Advanced Metering and Demand-Side Management

Introduction and Context

Electricity systems were traditionally operated according to one-way power flow: generators produced electricity, networks delivered it, and electromechanical meters recorded monthly consumption. Smart grids require a more interactive arrangement. Distributed generation, electric vehicles, storage, flexible loads, time-varying renewable output, and competitive electricity markets all require timely information and the ability to influence demand.

Advanced metering infrastructure (AMI) provides the communications, measurement, data-management, and control foundation for this interaction. Demand-side management (DSM) uses information, prices, incentives, automation, and efficiency measures to modify electricity consumption. Together, AMI and DSM can reduce peak demand, improve asset utilization, integrate renewable energy, reduce losses, and provide customers with better control over energy costs.

AMI is not merely a replacement for manual meter reading. It is a distributed cyber-physical system that connects meters, communication networks, utility applications, markets, customers, and regulators. Its design must therefore address engineering performance, economics, cybersecurity, privacy, reliability, interoperability, and social acceptance.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Explain the architecture and functions of AMI.
2. Distinguish AMI from automated meter reading and advanced metering infrastructure applications.
3. Describe meter-data management, validation, estimation, and editing.
4. Analyze dynamic tariffs and basic demand-response mechanisms.
5. Explain direct load control, automated demand response, and energy-efficiency programs.
6. Evaluate customer engagement, privacy, cybersecurity, and governance requirements.
7. Apply AMI and DSM concepts to a practical distribution-system case.

1. Advanced Metering Infrastructure Architecture

1.1 From conventional metering to AMI

Automated meter reading (AMR) generally supports one-way collection of periodic meter readings. AMI supports two-way communication, more frequent measurements, remote configuration, event reporting, outage notification, and in some systems remote service switching. The distinction is functional rather than purely technological: an AMI system must support operational and customer applications beyond periodic reading.

A typical AMI architecture contains five interacting layers:

Figure: Functional layers of an advanced metering infrastructure system.

1.2 Smart-meter functions

A smart meter commonly measures cumulative energy, interval energy, instantaneous demand, voltage, power factor, and supply interruptions. It may also maintain a time-of-use register, detect tampering, record phase imbalance, and communicate alarms. Measurement accuracy depends on the meter class, current-transformer characteristics, installation, temperature, time synchronization, and calibration.

The interval energy recorded during interval i is related to average power by:

where E_i is interval energy in kWh, P̄_i is average real power in kW, and Δt_i is interval duration in hours. This relationship assumes power is represented adequately by its interval average. For a 15-minute interval, Δt_i = 0.25 h; therefore a constant 4 kW load uses 1 kWh.

Meters must also distinguish energy import from export when customers have photovoltaic generation. Revenue-grade metering, bidirectional registers, and suitable tariff rules are required when exported energy is compensated.

1.3 Communications and interoperability

Communication choices include radio-frequency mesh, licensed radio, cellular networks, power-line communication, fiber, and hybrid arrangements. Selection depends on terrain, density, latency, ownership, installation cost, reliability, spectrum, and security. No single communication technology is universally optimal.

Interoperability requires common data models, standardized interfaces, secure authentication, clock synchronization, and clear ownership of data. Standards such as DLMS/COSEM, IEC 61850-related information models, and application-programming interfaces may support integration, but conformance testing and precise implementation profiles remain essential.

2. Meter-Data Management

2.1 The meter-data lifecycle

Meter data passes through a lifecycle: sensing, timestamping, local storage, transmission, ingestion, validation, estimation, editing, aggregation, retention, access, and eventual deletion. A meter-data management system (MDMS) provides the logical system of record for interval data, although billing, outage management, geographic information, and customer-information systems may retain related copies.

A sound MDMS should support:

• interval and register data;
• meter and service-point relationships;
• time-zone and daylight-saving treatment;
• versioned corrections and audit trails;
• quality flags and estimation methods;
• aggregation by feeder, transformer, customer, or tariff;
• role-based access and retention policies.

