted.Source: C. Bottom-Up Approach for Modeling · p. 3
  • Detailed bottom-up methods can require extensive consumer, appliance, and household data; this study bypasses that need with representative samples and statistical averages.The Capasso model requires detailed consumer-behavior data, while the Norwegian ERAD model also requires detailed house-design information.Source: C. Bottom-Up Approach for Modeling · p. 3

III. RESIDENTIAL BUILDING LOAD PROFILE

The study uses a bottom-up model to describe residential electricity loads by household type, addressing the lack of appliance-level composition and usage information in aggregated Singapore data. It treats seasonal variation as insignificant in Singapore and excludes weekday/weekend differences and uncertain events from the model.

III. RESIDENTIAL BUILDING LOAD PROFILE

  • Aggregated Singapore residential electricity data do not reveal individual household events or how appliance use composes the load.The study therefore simulates public-household loads by house type using a statistical bottom-up approach that considers each appliance.Source: §III · p. 3

A. Factors Influencing Energy Utilization

  • Prior studies associate yearly daily-use patterns with temperature and daylight, but this study excludes seasonal effects because those patterns change only slightly in Singapore.The authors characterize the resulting yearly consumption change as insignificant.Source: A. Factors Influencing Energy… · p. 4
  • The model ignores weekday/weekend load differences, although prior studies relate hourly fluctuations to consumer availability and activity.The authors defer this discrepancy to future work because of time and resource constraints.Source: A. Factors Influencing Energy… · p. 4
  • Abnormal weather and national or international events are excluded; the model instead targets average, universal residential consumption behavior.The passage gives popular television shows and commercial breaks as examples of events that can affect consumption.Source: A. Factors Influencing Energy… · p. 4

B. Mathematical Model for Household Electricity Consumption

The household load model builds a daily profile by simulating appliance activations and summing their consumption. Saturation levels represent appliance availability, while a starting-probability function governs when appliances turn on.

B. Mathematical Model for Household Electricity Consumption

  • Two repeating loops generate a one-day profile for each appliance and repeat the process until all household appliances are counted.The appliance loop generates an individual appliance curve; the household loop aggregates across the appliance list.Source: B. Mathematical Model for Household… · p. 4
  • Saturation level represents the probability that a particular appliance is available in a household, keeping availability statistics coherent across many households.The appliance list and saturation level together determine which appliances are represented in a household.Source: B. Mathematical Model for Household… · p. 4
  • Each appliance’s rated load is added during an activation cycle, and the household profile is the sum of individual appliance consumption.An appliance may be activated multiple times in a day according to its usage pattern.Source: B. Mathematical Model for Household… · p. 4
  • A starting-probability function determines appliance activation by time step, combining appliance saturation, hourly activity, daily starting frequency, and a step-size scaling factor.Higher hourly probability means a higher chance that an appliance is turned on.Source: B. Mathematical Model for Household… · p. 4

C. Simulation Procedure

The simulation checks each appliance for activation at every computational time step and adds standby use continuously where applicable. It is implemented in MATLAB under assumptions that simplify daily profiles and household behavior.

C. Simulation Procedure

  • At each time step, the model compares a random number with the appliance activation probability, turns the appliance off at the end of its cycle, and adds standby use continuously when applicable.The average on-cycle is taken from previous studies.Source: C. Simulation Procedure · p. 4
  • The daily average energy formula uses active and standby consumption parameters, appliance count, mean daily starting frequency, and average on-cycle.The passage identifies nominal and standby power in watts, total appliance count, starting frequency, and average on-cycle as formula quantities.Source: C. Simulation Procedure · pp. 4–5
  • The MATLAB simulation uses one representative day multiplied by 30 and assumes average on-cycles capture variable appliance behavior; weekday/weekend differences and time-of-use tariff effects are not modeled.Public-housing size is treated as an indicator of demographic and economic conditions, while extreme variations are excluded.Source: C. Simulation Procedure · p. 5

IV. LOAD PROFILE SIMULATION

The simulations use appliance characteristics and household-specific usage parameters to construct load profiles for different public-housing sizes. For one- or two-room flats, the modeled average is reported close to the EMA average.