2.2 Validation, estimation, and editing

Raw data cannot be assumed correct. A practical validation, estimation, and editing (VEE) process may include range tests, completeness tests, zero-consumption tests, comparison with neighboring intervals, meter-status checks, and cumulative-register consistency checks.

If an interval is missing, a simple estimate can use adjacent valid intervals:

Here Ê_t is the estimated energy for time t, and E_t−1 and E_t+1 are neighboring valid interval values. This method is appropriate only when the load is reasonably smooth and neither neighboring interval is affected by an event. Weather-sensitive or highly variable loads require more suitable methods, such as temperature-based regression, customer-specific historical profiles, or model-based estimation.

The energy-balance relationship is useful for distribution analysis:

The approximation reflects measurement error, timing differences, distributed generation, and estimation. Large unexplained residuals can indicate faulty meters, incorrect connectivity, theft, phase errors, or data-quality problems.

2.3 Data quality and forecasting

Important data-quality dimensions are completeness, accuracy, timeliness, consistency, uniqueness, and provenance. Forecasting applications may use interval histories, weather, calendar variables, occupancy proxies, and distributed-energy-resource output. Artificial intelligence can improve anomaly detection and short-term load forecasting, but models must be monitored for bias, concept drift, explainability, and inappropriate automated decisions. AI should complement engineering validation rather than replace it.

3. Tariffs and Economic Signals

A tariff converts electricity-system costs and policy objectives into customer charges or payments. Common forms include flat energy tariffs, block tariffs, time-of-use (TOU) tariffs, critical-peak pricing, real-time pricing, demand charges, and incentive payments.

For a simple two-period TOU tariff, customer cost is:

where C is the billing-period cost, p_o and p_p are off-peak and peak prices in currency/kWh, E_o and E_p are off-peak and peak energy in kWh, and F is a fixed charge. The equation assumes energy charges are the only variable charges and excludes taxes, demand charges, and export compensation.

Dynamic pricing can improve efficiency when customers or automated devices respond to prices. However, price variation transfers some risk to customers. Consumer protection may require advance notice, bill protection, opt-out provisions, default safeguards, and accessible communication.

Worked example: tariff response

A household uses 30 kWh during a peak period and 70 kWh off peak. The peak price is 0.30 currency units/kWh and the off-peak price is 0.12 currency units/kWh. Its variable cost is:

C = (30 × 0.30) + (70 × 0.12) = 9.00 + 8.40 = 17.40 currency units.

If 10 kWh of peak use is shifted to off peak, peak use becomes 20 kWh and off-peak use becomes 80 kWh. The new cost is:

C_new = (20 × 0.30) + (80 × 0.12) = 6.00 + 9.60 = 15.60 currency units.

The saving is 1.80 currency units, provided that shifting does not create additional consumption or inconvenience. The calculation does not claim that every appliance can be shifted; thermal comfort, process constraints, and rebound effects must be considered.

4. Demand Response and Load Control

Demand response (DR) is a change in electricity consumption from a baseline in response to prices, incentives, reliability needs, or direct instructions. It differs from energy efficiency: DR changes the timing or level of use, whereas efficiency provides the same service with less energy over an extended period.

The basic response can be represented as:

where B_t is baseline demand at time t and L_t is actual demand. Positive DR_t indicates reduction relative to the baseline. Baseline construction is difficult because the unobservable counterfactual demand must be estimated. Weather, occupancy, production, prior events, and customer behavior affect accuracy.

A useful performance measure is peak reduction:

where the two powers are measured or estimated at the same system condition. Baseline error can materially distort reported performance.

4.1 DR categories

Price-based DR includes TOU, critical-peak, and real-time prices. Customers decide how to respond, manually or through automation.

Incentive-based DR provides payments for availability, measured reduction, or emergency curtailment. Examples include interruptible contracts, capacity programs, and demand bidding.

Direct load control allows a utility or authorized aggregator to cycle or disconnect selected equipment, such as water heaters, air conditioners, or irrigation pumps. Customers must understand control duration, override rights, and compensation.

Automated demand response uses a signal, an energy-management system, and predefined customer rules. It is especially useful in commercial buildings and industrial facilities where manual response is slow.

4.2 Load-control technologies

A control strategy may shift, shed, modulate, or store load:

• Shift: move electric-vehicle charging to a later period.
• Shed: temporarily switch off nonessential lighting.
• Modulate: adjust HVAC setpoints or motor speed.
• Store: charge a battery during low-price periods and discharge during peaks.
• Substitute: use thermal storage or another energy source.

Control quality involves response time, rebound, comfort, process safety, availability, verification, and customer override. A system that reduces demand briefly but creates a larger later rebound may not reduce the daily peak.

5. Energy Efficiency and Customer Engagement

Energy efficiency measures reduce energy required for a service. Examples include LED lighting, insulation, efficient motors, heat pumps, variable-speed drives, improved controls, and building commissioning. AMI enables feedback, benchmarking, appliance-level disaggregation, and measurement and verification, but the meter alone does not produce efficiency.

A basic energy-saving calculation is:

where S is annual energy saving in kWh, P_old and P_new are equipment powers in kW, and h is annual operating time in hours. It assumes constant rated powers and identical service output; actual savings should account for operating schedules, controls, degradation, and interaction effects.

Customer engagement is more effective when information is timely, understandable, actionable, and trusted. Useful applications include:

• hourly or daily usage displays;
• comparison with a customer’s own historical use;
• alerts for unusual consumption or high prices;
• carbon-intensity information;
• appliance scheduling;
• personalized efficiency recommendations;
• prepaid or budget-management tools;
• community-energy and electric-vehicle coordination.

Disaggregation algorithms infer appliance behavior from aggregate meter data. These estimates are probabilistic and should not be presented as direct measurements unless supported by additional sensors. Customers should be able to correct inaccurate assumptions and control how recommendations are used.

6. Privacy, Cybersecurity, and Governance

Interval data can reveal occupancy patterns, sleeping periods, appliance operation, medical-equipment dependence, and business activity. Privacy therefore requires more than securing the communications link.

Important principles include data minimization, purpose limitation, informed consent, access control, retention limits, transparency, correction rights, and accountable sharing. Aggregated data may still permit re-identification when the population is small or behavior is distinctive. Techniques such as temporal aggregation, masking, access auditing, and privacy-preserving analytics can reduce risk, although they may reduce analytical value.

Cybersecurity controls should include secure device identity, strong authentication, encryption in transit and at rest, signed firmware, secure boot, key rotation, network segmentation, vulnerability management, incident response, backup, and recovery testing. Availability is critical: an attacker who disables meters or sends simultaneous load-control commands can create operational risk even without stealing data.

Governance must assign responsibility among the distribution utility, retailer, aggregator, meter manufacturer, cloud provider, customer, and regulator. Contracts should specify permitted uses, breach notification, subcontracting, data return, deletion, audit rights, and service continuity.

7. Applied Case: Illustrative Urban Feeder Program

Background

Consider an illustrative utility serving 40,000 customers on an urban distribution feeder. Summer air-conditioning demand creates a 20 MW evening peak. The utility plans to install AMI and avoid constructing a new transformer bank costing 4 million currency units.

Problem

The utility lacks interval visibility, has uncertain outage locations, and relies on broad emergency appeals. It must reduce the feeder peak without compromising medical customers, thermal comfort, or privacy.

Analysis

The utility identifies 12,000 households with controllable air-conditioning or water-heating loads, 3,000 electric vehicles, and 150 commercial buildings. It estimates that a diversified portfolio could provide:

The values are planning estimates, not guaranteed simultaneous reductions. Diversity, communication failures, weather, and customer override must be included in a reliability assessment.

Approach and solution

The utility deploys interval meters, a secure head-end, an MDMS, an outage-management interface, and a customer portal. It offers a TOU tariff, a critical-event rebate, and an opt-in automated-control program. Customers choose comfort limits and override rules. Medical and vulnerable customers receive protected status and alternative assistance.

Before each event, the utility estimates a customer baseline using comparable non-event days adjusted for weather. During events, it records actual interval demand, communications status, overrides, and rebound. Data are retained according to a published policy, and external researchers receive aggregated rather than customer-identifiable data.