IV. LOAD PROFILE SIMULATION

  • The simulation covers 1- or 2-room, 3-room, 4-room, and 5-room flats, with detailed profiles shown for the smallest and largest units and concise descriptions for 3- and 4-room flats.The 1- or 2-room profile is the lower-size example in the detailed presentation.Source: §IV · p. 5

A. Residential Appliance Data

  • Appliance inputs include household appliance lists and saturation, nominal and standby power, daily starting frequency, and operating time per cycle.Appliance saturation is treated as a reference and investigated further by household type.Source: A. Residential Appliance Data · p. 5

B. Load Profiling of One or Two Rooms Public Housing

C. Load Profiling of Five Rooms Public Housing

The modeled load profiles for five-room and three-room public flats are reported as close to EMA consumption figures.

C. Load Profiling of Five Rooms Public Housing

  • 447.8 kWh for the five-room flat is close to EMA’s 465 kWh average monthly consumption figure.The five-room profile is built using the reported starting frequency, saturation, and starting-probability data.Source: C. Load Profiling of Five Rooms Public… · p. 6

D. Load Profiling of Three Rooms Public Housing

  • 284.2 kWh for the three-room flat is close to EMA’s 278.2 kWh average monthly consumption figure.Only the daily load profile is shown; detailed three-room analysis is omitted for brevity.Source: D. Load Profiling of Three Rooms… · p. 6

E. Load Profiling of Four Rooms Public Housing

The four-room public-housing model produces an average load reported alongside EMA statistics, while NTU’s portal provides monitored household-use data at 30-minute intervals.

E. Load Profiling of Four Rooms Public Housing

A. Real-time Energy Portal at NTU, Singapore

  • NTU’s Customer Energy Portal monitored electricity use in 323 staff housing units every 30 minutes from June 2012.The units used smart meters; the portal’s automated metering infrastructure is shown in Fig. 9.Source: A. Real-time Energy Portal at NTU,… · p. 7

B. Validation

NTU’s measured average load is close to the cited public-housing average, and the paper reports similarities between modeled and measured profiles despite some pattern differences.

B. Validation

  • NTU’s measured profile has peaks at 07:30–08:30 and 20:30–21:30, with relatively low load in early morning and from 09:30 to 17:30.The text relates the morning and evening peaks to greater electricity-related activity when people are at home.Source: B. Validation · p. 7
  • The modeled three-room and measured NTU profiles have similar peak profiles and average energy use, though their patterns differ slightly by setting.The comparison is between public-housing Fig. 7 and university Fig. 10; the paper reports overall good similarity.Source: B. Validation · pp. 7–8, Figure 10 · p. 8
  • 220 kWh per unit in NTU monitored data is close to the 216 kWh average for 1–2-room and 3-room public housing combined.The average residential load in NTU on February 6, 2013, is shown in Fig. 10.Source: B. Validation · p. 7, Figure 10 · p. 8

VI. DISCUSSIONS

The discussion reports agreement with EMA statistics and NTU measurements, while noting the scope of the measured validation; the conclusion presents appliance-based load modeling for four dwelling types.

VI. DISCUSSIONS

  • NTU measurements support the modeled profiles, but the monitored campus units were mostly three-room flats and therefore serve as validation corresponding to that housing type.The discussion attributes the use of NTU data to the usual unavailability of detailed energy accounts for public housing across Singapore.Source: §VI · p. 8
  • Room count is associated with different load-profile patterns; five-room homes show midday peaks absent from the cited one- or two-room and three-room examples.The discussion links the five-room peaks to a higher probability of appliance use due to an expectedly higher number of inhabitants.Source: §VI · p. 9

VII. CONCLUSION

  • The study models 1–2-room, 3-room, 4-room, and 5-room housing using appliance characteristics, saturation by building type, power ratings, and use patterns.The models are simulated in MATLAB and compared with EMA average monthly usage statistics.Source: §VII · p. 9
  • The modeled profiles are reported as successfully verified against measured data from NTU’s web-based Customer Energy Portal.The conclusion identifies detailed household electricity-use modeling as potentially helpful to future research on energy-efficient building operations and smart grids.Source: §VII · p. 9
Loading 1508.02353v1…