Outcome

In the first season, the program produces a measured average reduction of 5.6 MW during the targeted peak, below the nominal 7.7 MW resource total but sufficient to defer the transformer project. Customer surveys show that automated controls are accepted when notification and override options are clear. The utility discovers that some meters have clock drift, causing billing and event-measurement errors; synchronization and quality checks are added.

Learning points

1. Resource nameplate totals are not the same as dependable peak reduction.
2. Baseline design is central to fair DR settlement.
3. Customer protection and usability affect participation as much as technology.
4. AMI value arises from integrated data and applications, not meters alone.
5. Privacy and cybersecurity must be designed before deployment rather than added afterward.

8. Challenges, Limitations, and Best Practices

AMI and DSM programs face high capital costs, vendor lock-in, rural communication gaps, meter replacement issues, interoperability problems, and uncertain customer participation. Dynamic tariffs can disadvantage customers who lack automation, flexible schedules, reliable internet, or suitable housing. Efficiency programs may produce rebound effects, while DR may shift rather than eliminate emissions or system stress.

Best practice is to begin with clear operational and customer outcomes, conduct a cost-benefit and equity assessment, pilot representative neighborhoods, define data-quality and cybersecurity requirements, test interoperability, and establish independent evaluation. Programs should report both gross and dependable reductions, customer impacts, opt-out rates, complaints, and distributional effects. Human review remains essential for safety-critical control and consequential customer decisions.

Contemporary Developments

Current development is moving toward grid-interactive efficient buildings, managed electric-vehicle charging, vehicle-to-grid operation, behind-the-meter batteries, transactive or flexibility markets, and distribution-level flexibility procurement. Edge computing can allow local decisions when communications fail. Federated or privacy-preserving learning may enable analytics without centralizing detailed customer data. These developments increase opportunity but also complicate coordination, market rules, interoperability, and responsibility for control actions.

Chapter Summary

AMI combines smart meters, communications, head-end systems, meter-data management, utility applications, and customer interfaces into a two-way cyber-physical platform. Its data must undergo validation, estimation, editing, secure storage, and controlled use. Dynamic tariffs and demand response can change when or how customers consume electricity, while energy efficiency reduces the energy required for a service. Load control may shift, shed, modulate, store, or substitute demand, but dependable performance depends on baselines, diversity, rebound, comfort, and override behavior. Customer engagement is strongest when information is actionable and transparent. Because interval data can reveal personal and commercial behavior, privacy, cybersecurity, governance, and equity are fundamental design requirements. Successful programs integrate technical performance with customer protection, measurement, and continuous evaluation.

Key Terms

Advanced metering infrastructure (AMI): Integrated meters, communications, data systems, and applications supporting two-way measurement and control.

Automated meter reading (AMR): Usually one-way collection of periodic meter readings.

Meter-data management system (MDMS): System that stores, validates, estimates, edits, and supplies meter data.

Interval data: Energy or demand recorded for defined time periods.

Head-end system: Platform that communicates with and manages field meters.

Time-of-use tariff: Tariff with prices that vary by predefined time periods.

Real-time pricing: Pricing that varies with current or near-current system conditions.

Demand response: Change in electricity use relative to an expected baseline.

Baseline: Estimated consumption that would have occurred without a DR event.

Direct load control: Authorized remote control of selected customer equipment.

Automated demand response: Rule-based or signal-based automatic customer response.

Energy efficiency: Reduction in energy required to provide a service.

Peak demand: Highest power demand over a specified period or system boundary.

Load shifting: Moving consumption from one time to another.

Rebound effect: Increase in later demand following a reduction or control event.

Data minimization: Collection and retention of only data necessary for a defined purpose.

Privacy-preserving analytics: Analysis designed to limit disclosure of identifiable information.

Interoperability: Ability of systems and devices to exchange and use information correctly.

Demand baseline error: Difference between estimated counterfactual use and actual would-have-used demand.

Distributed energy resource: Small-scale generation, storage, or controllable load connected to the distribution system.

Review Questions

Short Answer Questions

1. What distinguishes AMI from conventional automated meter reading?
2. State two functions of an MDMS.
3. Define demand response and explain the role of a baseline.
4. Give two examples of direct load control.
5. Why can interval meter data create privacy risks?

Long Answer Questions

1. Explain the layers of an AMI architecture and the information exchanged among them.
2. Discuss validation, estimation, and editing in meter-data management, including limitations of simple interpolation.
3. Compare time-of-use pricing, real-time pricing, and incentive-based demand response.
4. Explain how customer engagement and automation can support energy efficiency and load flexibility.
5. Evaluate the cybersecurity and governance requirements of a large-scale AMI deployment.

Analytical / Application Questions

1. A customer uses 24 kWh at a peak price of 0.28 currency units/kWh and 56 kWh off peak at 0.10 currency units/kWh. If 8 kWh is shifted from peak to off peak, calculate the original cost, new cost, and saving, ignoring fixed charges.
2. A feeder baseline peak is 18 MW and event demand is 15.3 MW. Calculate the percentage peak reduction and identify two reasons why the measured result may not equal dependable long-term capability.
3. Design a privacy-aware DR pilot for a mixed residential and commercial feeder. Specify the tariff or incentive, control technologies, baseline approach, customer protections, cybersecurity controls, and evaluation metrics.

Multiple Choice Questions with Answers

Question 1

Which feature most clearly distinguishes AMI from basic automated meter reading?

A. One-way monthly collection only

B. Two-way communication supporting operational and customer applications

C. Manual inspection of electromechanical meters

D. Use of printed billing statements

Correct Answer: B. Two-way communication supporting operational and customer applications

Question 2

A constant 4 kW load operates for 15 minutes. How much energy does it consume?

A. 0.25 kWh

B. 1 kWh

C. 4 kWh

D. 15 kWh

Correct Answer: B. 1 kWh

Question 3

What is the principal purpose of a meter-data management system?

A. Generate electricity

B. Store, validate, estimate, and supply meter data

C. Replace all distribution transformers

D. Control household behavior without authorization

Correct Answer: B. Store, validate, estimate, and supply meter data

Question 4

Under a time-of-use tariff, which action normally reduces a customer’s bill without reducing total service?

A. Shift consumption from peak to off peak

B. Increase peak consumption

C. Ignore the tariff schedule

D. Disable meter time synchronization

Correct Answer: A. Shift consumption from peak to off peak

Question 5

Demand response is normally measured relative to which quantity?

A. Transformer nameplate rating only

B. A customer’s estimated baseline demand

C. Annual utility revenue

D. The meter’s serial number

Correct Answer: B. A customer’s estimated baseline demand

Question 6

Which activity is an example of load modulation?

A. Moving EV charging to midnight

B. Switching off all customer supply permanently

C. Adjusting an HVAC setpoint during a peak event

D. Replacing a meter with a printed invoice

Correct Answer: C. Adjusting an HVAC setpoint during a peak event

Question 7

Which statement best describes energy efficiency?

A. It always shifts demand to another hour

B. It provides the same service using less energy

C. It is identical to emergency load shedding

D. It requires real-time pricing in every case

Correct Answer: B. It provides the same service using less energy

Question 8

Why is interval electricity data potentially sensitive?

A. It cannot be time-stamped

B. It may reveal occupancy and appliance-use patterns

C. It contains no information about behavior

D. It is useful only for transformer manufacturing

Correct Answer: B. It may reveal occupancy and appliance-use patterns

Question 9

If a feeder baseline is 20 MW and event demand is 17 MW, what is the percentage reduction?

A. 3%

B. 15%

C. 17%

D. 85%

Correct Answer: B. 15%

Question 10

Which is an appropriate privacy principle for AMI data?

A. Collect and retain every possible data item indefinitely

B. Share identifiable interval data publicly by default

C. Limit collection and use to defined legitimate purposes

D. Allow unauthenticated control commands

Correct Answer: C. Limit collection and use to defined legitimate purposes